SAS 2012 Proceedings

Proceedings for the 31st Annual Conference of the Society
for Astronomical Sciences
Joint Meeting of
The Society for Astronomical Sciences
The American Association of Variable Star Observers
Symposium on Telescope Science

Editors:
Brian D. Warner
Robert K. Buchheim
Jerry L. Foote
Dale Mais
May 22-24, 2012

Big Bear Lake, CA
Disclaimer
The acceptance of a paper for the SAS proceedings can not be used to imply nor should it be inferred as an endorsement by the Society for Astronomical Sciences or the American Association of Variable Star Observers of
any product, service, or method mentioned in the paper.
Published by the Society for Astronomical Sciences, Inc.

Rancho Cucamonga, CA
First printing: May 2012
Photo Credits:
Front Cover: NGC 7293, Alson Wong
Back Cover: Sagittarius Milky Way, Alson Wong
Table of Contents
PREFACE
CONFERENCE SPONSORS
II
SYMPOSIUM SCHEDULE
III
PRESENTATION PAPERS
PHOTOMETRY OF HUBBLES FIRST CEPHEID
IN THE ANDROMEDA GALAXY, M31
BILL GOFF, MATT TEMPLETON, RICHARD SABO, TIM CRAWFORD, MICHAEL COOK
BK LYNCIS: THE OLDEST OLD NOVA?
OR: ARCHAEO-ASTRONOMY 101
JONATHAN KEMP, JOE PATTERSON, ENRIQUE DE MIGUEL, GEORGE ROBERTS,

TUT CAMPBELL, FRANZ-J. HAMBSCH, TOM KRAJCI, SHAWN DVORAK,
ROBERT A. KOFF, ETIENNE MORELLE, MICHAEL POTTER, DAVID CEJUDO,
JOE ULOWETZ, DAVID BOYD, RICHARD SABO, JOHN ROCK, ARTO OKSANEN
PHOTOMETRIC MONITORING BY AMATEURS IN SUPPORT

OF A YY GEM PROFESSIONAL OBSERVING PROJECT
17
BRUCE L. GARY, DR. LESLIE H. HEBB, JERROLD L. FOOTE,

CINDY N. FOOTE, ROBERTO ZAMBELLI, JOAO GREGORIO,
F. JOSEPH GARLITZ, GREGOR SRDOC, TAKESHI YADA, ANTHONY I. AYIOMAMITIS
THE LOWELL AMATEUR RESEARCH INITIATIVE
25
DEIDRE ANN HUNTER, JOHN MENKE, BRUCE KOEHN,

MICHAEL BECKAGE, KLAUS BRASCH, SUE DURLING, STEPHEN LESHIN
LUNAR METEOR IMPACT MONITORING
AND THE 2013 LADEE MISSION
29
BRIAN CUDNIK
FIRST ATTEMPTS AT ASTEROID SHAPE MODELING
37
MAURICE CLARK
DIURNAL PARALLAX DETERMINATIONS OF ASTEROID DISTANCES

USING ONLY BACKYARD OBSERVATIONS FROM A SINGLE STATION
EDUARDO MANUEL ALVAREZ, ROBERT K. BUCHHEIM
45
ON THE MAXIMUM AMPLITUDE OF HARMONICS
OF AN ASTEROID LIGHTCURVE
59
ALAN HARRIS
PV CEPHEI AND GYULBUDAGHIANS VARIABLE NEBULA
65
DAVID BOYD
WHY VISUAL OBSERVATIONS OF MU CEPHEI ARE IMPORTANT
69
DAVID G. TURNER
ROAD (REMOTE OBSERVATORY ATACAMA DESERT):

INTENSIVE OBSERVATIONS OF VARIABLE STARS
75
FRANZ-JOSEF HAMBSCH
ER URSAE MAJORIS: A DWARF NOVA WITH SURPRISES
79
ENRIQUE DE MIGUEL, JOE PATTERSON, JONATHAN KEMP,

WILLIAM STEIN, GEORGE ROBERTS, T. CAMPBELL, FRANZ-J. HAMBSCH,
TOM KRAJCI, SHAWN DVORAK, ROBERT A. KOFF, ETIENNE MORELLE,
MICHAEL POTTER, DAVID CEJUDO, STEVE BRADY, KENNETH MENZIES
SMALL TELESCOPE SPECTROSCOPY OF EPSILON AURIGAE
89
JEFFREY L. HOPKINS
OBSERVATIONS USING A BESPOKE MEDIUM RESOLUTION FAST SPECTROGRAPH
97
JOHN MENKE
HIGH RESOLUTION SPECTROSCOPY FOR THE AMATEUR:

EXPERIENCES WITH THE LHIRES III SPECTROGRAPH
105
STANLEY A. GORODENSKI
EXTREMELY LOW-COST POINT-SOURCE SPECTROPHOTOMETRY (ELCPSS)
113
JOHN BEAVER, CHARLES CONGER
SPECTROSCOPIC ANALYSIS OF ALGOL DURING ECLIPSE CYCLE
121
JONATHAN BOYD, KODIAK DARLING, ELISE SPARKS, LAJEANA WEST, DOUGLAS WALKER
A FRESH LOOK AT THE ALGOL-LIKE ECLIPSING BINARY, AO SER
127
KEVIN B. ALTON, ANDREJ PRA
THE LIGHT AT NIGHT MAPPING PROJECT: LAN MAP 1, THE TUCSON BASIN
139
E.R. CRAINE, B.L. CRAINE, P.R. CRAINE, E.M. CRAINE
OBSERVING DOUBLE STARS
147
RUSSELL M. GENET, B.J. FULTON, FEDERICA B. BIANCO, JOHN MARTINEZ,

JOHN BAXTER, MARK BREWER, JOSEPH CARRO, SARAH COLLINS, CHRIS ESTRADA,
JOLYON JOHNSON, AKASH SALAM, VERA WALLEN, NAOMI WARREN,
THOMAS C. SMITH, JAMES D. ARMSTRONG, STEVE MCGAUGHEY,
JOHN PYE, KAKKALA MOHANAN, REBECCA CHURCH
AMATEUR IMAGE PIPELINE PROCESSING

WAYNE GREEN
USING PYTHON PLUS PYRAF
159
HIGH TIME RESOLUTION ASTRONOMY
OR HIGH SPEED PHOTOMETRY
165
GARY A. VANDER HAAGEN
PHOTON COUNTING ONE MORE TIME
177
RICHARD H. STANTON
TRACKING BOLIDES, 3D VISUALIZATION AND DATA
185
THOMAS G. KAYE, ROBERT CRAWFORD, MARK BOWLING, JOHN KALAS
TOOLS AND TECHNIQUES FOR MEASURING ASTEROID OCCULTATIONS

WITH DSLR AND CCD CAMERAS
191
JOHN E. HOOT
PAPERS WITHOUT PRESENTATION AND POSTERS
201
A VIRTUAL ASTRONOMICAL RESEARCH MACHINE IN NO TIME (VARMINT)
203
JOHN BEAVER
THE CONFUSING CASE OF 16666 LIROMA
209
ROBERT K. BUCHHEIM, JOHN RUTHROFF
STUDENT PROJECT AND CURRICULUM BASED ON

LIGHT AT NIGHT DATA COLLECTION
215
ERIN M. CRAINE, JENNIFER C. DEBENEDETTI
ENHANCING THE EDUCATIONAL ASTRONOMICAL EXPERIENCE

OF NON-SCIENCE MAJORS WITH THE USE OF AN IPAD AND TELESCOPE
219
ROBERT M. GILL, MICHAEL J. BURIN
THE ROTATIONAL PERIOD OF THE SUN

USING THE DOPPLER SHIFT OF THE HYDROGEN ALPHA SPECTRAL LINE
225
ROBERT M. GILL
FAST SPECTROMETER CONSTRUCTION AND TESTING
229
JOHN MENKE
ADILS: AN AMATEUR DUAL IMAGING LITTROW SPECTROGRAPH
233
WAYNE GREEN
HIGH ACCURACY RA/DEC ENCODERS FOR A CA 1890 GEM
234
WAYNE GREEN
A SINGLE BEARM POLARIMETER

JERRY D. HORNE
235
Preface
Preface
We welcome the American Association of Variable Star Observers (AAVSO) to the 31st annual SAS Symposium on Telescope Science. The first joint meeting between the Societry for Astronomical Sciences and the
AAVSO, held in 2009, was a tremendous success, bringing in presenters from around the world and in record
numbers. Current plans are for the AAVSO to make this a biannual affair, merging its annual Spring meeting in
even-numbered years with the SAS Symposium. The SAS Board of Directors is looking forward to a long and
rewarding partnership that brings together some of the worlds best backyard and professional astronomers to
share their experiences and expertise in the name of promoting astronomical research to advance our knowledge
of the Universe.
Not everyone wants to give a presentation but they still want to contribute to the Symposium by submitting a
paper for the proceedings. So, this year, partly because of the unprecedented number of papers received, the
proceedings include a section not just for poster papers but papers without presentation. We may try to do this
in future years as well but, as always, our first goal is to get presenters who also submit papers. The live presentations are vital for spurring discussion both during Q&A and in groups outside the formal paper sessions.
It takes many people to have a successful conference, starting with the Conference Committee. This year the
regular committee members are:
Lee Snyder
Robert Gill
Cindy Foote
Brian D. Warner
Dale Mais
Robert D. Stephens
Jerry L. Foote
Margaret Miller
Robert K. Buchheim
There are many others involved in a successful conference. The editors take time to note the many volunteers
who put in considerable time and resources. We also thank the staff and management of the Northwoods Resort
in Big Bear Lake, CA, for their efforts at accommodating the Society and our activities.
Membership dues alone do not fully cover the costs of the Society and annual conference. We owe a great debt
of gratitude to our corporate sponsors: Sky and Telescope, Software Bisque, Santa Barbara Instruments Group,
PlaneWave Instruments, Apogee Instruments, Inc., and DC-3 Dreams. Thank you!
Finally, there would be no conference without our speakers and poster presenters and those sitting attentively in
the audience (youre not doing your email right now, are you?). We thank them for making this one of the premiere pro-am events in the world.
Brian D. Warner
Robert K. Buchheim
Jerry L. Foote
Dale Mais
2012 April 27
Conference Sponsors
Conference Sponsors
The conference organizers thank the following companies for their significant contributions and financial support. Without them, this conference would not be possible.
Apogee Instruments, Inc.

Manufacturers of astronomical and scientific imaging cameras
http://www.ccd.com
Sky Publishing Corporation

Publishers of Sky and Telescope Magazine
http://skyandtelescope.com
Software Bisque
Developers of TheSky Astronomy Software and the
Paramount Telescope Mount
http://www.bisque.com
Santa Barbara Instruments Group

Makers of astronomical instrumentation
http://www.sbig.com
PlaneWave Instruments
Makers of the CDK line of telescopes
http://www.planewaveinstruments.com:80/index.php
DC-3 Dreams Software

Developers of ACP Observatory Control Software
http://www.dc3.com/
ii
Symposium Schedule
Symposium Schedule
The schedule is subject to change without notice.
Meeting Courtesies
Please turn off cell phones, pagers, and alarms during presentations.
Please be kind to the presenters and wait until breaks to check your email and browse the
Internet. Yeah, we know.
Please be considerate of those in the meeting room near the wall with the vendors room. If
visiting the vendors room during presentations, please keep voices down and do not use the
direct entrance between the vendors and meeting rooms. Enter the meeting room from the
hallway.
If you do not want your presentation to be recorded (video/audio) and/or not posted on the
SAS web site, please let us know before your talk begins.
Thanks!
SAS Program Committee
iii
Symposium Schedule
iv
Presentation Papers
Hubbles First M31 Cepheid Goff et al.
Photometry of Hubbles First Cepheid

in the Andromeda Galaxy, M31
Bill Goff
AAVSO
13508 Monitor Lane
Sutter Creek, CA 95685
b-goff@sbcglobal.net
Matt Templeton
AAVSO
49 Bay State Rd.
Cambridge, MA 02138, USA
Matthewt@aavso.org
Richard Sabo
AAVSO
Richard@theglobal.net
Tim Crawford
AAVSO
tcarchcape@yahoo.com
Michael Cook
AAVSO
michaeljcook@rogers.com
Abstract
In the 1920s Hubble discovered a variable star in the Andromeda Galaxy, M31 which later was revealed to be a
Cepheid type variable. Hubbles observations confirmed that M31 was indeed at remote star system separate
from our own Milky Way. We present here recent photometry of this object made during 2010 and 2011. These
observations were made by members of the AAVSO at the request of the Space Telescope Science Institutes
Heritage Project. Over 300 Rc, V and unfiltered observations have been made. These are the first observations
in current times to establish a period for the object. The period of this Cepheid as measured here is in agreement
with the earlier data from 1920s and 1950s.
1. Introduction
Edwin Hubble settled one of astronomys biggest
puzzles in 1925 when he published light curves of
Cepheid variables he had located in the Andromeda
Galaxy, M31. At the time astronomy was at a crossroads. Many astronomers believed that the nebulae
in the night sky were just parts of our own Milky
Way Galaxy, and others believed they were external
objects far away. There wasnt a good way to measure their distances
Hubble began monitoring M31 and a few other
galaxies with the largest telescope in the world of the
time, the 100-inch Hooker telescope on Mt. Wilson
in California. The great telescope as Hubble re-
ferred to it, was capable of resolving many individual

stars in that galaxy (Hubble, 1936).
Hubbles work hinged on the recently revealed
relationship between period and luminosity for Cepheid variables. Henrietta Leavitt had surveyed hundreds of variables in the Magellanic Clouds and noticed that the brighter Cepheids had longer periods.
Since they were nearly all at the same distance it
meant that absolute luminosity could be gauged. This
relationship later became know as the Leavitt Law.
Calibrating the Leavitt Law had significant interest and difficulties. There are very few Cepheids
close enough for parallax measurements. Shapely
used Cepheids in globular clusters, whose distances
were known through other means, to establish what
was seen as a good calibration. Hubble utilized this
calibration and noted that further revisions were expected to be of minor importance. After Hubble published his work on the galactic distance scale, it was
discovered there are two types of Cepheids. The cluster Cepheids used by Shapely were older, smaller and
fainter than the brighter classic Cepheids in the disk
of the galaxy and recalibrating the Leavitt Law
caused the size of the universe to double! (Percy,
2007).
2. Observations
The Space Telescope Science Institutes Heritage Project drew AAVSO to this object. They were
interested in re-imaging the target with Hubbles
namesake, Hubble Space Telescope, as a tribute to
his achievement and wanted to know if a current
ephemeris existed. As the object had not been observed by the AAVSO, an Alert Notice (Waagan,
2010) was published to invite observations so that a
light curve could be established. The object was reported at 19th magnitude by Hubble so, while a challenging target, it is not out of reach for amateurs with
CCD cameras and large-aperture scopes. Over the
course of the next several months, 214 observations
were made via CCD imaging. Most all observations
were done with Cousin R filter, but some were done
with a standard V filter. Total exposure times were
normally in the 60-minute range for one observation,
which were normally stacked groups of shorter exposures. Images were processed using standard available software packages and techniques such as flatfielding and dark subtraction. Results were then uploaded to the AAVSO database.
ObsCode
Name
Location
GFB
Goff
California
66
CMIA
Cook
Ontario, Ca
12
SRIC
Sabo
Montana
CTX
Crawford
Oregon
3. Analysis of 2010 Data

As noted above, Templeton et al. (2011) had reported on the analysis of the 2010 data and the
ephemeris for the object. A Fourier function had been
used with a result of 31.4 0.1 day period. This period was used to establish the ephemeris needed by
the Heritage Project.
In order to evaluate the two combined seasons of
data, we first needed to insure we agreed with
Templetons result. The 2010 data were downloaded
from AAVSO and then divided into separate sets by
observer. One observers (GFB) data were loaded
first into Peranso software, as it covered more than
two cycles of variation. Then the second observers
(HQA) data were loaded and used for a standard
magnitude. The two sets were adjusted by using a
Fourier phase diagram which overlays the data onto
one cycle, which made the differences appear obvious. Binning the data by the day also helped. The
program reports the standard deviations of the bins so
those days where the value was high indicated differences between the two sets. After the two sets were in
agreement, the remaining sets were loaded one at a
time and adjusted as well. The 2010 phased data are
shown in Figure 1.
Obs
Figure 1. Phased 2010 data. All observations are folded

onto two phases.
Observers for the 2011 data. The 2010 observer list is

reported by Templeton et al. (2011).
The observations are not on a standard photometric system though the observers did use a standard set
of comparison stars established by the AAVSO and
similar filters. This required that the observation be
adjusted to a common zero point. Templeton (2011)
reported on a technique to make these adjustments to
the 2010 data for the object.
Observations continued in the 2011 season by
four observers. They made an additional 89 observations all in Cousins R; and they too were processed as
above and added to the AAVSO database.
Figure 2. Light curve for 2010 data. Period is shown as

31.38 0.09 days.
We used Period04 software for our Fourier

analysis. The program uses a discrete Fourier Transform algorithm and provides least squares fitting for
a number of frequencies. Fitting the 2010 data resulted in a period of 31.38 0.09, which we considered a good result. Our light curve for the 2010 data
is shown on Figure 2. The light curve is not smooth
due to the fitting of multiple frequencies.
the Period04 software. The Fourier analysis and fitting proceeded as before, and as a final result, the
period was reported as 31.47 0.03 d. The light
curve for the combined 2010-2011 data is shown on
Figure 5. Templeton has suspected that by adding an
additional seasons observation the period error
would reduce, and this appears to be true.
4. Analysis of 2011 Data

The same process was repeated with the 2011
data. One observers (GFB) data set was first loaded
then the others loaded individually and zero pointed
using the phase diagram and binning until all were
loaded. The phase diagram is shown in Figure 3.
Figure 5. Light curve for combined 2010-2011 data. Period is shown as 31.47 0.03 days.
6. Discussion
Figure 3. Phased 2011 data. All observations are folded

onto two phases.
Templeton et al. (2011) noted that the first years

results were in agreement the results of Hubble and
Baade done in the 1920 and 1950 and that the object
is unchanged over this time span. We substantially
agree that the two-year results add support to that
result. We also note that Hubbles first results came
from 10 years of images from a dozen observers so
the greater span aided his result. Some members of
the team have expressed interest in continuing with
this target for another year. Gathering data on another
color band would be of value as Cepheids change
color during their cycle.
7. Acknowledgements
Figure 4. Light curve for 2011 data. Period is shown as

31.26 0.26 days.
The 2011 data set is a smaller one. Fourier

analysis and fitting the data with the Period04 program resulted in a period of 31.26 0.26 d. The light
curve is shown in Fig. 4. The reduced observations
have made it appear more erratic and the error is considerably higher.
5. Combined 2010-2011
The goal of continuing the observations through
the second year was to further refine the period over
the longer time span. Having two years of data now
at hand, we imported them both into one session with
The authors would like to thank Arne Henden for

his encouragement in undertaking this project during
the SAS conference in May 2010. Matt Templetons
paper of 2011 set a very high benchmark for our
work and his participation in this work have been
essential in directing our efforts here.
8. References
Hubble, E. (1939, 1982). The Realm of the Nebulae,
Yale University Press, New Haven, US.
Lenz, P., Breger, M. (2010). Period04, astronomical
time series analysis software.
http://www.univie.ac.at/tops/Period04/
Peranso 2010, Peranso, period analysis software.
http://www.peranso.com/
Percy, R. (2007). Understanding Variable Stars,

Cambridge University Press, Cambridge, UK
Templeton, M., Henden, A., Goff, W., Smith, S.,
Sabo, R., Walker, G., Buchheim, R., Belcheva, G.,
Crawford, T., Cook, M., Dvorak, S., Harris, B.
(2011). PASP 123, 1347.
Waagen, E.O. (2010). AAVSO Alert Notice 422 (2010
July 16).
BY Lyncis Kemp et al.
BK Lyncis: The Oldest Old Nova?

or: Archaeo-Astronomy 101
Jonathan Kemp
Gemini Observatory, Hilo, HI 96720
jk@155west.org
Joe Patterson
Department of Astronomy, Columbia University, 550 West 120th Street, New York, NY 10027
jop@astro.columbia.edu
Enrique de Miguel
CBA (Huelva), Observatorio del CIECEM, Matalascaas, 21076 Almonte, Huelva, Spain
edmiguel63@gmail.com
George Roberts
CBA (Tennessee), 2007 Cedarmont Dr., Franklin, TN 37067
Tut Campbell
CBA (Arkansas), 7021 Whispering Pine, Harrison, AR 72601
Franz-J. Hambsch
CBA (Mol), Oude Bleken 12,B-2400 Mol, Belgium
Tom Krajci
CBA (New Mexico), PO Box 1351 Cloudcroft, NM 88317
Shawn Dvorak
CBA (Orlando), AAVSO, 1643 Nightfall Dr., Clermont, FL 34711
Robert A. Koff
CBA (Colorado), Antelope Hils Observatory, 980 Antelope Drive West, Bennett, CO 80102
Etienne Morelle
CBA (France), Sirene ObservatoryLagarde dApt, France
Michael Potter
CBA (Baltimore), 3206 Overland Ave., Baltimore, MD 21214
David Cejudo
CBA (Madrid), Observatorio El Gallinero, El Berrueco, Madrid, Spain
Joe Ulowetz
CBA (Illinois), 855 Fair Lane, Northbrook, IL 60062
David Boyd
CBA (Wantage), 5 Silver Lane, West Challow, Wantage, OX12 9TX, United Kingdom
Richard Sabo
CBA (Montana), 2336 Trailcrest Drive, Bozeman, MT 59718
John Rock
CBA (Wilts), 2 Spa Close, Highworth, Swindon, Wilts SN6 7PJ, United Kingdom
Arto Oksanen
CBA (Finland), Verkkoniementie 30, Fi-40950 Muurame. Finland
Abstract
We summarize the results of a 20-year campaign to study the light curves of BK Lyncis, a nova-like star
strangely located below the 2-3 hour orbital period gap in the family of cataclysmic variables. Two superhumps
dominate the nightly light curves with periods 4.6% longer, and 3.0% shorter, than Porb. The first appears to be
associated with the star's brighter states (V~14), while the second appears to be present throughout and becomes very dominant in the low state (V~15.7). It's plausible that these arise, respectively, from a prograde apsidal precession and a retrograde nodal precession of the star's accretion disk. In 2011-2, the star's light curve
became indistinguishable from that of a dwarf nova in particular, that of the ER UMa subclass. No such transition has ever been observed in a cataclysmic variable. Reviewing all the star's oddities, we speculate: (a) BK
Lyn is the remnant of the probable nova on 30 December 101, and (b) it has been fading ever since, but has
taken ~2000 years for the accretion rate to drop sufficiently to permit dwarf-nova eruptions. If such behavior is
common, it can explain two other puzzles of CV evolution. One: why the ER UMa class even exists (because all
can be remnants of recent novae). And two: the relative space densities of short- and long-period CVs, and of
known nova remnants (set by the durations of these transient phases).
1. Introduction
BK Lyncis was discovered in the Palomar-Green
survey for objects with ultraviolet excess (Green et
al., 1986), and was listed as PG0917+342 in the preliminary catalog of cataclysmic-variable stars in that
survey (Green et al., 1982). A subsequent radialvelocity study confirmed the CV identification and
revealed an orbital period of 107.97 minutes (Ringwald et al., 1996). Two years of time-series photometry revealed superhumps in the star's light curve
large-amplitude waves interpreted as resulting from
apsidal precession of the accretion disk (Skillman and
Patterson, 1993; hereafter SP). These studies showed
only small variability in the range V = 14.5-14.7.
Thus the star became well-established as a novalike
variable, a class which would be unremarkable, except for the stars short orbital period. Of the few
hundred CVs known with orbital period below 2
hours, BK Lyn is the only novalike variable.
BK Lyn is also a good candidate as the oldest
old nova. Several studies of ancient Chinese records
have suggested that a nova appeared very close to its
position on 101 December 30 (His, 1958; Pskovskii,
1972; Clark and Stephenson, 1977), and Hertzog
(1986) concluded that BK Lyn is the remnant of
Nova Lyncis 101. This would certainly qualify as the
oldest old nova far exceeding the closest challenger, CK Vul = Nova Vulpeculae 1670 (Shara et
al., 1985).
Such curiosities have kept BK Lyn on our observing lists for years. In this paper we report the
results of many observational campaigns: spanning
20 years, and including ~400 nights and ~2500 hours
of time-series photometry. Among the several re-
wards, detailed here, was the star's spectacular transformation into a bona fide dwarf nova in 2012.
2. Observational Techniques
Essentially all the data reported here comes from
the Center for Backyard Astrophysics (CBA), a
global network of telescopes cooperating in campaigns of time-series photometry of variable stars.
Most of the observational techniques were discussed
in the second paper of our series (SP), but the network expanded in later years to include ~20 telescopes, spread sufficiently over the Earth to give very
long time series relatively untroubled by local
weather and daily aliasing. Our typical telescope is a
35-cm reflector, equipped with a CCD camera and
recording images every 60 s for many hours per
night. Most of the data is unfiltered (white-light, or
perhaps more correctly, pink, with an effective
wavelength near 6000 A) differential photometry,
although we always obtain some coverage in V light
to express results on a standard scale. Data from several telescopes are then spliced together to form a
one-night light curve, with minimal gaps. We take
advantage of overlaps in data to determine additive
constants which put all our measurements on one
scale (usually that of the most prolific or bestcalibrated observer). These constants are usually in
the range 0.01-0.05 mag, probably due to variations
in transparency and camera sensitivity. Most telescopes use the same comparison star, although we
also use data with other comparisons (requiring larger
and more uncertain additive constants) if there is sufficient overlap. In this case we used GSC 249600893, which is 3.6 arcmin NE from BK Lyn. On 5
Tale I. Observation Details
good nights in 2011, we measured that star to have

V = 13.897(8), B-V = 0.533(10).
Research programs on faint stars with small telescopes often use white light, to enable high time resolution and to improve signal-to-noise. In the case of
cataclysmic variables, it usually makes good astrophysical sense too, since the underlying sources of
light are broad-band emitters (accretion disk, white
dwarf). It is common practice to report magnitudes as
C (or often CV, though we will avoid this term
for obvious reasons): the result of differential photometry in clear light, added to the comparison star's
known V magnitude. This is also our practice. However, because the white-light passbands are typically
~4000 A wide, the effective wavelengths of the variable and comparison stars can easily be 500 A apart.
Therefore, C/CV magnitudes are not V magnitudes.
We nevertheless prefer the C/CV scale and use it
here, because it is our natural measurement scale, and
because it expresses the true changes in light very
accurately.
For archival purposes, a standard V magnitude is
more desirable. Our C magnitudes transform to V
magnitudes via
V = C + 0.37 (B-V)
which implies V = 0.20 in this case, where the
variable is assumed (and observed) to have B-V near
0.0. The latter assumption is pretty good for the great
majority of cataclysmic variables accreting at a high
rate.
Atmospheric extinction is significant for us, because the program stars are usually much bluer than
comparison stars (although we avoid very red stars,
which are the bane of all stellar photometry). We
know from experience that this differential extinction
amounts to ~0.04 mag/airmass for most CVs. Nevertheless, in the spirit of keeping human hands off the
data as much as possible, we usually make no correction for extinction.
The summary observing log in the five new observing seasons (adding to the two seasons reported
by SP) is given in Table 1. (A night denotes a timeseries of good quality lasting at least 3 hours.)
3. Seasonal Light Curves

During all years prior to 2011, the star lived up
to its billing as a novalike variable, showing quite
small excursions about a mean V = 14.7. This is also
consistent with the snapshot Roboscope data reported
by Ringwald et al. (1996), which found the star always in the range 14.6-14.7 on 116 nights in 1994
and 1995. And its consistent with the range (14.514.8) listed in the 1980s-90s versions of the General
Catalogue of Variable Stars (GCVS).
However, in 2011 the star clearly had excursions
to a fainter state (15.5), as well as a brighter state
(13.5). This seasonal light curve is shown in the upper frame of Figure 1. In 2012, the variations were
closely monitored over a long baseline, and revealed
a pattern which bore all the earmarks of a dwarf-nova
in particular, a dwarf nova of the ER UMa class.
This is shown in the middle frame of Figure 1.
Figure 2. Five nights in 2012, showing the basic pattern

of the negative superhump: large amplitude when the
star is faint and small amplitude when the star is bright.
The amplitude is roughly constant in intensity units.
Points are 10-minute averages over the original 1-minute
integrations.
Figure 1. Top: BK Lyn 2011 light curve. For the first time,
the star showed bright and faint states. Middle: BK Lyn
2012 light curve, showing the well-defined ER UMa
pattern of variability. Bottom: the contemporaneous
2012 light curve of ER UMa apparent quite a close
relative of BK Lyn. The two stars also show great similarity in their complex, yet virtually identical, pattern of
superhumps. In this figure, all points are averages over
1-3 superhump cycles.
In fact, by coincidence we were simultaneously

carrying out a long monitoring campaign on ER UMa
itself, and that seasonal light curve was practically
indistinguishable even in small details from that
of BK Lyn in 2012. This is shown in the lower frame
of Figure 1. In both cases, apparent superoutbursts
occurred every 40 days, followed by smaller excursions possibly normal outbursts repeating every
4-5 days. Furthermore, as will be discussed in the
next section, the two stars revealed a complex and
identical morphology of periodic signals (positive
and negative superhumps). Within the limits of data
now available, it appears that BK Lyn, after ~30
years of living as a novalike variable, became a dwarf
nova in 2011-2.
4. Nightly and Spliced Light Curves

Our main program was to study the time series for periodic signals. The first two seasons have
been published (SP: 1992 and 1993). BK Lyn then
stayed always near V=14.6, and showed rather stable
waves with a period slightly exceeding Porb: a positive superhump. The full amplitude was ~0.07 mag,
and the period excess was 4.6%, fairly typical of superhumps in CVs of comparable Porb.
10
Later years showed a much stronger signal,

which we illustrate in Figure 2, a 4-day segment obtained in 2012. This dominated all the later data
(1999-2012), and was especially strong, approaching
0.6 mag, when the star became faint. In this section,
we report on this and other signals in the various
years light curves.
Throughout this paper, we adopt a date convention of truncated Julian dates (JD = true JD
2,400,000), and a frequency convention of cycles
d-1 = c/d (for compactness, and as the natural unit of
frequency in programs affected by daily aliasing).
Since this paper is equally a story of classical and
dwarf novae, we also need a convention for describing their shenanigans: classical novae have eruptions,
whereas dwarf novae will be described as having
outbursts or maxima. Finally, we use the term
novalike to describe noneruptive and nonmagnetic
CVs whose spectrum, excitation, and MV are similar
to the prototype, UX Ursae Majoris. This is the most
common use of the term, although some authors use
it more expansively to describe any CV not known
to be a dwarf nova, old nova, or magnetic.
4.1 1999 Coverage

The 1999 coverage spanned 26 days, with good
instrumental magnitudes and a dense segment during
JD 51218-32. The power spectrum of this segment is
shown in the upper frame of Figure 3. The dominant
signals, by far, occur at 0.400(11) and 13.738(9) c/d;
these are sensibly phase-stable over the 26 days, and
have full amplitudes respectively of 0.17 and 0.13
mag. After subtracting these two powerful signals,
the time series shows weaker but significant features
at 12.770(10) and 25.560(11) c/d. Shown in the lower
frame of Figure 3, these latter signals are apparently a
manifestation of the positive superhump, familiar
from 1992/3. The strong signals are new, and are
very likely manifestations of a negative superhump, a fairly common phenomenon in CVs of
short Porb and high accretion rate. The mean waveforms are shown in the inset figures. The negative
superhump is closely sinusoidal, while the positive
superhump has a very strong second harmonic.
4.
Whenever a A sideband appears, a lowfrequency signal at A never appears.
In fact, for virtually all our data on all superhumping CVs, these rules appear to be strictly followed although data quality does not always yield
high confidence for frequencies below 2 c/d.
This terminology accords with a common modern interpretation, and will help our reportage in this
complicated story of periodic signals. A primer on
periodic-signal and superhump zoology in cataclysmic variables can be found in Appendix A of Patterson et al. (2002).
4.2 2002 Coverage
Figure 3. Top: power spectrum of BK Lyn in 1999, with

significant signals flagged with their frequencies in c/d.
The negative superhump dominates, along with its associated wobble signal. Bottom: the power spectrum
of the residuals, after these two strong signals are subtracted from the time series. A weak positive superhump
remains, with much of the power at the second harmonic (double-humped).
Unfortunately, the negative and positive superhumps are separated by a frequency close to 1 c/d, a
cruel blow to astronomers on our planet. Still, the star
never promised us a rose garden, and this long campaign with good alias rejection was able to separate
the two signals.
At the risk of injecting some interpretation into
matters of reportage (seldom wise, and we properly
hate ourselves for doing it), we will call the positive
superhump apsidal and express the frequency as
A and the negative superhump nodal and express the frequency as +N (where is the orbital
frequency). The 1999 signals are then N, A, +N,
22A, and 2A. In fact, for all years of observation, all detected frequencies obey these simple rules:
1.
2.
3.
All apparitions of N are in the form +N, +2N,

etc.
All apparitions of A are in the form A, 2A,
etc.
Whenever a +N sideband appears, a strong
low-frequency signal at N appears.
The 2002 coverage spanned 58 days, with a

dense segment during JD 52264-318. The powerful
signals occurred again at 0.394(2) and 13.740(2) c/d.
After subtracting these, weaker signals appeared at
25.521(2), 27.475(2), 12.760(2), and 13.344(2) c/d,
in order of decreasing power. These appear to be the
N, +N, 2(A), 2(+N), A, and signals in
order of decreasing power. Full analysis awaits a subsequent paper. But we note that these frequency assignments agree within the quite small errors, with
the basic underlying clocks taken to be the orbital
frequency, the nodal frequency (that of disk wobble, in the popular interpretation), and the frequency
of
apsidal
motion
respectively
at
= 13.344(2), N = 0.395(2), and A = 0.584(2) c/d.
4.3 2005 Coverage

The 2005 coverage spanned 20 days, when the
star stayed always close to V = 14.8. The runs were
sufficiently short that no reliable analysis was feasible for signals of very low frequency. But the power
spectrum showed the usual superhump signals at
13.740(4) and 12.751(4) c/d and their second harmonics, with full amplitudes of 0.140 and 0.024 mag,
respectively. The +N negative superhump is highly
sinusoidal, while the A signal has the distinctive
fast-rise-slow-decline pattern characteristic of positive superhumps (in virtually all cataclysmic variables).
4.4 2011 Coverage

The stars 2011 light curve covered 324 hours
and 58 nights. We defer detailed analysis to a later
paper. However, the seasonal light curve, seen in
Figure 1, clearly shows the star to have significantly
different brightness states both fainter and brighter
11
than anything seen (by us, or by anybody) in previous

years. We resolved to do a more thorough job in the
2012 campaign.
4.5 2012 Coverage

As of 2012 early April, our observing campaign
is ongoing. However, Table I shows details of the
coverage (~1000 hours and climbing), and Figures 1
and 2 show the seasonal and several-day light curves.
Full details await the end of the campaign, but weve
learned a lot from all this close scrutiny:
1.
The star now shows recurrent high and low

states. The high states are either bright and
long, or faint and short and follow the guidelines of supermaxima and normal maxima
familiar from the study of dwarf novae, especially the ER UMa class. [See Robertson et al.
(1995) for the standard treatment of this class,
and Osaki (1996) for a lucid theoretical account.] Supermaxima occur every 45 days, and
normal maxima occur every 5 days.
2.
An apsidal superhump grows suddenly to very

large amplitude (0.3 mag) at the beginning of
superoutburst, and decays slowly as the outburst does, over ~15 days. Its waveform initially shows the nearly universal fast-riseslow-decline pattern, but then mutates to a
double-humped shape after a few days.
3.
A negative superhump is present all the time,

with an amplitude which is practically constant
in intensity units (~0.6 mag at V = 16, 0.07
mag at V = 14). This signal appears to pay no
attention, in phase or amplitude, to any outbursts which may be occurring. And it always
dominates when the star is faint; the pattern
obvious in Figure 2 repeats exactly every time
the star shuffles between bright and faint
states.
5. Interpretation
Those are the facts. We now proceed to the challenge of interpretation. In that task we appeal to numerous elements of today's conventional wisdom
regarding the common interpretation, and in some
cases outright theory, of cataclysmic variables.
Of course, such reliance on theory and popular
interpretation always brings with it some hazard of
going wrong.
12
5.1 A Precessing and Wobbling Disk

The periodic signals (, N, A, and their children)
are the most straightforward, because they fit into a
detailed pattern of behavior which is familiar from
the study of numerous other CVs.
BK Lyn is a dwarf nova (nee novalike) with a
short orbital period, and outbursts showing a
long/short dichotomy. During all long outbursts of
such stars, a strong apsidal superhump (A) is
quickly born, and decays after a few weeks. Short
eruptions never hatch such signals. Stars like this are
called SU UMa-type dwarf novae. There are hundreds of such stars, and therefore thousands of eruptions sufficiently studied to test the universality of
these statements, and there are no known exceptions;
you can take it to the bank.
Present-day nomenclature also ordains subclasses. The superoutbursts of so-called WZ Sge stars
happen very rarely (P>10 years), while those of ER
UMa stars happen very frequently (P < 100 days).
Stars in between are just plain old SU UMas. In our
opinion, these subclasses do not reflect any essential
difference in physics, but merely accretion rate.
Osaki (1996; see his Figure 2) shows simply and lucidly why outburst recurrence rate should vary
smoothly with accretion rate; Patterson (2011, hereafter P11) shows ploddingly that observations bear this
out, and in particular that recurrence period scales as
(Mdot)-1 (Figures 7 and 11 of that paper).
Superhumps are a great distinguishing feature of
all such stars, and are sometimes taken to be a defining feature of the SU UMa class. Our enthusiasm
doesn't go quite that far, however. It is more general,
and more interesting, to say that (apsidal) superhumps inevitably result when stars of sufficiently
short Porb, and containing accretion disks, achieve
sufficiently high Mdot for a sufficiently long time.
Its then up to observers to determine what constitutes sufficient. Patterson et al. (2005) estimated
these sufficiency conditions as follows: Porb < 3.5
hours, Mdot 3x10-9 MO/yr, t ~ a few days.
In all respects that we have been able to measure,
or ponder, it certainly appears that BK Lyn's positive
superhump is a fully credentialed member of the
dwarf-nova family. In particular, it resembles in detail that of the ER UMa family, with behavior patterns reminiscent of other class members: V1159 Ori
(Patterson et al., 1995), V503 Cyg (Harvey et al.,
1995), and ER UMa itself (Gao et al., 1999). We take
that to mean, in the now-popular interpretation, that
in superoutburst, BK Lyn's disk becomes quickly
eccentric, with a line of apsides precessing prograde
at 0.58 c/d.
The negative superhump is more challenging to
understand, because this phenomenon has been seen
in only 10-20 CVs, and they have not shown the uniform behavior patterns nicely presented to us by positive superhumpers. We defer full analysis to a later
paper.
5.2 The Classical Nova

The guest star of 101 A.D. has been previously
discussed by Hsi (1958) and Clark and Stephenson
(1978), and especially by Hertzog (1986). Unlike
most guest stars, it is at relatively high galactic latitude (40 degrees), and can be placed with unusual
precision in the sky, since it is described as very close
to a star which all students of ancient records take to
be Alpha Lyncis. Hertzog argues that a compelling
case can be made for BK Lyn, a scant 29 arcminutes
away from Lyn. The modern-day BK Lyn is an
extremely unusual star the only short-period novalike among the ~1000 known CVs, and now the
only novalike to have morphed into a dwarf nova.
Positional coincidence with a very unusual object is
evidence of physical association. The modern-day
GCVS apparently accepts this, since it lists BK Lyn's
range of variability as 0.3-14.6.
The ancient records describe the star as not
bright, which, in the context of other brightness reports in those records, is taken to mean a magnitude
near zero. Outbursts of the modern-day dwarf nova
reach V = 13.8 (allowing for the difference between V
and C magnitudes), and outbursting dwarf novae are
pretty good standard candles with Mv = +4.6. Allowing an additional 0.2 mag for absorption on this
line of sight, we estimate BK Lyn to lie at a distance
of 600 pc. If the guest star is actually BK Lyn, then it
apparently rose to Mv = 9 (0.3 minus 0.2, with a
distance modulus m-M = 9). This is about right for a
moderately fast classical nova, and a cataclysmic
variable is exactly what's needed as a classical-nova
progenitor. Finally, we note that BK Lyn has just
executed the move which theorists have long predicted must occur for very old novae: it has settled
back into a dwarf-nova state. These arguments from
physics and brightness strengthen the argument from
positional agreement, and we conclude, following
Hertzog, that BK Lyn is very likely to be the remnant
of Nova Lyn 101.
5.3 BK Lyn as a Milestone for Nova

Evolution
There has long been a paradox or at least an
oddity in the census of cataclysmic variables. All
such stars should become classical novae as their
envelopes of accreted hydrogen grow to a certain
threshold, and the great majority of CVs (>90%) are
stars of short orbital period. Yet the great majority of

known old novae (~90%) are stars of long orbital
period. We still don't have a satisfactory explanation
for this oddity. However, identification of BK Lyn
with an ancient nova will shed considerable light
upon it.
In a classic study of old novae, mainly from the
20th century, Robinson (1975) found that novae have
the same brightness before and after eruption. But
since most known novae arise in stars of long Porb,
this finding only applies to that class. To express this
in terms of accretion rate, Robinson found that (longPorb) postnovae fade to ~10-8 MO/yr within 10-20
years after eruption... and then, within the limits of
data on photographic surveys, seem to have the same
brightness, and therefore accretion rate, a few years
before the eruption. The simplest interpretation is that
these stars are similarly bright throughout the long
interval between eruptions. We strongly suspect that
this is true because the present census of long-Porb
CVs, which are presumably the ancestors of future
novae, really does not contain any stars which substantially differ from this description. Among longPorb stars, prenovae, postnovae, and nova suspects all
look about the same (aside from fireworks associated
with the eruption itself: ejected gas and dust shells,
supersoft X-rays, etc.).
The situation for short- Porb stars should be radically different. Why? Because these stars are not
naturally entitled to accrete at 10-8 Mo/yr. Their only
reliable driver of mass transfer is GR, which only
provides 10-10 MO/yr. Therefore they have to wait
~100x longer to pile up enough matter to fuel another
eruption, and that may be plenty of time to cool sufficiently to join their natural compadres at short Porb
the garden-variety dwarf novae, with a quiescent MV
near +10 and brief outbursts every 100-1000 days.
In this scenario, BK Lyn is the product of a recent nova eruption probably the eruption of 101
December 30. With a mere 2000 years of cooling, the
white dwarf is still fairly hot, and the secondary star
sufficiently agitated to transfer matter at an unnaturally high rate near 10-8 MO/yr. This explains the
high temperature component seen in the ultraviolet
spectrum (35000 K; Zellem et al., 2009), and the
star's intrinsic brightness (<MV> = 4, Patterson,
2011). Somewhere between 2005 and 2011, the star
faded sufficiently to allow dwarf-nova outbursts to
occur.
This empirically sets an important and previously unknown timescale: 2000 years to resume life
as a dwarf nova, viz. of the ER UMa persuasion.
However, ER UMa stars are themselves quite rare;
P11 estimated that they comprise only 1% of the
population of short-period dwarf novae. If the ER
UMa stage lasts ~10000 years, the numbers work out
13
about right: 2000 for the BK Lyn era, 10000 for the
ER UMa era, and 1,000,000 for dwarf-nova normalcy. That satisfies the available space-density constraints, and allows the correct amount of time for
binaries accreting at 10-10 MO/yr to accumulate the
10-4 MO needed to trigger a nova eruption.
Shouldn't we then expect most short-period novae to be much fainter prior to eruption, contrary to
Robinson's study? Yes but that study concerned
mainly long-period novae. Schaefer and Collazzi
(2010) recently studied archival photographic magnitudes for a group of short-period novae, and found
them to be much fainter before eruption. So all the
evidence to date seems to be consistent with this account of relaxation following a classical nova in a
short-period CV.
6. Summary
1.
We began this long campaign to study the superhump signals of BK Lyn, but detailed analysis
still lies ahead. Both positive and negative superhumps appear to fit into the known categories
of similar objects and, in particular, strongly
resemble those of two known ER UMa stars:
V503 Cyg and ER UMa itself. Probably they
arise from apsidal advance and nodal regression
in an eccentric and tilted accretion disk.
2.
Seasonal light curves demonstrate that somewhere in the interval 2005-2011, the novalike
variable BK Lyn mutated into a card-carrying
dwarf nova of the ER UMa class, with superoutbursts occurring every 45 days. The dwarf-nova
standard-candle relation implies a distance of
about 600 pc.
3.
Arguments from positional coincidence, galactic

latitude, nova physics, and the rarity of shortperiod novalikes suggest that BK Lyn is the oldnova remnant of the guest star of 101 A.D.
4.
This suggests the following timescales for a

short-period dwarf nova's relaxation following a
classical-nova eruption: 2000 years as a novalike, 10000 years as an ER UMa dwarf nova
(gradually declining to an ordinary SU UMa),
and 1,000,000 years in true quiescence.
5.
This can explain why ER UMa stars exist (because they are remnants of recent novae), why
they're rare (because that phase is only 1% of a
nova's full eruption cycle), and why short-period
stars are rare among known old novae (because
they spend most of their quiescence very faint,
where they hide from our telescopes).
6.
If this story of smooth decline is correct, then the

superoutburst recurrence period a rather easily
determined number is very strongly affected by
time since the last classical-nova eruption. That
might be the second parameter (besides orbital
period) needed to understand the scatter in the
various Mdot-related quantities of dwarf novae
(see the figures in P11). For example, it could
explain the tremendous difference between the
frenetic RZ LMi (P~20 days) and the somnolent
WZ Sge (P~30 years) two stars of practically
identical Porb.
7.
You can learn an awful lot from a patient research program with 14-inch telescopes.
5.4 The Transition to Dwarf Nova: a
Singular Event?
Does the observed 2005-2011 transition to a
dwarf-nova state represent a singular event in BK
Lyn's postnova evolution? Well, maybe. Our 20-year
observation span, and the 6-year window for this
event, are short but not ridiculously short compared
to the putative 2000-year wait. Also, we obtain similar photometric coverage of many old novae and novalikes (dozens), and we have not observed such a
transition in any other star. So even as a singular
event, not to be repeated until the next nova cycle, it
does not seem wildly improbable.
But there is no need to hypothesize a singular
event. Secular decline in Mdot over a few thousand
years could easily be punctuated by small fluctuations about the temporary mean; indeed, many cataclysmic variables show small luminosity variations
on timescales of decades (Warner, 1988, Richman et
al., 1991). It's plausible that such fluctuations could
now be swinging the disk between states of stable
and unstable accretion. Indeed, that is our current
understanding of Z Cam stars: dwarf novae near that
threshold accretion rate, with their disks fluctuating
irregularly between stable and unstable. In this version of our scenario, BK Lyn could be considered the
first short-period Z Cam star.
If this latter version were correct, then we might
well see other ER UMa stars especially RZ LMi
with its whirlwind 20-day superoutburst cycle mutating temporarily into novalike variables, pausing
slightly in their inevitable decline towards a long and
simple life as a garden-variety dwarf nova.
14
7. References
Warner, B. (1988). Nature 336, 129.
Clark, D. H., Stephenson, F. R. (1977). The Historical Supernovae. Pergamon New York.
Zellem, R., et al. (2009) PASP 121, 942.
Gao, W., et al. (1999). Ap. J. 527, L55.

Green, R. F., Schmidt, M., Liebert, J. (1986). Ap. J.
S. 61, 305.
Green, R. F., Ferguson, D. H., Liebert, J. E., Schmidt,
M. (1982) PASP 94, 560.
Harvey, D. A., et al. (1995) PASP 107, 551.
Hertzog, K. P. (1986) Observatory 106, 38.
Honeycutt, R. K., Robertson, J. W., Kafka, S. (2011).
Astron. J. 141, 121.
Hsi, T.-T. (1958). Smithson. Contr. Astrophys. 2,
109.
Osaki, Y. (1996). PASP 108, 39.
Patterson, J., et al. (1995). PASP 107, 1183.
Patterson, J. (2011). MNRAS 411, 2695.
Pskovskii, Y. P. (1972). Sov. Astr. 16, 23.
Richman, H. R., Applegate, J. H., Patterson, J.
(1994). PASP 106, 1075.
Ringwald, F. A., Thorstensen, J. R., Honeycutt, R.
K., Robertson, J. W. (1996). MNRAS 278, 125.
Robertson, J. R., et al. (1995). PASP 107, 443.
Robinson, E. L. (1975). Astron. J. 80, 515.
Schaefer, B. E., Collazzi, A. (2010). Astron. J. 139,
1831.
Shara, M. M., Moffat, A. F. J., Webbink, R. F.
(1985). Ap. J. 294, 271.
Skillman, D. R., Patterson, J. (1993). Ap. J. 417, 298.
Thorstensen, J. R., Taylor, C. J. (2000) MNRAS 312,
629.
15
16
YY Gem Pro-Am Project Gary et al.
Photometric Monitoring by Amateurs in Support of a

YY Gem Professional Observing Project
Bruce L. Gary
Hereford Arizona Observatory,5320 E. Calle Manzana,Hereford, AZ 85615 USA
BLGary@umich.edu
Dr. Leslie H. Hebb
Department of Physics and Astronomy,Vanderbilt University, Nashville, TN 37235 USA
Leslie.Hebb@Vanderbilt.edu
Jerrold L. Foote
Vermillion Cliffs Observatory,Kanab, UT 84741 USA
jfoote@scopecraft.com
Cindy N. Foote
Vermillion Cliffs Observatory, Kanab, UT 84741 USA
cindyf@kanab.net
Roberto Zambelli
Societa Astronomica Lunae, 19030, Castelnuovo Magra, Italy
Robertozambelli.rz@libero.it
Joao Gregorio
Affiliation, Atalaia, Portugal
j.gregorio@netcabo.pt
F. Joseph Garlitz
Elgin Amateur Observatory, Elgin, Oregon 97827 USA
garlitzj@eoni.com
Gregor Srdoc
Physical Therapist, Trainer, Viskovo, Croatia
gregor@vip.hr
Takeshi Yada
The Shimane Nature Museum of Mt. Sanbe, Ohda, Shimane 694-0003, Japan
yada@nature-sanbe.jp
Anthony I. Ayiomamitis
Perseus Observatory, Nea Palatia-Oropou, Greece 19015
anthony@perseus.gr
Abstract
When a professional observatory is used in the study of an active star, it may be necessary to establish activity
level context, such as the stars flaring frequency, presence of star spots and brightness changes. Characterizing these aspects of star activity requires time-consuming observations, which can often be obtained by amateurs. This article describes coordinated photometric monitoring by a selected team of advanced amateurs in
support of a 10-day intensive spectropolarimetry observation of YY Gem, an M dwarf eclipsing binary. One of the
authors (LH) used the 3.6-meter Canadia-France-Hawaii Telescope in Hawaii for determining magnetic field
strength maps, and she initiated the request for photometric monitoring by amateurs. The observations are undergoing analysis, yet preliminary results demonstrate the value of a coordinated professional/amateur approach
in the conduct of this type of observing campaign.
17
1. Introduction
M-dwarf stars outnumber all other spectral classifications, so it is understandable that stellar modelers are anxious to understand why their models under-estimate M-dwarf radii by 5 or 10%. Could these
stars have magnetic fields that inhibit convection in a
way that swells their size? To find out, professional
astronomer Dr. Leslie Hebb led a team of investigators in obtaining 10 nights of telescope time on the
3.6-meter Canada-French-Hawaii Telescope (CFHT)
in Hawaii to conduct spectropolarimetry observations
of a pair of eclipsing M-dwarfs, YY Geminorum (YY
Gem).
A complex picture is starting to emerge where
three different phenomena, all related to magnetic
fields, are invoked to explain the observed radii of M
dwarf stars over the entire mass range. In particular,
the existence of large star spots near the poles of the
eclipsing binary component stars could be causing a
systematic overestimate of measured radii during the
light curve analysis procedure. This project involves
constructing a new eclipsing binary model that incorporates light curve data while constraining star spot
properties from a Doppler Imaging analysis of the
spectropolarimetric spectra (explained in Section 4).
This analysis will provide the most accurate radius
and temperature estimates of the component stars.
When combined with magnetic fields maps, we hope
to better understand if and how magnetic fields are
causing the radii of YY Gem to be larger than predicted by previous models.
In order for the CFHT observations to be fully
interpreted it was necessary to supplement these observations with high-precision light curves at many
wavelengths. These data are necessary for the eclipsing binary modeling and thus essential to the success
of the project. In addition, M-dwarfs are also prone to
flaring, perhaps due to magnetic field disturbances.
Star spot variations and flaring activity of the YY
Gem pair can be evaluated using light curve monitoring, with special emphasis on the 10-day CFHT observations (2012 January 4 to 13).
Professional telescopes can cost hundreds of dollars per observing minute, so the cost of including
light curve monitoring in the proposed YY Gem project would have been prohibitive. Instead, amateurs
were considered for providing this support. After all,
eclipsing binary light curves have been made by amateurs for decades and, since the advent of CCD cameras and computer-controlled telescopes, amateur
capabilities have improved greatly.
One of the authors (BLG) has a history of recruiting amateurs for special observing projects (e.g.,
Amateur Exoplanet Archive, Pro-Am White Dwarf
18
Monitoring), and he agreed to coordinate advanced

amateurs with experience in exoplanet light curve
measurements to support the YY Gem project, from
2011 October and to 2012 January.
2. Amateur Team
Continuous observing during the 10-day CFHT
observations would have been desirable, but since no
Earth-based site has clear skies all the time it is impossible for a ground-based network of observatories
to provide continuous coverage except on rare,
brief occasions. For example, if clear skies exist half
the time at three observing sites equally spaced in
longitude, they can be expected to provide complete
coverage of any target on only ~13% of dates. Even
if the sites have clear skies 2/3 of the time the complete coverage statistics increases to only ~23%. Another strategy is to use multiple sites at each longitude, but spaced apart so that their clear sky statistics
are uncorrelated. With two sites at each longitude,
and half of nights clear for each site, complete coverage can be expected for ~43% of dates. Therefore, it
is unrealistic to expect complete coverage for any
network of observatories when several dates are involved.
This monitoring project recruited 9 observers in
3 longitude regions: 4 in Europe, 4 in the US and one
in Japan, as summarized in Table I.
Longitude
Observer
Location
Filters
+024
Ayiomamitis
Greece
Bs
+014
Srdoc
Croatia
Rs
+010
Zambelli
Italy
Bs & V
-009
Gregorio
Portugal
-110
Gary
Arizona
griz
-112
J. Foote
Utah
V & Rc
-112
C. Foote
Utah
B & Ic
-118
Garlitz
Oregon
Bs
-227
Yada
Japan
Bs
Table I. Observer list and their filter assignments. Rs, Vs

and Bs are SBIGs RGB filters.
With only one observer at Pacific longitudes

there was no expectation of continuous coverage.
With 4 observers at two longitude regions, however,
good coverage was expected for ~ 2/3 of each date;
and indeed this was achieved. During the 10-dates of
interest 93 light curves were obtained, with a median
coverage of 18.3 hours (76%). During the preceding
3 months 31 LCs were obtained. Following the intensive 10-day observations two light curves were obtained (for refining the orbital period).
3. YY Gem and Reference Star

This is a difficult target for photometry because
YY Gem is only 71 arcseconds away from the bright
star Castor. The V-magnitudes for these two stars is
9.06 and 1.58; the difference of 7.5 magnitudes corresponds to a flux ratio of 980:1.
The RA/DE coordinates for YY Gem and the

reference star are 07:34:37.45, +31:52:10.2 and
07:34:26.32, +31:51:00.1. Meridian transit occurred
near local midnight in mid-January. Based on these
all-sky calibrations YY Gem is a very red star, with
B-V = +1.389 0.023, and the reference star is moderately red, with B-V = +0.970 0.026.
4. Previous Publications
Figure 1. YY Gem (blue cirlces) and Castor, 71 arc

apart, and nearby star used for reference (red square).
In all images Castor was saturated and bleeding. Most images also had pixel edge reflections in 4
directions. Some observers rotated their CCD assembly so that YY Gem was not close to the bleeding,
pixel reflection and spider diffraction spike features
(Fig. 1 has a rotation of 11 degrees). A nearby star
was used for reference for all filter bands. An all-sky
observation produced B, V, Rc and Ic magnitudes,
from which magnitudes for other bands were calculated.
YY Gem undergoes a 19.5-hour variation in
brightness, slightly greater at shorter wavelengths, so
in Table II, the magnitudes for YY Gem and the reference star are for phase 0.35.
Band
Wavelength
[nm]
B
V
Rc
Ic
Bs
Vs
Rs
g
r
i
z
441
540
624
800
450
540
640
478
620
755
910
YY Gem
Reference
Star
10.451
9.062
8.156
7.193
10.39
9.06
8.06
9.775
8.589
7.912
7.342
10.917
9.947
9.419
8.988
10.83
9.95
9.36
10.409
9.650
9.459
9.318
Table II. YY Gem and reference star magnitudes. Rs, Vs

and Bs are SBIG RGB filters.
Observations by Kron (1952) show a 40 mmag

amplitude (80 mmag peak-to-peak) variation of the
out-of-transit (OOT) V-band magnitude associated
with the orbital period (19.5-hours). Both transits had
depths of about 0.6 mag, so the transit geometry has a
grazing V-shape.
In 1971 Leung and Schneider (1978) observed
transits with V, R and I filters (Figure 8), and these
data showed OOT amplitude decreasing with wavelength. Torres and Ribas (2002), hereafter TR2002,
measured radial velocities spectro-scopically, reviewed previous mid-transit timing data to derive an
improved ephemeris, and modeled the 1971 light
curves using a star spot model. Their model solution
showed that both stars were virtually identical, with
radii of 0.62 solar and masses of 0.60 solar. Their
ephemeris is HJD = 2449345.112327(87) +
0.814282212(12) E, where E is an integer number
of orbits.
TR2002 found that the 1971 transit depths decrease with wavelength differently for the primary
transit (611, 596, 563 mmag for V, R and I) and secondary transit (544, 539, 544 mmag). Between primary and secondary transits the OOT level is constant. However, from phase 0.4 to 0.6 (including secondary transit) the OOT level fades ~ 45 mmag and
fully recovers by primary ingress (phase 0.9).
Qian et al. (2002), hereafter referred to as
Q2002, suggest that the orbital period exhibits a secular decrease, which might be due to mass loss due to
magnetic fields and stellar winds. Their ephemeris
for the primary transit is HJD= 2424595.8178(1) +
0.81428258(1) E 1.20(2) x 10-11 E2. This
ephemeris corresponds to a continuous period decrease of dP/dt = 1.08 10-8 day/year (-90 milliseconds/century).
Both the TR2002 and Q2002 ephemerides ignore
the possibility that changing star spots will affect
mid-transit timings, so their quoted SEs overlook this
source of systematic uncertainty.
5. Results of Present Observations

Normally, it is a trivial matter to assign one transit of an eclipsing binary as the primary. But when
19
the transits have almost the same depth, and when

star spots are present, it is possible that transit depth
relationships can change on yearly timescales. This
problem is greater when the period is short and the
interval between observations is large. The latest
published YY Gem ephemeris (Q2002) is based on
observations from 1971 to 1998, and since the orbital
period is ~0.814 days there are >5800 periods between 1998 and the present observations. Confusion
is compounded by the possibility that the period may
be decreasing, or is variable.
An assignment of primary was made for the
high quality 2011 November 27 transit, which occurred 8.1 0.2 minutes late compared to the
TR2002 ephemeris. If this transit were a secondary
then it would be occurring 578 minutes early. A provisional identification was needed for archiving purposes so this event was provisionally adopted as primary.
Figure 3. Depths of primary (red) and secondary

(blue) transits. The 1971 transit depths are shown with
large circles.
During the 2011 observations (56 nonoverlapping hours) only one small flare was observed. However, during the 10-day intensive observing dates (180 non-overlapping hours), 20 flares were
present (on 9 of the 10 days). Thats 1.1 0.2 flares
per 10-hour nominal observing session, versus 0.2
0.2 flares per 10 hours from 2011 October through
December. In other words, the 10-day intensive observing period occurred when YY Gem was actively
flaring. Flare activity was present at all phases, although 5 occurred between 0.69 and 0.83.
Figure 2. Typical light curve, showing a primary transit

followed by a flare.
A typical light curve (LC) is shown in Figure 2.

It was obtained by Gregor Srdoc (Croatia) on 2012
January 6 using a 0.30-m telescope with a Rs-band
filter. The information boxes state that transit depth is
538 1 mmag, transit length = 1.738 0.003 hours,
mid-transit occurs at 23.250 0.001 UT, or JD =
2455933.46875 0.00006 (with a BJD equivalent).
On this observing session overlapping transits lasted
21.1 hours, and three flares were observed.
An analysis of transit depths is shown in Figure
3. The deepest transits are the ones identified for the
current analysis as secondary. These transits vary
greatly with wavelength, in contrast with the shallower primary transits. Also shown in this figure
are the depths of the 1971 transits. It is curious that
the 1971 primary transit depths exhibit a similar behavior to the current secondary transits; and the
1971 secondary transit depths resemble the current
primary transits.
20
Figure 4. Light curve of the same transit shown in Fig. 2

but made with a Bs-band filter, showing that the flare
amplitude is greater at shorter wavelengths.
Flare rise and decay times can be estimated from

Fig. 2 data (using un-averaged data, with 8-second
spacing); from onset to peak takes 2.6 minutes, while
the one-half life decay time is 7.9 minutes. The decay
appears to be complete in 40 minutes. Each of the 21
flares observed exhibit an intensity that decreases
with wavelength.
Figure 5. Flare brightness versus wavelength for the

event of 2012 January 7, 02.1 UT.
ered there appears to be a residual brightness of ~ 80

mmag that persists for > 1 hour.
Out-of-transit (OOT) brightness varied in an interesting way as a function of orbital phase and wavelength. Establishing OOT shape is difficult due to
differences in extinction between the target star (YY
Gem) and the reference star. Since YY Gem is redder
than the reference star it will fade at a lower rate with
increasing air mass, leading to a bowl shaped LC.
Model-fitting LC observations involves removal of
this differential extinction effect. For stars that dont
vary in brightness during the LC the model fitting is
correctly identifying bowl shapes that are correlated with air mass and removing this air mass curvature (AMC) variation by adjusting a differential
extinction parameter. Provided that every LC uses the
same reference star, AMC should be the same for all
LCs, so the same value should be found for the differential extinction parameter for all such LCs.
However, if there is a real OOT variation with a
minimum near the middle of the LC the fitting procedure will attempt to remove this component of curvature as if it was being produced by a differential extinction unique to that LC. This would lead to removal, or an underestimate, of the actual OOT variation. Similarly, if the OOT really exhibits a maximum near the middle of the LC the fitting procedure
will reduce the amplitude of any real inverted bowl
OOT shape. In other words, the LC fitting procedure
will reduce the amplitude of any OOT variation that
is actually present. This problem would not exist if
the target and reference stars were the same color, but
with YY Gem being so extremely red there were no
similarly colored stars within a typical field-of-view
that could be used for reference.
Figure 6. Flare during a transit, observed by Zambelli in

Italy with Bs-band filter.
The flare in Fig.s 2 and 4 were observed by four

observers using 7 different filters, which allows a
determination of the flares spectral index to be
made over a wide range of wavelengths. This is
shown in Figure 5, where the logarithm of brightness
and wavelength are used to achieve a linear relationship. The slope of such a line is spectral index
which in this case is 3.2 0.3. This spectral index
requires that the flare emission have a non-thermal
origin, and it is reasonable to speculate that a magnetic disturbance is involved. Flaring activity was
maximum on the first of the 10 intensive observing
dates (Session #1, 2012 January 3/4), and gradually
subsided to zero on the last date (January 12/13).
Figure 6 shows the largest flare encountered,
which occurred during a transit (on the first of the 10days of intensive observations). After the flare recov-
Figure 7. Phase-folded g-band light curve for 8 dates,

using a 16-parameter model.
21
One solution for this problem is to simultaneously solve for both AMC and OOT shape, in a procedure that involves all LCs for a given filter band. A
more straightforward solution is to first estimate
AMC for a given filter band, based on all LCs for
that band, then re-do the model fitting using the average AMC, and finally phase-fold the resulting LCs.
This is a less elegant approach, but it is simpler to
implement, and it is the approach used for this analysis. Figure 7 is an example of a g-band phase folded
LC using this 3-step procedure.
correct, and assuming the OOT shapes are due to star

spots that persist for decades, then there must have
been a major change in star spots between 1971 and
2012.
Figure 9. Phase-folded z-band light curve for 8 dates,

using a 16-parameter model.
Figure 8. Phase-folded light curve for 3 bands, based on

measurements in 1971 (Leung and Schneider, 1978) and
model-fitted by Torres and Ribas (2002). The ordinate is
a magnitude scale. Phases of 0.4, 0.6 and 0.9 are indicated by vertical lines.
Notice in Fig. 7 that OOT shows a minimum at

phase 0.2, just after egress of the transit provisionally
identified in this analysis as primary. Compare the
OOT feature with what was found in a phase-folded
LC based on 1971 measurements, shown as Figure 8.
Each of the filter bands show a minimum just after a
secondary transit, at a phase of 0.6 or 0.7. As noted
by Leung and Ribas, who performed the model fit to
the 1971 data, all of the fade near the secondary transit occurs between phases 0.4 and 0.6, and a recovery
occurs between 0.6 and 0.9 just before primary ingress. Most of this OOT variation can be seen in the
present data, but with a 0.5 phase shift. In other
words, if the present data had an incorrect primary/secondary identification, the OOT shapes for
1971 and 2012 would exhibit a better match.
The 1971 phase-folded LC shows a slightly convex shape (brightening in the middle) between primary and secondary transit. If we reverse the primary/secondary identification for the 2012 OOT,
then it would exhibit a concave shape. If the primary/secondary assignments for the 2012 data are
22
Figure 10. OOT variation amplitude (1/2 peak-to-peak)

versus wavelength.
Figure 9 is a z-band version of Fig. 7. It exhibits

a much smaller OOT variation: ~8 mmag versus ~19
mmag at g-band. The phase-folded LCs at r- and iband (not shown) have OOT amplitudes of 15 and 14
mmag. Figure 10 shows OOT variation amplitude
versus wavelength.
Assuming the OOT variation is caused by star
spots they must have less temperature contrast at the
stellar atmosphere altitude where z-band emission
originates. This will constrain the atmosphere model
If star spot changes are not responsible for the

change in OOT pattern between 1971 and 2012, the
re-assigning of primary/secondary to the 2012 data
will require a period change. The implications of this
are important, for it would require speculations about
mass loss (Q2002). Before discussing this interesting
possibility lets consider the information present in
the 2012 LCs about P, and a possible discrepancy
between P for 2011/2012 and those derived by
TR2002 and Q2002.
From 2011 October to 2012 April a total of 127
LCs were obtained (31 in 2011, 93 during the 10-day
CFHT observations, and 2 following); 44 of these
included complete transits. Some of the mid-transit
times were measured with an accuracy of ~10 seconds. Figure 11 is a plot of differences between
measured mid-transit time and a best-fitting period
solution. For this plot it was assumed that the secondary transit occurs exactly mid-way between primary
transits, which doubles the number of transits available for such an analysis.
Figure 11. Mid-transit time differences versus date, using both primary and secondary transits, from 2011 October to 2012 April. The differences are with respect to a
solved-for period.
The best-fitting P, used for reference in Fig. 11,

is 0.81428330(46) day. This period was solved for
using chi-square minimization. The Q2002 ephemeris
calls for a P on 2012 January 7 of 0.81428372(2) day.
The difference between these two periods is +36 43
milliseconds, which means that the present data are
compatible with the Q2002 suggested P decay rate.
An alternative ephemeris is given by Q2002 (that
assigns weights to measurements slightly differently)
which assumes that P does not change: HJD =
2424595.8172 + 0.81428254 E. The difference
between the current P measurement and this constant P solution is 66 millisecond (no SE is given
by Q2002 for the constant P, but we may assume that
SE ~ 60 ms). The current P estimate is therefore insufficiently accurate to favor either of the Q2002
period solutions.
The P solution in Fig. 11 differs from the
TR2002 P by 35 49 millisecond. Therefore, the P
solution from the 2011/2012 transit timing data (Fig.
11) produce a P value that is compatible with the
TR2002 P solution. Whereas the P determinations are
compatible, the 2011/2012 transits all occur ~8.0
minutes later than predicted by the TR2002 ephemeris. The TR2002 P is assigned an accuracy based on
precision, and since star spot effects were not considered as a systematic source of uncertainty in their 73year database we suggest that the 8 minute offset for
all 2011/2012 transit events can be used to refine the
TR2002 ephemeris. Specifically, we suggest the following ephemeris: BJD = 2456024.67475(6) +
0.81428290(5) E.
We should ask if a primary/secondary reassignment to the 2011/2012 data is compatible with
TR2002 and Q2002 P solutions, and also compatible
with the P measurement from the present data. Lets
consider reassigning a 2012 transit from secondary
to primary status, and calculate the smallest change
to a linearly decreasing P that would be consistent
with this reassignment. Consider the LC on 2012
January 7 by Gregorio; mid-transit time is BJD =
2455934.69633(6), and this transits designation was
secondary. If we use the Q2002 ephemeris and
change the number of orbits for it by the smallest
fraction of an orbit that would make this transit correspond to a primary transit, then for this JD interval
the average P would be 0.81427179(2) day. Such a P
value is statistically incompatible with the Q2002
constant P, the Q2002 decreasing P (both by 400sigma) and the TR2002 P (400-sigma). Moreover, it
is incompatible with the P determined from the
2011/2012 data (21-sigma). We conclude that YY
Gem is not undergoing period changes by amounts
that have been measured. Instead, we suggest that
star spot changes on yearly to decade timescales distort the LC timings in ways that can be misinterpreted
as a changing ephemeris.
6. Spectropolarimetric Observations
Magnetic fields can be directly measured on low
mass stars through observations of the Zeeman Effect. This can be done when an atom or molecule is
in the presence of a magnetic field, once degenerate
energy levels undergo a small separation. This phenomenon can be detected in stars through both the
23
splitting and the polarization of the absorption features in observed stellar spectra.
In practice, the polarization signature is sensitive
to both the field strength and its orientation. Furthermore, as the star rotates, an individual magnetic feature on the surface will produce a slowly varying
distortion in the spectral line profile over time. Thus,
by obtaining a time series of polarized spectra over
the full range of rotational phases of a star it is possible to construct images of the large-scale component
of the surface magnetic field. This technique is referred to as Zeeman Doppler Imaging (ZDI) (Donati
et al., 1997; Semel, 1989). If a time series of unpolarized spectra is also obtained, a large-scale brightness
map of the stellar surface (i.e. star spot map) can be
derived through a Doppler Imaging (DI) analysis.
We obtained a series of circularly polarized spectra of YY Gem during the 10-night CFHT observing
run, densely covering all orbital (and rotational)
phases of the binary. We also obtained a corresponding series of unpolarized spectra. These data will be
used to derive magnetic field maps and surface
brightness maps of both M dwarf stars in the binary
using ZDI and DI analysis techniques.
The ultimate goal of the project to explore in detail how magnetic fields and the resulting surface
spots influence the structure and temperature of low
mass stars. It is now well established that theoretical
stellar evolution models underestimate the radii of M
dwarf stars by 5-10% and overestimate their temperatures by a few 100 degrees.
Preliminary maps of magnetic field and star
spots on both of the YY Gem stars have been produced, so we are confident that final results will contribute to our understanding of model adjustments
needed to account for M dwarf sizes.
7. Conclusion
When a professional astronomer is conducting
observations of an object that is likely to be photometrically variable it may be possible to rely upon a
network of advanced amateurs to document that variability. In the case of the eclipsing binary YY Gem,
spectropolarimetric observations were made at the
CFHT during a 10-day observing run, and on the
same dates a network of amateurs monitored flare
activity, transit depth, transit timing, transit shape and
OOT variability with ~76% coverage. Similar
photometric observations were made before and after
the 10-day run, which permitted flare activity context. These features were characterized as a function
of wavelength, from B-band to z-band.
The amateur photometry light curves show a
significant change in transit depth, depth versus
24
wavelength and OOT shapes, compared with 1971

light curves. Two explanations were considered: 1)
the orbital period has changed by one-half orbit during the past few decades, or 2) star spots have
changed significantly during that time in a way that
produces the above changes. We conclude in favor of
the second explanation. This was an unexpected finding and further illustrates the merits of extensive
photometric support of professional observations.
All light curves and additional results can be
found at a web page maintained by one of the authors
(BLG, http://brucegary.net/yygem/).
8. References
Donati, J.F., Brown, S. F. (1997) Zeeman-Doppler
Imaging of Active Stars. V. Sensitivity of Maximum
Entropy Magnetic Maps to Field Orientation.
Astron. & Astrophys, 326, 1135-1142.
Leung, K. C., Schneider, D. P. (1978). YY Geminorum I. Photometry and Absolute Dimensions. AJ,
83, 618-625.
Qian, S, Liu, D., Tan, W., Soonthornthum, B. (2002)
YY Geminorum: A Very Late Type Close Binary
with Possible Magnetic Stellar Wind. AJ, 124, 10601063.
Semel, M. (1989). Zeeman-Doppler Imaging of Active Stars. I Basic Principles. Astron. & Astrophys,
225, 456-466.
Torres, G., Ribas, I. (2002) Absolute Dimensions of
the M-Type Eclipsing Binary YY Geminorum (Castor C): A Challenge to Evolutionary Models in the
Lower Main Sequence. AJ, 567, 1140-1165.
Lowell Amateur Research Initiative Hunter et al.
The Lowell Amateur Research Initiative

Deidre Ann Hunter
Lowell Observatory
1400 West Mars Hill Road
Flagstaff, AZ 86001
dah@lowell.edu
John Menke
MenkeScientific
25000 Old Hundred Road
Barnesville, MD 20838
john@menkescientific.com
Bruce Koehn
Lowell Observatory
1400 West Mars Hill Road
Flagstaff, AZ 86001
koehn@lowell.edu
Michael Beckage
370 Clipper Way
Seal Beach, CA 90740
michael@discoverthestars.net
Klaus Brasch
krbrasch@earthlink.net
Sue Durling
900 Switzer Canyon Drive #101
Flagstaff, AZ 86001
sdurling@juno.com
Stephen Leshin
Sedona AZ
sleshin@yahoo.com
Abstract
Lowell Observatory has a rich history of amateur research, starting with its founder Percival Lowell. With this
SAS meeting, we are announcing the launch of a new initiative to involve amateurs in scientific research: The
Lowell Amateur Research Initiative (LARI). This program pairs qualified amateurs with professional astronomers
at Lowell Observatory to carry out collaborative research projects. I outline the process for becoming involved,
starting with a web-based interface for submitting applications, and the projects that are currently available. We
look forward to engaging with the amateur community and expect the program to grow in scope as we gain experience.
1. Introduction
Lowell Observatory has a long history of professional/amateur collaborations. Lowell himself was an
amateur, but collaborated with professional astronomers all over the world. With this meeting, we are
announcing the Lowell Amateur Research Initiative

(LARI), which enables research collaborations between amateur and professional astronomers. Although amateur-professional collaborations have
taken place at Lowell Observatory in the past, this
program provides a more formal means of promoting
25
the involvement of amateur astronomers in research

at Lowell.
2. Past Collaborations
Amateur astronomers have contributed significantly to research at Lowell. Here are a few recent
examples.
3. Current Projects
Lowell Observatory astronomers are currently
conducting several projects that would benefit from
participation by amateur astronomers. More are likely
to be added to this list as the program becomes established and as our scientific staff grows.
3.1 LITTLE THINGS (Deidre Hunter)

2.1 LONEOS
The Lowell Observatory Near-Earth Object
Search, directed by Ted Bowell, was aimed at discovering Earth-crossing asteroids. From 1993 to
2008, LONEOS discovered hundreds of NEAs and
honed their positions. Four amateurs Paul Johnson,
James Ashley, Robert Cash, and G. H. Bliss
searched LONEOS images for asteroids that were
missed by an automatic software procedure. They
discovered numerous objects too faint for the
LONEOS software, including 2003 SQ222, which
came within one-fifth of the distance to the moon
(Cash et al., 2003). Amateur Charles Juels also undertook follow-up observations to determine the orbits of hundreds of LONEOS discoveries.
LITTLE THINGS is a multi-wavelength study of

41 dwarf irregular galaxies. The goal of the project is
to understand how the tiniest galaxies in the universe
form stars from the gas between the stars
(http://www2.lowell.edu/users/dah/littlethings/index.
html). Determining the extent to which external interactions with other galaxies have affected their star
formation histories is very important.
An amateur with a large telescope, a high quality
CCD, and dark skies could contribute significantly by
obtaining ultra-deep, stacked images of the LITTLE
THINGS galaxies over a wide field of view. People
who have done this for other galaxies have found
amazing debris left over from galaxy-galaxy interactions or satellite cannibalism (see, for example,
Martnez-Delgado et al., 2012), and this has changed
our understanding of the factors influencing galaxy
evolution.
Figure 1. NEA 2003 SQ222 was discovered by volunteer

Robert Cash. These images show its motion over 41
minutes. The day before, 2003 SQ222 had passed just one
fifth of the Moons distance from Earth. Image obtained
from online Whats Happening Magazine, June 10, 2011;
article by Ted Bowell.
2.2 Mars
Amateur Jeff Beish worked with Leonard Martin
at Lowell Observatory on monitoring dust storms on
Mars (for example, Martin et al., 1991).
2.3 Comets
Amateur Joseph Marcus has analyzed sunlight
scattering from Sun-grazing comets (for example,
Marcus, 2007). David Schleicher at Lowell used his
analysis to build a full phase angle dust scattering
model for Comet Halley and determine the comets
emission properties.
26
Figure 2. Dwarf irregular galaxy IC 1613. What lies hidden to our eyes because it is faint or lies outside the
field of view of our current images? In this image of IC
1613, blue is far ultraviolet light from young stars, green
is green (V-band) light from middle-aged stars, and red
is atomic hydrogen gasthe fuel for new star formation.
The data are from the LITTLE THINGS project. The color
composite image was produced by Lauren Hill.
3.2 Asteroid Recovery (Bruce Koehn)

Data mining of old LONEOS frames could potentially lead to the discovery of new asteroids and
old asteroid positions that were not found the first
time and to improved photometry of all asteroids. A
test on a few frames yielded several asteroids that
were missed in the automated search.
3.3 Asteroid Light Curves (Bruce Koehn)

Amateurs with a modest telescope and a CCD
camera can participate in an asteroid light curve
campaign. The primary targets will be slowly rotating
asteroids since they are difficult for a single observer
to cover successfully. Additional targets will be Near
Earth Asteroids. Web page assistance would be
greatly appreciated for this project.
project would take images of these objects for purposes of astrometry. Because the objects are faint
(>18 for Centaurs and >20 for KBOs), this project
requires a fairly large telescope, large field view, and
dark skies.
3.6 Transiting Exoplanets

(Georgi Mandushev)
Help Georgi discover exoplanets, variable stars,
and other transient objects through data mining. He
has reduced photometric data for several fields near
the Milky Way. The fields are 6x6 degrees and contain tens of thousands of stars 10th magnitude and
fainter. The data are suitable for identifying and classifying variable stars, other objects, and unidentified
transiting planets.
4. How to Participate
3.4 Monitoring Low-mass Stars
(Evgenya Shkolnik)
Stars with masses less than about half the mass
of the Sun make up 75% of the Galaxys stars. An
amateur with a good photometer and an optical or H
filter will monitor young and active low mass stars to
determine their rotation periods as a function of age,
look for eclipsing binary stars, and look for evidence
of transiting giant planets. Stellar magnitudes are
primarily 12-13 mag in V-band.
Here are the steps for becoming involved in

LARI:
1. Go to the LARI web site:
http://www.lowell.edu/LARI_welcome.php.
2. Create a LARI login account.
3. Set up your research profile.
4. A committee at Lowell identifies an amateur
whose skills and interests appear to be well
matched to a particular project.
5. The astronomer contacts the amateur and they
get to work.
5. Acknowledgements
Figure 3. Light curve of the low mass star BD 22 5866,

with an eclipse period of 2.21 days. The difference in
starspot intensity from two different epochs is clear in
the light curve. From Shkolnik et al. (2008).
3.5 Orbits of Centaurs and Kuiper Belt

Objects (Larry Wasserman)
Centaur asteroids and Kuiper Belt Objects often
have relatively short orbital arcs, resulting in inaccurate sky-plane positions. The amateur involved in this
We appreciate the work that Antoinette Beiser

has put into organizing LARI and bringing the
Lowell Pro/Am Committee together. We also acknowledge Tom Vitron for providing text for the
LARI web site. We are grateful to Lowell Observatory and John and Meg Menke for funding participation in this SAS meeting.
6. References
Cash, R. A., Van Ness, M. E., Skiff, B. A. (2003).
(2003). 2003-RO3. Minor Planet Electronic Ciruclar.
27
Marcus, J. N. (2007). Forward-Scattering Enhancement of Comet Brightness. I. Background and

Model. International Comet Quarterly 29, 39-66.
Martin, L. J., Beish, J. D., Parker, D. C. (1991).
Martian Dust Storms in 1990. Bulletin of the
American Astronomical Society 23, 1217.
Martnez-Delgado, D., Romanowsky, A. J., Gabany,
R. J., et al. (2012). Dwarfs Gobbling Dwarfs: A
Stellar Tidal Stream Around NGC 4449 and Hierarchical Galaxy Formation on Small Scales. Astrophysical Journal Letters 748, L24.
Shkolnik, E., Liu, M. C., Reid, I. N., Hebb, L., Collier Cameron, A., Torres, C. A., Wilson, D. M.
(2008). BD -22 5866: A Low-Mass, Quadruplelined Spectroscopic and Eclipsing Binary. Astrophysical Journal 682, 1248-1255.
28
Lunar Meteor Impact Monitoring - Cudnik
Lunar Meteor Impact Monitoring

and the 2013 LADEE Mission
Brian Cudnik
Department of Chemistry & Physics
Prairie View A&M University
P.O. Box 519, M.S. 2230
Prairie View, Texas 77446
Abstract
For over twelve years, amateur and professional astronomers have been monitoring the moon regularly for the
tell-tale point-like flashes of impacting meteoroids. This endeavor started with my own visual sighting of a lunar
Leonid meteor impacting the dark hemisphere of the waxing gibbous Moon on 1999 November 18 (UT). Since
then, the Association of Lunar and Planetary Observers (ALPO) has established the Lunar Meteoritic Impact
Search (LMIS) section, which I have served as coordinator since early 2000, and we have logged over 80 lunar
meteor impact candidates. In this presentation I will provide a brief overview of lunar meteor monitoring techniques, including some of the equipment that people have been using in the field. I will outline how anyone can
become involved in this ongoing, fruitful study of this dynamic aspect of our solar system. Finally I will introduce
the upcoming LADEE moon mission and discuss how amateur astronomers can be involved in gathering useful
scientific data during the Perseid 2013 maximum, and afterwards, in support of this mission.
1. Introduction
Since the invention of the telescope, and even
before that, the occurrence of Lunar Transient Phenomena (LTP) has perplexed astronomers, challenging (especially in more recent centuries) the paradigm
that the moon is changeless and dead. LTP has
taken on many forms, including point flashes that
appear and disappear in a fraction of a second, but
have only been documented by single observers of
varying skill levels. There are relatively rare instances of small groups of people at the same telescope or the same observing site documenting LTP,
but no reports of two well-separated observers (by at
least 50 km), to my knowledge, have been recorded
until 1999.
Since 1999, when the first scientifically confirmed observations of lunar meteor phenomena were
recorded, astronomers at various levels have recorded, fairly regularly, the occurrence of point
flashes on the moon caused by meteoroid impacts.
These turned out to be much more common than lunar eclipses, but their random nature confounded
observations of them for centuries. A handful of systematic attempts at observing and cataloging lunar
meteoroid impacts have been attempted throughout
the 20th century, with one of the most notable being
the Association of Lunar and Planetary Observers
Lunar Meteor Search program which lasted from
1955 to 1965 but failed to provide simultaneous independent observations of lunar meteors. This is not
surprising given that the observations were visual at
the time, and it is extremely challenging to document

a lunar meteor strike solely by pure visual means. But
technology (and a little serendipity) would change
that near the dawn of the 21st century.
NASA is planning to launch the latest in a series
of spacecraft to the moon, in 2013. The spacecraft,
dubbed Lunar Atmosphere and Dust Environment
Explorer (LADEE) is scheduled to launch in 2013
July. Collaborations with universities and the amateur
astronomy community are sought for this worthwhile
project.
2. Historical Background
2.1 The First Confirmed Lunar Meteor
Impact Observations and the Formation
of ALPO-LMIS
Prior to the uneventful impact of Lunar Prospector in 1999 July, during the fall of 1998, Martin
Beech and Simona Nikolova, of the University of
Regina, made specific predictions about lunar
Leonids, published in their paper (Beech et al.,
1998). They realized that the moon would be entering
an especially dense filament of meteoroids during the
November Leonid meteor shower. David Dunham,
president of IOTA (International Occultation Timing
Association) coordinated an effort to videotape the
unilluminated side of the waxing gibbous Moon during the peak of activity November 18 (3h to 6h UT).
At that time, most setups involving video cameras
and telescopes were unable to take in the entire
29
Moon, so the job was partitioned into thirds. People

whose last names began with A through H would
monitor the northern third of the moon, etc. so as to
cover the entire darkened crescent of the waxing
Gibbous moon.
Not being equipped with video but having access
to a 14-inch f/11 Cassegrain, I visually monitored the
moon for impacts and spotted one at 4:46 UT. This
was subsequently confirmed on Dunhams video and
this confirmation motivated others to check their videos for additional impacts. Initially, six impact
events, showing themselves as point flashes of light,
were observed and confirmed. New software, developed by Peter Gural, of SAIC (Science Applications
International Corporation) for fireball detection was
used to located additional impact events on peoples
videotapes. This method resulted in five additional,
unconfirmed impact candidates being found.
Astronomical Association, more groups, individuals,

and organizations become involved toward the middle of the first decade of the 21st century. A group
from Japan, observing the moon during the Perseid
meteor shower of 2004, netted a confirmed Perseid
meteoroid impact, the first confirmed non-Leonid
lunar meteor. A year later, the Meteoroid Environment Office of NASA Marshall Space Flight Center,
Alabama, USA, began a regular program to monitor
the moon for two weeks per month with multiple
telescopes and video cameras. The first impact event
candidate was recorded in early 2005 November, a
likely Taurid meteoroid. Since then, this program has
recorded over 200 meteor impact events on the
Moon.
The observers who contributed to ALPO-LMIS
have, over the years, recorded some 80 viable candidates, but only a handful of these were scientifically
confirmed. In most cases, the observer was working
alone or happened to be looking at the moon for other
reasons when he/she is surprised by an impact event.
In recent years, however, with the collaboration with
NASA-MSFC, several of our active members have
recorded impacts that have been confirmed by NASA
and vice versa. Figure 2 shows an image of a lunar
Geminid impact that I recorded by video in 2010
December.
Figure 1. Image of Leonid impact A recorded on 18

November 1999 by David Dunham (image courtesy of
David Dunham).
In early 2000, the Association of Lunar and

Planetary Observers formed the Lunar Meteoritic
Impact Search section, for which I was made the coordinator (and remain so to this day). The section
enjoyed much activity in the first year of operation,
with many observers reported what would later be
deemed as spurious (non-impact) events caused by
retinal flashes or cosmic ray hits. By the second year,
the activity had tapered off, but enjoyed a brief resurgence in 2001 November when the Moon was once
again favorably placed to receive impacts from a
relatively dense ribbon of meteoroids from the
Leonid stream. Two confirmed and two probable
impact candidates were recorded with this event.
2.2 More Organizations get Involved

In addition to ALPO-LMIS, IOTA, the Italian
Geologic Lunar Researches Group, and the British
30
Figure 2. Image of a confirmed lunar Geminid impact,

recorded on 14 December 2010 by Brian Cudnik.
2.3 Watching Spacecraft Crash

The ALPO-LMIS, NASA-MSFC MEO and
other organizations were active in monitoring any
ground-based flashes or plumes produced by the impact of several spacecraft. The first was SMART-1
(European Space Agency) which was commanded to
crash-land on the Moon on 2006 September 3. A
campaign of ground-based amateur and professional
telescopes and observatories was carried out, and
nearly everyone observed a negative (i.e., no flash)
result. However, professional astronomers at the

Canada-France-Hawaii Telescope Corporation recorded a flash in the infrared (2,122 nm) with the
WIRCam wide-field infrared camera on the 3.6-meter
optical/infrared telescope on the summit of Mauna
Kea in Hawaii. The gif-animation posted on their
website,
http://www.cfht.hawaii.edu/News/Smart1/
shows nicely the flash followed by the resultant dust
cloud of material kicked up by the impact.
Nearly three years later, the Japanese Kaguya
spacecraft was commanded to crash into the moon on
2009 June 10 at 18:25UT. This was daylight for
North America so ALPO-LMIS did not coordinate
this one. Astronomers in Asia were able to observe
the impact flash resulting from this man-made event.
Five months later, in 2009 October, the twin impacts of LCROSS (Lunar CRater Observation and
Sensing Satellite) centaur upper stage rocket and
LCROSS itself were observed from ground-based
observatories all over the moon-facing hemisphere of
the Earth. Nothing was observed from the ground,
but LCROSS did capture plume images on its way to
its own demise (Figure 3). It also detected signs of
water (not the water itself, but two convincing signs
that water is present, from the profile of the near infrared spectrum recorded by LCROSSs spectrometer
to the ultraviolet signature of the hydroxyl (OH)
molecule formed when H2O encounters solar radiation and is disassociated into H and OH molecules.
3. Current Work on Lunar Meteors

3.1 Lunar Meteoroid Impacts and how to
Observe Them
Published in 2010, this first-of-its-kind volume
describes the various considerations that go into a
lunar meteor monitoring program. It is a collection of
work from various sources on all the stages involved
in this sort of endeavor. This book provides a useful
reference and training reference for those getting
started with lunar meteor monitoring. It also provides
resources for those who have been involved in astronomy for some time.
As the name of the book implies, it starts with
some background information about the cratering
history of the moon and the history of LTP sightings
from the 1500s onward. The LTP catalog it provides
focuses on events that resemble manifestations of
meteoroid impacts and includes only the events that
have the highest confidence ranking. Many of these
were observed by skilled observers, giving the events
significant credibility.
The second half of the book focuses on the tools
and techniques of lunar meteor observation. I focus
on the video-equipped observer, but I also do not
leave out the visual observer. I provide references
and resources to provide observers with what is
needed to carry out a fruitful campaign, and I share
the dates of some of the more significant lunar meteor opportunities for years to come.
3.2 Current and Ongoing Efforts
Figure 3. LCROSS image of impact plume generated by

the crash of a centaur upper stage rocket seconds earlier. This event was instrumental in providing conclusive
evidence for water ice existing in the permanently shadowed craters of the Moons polar regions. Image courtesy of NASA.
The ALPO-LMIS continues to coordinate a

world-wide effort in monitoring the moon for meteoroid impacts. It does so through its website, through
informal e-mail communication, papers at local conferences, and a list-server called lunarimpacts@yahoogroups.com. The level of activity and
participation varies from year to year depending on
the circumstances of the moons position during annual meteor showers.
NASA-MSFC also continues to monitor the
moon regularly and has, as of 2011 November, recorded 233 impact candidates. Most of these are from
sporadic meteors, but a number of annual showers are
represented in the impact list. They monitor the moon
about 10 nights per month, when the illumination is
between 10 and 55 percent. The distribution of impacts may be more a reflection of where the monitoring is centered (from roughly 30 degrees latitude
and from 20% radius from disk center along the lunar
equator to the limb).
31
Systematic imaging to study the ray structures of

the moon at various phases not limited to full
moon
Imaging for photometric and radiometric studies
Imaging of earthshine to gather information of
the Earths albedo over time
Filtered and unfiltered imaging of the moon
throughout its phase cycle to look for and characterize other forms of LTP
High cadence observations of lunar eclipses to
determine crossing times of lunar features by the
Earths shadow as well as estimates of contact
times of the shadow with the lunar disk
Figure 4. Map showing the locations of the 233 meteoroid impacts observed by NASA-MSFC from late 2005 to
late 2011. The clustering of impact sites visible in the
image may be more of a selection effect, refer to the text
for more details. This image is courtesy of NASA-MSFC
Meteoroid Environment Office.
In both the ALPO and the NASA cases, monitoring takes place during a one week period from a few
days after new until a day or two after first quarter,
then again a day or two before last quarter until a few
days before new. In these cases, the few days will
last between 3 and 5 days, depending on ecliptic angle at a given observing location. If an annual shower
is active, it will be monitored as part of the two week
monitoring period, with a possible extension beyond
the set periods to accommodate a shower peak or
extend the monitoring due to lingering shower activity.
Collaboration continues between NASA-MSFC
and ALPO-LMIS. In both cases, the preferred monitoring method involves video and in both cases the
video is analyzed by programs such as LunarScan or
UFOCapture to capture the transient flashes. Usually
one party discovers the event while the other party
confirms with their videotapes and vice versa.
3.3 CMOS Imaging

As part of our program to include new technologies (or at least technologies not widely used but potentially beneficial for research), we are also employing still imagery of the moon. We are using a CMOS
(Complementary Metal Oxide Semiconductor)
imager, assembled by Holland et al. (2008, 2009) to
do systematic imaging of the illuminated lunar disk
to look for several things (Cudnik et al., 2011a):
32
Figure 5. Ground-based image of the entire lunar disk

taken on 23 November 2010, obtained with a CMOS electronic detector and a Celestron NexStar 8-inch (20 cm)
Cassegrain telescope. Image Credit: Prem Saganti, Gary
Erickson, and Brian Cudnik.
We will be working on the above in conjunction

with other groups such as ALPO to maximize the
value of our collaborative efforts. The goal is to enable the determination of a viable cause for other
form of LTP in a similar manner as lunar meteor observations have been successfully used to validate
and explain the point flash form of LTP.
3.4 Observing Program Nuts and Bolts

We are always welcoming observations of the
moon in search of lunar meteoroid impacts. Although
we prefer the video method for observing and recording lunar meteors, we do not want to leave the
visual observer out. Visual observers may be such
due to personal preference or economics (financial
difficulty preventing the purchase of equipment for
video observation and data reduction). The advice I
offer visual observers is this: ensure you are well
rested and focused before your first observing ses-
sion, use a short wave radio and tape recorder and

blank moon map to record your observed candidates,
take frequent breaks after 10 or 15 minutes of observation to minimize fatigue (and resultant lost impact
events), and finally, if there is any doubt, call it out
anyway. Given the extremely transient nature of lunar meteoroid impacts, a very quick but noticed flash
may very well turn out to be a genuine event. This, of
course, is dependent on simultaneous video observation of the same region of the moon at the same time.
This brings me to video observations. The active
lunar meteor observer George Varros has an excellent
tutorial on his website, www.lunarimpacts.com,
which provides a handy walkthrough, from start to
finish, of a typical observing run. A typical setup
includes a telescope with a wide field of view, or a
focal reducer, a low light black and white CCD video
camera (like what people find with security vendors,
an example is a Watec 902H2 Ultimate) a video time
insertion / GPS device for recording time and geographic position, a recording device such as a digital
video camcorder or a portable DVD recorder, and the
means to store vast amounts of data. The equipment
is set up to monitor the moon for one or more hours
and the video is recorded for later analysis. One may
also record nearby stars for magnitude estimates of
any recorded impact candidate.
Once the observing run is complete, or during
the observing run if one is capable of doing so, the
data is streamed into a program called LunarScan 1.5
written by Peter Gural, which takes 1 GB chunks of
data and looks for impact candidates. The video
stream is broken up into these chunks using a program called Virtual Dub. These chunks are then
scanned frame by frame to look for meteoroid impact
point flashes. Once a set of candidate impacts are
found by the software, it is up to the user to weed out
transients such as video noise or cosmic ray hits.
After a list of likely candidates, including date,
time, position, estimated magnitude, etc. is generated,
this information is reported to NASA-MSFC and
ALPO-LMIS. The observing team at NASA-MSFC
may have been observing at the same time and can
provide confirming observations of one or more of
the impact candidates; the ALPO-LMIS serves to
coordinate the former along with interested individuals and groups, they may be able to contact someone
else observing at the same time to obtain a confirmation of one or more of the events observed.
3.5 Lessons Learned

We have come a long way as far as lunar meteor
monitoring is concerned. The technology has greatly
improved since the early Leonid days of late 1999,
and over twelve years of experience on many fronts
have improved the techniques and approaches to lunar meteor observation. These are quite common
events, but are randomized enough and of extremely
short duration (20 to 100 milliseconds) so as to make
them a significant challenge to observe.
The presence of spurious events such as cosmic
ray hits were quickly found to mimic real lunar meteor impact flashes, but closer inspection of these
transients revealed subtle differences between these
and true impact events. These differences include a
fuzzy versus sharp flash profile, the flash appearing
over multiple frames versus a single frame, etc.
While these differences are not always easily discernible, knowing them did go a long way in helping
to narrow down which events were more likely to be
genuine events and which were not.
The best, most foolproof means of determining
whether an event is a genuine impact event remains
this: to use two independent setups spaced at least 30
km apart observing at the same time to document the
same event, within less than a second difference in
event time and less than 5 difference in position on
the moons disk.
4. Upcoming Project: Observations in

Support of the LADEE Mission
The Lunar Atmosphere and Dust Environment
Explorer (LADEE, Fig. 6) is a NASA spacecraft that
will orbit the Moon and explore the dust and atmosphere environment at the moon. The mission is currently set for launch on a Minotaur-V launch vehicle
in early 2013 July. After a 32 day cruise phase and a
30 day checkout period, the satellite will be ready to
take data by mid-September. Unfortunately this is too
late for a favorable Perseid encounter with the moon,
which was originally going to be the focus of the
collaborative effort. However, sporadic meteor impacts, regularly observed, will become the focus of
observational efforts during the 60 day mission.
The purpose of supplementing the LADEE mission with ground-based observations is to identify
distinct impact events that are expected to kick up
dust and inject it into the near-moon environment,
replicating in some ways the days of Apollo 17 when
sketches made by Eugene Cernan from lunar orbit
showed some unusual optical phenomena minutes to
seconds before orbital sunrise. Most noticeable of
this is a set of crepuscular rays that seem to be caused
by scattering attributed to particles or molecules of
some type. In the airless environment of the moon,
one would not expect to see any such phenomena.
This leads to the question of what is replenishing the
dust or material and how does it remain suspended
above the lunar surface.
33
time at the University, and candidates will be logged,

evaluated, reported, and catalogued.
In addition to the universitys efforts, I will be
contacting existing and new collaborators to encourage them to put forward the effort to monitor the
moon as often as they are able to each month, in support of existing regular campaigns. Ideally I would
like to have a global network, monitoring the moon
continuously (24 hours a day) for five days waxing
and five days waning. But weather and a host of other
influences will likely curtail this, but we will observe
as much as we can.
The monthly 10-day campaign will begin sometime later in 2012 and will continue prior to the
launch of LADEE, and will continue through the
science phase and afterwards. Impacts will be catalogued with me as coordinator of ALPO-LMIS and
with NASA-MSFC, the Meteoroid Environment Office as part of their ongoing program. Sporadic meteoroid impacts happen fairly regularly and are expected to provide stimuli for, as well as an agent to
maintain, the existing dusty environment in the vicinity of the moon. Meteoroid impacts are also thought
to play a role in the replenishment of the lunar sodium exosphere (Bellot Rubio, et al., 2000).
Figure 6. The LADEE spacecraft at the Moon, artists
conception. Image courtesy of NASA.
To answer these questions, Dr. Brian Day of

NASA-Ames has requested ground-based observations in collaborative support of the mission. These
will be observations of lunar meteoroid impacts,
made by interested amateur astronomers and participants at universities. The idea would be to characterize the source of the dust aloft by correlating increased fluxes of dust with impact events. If enough
of these correlative events are documented, this effort
will go a long way in answering the questions about
what the Apollo astronauts (and unmanned Surveyor
spacecraft) saw from the moon: the crepuscular rays
and the horizon brightening just prior to sunrise.
In preparation for this campaign, I will be upgrading the astronomical aspect of the department of
Chemistry and Astronomy at Prairie View A&M
University. The plan is to employ our physics majors
to make the routine observations of the moon to
match what NASA-MSFC is doing: for 10 days per
month, monitor the moon (between 10% and 55%
illumination) with a video camera and telescope system in search for meteoritic impacts. I plan to upgrade existing equipment to handle the expected
heavy data load, and I plan to enable a semiautomated system to be run for each observing session. Data will be collected and analyzed in near real
34
5. Conclusion
Professional-Amateur (pro-am) collaboration is
enjoying a sort of golden age as professionals willingly seek out the assistance of amateurs in their research endeavors. The professionals have the training
and the ideas, as well as the funding through grants
and contracts, while the amateurs have the time and
the equipment that could benefit a professionals project. Lunar meteor astronomy is an excellent example
of pro-am collaboration at its finest.
The Lunar Meteoritic Impact Search section of
the Association of Lunar and Planetary Observers
(referred to as ALPO-LMIS in this paper) has been
coordinating the activities of interested amateur and
professional astronomers since 2000 and has resulted
in over 80 impact candidates being documented. A
professional group at NASA, at the Marshall Space
Flight Center (referred to as NASA-MSFC in the
paper) Meteoroid Environment Office, coordinated
by Dr. Bill Cook, has been regularly monitoring the
moon for meteoroid impacts since late 2005 and as of
2011 November, have logged 233 impact candidates,
most of which have been confirmed. Both groups
have been working together for the last six years on
lunar meteor observation and collaboration.
A NASA mission to the moon, to be launched in
2013 July, is the Lunar Atmosphere and Dust Environment Explorer (LADEE), which is expected to
better our understanding of the lunar dust environment and connections between suspended dust and
meteoroid impacts. ALPO-LMIS has been asked to
collaborate with them during the mission timeframe
(late summer to early fall 2013) to observe meteoroid
impacts on the moon in order to help correlate increases in dust abundance with meteoroid impacts.
This partnership will be an excellent one to further
our understanding of the near-moon environment,
including the moons exosphere and changes to it
related to impacts.
Anyone who is interested in contributing to the
LADEE portion of the campaign, or the general lunar
monitoring campaign is encouraged to contact me or
visit www.alpo-astronomy.org for more information.
6. Acknowledgements
I want to thank Robert Buchheim for encouraging me to share with the Society of Astronomical
Sciences the work I am helping to coordinate. I am
thankful for Dr. Premkumar Saganti, my colleague at
Prairie View A&M University for his support and
encouragement. I also thank Brian Day for providing
the opportunity for collaboration between a worthwhile space-based mission and a worthy groundbased observing program. Finally, many thanks go to
Mr. Doug Holland for providing the CMOS detector
and all the help in its operation.
7. References
Beech, M. and Nikolova, S. (1998). Leonid flashersmeteoroid impacts on the Moon. Il Nuovo Cimento,
Note Brevi 21C, 577-581.
Holland D.. et al. (2008). U.S. Patent-7411198.
Holland D., et al. (2009). U.S. Patent-7521682.
Cudnik, B. M., Saganti, P. B., Erickson, G. M.
(2011). Imaging Detectors in Planetary and Space
Science. 42nd Lunar and Planetary Science Conference, 2011 March 711 at The Woodlands, Texas.
No. 1608, p. 1543.
Bellot Rubio, L. R., Ortiz, J. L., Sada, P.V. (2000).
Observation and Interpretation of Meteoroid Impact
Flashes on the Moon. Earth Moon and Planets,
82/83, 575-598.
35
36
Asteroid Shape Modeling - Clark
First Attempts at Asteroid Shape Modeling

Maurice Clark
Physics Department, Texas Tech University
5200 Broadway Lubbock TX 79411
maurice.clark@ttu.edu
Abstract
Results are presented from initial attempts at asteroid shape-modeling. Three asteroids were chosen in this
study: 1708 Polit, 2036 Sheragul and 3015 Candy. Observations used in this study were made by the author, at
a variety of locations, most recently at the Preston Gott Observatory at Texas Tech University.
1. Introduction
2. The Observatories
Having worked on asteroid photometry for some

10 years and built up an archive of more than 100
well-characterized lightcurves, I decided early in
2011 to start re-observing some asteroids with the
intention of using the data to model their 3dimensional shape.
Three asteroids were chosen for this first attempt: 1708 Polit, 2036 Sheragul, and 3015 Candy.
All three are main-belt asteroids for which I had very
good lightcurve data. 1708 Polit, (Clark, 2011) and
3015 Candy (Clark, 2007) had been observed in 2005
while 2036 Sheragul had been observed in 2003 and
2010 (Clark 2004; Clark, 2011).
Period analysis was also done in MPO Canopus,
which implements the algorithm developed by Alan
Harris (Harris et al., 1989). Differential magnitudes
were calculated using reference stars from the
USNO-A 2.0 catalog and the UCAC2 catalog. Shape
models were derived from the lightcurve data using
MPO LCInvert.
Rosemary Hill Observatory (MPC code 831) is

owned by the University of Florida, Gainesville, and
is located near Bronson, Florida, at an elevation of 51
m. The main instrument at Rosemary Hill is a 0.76-m
Tinsley reflector. For these observations, the instrument was used at its f/5 Newtonian focus with an
SBIG ST-7E CCD camera.
The Preston Gott Observatory is the main astronomical facility of the Texas Tech University. Located about 20 km north of Lubbock, the main instrument is a 0.5-m f/6.8 Dall-Kirkam Cassegrain. An
SBIG STL-1001E CCD was used with this telescope.
Bucknell Observatory is located on the campus
of Bucknell University in Lewisburg, Pennsylvania.
The elevation is 171 m. The telescope used for these
observations was a 0.35-m Celestron working at f/7
with a focal reducer. The CCD camera was an SBIG
ST-9E.
Year
Date Range
Solar Long
Obs
Sess
Err
Amp
Err
2005
Jan 05 Mar 06
133-164
Rosemary Hill
Per (h)
7.507
0.001
0.4
0.1
2011
May 06 Jun 22
224-240
Preston Gott
7.5085
0.0008
0.35
0.1
Per (h)
Table I. Observational data for 1708 Polit.

Year
Date Range
Solar Long
Obs
Sess
2003
Jul 29 Aug 04
232-238
Bucknell
5.41
Amp
Err
0.01
2010
Oct 31 Nov 28
24-36
Preston Gott
2012
Jan 18 Jan 29
54-67
Preston Gott
Err
0.58
0.02
5.413
0.001
0.95
0.02
5.4120
0.0002
0.60
0.02
Table II. Observational data for 2036 Sheragul.

Year
Date Range
Solar Long
Obs
Sess
Per (h)
Err
Amp
Err
2005
Dec 22 Dec 29
45-52
Rosemary Hill
4.6248
0.0003
1.05
0.05
2011
May 06 Jun 22
209-225
Preston Gott
4.6249
0.0002
0.85
0.05
Table III. Observational data for 3015 Candy.
37
3. Observations and Data

Tables I-III summarize the observations used in
this study. Included in the tables are the dates over
which observations were obtained, the heliocentric
longitude of the asteroid, the observatory at which the
observations were obtained, the number of observing
sessions in which useful data was obtained, the derived period and its estimated uncertainty, and finally
the amplitude of the lightcurve and its estimated uncertainty. The resulting lightcurves are shown in Figures 1-7.
Figure 3. 2005 lightcurve for 3015 Candy
Figure 1. 2005 lightcurve for 1708 Polit
Figure 4. 2011 lightcurve for 3015 Candy
Figure 2. 2011 lightcurve for 1708 Polit
Figure 5. 2010 lightcurve for 2036 Sherugul
38
4.1 General Procedure
4. Analysis
It should be noted that, because of the nature of
the lightcurve inversion process, solutions for asteroids that have low-inclination orbits (in reference to
the ecliptic plane) usually come in pairs where the
latitude (b) of the pole, referenced to the ecliptic
plane, is nearly the same but the longitudes () of
differ by about 180.
Another possibility is for two solutions where
the longitudes differ by about 180 but the latitudes
are somewhat mirrored, i.e., one is positive indicating prograde rotation while the other is negative
indicating retrograde rotation. In this case, the absolute values of the latitudes are not always similar. In
some cases, four solutions are found, combining the
two possibilities given above.
For each asteroid, the lightcurve data outlined in

Tables I-III were used in LCInvert to find a probable
synodic period. A sidereal period was then derived,
again using LCInvert. This period was applied in a
pole search that generated 310 solutions using discrete, fixed longitude-latitude (, ) pairs but allowing the sidereal period to float.
This produces a pole search plot where the
log(chi-square) value found in the search for each
discrete pair is color-coded to help look for possible
solutions. In the plots, deep blue represents the lowest chi-square value and deep red (maroon) represents
the highest value. All other values are scaled between
dark blue and bright red.
An ideal solution has only one small region of
deep blue, a small region of bright red, with the rest
of the plot a mix of colors from aqua to yellow. There
may be two regions of blue within the sea of midcolors, indicating the two possible pole solutions discussed above.
If the plot is awash in blues, then the solution is,
essentially, meaningless since this means that almost
any solution was as good as any other.
Once one or more probable solutions were
found, a new search was made using one of the solutions as the starting point, this time allowing the longitude and latitude to float as well as the period. For
the final results, the error for the poles is approximately 10 while the period solutions have errors on
the order of 2-3 units in the last decimal place.
4.2 1708 Polit

The pole search plot for 1708 Polit is show in
Figure 8. This shows a general solution near
(30, 10) and a second one near (210, +40).
Figure 8. Plot of the log (chi-square) values for 1708

Polit. Dark blue represents the lowest chi-square value.
The final result showed two possible solutions:

(27, 33, 7.50837340 h) and (210, +47,
7.50832256 h) The models for these are shown in
39
Figures 9 and 10. The preferred solution is the one at

(210, +47), or a prograde rotation. Supporting this
preference is the derived axis of rotation. As visible
in Figures 9 and 10, the spin axis is much more
closely aligned with the shortest axis of the asteroid
with the (210, +47) solution, a much more likely
scenario. Also supporting the preferred solution is
that the models lightcurve more closely matches the
observed lightcurve, particularly for the 2011 observations. Examples of this are shown in Figures 11-14.
Figure 12. Correspondence between model (black) and
observed (red) lightcurve of 1708 Polit for solution
(27, 33) for 2011 May 6.
Figure 9. 3-D model of

(27,33, 7.50837340 h).
1708 Polit for
solution

(210, +47) for 2005 January 9.
Figure 10. 3-D model of 1708 Polit for solution

(210, +47, 7.50832256 h).

(210, +47) for 2011 May 6.
4.3 2036 Sheragul

observed (red) lightcurves of 1708 Polit for solution
(27, 33) for 2005 January 9.
40
The pole search plot for 2036 Sehragule is

shown in Figure 15. This indicates one general solution near (130, +50). There is a possible second
near (310, 30), but it is not very obvious.
The final result showed two possible solutions:
(134, +38, 5.41283557 h) and (314, 46,
5.41282858 h) The models for these are shown in

Figures 16 and 17.
The solution at (134, 38) is slightly preferred
since the model fits the observed lightcurves slightly
more closely. However the difference is minimal.
Examples of the model fit with observed lightcurves
are shown in Figures 18-21. Given the clear ambiguity in the results, more observations at future oppositions are needed.
Figure 18. Correspondence between model and observed lightcurve of 2036 Sheragul for solution
(134, +38, 5.41283557 h) for 2010 October 31.

Sehragul. Dark blue represents the lowest chi-square
value.
(314, 46, 5.41282858 h) for 2010 October 31.
Figure 16. 3-D model of 2036 Sheragul for solution

(134, +38, 5.41283557 h).
(134, +38, 5.41283557 h) for 2012 January 18.
Figure 17. 3-D model of 2036 Sheragul for solution

(314, 46, 5.41282858 h).
41
(314, 46, 5.41282858 h) for 2012 January 18.
Figure 23. 3-D model of 3015 Candy for solution

(103, +17, 4.62500318 h).
4.4 3015 Candy

The pole search plot for 3015 Candy is shown in
Figure 22. This shows one general solution near
(110, +15). However there is no strong solution as
shown by the large area of blue. This indicates that
the pole direction and therefore the shape are poorly
constrained.
Figure 24. 3-D model of 3015 Candy for solution

(277, 79, 4.62498205 h).

Candy. Dark blue represents the lowest chi-square
value.
Two possible solutions were found in the secondary search: (103, +17, 4.62500318 h) and
(277, 79, 4.62498205 h).
Examining the resulting models (Figures 23 and
24), there is little to choose between them. The
(103, +17) solution slightly more closely matches
the observed lightcurve. However, the difference
appears minimal. Examples of this correspondence
are shown in Figures 25 and 26.
This ambiguity is not surprising given how
poorly the pole direction was constrained. As a result,
the models presented here have a low confidence.
Probably the main cause of this uncertainty is that the
observations of the asteroid were made 180 apart in
heliocentric longitude. Further observations at future
oppositions are required to determine which, if either,
of these models reflect reality.
42
Figure 25. Correspondence between model and observed lightcurve of 3015 Candy for solution
(103, +17, 4.62500318 h) for 2011 November 6.
Figure 26. Correspondence between model and observed lightcurve of 3015 Candy for solution
(277, 79, 4.62498205 h) for 2011 November 6
5. Conclusions
Converting lightcurves into 3-D models of these
asteroids has been an interesting experience, despite
the steep learning curve. It is certainly a worthwhile
endeavor and something I would encourage other
asteroid photometrists to become involved in. Of the
three asteroids studied in this paper, the results for
1708 Polit have the highest confidence, while those
of 3015 Candy are very poorly constrained. Further
observations at other heliocentric longitudes are
planned to further refine the models presented here.
6. Acknowledgements
I would like to thank Mikko Kaasalainen and
Brian Warner for answering all my email questions
regarding lightcurve inversion and Brian Warner for
all of his work with the program MPO LCInvert.
.
7. References
Clark, M. L. (2004). Rosemary Hill Observatory
Photometry of Asteroids 2036 Sheragul And (21652)
1999 OQ2. Minor Planet Bulletin 31, 15-16.
Clark, M. L. (2007). Lightcurve Results for 1318
Nerina, 2222 Lermontov, 3015 Candy, 3089 Oujianquan, 3155 Lee, 6410 Fujiwara, 6500 Kodaira,
(8290) 1992 NP, 9566 Rykhlova, (42923) 1999
SR18, and 2001 FY. Minor Planet Bulletin 34, 1922.
Clark, M. L. (2011). Asteroid Lightcurves from The
Preston Gott And Mcdonald Observatories. Minor
Planet Bulletin 38, 187-189.
43
44
Asteroid Distances from Diurnal Parallax Alvarez et al.
Diurnal Parallax Determinations of Asteroid Distances

Using Only Backyard Observations from a Single Station
Eduardo Manuel Alvarez
Observatorio Los Algarrobos, Salto, Uruguay
Costanera Sur 559, Salto 50.000, URUGUAY
olasu@adinet.com.uy
Robert K. Buchheim
Altimira Observatory
Coto de Caza, CA 92679, USA
Bob@RKBuchheim.org
Abstract
A method of using diurnal parallax for determining the distance to asteroids has been developed, which provides
excellent accuracy using an amateur-level telescope and CCD, and a surprisingly simple set of observations.
Data from two consecutive nights is sufficient, obtained at the beginning of each night, and at each culmination.
Just those few data points proved to be enough to allow computing accurately (better than 5%) the distance to
the asteroid. Results for several asteroids will be shown.
1. Introduction
The direct comparison of two observations of a
relatively near object obtained at exactly the same
time from two different stations, where it shows some
angular displacement (parallax) with respect to the
distant background stars, is a well-known exercise in
positional astronomy (astrometry). In fact, as early as
1672 Cassini was able to demonstrate its full validity
by achieving quite good results for Mars parallax
although the extensive travel, and years of preparation and data reduction that this required are testimony to just how difficult a proposition it was!
The advent of photography made this parallax issue far more practical, and the use of CCD cameras
made it far more accurate.
Twenty years ago Ratcliff and colleagues reported that they had successfully made parallax observations of two asteroids by this method (Ratcliff et
al., 1993). Those pioneering measurements were accomplished after a careful and long preparation, involving the joint and synchronized work of two research observatories. Several years later the measurement of asteroid parallax by means of small telescopes was addressed in detail for Near Earth Objects
(NEO) (Smallwood et al., 2004), who exclusively
referred to simultaneous measurements obtained from
two stations.
In contrast, the method of diurnal parallax requires only a single observing site, and uses a baseline that is provided by the rotation of the Earth. This
is also a historically significant and a simpler method
to use (compared to the two-station parallax). It

seems to have been first successfully applied to the
solar system by Flamsteed in 1672, to measure the
parallax of Mars (Forbes, 1976) and later by Lindsay
and Gill, when they measured the parallax of Juno in
1874 (Lindsay and Gill, 1877).
The possibility that amateurs could use the diurnal parallax effect for determining Solar System distances from CCD images obtained from just one single observing point has been recently discussed in
practical terms by Vanderbei and Belikov (2007) and
Buchheim (2011). However, both approaches present
drawbacks the first one requiring precise orbital
data which is not obtained by the observer himself;
the latter only dealing with a particular case.
The question that we posed was whether a general self-contained practical method for determining
asteroid distances via parallax by means of CCD images obtained from just one vantage point could be
possible. By self-contained we emphasize the condition that all required data for the distance computation must be obtained from direct observations not
from third party data. (If one requires extra technical
information for determining an objects distance, the
used method seems to us to be a bit handicapped.)
The idea was to develop a method that could accurately determine a Solar System objects distance
from data gathered by just one observer. The good
news is that we confidently think we have succeeded.
45
2. The Diurnal Parallax Effect

Parallax is a widely-understood and easily demonstrated phenomenon, as well as its derived practical benefit to allow determining the distance to an
object without directly measuring it. In astronomy,
this fits like a glove. It has been used to determine the
size of the Earth (Eratosthenes), the scale of the Solar
System (during Venus transits) and the distance to
near-by stars (Hipparcos satellite).
In theory, once the parallactic shift of the apparent position of a distant object as observed from the
ends of a baseline of known length is measured, then
its distance can be readily determined. In practice,
taking into account that in most astronomical applications the parallax angle is very small, the only way to
improve the accuracy of the distance determination is
by means of enlarging the baseline.
As already mentioned, parallactic measures simultaneously performed from two conveniently far
away observatories have successfully yielded asteroid
distances. The big question here is: can one single
observer determine accurate asteroid distances by
taking advantage of his relative displacement to the
target due to Earths rotation?
Placed on the Earths surface, any observer is actually moving along a perfectly circular track, several
thousand kilometers in radius, making one complete
circle in almost 24 hours (sidereal period). At any
given time, the observer sees the asteroid from a
slightly different perspective compared to a fictitious
observer placed at the center of the Earth. Those are
the vantage points that play this particular parallax
game. This variable parallax effect has nothing to do
with the fact that the asteroid is also moving with
respect to the Earth it only refers to the different
(and constantly changing) spatial position of the observer relative to the center of the Earth due to the
Earths rotation.
Assuming the asteroids distance is fixed with
respect to the Earths center except for NEO asteroids, a general pertinent assumption for time periods
as short as half day to a couple of days the time
variability of the parallax angle due to observer location on the rotating Earth can be exclusively attributed to his changing position relative to a fixed direction in the equatorial plane. In other words, the parallax angle relative to the plane that contains the axis of
rotation never changes.
This is the basis of the diurnal parallax definition
also called the east-west parallax. As shown in Figure 1, the diurnal parallax effect is the variable parallax due to the observers daily rotation with respect to
the Earths center. Its clear that at the moment of the
targets culmination (when the target happens to be
46
on the observers Meridian) the diurnal parallax angle

becomes null, while achieving its maximum value 6
hours either before or after transit.
target
R=
distance
to target
baseline
t
Observer
Figure 1 : Geometry of the diurnal parallax effect. As the

Earth rotates, any observer on its surface sees the targets parallactic angle constantly varying, from a
maximum value occurring whenever his Meridian is at
right angle to the plane containing the Earths rotation
axis and the target (some 6 hours prior or after transit
time) to a null value whenever the target is placed on his
Meridian (target culmination).
The diurnal parallax becomes the angle that at

any given moment has the target on its vertex, the
observer on one side and the center of the Earth on
the other side, projected to the equatorial plane.
Therefore, by definition, at any given time t the diurnal parallax angle becomes
(t ) = [ RAtopo (t ) geo (t )] cos (t )
(1)
where RAtopo is the objects topocentric (measured

from the observer location) right ascension, geo is the
objects geocentric (measured from the center of the

Earth) right ascension, and is the objects declination (assumed to be practically the same as measured
either from the observers place or from the center of
the Earth).
Once this angle is determined, and knowing the
length of the corresponding projected observational
baseline B, the distance to the asteroid R can be readily obtained by making use of the small-angle approximation:
B (t )
(t )
(2)
where B and R are expressed in the same units and

the angle is in radians. From Figure 1 the projected
baseline is
B (t ) = R E cos cos(t )
(t ) = geo (t ) 0
where geo(t) is the geocentric right ascension of the

asteroid at time t, and 0 is the geocentric right ascension at the moment of the asteroid culmination. R, ,
and are functions of time.
Considering that the observer sees the asteroid
not from the Earths center but from a different vantage point (as shown in Figure 2), necessarily the
asteroid coordinates that he measures at any given
time are (slightly) different from corresponding geocentric coordinates.
(3)
target
where RE is the radius of the Earth, is the latitude of

the observer, and is the angular sidereal rotation
rate of the Earth. Therefore, all that matters now is
how to accurately determine the elusive diurnal parallax angle .
3. The Model
The geometry of our model is shown in Figure 2.

The non-rotating coordinate frame is attached to the
center of the Earth (geocentric frame). It is oriented
so that the +Z-axis points toward celestial north, and
the +X-axis points toward the position where the asteroid is at time t = 0. Because of the symmetry of the
diurnal parallax effect we strategically select the beginning of the time parameter to coincide with the
moment of the asteroid culmination. Hence,
(4)
where T is the clock time (this might be civil time

or UT, depending on the preference of the observer)
and T0 is the exact time of the asteroid culmination
(in the same timing reference).
At any time t, the geocentric position of the asteroid (which does move on the sky) is given by the
vector R(t):
x R (t ) R cos cos
R(t ) = y R (t ) = R cos sin
z R (t )
R sin
(5)
observer
t = T T0
(6)
Z= Rsin
R=
where R is the geocentric distance of the asteroid, is

the geocentric declination of the asteroid, and is
the angular difference in the geocentric right ascension of the asteroid measured from t = 0, thus becoming
Rcos
Figure 2: The spatially fixed geocentric frame. Its centered at the Earth center, the Z-axis coinciding with the
Earth rotational axis and the X-axis pointing towards the
direction that makes the plane XZ to also contain the
target at the moment of its culmination.
At any time t, the geocentric position of the observer as he is carried around by the Earths rotation
is given by the vector r(t) (not shown in Figure 2).
Assuming a spherical Earth, the vector r(t) is:
x r (t ) RE cos cos(t )
r (t ) = y r (t ) = RE cos sin(t )
z r
RE sin
(7)
where RE is the radius of the Earth, is the latitude of

the observer, and is the angular sidereal rotation
rate of the Earth.
The vector D(t) from the observer to the asteroid
thus becomes:
47
D (t ) = R (t ) r (t )
RA(t ) = RAtopo (t ) RA0
(8)
so that
x D (t ) R cos cos RE cos cos(t )
D(t ) = y D (t ) = R cos sin RE cos sin(t )

z D (t )
R sin RE sin
(9)
target
(11)
where RAtopo(t) is the topocentric right ascension of

the asteroid at time t, and RA0 is the topocentric right
ascension at the moment of the asteroid culmination.
D, Dec, and RA are functions of time.
From (10), dividing the yD by the xD coordinates
of the vector D(t) and substituting the corresponding
values from (6) it becomes
yD
X cos sin sin(t )
= tan( RA) =
xD
X cos cos cos(t )
(12)
where
observer
Dec
X =
RA
z
(13)
plane containing observers path,

parallel to Equator
y
x
Figure 3: The moving topocentric frame. Its centered at

the observers location on the Earths surface, but at
any time its three axes are parallel to corresponding
axes of the geocentric frame. Hence, the XY plane is
always parallel to the Equatorial plane, and the XZ plane
happens to coincide with the Meridian at the moment of
the targets culmination.
Figure 3 shows a non-rotating coordinate frame

centered on the observers location on the surface of
the Earth (topocentric frame). It is oriented so that
both the +Z-axis and the +X-axis are parallel to corresponding axes in the geocentric frame. Hence, at
time t = 0 the +X-axis also points toward the position
where the asteroid is, which happens to be placed at
the Meridian.
At any time t, the topocentric position of the asteroid is given by the vector D(t):
x D (t ) D cos( Dec) cos(RA)
D(t ) = y D (t ) = D cos( Dec) sin(RA)

D sin( Dec)
z D (t )
(10)
where D is the topocentric distance to the asteroid,

Dec is the topocentric declination of the asteroid, and
RA is the angular difference in the topocentric right
ascension of the asteroid measured from t = 0, thus
becoming
48
R
R E cos
Besides the goal distance R, Equation 12 still includes other unknown time-dependent parameters (
and ) basically related to the actual asteroid motion on the sky. In order to simplify the asteroid distance computation, now we are going to make several
important assumptions.
First of all, we assume that the asteroids geocentric distance R is constant over the interval of a few
days during which we make our measurements. Of
course this is not true, but it is a pretty good approximation the geocentric distance to the asteroid
does constantly change but slowly, so that over an
interval of a couple of nights we can pretend that it is
constant.
The next key assumptions relate to the asteroids
orbital motion. They are:
The geocentric rates in right ascension and declination will be assumed to be constant (linear motion) over an interval of a couple of days.
We will estimate these orbital rates by using
topocentric measurements at the moment of the
asteroid culmination (when diurnal parallax is
null).
We will ignore any differences in the asteroid
topocentric and geocentric declinations at the
moment of transit (in practice, the real difference
is usually insignificant).
From these three assumptions, at least in first
approximation the asteroid orbital motion (geocentric
coordinates) can be determined at any time from
topocentric measurements as:
geo (t ) 0 + t = RA0 + t
(14)
geo (t ) 0 + t Dec 0 + t
(15)
where the asteroid right ascension constant rate is

defined as
RA0 2 RA01
T0 2 T01
(16)
and the asteroid declination constant rate is defined

as
Dec 0 2 Dec 01
T0 2 T01
(17)
where RA01 and RA02 are the asteroid topocentric

right ascensions respectively at the moment of the
first (T01) and second (T02) transit times, while Dec01
and Dec02 are the asteroid topocentric declinations
respectively also at the moment of the first (T01) and
second (T02) transit times.
Thus substituting in Equation 12 it becomes
tan( RA) =
X cos( Dec 0 + t ) sin( RA0 + t ) sin(t )

X cos( Dec 0 + t ) cos( RA0 + t ) cos(t )
(18)
This is the fundamental equation of our model. Once
the key parameters T0 and RA0 become known, the
left term is directly measurable by the observer at any
given time, while in the right term except for the
geocentric asteroid distance R all of the remaining
parameters are either known a priori (RE, , ) or are
also directly measurable by the observer (Dec0, , ).
In consequence, the asteroid distance can be determined by applying an iterative process to a set of
observations as follows:
1.
From observed topocentric coordinates the asteroid linear orbital motion is determined, and
hence the asteroid geocentric coordinates can
be estimated at any given time.
2.
Knowing the asteroid orbital motion and assuming a certain (guessed) asteroid distance,
the model predicts for any given time what the
topocentric right ascension has to be.
3.
The distance value that best minimizes all RA

residuals (observed minus model predicted
topocentric right ascensions) for the given set

of observations becomes the asteroid distance.
4. Determination of the Key Parameters

T0, RA0 and Dec0
The asteroids exact local transit time T0 has to
be known as precisely as possible. An error of just
one second does make difference. This fact imposes a
hard restriction on the timing accuracy and therefore
the easy shortcut of getting T0 from published asteroid ephemerides is not an option.
Fortunately, T0 can be easily determined by the
observer at the required accuracy. The way to do this
is by taking a set of images while the asteroid is
around transit time (preferably from a couple of minutes before up to a couple of minutes after transit
time). For each image, the corresponding time has to
be accurately known and target astrometry has to be
precisely measured. Then, the difference between
measured RA of the asteroid to Local Sidereal Time
can be determined for each observation time of the
set of images. Finally, by using the Intercept function
on a spreadsheet the time of transit T0 and corresponding RA0 and Dec0 can be precisely obtained.
Having determined from observations made on
two proximate nights (the adjacent, the better) corresponding parameters T0, RA0 and Dec0, then the asteroids geocentric orbital motion in right ascension
and declination can be conveniently represented, at
least in first approximation, by linear rates and
just derived from those six parameters.
5. The RA Geocentric Linear

Assumption
The quality of the diurnal parallax angle determination is based on two main subjects: (a) the accuracy of RA topocentric values actually measured, and
(b) the accuracy of RA geocentric values estimated
for the required parallax computation.
As already discussed, our (self-contained) model
uses a linear approximation for the unknown RA
geocentric values, so that asteroid distances can be
truly determined from just the observers data. However, replacing a curve by a linear approximation
always introduces some error.
In case the time frame we consider is just two
consecutive nights and the target asteroid is near opposition, then the asteroid does have an almost linear
motion against the background stars and our linear
assumption is totally pertinent even though its also
the time of greatest motion in right ascension. In a
more general situation, or for a longer time frame
(lets say 3-4 days) the real asteroid trajectory in-
49
creasingly departs from a linear motion. Could the

error derived from the linear simplification still be
manageable in a more general context?
The problem to solve is to somehow minimize
the error derived by the replacement of the unknown
general RA geocentric motion curve by a linear approximation. The linear approximation is the uniform
rate that the observer is able to derive from the only
two points (T01, RA01 and T02, RA02) of the real RA
geocentric curve that he can actually measure on just
two proximate nights.
Of course, if the observer could be able to observe along several nights he would collect several
pairs of (T0 , RA0) data points, which in turn makes it
possible to derive a more accurate variable (timedependent) rate for the real RA geocentric motion
instead of a simple uniform rate, finding out a first,
second or even third order analytical expression for
(t) . The same goes in case the observer could know
such data from accurate target ephemerides. We are
dealing here with the most likely and easiest case
where the observer dedicates just two (partial) nights
to find out the distance to a particular asteroid exclusively by means of his own data.
The graph in Figure 4 poses the question. If the
real RA geocentric curve were to be replaced by just
a linear approximation, then along one single observing session any new RA geocentric values would
be either underestimated or overestimated by the
simplification (the larger the time apart from the transit time T0, the worse). Also for the same observing
session, all RA geocentric values prior to the transit
time would be replaced in the opposite way as those
RA geocentric values after the transit time (if the
former became underestimated, the latter resulted
overestimated, or vice versa).
It becomes clear that working with a linear RA
geocentric rate and using data from just one observing session obtained entirely from one side of the
Meridian (either only prior or only after transit time)
will achieve the largest error. To minimize diurnal
parallax errors derived from the use of data from only
one observing session it is necessary to observe the
target for a long time both sides of the Meridian.
However, we found out an elegant simplification. Instead of observing the asteroid for a full night
from dusk to dawn, and also at transit time on the
following day, as shown in Figure 5 basically the
same exactitude can be achieved if the target is observed on the two nights just from the same side of
the Meridian.
50
Figure 4: The problem with the linear assumption. Replacing a real curve by a linear approximation introduces errors in a particular direction previous to the
point where the segment actually coincides with the
curve, and in the opposite direction after such coinciding point.
Figure 5: Replacing the real RA geocentric curve by a

linear approximation makes the new RA values at any
time previous to transit time (T0) to be wrong in one direction for the first session (underestimated as shown in
the red area for Session #1) but wrong in the opposite
direction for the following session (overestimated as
shown in the green area for Session #2).
Therefore, despite the fact that the RA geocentric

values for each session necessarily are going to be
inexact (the larger the curvature, the greater the error), if the observer only collects data either before or
after the transit time, but not simultaneously from
both sides, then the error introduced by the linear RA
geocentric simplification tends to cancel out.
Using data from just one side of the Meridian not
only means the observer does not need to work all
night long (not a minor issue for amateurs committed
to full-day jobs) but also implies that the asteroid no
longer has to be near opposition (it could transit at
any time during the night). Of course far from opposition the asteroid will be fainter and may follow a
much more complicated trajectory thus most likely
decreasing the properness of the linear rate approximation but on the other hand (a) the asteroid could
be observed for a longer time frame than the 4-5
hours from dusk to midnight, or from midnight to
dawn, and consequently (b) the asteroid could also be
observed for longer periods while comparatively at

higher altitudes on the sky.
trend for errors to cancel out if working with observations from just one side of the Meridian.
We have applied the model to five different asteroids, trying to cover a wide distance spectrum.
One close and fast moving NEO asteroid was observed having a large diurnal parallax angle (greater
than 120 arcsec), three asteroids were observed at
intermediate distances, and one last asteroid was observed at such a challenging far distance that its diurnal parallax angle (less than 3 arcsec) was as small as
the local atmospheric blurring effect (seeing).
All observations were performed from Observatorio Los Algarrobos, Salto, Uruguay (MPC I38) at
latitude -31 23 33, using a 0.3-m telescope and
CCD camera yielding an image scale of 1.9
arcsec/pixel. MPO Canopus was the software used
for astrometry.
All observed asteroids happened to be relatively
close to their opposition date. We observed our five
targets at least on one pair of consecutive nights, accumulating 10 pairs of such sessions. Except for one,
all individual observing sessions lasted more than 4
hours. For each target we obtained a series of images
taken about 15 minutes apart, except at each opening
session and close to target culminations when several images were taken continuously in order to improve derived astrometry accuracy. On some occasions we not only imaged the target for many hours
prior to its culmination, but also for many hours after
having past the Meridian as well.
6.1 Asteroid 8106 Carpino

This main belt asteroid was our first target, observed on two consecutive nights (2011, April 6 and
7) while exactly at opposition (April 7th) when it
appeared with a visual magnitude of 15.5 and a phase
angle of 9.9 degrees. Almost all observations of both
sessions were made prior to transit time. Figure 6
shows the measured parallax angles assuming a linear
RA geocentric motion derived from data collected on
those nights.
Technical data for each session and corresponding values are summarized in Table I.
The distance derived from each session data was
excellent (better than 4%). As expected, one value
was underestimated while the other was overestimated. The found distance derived from the two sessions was remarkably good (2.5%). This proves the
full validity of the RA constant rate assumption for
the asteroid motion at opposition and the predicted
Parallax angle
(arc-sec)
6. The Model At Test
Elapsed time, hr
Figure 6: Observed 8106 Carpinos diurnal parallax by

assuming asteroid RA linear motion, from data obtained
on two consecutive nights (2011 April 6 and 7), which
happened to coincide with its opposition date.
8106 Carpino
Observation date (UT)
Lapse to/from opposit (day)
True Distance (AU)
Phase Angle (deg)
Visual magnitude (mag)
Observed lapse before T0 (h)
Observed lapse after T0 (h)
Used RA rate
Found Parallax (arcsec)
Found Distance (AU)
Relative error (%)
Session #1
Session #2
2011-04-06
-1
0.974
9.94
15.53
4.0
0.7
2011-04-07
0
0.972
9.91
15.53
4.8
0.7
Sess. 1&2
0.973
Constant: = -0.010151 arcsec/sec

7.88
7.41
7.52
0.951
1.011
0.997
-2.4
+4.0
+2.5
Table I: 8106 Carpino Results
6.2 Asteroid 819 Barnardiana

This main belt asteroid was observed on three
pairs of consecutive nights (2011, May 29-30, June
1-2, and June 11-12), that is, increasingly apart from
opposition (May 19th) when it appeared with a visual
magnitude of 13.6 and a phase angle of 5.2 degrees.
We tried on this target several different and mixed
combinations of prior/after transit observations. Figures 7-9 present the measured parallax angles for
each pair of consecutive nights assuming a linear RA
geocentric motion derived from data collected on
corresponding nights.
Figure 7: Observed 819 Barnardianas diurnal parallax

by assuming asteroid RA linear motion, from data obtained on two consecutive nights (2011 May 29 and 30),
about ten days past opposition.
51
Table II: Conditions and results for 810 Barnardiana

by assuming asteroid RA linear motion, from data obtained on two consecutive nights (2011 June 1 and 2),
some 14 days past opposition.
toriously in Figure 9. Despite the diurnal parallax

effect being a totally symmetric phenomenon, the
uniform RA rate assumption introduces errors that
make the angle to appear increased at one side of the
Meridian and diminished at the other, so that the
shape distortion becomes worse as the asteroid trajectory increases its curvature, i.e., as the asteroid gets
farther from opposition.
Having observed this target on six sessions distributed half-month apart, from our own data we were
able to find out a much better representation of the
actual RA geocentric motion of the asteroid. Figures
10-12 show the same collected data as before, but
this time a variable (third order) rate has been used
for the RA geocentric trajectory. One glance at Figures 9 and 12, which are based on the same data, is
enough for understanding the matter.

by assuming asteroid RA linear motion, from data obtained on two consecutive nights (2011 June 11 and 12),
It seems quite evident that for each pair of obtained parallactic curves, the shape of the first compared to the second become increasingly deformed.
This is due to the linear RA geocentric assumption,
which gradually loses validity as the asteroid moves
away from opposition. This is made manifest by observing how the linear rate changed from an almostsimilar value for the first pair of nights compared to
the last value respectively, 0.0105, 0.0098 and
0.0065 arcsec/sec, that is, the linear rate for the last
pair was only 60% of the first one, thus approaching
the highly curved part of its orbital trajectory when
the asteroid switches between prograde and retrograde motion. For those sessions that covered a large
time on both sides of the Meridian, some deformation
of the diurnal parallax curves becomes evident no-
52

by assuming asteroid RA variable (3rd order) motion,
from data obtained on two consecutive nights (2011 May
29 and 30), about ten days past opposition.

from data obtained on two consecutive nights (2011
June 1 and 2), some 14 days past opposition.

from data obtained on two consecutive nights (2011
June 11 and 12), some 24 days past opposition.
Technical data for each session and corresponding values are summarized in Table II.
Using the linear approximation we obtained mediocre results if considering only data from each
night; however, outcomes from considering the joint
data of each pair of nights were again excellent (better than 3.5%) for the first and second pairs (those
closer to opposition), although poor (17%) for the last
one (24 days past opposition). On the other hand,

using a quadratic approximation for the whole considered period we obtained very good results (better
than 5%) either for the individual sessions as well as
for the pairs, except for the first one (10%). Logically, the variable rate expression rectifies the problem where the trajectory wasnt uniform, and worsens where the motion was actually practically linear.
One last fact worth mentioning is that the two larger
errors from combined nights also occurred for those
pairs (first and third) when one of the sessions covered a large span on one side of the Meridian not
covered by the other.
6.3 The Far Away Asteroid 414 Liriope

This challenging outer main belt asteroid was
observed on three pairs of consecutive nights (2011,
August 30-31, September 1-2, and September 5-6),
that is, increasingly apart from opposition (August
25th) when it appeared with a visual magnitude of
14.5 and a phase angle of 2.5 degrees. We also tried
on this target several different and mixed combinations of prior/after transit observations. Unfortunately, the third session was aborted half hour prior
transit time due to cloud covering. Figures 13, 14 and
15 present the measured parallax angles for each pair
of consecutive nights assuming a linear RA geocentric motion derived from data collected on corresponding nights.
The computation of the exact transit time for the
third session (2011-09-01) couldnt be done as usual
(from data collected around asteroid culmination
from that session) so that it had to be derived from
previous and following observations. The quality of
the obtained data is clearly inferior to those from previous targets, something totally expected given the
small parallactic angles now in play.
Technical data for each session and corresponding values are summarized in Table III.
Table III: Circumstances and results for 414 Liriope
53
Despite the small parallax angle approaching

the frontier where measurement reliability becomes
degraded outcomes from the linear assumption
were still very good (around 5%) for the first pair of
sessions and even better (better than 3.5%) for the
other two pairs.
Figure 15: Observed 414 Liriopes diurnal parallax by
on two consecutive nights (2011 September 6 and 7),
6.4 Asteroid 1660 Wood

on two consecutive nights (2011 August 30 and 31),
This main belt asteroid was observed on four

consecutive nights (2012, January 23-24-25-26) prior
to opposition (January 29th) when it appeared with a
visual magnitude of 13.4 and a phase angle of 24.6
degrees. Sessions were relatively short (all under 5.5
hours, while the 3.7 hour session from 2012-01-24
was the shortest one for this work). Figures 16 and 17
present the measured parallax angles for each pair of
consecutive nights assuming a linear RA geocentric
motion derived from data collected on corresponding
nights.

on two consecutive nights (2011 September 1 and 2),
Figure 16: Observed 1660 Woods diurnal parallax by
on two consecutive nights (2012 January 23 and 24),
some 5 days prior to opposition.
Table IV: Circumstances and Results for 1660 Wood
54
obtained) not previously dealt with. Figure 18 presents the measured parallax angles for the pair of
consecutive nights using the linear RA geocentric
motion assumption, derived from data collected on
these nights.
Figure 17: Observed 1660 Woods diurnal parallax by

on two consecutive nights (2012 January 25 and 26),
some 3 days prior to opposition.
Despite working with short-span data, outcomes

for combined sessions were very good (5.8% and
1.3%). The fact that we got (twice) data from three
nights in a row allowed us to compare results for the
same central night but derived from two different
uniform rates one applying the linear motion from
the adjacent first and central nights, and another from
the also adjacent central and last nights of the trio.
For the 2012-01-25 session we obtained a distance of
0.885 AU (+6.8%) and 0.781 AU (-5.8%) if we applied the different linear rates measured for each pair
of days of the trio (respectively, -0.007404 and 0.007315 arcsec/sec). That is, a mere 1.2% in difference in the assumed uniform RA geocentric rate
caused a 12.6% difference in the derived asteroid
distance, for data just 4 days away from opposition.
6.5 The Close Asteroid (162421) 2000 ET70

This NEO (Athen) asteroid was observed on two
consecutive nights (2012, February 22-23) coinciding
with its opposition (February 22nd) when it appeared
with a visual magnitude of 13.2 and a phase angle of
40.7 degrees. The rapid sky displacement of this
neighbor while visiting us at opposition (-0.2644
arcsec/sec) posed some practical problems (varying
star background context from where astrometry was
Figure 18: Observed (162421) 2000 ET70s diurnal parallax by assuming asteroid RA linear motion, from data
obtained on two consecutive nights (2012 February 22
and 23), which happened to coincide with its opposition
date.
Besides the rapid sky motion, the closeness of

the asteroid also implied a large diurnal parallax effect. Hence, absolute measurement errors comparatively became less important, which neatly explains
the cleanness of the obtained parallactic curves. Considering that except at the beginning and at transit of
each session each data point was obtained at approximately the same intervals, the curves not only
show how the parallactic angle actually changed with
time, but also its variable rate of change.
The asteroid distance obtained from the combined session was quite good (6%), derived from the
two sessions with large individual errors (-10.4 and
Table V: Circumstances and Results for (162421) 2000 ET70
55
+20.0%). One important matter for this close flyingby target is that at opposition, its geocentric distance
was changing by about 4% per 24-h, which obviously
violates the crucial constant distance assumption in
our model.
6.6 Discussion of Results
Table VI: Distance percent errors for each one of the

combined consecutive sessions
Table VI summarizes final asteroid distances derived from observational data. Distances from combined consecutive nights' data were very good, even
for the challenging 2.5 AU far away asteroid. Some
measurements for 819 Barnardiana gave the largest
errors, due to the fact that observations at first and
third pairs of sessions were collected on both sides of
the Meridian. The NEA asteroid gave the worst case,
but the obtained 6.1% error distance was also surprisingly good, given its rapidly-changing distance.
The adopted asteroids geocentric rate in RA is
critical for an accurate determination of its distance
derived from the diurnal parallax effect. Results are
very sensitive to the RA geocentric rate that was actually used. Depending on asteroid distance and orbital position, a tiny percent change in the assumed
RA rate can change the calculated asteroid distance
by far more than ten times such percent.
The linear RA geocentric assumption has been
proven useful and effective. Results derived from the
more accurate variable rate model became marginally superior in the case when 819 Barnardiana was
far away from opposition. On this scenario, the a
single observer can still accurately determine the distance to an asteroid all by his own by observing the
target at transit along a handful of nights, so that he
can obtain a good model of the changing RA geocentric rate.
7. A Much Simpler Stratagem

We also applied our data to Equation 1. Assuming that the distance to a far away target remains constant, and that there is almost no variation in declina-
56
tion for the short considered daily interval, the diurnal parallax angle varies in a sinusoidal way as
(t ) A sin( T t )
(19)
where the amplitude A is the maximum diurnal parallax angle (T) is the angular rate of the sine curve and
the time variable (t) is selected exactly the same way
as before (its beginning coincides with the moment of
the asteroids culmination).
The maximum diurnal parallax angle A corresponds to the parallactic value 6 hours either before
or after transit, and becomes the key value for finding
out the distance. In theory, it could be possible to
directly measure it but it requires the asteroid to be
far way from opposition (which poses the problem of
the variable RA geocentric rate); in practice, the difficulty of accurately measuring such a small angle
(usually less than 10 arcsec) turns the direct attempt
almost impossible. However, it is possible to determine it indirectly.
From equations 1 and 19 we have
[ RAtopo (t ) geo (t )] cos (t ) A sin( T t )

(20)
From our data, at the time of each observation
we know the asteroid topocentric coordinates, as well
as corresponding RA geocentric value. The angular
rate of the sine curve can be readily derived from our
observations at two consecutive transit times, becoming
T =
2
T02 T01
(21)
Therefore, by finding the A value that best

minimizes all residuals in Equation 20 for a given set
of observations (the same process applied in our previous model to Equation 18, this time trying to determine the sinusoidal curve that best fits the left-side
data term in Equation 20) the maximum parallactic
angle is determined and yields the distance to the
object once the baseline (REcos) is applied to Equation 2.
Asteroid distances obtained by using the simplification of Equation 20 applied to exactly the same
observational data already presented were virtually
identical to those obtained through our more rigorous
model. This wasnt a surprise as both models are basically solving the same geometry, but anyway it was
a general validity for Equation 19.
Finally, we tried an even more intrepid bet. Instead of dealing with a large set of observations for
finding the amplitude of the parallactic sine curve, we

tried using only four data points: those from the beginning and around transit time for each of two consecutive nights. This ultimate simplification to an
already simple method proved to be totally pertinent,
as results were only marginally poorer than those
obtained from our (much more complex) model. Just
four data points strategically selected (those at zerolevel parallactic angle and those as large as possible
from the same side of the Meridian) are sufficient for
accurately finding out the amplitude of the diurnal
parallax effect.
In order to improve the quality of the results, in
practice the data-points far away from culmination
should be obtained as a set of several consecutive
images the same procedure as for accurately determining transit times.
Distance percent errors from combined consecutive sessions, by means of our model, the sine curve
simplification, and the 4-Point shortcut (as we have
named it) are shown in Table VII.
The innovative 4-Point shortcut for determining
an asteroid distance is self-contained, extremely simple (only requiring two observations on consecutive
nights), and yields very good accuracy (around 5% or
better) on condition the asteroid is fairly close to opposition. If that is not the case, some more data is
required (observations from a few more nights, but
only done at transit times).
8. Range of Distances
By means of a modest 0.30-m telescope we have
successfully tested the 4-Point shortcut method for
determining asteroid distances up to 2.5 AU which
implied dealing with a parallactic angle about 3
arcsec without any problem. Working with similar
sized, good optics telescopes, in principle we dont
see any limitation preventing the detection of a ten
times lesser parallactic angle if due care, proper image scale, and favorably seeing conditions are all in
play.
From equations 2 and 3, the corresponding distance in AU for a parallactic angle of 0.3 arcsec
equals 29.2 times the cosine of the observers latitude. Therefore, we think the distance determination
of bright enough Solar System objects up to 20-25
AU by means of a backyard telescope applying the 4POINT shortcut is a feasible task although a certainly challenging one. We are planning some experiments to see if we can measure the distance to
Pluto with this method.
Table VII: Comparison of Results from Three Methods
9. Stellar Parallax
The distance to nearby stars can be accurately
determined by measuring their parallax angle while
the Earth orbits around the Sun. The geometry of this
situation is exactly the same as drawn in Figure 1,
where now the circumference represents the orbit of
the Earth. As the Earth orbits around the Sun, any
observer on its surface sees the stars parallactic angle constantly varying, from a maximum value
occurring whenever the Earth is at right angle to the
plane perpendicular to the Ecliptic which also contains both the Sun and the star (some 3 months prior
or after star due opposition) to a null value whenever
the Earth happens to also be placed on such plane
(star opposition).
If the beginning of the time parameter is strategically selected to coincide with the moment of the star
opposition, then our reasoning for the diurnal parallax effect is completely valid for the star parallax
case and therefore all the equations of our model
can be accordingly translated. The former exact time
of the asteroid culmination now becomes the exact
time of the star opposition; the former unknown geocentric motion of the asteroid now becomes the unknown proper motion of the star.
57
Therefore, the 4-Point shortcut for the distance

determination of a nearby star should work all the
same, except this time to collect the four data points
will take 15 months instead of just 30 hours.
Considering that the parallax angle of nearby
stars is within reach of backyard observers (Sirius has
a parallax of about 0.38 arcsec), once again the determination of their distance by means of the 4-Point
shortcut should be a challenging but feasible task
although not especially suitable for the impatient
observers.
10. Conclusions
For asteroids (or other Solar System objects) our
method for determining their diurnal parallax and
hence their distance is both simple and sensitive, only
requiring two measurements on two proximate nights
(the more adjacent, the better) from just one single
observing point. On each night one of the measurements must be done around the time of culmination
and the other as far from culmination as possible, in
order to maximize sensitivity; but on both sets of
measurements should be taken from the same side of
the Meridian.
Achieving a distance accuracy of 5% for an asteroid close to opposition, derived from nothing other
than just backyard observations from a single station
during a short time interval about 30 hours is remarkable and totally feasible. The cherry on the top
is that this goal a kind of amateur fantasy not so
long ago only demands two sets of asteroid images,
each covering not more than ten minutes, taken on
two consecutive nights.
11. References
Buchheim, R. K. (2011) A Modern Incarnation of
Tychos Diurnal Parallax Method. in Proceedings
for 31st Annual Symposium on Telescope Science
(Warner et al., eds.) pp. 109-114. Society for Astronomical Sciences, Rancho Cucamonga, CA.
Forbes, E. G. (1976). Early Astronomical Researches of John Flamsteed. Journal for the History
of Astronomy 7, 124-138.
Lindsay, L., Gill, D. (1877) Note on the results of
heliometer observations of the planet Juno, to determine its diurnal parallax. Monthly Notices of the
Royal Astronomical Society 37, 04/1877, 308.
Ratcliff, S. J., Balonek, T. J., Marschall, L.A., DuPuy, D. L., Pennypacker, C. R., Verma, R., Alexov,
A., Bonney, V. (1993). The measurement of astro-
58
nomical parallaxes with CCD imaging cameras on

small telescopes. American Journal of Physics 61,
208-216.
Smallwood, L. M., Katz, D. M., Richmond, M. W.
(2004). Near Earth Objects: A Brief Review and A
Student Project. American Journal of Physics 72,
264-279.
Vanderbei, R. J., Belikov, R. (2007). Measuring the
Astronomical Unit. Sky and Telescope January
2007, 91-94.
Maximum Amplitudes of Harmonics - Harris
On the Maximum Amplitude of Harmonics

of an Asteroid Lightcurve
Alan Harris
MoreData!
La Caada, CA USA
harrisaw@att.net
Abstract
Most asteroid lightcurves are dominated by the second harmonic of the rotation period, caused by elongated
shape. However, if the shape is not very elongate, other harmonics may dominate, leading to ambiguity of which
is the true rotation period. It is argued from geometry that at low phase angle, harmonics other than the second
with amplitude exceeding ~0.4 are nearly impossible, so lightcurves with larger amplitude than that suggest a
unique period dominated by the second harmonic, unless the spin is complex, non-principle axis rotation. On the
other hand, lightcurves with amplitude less than 0.2-0.3 magnitudes can be dominated by other harmonics, especially the 4th and 6th, so the period may be ambiguous unless odd harmonics can be found to identify the true
rotation period. Examples are presented of each, i.e., low and high amplitude ambiguities.
1. Geometric Rotational Variation Due

to Shape
0.2
3,6
0.4
Magnitude
In the limit of geometric scattering at zero phase

angle and no albedo variegation, the brightness of an
object is just proportional to the projected cross sectional area. Consider various possible shapes as regular polygonal equatorial cross section; either prisms
or cones with the same number of sides yield the
same rotational variation of cross section. Tables I
and II list the maximum range of cross section that
comes from each shape, and the amplitude in magnitudes as such a figure rotates, which is just
4
2
0.6
0.8
1.0
1.2
1.4
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Rotation phase
A = 2.5log(CSmin/CSmax)
(1)
where CSmin is the minimum cross section and CSmax

is the maximum cross section.
Note that concave faces do not increase the
variation in cross section, hence amplitude of variation; convex faces reduce it. Thus the polygonal
shapes considered represent the maximum variation
of cross section as the figure rotates, and hence, subject to the geometric scattering approximation, the
maximum amplitude of the primary harmonics due to
that shape. Figure 1 illustrates the lightcurves that
result from each of the polygonal shapes.
Figure 1. The rotational lightcurve of each of the polygonal figures considered, at zero phase angle and
geometric scattering. A two-sided polygon is just a
flat plate or straight line and has an infinite aspect ratio,
hence an infinite amplitude lightcurve. Practically speaking, we see small asteroids of generally prolate form
with axis ratios up to around 4:1, yielding lightcurves of
1.5 magnitudes or so. Lightcurves dominated by higher
order harmonics (4, 6, 8, 10) have successively lower
maximum amplitudes.
We know from symmetry that there are only

even harmonics in this approximation. Shapes with
odd numbers of sides (3, 5, 7, ) produce primary
harmonics of twice that number, i.e., 6, 10, 14. Figures of even numbers of sides (2, 4, 6, 8, ) produce
primary harmonics equal to the number of sides, i.e.,
2, 4, 6, 8. In the left table above, I have calculated the
amplitude of variation of the given harmonic that
results from these figures. Actually, the amplitudes
listed correspond to the maximum vs. minimum cross
sectional area of the figure. A Fourier analysis of the
59
Table I. Number of primary harmonics for a shape with a given number of sides, the ratio of the min//max cross section, and the maximum amplitude.
Table II. The maximum amplitude for a given primary harmonic.
cross sectional area under uniform rotation would

yield a series of harmonics (e.g., n, 2n, 3n, ); the
lowest order harmonic, n (an even number), would
have a lower amplitude than listed, but a full composition of the harmonics would yield the amplitude
listed, with n pairs of extrema per rotation cycle. The
right hand table just re-arranges things by harmonic
order. Note that the 3 and 6 sided figures (triangular
or hexagonal) both yield maximum amplitudes of the
sixth harmonic, and the same value, 0.156 magnitudes. It should also be noted that the above amplitudes are full-range amplitudes, maximum magnitude
minus minimum magnitude, which is double what is
meant by amplitude in harmonic analysis. What is
to be noted here is first, the amplitude of the second
harmonic has no geometric limit. In practice, it is
about equal to the prolate axis ratio of the equator.
For a rubble pile rotating slow enough that centrifugal force is not a major component, this has a
limit of around 4:1 (Harris et al., 2009), corresponding to an amplitude of about 1.5 magnitudes, and
indeed thats about the maximum we see among asteroids at low phase angle. In cases where the equatorial profile is not particularly elongate, that is, where
higher shape harmonics dominate, we can consider
the maximum amplitude that can result, which is
what I have done above. A square profile can result
in a 4th harmonic lightcurve of more than 0.3 magnitudes. Either a triangular or hexagonal profile can
produce a 6th harmonic amplitude of about 0.15 magnitude. The maximum of still higher harmonics becomes rapidly smaller, such that it is rare to see any
60
harmonic beyond about order ten with a statistically

significant amplitude.
I next examine two examples, one where the amplitude is so low that ambiguity is possible, and the
other such a large amplitude that more than two extrema per rotation cycle is essentially impossible. In
the latter case, we are forced to conclude that the asteroid is in a state of non-principal axis rotation.
2. 5404 Uemura
The asteroid (5404) Uemura was observed extensively from Modra in 2008 and again in 2009.
Figure 1 shows the photometric data from the two
apparitions plotted for the half-period of 1.7 hours,
and the double-period of 3.4 hours. Even the shorter
period shows three pairs of extrema, and the longer
period has six. Unfortunately, the longer period contains almost no odd harmonics, and the amplitude is
such that the longer period cannot be ruled out on
account of its amplitude: the amplitude of ~0.12 is
possible, even at sixth harmonic. This case was particularly provocative because the shorter period, 1.7
hours, is well shorter than the spin barrier for a
rubble pile of normal rock density, and this object
is well larger (~ 8 km diameter) than any other asteroid spinning faster than the rubble pile spin barrier
(Harris and Pravec, 2006). Thus one is faced with a
dilemma: the shorter period is indeed profound, but
could not be ruled out based on these observations.
When it was re-observed in 2009 (Figure 1c and 1d),
essentially the same lightcurve was obtained. There is
just a small hint of asymmetry, that is odd harmonics,

which repeats both years, thus leading us to prefer the
longer period solution over the shorter period. Nevertheless, it is a cautionary lesson in the difficulty of
determining the true period when the amplitude is
low. Many other examples exist among larger asteroids, e.g. (53) Kalypso (Pilcher and Pray, 2010), but
usually neither the half nor double period is particularly provocative as it is in the case of Uemura, so
resolving the ambiguity is not so urgent.
We can further note that ambiguities like this can
sometimes be resolved with observations at larger
phase angle. The geometric approximation that

leads to no odd harmonics is not valid at non-zero
phase angle, and in fact odd harmonics often appear
enhanced as the phase angle increases. So it can be
valuable to break the ambiguity by observing at the
largest practical phase angle. Another important factor is to attempt to observe at two or more widely
differing viewing geometries. Unfortunately, in the
case of (5404) Uemura, the viewing direction in 2009
was almost exactly 180 around from that in 2008, so
the projected geometry was nearly the same. Uemura
was observed again in early 2011 at an aspect sub-
Figure 2. (a) 2008 data plotted with shorter period of 1.7 hours. (b) 2008 data plotted with double period of 3.4 hours.
(c) 2009 data plotted with shorter period of 1.7 hours. (d) 2009 data plotted with double period of 3.4 hours. Note in
both years, the two closely spaced minima at 0.4 phase in (b) and 0.9 phase in (d) are nearly equal, while the opposite two minima, at 0.9 in (b) and 0.4 in (d), are unequal in the same way. Likewise, the two maxima near 0.2 in (b)
and 0.6 in (d) are nearly equal, while the other pair of maxima are slightly unequal, again in the same pattern both
years. These subtle differences, well below the noise level of any single night lightcurve but repeating in the two
apparitions, lead us to favor the longer period.
61
Figure 3. (a) The lightcurve of 2010 RC130 phased for a period of 8.687 hours. The fit is not exact, but mainly has an
implausibly large amplitude for more than two minima per rotation cycle. (b) The complex rotation fit, phased to the
longer of the two principal periods. The model fit lines show the variation cycle-to-cycle due to the second noncommensurate frequency; the plotted points across the bottom show the fit residuals.
stantially off the line of the 2008 and 2009 aspects,

and indeed showed a slightly greater asymmetry,
confirming the longer period.
3. 2010 RC130
The second example is the asteroid 2010 RC130,
an NEA that is small enough (about 100 m diameter)
that super-fast rotation is not unexpected. The lightcurve of this object revealed variations of about one
full magnitude, but an apparent repeat period of about
9 hours, featuring three deep minima of about 1 magnitude and several other shallower minima in between. Figure 3(a) displays the data taken at Lowell
Observatory phased to a period of 8.687 hours. The
fit is not perfect, but more importantly, the large amplitude precludes a rotation period with more than
two such deep minima. We therefore suspected a
complex tumbling rotation state, which a primary
62
period perhaps about 1/3 of the 8.7 h period, but

nearly harmonic with that period such as to produce a
beat pattern that repeats approximately every 8.7
hours. Pravec ran a dual-frequency fit appropriate for
NPA rotation, and indeed found a satisfactory fit with
periods of 2.8915 and 1.7378 hours. Figure 3b is that
solution, including the Lowell data plus observations
from other observers, phased to the longer period. As
noted, even if the fundamental rotation period is the
shorter one it is not unexpectedly short for an object
this small.
4. Conclusions
Deriving correct rotation periods from asteroid
lightcurves is often difficult, especially when confronted with the possibility of non-principal axis rotation, super-fast rotation (P < 2.2 hours or so), and/or
low amplitude of variation. The constraints on ampli-
tude of variation versus harmonic order can help resolve ambiguities. A further complication we have
not discussed here is the substantial population of
binary asteroids among objects less than ~10 km in
diameter, which often show mutual eclipse events,
with orbit periods not synchronous with the rotation
period(s). All of these complications illustrate the
need for extensive sets of high quality data to extract
correct results. On the positive side, these complexities can lead to deeper understanding of asteroid
structure beyond just measuring the rotation period.
5. Acknowledgements
I want to thank those who contributed data for
5404 Uemera and 2010 RC130: P. Birtwhistle, E.
Bowell, A. Carbognani, E. W. R. Cooney, . Gajdo,
A. Gald, J. Gross, D. Higgins, K. Hornoch, B.
Koehn, P. Kunirk, B. Skiff, D. Terrell, J. Vilgi,
and B. D. Warner
This work was supported by NASA grant
NNX10AL35G and NSF grant AST-1032896.
6. References
Harris, A. W., Pravec, P. (2006). Rotational properties of asteroids, comets and TNOs. Proc. IAU
Symp. 229, 439-447.
Harris, A. W., Fahnestock, E. G., Pravec, P. (2009).
On the shapes and spins of rubble pile asteroids.
Icarus 199, 310-318.
Pilcher, F., Pray, D. P. (2010). Minor Planet Bul. 37,
75-76.
63
64
PV Cephei - Boyd
PV Cephei and Gyulbudaghians Variable Nebula

David Boyd
British Astronomical Association, Variable Star Section
5 Silver Lane, West Challow, Wantage, OX12 9TX, United Kingdom
drsboyd@dsl.pipex.com
Abstract
The Herbig Ae star PV Cephei is embedded in a dense molecular cloud which also contains Gyulbudaghians
Variable Nebula. While the variability of PV Cephei has been studied intermittently over the past 35 years, little
attention has been paid to measuring the variability of the nebula. We report on photometric observations of both
objects over the past two years to investigate if there is a relationship between variations of the star and the nebula.
1. Pre-main-Sequence Stars
Young pre-main-sequence stars are intrinsically
variable. They are still in the process of contracting
onto the main sequence from dense molecular clouds
under the influence of gravity. As a young star contracts, it spins up and material not yet drawn into the
star forms an accretion disc around it. Material from
the disc falling onto the surface of the embryonic star
causes apparently random fluctuations in the stars
brightness. At the same time the initially weak magnetic field within the gas cloud is concentrated within
the contracting star which often results in an active
chromosphere, magnetically channelled accretion
from the disc onto the stars surface, and material
being ejected along the stars rotational axis in the
form of bipolar conical winds and collimated jets
(Romanova et al., 2009). The jets may collide at high
speed with clumps of gas and dust in the surrounding
molecular cloud creating the shock features known as
Herbig-Haro objects. Pre-main-sequence stars with
mass less than about two solar mass are classified as
T Tauri stars, those between two and eight solar mass
as Herbig Ae stars.
2. PV Cephei
PV Cephei is a fairly typical Herbig Ae star,
spectral type A5e, less than a million years old,
~500pc distant, with ~3.5 solar mass and luminosity
at its brightest approaching 100 times that of the Sun.
It is still in the process of contracting out of the molecular cloud which surrounds the young star. It has a
rather massive circumstellar disc, containing about
20% of the mass of the star (Hamidouche, 2010),
with an inclination of about 80 degrees relative to our
line of sight so we are looking at the disc almost
edge-on (Gomez et al., 1997).
Material from the disc accreting onto the surface

of the star causes periods of increased brightness
(Lorenzetti et al., 2011). Obscuration by either the
edge of the disc or dust close to the star may contribute to occasional fading of the star (Elek and Kun,
2010; Kun et al., 2011).
Measurements of red plates obtained with the
Crossley reflector at Lick Observatory in 1977-1979
showed the star brightening to magnitude 11 before
fading to magnitude 16 (Cohen et al. 1981). Little
attention was then paid to the star until 2005 when it
was observed at magnitude 12 (Elek and Kun, 2010).
Between 2007 and 2010 it varied in the range magnitudes 15 to 18 (Kun et al., 2011, Lorenzetti et al.,
2011). A search using adaptive optics failed to reveal
a close binary companion to PV Cephei, which could
act as a possible cause of variability (Connelley et al.,
2009).
PV Cephei is located near the edge of a molecular cloud. Outflow from the star towards the north has
cleared a conical cavity in the cloud through which
the stars light illuminates the surrounding interstellar
medium (Levreault, 1984; Arce and Goodman,
2002). Jets emitted by the star have created several
Herbig-Haro objects to the north and south of the star
as they plough into the surrounding interstellar medium (Reipurth et al., 1997). The northern jet is blueshifted and the southern jet red-shifted indicating
they are tilted out of the plane of the sky by about 10
degrees. The jets are also precessing in a helical
manner with a period of approximately 8300 years
(Gomez et al., 1997). From the positions of the Herbig-Haro objects, estimates of the velocity of the jets,
and the known distance to the star it is estimated that
new Herbig-Haro objects have been produced by
major emissions from PV Cephei at approximately
2000 year intervals (Reipurth et al., 1997).
65
PV Cephei - Boyd
3. Gyulbudaghians Nebula
4. New Observations
Light from PV Cephei shining on the inside of

the cavity in the surrounding molecular cloud cleared
by the northern outflow creates the variable reflection
nebulosity variously known as RNO125, GM29 or
Gyulbudaghians Nebula. The opposing outflow is
directed into the molecular cloud and its light is heavily attenuated to the point that it is barely visible.
The British Astronomical Associations Deep

Sky Section has an observing programme to monitor
the changing brightness of variable nebulae associated with young stars. As a contribution to this programme, I began monitoring PV Cephei and Gyulbudaghians Nebula in 2010 March using the equipment
listed above. I measured the brightness of PV Cephei
by conventional CCD photometry using an R filter
because of its predominantly red colour. I took care
to measure the sky background outside the area of the
nebula. I used a square aperture sufficiently large to
encompass the nebula at its brightest to measure the
total light intensity from the nebula plus sky background. The level of sky background was estimated
by measuring the light intensity in a similar square
offset to the east of the star to avoid contamination by
the opposing jet hidden within the molecular cloud.
The sky background was subtracted from the nebula
plus sky background to give a measurement of the
light emission from the nebula alone and this was
converted to a magnitude using the usual formula.
This method has been used consistently throughout
the project keeping all aperture sizes constant. The
zero point of the nebula magnitude scale was chosen
such that the mean magnitudes of the star and the
nebula were the same for convenience of graphical
representation.
Figure 1. 33 min exposure of the field around PV Cephei.
Figure 1 is a 33-min stacked exposure of the

field around PV Cephei taken with a 0.35-m SCT,
SXVR-H9 CCD camera, and an R filter on 2011 August 19 when the fan-shaped reflection nebula was
relatively bright. It also shows some of the HerbigHaro objects close to the star.
Although several papers have reported measurements of the variability of PV Cephei, the changing
brightness of the nebula has only been commented on
qualitatively based on its appearance in photographs
and more recently CCD images. For example: The
rapid variability of the nebula indicates that changes
in its luminous structure must result from variable
illumination. (Gledhill et al., 1987).
While it has been assumed that the source of this
varying illumination is the variability of PV Cephei,
no direct evidence has so far been presented to support this assumption. In particular, no one appears to
have systematically measured the varying brightness
of the nebula and correlated this with the variability
of the star.
66
Figure 2. Light curves of PV Cephei and Gyulbudaghians Nebula.
The stars high declination means it can be followed throughout the year from the UK. During the
following two years PV Cephei varied between magnitudes 15.8 and 17.1. Over the same interval, the
nebula followed a broadly similar pattern with two
bright episodes separated by a deep fade (Fig. 2).
PV Cephei - Boyd
A search for periodic signals in the light curves

of PV Cephei and the nebula failed to reveal any significant signals with period between 2 and 200 days.
Figure 3 shows images of the field around PV
Cephei taken at various times between 2010 April
and 2012 March to illustrate the variability of the
nebula. Each image is typically a stack of 25-30 one
minute exposures
5. Correlated Variation
To investigate whether there is a relationship in
time between these two light curves, a Discrete Correlation Function (Edelson and Krolik, 1988) was
computed adding a variable delay to the light from
the star. Maximum correlation occurs when the stars
light is delayed by 27 2 days relative to that of the
nebula. Figure 4 shows the two light curves superimposed with this delay.
Figure 3. Images of Gyulbudaghians Nebula at intervals between April 2010 and March 2012.
67
PV Cephei - Boyd
Connelley M. S., et al. (2009). Astron. J. 138, 1193.

Edelson, R. A., Krolik, J. H. (1988). Ap. J. 333, 646.
Elek E., Kun M. (2010). Journal of Physics Conference Series 218, 12021.
Figure 4. Superimposed light curves of PV Cephei (delayed by 27 days) and Gyulbudaghians Nebula.
For the first time this provides evidence in support of a direct correlation between changes in the
light output of the star and the brightness of the nebula. However, while the major increases and decreases of the two light curves are consistent, the
amplitudes of these changes do not follow each other
with the same degree of consistency. In particular,
the increase in brightness of the star in the latter part
of 2011 was considerably less than the corresponding
increase in brightness of the nebula.
A possible explanation for this arises from consideration of the mechanisms involved in these
changes. It is likely that a steady brightening of the
star is caused by a major accretion event. This in turn
causes the nebula to brighten. If either the edge of the
circumstellar disc or the dust clouds which are in
continual movement around the star happen to partially obstruct the light it emits in our direction, this
may reduce the apparent increase in the stars brightness. Movement of dust around the star may also
create a shadowing effect and lead to the varying
pattern of illumination we see in the nebula.
It is interesting to calculate the light travel time
from the star to the nebula to see if this is consistent
with the observed delay between their respective light
curves. Using the published distance to the star, the
apparent inclination of the outflow, the distance in
images from the star to the nebula and the image
plate scale, this light travel time works out at approximately one month.
Only continuing observation will reveal whether
this apparent correlation is maintained in the future.
6. Acknowledgements
This research has made use of NASA's Astrophysics Data System Bibliographic Services.
7. References
Arce H. G., Goodman A. A. (2002). Ap. J. 575, 911.
Cohen M., et al. (1981). Ap. J. 245, 920.
68
Gledhill T. M., et al. (1987). Monthly Notices of the

Royal Astronomical Society 229, 643.
Gomez M., et al. (1997). Astron. J. 114, 265.
Hamidouche, M. (2010). Ap. J. 722, 204.
Kun M., et al. (2011). Monthly Notices of the Royal
Astronomical Society 413, 2689.
Levreault R. M. (1984). Ap. J. 277, 634.
Lorenzetti D., et al. (2011). Ap. J. 732, 69.
Reipurth B., et al. (1997). Astron. J. 114, 2708.
Romanova, M. M., et al. (2009). Monthly Notices of
the Royal Astronomical Society 399, 1802.
Visual Observations of Mu Cephei - Turner
Why Visual Observations of Mu Cephei are Important

David G. Turner
Saint Marys University
Halifax, Nova Scotia, B3H 3C3, Canada
turner@ap.smu.ca
Abstract
Our understanding of Type C (M supergiant) semiregular variables such as Mu Cep, Betelgeuse, S Per, BC Cyg,
etc. is limited to some extent by a lack of regular and accurate observations for the stars. Bright objects like Mu
Cep, in particular, are frustrating to observe because of their long periods of variability, almost two and a half
years for Mu Cep, a lack of reference standards lying in close proximity, and red colors, which affect visual perceptions of brightness and/or photometric reductions. Yet such observations are essential for proper phasing of
spectroscopic studies, which are beginning to reveal intriguing insights into the nature of their variability.
1. Introduction
decades. Yet as a group they are also relatively unstudied, for obvious reasons: periods of variability
spanning anywhere from 100 to 900 days, low amplitudes of variability (for most), complex spectra
dominated by TiO molecular bands, and photometric
brightnesses that are a challenge to calibrate reliably
on a common system. That is where simple eye estimates may play an important role.
2. Current Knowledge
Figure 1. The SRC variable S Persei according to the

Digitized Sky Survey (Moncrieff 2011).
Variable M-type supergiants such as S Per (Fig.

1) belong to a small group of long-period variables,
referred to as Type C semiregulars (SRCs), that are
intrinsically interesting for several reasons. They are
identified as likely precursors to core collapse supernovae, typically Type II (possibly within our lifetimes), and have been identified (Pierce et al., 2000;
Jurcevic et al., 2000) as potentially valuable extragalactic distance indicators because of their high luminosities, brighter than that of Cepheids, and the fact
that, like Cepheids, they appear to obey their own
period-luminosity relation (Stothers, 1969; Moncrieff, 2011). They have actually been used as distance indicators for Local Group galaxies for several
So few researchers specialize in and actively observe Type C semiregulars that even the description
of their particulars given in the General Catalogue of
Variable Stars is misleading. Photographic light
curves for CL Car and IX Car presented in Figs. 2
and 3 illustrate some of the main characteristics. True
Type C semiregulars, unlike their M supergiant cousins, the LC irregular variables, display highly regular
light variations, which can be attributed to pulsation
(Stothers, 1969; Turner et al., 2012). They often display superposed long-term variations in mean brightness of unknown origin. Perhaps starspot cycles occur as well? The light amplitudes depend upon the
effective temperature of the stellar photosphere and
the luminosity of the star, with the cool, luminous
supergiant variables displaying the largest light amplitudes (Moncrieff, 2011). The SRC variable IX Car
is a most curious case, since its light amplitude appears to increase as the star brightens.
The periods of variability attributed to individual
objects vary from source to source in the literature,
and probably depend upon the method adopted to
measure them. Fourier techniques do not always
work well for SRC variables. For the regular light
variations of the intriguing SRC variable BC Cyg
(Fig. 4), for example, the author developed a variant
of existing techniques in which measured times of
69
Figure 2. Eye estimates for CL Car (spectral type M5 Ia)

from photographic plates in the Harvard collection. The
blue light variability has a dominant period of 512 days
superposed upon gradual changes in mean brightness.
Figure 4. Eye estimates for BC Cyg (spectral type M3-5
Ia) from photographic plates in the Harvard and
Sternberg collections. The blue light variability has a
dominant period of 692.8 days established from
observed times of light maximum.
Figure 3. Eye estimates for IX Car (spectral type M2 Iab)

from photographic plates in the Harvard collection. The
blue light variability has a dominant period of 357 days,
with the amplitude appearing to increase as the star
brightens.
Figure 5. Estimated times of light maximum as a
function of cycle count for BC Cyg, where points
represent maxima measured photographically and plus
signs from AAVSO visual estimates.
light maximum are correlated with cycle number to

establish the mean period (Turner et al., 2012), as in
Fig. 5. BC Cyg varies with a mean period of 692.8
days, but individual times of light maximum appear
to deviate from the predicted times by as much as
100 days on average (Fig. 6). Much of the scatter is
probably caused by deviant observations used in the
analysis (see recent AAVSO visual observations in
Fig. 7), but a small proportion of the scatter is consis-
70
tent with what is observed for random fluctuations in

period for pulsating variables (Turner et al., 2009).
The establishment of an accurate period of variability for BC Cyg was an essential step towards understanding the physics of the variability in the star.
A reliable ephemeris for BC Cyg permitted an accurate phasing of older and more recent spectroscopic
and spectrophotometric observations of the star
(White and Wing, 1978; Turner et al., 2012).
Figure 6. O-C residuals for BC Cyg from measured times

of light maximum, with symbols as in Fig. 5.
Figure 7. Recent visual estimates for BC Cyg from the

AAVSO database. The large scatter makes it difficult to
establish the periodicity of the star and times of light
maximum reliably.
Many Type C semiregulars are members of open

clusters. S Per and several SRC variables around it
are members of the young cluster Chi Persei (NGC
884), for example, BC Cyg is an outlying member of
the extremely young cluster Berkeley 87, and Mu
Cep is a member of the young cluster Trumpler 37.
At least 17 of the known SRCs, about a third of the
existing sample, are identified as belonging to Galactic clusters (Moncrieff, 2011).
The location of the stars in their cluster H-R diagrams implies masses for them of order 15-20 solar
masses, with indications of higher turnoff point
masses for the associated clusters. It appears that the
stars have undergone considerable mass loss since the
main-sequence stage prior to reaching their current
state. The final stages of nuclear energy generation in
the cores of these stars involve fusion of heavy elements through alpha particle capture, and are predicted to occur fairly quickly, in some cases over
intervals shorter than the pulsation periods of the
stars. The possibility that a known Type C semiregular variable might soon explode as a supernova cannot be discounted, in which case all information
about its light signature prior to the event would be of
significant scientific value.
It is here that visual observations of SRC variables become important. The literature contains very
few photoelectric or CCD observations of such stars.
A long-term photoelectric monitoring of the Type C
semiregulars Alpha Ori and TV Gem by Wasatonic et
al. (2011) is an exception, although future and current ground-based surveys should fill the void. However, all estimates of reddening and luminosity for
known SRC variables must currently rely upon visual
estimates, since high precision light curves are simply unavailable in most cases. A particularly challenging case is that of Mu Cephei.
3. Mu Cephei
Figure 8. Older visual brightness estimates for Mu Cep

from the AAVSO database for dates normalized to HJD
2400000.
That in turn revealed that the systemic velocity

of the star as well as its temperature and luminosity
class variations are correlated directly with its brightness variations, in much the same fashion as in pulsating classical Cepheids (Turner et al., 2012). Apparently the dominant variability in such stars is the
result of radial pulsation, as argued years ago by
Stothers (1969).
Mu Cephei is one of the brighter SRC variables,

and is bright enough and close enough spatially to the
Cepheid Delta Cephei that both can be observed simultaneously, either through binoculars or without
optical aid. The star exhibits a relatively small light
amplitude, however, which means that the collection
of AAVSO observations for the variable (Figs. 8 and
9) is particularly difficult to analyze either for periodicity or mean brightness. Moncrieff (2011) used
Fourier techniques to find a range of best-fitting periods for the star covering values from 747 days to
1116 days, with an optimum value of 844 days for all
AAVSO estimates. The resulting phasing of available
spectroscopic and spectrophotometric measures (radial velocities, spectral types) with that period was
71
particularly troublesome (Moncrieff et al., 2010;

Moncrieff, 2011), implying that a different period is
needed to match all available observations.
period of variability for Mu Cephei, but the membership of the star in the cluster Trumpler 37 implies that
this variable M supergiant is a valuable calibrator for
the SRC period-luminosity relation, particularly since
it is the longest period Type C semiregular defining
the relationship. It is at long pulsation periods that
extragalactic SRC variables are most easily detected
and reliably observed photometrically. Nailing down
an accurate value for the pulsation period of Mu Cep
could have implications for our knowledge of the
extragalactic distance scale, which, at present, is essential for understanding the nature of the universe
on large scales.
4. Future Observations
Figure 9. More recent visual brightness estimates for Mu

Cep from the AAVSO database for dates normalized to
HJD 2400000.
Figure 10. Recent visual brightness estimates for Mu

Cep by the author (black plus signs) and from the
AAVSO database (red plus signs) relative to older photoelectric and Hipparcos data (filled circles) phased to a
period of 787 days.
There are a few photoelectric observations for

the star available in the literature, as well as more
than three years of monitoring from space by Hipparcos. Those observations were combined with recent visual observations of the star by the author to
infer a possible best-fit period of 787 days for the
star, with considerable uncertainty. The data phased
to that period are plotted in Fig. 10, and include some
recent visual estimates from the AAVSO database.
The light curve is probably less well defined than
indicated by the observations, although the recent
steady drop in brightness is real. At least another year
or two of monitoring is needed to establish the period
accurately. It may even be undergoing irregular
changes. It seems like such a minor parameter, the
72
While it is true that observations of all Type C

semiregular variables are valuable, a strong case can
be made for the current need of observations for Mu
Cephei. The period of variability for the star is still
very uncertain, and the scatter in recent AAVSO observations for it (Fig. 10) suggests a need for improving the accuracy of observers worldwide. The great
visual observer Johann Friedrich Julius Schmidt got
practical assistance from Argelander in the 1800s at
the start of his long-term program of monitoring the
brightness of Polaris (Schmidt, 1857), and such mentoring would seem to be of particular value in observations of red variables like the SRCs.
Most observers are aware of the Purkinje effect
when observing red stars at low light levels, since the
stars may appear fainter than they actually are because of the changing color sensitivity of the eye at
different light levels. Such effects may be present in
some older AAVSO estimates for the brightness of
BC Cyg, for example (Turner et al., 2006), since it
was often observed to be about 0.5 magnitude fainter
than expected several decades ago. In the case of Mu
Cep an opposite effect may be present (Fig. 10), since
some observers appear to overestimate the brightness
of the star, perhaps through overcompensation for the
Purkinje effect. The red variables have long been a
major component of AAVSO programs (Williams
and Saladyga, 2011), but the method of averaging
observations by individual observers into mean light
curves of higher precision may need improvement in
the case of M supergiant variables. A concerted program to recalibrate observers of SRC variables onto a
common, accurate system linked to the Johnson UBV
system may be needed if we are to advance our understanding of what is truly an intriguing and valuable class of variable star.
All observations of Mu Cephei would be welcomed by this researcher, who is willing to provide
advice on different techniques to try. The precision of
eye estimates is best when the object being observed

is near the limit of vision (Turner, 2000), and the
author has found that observing bright stars through
binoculars that have been defocused enough to place
either the program star or the comparison star at the
limit of detectability provides more reliable magnitude estimates than is the case for in-focus images.
Experience in making eye estimates of brightness for
stars on out-of-focus plates in the collection of the
Harvard College Observatory is sufficient to confirm
the validity of such an approach. Perhaps more experienced visual observers have their own tricks to
share?
of the Semiregular M Supergiant Variable BC

Cygni. Publ. Astron. Soc. Pacific 118, 1533.
Turner, D. G., Percy, J. R., Colivas, T., Berdnikov, L.
N., Abdel-Sabour Abdel-Latif, M. (2009). Stochastic Processes in Yellow and Red Pulsating Variables.
In Stellar Pulsation: Challenges for Theory and Observation. AIP Conference Series 1170 (J. A. Guzik,
P. A. Bradley, eds.). p. 167.
Turner, D. G., Moncrieff, K. E., Short, C. I., Wing,
R. W. (2012). The Nature of Variability in M Supergiants: The Forgotten Type C Semiregulars. Astrophys. Space Sci. Proceedings, in press.
5. Acknowledgements
The author is grateful to Kathleen Moncrieff for
permission to use some of her thesis work in this
study, and acknowledges with thanks the variable star
observations from the AAVSO International Database that were contributed by observers worldwide
and used in this research.
6. References
Jurcevic, J. S., Pierce, M. J., Jacoby, G. H. (2000).
Period-Luminosity Relations for Red Supergiant
Variables, I. The Distance to M101. Mon. Not. Roy.
Astron. Soc. 313, 868.
Wasatonic, R. P., Guinan, E., Engle, S. (2011). RedEye Astronomy: 15 Years of V-band and Near-IR
TiO Photometry of the Red Supergiants Alpha Orionis And TV Geminorum. Bull. Am. Astron. Soc.
43, AAS 218, #322.05.
White, N. M., Wing, R. F. (1978). Photoelectric
Two-Dimensional Spectral Classification of M Supergiants. Astrophys. J. 222, 209.
Williams, T. R., Saladyga, M. (2011). Advancing
Variable Star Astronomy: The Centennial History of
the American Association of Variable Star Observers.
Cambridge University Press, Cambridge.
Moncrieff, K. E. (2011). Improving Our Understanding of SRC Variable Stars. Ph.D. Thesis, Saint
Marys University.
Moncrieff, K. E., Turner, D. G., Short, C. I., Bennett,
P. D. (2010). Spectral Type and Radial Velocity
Variations in Three SRC Variables. Odessa Astron.
Publ. 23, 86.
Pierce, M. J., Jurcevic, J. S., Crabtree, D. (2000).
Period-Luminosity Relations for Red Supergiant
Variables, I. The Calibration. Mon. Not. Roy.
Astron. Soc. 313, 271.
Schmidt, J. F. J. (1857). Uber veranderliche Sterne.
Astron. Nachr. 46, 293.
Stothers, R. (1969). On the Pulsation Hypothesis for
Massive Red Supergiants. Astrophys. J. 156, 541.
Turner, D. G. (2000). Observing Bright Cepheids
without Optical Aid. J.A.A.V.S.O. 28, 116.
Turner, D. G., Rohanizadegan, M., Berdnikov, L. N.,
Pastukhova, E. N. (2006). The Long Term Behavior
73
74
ROAD Observations of Variable Stars - Hambsch
ROAD (Remote Observatory Atacama Desert):

Intensive Observations of Variable Stars
Franz-Josef Hambsch
VVS, AAVSO, BAV, GEOS
Oude Bleken 12, B- 2400 Mol, Belgium
M31@telenet.be
Abstract
The author discusses his new remote observatory under pristine skies and the intensive observations of variable
stars he is accomplishing. The stars under investigations are mainly cataclysmic variables in request of AAVSO,
CBA and VSNET alerts as well as other type of stars like RR Lyrae stars. Examples of dense observations of
different cataclysmic variables as well as a RR Lyrae star are presented. The focus goes to the first bright outburst of SV Ari (Nova Ari 1905) since its discovery as well as the first outburst of UGWZ candidate BW Scl. Also
results for VW Hyi, another cataclysmic variable will be shown. Furthermore an intensively observed RR Lyrae
star will be highlighted.
1. Introduction
It is an amateur astronomers dream to observe
under pristine dark and clear skies nearly every night
like at the sites where the professional astronomical
observatories are located. Such a dream normally
never comes true. However, modern techniques and
infrastructures in most countries make it possible
nowadays to observe from remote sites using off the
shelf technology. The author installed a remote observatory under the dark skies of the Atacama Desert
close to the town of San Pedro de Atacama. The telescope is housed at SPACE (San Pedro de Atacama
Celestial
Exploration;
http://www.spaceobs.com/index.html). The owner is
an amateur astronomer, too who formerly worked at
ESO at the big telescope sites in Chile. In 2003 he
started SPACE, which has been extended to telescope
hosting since a couple of years.
I got in contact with him in 2009 and decided to
put my observatory at his place.
Unfortunately, delivery of the telescope took
much longer than anticipated and only in 2011 July,
the dome, mount and telescope could be installed.
Since 2011 August 1, the remote observatory is producing data every clear night. So far in the less than
8.5 month of operation this amounted to about 220
data taking nights. Not bad, is it not?
I am collaborating with J. Patterson from the
Center of Backyard Astrophysics (CBA) and T. Kato
from the VSNET email alerts on the observation of
cataclysmic variable stars.
Furthermore I am interested in observing RR Lyrae stars. Here, I collaborate with the Groupe Europeen dObservations Stellaires (GEOS). In Belgium I am member of the Werkgroep Veranderlijke
Sterren (WVS) on High Amplitude Delta Scuti

(HADS) stars. I am also member of AAVSO and the
German Bundesdeutsche Arbeitsgemeinschaft fr
Vernderliche Sterne (BAV).
During the many clear nights a lot of data is being gathered on many stars. Since the space here is
limited I restrict myself to some highlights, the outburst of SV Ari, BW Scl and VW Hyi. Also a RR
Lyrae type star with a strong Blazhko effect will be
mentioned.
2. Observatory
The remote observatory in Chile houses a 40-cm
f/6.8 Optimized Dall-Kirkham (ODK) from Orion
Optics, England. The CCD camera is from FLI and
contains a Kodak 16803 CCD chip with 4kx4k pixels
of 9 m size. The filter wheel is also from FLI and
contains photometric BVI filters from Astrodon.
In Belgium, where I live, I have a roll-off-roof
observatory in my backyard, housing three scopes for
variable star observations if it ever gets clear. Belgium is neither famous about its weather nor its light
pollution, nevertheless there is still some room for
interesting observations. Those telescopes are a Celestron C14 and C11 and a Meade ACF14. The telescopes are equipped with two SBIG ST8XME and an
STL-11kXM CCD camera using BVRI photometric
filters.
To make the telescope park complete I also collaborate with AAVSO in sharing an AAVSONet telescope (a Meade 30-cm SCT) with an SBIG ST9XME CCD.
75
tic as it seems that many brightness determinations of

M. Wolf were about 2 magnitudes too bright.
Fig. 1 Photo of the remote telescope installation in Chile.
Fig. 1 shows an image of the remote telescope in

Chile. It is housed in a clamshell dome, making easy
movement of the telescope possible without the need
to follow with a shutter of a normal dome.
Images of a nights session are either acquired
with ACP or CCDCommander automation software.
Further analysis in terms of determination of the
brightness of the stars is done using a program developed by P. de Ponthierre (2010). The data are then
finally submitted to the AAVSO.
Fig. 2 Light curve of the observations of SV Ari over a

period of more than two weeks.
Fig. 2 shows the observations I took of this star

during a period of more than two weeks. During this
period the star dropped more than one magnitude in
brightness. It is also clearly visible from this figure
that individual nights show brightness variations of
about 0.3 magnitudes.
3. SV Aries (Nova Aries 1905)

As a first example, I show the results of the
campaign on SV Ari. SV Ari was discovered on 1905
Nov. 6 by M. and G. Wolf in Heidelberg, Germany,
at a photographic magnitude of 12.0. It was reported
that it had brightened from mag 22.1. Kimpel and
Jansch
reported
a
possible
sighting
in
1943 September at a magnitude of 15.7, but this was
not confirmed. No brightness increase has been observed for this star ever since.
Then on 2011 August 2.788 UT, R. Stubbings
observed the field of SV Ari and saw an object at
magnitude 15.0. He sent an alert via the mailing list
cvnet-outburst around to ask for confirmation. Also
via the VSNET mailing list the information of the
outburst of SV Ari was spread. This lead to confirmation by G. Masi (2011) and R. Fidrich (2011). I saw
those alerts via VSNET and since G. Masi immediately took a time series and observed superhumps, I
also decided to go after this star and started a time
series at Aug., 2.844 UT just 1.34 h after the initial
discovery. A first analysis of the data of G. Masi by
the Kyoto team (Ohshima et al., 2011) showed that
the star is probably an SU UMa type dwarf nova with
a superhump period of about 1.54 h. The present outburst seems also not as bright as the one during the
discovery of magnitude 12. However, another mail
from T. Kato (2011) on VSNET reported that probably the original brightness estimate was too optimis-
76
Fig. 3 Light curve of the observations of SV Ari over a

period of more than two weeks.
Fig. 3 shows the time series observation during

the first night. Clearly a superhump of 0.3 magnitudes is visible. The magnitude of the superhumps
reduced to about 0.15 magnitudes after a couple of
days.
Based on the analysis by the Kyoto team, there
have been distinct stages in the evolution of superhumps in SV Ari. The mean period before Aug. 4
was 0.05574(18) d, which has since shortened to
0.05519(5) d. Later data also showed that SVAri was
gradually declining (~0.05 mag / day). This seemed
very slow for an ordinary SU UMa type dwarf nova.
Thus probably this object may be a WZ Sge type
dwarf nova. This information is based on e-mail exchanges via VSNET. Towards the end of 2011 August, SV Ari had dimmed towards magnitude 18.
4. BW Sculptoris
Another example of an intensively followed star
is BW Sculptoris, which went in its first ever observed outburst on October 21. BW Scl is also a cataclysmic variable star. The AAVSO published on the
same day a special notice #261 mentioning the outburst of BW Scl. It was visually observed by M. Linnolt on Oct. 21.3146 at a magnitude of 9.6 (vis). The
outburst was confirmed by A. Plummer at mag. 9.4
(vis). The star has conflicting classifications in literature and is probably a WZ Sge-type dwarf nova. On
Oct. 25 AAVSO issued an Alert Notice 449 concerning this outburst.
Fig. 4 Light curve of the observations of BW Scl over a

period of nearly three month..
The star was followed by me already in the preoutburst phase. However, I missed the outburst as I
thought the star is not doing much so stopped observations on Oct. 14, just a week before the outburst
took place. Of course I restarted observations immediately after the news was spread and followed the
star over a period of 2.5 month. Fig. 4 shows the development of the brightness over the full observing
period.

period of nearly three month.
After a few days nice superhumps developed of

about 0.25 mag as can be seen in Fig. 5. After a week
into the outburst the early superhump period was
determined to be 0.054308(3) by Ohshima et al.
(2011a). After a rapid declining phase from mag.
11.7 to about mag. 14 starting at about JD 2455877
again nice superhumps of 0.3 mag developed.
5. VW Hydri
On request by professional astronomers, I started
observations of VW Hyi just as it went into superoutburst, although that was a surprise to the pros as the
superoutburst was not expected yet. My observations
triggered satellite observations of the star
Fig. 6 Observations of VW Hyi during superoutburst and

normal outburst.
VW Hyi is a popular cataclysmic variable in the

Southern sky. Many studies have been performed
about this star; see e.g. AAVSO (2010). During quiescence the star is at magnitude 14.4. Normal outbursts happen on average every 27.3 d and last about
1.4 d. The superoutburst happens on average every
179 d and lasts for about 12.6 d. Fig. 6 shows impressively this behavior of VW Hyi with one superoutburst and two normal outbursts.
Fig. 7 Observations during one night at the top of the

superoutburst.
77
The star develops strong superhumps during its

outburst as can be seen in Fig. 7. The variation of
those humps reaches 0.5 magnitudes.
6. Example of an RR Lyr Star:

V1820 Ori
This type of variable is named after the prototype, the variable star RR Lyrae in the constellation
Lyra. RR Lyr stars are pulsating horizontal branch
stars with a mass of around half of our Sun's. Their
period is short, typically less than one day.
My interest in observing RR Lyr stars is on one
hand due to the short period of those stars. So within
one night you can see already quite a change in
brightness. On the other hand the stars also show
some brightness modulations which is known as the
Blazhko effect. Back in 1907, S. Blazhko observed
this effect for the first time on the star RW Dra
(Smith, 2004). The Blazhko effect is to date not
really understood and needs further observational
campaigns. Recently due to the Kepler and CoRot
satellite missions more insight into this phenomenon
has been gained as the satellites can of course observe the stars continuously, which is impossible for
Earth-bound observations. Nevertheless observations
from Earth are also very valuable as can be seen in
many publications on this subject in the astronomical
literature. Fig. 8 shows the phase diagram of the RR
Lyr star V1820 Ori, which has been observed from
Chile over a full season (more than 3 month). It is
obvious from the figure that the light curve is not
regular, the maximum brightness is changing over
more than 0.5 mag. And also the moment of maximum time is changing. So a rather complex light
curve is the result. Presently the data are under analysis and the results are intended to be published.

period of nearly three month.
78
7. Conclusion
The remote observatory under pristine skies in
the Atacama Desert opens up great possibilities to
observe variable stars. Intensive follow-up observations over many days, weeks or even months are possible due to the stable weather conditions. The given
examples show impressively what is possible. Collaborations are searched for in order to contribute to
scientific research of common interest.
8. References
AAVSO (2010) Variable Star of the Month.
http://www.aavso.org/vsots_vwhyi
de Ponthierre, P. (2010) LesvePhotometry software at
http://www.dppobservatory.net/AstroPrograms/Softw
are4VSObservers.php
Fridrich, R. (2011). e-mail to cvnet-outburst.
Kato, T. (2011). vsnet-alert e-mail message 13544.
Masi, G. (2011). vsnet-alert e-mail message 13538.
Ohshima, T. (2011). vsnet-alert e-mail message
13543.
Ohshima, T. (2011a) vsnet-alert e-mail message
13799.
Smith, H. A. (2004). RR Lyrae Stars. Cambridge
University Press. p. 103. ISBN 0521548179.
ER Ursae Majoris de Miguel at al.
ER Ursae Majoris: A dwarf nova with surprises

Enrique de Miguel
CBA (Huelva), Observatorio del CIECEM, Matalascaas, 21076 Almonte, Huelva, Spain
Departamento de Fsica Aplicada, Universidad de Huelva, 21071 Huelva, Spain
edmiguel63@gmail.com
Joe Patterson
Department of Astronomy, Columbia University, 550 West 120th Street, New York, NY 10027
jop@astro.columbia.edu
Jonathan Kemp
Gemini Observatory, Hilo, HI 96720
jk@155west.org
William Stein
AAVSO and CBA, Las Cruces, NM
George Roberts
CBA (Tennessee), 2007 Cedarmont Dr., Franklin, TN 37067
T. Campbell
CBA (Arkansas), 7021 Whispering Pine, Harrison, AR 72601
Franz-J. Hambsch
CBA (Mol), Oude Bleken 12,B-2400 Mol, Belgium
Tom Krajci
CBA (New Mexico), PO Box 1351 Cloudcroft, NM 88317
Shawn Dvorak
AAVSO, 1643 Nightfall Dr., Clermont, FL 34711
Robert A. Koff
CBA (Colorado), Antelope Hils Observatory, 980 Antelope Drive West, Bennett, CO 80102
Etienne Morelle
CBA (France), Sirene ObservatoryLagarde dApt, France
Michael Potter
CBA (Baltimore), 3206 Overland Ave., Baltimore, MD 21214
David Cejudo
CBA (Madrid), Observatorio El Gallinero, El Berrueco, Madrid, Spain
Steve Brady
AAVSO, 49 Bay State Rd, Cambridge, MA 02138
Kenneth Menzies
Tigh Speuran Observatory, Framingham, MA
79
Abstract
ER Ursae Majoris is a relatively bright member of the SU UMa-type dwarf novae that undergoes frequent eruptions of amplitude ~2 mag every ~4 d. In addition to these ordinary dwarf-nova outbursts, ER UMa shows
brighter eruptions (superoutbursts) lasting ~20 d with a recurrence time of ~44 d. As other members of the SU
UMa class, ER UMa displays a characteristic periodic modulation (superhumps) in its light curve during the
course of a superoutburst, the period being a few percent longer than the orbital period of the underlying binary
system. A routine photometric patrol of ER UMa during its January 2011superoutburst became much more interesting after noticing the presence of an additional modulation in its light curve, with a period slightly shorter than
the orbital period. Similar photometric waves (known as negative superhumps) have already been observed in
other cataclysmic variables, mainly in nova-like systems, but only in a small handful of SU UMa-type stars during
quiescence, and never before during a superoutburst. We report in this communication the results of a worldwide photometric campaign that was orchestrated right after the discovery of negative superhumps in ER UMa,
and provide a thorough analysis of the data obtained from more than 10 stations of the Center of Backyard Astrophysics (CBA) network, which include more than 1000 hours of photometric data over a 4-month baseline on
2011, and a similar (still in progress) dense coverage in 2012.
1. Introduction
On 2011 January 19, an e-mail sent by the
VSNET network (vsnet-alert 12671) reported that ER
Ursae Majoris (ER UMa) was likely undergoing a
superoutburst, as the visual estimate of V = 12.6 from
Gary Poyner seemed to suggest. As cataclysmic variable devotees probably know, this is not the type of
news that calls for immediate change of observing
plan in the middle of the night: after all, ER UMa is
known to be a very active dwarf nova that pays short
visits (~3 d) to its low state before resuming its eruptive activity with outbursts of amplitude ~2.53 mag.
Such a state of frenetic activity is characteristic of a
handful of dwarf novae that are collectively known as
ER UMa stars.1 These stars constitute a small group
of the broader SU UMa class of dwarf novae. Most
SU UMa stars have orbital periods shorter than ~2.2
hr. They all exhibit two distinct types of eruptions:
short normal outbursts lasting a few days; and longer,
less frequent, and slightly brigther superoutbursts
with typical duration of ~1520 d.
All SU UMa stars show distinctive periodic oscillations in brightness during superoutburst, with
periods a few percent longer than the orbital period.
These characteristic light modulations have been
given the name of ordinary (also known as common,
apsidal, or positive) superhumps. As a general rule,
ordinary superhumps in SU UMa stars are exclusively observed during superoutburst, and are extinguished once the superoutburst ends and the system
returns to its quiescent state.
Superoutbursts recur in time at a rather regular
pace usually, even more regularly than the se1
Sometimes, also known as RZ LMi stars. Accredited members of this subgroup are, in addition to ER UMa, DI UMa (Kato et
al., 1996), RZ LMi (R95, Nogami et al., 1995), V1159 Ori (R95,
Patterson et al. ,1995), and IX Dra (Ishioka et al., 2001; Olech et
al., 2004)
80
quence of normal outbursts occurring between consecutive superoutbursts. In usual SU UMa stars, the
recurrence time (supercycle) is normally in the range
1501000 d. But there are outliers. Stars of the WZ
Sge subclass undergo very infrequent superoutbursts
at intervals of decades. ER UMa is also an outlier but
exactly for the opposite reason.
ER Ursae Majoris (PG 0943+521) was discovered in the Palomar Green survey for blue objects
(Green et al., 1986), and later identified as a cataclysmic variable (Robertson et al., 1995, R95 hereafter; Misselt and Shafter, 1995; Kato and Kunjaya,
1995; Zhao et al., 2006). Based on a photometric
study consisting on 454 exposures on 145 nights during the interval 1993 October 6 to 1995 January 5,
R95 concluded that ER UMa exhibited eruptions with
the typical long (and brighter) and short (and slightly
fainter) dichotomy of SU UMa systems. This identification was confirmed by the detection of superhumps with periodicity Psh = 94.41(3) min (Kato et
al., 2003). This is just 3% longer than the orbital period Porb = 91.67(4) min (Thorstensen et al., 1997),
an excess value typical for SU UMa stars. However,
with a supercycle period of 42.95(6) d and normal
eruptions taking place every 4.4(1) d (R95), ER UMa
cannot be simply considered as a typical SU UMa
dwarf nova. This high outburst frequency is the main
observational hallmark of ER UMa stars, all having
short supercycles (1950 d), which testifies to a very
large accretion rate and high intrinsic luminosity.
Our observational program was initially conceived as a long-term project aiming to investigate
possible minor differences between superoutbursts on
the same object. ER UMa, being bright enough, and
having long-lasting and frequent superoutbursts,
seemed to be a good choice; and that alert on 2011
Jan 19, the right excuse to start the observations. The
data obtained that night confirmed that ER UMa was
in superoutburst, exhibiting the usual superhump-like
modulations, but with a period of ~89.6 min, definitely shorter than Porb. In the cataclysmic variable
jargon, photometric waves with periods slightly less
than Porb are known as negative superhumps.2 They
are definitely less common than ordinary superhumps, and are almost exclusively observed in novalike cataclysmic variables.
The campaign that was subsequently launched is
still in progress, and from the very beginning, ER
UMa has been gifting us with daily negative superhumps. Here we report on a preliminary analysis of
the main features found after more than 2000 hr of
photometric coverage, paying particular attention to:
(1) the global light curve; and (2) the characterization
of the negative superhump in ER UMa and its stability.
from adding the V magnitude of the comparison star

to our differential magnitudes.4 Heliocentric corrections were applied to all observation times. Hereafter,
time will be referred to in the truncated form HJD
HJD (true) 2455000 and, similarly, dates in truncated JD format.
2. Observations
All observations reported here were performed
by contributors to the Center of Backyard Astrophysics (CBA), a worldwide network of observing stations equipped with small telescopes in the 0.250.40
m range.3 A total of 12 CBA stations contributed to
this campaign: 4 in Europe (2 in Spain, 1 in Belgium,
and 1 in France), and 8 in America (7 in the USA,
and 1 in Chile).
The 2011 season included time-resolved
photometry on 97 nights over a 131 day baseline,
from January 19 to May 30, totaling 1190 (~900 nonoverlapping) hours. The 2012 campaign was started
on 2011 December 4 and is still in progress. By the
time of writing, time-resolved photometry has been
performed on 91 nights over a 123 day baseline, this
amounting to more than 1200 hours of photometric
data.
Most of the observations were carried out unfiltered, with integration times in the range 4060 s.
Raw images were dark-subtracted and flat-fielded
before performing differential photometry relative to
a nearby star. There was no unique choice of comparison star, but GSC 349300920, with V =
14.203(2) and B = 14.833(4) according to a photometric calibration by Henden (2010), was the most
commonly used. The dense coverage in both seasons
ensured temporal overlapping data sets, which allowed us to bring all data under a common magnitude
scale by adding appropriate offsets. This procedure is
not expected to introduce errors larger than 0.030.04
magnitudes. All magnitudes reported here are formed
Figure 1. Light curves of ER UMa from 2011 January 19

to May 30 (2011 season) and from 2011 December 4 to
2012 April 4 (2012 season). Data points are mean values
of nightly time series.
3. Global Light Curve

We show in Fig. 1 the overall light curves obtained in the 2011 and 2012 seasons. Data points included in this figure correspond to the averages over
each individual time series. As more than 90% of the
data sets consisted on series spanning more than 3 hr,
each data point in the figure is expected to be unaffected by variations at short timescales. Except for
the last segment of the 2011 season and the early
stages of the 2012 season, ER UMa was followed
essentially on a daily basis. The supercycle was sampled uniformly in both seasons, with 40% of the
data collected during superoutbursts. This is consistent with the expected time ER UMa expends in each
of the different stages along a supercycle. Inspection
of Fig. 1 shows that ER UMa exhibits the expected
sequence of long-lasting, bright eruptions, followed
by a series of much shorter, slightly fainter normal
outbursts. We now turn to discuss these features in
more detail.
A plausible explanation of the negative superhump clock is

the retrograde precession of a tilted accretion disk (Patterson et al.,
1993, 1997).
3
More details relative to the CBA can be found at
http://cbastro.org
For enthusiasts of standard magnitudes, we have estimated

that V magnitudes follow from our magnitudes by adding 0.22.
81
Figure 2. Expanded portions of

peroutbursts in ER UMa.
Fig. 1 showing su-
As can be seen in Fig. 1, our data are able to

capture three superoutbursts in their entirety, one in
2011 (11SO2), and two in 2012 (12SO2, and 12SO3).
These are shown in Fig. 2 on an expanded scale. For
two additional superoutbursts (11SO1 and 11SO3),
we missed the initial stages, but we managed to cover
most of the plateau and fading stages of the eruption.
For the purpose of discussion, let us take the JD
925944 superoutburst (12SO2, middle panel of Fig.
2) as reference: it starts from quiescence at 15.44
mag; reaches supermaximum in 3.2(2) d while
82
brightening 2.5(1) mag; enters a smooth, non-linear

fading stage, where the brightness decreases by
1.02(8) mag over the subsequent 12.1(3) d; and drops
rapidly toward quiescence, the brightness decreasing
by 1.47(8) mag on the following 3.7(2) d. The eruption has been alive for 18.9(1) d, a value that is consistent with the estimate of 20 d quoted by Kato and
Kunjaya (1995). The overall shape of the light curve
remains very similar from one superoutburst to the
next. The rise to maximum brightness spans ~3.5 d,
and seems to occur at a non-constant rate much
faster over the first 12 d. On average, the brightness
at supermaximum is 12.94(4) mag. The duration of
the plateau stage of all superoutbursts is remarkably
similar, 11.4(3) and 12.0(2) d for 11SO2 (top panel
of Fig. 2) and 12SO3 (lower panel of Fig. 2), respectively. These values are to be compared with the duration of 13 days reported by R95. The difference is
small and probably of no physical significance. Assuming 11.4 d for the duration of the two partially
covered superoutbursts in 2011, we can estimate that
supermaxima occurred around JD 576.1 (for 11SO1)
and JD 674.2 (for 11SO3) to within 0.5 d. This will
be useful for later discussion in the next section.
A closer inspection of Fig. 2 shows two subtle
differences among the fully covered superoutbursts.
Firstly, a normal outburst seems to interrupt the rapid
fading branch of 11SO2; secondly, 12SO3 is not triggered from quiescence, but before the preceding
normal outburst had relaxed to the quiescent state. As
a result, both superoutbursts are slightly shorter than
our reference superoutburst by 12 d.
The main eruption is followed by a sequence of
5 normal outbursts. This is definitely at variance with
the sequence of 7 outbursts at intervals of 4.4(1) d
reported by R95. We find no definite recurrence time
for the normal outbursts in our data. Times between
consecutive outbursts tend to be clustered around 6 d,
but we find values anywhere in the range 5.510 d. It
is noteworthy that the interval between outbursts
seems to stretch over a supercycle. This is not at the
cost of the duration of individual outbursts which
almost invariably involve 4 days but is seemingly
due to the tendency of the system to remain in its
faint state slightly longer as the supercycle proceeds
and the system approaches the next superoutburst.5
This being said, we should note that ER UMa probably

does not have a truly state of quiescence, but short periods -typically 1-3 d, and rarely as long as 6-7 d-- where it wanders
around a mean magnitude of 15.55. Something qualitatively similar is found for the other ER UMa stars.
Figure 3. Phased light curves of the superoutbursts in

the 2011 and 2012 seasons folded on P = 49.3 d (2011)
and P = 53.8 d (2012).
What about the supercycle period? The data in

Fig. 1 suggest a value of ~50 d between consecutive
superoutbursts in 2011 and of ~55 d between the only
two main eruptions registered in 2012. These values
are significantly longer than those reported previously (R95, Kato and Kunjaya, 1995), and call for a
more detailed analysis. A power spectrum of the data
pertaining to superoutburst stages shows a strong
peak at 49.3(3) d in 2011 and at 53.8(1.3) d in 2012.
When folding the data from each season on the corresponding period the resulting phased light curves
show little dispersion, as can be appreciated in Fig. 3.
As one may anticipate, no single period can bring
superoutbursts in both seasons into a neat, phased
light curve. Further insight into the time variations of
the supercycle period can be obtained from an O-C
analysis. As a fiducial marker, we choose the time at
which the plateau stage ends (or, equivalently, the
time at which the ensuing rapid decay to quiescence
takes place). The choice follows from practical reasons: this feature is observed in all superoutbursts,
and can be timed with a precision of ~0.2 d. The linear ephemeris is
2455585.5(1.5) + 51.0(3) E
(1)
The OC deviations from this ephemeris are

shown in Fig. 4. Assuming a parabolic OC diagram,
we find dP/dt = 1.5(2) x 102, to be contrasted with
the value of 4x103 reported by R95. Although consistent with our data, this period change must certainly be non monotonic. We may conclude at this
point that the supercycle seems to be stretching in ER
UMa at a rate yet to be convincingly determined. One
would be tempted to conclude that ER UMa is telling
us that it is becoming an SU UMa star, but the baseline of our observations is insufficient to draw any
definite conclusion.
Figure 4. O-C deviations of the times of superoutbursts

in ER UMa using the linear ephemeris given in Eq. (1).
4. Periodic Signals
As stated in the Introduction, the ultimate reason
that triggered this extensive campaign on ER UMa
was the discovery of a strong periodic signal slightly,
but clearly shorter than the 91.67 min orbital signal.
We turn next to characterize this signal and its stability throughout the 2011 and 2012 seasons. In the
process, we also analyze the existence of other possible periodic signals.
A common feature to all light curves is the presence of prominent oscillations of varying amplitude
at all phases of the supercycle, sometimes as large as
0.8 mag. This has been a daily constant during 2011
and 2012 with no exceptions. Not even one single
night. Some light curves were seen to exhibit a seemingly simple pattern, as it is the case at supermaximum and 23 d thereafter (top frame in Fig. 5). But
in most instances, nightly light curves were complex,
without a visually clear regular pattern, and largely
changeable from night to night (see lower frame in
Fig. 5). In general, amplitudes (in magnitude units)
were larger as the system was fainter.
4.1 Frequency analysis

We performed a Fourier analysis using the Period04 (Lenz and Breger, 2005) program. This package allows identification of frequencies and sequential prewhitening of selected frequencies by a leastsquare technique. In order to emphasize the actual
changes in signal strength, we normally make use of
intensity units rather than magnitudes in the analysis.
Relative intensities, I/Io are formed from magnitudes,
m, according to
m mo = 2.5log(I/Io)
(2)
83
where mo = 16 mag is arbitrarily chosen here to set

the unit of intensity. The mean intensity was removed
from each nightly light curve, as it was any obvious
trend.
Figure 5. Upper frame: 2012 February superoutburst,

dominated by positive superhumps. Lower frame: ER
UMa at quiescence. The apparent modulation corresponds to negative superhumps.
In order to gain in frequency resolution, we

formed sets of long light curves over different intervals of evolution. The sets corresponding to superoutbursts included photometry obtained over all
stages (whenever observed), and therefore contain
data pertaining to the rise and decay stages in addition to data along the main plateau. With the convention introduced before, these data sets are labeled
11SOn, and 12SOn, with n = 1,2,3 representing each
of the supercycles observed in 2011, and 2012, respectively. We also formed light curves for each individual normal outburst. In this case, each light
curve included all data from the start of a given outburst to the start of the next, thus including photometric data obtained while ER UMa was at quiescence.
These data sets are referred to as 11NOn-m, with m =
1, 2, labeling the outburst during supercycle n in
2011, and similarly for normal outbursts in the 2012
season. We refer to tables 2 and 3 for further details.
As a useful reference, we recall that the orbital frequency is = 15.708 cycles/day (abbreviated in the
following as c/d), the orbital period being Porb =
91.67 min (Thorstensen et al., 1997).
84
4.2 The 2011 season

The frequencies of the periodic signals found in
each data set over the 2011 season have been included in Table I, along with the amplitude of the
corresponding signal. Amplitudes are in units of intensity, Io, as defined in Eq. 2, and represent half of
the full variation of the signal (sometimes, called
"semi-amplitude"). These values follow from a leastsquare optimization procedure implemented within
the Period04 package. Occasionally, the output amplitudes were found to be unreliable. This happened
particularly nearby supermaxima, where the oscillations in the photometric wave (in intensity units) are
of large amplitude and the least-square procedure
converged poorly. In these cases, we simply give in
Tables I and II an estimate from the light curves.
The power spectrum of the 2011 January superoutburst (11SO1) is dominated by a strong peak at
16.07(2) c/d, (associated period of 89.6(1) min). A
peak at 32.13(3) c/d, corresponding to the first harmonic of the fundamental signal, is also visible in the
spectrum at higher frequencies. No other peak of
physical significance is observed in the spectrum.
This is surprising, as the period of ordinary superhumps in ER UMa is known to be significantly
longer (94.4 min; Kato et al., 2009). Having a periodicity about 2.1% shorter than the orbital period,
the detected signal must correspond to negative superhumps. Negative superhumps had been detected
before in some SU UMa stars (Patterson et al., 1995),
but they had never before been observed during the
course of a superoutburst. The amplitude of negative
superhumps (in intensity units) was seen to remain
essentially constant along the superoutburst.
Also seemingly intriguing is the complete absence of positive superhumps during the 2011 January superoutburst. Even more striking is the fact that
negative superhumps survived the superoutburst and
were detected at all phases of the supercycle. The
peak-to-through amplitude of these oscillations was
as large as 0.8 mag when ER UMa was in its low
state, but the amplitude was not seen to change appreciably when expressed in intensity units (see values included in Tables I and II). Also the mean frequency of negative superhumps remained nearly constant after the superoutburst, with values wandering
in the range 16.0116.10 c/d.
The fundamental signal in the power spectrum
during the following superoutburst in March 2011
(11SO2) occurs at 16.05(1) c/d, and corresponds once
again to negative superhumps. A second peak is also
found at 15.23(2) c/d, fully consistent with the frequency of positive superhumps. When the power
spectra of individual nights are analyzed, we con-
label
11SO1
11NO1-1
11NO1-2
11NO1-3
11NO1-4
11NO1-5
11SC1
11SO2
11NO2-1
11NO2-2
11NO2-3
11NO2-4
11NO2-5
11SC2
11SO3
11NO3
11SC3
JDi-JDf
581-590
591-597
598-603
604-409
610-614
615-621
581-621
622-638
639-643
644-650
651-656
657-661
662-671
622-671
672-688
689-711
672-711
nights
8/10
7/7
6/6
6/6
5/5
7/7
39/41
16/17
5/5
7/7
5/6
5/5
5/10
43/50
11/17
5/23
16/40
hours
56.52
52.45
76.94
65.28
34.25
103.17
388.61
208.14
47.60
73.29
26.94
39.17
31.89
427.03
69.69
21.29
90.98
-, A16.07(2), 0.363
16.01(2), 0.389
16.03(2), 0.345
16.06(2), 0.358
16.04(2), 0.365
16.10(2), 0.307
16.051(9), 0.271
16.08(2), 0.349
16.02(2), 0.412
16.03(2), 0.324
16.05(3), 0.380
16.08(2), 0.456
16.08(3), 0.256
16.052(12), 0.238
16.07(2), 0.291
16.04(3), 0.381
16.057(20), 0.226
+, A+
15.23(2), ~2.5
15.253(12), ~2.5
15.16(2), 0.114
15.137(20), 0.182
Table I: The frequencies and amplitudes (in intensity units)of the photometric signals exhibited by ER UMa on
2011 over different intervals of evolution.
clude that positive superhumps developed only when

the supermaximum was reached, the amplitude being
~510 times as large as those of their negative counterparts. The amplitude decreased nearly exponentially, and 4 days after supermaximum, no trace of
positive superhumps was found in the power spectra.
By contrast, negative superhumps were persistently
observed all nights.6 Incidentally, we note that this
may explain why positive superhumps were not detected during the previous superoutburst. As argued
in the previous section, the supermaximum of the
2011 January superoutburst is estimated to have occurred around JD 576, 5 days before our first observation (on JD 581).
label
JDi-JDf
nights
hours
12SC1
900-926
14/27
67.59
12SO2
927-944
17/18
172.74
12NO2-1
945-950
5/5
34.82
12NO2-2
951-960
9/10
96.12
12NO2-3
961-968
5/8
65.86
12NO2-4
969-976
5/8
38.88
12NO2-5
977-980
5/5
48.18
12SC2
927-980
46/54
456.60
12SO3
981-997
17/17
216.43
12NO3-1
998-1003
6/6
96.41
6
To be more precise,
they were not observed
12NO3-2
1004-1012
9/9 at supermaxi94.77
mum, nor one day later, but this is most likely due to the over12NO3-3
7/7 during those
69.92
whelming
energy content1013-1019
of the positive superhumps
24-48
hours. As negative981-1019
superhumps were clearly
detected in477.53
the
12SC3
38/38
The power spectrum during the 2011 April superoutburst (11SO3) showed, once again, two main
peaks at 16.07(2) c/d (negative superhumps) and
15.16(2) c/d (positive superhumps), with negative
superhumps being again the winner.
4.3 The 2012 season

A similar frequency analysis was implemented
for the 2012 season. Frequencies and amplitudes of
the periodic signals have been collected in Table II.
The season was started early in 2011 December and
ER UMa underwent the first superoutburst of the
season on 2012 January 1. Data collected during
-, A16.07(2), 0256
16.08(2), 0.340
16.02(2), 0.303
16.04(2), 0.312
16.06(3), 0.340
16.09(2), 0.256
16.16(3), 0.315
16.075(14), 0.260
16.06(2), 0.292
16.02(2), 0.268
16.02(2), 0.293
16.05(3), 0.328
16.040(13), 0.260
+, A+
15.24(2), ~3.2
15.281(14), ~3.2
15.25(2), ~3.5
15.226(13), ~3.5
power spectra before supermaximum even during the rising

Table II: As in table I, but for the 2012 season.
branch, and 2 days after supermaximum, we have no reasons to
believe that they magically disappeared at supermaximum only to
reappear (again, magically) 2 days later.
85
2011 December have been grouped into a single light

curve (12SC1) for frequency analysis. The spectrum
during that time segment is dominated by a peak at
16.07(1) c/d: negative superhumps did not seem to
have disappeared.
The power spectra during the 2012 January
(12SO2) and 2012 February (12SO3) superoutbursts
are essentially the same as those corresponding to
superoutbursts 11SO2 and 11SO3: two main signals
at 16.08(1) c/d and 15.24(2) c/d associated to negative and positive superhumps, respectively. A conspicuous peak at or near frequency 16.07 c/d was the
norm in all power spectra.
quencies.7 Rather than timing individual maxima, we

proceeded to extract one timing per night. For this
purpose, we first identified all relevant frequencies of
each nightly curve, fitted the light curve to a sinFourier series using the least-square technique embedded in the Period04 package, and determined one
epoch of maximum associated with the nightly value
of -. In total, we extracted 93 and 88 timings from
the 2011 and 2012 campaigns, respectively. A linear
regression yields
2455581.503(2) + 0.062269(2) E
(3)
for the 2011 season, and
5. Stability of Negative Superhumps

2455900.661(4) + 0.062282(3) E
Up to this point, we have provided strong evidence of the presence of negative superhumps in ER
UMa at all phases of the supercycle. While positive
superhumps are only excited during superoutburst,
the essentially constant amplitude of the negative
superhumps testifies to a weak dependence, if any,
with the particular outburst state of the system.
What about their stability? During superoutbursts, the period of positive superhumps in SU UMa
stars is known to be unstable, with characteristic
times in the range |dP/dt| 525 x10-5 (Kato et al.,
2009). This wandering is usually manifested in the
power spectrum of a superoutburst in the form of
peaks close to the fundamental signal, typically to
within a distance of 0.15 c/d, and they show up in
the spectrum if the resolution is sufficiently high
(otherwise, the result is a broadening of the fundamental peak). Side peaks around the frequency of
negative superhumps, -, were seen in most of the
spectra analyzed in the previous section, which points
to a likely variation of - with time. Inspection of the
values of - included in Tables 2 and 3 seems to suggest that some wandering do exist, but also that the
mean - does not change from one superoutburst to
the next. If we stopped at this point, we would be
allowed to conclude that the period of negative superhumps remains essentially constant (at least, during superoutbursts), and varies slightly afterwards.
We decided, however, to investigate this point further.
Time variations of - were analyzed using an
OC analysis of the times of maxima of the negative
superhump waves. In principle, timing these maxima
poses no major problem when the oscillations are
well defined. In the case of ER UMa, most nightly
light curves show secondary peaks at all sort of fre-
(4)
for the 2012 season, the corresponding frequencies

being - = 16.0594(5) c/d and - = 16.0560(8) c/d,
for each season. Thus the negative superhump in ER
UMa is very stable, with a characteristic period
change of dP/dt = (41) x 10-8. This is equivalent to
an increase of 1.3 seconds per year, certainly a very
stable clock.
The tenacious reader may still wonder how the
apparent wandering of - can be reconciled with the
alledged stability of negative superhumps in ER
UMa. The clue is provided in Fig. 6. We plot in that
figure the OC deviations from the ephemeris given
in Eqs. (3) and (4). According to the figure, the period of the negative superhump varies at shorter timescales, but these variations are cyclical (almost sinusoidal), and so they essentially vanish when averaged
over each season. But what we find most striking is
the fact that the frequency of these cyclic variations
coincides with the frequency of the supercycle. Assuming a sinusoidal variation of the OC deviations
shown in Fig. 6, the best fits are achieved with periods 48(4) d and 55(8) d for the 2011 and 2012 seasons, in full consistency with the supercycle length
found in each season. It is noteworthy that the minima in the OC diagrams correspond to superoutburst
stages. Also, we note that this modulation keeps coherence from season to season, which provides further evidence of the intimate interplay between period modulations of the negative superhump and supercycle.
7
They resemble low-energy fluctuations -thus becoming
more apparent when ER UMa is in its low state- but show no coherence. They are definitely non periodic and probably related with
flickering.
86
quence of 7 outbursts at regular intervals of

4.4(1) d reported for ER UMa during 19931995. These outbursts tend to be increasingly
brighter as the supercycle proceeds, and have a
rather constant duration of 4 d.
3. All observed superoutbursts had a similar pattern. The amplitude of the eruption is 2.60 mag
(in white light), with the system rising from
15.55 to 12.95 mag (in white light) in ~3.5 d.
The brightness increases at a lower rate than in
the case of normal eruptions. After supermaximum, the brightness decreases slowly and nonlinearly over the next 11.8(4) d at a mean rate
of 0.085 mag/day. Afterwards, the system returns to quiescence in ~3.8 d at a rate of ~0.4
mag/day. The mean duration of the eruption is
19.1(5) d.
4. Our data are consistent with a lengthening of
the supercycle. It was found to be 49.3(3) d in
2011, and 53.8(1.3) d on 2012. The supercycle
was of 44 d in the 1993-1995 photometric study
reported by R95. If this is a secular variation,
we may be witnessing a transition of ER UMa
to a SU UMa-like star, with much lower accretion rates and, subsequently, longer waiting
times between superoutbursts. However, the
current information at our disposal is limited
and the above conclusion is not fully warranted.
Probably, a simpler explanation is that the supercycle in ER UMa stars is not very stable, but
subject to cyclic (or irregular) variations.
Figure 6. O-C deviations of the times of maxima of negative superhumps.
6. Summary
1. We report extensive time-resolved photometry
of ER UMa carried out on 2010 and 2011. The
global light curve shows the sequence of superoutbursts and interspersed outbursts expected for a highly active cataclysmic variable.
2. In all cases, ER UMa was seen to display 5
normal outbursts before a new superoutburst
was triggered, with no clearly defined regularity. This is at variance with the reported se-
5. For reasons yet to be known, ER UMa is systematically exhibiting since 2011 a highly stable signal with a period shorter than the orbital
period. We ascribe this modulation to negative
superhumps. The seasonal value of the period
has increased from 89.667(3) min in 2011 to
89.686(4) min in 2012. This negative superhump is present at all phases of the supercycle,
and its amplitude in intensity units is essentially
constant. The period of the negative superhump
exhibits cyclic variations modulated by the supercycle.
7. Acknowledgements
We welcome this opportunity to acknowledge
the work of observers of the Center of Backyard Astrophysics (CBA) who contributed to this and related
campaigns.
87
8. References
Gao, W., Li, Z., Wu, X., Zhang, Z., Li, Y. (1999). AP
527, L55.
Green, R. F., Schmidt, M., Liebert, J. E., (1986).
ApJS 61, 305.
Henden, A. (2010). AAVSO Database.
Ishioka, R., Kato, T., Uemura, M., Iwamatsu, H.,
Matsumoto, K., Martin, B. E., Billings, G. W,. Novak, R. (2001). PASJ 53, L51.
Kato, T., Kunjaya, C. (1995). PASJ 47, 163.
Kato, T., Masuda, S., Nogami, D. (1995). IBVS 4233.
Kato, T., Nogami, D., Baba, H. (1996). PASJ 48,
L93.
Kato, T., Nogami, D., Masuda, S. (2003). PASJ 55,
L7.
Kato, T., et al. (2009). PASJ 61, S395.
Lenz, P., Breger, M. (2005). Commun. Asteroseismol.
146, 53.
Misselt, K.A., Shafter, A.W. (1995). Astron. J. 109,
1757.
Nogami, D., Kato, T., Masuda, S., Hirata, R., Matsumoto, K., Tanabe, K., Yokoo, T. (1995). PASJ 47,
897.
Olech, A., Zloczewski, K., Mularczyk, K., Kedzierski, P., Wisniewski, M., Stachowski, G. (2004). Acta
Astron. 54, 57.
Patterson, J., Thomas, G., Skillman, D.R., Diaz, M.
(1993). ApJS 86, 235.
Patterson, J., Jablonski, F., Koen, C., ODonoghue,
D., Skillman, D. R. (1995). PASP 107, 1183.
Patterson, J., Kemp, J., Saad, J., Skillman, D. R.,
Harvey, D., Fried, R., Thorstensen, J. R., Ashely, R.
(1997). PASP 109, 468.
Robertson, J. W., Honeycutt, R. K., Turner, G. W.
(1995). PASP 107, 443.
Thorstensen, J. R., Taylor, C. J., Becker, C. M., Remillard, R. A. (1997). PASP 109, 477.
88
Zhao, Y., Li, Z., Wu, X., Peng, Q., Zhang, Z., Li, Z.
(2006). PASJ 58, 367.
Epsilon Aurigae Spectroscopy - Hopkins
Small Telescope Spectroscopy of Epsilon Aurigae

Jeffrey L. Hopkins
Hopkins Phoenix Observatory
7812 West Clayton Drive
Phoenix, Arizona USA 85033-2439
phxjeff@hposoft.com
Abstract
With the availability of professional quality and reasonably priced spectrographs such as the Star Analyser, for
low-resolution work, and the Lhires III with a 2400 l/mm grating, for high-resolution work, along with easy to learn
and use spectral processing software such as RSpec, what was once well beyond the backyard astronomer is
now well within the means of the advanced amateur astronomer allowing them to do serious astronomical spectroscopic work. The 27-year eclipse of Epsilon Aurigae is over until 2036, but there is still much going on out-ofeclipse that warrants continued observations. This paper will describe low and high-resolution spectroscopy of
Epsilon Aurigae done at the Hopkins Phoenix Observatory. In particular five prominent hydrogen Balmer and
sodium D lines will be examined.
1. Introduction
3.
A trimmed and calibrated line profile of the

first order spectrum.
The purpose of this paper is to explore the hydrogen Balmer lines and sodium D lines of Epsilon
Aurigae. Both low and high-resolution spectroscopy
is used.
The exploration of these spectroscopic lines is an
excellent learning exercise as well as a chance to discover some interesting features and changes. The
low-resolution spectroscopy provides a complete
visible spectrum picture. High-resolution spectroscopy provides a detailed look at each spectral region
of interest.
While the recent eclipse of Epsilon Aurigae is
over, the star system is constantly changing. The
hydrogen alpha (H) line has been studied in detail
and has provided tantalizing spectra of an emission
line horn dance. The hydrogen Balmer lines H, H,
H, and H have been neglected. The H and sodium
D lines have ample telluric and neon lines for
wavelength calibration. The shorter wavelength lines
are much more challenging and require a
considerable changing of techniques.
4.
An instrument wavelength response calibrated

line profile.
2.2 Low-Resolution Equipment
Zero and First Order Spectra
2. Low-Resolution Spectroscopy
Star Analyser Spectrograph (Shelyak Instruments)
2.1 Introduction
For low resolution spectroscopy four things are
sought
1.
A low-resolution zero and first order spectrum

image of epsilon Aurigae.
2.
A low-resolution first order spectrum image of

Epsilon Aurigae.
A Star Analyser spectrograph was used for the

low-resolution work. This unit is the size of an eyepiece filter and screws into the nosepiece of a CCD
camera for zero order spectra and the front of a star
diagonal for just first order spectra imaging. The
spectra should be horizontal with the zero order spectrum to the left and the first order spectrum with the
shorter wavelength (blue part) to the left and the
89
longer wavelength red part to the right. The Star Analyser must be rotated relative to the cameras CCD
chip to obtain the correct orientation.
A Meade 8 LX90 telescope was used along
with a Meade DSI Pro II monochrome CCD camera
to image the spectra.
2.
A trimmed and wavelength-calibrated (including heliocentric correction) line profile of each

spectrum.
3.
A normalized line profile for each spectrum.
4.
Precise measurements of the wavelength of

each line.
5.
Precise measurements of the equivalent width

of each line of interest.
3.2 High-Resolution Equipment
Lhires III Spectrograph (Shelyak Instruments)

Meade 8 LX90 Telescope and DSI Pro II Camera
2.3 Wavelength Calibration

There are multiple techniques for the wavelength
calibration. For low resolution the zero order spectrum and one identifiable first order spectrum line
can be used. For a first order spectrum image, two
lines identified lines can be used.
3. High-Resolution Spectroscopy
3.1 Introduction
For high-resolution spectroscopy five things are
sought
1.
High-resolution spectrum images of the H,

H, H, H, H and sodium D lines of epsilon
Aurigae.
90
Meade 12" LX200 Telescope, Lhires III Spectrograph and

CCD Cameras
For high-resolution work the built-in neon calibrator of the Lhires III provides the easiest set of reference neon lines. The disadvantage is for high-
precision work just flipping the calibrator in and out

of the light path can introduce slight physical stress to
the spectrograph and thereby introduce slight changes
in the wavelength relationship. An alternate approach
is to use an external neon ring on the front of the telescope to add the neon lines to the same spectrum image as the main object. For the spectral areas where
there are telluric lines, these can be used to precisely
calibrate wavelengths. A final method is to use various elemental gas discharge tubes on the bench to
provide calibration lines. This is particularly important in the shorter wavelength Balmer lines. In some
cases the neon lamps provide lines, but are not neon
lines. These lines can be calibrated and used, however.
4. Spectrum Processing Software

4.1 Introduction
the Profile window can be dragged top the left to

expand it.
4.2 Low Resolution

4.2.1. Wavelength Calibration
There are several methods for calibrating lowresolution line profiles. These include two points,
zero order and one point, one point and known dispersion and multi point non-linear calibration.
4.2.2. Instrument Response
While it is not required for wavelength determination or element identification, for the line profile to
look anything like a standard stars profile, the profile
must be corrected for the instruments spectral response.
4.3 High Resolution

4.3.1. Wavelength Calibration
RSpec Initial Screen
Spectral images are first processed like regular

astronomical images. Exposures are set to stay within
the cameras linearity range. Dark frames are subtracted for exposures greater than 1.0 second. The
image is then loaded into a spectrum processing
software application. Because of its ease of use for
this project the commercial program RSpec was chosen. Upon opening RSpec two windows are presented, the Image/Spectrum window and the Profile
window. The spectrum image will be in the Image
window. The sky background can be subtracted and
the image orientation adjusted if needed. Two orange
horizontal lines bracket the spectrum. These can be
dragged up or down to precisely bracket the image.
Whatever is between these lines has its pixel columns
summed and a line profile created and displayed in
the Profile window. A calibrated line profile of a
spectrum is needed if empirical data are to be produced from it. Once a line profile has been created
There are no zero order spectra seen with a highresolution spectrum. Calibration of high-resolution
line profiles is limited to two points and multi point
calibration. The dispersion method is not recommended. Because resolution in the tenths of an angstrom can be achieved, it is very important to correct
for the Earths motion relative to the object of interest. This is known as heliocentric correction. As the
Earth travels in its orbit around the Sun, the radial
velocity relative to a stationary astronomical object
outside the solar system can vary from +67,000 MPH
to 67,000 MPH. In addition, the Earths rotation can
contribute an additional velocity of up to 1,042
MPH. The heliocentric correction compensates for
this motion by calculating a wavelength shift that is
added or subtracted from a lines wavelength. RSpec
does not have this capability so the French Freeware
program VSpec was used for this calculation.
4.3.2. Normalization
To calculate the equivalent width of a line, the
line profile must be normalized. To normalize a line
profile an average value for the continuum in the area
of the line of interest must be determined. That value
is then divided into the line profile causing the continuum to be centered on a value of 1.0.
4.3.3. Equivalent Width
Once a line profile has been normalized the
equivalent with of a line can be determined. The
91
equivalent width is an indication of the power of the

line. Its the lines area. The equivalent width is the
width in angstroms for the area of a rectangle from
the 1.0 level to the zero level equivalent to the lines
area. Studying the equivalent width of a line as it
changes over time can give insights as to what is going on in a star system.
Equivalent Width Calculation
5. Observations
5.1 Introduction
5.2 Low-Resolution Spectroscopy

For low-resolution work images of Epsilon
Aurigaes spectra were taken to include the zero and
first order spectra and also just the expanded first
order spectra.
Equipment:
Telescope:
Spectrograph:
Slit:
Grating:
Detector:
Observation
Double Date:
Local Time:
Universal Time:
Decimal UT Day:
JD:
Spectral Region:
Exposure:
Fits File:
Results
Air Mass:
Rv:
8 LX90, f/10
Star Analyser
N/A
100 l/mm
DSI Pro II
24/25 January 2012
18:53 MST
01:53
25.07847
2,455,951.5785
Visible Spectrum
1.0 Second
Eaurnz.fts
1.1478
19.89 k/s
Spectral observations of Epsilon Aurigae were

done with both low and high-resolution equipment.
All the hydrogen Balmer lines can be seen in the lowresolution spectra. For high-resolution spectra there
are images for each of the first 5 lines and Na D lines.
Zero and First Order Spectra Line Profile
Hydrogen Balmer Lines
Observatory Log (General)

Observer:
Jeff Hopkins
Observatory:
Hopkins Phoenix Observatory
Latitude:
33d 30' 06'
+33.5017 decimal
Longitude:
+112d 13' 23"
112.229 decimal
Object:
Epsilon Aurigae
RA (2000):
5h 01m 58.1s
Dec (2000):
+43d 49' 24"
92
First Order Spectrum Calibrated Line Profile
RSpec has the capability of generating a color

pseudo spectrum of a calibrated line profile.
5.3.2. Epsilon Aurigae Hydrogen beta

Epsilon Aurigae Pseudo Spectrum
5.3 High-Resolution Spectroscopy

For high-resolution work, the five most prominent hydrogen Balmer lines (H, H, H, H and H)
and sodium D lines of Epsilon Aurigae were observed. All Exposures were 480 seconds (8 minutes)
with dark frames subtracted
Equipment:
Telescope:
Spectrograph:
Slit:
Grating:
Imaging CCD:
Guiding Camera:
12" LX200 GPS

Lhires III (2400 l/mm)
22 m
2400 l/mm
Orion StarShoot G3 mono
DSI Pro II
5.3.1. Epsilon Aurigae Hydrogen alpha

Observation
Double Date:
17/18 February 2012
Local Time:
19:47 MST
Universal Time:
02:47
Decimal UT:
18.11597
JD:
2,455,975.6160
Spectral Region:
H
6563
Reference Wavelength: 6562.81
Micrometer Setting:
2010
Fits File:
EaurHalpha.fts
Results
Wavelength Center:
b = 6560.98
a = 6562.84
r = 6564.91
Heliocentric Correction: 0.576
RV (km/s):
26.33 k/s
Air Mass:
1.0166
Equivalent Width:
b = 0.006
a = -0.484
r = 0.036
VR (If applicable):
0.167
Epsilon Aurigae H Calibrated Line Profile
The region was a bit more difficult to work

with. The neon lines must be used for the wavelength
calibration here.
Observation
Double Date:
Local Time:
Universal Time:
Decimal UT Day:
JD:
Spectral Region:
Reference Wavelength:
Micrometer Setting:
Fits File:
Results
Wavelength Center:
Heliocentric Correction:
RV (km/s):
Air Mass:
Equivalent Width:
VR (If applicable):
18/19 February 2012

19:16 MST
02:16
19.09444
2,455,976.5944
H 4861
4861.350
1466
Eaur Hbeta.fits
4862.10
0.429
26.45 k/s
1.019
0.868
N/A
5.3.3. Epsilon Aurigae Hydrogen gamma

As with the H line the H line must also rely on
the neon reference lines for the wavelength calibration.
Observation
Double Date:
Local Time:
Universal Time:
Decimal UT Day:
JD:
Spectral Region:
Micrometer Setting:
Fits File:
Results
Wavelength Center:
RV (km/s):
Air Mass:
21/22 February 2012

19:32 MST
02:32
22.10556
2,455,979.6056
H 4340
4340.470
1320
EaurHgamma.fts
4343.62
0.39
26.93 k/s
1.0167
93
Equivalent Width:
VR (If applicable):
3.022
N/A
5.3.4. Epsilon Aurigae Hydrogen delta

As with the H and H lines the H line must
also rely on neon line spectra for wavelength calibration. The problem is that there are no usable neon
lines in the area of the H line. It turns out that the
neon lamps have contamination in them. As a result
if a neon lamp, either the built-in neon calibrator or
an external neon ring, is used lines will be seen that
are not neon lines, but if calibrated can still be used.
These may be mercury and/or argon lines. If the
H line is moved to the far left of the computer window three lines at the far right can be used for the
wavelength calibration. The reference wavelengths
are 4149.42 , 4144.55, 4132.08 .
Observation
Double Date:
15/16 February 2012
Local Time:
20:28 MST
Universal Time:
03:07
Decimal UT Day:
16.14444
JD:
2,455,973.6299
Spectral Region:
H
4102
Reference Wavelength: 4101.740
Micrometer Setting:
1250
Fits File:
EaurHdelta.fts
Results
Wavelength Center:
4102.05
RV (km/s):
26.02 k/s
Air Mass:
1.0256
Equivalent Width:
1.285
VR (If applicable):
N/A
5.3.5. Epsilon Aurigae Hydrogen epsilon

Again the problem is that there are no usable
neon lines in the area of the H line. Again it turns
out that the neon lamp contamination provides some
useful reference lines. If a neon lamp is used, either
the built-in neon calibrator or an external neon ring,
the contamination will produce reference lines that
can be calibrated and used for the wavelength calibration.
Observation
Double Date:
Local Time:
Universal Time:
Decimal UT Day:
JD:
Spectral Region:
Micrometer Setting:
Fits File:
Results
Wavelength Center:
RV (km/s):
Air Mass:
Equivalent Width:
VR (If applicable):
12/13 February 2012

21:07 MST
04:07
13.12986
2,455,970.6715
H 3970
3970.07
1215
EaurHepsilon.fts
3970.05
0.336
25.39 k/s
1.0167
1.271
N/A
94
5.3.6. Epsilon Aurigae Sodium D Lines

The second spectral area of most interest, behind
the hydrogen alpha region is the sodium D line region. The two sodium D lines are very prominent.
Observation
Double Date:
Local Time:
Universal Time:
Decimal UT Day:
JD:
Spectral Region:
Micrometer Setting:
Fits File:
Results
Wavelength Center:
17/18 February 2012

19:30 MST
02:30
18.10417
2,455,975.6042
Na D Lines 5892
5889.950/5895.929
1782
Eaur NaD.fts
D1 5892.41
D2 5898.29
RV (km/s):
26.30 k/s
Air Mass:
1.0171
Equivalent Width:
D1 1.106
D2 0.821
VR (If applicable):
N/A
Epsilon Aurigae Na D Calibrated Line Profile
6. High-Resolution Data Summary

RJD
5,975.6178
5,975.6178
5,975.6178
Micro Region c
2010
Hv 6560.98
2010
Ha 6562.84
2010
Hr 6564.94
EW
0.01
-0.48
0.04
Note: Reduced Julian Date RJD = JD 2,450,000

Micro is the Lhires III Micrometer setting
Hv is the H violet emission line
Ha is the H absorption line
Hr is H the red emission line
c is line center wavelength
EW is Equivalent Width
7. Conclusions
It has been demonstrated that with even very
simple and inexpensive equipment and software an
observer can get started into the world of small telescope astronomical spectroscopy. By adding a highresolution spectrograph, e.g., the Lhires III with a
2400 l/mm grating, high quality and high-precision
spectroscopy can be done. Scientifically useful data
can be produced e.g., wavelength calibrated line profiles. Equivalent widths of various lines can be determined and plotted over time to see changes in the
star system. Knowing calibration lines in various
areas of the visible spectrum can help wavelength
calibration. The built-in neon calibrator of the Lhires
III helps, but some areas are void of neon lines.
Sometimes fluorescent lights can be used with the
mercury lines. For the sodium D lines an inexpensive
low-pressure sodium light can provide a very accurate wavelength calibration. A set of gas discharge
tubes for various elements can also enhance the bag
of tricks for wavelength calibration.
During the observation of the Balmer line regions of the spectrum of Epsilon Aurigae, other unknown lines were seen. Some of these are very strong
and worth further investigation. It may require the
use of a lower resolution spectrograph so more of the
spectrum can be viewed. This would help in identifying the unknown lines. Once identified highresolution spectroscopy could be done to examine
those lines in detail. Continued observations out-ofeclipse may help in unraveling the mystery of Epsilon Aurigae. Not all the action is limited to the main
eclipse. There is still a lot of action going on right
now.
5,976.5962 1466
4862.10
-0.43
8. Acknowledgements
5,979.6070 1320
4343.62
-3.02
I wish to thank the following people:
5,973.6464 1250
4102.05
-1.29
Dr. Douglas Walker, Estrella Community College,

for his encouragement.
5,970.6321 1215
3970.05
-1.27
NaD1 5892.41
NaD2 5898.29
-1.11
-0.82
5,975.6060 1782
5,975.6060 1782
Tom Field for providing the excellent, easy to learn

and use spectroscopy processing software RSpec.
95
Olivier Thizy, of Shelyak Instruments, for making

excellent spectroscopic equipment available and for
his enthusiasm, which is contagious.
Dr. Robert Stencel (Dr. Bob), University of Denver,
for his encouragement, inspiration and for being patient with all my questions. Our December 2010 observing session on Mt. Hopkins with the infrared
spectroscopy using the MIRAC4 mid-infrared spectrograph and 8-meter telescope was inspirational.
Dr. William F. Ketzeback, Apache Point Observatory, New Mexico, for his spectroscopic work on
Epsilon Aurigae and his kind assistance in answering
my questions about the sodium D line changes.
9. References
Hopkins, J.L. Stencel, R.E. (2009). Epsilon Aurigae
Hydrogen Alpha Emission Line Variation Horn
Dance. Proceedings for the 28th Proceedings for the
28th Annual Conference of the Society for Astronomical Sciences. (Warner et al., eds). Society for Astronomical Sciences, Rancho Cucamonga, CA.
Hopkins, J. L. (2012). Small Telescope Astronomical
Spectroscopy. HPOSoft. Phoenix, AZ.
http://www.hposoft.com/SB/SpectroBook.html
96
Using a Bespoke Spectrograph - Menke
Observations Using a
Bespoke Medium Resolution Fast Spectrograph
John Menke
22500 Old Hundred Rd
john@menkescieentific.com
Abstract
Designing and building a medium resolution (R = 3000) spectrograph was the relatively easy part. The really
challenging part is learning how to use it: learning the characteristics of the spectrograph, choosing the right kind
of astronomical problems, learning the best methods of taking data, and figuring out how to analyze the results. I
have used several observing projects to commission this system, including measuring the Doppler shifts in
several WUMa type stars. I will briefly describe the spectrograph but discuss in more detail the early experiences
of using it.
1. Introduction
2. DSS7 Experience
After four years of using a low resolution R=100

SBIG DSS7 spectrograph, I am still in awe of what it
can do. But in this paper I describe exploratory work
with a new bespoke (custom) Fast Spectrograph
(FS1) of R = 3000 some 30x greater than the DSS7.
The price of high resolution is low signal level!
A spectrometer resolution of R = 3000 implies
roughly a x3000 reduction in s/n at each pixel, so a
five minute exposure of say 16 mag now only reaches
about 7 mag., and probably only 6 mag with real optical elements!
RGB and similar filter combinations do allow
some ability to distinguish among the colors in the
light from an astronomical object. RGB filters have a
wavelength resolution of about 10 or soat about
5000A the bandwidth is about 500A, but already we
begin to sacrifice s/n for information. But for greater
resolution, we move to ever higher resolution spectroscopy: We always want more, more, more,
whether it is light gathering or resolution, ice cream,
or Scotch.
The design and construction of the FS1 is given
in detail in a paper on our website (Menke, 2012) and
in a brief poster at this conference. After completing
construction and first light with the new FS1, I
embarked on a series of observations designed to
explore the capability of the instrument and to help
me learn how to use it. Of course, identifying and
conducting useful research opportunities is the real
next step.
Some may believe that the low resolution of

the DSS7 prevents useful work. However, in past
years, I have used it to learn spectroscopy observational and analytic techniques. Observations included
performing coarse stellar classification from the
shape of the spectrum, observing a variety of objects
including planets, comets, planetary nebula, and novae. The most notable of these were observations of
nova U Scorpii as reported at SAS in 2010.
In previous papers and on our website, I have detailed some of the limitations of the DSS7 and effective ways to overcome them (Menke, 2011). One of
the limitations of the DSS7 is (as with most spectrometers) that it is designed for an f/10 optical system. However, my 18-inch Newtonian is f/3.5, thus
requiring a properly matched Barlow lens. This extends the optical train, and of course, enlarges the
stellar image on the slit, thus usually requiring use of
a larger slit with reduced resolution and very tight
tracking because there is no real time guiding in the
DSS7. The second limitation is, of course, the low
resolution. .
3. Fast Spectrometer FS1

After several years of using the DSS7-and by no
means exhausting its possibilities-I decided to obtain
a higher resolution unit in the R = 2-6000 range and
suitable for my f/3.5 Newtonian. A particular attraction of higher resolution is the possibility of measuring astronomical Doppler shifts. While some galactic
and nova Doppler shifts can be observed with the
DSS7, velocities of planets are in the 10s of km/s,
semidetached binaries are in the low 100s of km/s.
97
100 km/s corresponds to about a Doppler shift of

about 1:3000 or 2 at 6000 . To accomplish a 10%
precision implies the ability to measure to about 0.2
.
After searching the market, and finding that
nearly all units were designed for f/7-f/10 and had
other features I did not like, I decided to roll my
own (plus I like building things). The design goals
included:
Native f/3.5 design (but will easily accept slower
telescopes)
Physically compact, and well balanced
Inexpensive (approx $500)
Construction requiring only basic machining
skills
Adaptable, easy to modify
Accept almost any CCD camera (I use an SBIG
ST1603)
R = 3000 with possibility to go higher, and with
high stability
Built-in calibration (e.g., neon bulb or other)
Built-in guiding, including offset guiding (i.e.,
not using the target) using Starlight Express
Lodestar or SBIG SGi
Fig. 1 shows the FS1, while Fig. 2 shows the Littrow design (in which the light comes back off the
grating at the same nominal angle as incoming) that
allows for a single collimating lens that is more compact. The many design considerations are detailed in
our website paper (Menke, 2012a). Construction and
alignment was relatively straightforward using several red lasers to trace rays through the unit and neon
bulb and sodium flame sources to measure resolution
as described in the reference.
Fig. 2. Schematic Diagram of FS1
Once turned on the sky, it was obvious that this

was a very different instrument from the DSS7! The
integral real time guiding meant that keeping the star
on the slit, or even precisely in the middle of the slit,
was now easy-no custom software needed. The very
first night showed lots of solar lines from the moon,
splitting sodium D-lines, and even (incorrectly)
seemed to show the rotation of Jupiter. And it
showed that the expected mag 8 was indeed about the
practical limit with good S/N.
After the first night euphoria, the challenges
were to measure the characteristics of the FS1 (and
making any necessary modifications) while performing real observations that would help me learn how
to use it and how to interpret the data.
4. Studies of FS1 Performance
Fig. 1. Picture of FS1 Spectrograph
98
The first real world task was to determine how

well the tracking worked, not only for short data runs,
but also for long runs while the telescope shifted in
direction. In general, the guider kept the target within
the slit to a few tenths of a 8- pixel: guiding was a
cinch. Another question was whether there were special demands for the focus of the telescope. Due to a
fault in the design of the telescope, its focus shifts
when low in the sky, especially toward E/W; however, this did not affect the spectrograph more than
expected and defocus could be easily detected and
compensated.
There were many questions about wavelength

measurement: resolution, dispersion, stability, repeatability, etc. Measurements quickly using the ST1603
camera (ST8 size chip) showed that the wavelength
span is approximately 600 at ~0.4 per 9- pixel
dispersion. The remote stepping motor controlled
operation of the grating sets the coarse energy scale.
When setting the energy scale using the stepping motor drive, the repeatability is about 10 (corresponding to approximately 1 arcminute of grating rotation).
While the spectrometer has a resolution of about 2 ,
it is important that the stability exceed this so that
wavelength changes can be accurately followed.
Studies discussed below showed that the stability
over a data run is normally about 0.2 , thus allowing Doppler shift and related work more than 10
times finer than the resolution.
As I expected, it is essential to establish the
wavelength scale using either the internal calibrator
(usually Neon, fluorescent, or other source) or a spectral feature. One limitation of using the calibrator is
that it does require a special exposure, i.e., during
long data runs, the sequence must be interrupted either manually or via special software to obtain a calibration image. Calibrating on a feature of the spectrum being taken is often feasible, depending on the
wavelength range being used and the object being
observed. In some cases sky background can provide
a line; however, the most useful for many of my observations in the 6200-6800 band has been use of
the O2 absorption at 6283 which I used to measure
the wavelength stability of the spectrometer.
A particular concern was to measure the stability
of the wavelength calibration, especially over long
observing runs. The most obvious potential problem
is mechanical shifts are due to orientation changes as
the telescope tracks the star, temperature, etc. However, there a more subtle issue is the stability of the
target position in the slit.
Depending on the target, the slit of the spectrograph has two major purposes. If the target is a star
(point source) focused at the plane where the slit
would go, and if the star image is placed and held on
precisely that spot, then the spectrum will be formed
even though the slit is not present and the wavelength
scale will be set and stable. If the star is faint, then
the sky background light may contaminate the spectrum, but providing a slit reduces this. If the target
cannot be kept stable in position, the wavelength
scale will change but a slit at least limits this change.
Of course, a slit also allows a spectrum to be formed
from a diffuse target.
The point of this discussion is that good guiding
stability allows a stability of the wavelength calibration that can be substantially tighter than the resolution calculated from the slit width, and indeed, tighter
even than the FWHM of the target image would indicate.

To test wavelength stability, Fig. 3 shows the result of a long run of taking neon calibrator spectra
every ten minutes. The graph shows a variation of
nearly 5 , which of course, should not happen. I
traced this instability to a minor slop in the grating
holder that was cured with a 5-cent spring.
Fig. 3. Wavelength Stability Test (Neon)
Just as challenging as learning how the spectrograph operated was learning how to analyze and interpret the spectral. To accomplish these goals, I undertook two major observational programs as discussed below:
Measuring the Doppler shift from a pulsating
star. The star chosen was Beta Cephei, the namesake of this class. It is bright and hot: 3.3 mag
and B spectral class, so there is plenty of light.
The pulsations will cause the lines to swing back
and forth over a few tenths of Angstrom.
Measuring the Doppler shift for several semidetached eclipsing binaries (including WUMa
7.8mag) having Doppler shifts of 1-200 km/s
(about 2 ). With an 8-hr period, WUMa would
show splitting of the lines every four hours.
5. Data Processing Methods

I will describe the observing and data analysis
used for the Beta Cephei project, and later the modifications needed for the WUMa observations.
The FS1 operates with automatic guiding using
the internal beam splitter (pellicle) and Starlight
Xpress Lodestar guide camera that has an electronic
shutter system. In general, guide exposures can be 10
sec or longer if needed, however, this bright target
star allowed <1 sec exposures to avoid saturation.
99
The guider easily kept the target on the slit to within

the desired 0.1-0.2 pixel rms.
Spectral images were five minutes duration, for
Beta Cephei typically reaching about 1/3 of saturation, and were taken using the MaximDL in sequence
mode.
Each morning, after a night of automated data
collection, the archived images were downloaded
from the observatory (about 400 ft away). I use
Maxim and Visual Basic for the image manipulation
and reading the images, then Excel for analysis. This
allows me to easily tailor the spreadsheet to perform
the analysis I need and the graphic or other outputs I
desire. Plus, the spreadsheet itself records all the arithmetical steps so that I can later review and verify
what was done. As you will see later in the WUMa
analysis, this process also allows easy simulation of
data so that one can explore alternative analytic
schemes with known data.
and subtracts the associated background. The program also automatically averages a specified number
of successive images, and has the ability to reject
images with poor s/n while maintaining a uniform
cadence (timing) of the averaged images. The resulting averaged images (typically 10-30) in the batch
are written into a single numerical array. I call the
entire array the dataset for the night, while each averaged spectral image is a data subset.
Each array is copied into a spreadsheet for that
nights run where the array is displayed as n spectral image columns. By using a standardized spreadsheet, the data automatically display into the desired
graphs. The standard spreadsheet then repeats the
array to the right to allow an adjustable wavelength
calibration for each data subset. Pasting the data into
the sheet takes about 20 sec, and all the results are
available immediately. Fig 5 shows a typical result of
an observing run, with each data subset vertically
offset for easy viewing.
Fig. 5. Typical Spectra Display (Beta Cephei)
6. Beta Cephei Observations
Fig. 4. Spectrum Processing Control Window
As noted, image manipulation was done in

MaximDL. Due to the volume of data, I wrote a VB
program (Fig. 4) to control Maxim that would allow
processing scores of images at a time using simple
checkboxes to set the manipulations needed. In this
case, the spectral image manipulation included dark
and bias calibration and rotation of the image to align
the spectral band. The program measures the vertical
center of the spectral band in the image, averages the
value at each wavelength interval (horizontal pixel),
100
The BCep program data became vital data for

verifying FS1 operation and for developing the analytic process.
Beta Cephei pulsates in brightness with a period
of 4.57 hr and an amplitude of about 0.02 mag. The
pulsation is driven by internal processes that result in
expansion/contraction of the stars outer surface
whose Doppler shifts can be detected. The line of
sight velocity of the oscillation is only about 2030km/s, corresponding to a total wavelength shift of
0.4-0.6 , which is much smaller than the 2 resolution of the spectrometer. The brightness of the star
provides plenty of light, however, so that exposures
of only five minutes still have sufficient s/n to allow
precise determination of the wavelengths of the absorption lines. Thus, if the instrument calibration is
stable, or if each exposure can be calibrated, then this
instrument would be able to measure the Doppler

signal.
Fig. 6 shows the Telluric O2 line at 6283 from
a Beta Cephei observing run. One can see that the
lines are unchanged in wavelength during the run to
better than 0.2 . And as expected, the relative depth
of the 6183 line does vary during the run as the light
travels through differing thickness of atmosphere,
while the lines of astronomical sources do not. The
stability results show that long runs with only occasional calibrations with the Neon (or other) calibrator
are feasible.
Fig. 7. Beta Cephei Ha Oscillation
Fig. 6. Stability Verification (Beta Cephei)
So, is the Doppler oscillation of Beta Cephei

visible with this spectrograph? Fig 7 and Fig. 8 show
the 6563 H and 6683 H lines from the same data
set. The spectra are vertically offset for easy visibility, and represent 15 minute averages (period of the
pulsation is 4.5 hrs). Clearly, the H line shows a
wavelength shift during the run of about
0.6 . However, the HeII line at 6283 shows a
more complex behavior as both shape and apparent
wavelength change hour to hour. Examining the data
makes one suspect not only similar wavelength
changes as with H, but also the strengthening and
weakening of nearby absorption lines as the star pulsates. In fact, there are additional HeII lines nearby,
at 6678 and 6681, that appear to be varying in
strength as the star pulsates.
These changes in line shapes have been studied
since the 1950s as giving insights into the internal
process of this class of stars. What is gratifying is that
homemade equipment can allow its study by an amateur in 2012! If tracking these or similar variations
would be useful to a Pro, this Am would be happy to
track this for many nights!
Fig. 8. Beta Cephei He Oscillation
7. WUMa Observations
WUMa is a semi-detached eclipsing binary with
an 8.00 hr period. The variable nature of the star was
discovered in 1903 and its binary nature in 1919
when the first Doppler measurements were published.
However, the measurements are surprisingly difficult
to make accurately and have been refined ever since.
As the stars orbit one another, they almost totally eclipse one another. During the eclipses, first
one, then the second star is the only one seen. At such
times, the single star spectrum will show only the
system Doppler shift due to the Sun-WUMa relative
velocities and the line of sight movement of the front
star relative to the system barymetric center: Both of
these shifts are relatively small. At quadrature, however, one will see the spectra of the two stars, but
with each displaced in wavelength relative to the
eclipsed wavelengths. Observing this split sounds
easy but turns out to be not so, especially not with
marginal 2 resolution on a relatively faint star.
101
While the 7.7 mag of the star greatly restricts the

available signal, it proved to be a good test for learning how to handle poor s/n while seeking to measure
small effects! Although there are more absorption
lines available in bluer wavelength bands, for most of
this work I chose to continue in the 6100-6800 band
due to the availability of the Ha line from the source
and the 6183 telluric line.
Given the full 8 hr cycle of the WUMa, one can
combine about 6 five minute images without introducing more than 10% error in the Doppler velocity.
Obviously, it is desirable to perform the averaging so
that the averages are centered on the orbit quadratures, thus requiring accurate knowledge of the
ephemeris. Caution: many standard tables of
ephemeri are outdated: not only are some periods in
error, but also the periods change over years and decades. For most of the interesting binary stars, Bob
Nelson and AAVSO maintain a valuable updated
catalog on the web (Nelson). I also found it sometimes useful and reassuring to verify the timings by
performing the necessary photometry.
Even with 30 min averaging, the WUMa data
were rather noisy. In some runs (depending on the
quality of focus, night clarity, etc.), the deep Ha line
shows the splitting very clearly, as shown in Fig 9
along with a double Gaussian fitted to the points.
undisplaced. Then using the target spectrum, one

computes the correlation coefficient (CC) between
the two, thus measuring the relative similarity of the
two. If one then calculates the CC for a range of displacements (e.g., 15 ) between the two spectra
and graphs the result, the value of the CC vs. displacement wavelength will actually be an accurate
graph of the shape of the split spectral lines!
Professional astronomers have spent much effort
refining this general approach and developing similar
analytic methods; however, this one is easy to implement in Excel.
Fig. 9. WUMa Line Split

Fig. 10. Synthesized Spectra-Low Noise
Using only Ha lines wavelength measurements

from quadrature data averaged over three nights, calculations yielded WUMa star velocities of 260 and
77 km/sec. This result may be compared with other
measurements from the literature such as those of
Worden (Worden, 1977) that showed 243 and 131
km/s. The relatively impressive agreement may be
fortuitous, given the data noise levels and the early
stage of this analysis.
However, because most of my data were not as
good as these sets, there is need for a better approach
that takes advantage of all the lines in the spectrum
even though affected by noise. One answer is to obtain a reference spectrum of a similar class star with
102
It was fairly easy to prove that the method was

workable by generating a synthetic spectrum in Excel. The spectrum included Gaussian noise and Gaussian shaped lines of adjustable width and spacing.
Fig. 10 and 11 show low and high noise spectra, and
the CC results. This test shows the ability of the
method to pull data out of noise and the feasibility of
spreadsheet implementation.
Using such a reference, and using quadrature data

from a noisy WUMa observing session, Fig 12 shows
a typical set of promising CC results for the four
quadratures from early tests of this method.
With hundreds of data sets from many nights of
data covering the entire orbital period (not just quadratures), one is tempted to massage it forever, or until
one is exhausted or the computer crashes. While it is
clear that substantial refinement of the methods are
needed, the results already show that the general approach is correct and that this work is feasible for
amateur equipment and skills.
8. Conclusions and Implications for SAS

Members
Fig. 11. Synthesized Spectra-High Noise
Fig. 12. Example Correlation Coefficient Results
The reference spectrum must of course have the

same dispersion as the observed spectrum. Thus, I
have used as reference spectra my own measures of
stars or the sun; however, the reference spectrum that
was most effective was generated from the website of
R.O.Gray (Gray). Using the SPECTRUM program
from that site, one can manufacture a digital reference spectrum of desired line widths and dispersion.
This work has already shown that a shop-built

spectrometer can be used for many advanced observing projects requiring rather stringent instrument stability. The initial observing projects have been a challenge to analyze, but the results already show the
validity of the FS1 design and the promising work
that can be performed.
There are many additional targets for this spectrometer including the planets, nebulae, novae, and
following the shape changes in particular spectral
lines. As an example of the latter, Christian Buil has
led a major effort in this area, mostly following Be
stars (these are hot B-stars with hydrogen emission
features). Because we amateurs can spend as much
time on a subject as we wish, time series spectroscopy would appear to be a very fruitful area for us.
However, it will likely take a substantial expansion of
the Pro-Am programs to identify research areas suitable for amateurs with this type of equipment and this
level of skill.
When Dale Mais several years ago gave several
talks on his spectroscopic work, he wowed me. And
intimidated me with his encyclopedic knowledge of
astrophysics, of this and that spectral line, and of the
significance to astrophysics. But if one is not in a
university setting, then even with the Internet, how
does an amateureven one with some knowledge of
the physics of instruments and stars and research-ever get started? The Internet helps, but many papers
of interest are essentially unavailable to the individual, thus making self-education in advanced spectroscopy difficult. In contrast, there are many websites
and other resources that help one learn advanced
photometric techniques, develop Pro-Am contacts,
submit and publish data, etc.
That may be changing. As more of us move into
spectroscopy, there will be more cross-fertilization
among amateurs. And we are seeing an increase in
organized Pro-Am work witness AAVSO and espe-
103
cially the Lowell initiative that I hope will develop

programs to build advance skills in spectroscopy. In
the long run, these programs should help increase the
ease of establishing Pro-Am collaborations and encourage amateur driven research (i.e., non-Pro-Am)
as well.
So let me leave you with a dual challenge
To the Pros: Seek out amateurs who can contribute to spectroscopic astronomical research, both
in support of your particular study area or to
broader research objectives. It would seem that
amateurs might be particularly helpful in long
time series spectroscopy.
To the Ams: Seek out the new challenges of
spectroscopic research, and to learn how to do
itit is, after all, The Dark Side
9. References
Gray, R. O. (2012).
http://www1.appstate.edu/dept/physics/spectrum/
spectrum.html
Menke, J. L. (2011). Spectroscopy Techniques I and
II and others.
Menke, J. L. (2012). Construction of a Fast f/3.5
Medium Resolution Spectrometer.
http://www.Menkescientific.com
Nelson (2012). Updated O-C files for eclipsing binaries. http://www.aavso.org/bob-nelsons-o-c-files
Worden, S. P., Whelan, J. (1977). W Ursae Majoris:
The parameters of a Contact Binary. Mon. Not. R.
Ast. Soc. 163, 391-400.
104
Spectrosopy with the Lhires III - Gorodenski
High Resolution Spectroscopy for the Amateur:

Experiences with the LHIRES III Spectrograph
Stanley A. Gorodenski
Blue Hills Observatory
9440 E. Newtown Avenue
Dewey, Arizona 86327
stanlep@commspeed.net
Abstract
This paper describes the authors experience with the high resolution LHIRES III spectrograph and other equipment used. It discusses mechanical improvements made that may have increased the calibration accuracy of the
spectrograph, problems with guiding, and the need to take flat fields. It also briefly mentions the freeware software used and the types of computer programs written by the author to aid in the reduction and analysis of the
spectra. An assessment is made of the method for determining equivalent width the author described in the 2011
issue of the SAS News. It finishes by illustrating the ability to study binary stars, such as V1143 Cyg, with the
LHIRES III, and discusses some interesting results that were obtained on Epsilon Aurigae. The evolution of a
split line centered at around 5853 Angstroms is mentioned, as well as other aspects of the Sodium D Lines region, such as the constancy of separation between the two lines.
1. Introduction
2. Equipping the Observatory
I have a two story 16x16 observatory in

Dewey, Arizona, about 7 miles from Prescott, Arizona in a straight line, at an elevation of 5,140 ft. The
observatory and dome are my own designs and I constructed them completely by myself (except for the
concrete and block work). For a long time all I had in
it was a 12.5 Dall-Kirkham made by a machinist
friend in the 1970s (see http://users.commspeed.
net/stanlep/homepagens.html for more information
about the observatory and the Dall-Kirkham). Although an excellent telescope for visual observing, to
use it for serious science would have required some
major retrofitting. Instead of doing this I purchased a
16 Meade LX200R in 2006 (see SAS Newsletter
Vol 5, No 2 for more information about the Meade
and its installation in the observatory).
For many years I wanted to get into photometry.
However, building the observatory and dome was a
major project taking many years that prevented me
from doing this at the same time. With good intentions I had purchased a used Starlight-1 photometer
but it would have been too difficult to do photometry
with the Dall-Kirkham. I have always had an interest
in spectroscopy and the 2005 issue of the SAS Newsletter, Vol 3, No 3, had an ad for the newly introduced LHIRES III spectrograph as a knock down kit.
It had a good introductory price and so I purchased
and assembled it. Shortly after this I purchased the
Meade 16. Now I was ready to do some science.
To attach the spectrograph to the telescope, I attempted to use the Meade electronic focuser that
came with the Meade 16 but it had a number of
problems: 1) It was not strong enough to lift the
weight (about seventy pounds) of the instrument
package consistently at all times, 2) the 2 diameter
nose of the spectrograph could not be tightened down
enough in the draw tube to prevent wobble, and 3)
the draw tube wobbled inside its housing.
Figure 1.
I felt wobble could be a potential problem for the

following reason. Figure 1 is a diagram of a slit with
four incident light rays. If guiding is not perfect and
the target star drifts parallel to the slit (the 2nd and 4th
light ray from the front), all it will do is spread the
image of the spectrum over a larger area on the CCD
105
chip. Instead of it being a narrow bright line it will be

wider with a reduced intensity peak. This will result
in having to increase integration time to reach the
ideal saturation level. If the star drifts in declination
(i.e., in a direction perpendicular to the slit, the 3rd
light ray from the front), all it will do is result in light
loss, again resulting in having to increase integration
time. However, a wobble, I think, changes the angular direction of the light ray through the spectrograph
(the first light ray in Figure 1) and I think this could
cause problems with getting a high quality spectrum.
I do not know if this is a significant concern or if it
has any detrimental effect at all, but even if there is
no effect I like things to be rigidly attached. I replaced the Meade focuser with a v-grooved adapter I
machined (see Figure 2).
Figure 3.
Figure 1.
The spectrograph nose slides over it as shown in

Figure 3. This design completely eliminated all wobble (note the four stainless steel allen head screws.
These screws have tapered ends to match the vgrooves of the adapter). In Figure 3 is also shown the
ST-8XME imaging camera, and the Meade DSI I
guiding camera.
With this setup I had to use the telescope manual
focus to focus the target star for guiding. With increased use, manual focus (the long bottom knob at
the back of the OTA in Figure 3) started feeling
rough in one direction. Consequently, I felt I needed
an electronic focuser to eliminate having to use manual focus. This would avoid further wear and possible
future problems. However, I also wanted an electronic focuser so that I could lock the primary mirror
of the 16 telescope and thereby eliminate mirror
movement as a possible source of spectral error. For
those who are not familiar with this (poor?) design
feature of the Meade telescope, the mirror lock is the
upper knob at the back of the OTA in Figure 3.
106
Figure 4.
Some of the focusers I found on the internet that

could handle a heavy load were Crayford types.
These were fairly expensive but aside from the cost, I
do not like the Crayford design because not only is
there still the possibility of slippage even though they
are designed to handle a heavy load (the manufacturers of the Crayfords admit this), the intense pressure
the roller(s) has(have) to be under does not appeal to
me. I came across Starlight Instruments which makes
one that is relatively inexpensive and uses a helical
gear. A helical gear will not slip and it is not under
intense pressure. I bought the Feathertouch FTF
3015-B-A, and the add on developed by Starizona

which gives it the electronic focus capability. The
electronic focus developed by Starizona works great.
Figure 4 shows what the focuser looks like with the
instrument package attached.
The Starlight focuser is a real work of art, but I
discovered a number of problems. One is that the
spectrograph still wobbled in the draw tube eyepiece
holder, like it did in the Meade focuser, even though I
tightened down the spectrograph nose as much as I
dared. To correct this problem I removed the compression ring from the draw tube eyepiece holder and
replaced the thumb screws with tapered allen head
stainless steel screws. Then I drilled three tapered
holes near the base of the nose of the spectrograph.
These tapered holes are offset so that when the
screws are tightened the front plate of the spectrograph is pulled against the back surface of the focuser, thereby, eliminating all wobble. Figure 5
shows the modification I made to the nose of the
spectrograph.
happens I have to again reset the rotational orientation of the spectrograph.

Finally, I am not happy about weight of the Starlight focuser. The aluminum adapter I machined
(Figure 2) to replace the Meade electronic focuser
weighs only 2/3 of a lb. The Starlight focuser weighs
an additional four lbs, a 7-fold increase in weight. To
make matters even worse, the focuser adds another 3
or so to the back of the OTA and this increases the
moment of the instrument package. This means more
counterbalance weight is needed. The adapter I machined is the best solution to the wobble, but, of
course, it doesnt have electronic focus.
Figure 6.
Figure 5.
After eliminating this problem, I discovered that,

just as with the Meade focuser, the draw tube wobbles in its housing. However, this wobble can be
eliminated by tightening a (one) brass knob (the 2nd
brass knob from the back of the telescope in Figure
4). I thought I would have to loosen this knob each
time I focus, but according to Starlight I can tighten
the knob to the point where it just eliminates wobble
without damaging the focuser, or the Starizona focus
motor.
Another problem is the mechanism to lock the
focuser in a desired rotational position around the
optical axis. It is fixed in place by finger tightening
three large brass knobs (two of them can be seen in
Figure 4). However, because the brass knobs are nylon tipped, accidentally bumping the spectrograph
can cause the rotational setting to shift. When this
There are a few modifications I made which may

have improved the calibration accuracy of the target
spectrum. To calibrate the target spectrum, one needs
a calibration spectrum immediately before and after
the acquisition of the target spectrum. The calibration
spectrum in the LHIRES III is obtained from a neon
lamp built into the spectrograph. To get a calibration
spectrum requires manually positioning the neon
lamp over the slit by turning a knob at the front of the
spectrograph. This mechanical act can upset the mechanical stability of the spectrograph and result in a
shift of the calibration spectrum relative to the target
spectrum. To reduce this potential affect, some owners of the LHIRES have added a stepper motor to
turn the knob. I do not have the electronics capability
to do this, but I noticed that I could more easily move
the lamp from the back of the spectrograph with an
allen wrench. Figure 6 shows a permanently mounted
allen wrench for this purpose.
Another change I made was replacing the adapters that attach the ST-8 camera to the spectrograph. I
felt the ones that came with the unit were too light
duty with too many treaded parts and so I machined
my own. The machined adapters are the bare (nonanodized) aluminum parts in Figure 7.
107
of a sample of 51 spectra only 9 were off by more

0.05 A, and 33 were only off by no more than 0.03A,
the rest falling in between 0.03A and 0.05A.
3. Software
Figure 7.
Figure 7 shows the camera is attached to the

spectrograph with four tapered #10 brass screws. The
v-groove was offset a small amount so that when the
brass screws are tightened they push the male adapter
(the adapter attached to the spectrograph) against the
back of the female adapter (the adapter attached to
the camera), thus eliminating any possibility of
movement.
I do not know if the changes I made improved
calibration accuracy but I think they have. It has been
stated in the Spectro-L discussion group (SpectroL@yahoogroups.com) that one can at best expect no
more than a 0.1 of an Angstrom accuracy using the
internal neon lamp. I am getting better than this. Out
Figure 8.
108
I use CCDSOFT to image the spectrum and do

the dark subtraction. Currently, I am using an old
version of IRIS, ver 5.57, to reduce the spectra. This
includes flipping the spectra so the blue is on the left
side, making tilt and slant corrections, removing the
sky background, and optimizing the final spectrum
for input into VSPEC. These reductions in IRIS involve much more than the quick description just
given. I use Audela to get the Gaussian line center
estimates of the neon spectral lines (the calibration
lines). I use VSPEC to calibrate the target spectrum
(usually a multiline, i.e., at least three neon lines,
calibration using the neon line center estimates from
Audela), correcting the spectrum for instrumental
response, fine tuning the calibration with telluric
lines, removing the telluric lines after fine tuning, and
performing a heliocentric correction. I keep a record
in an EXCEL spreadsheet (Figure 8) of all my steps
in reducing a spectrum so that I can reproduce a reduction if needed. The final step in VSPEC is to create a TXT file of the fully processed, i.e., reduced,
spectrum. The .txt file is input into SPSS programs I
write for analysis, graphing, computing equivalent
widths, and other statistics. SPSS has very good pro-
gramming and statistical capabilities. I have been

using SPSS for over 25 years and so I am knowledgeable of its capabilities. IRIS and VSPEC are
freeware. SPSS is not but I purchased it at a good
price years ago.
4. Data Analysis
I use an SPSS computer program I wrote to estimate equivalent width (EW). The EW estimation
method I use is the one I developed as described in
the SAS Newsletter Vol 9, No 2. To briefly describe
it, the first step is to identify those parts of the spectrum that are obviously not noise (such as absorption
and emission lines). Once these are identified their
intensity values are set to system missing in the SPSS
program. What is left is considered to be noise for
estimating a continuum.
The second step is to select a range of noise values around the line (whose EW is being estimated) to
estimate a continuum with a polynomial least squares
regression model. Generally, I have found it is not a
good idea to use a large range on either side of a line.
Why it is not a good idea is shown in Figure 9.
Figure 9.
Figure 9 are plots of what were deemed to be

noise for three spectra. The spectra were taken within
a half hour of each other, all had the same integration
time and camera cooling, and each one was integrated on a different area of the CCD chip. To estimate the continuum for the Sodium D lines (the vacant area between the 2nd and 3rd lines from the left)
one could use the entire range of the spectrum. However, upon a study of the graph it can be seen minor
differences exist in the noise levels prior to the first
line, and the noise values are quite variable between
the third and fourth lines (from the left) in intensity
and shape. Also, the group of noise values at the exLine
5853A
Na D1
Na D2
5978A
yyyy
2010
2010
2010
2010
m
3
3
3
3
d
4
4
4
4
Time
03:36:25
03:36:25
03:36:25
03:36:25
Min
0.0422
1.2957
1.2032
0.0679
EW
0.0560
1.3247
1.2331
0.0836
treme right end of the graph could potentially have a

large regression model leveraging effect and, thereby,
result in a poor estimate of a Sodium D lines continuum. Consequently, for estimating a continuum for
the Sodium D lines in this example, I would use the
noise values between the 2nd and 3rd lines. By extension, it should be apparent that, generally, it would
not be a good idea to use the same range of data to
estimate one continuum for two or more lines (one
exception being the Sodium D lines).
Once a continuum is estimated I determine the
beginning and end points of a line (from an overlay
graph of the estimated continuum and actual data
points) and then run the complete SPSS program to
get an output of equivalent widths and 95% confidence limits.
As can be seen from an example output titled
Equivalent Width Data for Epsilon Aurigae, my
SPSS program can estimate the EW for up to four
lines (the 5853A line, the two Na D lines, and the
5978A line) at a time. I could modify my program to
do more lines if the need ever arises.
I like this method for computing EW because it
gives me greater flexibility than exists in some spectroscopic software, it has greater statistical validity,
and it can be better defended than some of the other
simpler methods. However, all these presumed advantages may be swamped out by a small degree of
subjectivity in telluric line removal, the somewhat
subjective determination of the beginning and end of
a line, other random effects near a line that can affect
continuum estimation, improper (or no) flat fielding,
the resolution of the LHIRES, and the fact that the
continuum estimate, at the non-professional level, is
usually somewhat a guess to begin with. It may be
some of the simpler and quicker methods to compute
equivalent width may produce equally good estimates. This is something I have not looked into.
I have looked at many spectra for the same star
taken on the same night on different parts of the CCD
chip (with equal integration times and cooling temperatures) and for the most part the shapes of the
spectral profiles agree very close with one another.
At times there are some minor disagreements. I am
not sure what the causes of these are, but it could
reflect the need to take flat fields. Currently, I have
the slit on my spectrograph set at 22 microns, which
is very narrow. Because of this small size I was not
Max
0.0698
1.3536
1.2631
0.0993
Start End
416
432
684
715
729
761
1364 1381
/pix
0.132
0.132
0.132
0.132
S.D.
2325.54
2418.27
2418.27
2291.98
All
360
336
336
84
Miss
152
176
176
46
Table I. Equivalent width data for epsilon Aurigae. The Min/Max values are the lower and upper limits for EW.
Start/End are the line start/end values. All gives the number of all cases while Miss gives the number of missing
cases.
109
able to get enough light through the spectrograph, in

a reasonable integration time, for a flat with two 90
watt Halogen lamps illuminating a white surface.
However, even without a flat correction nearly all
spectra taken on the same night are generally in very
good agreement with each other and I am getting
good results. I will continue to work on getting a flat
but it seems there is some contention, at least among
some of the non-professional spectroscopists, as to
whether a flat is necessary and whether a flat could
introduce unwanted effects by subtracting an effect
that is not present under the actual conditions of getting a star spectrum.
ment amazed me, because, as I said, I was only

reaching about 5% of saturation.
Figure 11 is the classic P Cygni profile of P Cyni
itself. The noise in this profile is unbelievably low. I
think this is partly the result of a scale effect caused
by the emission line being over four times the magnitude of the continuum level.
Figure 11.
Figure 10.
A big problem is guiding on faint stars, but even

on brighter stars it is hard for me to guide accurately
enough to get a tight bright line. I have tried a number guiding software that support the Meade DSI I
camera. Meades Envisage software is unstable and
frequently crashes. MaximDL worked good but I
could not justify spending $565 for a package that
will only be used for guiding. I tried Astroart but I
was unable to center the guide star (which is also the
target star) over the slit. I am currently using PHD
and it seems to work okay but on a faint star I frequently get a message saying the signal to noise is
low. A lot of times when this happens I lose guiding.
I am restricted to guiding software that support the
Meade DSI cameras which is a problem. At some
time in the future I will have to break down and get
another, and better, guiding camera.
Now I would like to show some of the things that
can be done with the LHIRES. Figure 10 are spectra I
took of the binary V1143 Cyg. Because it is a 5.9
magnitude star, I had to integrate an hour with the 22
micron slit width, but even with an hours integration
time the peak intensity of the spectrum was still only
about 5% of saturation. I was not happy with the
spectra because I could see more noise in their profiles than I am used to seeing on much brighter stars
like Epsilon Aurigae. However, Dr. Dirk Terrell analyzed them, came up with a velocity estimate of 131
km/sec, and said this shows I can do pretty good
work with the system I have. This favorable assess-
110
Figures 12 and 13 are two graphs in press for the

special edition the JAAVSO on the Epsilon Aurigae
campaign.
In Figure 12, a graph of equivalent widths of the
Na D1 line, a number of things can be seen. First, the
equivalent widths are at a low point at mid-eclipse.
This is expected because there is less disk material at
this point that contributes to the absorption line. After
mid-eclipse the EWs gradually increase until a high
point is reached at 3rd contact. This again is expected.
After 3rd contact there is a decline, and this is expected since the disk is now on its way to clearing the
primary star. Far after 4th contact, when the primary
star should be clear of the disk, the EWs are
still significantly non-zero. This indicates the disk
material does not end at forth contact, but continues
significantly further.
Figure 12.
The low point prior to March 28, 2010 is hypothesized to be either the result of a void in the disk,
or a ring structure. Evidence for the latter has been
seen by some astronomers (Leadbeater et al., 2010,
Seebode et al., 2011). The sharp downturn in radial
velocities in Figure 13 also confirms this interpretation.
Figure 13.
The shape of the radial velocity (Figure 13)

curve from about May 6, 2010 to about November 1,
2010 is what is expected of a disk with a central
clearing around a primary object. The predicted date
of mid-eclipse was August 4, 2010, although this will
probably change as more data comes in. Based on the
two Sodium D lines and taking into account that the
Epsilon Aurigae system is moving toward Earth at
about 2.5 km/sec, I came up with a mid-eclipse date
of August 17, 2010. Finally, the continuation of a
significant non-zero radial velocity after 4th contact
supports the equivalent width interpretation that the
disk continues well after 4th contact.
Now I want to show some stuff that is not in the

JAAVSO paper. I came up with four different methods to estimate line centers for the D2 and D1 lines.
There were two categories. One was based on the
beginning and end points of the line that were used to
compute equivalent width. The other was a visual
estimate in VSPEC of the beginning and end of a
line. Within each category there were two line estimation methods: the wavelength center of the line
barycenter, and the wavelength center of the VSPEC
Gaussian fit to a line. Figure 16 contains four graphs,
one for each of the four different methods of estimating line center, of the wavelength difference between
the D1 and D2 lines vs. time.
Based on my understanding of quantum mechanics (which is very little), I would expect the separation between the two Sodium D lines to be constant.
However, it can be seen that a second order polynomial model fits both Gaussian estimates, and a linear
trend fits the visual-barycenter estimate. This indicates the separation between D2 and D1 is not constant but changes over time. However, a statistically
significant curve could not be fit to the EWbarycenter estimate. This indicates the D lines separation is constant. Maybe some kind of line measurement bias is present in the other three estimates
that result in a false change over time. The
Figure 14.
Figure 16.
Line
Na D2
Na D1
Figure 15.
One thing I observed during my participation in

the Epsilon Aurigae campaign was a split line at
about 5853 Angstroms that can be seen in Figure 14.
Figure 15 is an equivalent width graph of the 5853A
line. From Figure 15 the split line event is estimated
to start on January 27, 2011 and end on May 8, 2011
which gives a duration of 101 days.
Method
EW Barycenter
EW Gaussian
Visual Barycenter
Visual Gaussian
Barycenter
EW Gaussian
Visual Barycenter
Visual Gaussian
Mean
5889.79
5889.82
5889.80
5889.82
5895.77
5895.79
5895.78
5895.79
Std Dev
0.3152
0.3033
0.3028
0.3032
0.3093
0.3084
0.3072
0.3085
Table II. Standard deviations of the line estimation

methods.
EW-barycenter estimation method could be unbiased

and be giving the correct relation, which is no relation, but it has the highest variance among all the
methods. This higher variability could mean the EW-
111
barycenter estimate was just too variable to detect a

small linear relation or for a bias to be seen.
I wanted to show this to demonstrate the care one
has to exercise in interpreting results. More than
likely the change over time shown in three of the
methods is the result of a bias in line center estimation, but wouldnt it be something if I, with an
LHIRES III spectrograph and a relatively inexpensive Meade telescope, discovered a major flaw in
Quantum Mechanics? On the 0ther hand, maybe
there is a quantum mechanical explanation.
5. Conclusion
In conclusion, I have had a very good experience
using the LHIRES III high resolution spectrograph. I
think this may in part be because the seeing is pretty
good at my observatory resulting in low spectral
noise, I am in a desert area and so I do not have a
major problem with telluric lines, my introduction to
spectrometry was on a bright star (Epsilon Aurigae),
because my introduction was on a bright star I was
able to set the slit width near the Nyquist minimum to
give me maximal resolution without adversely affecting integration time, and I have a good imaging camera, a Class I ST-8XME. Generally, it seems deviations from ideal seeing, ideal mechanical aspects, and
ideal guiding has been very forgiving. If one reads all
the very technical advice and very technical solutions
to problems given in the Spectro-L discussion group
one could quickly become overwhelmed and conclude spectroscopy is not for them. Maybe the best
advice is to get good equipment (especially the imaging camera), just do it, and, at least at first, concentrate on the brighter stars.
6. References
Leadbeater, Robin; Stencel, Robert E. (2010). Structure of the Disc of Epsilon Aurigae: Spectroscopic
Observations of neutral Potassium during Eclipse
Ingress. arXiv:1003.3617v2 [astro-ph.SR].
Seebode, et al. (2011) Ring-like Structures around
Epsilon Aurigae Companion. 217th AAS Meeting,
Poster Paper 257.08.
112
ELCPS Spectrophotometry Beaver et al.
Extremely Low-Cost Point-Source Spectrophotometry

(ELCPSS)
John Beaver
University of Wisconsin Fox Valley
1478 Midway Rd.
Menasha, WI 54952
john.beaver@uwc.edu
Charles Conger
University of Wisconsin Plattville
1478 Midway Rd
Menasha, WI 54952
congerc@uwplatt.edu
Abstract
We describe preliminary tests of a low-cost method for obtaining, reducing and calibrating stellar spectra. Instead
of a post-focus spectrometer, we use an inexpensive, low-dispersion objective grating made by printing onto
acetate with a laser printer, coupled with a low-end digital SLR camera as the detector. Although originally intended for educational use, we consider the possibility of using this technique to obtain accurately-transformed B
and V magnitudes of stars without the need of an expensive photometric filter and filter wheel system. The results of two nights of observations of several bright stars are presented. Future plans are presented for more
tests using a wider range of gratings, telescopes and detectors, and more advanced observing techniques that
are likely to produce higher-quality data. But we show that even the crude observing techniques used for the test
data can produce calibrated B and V magnitudes. General methods for reducing and calibrating the data are
described, and some of the educational uses for ECLPSS are also considered.
1. Introduction
The usual method for taking spectra of stars and
other astronomical objects is to use a telescope to
gather and focus the light, and then allow the light to
pass into a spectrograph attached to the eyepiece end
of the telescope. The focused light passes through a
slit, where it is dispersed by a diffraction grating into
different wavelengths and then detected by a CCD
array digital detector. A spectrograph such as this,
even a relatively low-cost one, can be quite expensive
for an individual amateur astronomer or even a high
school or small-college astronomy program.
Extremely Low-Cost, Point-Source Spectrophotometry (ELCPSS) is an attempt to bring stellar spectrophotometry to those who otherwise would have
little opportunity. We do this by using a simple lowdispersion diffraction grating, produced for almost
nothing by printing a pattern of lines onto acetate
with an ordinary laser printer. This grating is placed
over the objective of a typical small amateur telescope and an ordinary low-end commercial digital
SLR camera (DSLR) is used as the detector. Thus the
entire telescope becomes the spectrograph, and all of
the equipment is of the sort a typical small college,
high school or amateur astronomer may already own

(Beaver and Robert, 2011a).
These slitless spectra produced by ELCPSS are
of rather low resolution, since the spectral resolution
is limited by the observed size of the star image, and
this is degraded by the acetate base of the objective
grating itself. The point of ELCPSS is not to produce
research-quality spectra in and of themselves. Rather,
we hope to inexpensively obtain low-resolution spectra that are nonetheless fully wavelength and fluxcalibrated, and using much of the same procedure as
at any major observatory. This means ELCPSS has a
fairly clear educational use. But we also show that
ELCPSS may be useful as a roundabout but inexpensive method for obtaining calibrated BV (and perhaps
BVR) stellar magnitudes.
Here we present the results of two nights of
ELCPSS data. We demonstrate proof of concept regarding using it as a back door to stellar broadband
photometry, and we consider its potential as an educational tool. We also lay out a strategy for making
the rather complex data reduction and calibration
process more accessible to the uninitiated. Future
work to be done is outlined, and collaborations are
solicited.
113
2. Useful Numerical Relations for

Objective-Grating Spectroscopy
For objective-grating spectroscopy, the scale of
the spectrum on the detector (in angstroms per pixel,
for example) is set only by the line spacing of the
grating, the focal length of the telescope and the
physical size of the detector pixels. These data, combined with the number of pixels on the detector in the
dispersion direction, then fix the possible range of
wavelengths that can be imaged for a particular combination of grating, telescope and detector. We define
the following quantities, with units chosen to give
typical values near unity:
d
m
p
f
grating spacing, in hundredths of an inch.

order number of the spectrum.
pixel size on the detector, in microns.
focal length of the telescope, in meters.
With these definitions, a given number, N, of

pixels in the dispersion direction, corresponds to a
given range of wavelength in Angstroms, , given
by:
tector. For the data presented here, we estimate the

size of the direct star image through the grating to
have been about 10 arcsecond, and this corresponds
to a spectral resolution element of about 60 Angstrom
for our grating (d = 0.5).
3. Observations
The observations presented here were carried out
on two nights, 2010 June 7 (Night 1) and 2010 October 1 (Night 2). All observations were made from
Appleton, Wisconsin, with moderately light-polluted
skies (naked-eye limit m = 4.5). Both nights were
judged to be photometric. The observing log for the
observations is given in Table I. For reasons discussed below, multiple short exposures were combined to give the total exposure times listed in Table
I. The individual exposures were all between 1 and
30 seconds. Preliminary results from Night 1 were
previously published (Beaver and Robert, 2011a;
Beaver and Robert, 2011b), but errors in the reduction and calibration process have been corrected for
the results presented here.
Star
Night
Exp. (s)
(1)
Lyr
1.49
12
As an example, for the instrumentation used in

this paper the pixel size was 7.88 micron, the grating
was 200 lines/inch (d = 0.5) and the focal length of
the telescope was 2.0 m. And so for this case
/N = 5.0 Angstroms/pixel for the first order spectrum (m = 1). As another example, to get everything
from the direct image of the star (m = 0) to 7000
Angstroms in the first order onto the detector, solving
Equation 1 for N shows that about 1400 pixels are
needed for this combination of grating, detector and
telescope.
The lack of a slit ties the spectral resolution directly to the angular resolution. For a given angular
resolution element, R, in arcseconds, the corresponding wavelength resolution element, R, in Angstroms,
is given by:
UMa
1.10
20
Dra
1.22
68
Lyr
1.09
20
Lyr
1.17
22
58 Aql
1.44
210
Cyg
1.10
60
Her
1.67
105
=2.540
R =12.3
dpN
mF
d R
m
(2)
Since R is set by the seeing, tracking, and especially the degrading of the image by the acetate grating, this means that one would want the smallest
value of d possible, so as to maximize the wavelength
resolution (i.e. minimize the size of the resolution
element), even if this means the data is oversampled.
But of course we also want d chosen such that the
combination of telescope focal length and detector
size puts the desired wavelength range onto the de-
114
Table I. The observing log for the stars observed for this
paper.
For a post-focus slit spectrograph the spectrum

automatically appears at the same location on the
detector each time. But with an objective-grating, the
spectrum could in principal appear anywhere on the
detector, depending on the pointing of the telescope.
And so if one is to be able to relate a given position
in the observed spectrum to wavelength, either there
must be a reference for wavelength in the image itself
or there must be a way to exactly repeat the positioning of the star image in the telescope (and thus the
location of the spectrum) from frame to frame.
Thus there are two obvious approaches. For what
we call zero-order referenced (ZOR) spectra, we use
the zero-order spectrum (the same as the direct image
of the star) as a wavelength reference for each frame
separately. Thus, for ZOR spectra, both the zero and

first orders must be present on each frame, but there
is no need to ensure that each exposure is taken with
the same positioning of the object. Since the zeroorder, direct image of the star is much more intense
than the first-order spectrum, short exposures must be
used so as to not overly saturate the zero-order spectrum. With proper guiding, however, one could in
principal take a series of exposures for each object
long exposures to record the first-order spectrum, and
short exposures to accurately determine the location
of the zero-order spectrum (although one would still
need to worry about scattered light and blooming
from the saturated pixels).
The other approach, zero-order guided (ZOG), is
to guide directly on the zero-order spectrum during
the exposure, perhaps with an off-axis guider or by
using the separate guide chip included with some
astronomical CCD cameras. The ZOG technique
would seem to have a big advantage over ZOR spectra, as the much brighter zero-order spectrum is out
of the frame. But ZOG gives fewer choices of dispersion, given a particular off-axis guiding arrangement.
For example, if one is using a fixed-prism off-axis
guider, then the correct grating to put the desired
maximum wavelength at the far edge of the detector
will not necessarily put the desired minimum wavelength at the near edge. Put another way, a choice of
grating that places the desired wavelength range
across the detector may not at the same time put the
direct image of the star in the guiding field.
For the spectra presented here, we have used the
ZOR technique, in part to first test the efficacy of the
crudest and lowest-cost observing techniques. A Celestron 8 (with uncoated corrector plate) was used
with a 200 line/inch objective grating and a Nikon
D40 SLR shooting in RAW mode (with noise reduction on) at ISO 1600. Exposure times were limited by
the tracking of the (unguided) telescope and the fact
that we could not too-badly saturate the zero-order
spectra, as they are needed as a reference for the
wavelength calibration.
We took the standard array of basic CCD reduction frames, including bias, dark and flat field. Flat
fielding was accomplished by means of observations
of a white screen illuminated by an old slide projector. We also used twilight sky flats to produce a final
illumination correction. These frames were all taken
with the objective grating in place. Since the sources
of light for the flat field frames were diffuse over the
entire field, so too was the spectrum produced, the
zero and first order spectra mixing over the entirety
of the image so as to produce images very much as if
the grating were not present at all. Dark frames were
taken implicitly for the longer exposures, as it is a
property of the noise reduction feature of this par-
ticular DSLR to take a dark frame immediately after

the exposure and to automatically subtract it. This
feature can be turned off, but it was left on for these
data. The built-in noise reduction also performs some
median filtering, and this would have been a concern
except that the spectral data is grossly oversampled in
wavelength, and so this could not have removed
real features of the spectra.
In addition to these basic CCD imaging calibration frames, we have taken others related directly to
the calibration of the spectra. For wavelength calibration we observed a low-pressure sodium light, occluded by a screen except for a small pinhole, placed
about 30 m from the telescope. A similar arrangement was used to make an artificial point source of a
300-W, 3100-K color-temperature quartz lamp. This
was needed to flatten the spectra somewhat in order
to better accomplish the final flux calibration.
4. Data Reduction and Calibration

The RAW Nikon images were extracted to FITS
format (Wells et al., 1981) with the low-cost proprietary
software
package
ImageTOOLSca
(http://arnholm.org/astro/software/ImageTOOLSca/).
This program effectively converts many consumer
digital RAW formats to FITS, and writes a proper
header with standard fields such as exposure time and
date and time of observation. The data were extracted
summing the three separate RGB mask images to
single-band, 16-bit FITS images.
We used IRAF (Tody et al., 1986; Tody et al.,
1993) to reduce and calibrate the data. The basic image reductions, incorporating bad-pixel repair, bias
subtraction, flat-fielding (using the white-screen
flats), and illumination correction (using twilight
flats), were carried out in the usual way with the
IRAF task CCDPROC.
The 1-d spectra were extracted from each 2-d
image using the APEXTRACT package of IRAF,
which fits a function to a trace of the spectrum on the
2-d image. Care was taken to use an aperture large
enough to account for differences in tracking and
focus, except for the extractions of the double star
Cyg, which required a smaller extraction aperture
than we would have liked.
A dispersion in Angstrom/pixel was determined
by measuring the separation in pixels of the zero and
first order spectrum of the sole line in the observed
sodium spectrum. This dispersion was then applied to
the extracted 1-d spectra by directly editing the image
headers. The zero-point for the wavelength calibration was determined separately for each spectrum
using the interactive IRAF task SPLOT. Only a simple linear dispersion relation was applied, since for
115
ZOR spectra nothing else would be meaningful given

the fact that the individual spectra were at different
locations on the detector.
A multi-order cubic spline was fit to the quartz
lamp spectrum and all of the stellar spectra were divided by this spectrum in order to remove mediumscale variations before attempting flux calibration.
This greatly facilitates the process of flux calibration
because the instrumentation introduces medium-scale
features in the spectrum due to, for example, the
RGB pixel mask and interference coatings on the
intervening optics (for example, the IR and UV
blocking filter Nikon places over the CCD).
Flux calibration was performed with the IRAF
tasks STANDARD, SENSFUNC and CALBRATE,
using Lyr as the sole spectrophotometric standard.
For the second night, only the second observation of
Lyr was used for the calibration, as the first observation was inconsistent in overall scaling. This was
the first observation of the night and the observing
log noted that dewing may have been a problem for
that observation (and this is consistent with the
measured flux level being too low for that observation). No extinction curve was determined from the
data; instead the mean extinction curve for KPNO
was applied.
5. Results
We compare our observed, calibrated spectra to
catalog spectra from the STScI HST Calibration Database Systems (CDBS) Stellar Spectral Atlases
(STScI, 2012a). Figure 1 shows, plotted along with a
Kurucz model-atmosphere Lyr spectrum (Colina,
1995), the three Lyr observations and the observations of Uma, also an A0V star. Figure 2 shows the
observed spectrum of 58 Aql, along with a comparison spectrum of that star from the Bruzual-PerssonGunn-Stryker (BPGS) Atlas of the CDBS. Figure 3
shows the observed spectra of Cyg A and B, along
with CDBS spectra of stars of similar spectral type.
Figure 4 shows the observed spectrum of Her, along
with a spectrum of that star from the Wisconsin
Halfwave Polarimeter (HPOL), retrieved from
http://archive.stsci.edu, and Figure 5 shows the observed spectrum of Dra, compared to a BPGS catalog spectrum of a star of similar spectral type. The
CDBS spectra were all obtained via the STScI Specview software package (STScI, 2012b). Finally, Figure 6 shows the flux-calibrated spectrum of the observed quartz lamp, compared to a 2900 K blackbody, in rough agreement with the rated color temperature of 3100 K for the bulb used.
116
Figure 1: A Kurucz model atmosphere of Lyr, along

with our three Lyr observations and the A0V star
UMa.
Figure 2: Our observed spectrum of 58 Aql, along with a

comparison spectrum of that star from the BPGS atlas.
Figure 3: Our observed spectra of Cyg A and B, along

with CDBS spectra of stars of similar spectral type.
Figure 4: Our observed spectrum of Her along with a

spectrum of that star from the Wisconsin Halfwave Polarimeter.
noticeable error in absolute calibration, appearing too

faint by the equivalent of a few tenths of a magnitude. This is not surprising as the observing log for
that particular observation reports dewing problem? The rest of the observed spectra agree well
with the comparison spectra, with the exception of
Cyg B, which shows a noticeable error in relative
flux calibration, appearing too red. We suspect this
may be due to the difficulty in extracting this spectrum without contamination from its much brighter
close neighbor, Cyg A.
With the exception of Lyr, the comparison
spectra presented in these graphs have all had a scale
factor applied to best match the observed spectra.
This is because the comparison spectra we found had
only relative flux calibration, and in some cases we
could only find comparison spectra of the same spectral class (but not the same star). We will see in the
next section, however, that we have good reason to
believe the absolute flux calibrations of the spectra
are reasonable, with the exceptions of Cyg B and
the first Night-2 Lyr spectrum, as previously noted.
6. ELCPSS as a Back-Door to BroadBand Photometry
Figure 5: Our observed spectrum of Dra, compared to

a BPGS catalog spectrum of a star of similar spectral
type.
Figure 6: The flux-calibrated spectrum of the observed

quartz lamp, compared to a 2900 K blackbody.
We note that Lyr was used as the spectrophotometric standard for both nights, and so the agreement between these spectra and the catalog spectrum
is unremarkable. The first Lyr spectrum from Night
1, however, was not used for the flux calibration, and
so its agreement (or lack thereof) is meaningful. This
spectrum shows good relative flux calibration, but a
Broad-band B and V filter photometry takes

complex spectral information over a range of nearly
3000 Angstrom, and compresses it into just two
numbers. The success of broad-band filter photometry depends on the spectral response of the instrument-detector system, and the spectrum of the observed source, being similar enough to that which set
up the standard system. If not, then a one-to-one correspondence between the instrumental and standard
magnitudes may not be obtainable.
Even a low-resolution spectrum contains far
more information than a set of broad-band photometric measurements. In particular, the problems of flux
calibration and extinction correction for spectra are in
many ways simpler than the corresponding problems
of transforming instrumental broad-band magnitudes
to standard magnitudes.
Thus once a spectrum is accurately wavelength
and flux calibrated, a standard magnitude can be extracted simply by convolving the appropriate magnitude bandpass function with the spectrum.
We have done this for the spectra here presented
using the IRAF task SBANDS, and the filter profiles
for B and V from Ausienis and Straiys (1969).
These profiles are normalized, and so we used the
filter extractions of our standard star ( Lyr) to set the
zero points for the magnitudes. Table II shows the
results of these extracted magnitudes, as well as the
values from SIMBAD for comparison. As can be
117
seen, even with only one observation of a single

standard star and the use of a mean extinction curve
for a different site, the values are all to within 0.2
magnitude in V and 0.13 in B-V, except for Cyg B
and the first Lyr spectrum from Night 2, as already
noted.
Star
V'
B-V
B'-V'
Lyr
1.09
0.27
0.03
0.01
0.00
58 Aql
1.44
5.45
5.63
0.18
0.08
Her
1.67
2.71
2.80
0.69
0.63
Dra
1.22
2.23
2.23
1.65
1.52
UMa
1.10
2.38
2.43
0.01
0.04
Cyg A
1.10
3.20
3.09
1.19
1.09
Cyg B
1.10
5.23
5.09
0.16
-0.06
Table II. The extracted V and B-V magnitudes for the

program stars, compared to the values from SIMBAD
(the primed columns).
In principal, one should be able to do much better by adopting a hybrid technique. Instead of treating
these extracted magnitudes as the final values, think
of them instead as raw instrumental magnitudes for a
program of broad-band photometry that includes observations of broad-band photometric standards as
well as the spectrophotometric standards used for the
spectroscopic flux calibrations.
Even using one spectrophotometric standard star
and a generic mean extinction curve to establish the
calibration, the extracted magnitudes should still be
closer to the standard system than many detectorfilter-instrument combinations in use. Thus one could
in principal use ELCPSS to observe a regular program of photometric standards in order to determine
accurate standard magnitudes from these almostcalibrated extractions. The extracted magnitudes
should transform very smoothly to the standard system since they are almost there to begin with.
This would also solve the problem of a relative
lack of bright spectrophotometric standards (as compared to broad-band photometric standards), thus
allowing ELCPSS to be useful for differential photometry.
7. Educational Uses of ECLPSS

Both the process of carrying out ELCPSS and
the calibrated spectra that result can be of educational
use. We choose to illustrate this with an example.
Although we present this as a hypothetical case, we
are attempting to actually carry out much of this at
the University of Wisconsin Fox Valley, a small
two-year campus with a primary mission of transfer
118
to one of the four-year campuses of the University of

Wisconsin system.
A beginning astronomy major starts a project to
record and calibrate ECLPSS spectra, with the goal
of compiling an atlas of stellar spectra. Some of the
spectral types have already been recorded by a student from the previous year. The current student adds
to the previous work, getting spectra of multiple stars
for some spectral types. In the process, she has to
plan and execute an observing run, and then reduce
and calibrate the data with IRAF using a preconfigured virtual machine environment set up especially for this purpose (see Section 8). Although all of
this is done with modest equipment, the basic observing and data reduction techniques and the software
used are all likely to be similar to what she will eventually use in graduate school.
Although the data themselves may not contribute
directly to basic research, the results still have a purpose beyond that of a mere student exercise. This is
because the spectra are to be used for introducing the
concept of spectral classification to the beginning
Survey of Astronomy courses for non-science majors.
Instead of the students learning about stellar spectra
from examples in a textbook or on the Internet, obtained by a process to which the students have no
direct connection, they are presented with data obtained locally by one or more of their peers.
In fact, the astronomy major makes a presentation to the class, explaining how the spectra were
taken, and the actual telescope used is set up for the
students. The Survey of Astronomy students gain a
learning experience that has a visible and immediate
connection to their experience, and the astronomy
major is motivated by contributing to their learning
with data she gathered herself. This is service learning in the true sense of the term.
8. Facilitating the Reduction and

Calibration Process
Since the goal of ELCPSS is to make spectrophotometry more accessible, the complex process of
reducing and calibrating the spectra is a concern. To
address this, we have developed a virtual machine
(VM) with an open-source operating system (Ubuntu), with all of the necessary software (also free
and open-source) pre-configured and including
documentation and tutorials, for reducing and calibrating ELCPSS spectra, along with other common
astronomical research projects (Beaver, 2012). The
goal is to provide a self-contained researchcomputing environment that can be shared between
users, thus sidestepping for beginners the difficult
process of establishing such a computing environment from scratch.

Since the virtual machines can be portable and
cross-platform, individual VMs can be customized
for particular combinations of projects and users. We
see this, then as a key component to making ELCPSS
practical and useful. For example, the coauthor of
this paper had no prior training in astronomy before
beginning this project, and he learned much of the
process via a virtual machine specifically developed
for this task.
9. Extending the Wavelength Range

The usable wavelength range for the current set
of observations is from 4000-6500 Angstrom. The
short wavelength cutoff is mostly from the small
amount of light output at short wavelengths by the
spectral flat-fielding lamp used (see above), but also
from absorption by the corrector plate and the bandpass filter built into the CCD camera.
The long wavelength cutoff is mostly due to the
IR blocking properties of the CCD bandpass filter,
and it is possible to have this filter removed and replaced with a broad-spectrum window, extending the
spectral response out to much longer wavelengths.
This would possibly allow for extraction of the Rband as well as B and V.
The traditional complication of near-IR spectroscopy is that a diffraction grating produces a secondorder short-wavelength spectrum that overlaps the
first-order, long-wavelength spectrum. The usual
method of dealing with this is to take a second spectrum through an order-separating filter that blocks the
short-wavelength first-order spectrum. Since the
point of ELCPSS is that it is simple and low-cost, the
introduction of a filter-wheel mechanism and a specialized order-separating filter is counter to the goals
of the project.
There may be another approach however. Since
the spacing of the lines in the grating is relatively
large compared to a traditional grating, it may be
possible to execute control not only over the spacing
of the lines, but also over their actual transmission
profile. This raises the possibility of making a grating
that produces no significant second-order spectrum in
the first place, thus eliminating the need for an orderseparating filter and allowing for one to record a
spectrum from 3500-1050 Angstrom in one exposure.
Thus one could in principal obtain BVR magnitudes
with a single spectrum.
The correct transmission profile for a grating that
produces only odd orders (and thus no second-order
spectrum) is that of equally-sized adjacent dark and
clear bands (Jenkins and White, 1950). Indeed, the
100 line/inch grating we produced already approximates this, as it was printed from a digital file with
this pattern, and the 600 dpi laser printer was able to
approximate this pattern. We believe that with a highend inkjet printer, it may be possible to produce gratings that effectively suppress the second-order spectrum. Thus ELCPSS could perhaps be used as a
roundabout means for extremely low-cost BVR photometry (ELCBVR).
10. Conclusions
We find that the use of a low-cost objective grating on a small telescope, coupled with a low-end
consumer DSLR camera, can yield wavelength and
flux-calibrated spectra. Standard B and V magnitudes
can be extracted from the spectra, and our early results suggest this may be a useful way in some circumstances to obtain accurate B and V stellar magnitudes at very little cost in equipment.
Although the process itself has clear educational
uses, much work remains to be done to determine
whether ELCPSS could find practical use as a tool
for extending broad-band photometry to users who
otherwise would be limited to visual magnitude estimates. Further tests need to be done using ZOG techniques, and the practical limits in magnitude and precision need to be determined, for both absolute spectrophotometric calibration and for hybrid techniques that use ELCPSS to obtain nearly-calibrated
instrumental magnitudes with broad-band photometric standards providing the final calibration. Experiments should be carried out with other grating
materials and printing methods, and a wider array of
telescopes and detectors should be tested. Individuals
wishing to collaborate on the ELCPSS project should
contact the authors.
11. Acknowledgements
We engaged in many helpful discussions with
Doug Fowler. This work was supported in part by the
following grants:
UW - Fox Valley Foundation Grant, 2010
UW - Fox Valley Professional Development

Grants, 2010, 2012
UW Colleges Department of Computer

Science, Engineering and Physics Professional
Development Grants 2010, 2012
119
12. References
Ausienis, A., Straiys, V. (1969). Soviet Astronomy
13, 316.
Beaver, J. E. (2012). A Virtual Astronomical Research Machine in No Time (VARMiNT) in Proceedings for the 31st Annual Symposium on Telescope
Science (Warner et al., eds.). pp. 203-208. Society for
Astronomical Sciences, Rancho Cucamonga, CA.
Beaver, J. E, Robert, D. M. (2011a). A CCD Spectrograph for One Dollar. in Earth and Space Science: Making Connections in Education and Public
Outreach, ASP Conference Series, 433, 425.
Beaver, J. E., Robert, D. M. (2011b). A CCD Spectrometer for One Dollar. in Cosmos in the Classroom 2010: A Hands-on Symposium on Teaching
Introductory Astronomy. (A. Fraknoi, ed.). C1. Astronomical Society of the Pacific.
Colina, L. (1995). CDBS Kurucz Stellar Atmosphere Atlas. STScI OSG# Instrument Science Report
SCS CAL-006.
Jenkins, F. A., White, H. E., (1950). Fundamentals of
Optics. p. 310, 327. McGraw Hill, New York.
Space Telescope Science Institute (2012a). Calibration Database System (CDBS).
http://www.stsci.edu/hst/observatory/cdbs/astronomic
al_catalogs.html. Retrieved, 11 April 2012.
Space Telescope Science Institute (2012b). Specview.
www.stsci.edu/institute/software_hardware/specview.
Retrieved, 11 April 2012.
Tody, D. (1986). The IRAF Data Reduction and
Analysis System. in Proc. SPIE Instrumentation in
Astronomy VI. (D. L. Crawford, ed.). 627, 733.
Tody, D. (1993). IRAF in the Nineties. in Astronomical Data Analysis Software and Systems II,
A.S.P. Conference Ser. (R. J. Hanisch, R. J. V. Brissenden, J. Barnes, eds.) 52, 173.
Wells, D. C., Griesen, E. W., Harten, R. H. (1981).
FITS - a Flexible Image Transport System. Astronomy and Astrophysics Supp. Series 44, 363370.
120
Algol Spectroscopy Boyd et al.
Spectroscopic Analysis of Algol during Eclipse Cycle

Jonathan Boyd, Kodiak Darling, Elise Sparks, Lajeana West
Students
Estrella Mountain Community College, Phoenix, Arizona
Douglas Walker
Adjunct Faculty
Estrella Mountain Community College, Phoenix, Arizona
Douglas.Walker@estrellamountain.edu
Abstract
Algol, within the Perseus constellation, is referred to as the Winking Demon Star due to its varying apparent
magnitude and its representation of the Gorgon Medusa. Every 68.75 hours its light dims suddenly, and brightens again over a ten-hour period. Further observation shows a small dip in light output halfway in between the
large dips, indicating that Algol is an eclipsing binary star system. Detailed inspection of the spectrum indicates
that Algol is also a spectroscopic binary. Algol consists of a 3 solar diameter B8V star and a 3.5 solar diameter
K0IV in very close orbit around each other. This project investigated the spectral characteristics of Algol A and B
during the primary eclipse cycle. Low-resolution spectroscopy of the eclipse cycle was imaged over several
nights in order to investigate any changes in the emission line profile of the star system. This work lays the foundation for future studies in low-resolution spectroscopy of the Algol and Algol-type systems.
1. Introduction
1.2 Special Projects Course Structure
1.1 Background
Estrella Mountain Community College is part of

the Maricopa County Community College System.
This system consists of ten colleges, two skill centers. and numerous education centers, all dedicated to
educational excellence, and to meeting the needs of
businesses and the citizens of Maricopa County. Each
college is individually accredited, yet part of a larger
system the Maricopa County Community College
District, which is one of the largest providers of
higher education in the United States.
EMCCs special mathematics course, designated
MAT298AC, is a student driven research course focused on the application of learned techniques in the
fields of astronomy, applied mathematics, and statistics. Students study with a large array of goals in
mind; a common goal amongst the student body is to
gain the knowledge necessary for work in a gamut of
fields and professions. MAT298AC offered a unique
opportunity within this environment to gain a particularly advantageous amount of experience in primary
research and data analysis. The class, having been
reclassified as AST298A, now has a greater focus on
the techniques and skills needed for astronomy specific activities. At the start of the semester the class
had a wide variety of possible research projects to
choose from, each with its own unique obstacles and
avenues of experience. The field of study chosen,
spectroscopic analysis of the Algol system during its
primary eclipse, was both within the capabilities of
EMCC equipment and the classes' experience in the
Since the fall of 2010, Estrella Mountain Community College (EMCC) has offered a series of special mathematics project courses focused on applied
research for the undergraduate student. The first series of courses, offered in the fall 2010 and spring
2011 semesters, engaged the students in observing
and recording visual measurements of double stars,
selected from the US Naval Observatory maintained
Washington Double Star Catalog. This research oriented course was extended in the fall 2011 semester,
in a learning community which combined an introductory astronomy course with the special mathematics projects course. This approach was intended to
extend the students exposure to observational astronomy and data collection and analysis. The success of this sequence resulted in a special astronomy
projects course being offered for spring 2012, focusing on spectroscopy. The objective of this course was
to introduce the student to spectroscopy and its applications to stellar observation and analysis. The intent
of this course offering was to give students the opportunity to conduct original astronomical research with
the application of mathematics to data collections and
analysis.
121
astronomical sciences. This study resumes work in

the field of binary star research, further augmenting
EMCC's fledgling contributions to this field. With
this in mind, the students of AST298A began their
research project with the following objectives:
difficulty of resolving Algol C as an individual star, it

is often overlooked and incorporated with light emitted from Algol B.
to gain further knowledge of the nature of the individual stars within the Algol system
to use the knowledge gained to make inferences
about the Algol system
to provide data for more accurate models of binary stars
to establish a foundation for further Algol research
to expand upon individual knowledge of research
processes and analysis
to provide an avenue for the students to attempt
publication of their work in a recognized journal
1.3 Algol System

Algol, also known as Beta Persei, is the most
famous eclipsing binary star system, because it can
be distinguished with the naked eye. It is found in the
constellation of Perseus, named for the ancient hero
who killed the gorgon Medusa. Because of the stars
dramatic eclipse cycle and the star's position on the
disembodied head of Medusa, it has been nicknamed the Winking Demon star.
Algol is an eclipsing trinary star system, determined to be 93 light years from Sol. It is one of the
first eclipsing variable stars to ever be discovered.
The main star (Algol A) is a spectral class B8V main
sequence star with a mass 3.5 times greater, a diameter 3 times greater, and an absolute brightness 100
times brighter than our Sun. The secondary star (Algol B) is a spectral class K2IV type subgiant that
measures about 0.8 solar masses, 3.5 solar diameters,
and an absolute brightness 3 times brighter than our
Sun. Algol A and B are a close binary pair, about 8
solar diameters apart, with a period of 2.87 days. The
primary eclipse happens when the less massive Algol
B passes in front of the more massive Algol A. The
eclipse cycle lasts for a period of about 10 hours. At
the eclipse's minima, the luminosity changes from an
apparent magnitude of 2.1 to an apparent magnitude
of 3.4. There is a faint third star (Algol C) in the Algol system, which is a spectral class F1 main sequence star. Algol C measures 1.7 solar masses, with
a diameter slightly smaller than our Sun and an absolute brightness of about 4 times that of our Sun. Algol
C orbits the inner two stars at a distance of about 2.69
AU with a period of about 1.86 years. Due to the
122
Figure 1. Diagram of Algols eclipse cycle.
2. Equipment & Resources

2.1 Telescope
Equipment for observing sessions consisted of a
Celestron 11-inch Schmidt-Cassegrain telescope with
a 2800mm focal length. The telescope was used in
conjunction with a Meade DSIPro CCD camera fitted
with a diffraction grating to image the Algol system.
The telescope was kept on target through the use of a
heavy-duty (CG-5GT) computerized German Equatorial Mount.
The Celestron 11-inch Schmidt-Cassegrain telescope is equipped on a mobile tripod base which
made transport to and from a remote location observing site relatively quick and easy.
Figure 2. Celestron Schmidt-Cassegrain telescope and

student, Jonathan Boyd.
2.2 Star Analyzer

The Paton Hawksley Star Analyser 100 was used
to provide the spectra of the Algol system. The Star
Analyser 100 was designed specifically for amateur
astronomical spectroscopy. It can be mounted on the
telescope similarly to other 1.25-inch filters and will
work with most cameras. It can image dim objects

and display the zero-order (star) along with the spectrum. This makes it easy for spectra to be calculated.
The Star Analyser was mounted on the end of the
CCD camera.
line profiles, for identifying the compounds that correspond to the spectral features within the calibrated
graph.
Figure 5. Screenshot of RSpec software in use.
2.5 Observing Site
Figure 3 Hawksley Star Analyser 100
2.3 Meade DSIpro CCD Camera

The Meade DSIpro CCD camera is designed to
help novice and amateur astronomers capture the
images of a variety of objects throughout the night
sky. By attaching the Star Analyser to the camera, the
spectral images created were captured for later analysis.
Figure 6. Student Jonathan Boyd, guest Melissa Boyd,
and instructor Doug Walker at Litchfield Park location.
Figure 4. Meade DSIpro CCD Camera
2.4 RSpec Real-Time Spectroscopy Software

RSpec is a new form of software that allows astronomers to take raw spectrum images and create a
calibrated profile in real time. The software allows
the user to analyze the composition of various stellar
objects ranging from stars, to nebula emission lines,
to detecting the red shift of a receding quasar. It also
provides a library of reference series including profiles for all star spectral types and a small variety of
Figure 7. Students Elise Sparks, Lajeana West, and Kodiak Darling at Buckeye location.
Observations were taken at two separate coordinates. The first quarter of the eclipse cycle was imaged at 33.503049 N, 112.363466 W, in Litchfield
Park, Arizona, with medium levels of light pollution.
The other part of the eclipse cycle was imaged at
123
33.365849 N, 112.577612 W in Buckeye Arizona,

with low levels of light pollution. Ambient temperature at ground level was between 14-25 degrees Celsius.
Reduction and analysis of raw images was accomplished through the use of RSpec v1.5. The library of reference spectra for B8V and K0IV stars
included in Rspec 1.5 was used as a calibration and
analysis tool throughout the research process. Adobe
Photoshop CS5 was used as an aid in the analysis of
changing spectral features as the eclipse cycle progressed.
3. Observations
3.1 Observation Procedure
Imaging of the eclipse cycle was broken into two
segments due to the Algol system setting on the horizon during the latter portion of the eclipse cycle. The
first imaging session took place on 2012 February 28.
The second imaging session took place on 2012
March 2 with the third and final imaging session taking place on 2012 March 3.
Figure 8. Laptop image of spectrum of Algol
The telescope was set to track on the Algol system, and the CCD camera mounted on the viewpiece; the CCD camera was connected to its accompanying computer software, which was used to set
image parameters and capture images. Images were
taken at 15 minute intervals, beginning immediately
before Algol B began its eclipse of Algol A, and ending directly after the conclusion of the eclipse cycle.
Images were taken through the use of the CCD camera and diffraction filter described above. The Images
were saved as FTS files and analyzed in their raw
format with the RSpec computer program.
3.2 Observing Session 1

The first night of observation encompassed the
evening February 28 to the morning of February 29.
The minimum of the eclipse was scheduled for 01:46
on February 29, which meant the eclipse would begin
at 20:46. The first image was taken at 20:00 then
further images were taken at 15 minute intervals.
This allowed for a couple images outside of the
eclipse to be considered as a baseline image.
February 28-29 had been chosen for its ideal
time frame. The eclipse would begin shortly after
sunset and finish approximately at sunrise. However,
at the beginning of the eclipse cycle, the star was
almost directly overhead. Therefore, due to the stars
movement across the sky, the images became unclear
as the star approached the horizon. Images were captured until 23:30, approximately 30% completion of
the eclipse. The star was no longer within view after
23:37.

To continue with the eclipse cycle, another night
of observation would be required. The evening of
March 2 to the morning of March 3 was chosen because the minimum would be at 22:36. The eclipse
would begin at 17:36, but by the time images could
be captured, beginning at 19:30, the star would be at
approximately 25% completion. This allowed for
overlap with the previous session of observation.
In the hopes of keeping view of the star system
for a longer time frame, the location had been
changed from Litchfield Park to Buckeye, in order to
lower light pollution and the number of ground level
obstructions. Images were collected through the
minimum and until 00:30, approximately 70% completion of the eclipse cycle.

Observation 3 was conducted outside of the primary eclipse, in order to establish a set of data
against which spectra changes during the primary
eclipse could be compared to. Observations were
undertaken at the original Litchfield Park location,
initiating at 19:45 on March 3rd and ending at 22:15
the same night.
4. Data Collection and Reduction

4.1 Measurement Data
During observing sessions, imagery was captured at 15 minute intervals on the quarter hour. Each
124
set of observation data was archived in separate folders that were organized according to date of observation. The raw profile line image is shown in Figure
12. During the early part of the observing run, the
line profile was relatively clean, but as Algol began
to approach the horizon, atmospheric extinction began to take its toll: the line profile became very noisy,
as shown in Figure 10.
Figure 11. Instrumented corrected profile on February 28

at 22:00, includes Hydrogen Balmer lines.
5.2 Analysis Observing Session 2 (Mar 2)

Data from observing session 2 consists of 49 individual profiles spanning the time from 20:00 until
00:15. In order to validate the quality of the data being collected, the response curve during the same
portion of the eclipse cycle for observing session 1
and 2 were compared. As shown in Figure
5.3 Analysis Observing Session 3 (Mar 3)
Figure 9. Algol line profile during early part of observing

session.
22:15. This observing run was conducted to obtain
additional baseline data outside of the eclipse cycle.
5.4 Summary Analysis of Eclipse

To establish a baseline of instrumentation stability throughout the eclipse, two profiles were taken at
the same time within the eclipse cycle. The first profile was from February 28 and the second one was
from March 2 at 25% completion within the eclipse
cycle. As shown in Figure 12, the profiles from both
lines align, establishing stability within the data collection between the two separate nights
Figure 10. Algol profile during latter part of observing

session.
The data reduction process consisted of reviewing the quality of the line profiles and producing instrument response curves for all good profiles.
5. Data Analysis
5.1 Analysis of Observing Session 1
(Feb 28)
23:30. The Hydrogen Balmer absorption lines were
the primary focus of the analysis.
Figure 12. Line profiles from March 2 at 20:00 and February 28 at 11:15: 25% completion of the eclipse cycle.
A comparison of line profiles covering the preeclipse from observing session 1 with line profiles
during the first quarter, last quarter from sessions 1
and 2, and from outside the eclipse from session 3
shows the changes in the Hydrogen Alpha, Beta,
Gamma and Delta absorption lines as shown in Table
I and Figures 13, 14, 15, and 16 respectively.
125
Table I. Measurement Data Spreadsheet.
Figure 16. Hydrogen Delta.
6. Conclusion
Figure 13. Hydrogen Alpha.
Variations in spectral features at this point may

be attributed to several possible factors, which are
currently under further investigation. Additional data
and updated conclusions are available upon request.
Throughout the data analysis process, student
familiarity with the RSpec program and associated
equipment has increased to a point which will allow
for more thorough and complex research initiatives in
the future. Using the data of this paper as a launching
point for future lines of inquiry, EMCC special projects courses may be able to contribute to the current
available body of data on the Algol system in a significant way.
7. Acknowledgments
Figure 14. Hydrogen Beta.
The authors thank Tom Fields for his guidance and

assistance with the use and acquisition of the RSpec
program; to Doug Walker for the use of his equipment, expertise, and guidance. A very special thanks
goes to Jeff Hopkins of the Hopkins Phoenix Observatory for his guidance and assistance at every step of
the way. Finally, we thank the Estrella Mountain
Community College administration for their enthusiastic support.
8. References
Rspec reference libraries
NASA.gov
The SAO/NASA Astrophysics Data System
Figure 15. Hydrogen Gamma.
126
Sky and Telescope Magazine star charts
AO Ser Alton et al.
A Fresh Look at the Algol-like Eclipsing Binary, AO Ser

Kevin B. Alton
UnderOak Observatory
70 Summit Ave., Cedar Knolls, NJ 07927
mail@underoakobservatory.com
Andrej Pra
Dept. of Astronomy & Astrophysics
800 Lancaster Ave, Villanova University, Villanova, PA 19085
aprsa@villanova.edu
Abstract
CCD images acquired in B, V, and Ic passbands between July and August 2008 and from June to August 2010
were used to revise the ephemeris and update the orbital period for AO Ser. Analysis of the O-C diagram using
all available time-of-minima data revealed a continually increasing orbital period for at least 68 years. Thereafter,
a sudden decrease in orbital period (~0.84 sec) most likely occurred during the first few months of 1998; potential
causes for this abrupt jump and other alternating changes in period are discussed. Although the spectral classification of the primary star has been long considered A2V (Teff ~8800 K), B-V data from this study as well as color
index estimates from several other recent survey catalogs suggest a significantly cooler A7V (Teff ~7800 K) primary. A semi-detached Roche model of the binary produced theoretical fits matching light curve data in all passbands. AO Ser is reported to be an oscillating Algol-type variable star referred to as an oEA system. Using Fourier methods, lightcurve analysis in this study did not convincingly reveal any underlying periodicity other than
that expected from the dominant orbital period.
1. Introduction
AO Ser is a short period (<1 day) and bright Algol-type binary suitable for study by amateur and
professional astronomers alike. After its discovery in
1935 by the prolific observer Cuno Hoffmeister, the
publication record for AO Ser was very sporadic with
only a handful of times-of-minima (ToM) available
for each of the next three decades. These mostly included photographic as well as visual timings. Only
since 1969 has a steady flow of ToM values been
available but no full light curve had been published
until Zavros et al. (2008). AO Ser not only varies
extrinsically by mutual eclipses but evidence is
mounting that it also varies intrinsically by Sct-type
pulsation of its primary component (Kim et al., 2004;
Zavros et al., 2008). This new class of oscillating
Algol-type variable stars is now designated as an
oEA variable (Mkrtichian et al., 2004). More recently, the multiband (V and R) photometric properties and period variations of AO Ser were published
by Yang et al (2010).
AO Ser (BD+172943) varies in magnitude (V)
between 10.8 and 12.2; the first modern orbital period (0.87934745 d) was reported by Koch (1961).
This
target
is
favorably
positioned
(J2000 = 15h58m18s.408, J2000= +17169.96) for
mid latitude backyard observers in the Northern
Hemisphere with a clear view of Serpens Caput. Its
spectral type (A2) reported by Brancewizc and

Dworak (1980) is consistent with what would be
expected for an Algol-like binary system, however,
evidence from this study and several survey catalogs
(eg. Tycho-2, USNO-A2.0, USNO-B1.0, and
2MASS) suggest a much cooler A7 system. This difference from all other published reports on this binary system will be discussed herein.
2. Methods, Observations and Data

Reduction
2.1 Astrometry
Images of AO Ser were matched against the
UCAC3-based standard star fields provided in MPO
Canopus (V10.3.0.2 Bdw Publishing, Inc.). This
automatch feature generates a star chart centered
on the putative center of the image and then matches
the charts center, rotation, and scaling to the image.
Plate constants are internally calculated which convert X/Y coordinates of a detected object to a corresponding RA and declination.
2.2 Photometry
CCD observations at UnderOak Observatory
span from 2008 July 25 to August 22 and then again
from 2010 June 19 to August 31. Equipment included
127
AO Ser Alton et al.
a 0.2-m catadioptric telescope with an SBIG ST

402ME CCD camera mounted at the primary focus.
The KAF 0402ME camera array (765510) comprised of square pixels (9 m) produces a
~914 arcmin field-of-view (FOV) with an image
scale of ~1 arcsec/pixel with this optical assembly.
The CCD camera was maintained at 0C with thermoelectric cooling. Automated multi-bandwidth (B,
V, and Ic) imaging was performed with SBIG photometric filters manufactured to match the Bessell prescription. During each 2008 session, an image in each
bandpass was sequentially collected every 45 s; data
acquisition and reduction (raw lights, darks, and flats)
for this camera system has been described elsewhere
(Alton, 2010). The other campaign which started in
2010 employed essentially the same equipment except that the acquisition time for each filter was increased to 60 s. Instrumental readings were reduced
to catalog-based magnitudes using the MPOSC3 reference star fields built into MPO Canopus. The MPO
Star Catalog (MPOSC3) is a hybrid assemblage
which includes a large subset of the Carlsberg Meridian Catalog (CMC-14) as well as from the Sloan
Digital Sky Survey (SDSS). Almost all stars in
MPOSC3 also have BVRI magnitudes derived from
2MASS J-K magnitudes; these have an internal consistency of 0.05 mag for V, 0.08 mag for B,
0.03 mag for Ic, and 0.05 mag for B-V (Warner,
2007).
minimize error due to differential refraction and color

extinction.
Figure 1. Constant relative magnitude (Ic-band) exhibited

by comparison stars during a typical AO Ser photometric session (2008 Aug 02).
3.2 Folded Light Curves and Ephemerides

Individual photometric values in B (n = 1573), V
(n = 1598), and Ic (n = 1580) were combined by filter
to produce light curves that spanned 4 weeks of imaging in 2008 and 10 weeks in 2010. A period solution from all folded 2008 and 2010 datasets was calculated by MPO Canopus. Accordingly, the linear
ephemeris (Eq. 1) for the heliocentric primary minimum was initially determined to be:
2.3 Lightcurve Analyses

Light curve modeling was performed using Binary Maker 3 (Bradstreet and Steelman, 2002),
PHOEBE 0.31a (Pra and Zwitter, 2005), and
WDwint Version 5.6a (Nelson, 2009), all of which
employ the Wilson-Devinney (W-D) code (Wilson
and Devinney, 1971; Wilson, 1979). Geometric renderings were produced by Binary Maker 3 (Bradstreet and Steelman, 2002).
3. Results and Discussion

3.1 Ensemble Photometry
Five comparison stars in the same field of view
with AO Ser were used to calculate the relative
change in flux and standard magnitudes (Table I).
Over the duration of each session, comparison stars
did not exhibit any variable behavior so Comp Cavg
values remained constant over each observation period (Figure 1). The mean standard magnitude for
each comparison star varied between 0.015 (V and
Ic) and 0.03 (B) mag. During the entire campaign
the airmass for all observations was kept under 2 to
128
Min I = HJD 2454664.54963

+ 0.8793429 ( 0.0000013) E
(1)
This result is in general agreement with the orbital period reported over the past 5 decades from
other investigators. The period determination was
confirmed with periodograms produced (Peranso
v 2.1, CBA Belgium Observatory) by applying periodic orthogonals (Schwarzenberg-Czerny, 1996) to
fit observation and analysis of variance (ANOVA)
which was used to evaluate fit quality.
ToM values were estimated by Minima (V24b,
Nelson, 2007) using a simple mean from a suite of
six different methods including parabolic fit, tracing
paper, bisecting chords, Kwee and van Woerden
(1956), Fourier fit, and sliding integrations (Ghedini,
1981). Four new secondary (s) and two new primary
(p) minima were detected during this investigation.
Since no obvious color dependencies emerged, the
timings from all three filters were averaged for each
newly determined ToM. Residual values (O-C) were
estimated from a complete set of visual (vis), photographic (pg), photoelectric (pe) and CCD timings
reported over the past 80 years. Calculations were
AO Ser Alton et al.
MPOSC31
Star Identification
RA
MPO3 J15575952
15h57m59s.52
+171926.3
10.843
MPO3 J15580243
15h58m02s.43
+172113.4
11.985
MPO3 J15581300
15h58m13s.00
+171642.5
12.852
12.158
DEC
B mag
Ic mag
(B-V)
10.41
9.889
0.433
11.547
11.021
0.438
11.391
0.694
V mag
MPO3 J15581841
15 58 18 .41
+171610.0
11.667
11.208
10.661
0.459
MPO3 J15581990
15h58m19s.90
+171945.4
12.518
11.50
10.445
1.018
MPO3 J15582353
15h58m23s.53
+171839.8
12.876
12.214
11.474
0.662
MPOSC3 is a hybrid assemblage which includes a large subset of the Carlsberg Meridian Catalog
(CMC-14), the Sloan Digital Sky Survey (SDSS), and 2MASS Catalog.
Table I. Astrometric Coordinates and Estimated Color of AO Ser (MPO3 J15581841) and Five Comparison Stars
in Same Field of View
initially based upon the linear elements (Eq. 2) defined by Kreiner (2004):
Min I = HJD 2452500.3991 ( 0.0007)
+ 0.8793410 ( 0.0000004) E
increasing linearly with time as suggested by the

positive coefficient (a2) for the quadratic term, x2.
Algol-type binaries typically consist of a hot mainsequence massive primary well within its Roche lobe
and a less-massive subgiant companion filling its
Roche lobe. Gas streams from the secondary to the
primary result in mass exchange and barring any
other competing phenomena, the net effect is a
change in the orbital period. In this case, the orbital
period rate of increase (p/p = 2a2 =1.25 10-10) of
this system, which is equivalent to a period increase
rate of dP/dt = +0.004479 sec yr-1, lasted for at least
68 years but suddenly slowed to a standstill as judged
by the break in O-C residuals (Figure 2) around cycle -1855. The intersection of Q+S and L1 suggests
this change which is centered at the beginning of
1998 probably occurred sometime between mid 1996
and mid-1999. This cannot be attributed to the shift
from vis-pg to ccd-pe recordings since instrumentderived ToMs regularly reported after cycle 118.5
continue in a similar fashion until another break is
observed near cycle 2000. This characterization of
AO Ser is in contrast to that described by Yang et al.
(2010) who only report a period decrease since 1981.
This obvious difference from our findings is dis-
(2)
The O-C diagram (Figure 2) for AO Ser is complex and characterized by parabolic and possibly alternating changes exhibited by many other late type
Algol-like binary systems such as U Cep (Manzoori, 2008), Y Leo (Pop, 2005), TY Peg (Qian,
2002), and X Tri (Qian, 2002). Regression analysis
using a scaled Levenberg-Marquardt algorithm (Press
et al., 1992) as implemented in QtiPlot (v0.9.8 2010)
revealed that the O-C data from the initial timing in
1930 until the last twelve years could be fit (r2 > 0.99)
by a quadratic expression (Eq. 3) modulated with a
sinusoidal term:
c + a1x + a2x2 + a3sin(a4x + a5)
(3)
Accordingly, the coefficients ( error) for each

solved term in the above (Eq. 3) are provided in Table II. From the parabolic component (c + a1x + a2x2)
it is apparent that the orbital period has been slowly
Curve
a1
1.59310-3
-8.67910-7
(0.29010-3)
(0.228)
-6.50310-3
3.04610-6
L1a
L2a
a2
a3
a4
a5
(1.08810-3) (0.418910-6)
Q+Sb
1.66710-2
(0.13610-2)
+8.91110-6
(0.23710-6)
+6.24010-11
(0.76010-11)
Q+Sc
1.66710-2
(0.13610-2)
+8.91110-6
(0.23710-6)
+3.74810-3
+6.24010-11
4.24910-4
3.209
(0.76010-11) (0.59310-3) (0.20610-4) (0.244)
a: Linear least squares fit: (O-C) = c + a1x

b: Non-linear regression fit (O-C) = c + a1x + a2x2
c: Non-linear regression fit (O-C) = c + a1x + a2x2 + a3sin(a4x + a5)
Table II. Coefficients (error) From Regression Analysis of O-C vs Cycle Number Data for AO Ser
129
AO Ser Alton et al.
cussed further on in this section along with the potential for a light time effect which they also had proposed.
where
m =
and
2/
The abrupt change in orbital period likely took

place within a 3 year window as suggested by the
O-C diagram (Figure 2). Assuming the solar mass of
a putative A7V primary approximates 1.82 M
(Harmanec, 1988), then from the mass ratio
(qph=0.235) determined herein, the calculated mass
loss is 6.85110-7 My-1. Averaged over three years
this translates into nearly two-thirds of the Earths
mass. Alternatively from Miku et al. (2002), the
associated change in angular momentum can also be
estimated from the equations for total angular momentum (L) and Keplers law:
=
2
+
3 3 1+
1/
+ 1
3 1+
2/
1
3
/ ,
where q = m2/m1 and P is the period of AO Ser.

Figure 2. Nonlinear quadratic+sine term fit (Q+S) of residuals (O-C) vs cycle number for AO Ser observed from
1930 March to 1998 June 17. The inset figure shows a
breakpoint centered near cycle -1855 (1998 Feb-March)
and the straight line linear fit of data (L1) through cycle
1983.5 (mid-2007). Near term data (L2) were also fit using simple linear least squares analysis. The bottom
panel represents a composite plot of all (O-C)2 residuals
from fitting Q+S, L1, and L2. All ToM values and references used to prepare this figure can be obtained by
request (mail@underoakobservatory.com).
Calculations based on linear coefficients from

the quadratic + sine term (Eq. 4) and simple least
squares (Eq. 5) fits resulted in the following
ephemerides, respectively:
Min I = HJD 2455666.9548 ( 0.0013)
+ 0.8793499 ( 0.0000002) E (4)
Assuming that the angular momentum change dL

can only be zero or a negative value, then in order to
produce a period change of dP/P = 1.1110-5 the total
mass lost from the system had to be at least 3.23
1024 kg or slightly more than one half (~54%) of
Earths mass. The corresponding magnitude of mass
loss estimated by either equation 6 or 7 is significant
but nonetheless not unprecedented (1.62610-4 to
9.510-8 M y-1) for semi-detached binaries (van
Rensbergen et al., 2011).
Due to the continually changing nature of the
O-C diagram, a new ephemeris equation (Eq. 8)
based upon a linear least squares fit (L2) of near-term
data (Table III) collected over the past four years
(cycle 1905 to 3601) was calculated:
Min I
Min I = HJD 2455666.9045 ( 0.0003)

+ 0.8793401 ( 0.0000002) E (5)
The orbital period difference (-9.778810-6 days)
between equations (Eq. 4) and (Eq. 5) corresponds to
a rather large decrease (0.844 sec) over a relatively
short period of time. The two most likely causes for a
decrease in period are mass exchange between the
components and/or angular momentum loss from the
binary system. A rate of conservative mass transfer
can be predicted (Eq. 6) as follows from Kwee
(1958):
/
130
1 2
(6)
= HJD 2455,666.9105( 0.0011)

+ 0.8793441 ( 0.0000004) E (8)
As such, if the complex O-C behavior of this

system continues unabated, revised ephemerides for
AO Ser will need to be calculated on a more regular
basis. The composite residuals from the quadratic +
sine term fit (Q+S) and simple least squares fit (L1
and L2) are shown in the bottom panel of Figure 2.
No additional underlying periodicity in the (O-C)2
diagram was uncovered with Peranso using Fourier
(Lomb-Scargle) and statistical (ANOVA) methods.
AO Ser Alton et al.
In a recent publication, Yang et al. (2010) attempted to make a case for decreasing period and a
third body with an incomplete set of data which exTime of Minimum
( 2,400,000)
Type
Cycle
Number
cluded all pg and visual (n = 300) plus other CCD

readings (n = 10) published from 1941 through the
middle of 2008. It is unfortunate that this rich set of
(O-C)1
(O-C)2
Reference
-0.000805
-0.000700
54175.5429
1905
54186.5371
1917.5
0.001632
-0.000662
54201.4835
1934.5
-0.000765
-0.000610
54211.5959
1946
-0.000786
-0.000575
54213.3545
1948
-0.000868
-0.000569
54220.3893
1956
-0.000826
-0.000545
54220.3906
1956
0.000494
-0.000545
54224.7859
1961
-0.000901
-0.000529
54242.3741
1981
0.000459
-0.000469
54244.5717
1983.5
-0.000274
-0.000461
54491.6646
2264.5
-0.002195
0.000395
54574.7651
2359
0.000581
0.000683
54581.3604
2366.5
0.000823
0.000706
54599.3865
2387
0.000433
0.000768
54610.8172
2400
-0.000300
0.000808
54612.5770
2402
0.000818
0.000814
54614.3375
2404
0.002636
0.000820
54628.4051
2420
0.000780
0.000869
54685.5625
2485
0.001015
0.001067
54696.5546
2497.5
0.001353
0.001105
54926.5017
2759
0.000781
0.001901
10
54928.2612
2761
0.001599
0.001907
11
54928.2614
2761
0.001799
0.001907
11
54929.1397
2762
0.000758
0.001911
11
54929.1400
2762
0.001058
0.001911
11
54932.2170
2765.5
0.000364
0.001921
11
54932.2181
2765.5
0.001464
0.001921
11
54939.6925
2774
0.001466
0.001947
12
54952.8851
2789
0.003951
0.001993
13
54968.7113
2807
0.002013
0.002048
12
54974.4272
2813.5
0.002196
0.002067
55259.7737
3138
0.002542
0.003056
14
55316.4953
3202.5
0.006647
0.003252
15
55327.4831
3215
0.002685
0.003290
15
55366.6142
3259.5
0.003157
0.003426
55395.6358
3292.5
0.006504
0.003526
55436.5206
3339
0.001938
0.003668
55666.9103
3601
0.004259
0.004466
16
(1) Baldwin and Samolyk, 2007 (2) Brt et al., 2007 (3) Borkovits et al., 2008 (4) Parimucha et
al., 2009 (5) Samolyk, 2008 (6) Brt et al., 2008 (7) Hbscher et al., 2010 (8) Lampens et al., 2010
(9) Present study (10) Doru et al., 2009 (11) Yang et al., 2010 (12) Samolyk, 2010 (13) Diethelm,
2009 (14) Samolyk, 2011 (15) Doru et al., 2011 (16) Diethelm, 2011.
Table III. Near Term Recalculated Residuals (O-C)2 for AO Ser Following Simple Linear Least Squares Fit of (O-C)1
and Cycle Number Between 2007March16 and 2011April15.
131
AO Ser Alton et al.
data was not considered since out of 300 additional

visual and pg observations, we found that only four
values needed to be rejected as spurious. In contrast
when all data are evaluated, residuals for the first
seven decades clearly describe an upwardly turned
parabola (Q), namely behavior consistent with conservative mass transfer and a continually increasing
period. Additionally on this quadratic curve, there is
evidence for underlying cyclical changes based upon
non-linear regression analysis (Eq. 3). The amplitude
(0.003748 0.000593 d) of the periodic oscillation is
defined by a3, the coefficient of the sine term. Assuming for the moment that this behavior is associated with a third body, then according to the relationship:
P3 = 2P/,
where the angular frequency, , corresponding to
a4 = 4.249 (0.206)10-4, its Keplerian orbital period
would be ~36 y. This estimate is more than twice as
long as the periodic value (~17 y) reported by Yang
et al. (2010) using an incomplete dataset covering the
time span between 1981 and 2009.
Analysis of the complete set O-C residuals available between 1930 and 2011 paints a very different
behavioral picture for this binary system. We agree
there is a modicum of evidence that suggests cyclic,
if not sinusoidal behavior in the O-C diagram but
only up to 1998. Thereafter, between 1998 and 2011
the O-C trace is curvilinear but not necessarily sinusoidal for this system. It is possible that the event(s)
responsible for the sudden change in orbital period in
1998 have masked our ability to tease out the putative
36 year periodicity observed during prior epochs.
Nonetheless, since the gravitational influence of
a third body cannot be simply turned off, we believe
that evidence for the presence of a third body in this
system is inconclusive. Moreover, during attempts to
fit light curve data to a Roche type model with our
data, neither the orbital eccentricity (e) nor third light
(l3) rose above a level different than zero when included as variable parameters.
This begs the question as to what other phenomena might lead to cyclic, but not necessarily, persistent sinusoidal behavior. Hall (1989) noted that Algol-like systems which exhibit alternating variability
in their OC diagrams are almost exclusively those
possessing a late-type secondary which to a high degree of certainty are chromospherically active. Furthermore, earlier type Algols composed of two radiative-envelope companions yield much simpler O-C
diagrams rarely demonstrating cyclic behavior (Budding and Demircan, 2007). The secondary in AO Ser
is likely a rapidly rotating late type subgiant and
132
therefore possesses a dynamically active lobe-filling

convection envelope.
A plausible alternative to the light travel time effect (LITE) from additional unseen mass involves
cyclic variability arising from magnetic activity of
the secondary (Applegate, 1992). In addition, Biermann and Hall (1973) proposed a model where the
secondary experiences abrupt episodes of mass transfer which orbits the primary as a disk. This leads to a
temporary reservoir of orbital angular momentum
where P decreases but at later time, angular momentum is returned by accretion during which P increases.
This scenario may have played out with AO Ser
starting with the sudden break observed in the O-C
diagram in early 1998 followed by a period decrease
which begins to reverse direction around cycle 2000
(mid-2007). At this time in the absence of any supporting data from a spectroscopic study or speckle
interferometry it would be speculative to assign any
single reason or combination of phenomena that
would unequivocally include LITE as an explanation
to this very complex O-C diagram for AO Ser.
3.3 Light Curve Synthesis
Figure 3. Folded (2008 July-August and 2010 JuneAugust) CCD light curves for AO Ser captured in B, V,
and Ic bandpasses.
Folded light curves (Figure 3) comprised of all

observations (relative magnitude) in B, V, and Ic,
show that minima are separated by 0.5 phase. Collectively, mode 5 (semi-detached; secondary fills Roche
lobe), synchronous rotation and circular orbits were
selected for modeling by PHOEBE. Each model fit
incorporated individual observations assigned an
equal weight of 1. Bolometric albedo (A1 = 1) and
gravity darkening coefficients (g1 = 1) for the primary
component were based on theoretical considerations
for radiative stars reported by Eddington (1926) and
von Zeipel (1924), respectively. As mentioned previously, third light (l3) did not rise above a value differ-
AO Ser Alton et al.
ent than zero when adjusted. Bolometric albedo

(A2=0.5) and gravity darkening coefficients
(g2 = 0.32) for cooler stars with convective envelopes
were assigned as determined by Rucinski (1969) and
Lucy (1967), respectively. After any change in Teff,
logarithmic limb darkening coefficients (x1, x2, y1, y2)
for both stars were interpolated from PHOEBE
(Van Hamme, 1993). Once an approximate fit was
obtained, differential corrections (DC) were applied
simultaneously to photometric data in all filters.
3.3.1. Binary Model
As mentioned earlier the spectral type reported
by Brancewizc and Dworak (1980) was A2 while the
color index (B-V = 0.1) listed by Kreiner et al.
(2004) suggests an even hotter primary star. Yet, evidence from this study which is the first to simultaneously acquire AO Ser light curves in B and V passbands, as well as data from multiple survey catalogs
point to a much cooler system (Table IV).
The B-V value (0.247) observed in this study
was taken during Min II when spectral contamination
by the secondary was least likely to bias the result. In
addition, maximum galactic reddening (E(B-V) =
0.0381) in this region of the sky (Schlegel et al.,
1998) is unremarkable and therefore no adjustment
has been made to obtain the associated intrinsic color
index for AO Ser.
On balance, we adopted an A7 spectral classification for AO Ser which is considerably cooler than
that reported by all previous investigators; the effective temperature of the primary was estimated to be
7800 K based on the tables from Flower (1996). Values for q, 1, 2, and i reported by Zavros et al.
(2008) were used as a starting point for an unspotted
fit (V bandpass) using Binary Maker 3. A1, A2, g1, g2,
q, and T1 were fixed parameters whereas 1, T2,
phase shift, x1, x2, and i were iteratively adjusted until
a reasonable fit of the observed relative flux by the
model was obtained.
Thereafter, the model fit was refined with
PHOEBE while using DC to achieve a simultaneous
minimum residual fit of all (B, V, and Ic) photometric
observations. In mode 5, 2 is constrained and as
Survey
(B-V)a
Spectral Class
such is automatically calculated with any change to

this parameter. The initial W-D model fit was marginally acceptable in all passbands which led to further adjustment of variable parameters. Early investigators of this Algol like system had used a mass ratio
(q = m2/m1) of 0.45, presumably based on a tabulation of computed parameters for eclipsing binaries
published by Brancewicz and Dworak (1980). However, in the absence of any radial velocity data for
this binary system, it was deemed appropriate to allow this physical element to also vary freely.
An improved solution quickly emerged at a significantly different value, so it was decided to carefully examine this parameter under more controlled
conditions. Mass ratio (q) was fixed over a range of
0.1 to 0.8 and a series of fits allowed to converge.
The curve generated by plotting 2 as a function of
the putative mass ratio showed a shallow minimum
between q = 0.220 and 0.260 (Figure 4). The
mid-point (0.240) was nominally assigned as the
starting q-value during Roche modeling.
Figure 4. Search for mass ratio (q) providing the best

2
simultaneous light curve fit (Chi ) using data from all
passbands (B, V, and Ic). The shallow minimum is essentially the same between 0.22 and 0.26 q.
Elements for AO Ser obtained using PHOEBE

are provided in Table V while light curve fits for
each filter band (B, V, and Ic) are respectively illustrated in Figures 5, 6, and 7. A 3-dimensional rendering produced using Binary Maker 3 depicts a spatial
Tycho-2
USNOB1.0
USNOA2.0
2MASS
ASCC2.5 V3
Present Study
0.219
0.195
0.235
0.459
0.266
0.247
A7
A7
A7
F5
A8
A7
a: Not corrected for galactic reddening E(B-V) ~ 0.0381

b: Estimated from Fitzgerald (1970)
Table IV. AO Ser Color Index (B-V) From Several Survey Catalogs and the Present Study
133
AO Ser Alton et al.
Parameter
T1 (K)a
T2 (K)b
q (M2/M1)
A1a
A2a
g1a
g2a
1 b
2
i b
Present Study
Zavros et al. 2008
Yang et al. 2010
7800
4412 (33)
0.235 (3)
1.0
0.5
1.0
0.32
3.56 (3)
2.318
86.8 (4)
8770
4858-5137c
0.45
1.0
0.5
1.0
0.32
4.625-4.936c
2.778
80.12-81.18c
9480
4786 (11)
0.220 (2)
1.0
0.5
1.0
0.32
2.597 (11)
2.282
87.62 (17)
a: Fixed elements during DC

b: Error estimated from heuristic scanning of the 2 surface (Pra and Zwitter, 2005)
c: Range since V, R, and I not simultaneously fit during W-D modeling
Table V. A Comparison of Selected Photometric Elements for AO Ser Obtained Following Roche Model Curve Fitting
model for AO Ser (Figure 8). As might be expected

from adopting an A7 primary for AO Ser instead of a
hotter A2, most of the physical and geometric elements determined in this study are quite different
from those most recently reported by Yang et al.
(2010) and Zavros et al. (2008). One other revealing
difference is related to using a theoretical q value
(0.45) from the literature (Zavros et al. 2008), as opposed to empirically determining a photometric q. As
such, our value (qph = 0.235) agrees reasonably well
with the value reported by Yang et al. (2010) of qph =
0.220). Based upon a compilation of absolute dimensions of eclipsing binaries (Harmanec, 1988), the
mean stellar mass for an A7 main sequence binary
star (1.82 M) was used to estimate the mass of the
secondary and the semi-major axis (a = 5.055) in
solar units. The associated mass (0.428 M) of the
secondary and best fit temperature for T2 (4412 K;
log 3.645) of the putative sub-giant suggested a spectral type between K1 and K3 (Harmanec, 1988;
banolu et al., 2006; Surkova and Svechnikov,
2004). Clearly, further spectroscopic analyses will be
necessary to solidify the spectral classification of this
Algol-like system.
offset by 1 cycle per day and likely result from daily

sampling aliases; results are shown for B-band, however, similar findings were observed in V and Ic. Data
subsets intentionally stripped of measurements
around the primary minimum or residuals remaining
after Roche modeling were no more revealing of any
underlying pulsations. Successive pre-whitening did
not expose any other statistically meaningful (s/n > 4)
oscillations in the frequency range (5-80 d-1) expected for Sct-type pulsations (Breger, 2000) according to the criterion proposed by Breger et al.
(1993). It is entirely possible, however, that intrinsic
oscillations were not detected due to insufficient precision during photometric sampling given that low
amplitude Sct-type pulsations may not exceed
0.06 mag (Christiansen, 2007).
3.3.2. Oscillating Frequencies of the Primary Star

Recent evidence suggests that AO Ser not only
varies extrinsically by mutual eclipses but also intrinsically by Sct-type pulsation of its primary component (Kim et al., 2004; Zavros et al., 2008). To this
end, we investigated the possibility that such oscillation could be detected in our data. Frequency analysis
by classical Fourier methods with Period04 (Lenz
and Breger, 2005) and Peranso did not convincingly
expose any underlying periodicity other than that
expected from the dominant orbital period. The spectral window (Figure 9) exhibits strong side-bands
134
Figure 5. AO Ser light curve (B-band) synthesis determined by PHOEBE after simultaneous fit of all photometric data from three passbands (B, V, and Ic). Roche
model residuals are adjusted by a fixed amount to display them in the same chart.
AO Ser Alton et al.
4. Conclusions
Figure 6. AO Ser light curve (V-band) synthesis determined by PHOEBE after simultaneous fit of all photometric data from three passbands (B, V, and Ic). Roche
Figure 7. AO Ser light curve (Ic-band) synthesis determined by PHOEBE after simultaneous fit of all photometric data from three passbands (B, V, and Ic). Roche
Figure 8. Three-dimensional rendering of the AO Ser

binary system at phase 0.25.
CCD (B, V, and Ic filters) photometric readings

were used to revise the orbital period for AO Ser and
calculate an updated ephemeris. A detailed period
analysis using O-C residuals generated from observations spanning 80 years revealed a very complex pattern of changes. Following at least seven decades of a
constantly increasing orbital period, a sudden decrease (0.844 s) probably occurred around the beginning of 1998. Assuming no mass loss from the system and conservation of angular momentum, on
average the minimum amount of stellar material
transferred during that period exceeded 50% the mass
of Earth.
Since then, the O-C residuals suggest that
AO Ser is presently undergoing further changes
caused by a yet to be determined phenomena or combination of effects. Evidence points to cyclic, if not
sinusoidal behavior in the O-C diagram but only up
to 1998; however, this same cyclic periodicity (~36
y) is not obvious within subsequent O-C residuals
available for this system between 1998 and 2011.
At this time in the absence of any supporting
data from a spectroscopic study or speckle interferometry it would be too early to assign any single reason or combination of phenomena that would unequivocally include the light travel time effect as an
explanation to this very complex O-C diagram for
AO Ser.
A Roche model based on the W-D code provided
a simultaneous theoretical fit of light curve data in
three colors (B, V, and Ic). The photometric mass
ratio estimated (q = 0.235) from W-D analyses is at
odds with a theoretically computed value (0.45) published in 1980 (Brancewicz and Dworak), but consistent
with
that
recently
published
by
Yang et al. (2010). Despite lacking critical spectroscopic or radial velocity data for this binary system,
the light curve analysis herein based on a revised B-V
color-index value for AO Ser provide strong evidence
for an Algol-type binary system composed of an A7V
main-sequence primary and K1 to K3 subgiant secondary. This classification is much cooler than previously considered by all other investigators. Finally,
Fourier analysis did not convincingly reveal any underlying periodicity other than that expected from the
dominant orbital period. However, it is possible that
the magnitude of any putative Sct-type pulsation is
below the limits of detection for the equipment used
in this investigation.
135
AO Ser Alton et al.
Borkovits, T., Van Cauteren, P., Lampens, P., et al.

(2008). New and Archive Times of Minima of
Eclipsing Binary Systems. IBVS 5835, 3.
Bradstreet, D. H., Steelman D. P. (2002). Binary
Maker 3.0 An Interactive Graphics-Based Light
Curve Synthesis Program Written in Java. Bull.
A.A.S. 34, 1224.
Brancewicz, H. K., Dworak, T. Z. (1980). A catalogue of parameters for eclipsing binaries. Acta
Astron. 30, 501.
Figure 9. Spectral window for AO Ser exhibits strong
side-bands offset by 1 cycle per day resulting from daily
sampling aliases; results are shown for B-band residuals following Roche modeling, however, similar findings
were observed in V and Ic.
5. Acknowledgements
Special thanks are due to the NASA Astrophysics Data System hosted by the Computation Facility
at the Harvard-Smithsonian Center for Astrophysics
for providing convenient access to published literature. This research has also made use of the NASA/
IPAC Infrared Science Archive, which is operated by
the Jet Propulsion Laboratory, California Institute of
Technology, under contract with the National Aeronautics and Space Administration. The contribution
of all the dedicated folks contributing to or managing
the variable star time-of-minima data housed at the
AAVSO, B.R.N.O., BBSAG, VSOLJ, and IBVS
websites is gratefully acknowledged. This investigation has also made use of the SIMBAD database,
operated at CDS, Strasbourg, France and the Bundesdeutsche Arbeitsgemeinschaft fr Vernderliche
Sterne (BAV) website in Berlin, Germany.
Brt, L., Zejda, M., Svoboda, P. (2007). B.R.N.O.

Contributions #34. Open Eur. J. Var. Stars 74, 65.
Brt, L., melcer, L., Kukov, H., et al. (2008).
B.R.N.O. Times of minima. Open Eur. J. Var.
Stars 94.
Breger, M. (2000). Asteroseismology of Delta Scuti
Stars. Balt. Astron. 9, 149.
Breger, M., Stich, J., Garrido, R., et al. (1993). Nonradial Pulsation of the Delta-Scuti Star BU Cancri in
the Praesepe Cluster. Astron. Astrophys. 271, 482.
Budding, E., Demircan O. (2007). Introduction to
Astronomical Photometry. Cambridge University
Press, New York.
Christiansen, J. L., Derekas, A., Ashley, M. C. B., et
al. (2007). The first high-amplitude Scuti star in
an eclipsing binary system. MNRAS 382, 239.
Diethelm, R. (2009). Timings of Minima of Eclipsing Binaries. IBVS 5894, 7.
6. References
Diethelm, R. (2011). Timings of Minima of Eclipsing Binaries. IBVS 5992, 16.
Alton, K. B. (2010). A Unified Roche-Model Light

Curve Solution for the W UMa Binary AC Bootis.
JAAVSO 38.
Doru, S. S., Erdem, A., Dnmez, A., et al. (2009).

New Times of Minima of Some Eclipsing Binary
Stars. IBVS 5893, 2.
Applegate, J. H. (1992). A mechanism for orbital

period modulation in close binaries. Ap. J. 385,
621.
Doru, S. S., Erdem, A., Aliavu, F., et al. (2011).

CCD Times of Minima of Some Eclipsing Variables. IBVS 5988, 2.
Baldwin, M., Samolyk, G. (2007). Observed Minima Timings of Eclipsing Binaries No. 12. AAVSO,
Cambridge, MA.
Eddington, A.S. (1926). The reflection effect in

eclipsing variables. MNRAS 86, 320.
Biermann, P., Hall, D. S. (1973). Astron. Astrophys.

27, 249.
136
Fitzgerald, M. P. (1970). The Intrinsic Colours of

Stars and Two-Colour Reddening Lines. Astron.
Astrophys. 4, 234.
AO Ser Alton et al.
Flower, P. J. (1996). Transformations from Theoretical Hertzsprung-Russell Diagrams to ColorMagnitude Diagrams: Effective Temperatures, B-V
Colors, and Bolometric Corrections. ApJ, 469, 355.
Ghedini, S. (1981). A method for evaluating the
epoch of minimum of an eclipsing variable - the sliding integrations. Societa Astronomica Italiana 52,
633.
Hall, D. S. (1989) The relation between RS CVn
and Algol. Space Science Reviews 50, 219.
Harmanec, P. (1988). Stellar masses and radii based
on modern binary data. Bull. Astron. Inst. Czechosl.
39, 329.
Hoffmeister, C. (1935). 162 neue Verderliche.
Astron. Nachr. 255, 401.
Hbscher, J., Lehmann, P. B., Monninger, G., et al.
(2010). BAV-Results of Observations Photoelectric Minima of Selected Eclipsing Binaries and
Maxima of Pulsating Stars. IBVS 5918, 10.
banolu, C., Soydugan, F., Soydugan, E., Derviolu, A. (2006). Angular momentum evolution of
Algol binaries. MNRAS 373, 435.
Koch, R. H. (1961) Times of minimum light for
several eclipsing binaries. Astron. J. 66, 35.
Kim, S.-L., Kang, Y. B., Koo, J.-R., et al. (2004).
Discovery of a short-periodic pulsating component
in the Algol-type eclipsing binary system AO Ser.
IBVS 5538, 1.
Lucy, L. B. (1967). Gravity-Darkening for Stars

with Convective Envelopes. Z. Astrophys. 65, 89.
Manzoori, D. (2008). The O-C curve analysis and
simultaneous light curve solutions of classical Algol
system U Cephei. Ap&SS 318, 57.
Miku, H., Dintinjana, B., Pra, A., et al. (2002).
Period Change and Surface Activity of the Eclipsing
Binary UV Leonis. IBVS 5338.
Mkrtichian, D. E., Kusakin, A. V., Rodriguez, E., et
al. (2004). Frequency spectrum of the rapidlyoscillating mass-accreting component of the Algoltype system AS Eri. Astron. Astrophys. 419, 1015.
MPO Canopus 1996-2008. MPO Canopus and
PhotoRed Installation Guide and Reference Manual.
Bdw Publishing, Colorado Springs, CO 80908.
Nelson, R. H. (2007). Minima2002, 2006: Astronomy Software by Bob Nelson URL:
http://members.shaw.ca/bob.nelson/software1.htm.
Nelson, R. H. 2009, WDwint Version5.6a: Astronomy Software by Bob Nelson URL:
http://members.shaw.ca/bob.nelson/software1.htm.
Parimucha, ., Dubovsk, P., Baludansk, D., et al.
(2009). Minima Times of Selected Eclipsing Binaries. IBVS 5898, 8.
Pop, A. (2005). On the Orbital Period Modulation of
the Eclipsing Binary System Y Leonis. Proceedings
of ASP Conference Series 335, 263.
Kreiner, J. M. (2004). Up-to-Date Linear Elements

of Eclipsing Binaries. Acta Astron. 54, 207.
Press, W., Teukolsky, S. A., Flannery, B. P., Vetterling, W. T. (1992). Numerical Recipes in
FORTRAN. Cambridge Univ. Press. Cambridge.
Kwee, K. K. and van Woerden, H. (1956). A

method for computing accurately the epoch of minimum of an eclipsing variable. B.A.N. 12, 327.
Pra, A., Zwitter, T. (2005). A Computational Guide

to Physics of Eclipsing Binaries. I. Demonstrations
and Perspectives. Ap. J. 628, 426.
Kwee, K. K. (1958). Investigation of variations in

the period of sixteen bright short-period eclipsing
binary stars. B.A.N. 14, 131.
Qian, S. (2002). Orbital Period Studies of Two Algol-Type Eclipsing Binary Systems: TY Pegasi and
X Trianguli. PASP 114, 650.
Lampens, P., Kleidis, S., Van Cauteren, P., et al.

(2010). New Times of Minima of 36 Eclipsing Binary Systems. IBVS 5933, 3.
Rensbergen, W. van De Greve, J. P., Mennekens, N.,

et al. (2011). Mass loss out of close binaries. II.
Astron. Astrophys. 528, A16.
Lenz, P., Breger, M. (2005). Period04 User Guide.

Commun. Asteroseismol. 146, 53.
Rucinski, S. M. (1969). The Proximity Effects in

Close Binary Systems. II. The Bolometric Reflection
Effect for Stars with Deep Convective Envelopes.
Acta Astron. 19, 245.
137
AO Ser Alton et al.
Samolyk, G. (2008). Recent Minima of 184 Eclipsing Binary Stars. JAAVSO 36, 186.
Schlegel, D., Finkbeiner, D., Davis, M. (1998). Application of SFD Dust Maps to Galaxy Counts and
CMB Experiments. Wide Field Surveys in Cosmology.14th IAP Meeting, May 26-30, Paris.
Schwarzenberg-Czerny, A. (1996). Fast and Statistically Optimal Period Search in Uneven Sampled Observations. Ap. J. 460, L107.
Surkova, L. P., Svechnikov, M. A. (2004). Semidetached eclipsing binaries (Surkova+, 2004). VizieR On-line Data Catalog: V/115.
van Hamme, W. (1993). New limb-darkening coefficients for modeling binary star light curves. Ap.
J. 106, 2096.
von Zeipel, H. (1924). Radiative equilibrium of a
double-star system with nearly spherical components. MNRAS 84, 702.
Warner, B. D. (2007). Initial Results of a Dedicated
H-G Project. Minor Planet Bulletin 34, 113.
Wilson, R. E., Devinney, E. J. (1971). Realization
of Accurate Close-Binary Light Curves: Application
to MR Cygni. Ap. J. 166, 605.
Wilson, R. E. (1979). Eccentric orbit generalization
and simultaneous solution of binary star light and
velocity curves. Ap. J. 234, 1054.
Yang, Y.-G., Hu, S.-M., Guo, D.-F., Wei, J.-Y., Dai,
H.-F. (2010). Photometric Properties for Selected
Algol-type Binaries. II. AO Serpentis and V338 Herculis. Astron. J. 139, 1360.
Zavros, P., Tstantilas, S., Rovithis-Livaniou, E.
(2008). AO Serpentis: Observations and Interpretations. Romanian Astronomical Journal Suppl. 18,
113.
138
Light at Night Mapping Project Craine et al.
The Light at Night Mapping Project:

LAN MAP 1, the Tucson Basin
E.R. Craine
Western Research Company, Inc. and STEM Laboratory, Inc.
3275 West Ina Road, Suite 217, Tucson, AZ 85741
ercraine@stemlab.org
B.L. Craine
Western Research Company, Inc. and STEM Laboratory, Inc.
3275 West Ina Road, Suite 217, Tucson, AZ 85741
blcraine@wrc-inc.com
P.R. Craine
STEM Laboratory, Inc.
Tucson, AZ 85741
prcraine@gmail.com
E.M. Craine
Tucson, AZ 85741
ecraine@dakotacom.net
Abstract
Tucson, Arizona, once billed as the Astronomical Capital of the World, has long been home to at least ten major
astronomical institutions and facilities. The region also hosts numerous productive amateur observatories and
professional-amateur astronomical collaborations. In spite of the implementation of progressive night time lighting codes, the continued growth of the region has arguably deprived Tucson of its title, and threatens the future
of some if not all of these facilities. It has become apparent that there are several difficulties in regulating this
lighting environment. It is not easy to model the actual effects of new or changed lighting fixtures, there are compelling economic conflicts that must be considered, and adherence to various guidelines is often ignored. Perhaps the most fundamental problem is that there have historically been no comprehensive measures of either
light at night or sky brightness over the extended growth areas. What measurements do exist are inhomogeneous and poorly accessible spot measurements at some observatory sites. These have little to tell us about the
actual light distributions in the overall region, and rarely are informative of the specific light sources that offend
the observatory sites. Tucson remains, for the time, an important astronomical resource. Because of its astronomical and lighting code circumstances, it is an interesting and valuable laboratory for studying these issues. In
this paper we introduce an innovative new 5-year project to comprehensively map both sky brightness and associated artificial lighting over extended areas of development in the vicinity of important astronomical institutions.
We discuss the various vectors employed in data collection; we outline the protocols used for each methodology,
give examples of the data collected, and discuss data analysis and conclusions. This program has been underway since January 2012, and has already produced results of interest to professional and amateur astronomers
alike.
1. Introduction
During the past 15-20 years there has been a
steady increase in collective sensitivity to issues surrounding the quality of artificial light at night. Prior
to this time, relatively little thought was given to
these issues, but now whole communities are enacting lighting codes directed at requiring efficient, effective, and non-intrusive artificial lighting. In many
cases the net results of these codes are not well un-
derstood, nor is it clear how to measure compliance,

or identify and verify mitigation procedures. A common approach to planning is to implement unverified
computer models as a best guess at the resultant impact of lighting projects, but systematic, high resolution measurements of real effects are rare or nonexistent.
The Light at Night Mapping Project (LAN
MAP) is a unique program to address this problem.
The prototype LAN MAP project is focused on the
139
Tucson, AZ urban area as a demonstration of the utility of actual measurement of community wide lighting output, and is known as LAN MAP 1.
An approximately 1,000 square mile area, centered roughly on Davis Monthan AFB, will be
mapped from the air by collecting sky brightness data
in the survey. These data will be collected using
STEM Laboratory, Inc. (STEM) Sky Brightness Meter (SBM) and LAN Monitor (LANM), single channel and multispectral photometers. The data collection will be targeted on multiple clear, moonless,
nights and the airborne data will be both up- and
down-looking measurements.
The LAN MAP 1 survey region is shown in
Figure 1. The data collection from aircraft will be at
an altitude of 8,000 ft MSL. This will allow safe
clearance of the mountains in the flight operations
area, and will provide data also relevant to many
astronomical observatories in the surrounding area.
The airborne data will be supplemented by both
mobile and static ground-based data. These data can
be used to scale the airborne observations to
comparable ground intensities and allow for future
comparisons from ground based measurements.
Each survey session will be repeated periodically
for the projected five year duration of the project
resulting in the most detailed and comprehensive
LAN survey analysis yet undertaken.
We will also obtain limited ground based
photography of some of the encompassed light
installations, both to illustrate the good and bad
installations and to show the changes in lighting over
the project time line.
2. Background
Tucson, Arizona is a southwestern desert community that grew from a population of 120,000 in
1950 to 220,000 in 1960, by which time Kitt Peak
National Observatory had been installed west of the
town. In 2010 the city population was about 546,000
and the urban metropolitan area was over 1.02 million, with a current annual growth rate of about 4%.
This is now a sprawling community encompassing numerous suburban areas and adjacent cities. City
lights, once confined to a few square miles, now
spread over an area well in excess of 1,000 square
miles.
The city is largely situated on the floor of a desert basin (about 2,600 ft MSL), nearly completely
ringed by mountain ranges of up to about 7,000
8,000 ft MSL. Many of these mountain peaks host
facilities and multiple telescopes, of approximately a
dozen major astronomical observatories.
140
In addition to the professional facilities, there are

numerous privately owned observatories, many engaged in productive astronomical research, scattered
throughout the region. Some of these reside close to
the several National Forest, State Park, and wildlife
preserve lands in the region. All are concerned with
increases in LAN that accompany the population
growth and expanded areas of development.
Figure 1. FAA Aeronautical Chart, Phoenix Sectional,

Tucson, AZ surroundings, showing the LAN MAP 1 flight
grid throughout the Tucson basin.
The LAN MAP 1 project is using multiple mobile photometers to provide the first comprehensive,
quantitative monitoring of LAN in the Tucson Basin
area that impacts all of these light sensitive activities.
3. SBM and LANM Photometers

Mobile, temporal sky brightness measurements
benefit from access to an integrated photosensor,
Global Positioning System (GPS) receiver, data logger, microcontroller and power source. Such a system
has further value if it can operate in an automatic
autonomous mode. Compact size and light weight are
additional advantages. With this specification in hand
the design and construction of a family of such units
was undertaken by STEM Laboratory (Tucson, AZ).
We briefly summarize three members of the family in
this section: SBM Model 4, LANM Model 5, and

LANM Model 6.
3.1 SBM Model 4

The SBM Model 4 is the first of the photometers
that STEM completed beyond the prototype stage,
and used in a variety of mobile ground and air based
sky brightness data collection projects. This unit uses
a single light-to-frequency silicon photodiode detector. In its stock form it is optically filtered with an
IRC-20 infrared cut filter (Sunex), as shown in Figure
2. The unit operates with an integrated GPS module
which is polled every two seconds to yield date, time,
geographic coordinates, and altitude. The GPS output
is a standard National Marine Electronics Association
(NMEA) GPS data suite; c.f. SIRF (2005) and
NMEA (2011). Simultaneously, the photodiode is
sampled for light intensity (mag/arcsec2 and cd/m2).
These quantities are coordinated in the onboard microcontroller and written to an onboard micro-SD
card writer, from which they can be transferred to a
project data analysis computer upon completion of
the data collection flight.
3.2 LANM Model 5

The Light at Night Monitor, LANM Model 5, is
built on the same platform as the SBM Model 4, but
has two of the silicon photodiode detectors that are
read simultaneously. The GPS data polled are the
same as the SBM Model 4, with the exception of the
inclusion of the speed of the vehicle carrying the
LANM Model 5. Also, there are two channels of
photometric data rather than just one. The optical
filtration is generally an IRC-20 infrared cut filter;
otherwise the spectral response is that of the silicon
photodiode. Another distinguishing feature of the
LANM Model 5 is that the two photodiodes are
aligned with a separation of 180o on the optical axis,
ensuring that they are looking in opposite directions.
For purposes of airborne operations, this ideally results in one detector looking at the zenith and the
other simultaneously observing the nadir. Thus, the
LANM Model 5, operated in airborne mode, provides
a good contrast between sky brightness overhead, and
the distribution of light intensity on the ground that is
scattered upward to contribute to sky brightness.
Figure 2. The spectral response curve of the SBM Model

4 Sky Brightness Meter and the LANM Model 5 photometer.
3.3 LANM Model 6

The LANM Model 6 is electronically identical to
the LANM Model 5. The distinguishing feature of the
LANM Model 6 is that the two detectors are aligned
to look simultaneously at the same target field, and
the optical filtration is different from the other units.
We are using the LANM Model 6 somewhat as a
custom system where we can insert specific optical
filters of interest. Our early work involves two filter
combinations: 1) IRC-21 infrared cut + Wratten 47
and 2) IRC-21 infrared cut + Wratten 58; yielding
blue (B) and visual (V) band data for each scene, as
shown in Figure 3. The LANM Model 6 is thus most
useful for defining a color (such as B-V) for a relatively bright area (since the introduction of additional
optical filters reduces the sensitivity of the photometric system).
Figure 3. The spectral response curve of the LANM

Model 6 photometer.
141
4. Aircraft Photometer Installations

The aircraft used for collecting the initial airborne data was a Mooney M231. Subsequent flight
programs were conducted with a Cessna 182Q. Both
are high-performance aircraft capable of cruise
speeds over 150 KIAS (approximately 170 mph).
Such speeds can impose high aerodynamic loads on
externally mounted devices and space the data points
over a widely dispersed area. A speed of approximately 100 mph over the ground has been found to
work well. This requires the airspeed to be adjusted
to the particular atmospheric conditions on each
flight. The greatest effect comes from the winds,
which may be much higher at altitude than at ground
level. If the wind direction is parallel to the grid legs
then the total groundspeed variation will be twice the
wind speed. This can easily introduce variations of
data point spacing of 20% if not corrected. An additional source of variation comes from the fact that the
groundspeed will be higher for a given indicated airspeed as the airplane is operated at higher altitudes.
High-performance aircraft have been found to be
particularly well suited to gathering data because they
can quickly cover the distance between the base airport and the survey area and because they generally
have sufficient fuel capacity for long duration flights.
They have tended to be more stable than lighter aircraft, which eases pilot fatigue and yields higher
quality data.
The photometers must be mounted to the airplane such that no part of the aircraft structure obscures the detectors field of view. All aircraft approved for night operations carry a variety of lights
which have the potential to impact the measurements.
The lights are usually arranged such that they will not
affect the pilot's night vision, so mounting the photometer near the crew cabin can help minimize the
effect of aircraft lighting. The surface of the airplane
can be sufficiently reflective to bounce light (ambient
or aircraft) into the photometer's field of view so care
must be taken in planning the photometer mounting.
Two installations have been used: a cantilever
mount and an in-wing mount. The cantilever mount is
a dedicated bracket which bolts to the bottom of the
wing, near the leading edge, and holds the photometer ahead of the wing. This allows the photometer
sensors to face upward and downward without interference from the airframe. The drawbacks are that the
bracket must be carefully designed to prevent adverse
aerodynamic effects on the airplane. The bracket itself is subjected to aerodynamic forces as well as
maneuvering forces (G-forces). The maximum operating speed of the airplane is greatly limited with a
cantilever mount in place. Additionally the airplane
142
must be flown very gently. The existing cantilever

mount has been tested at airspeeds up to 105KIAS
and forces of +1.3/-0.5G.
The in-wing mount is particularly well suited for
down-looking installations, though we have engineered designs for upward looking installations as
well. As currently implemented, such a mount introduces no aerodynamic changes and does not impact
the aircraft performance. This can be useful for making measurements at long distances from the aircraft
base of operations since the full cruising speed of the
airplane can be utilized.
Figure 4. LANM Model 5 and LANM Model 6 mounted on

a C-182Q aircraft wing. The former is on the cantilevered
mount, and the latter inside the blue wing panel.
In both cases the photometers must be mounted

so that they will be oriented vertically during flight.
This means that the dihedral angle of the wings must
be compensated for, and average angle of attack in
cruise phase flight must be considered. Note that this
may be different from the level attitude when the
airplane is on the ground. It is not sufficient to simply
adjust the mount to level the SBM while on the
ground. The external mounting bracket has been
aligned with the airflow streamlines at the cruise
condition to minimize aerodynamic loads.
We have simplified determination of angular
displacements of the photometers during data collection by developing a three axis accelerometer device
called the Time-Orientation-Position (TOP) Logger.
This device automatically logs time, magnetic heading, pitch, roll, latitude, and longitude of the aircraft.
This data stream can be integrated with the photometry data stream to indicate when the photometers are
looking off target, and those data points can be
removed from the analyzed data set.
5. LAN MAP 1 Protocols

Details of the airborne flight protocols generally
employed by STEM can be found in Craine and
Craine (2011). We summarize the key sequences for
LAN MAP 1 below:
The flight path for the survey segment to be
flown is programmed into a ForeFlight moving
map navigation system.
shown here were obtained at 8,000 ft MSL. Flights

were timed to ensure that the aircraft arrived at the
data collection Initial Point (IP) shortly after astronomical twilight. The flights were conducted in the
complete absence of moonlight, and were completed
prior to Pima County lighting code curfew for commercial lighting.
In Figures 5 and 6 we show sky brightness isophotal maps plotted on Google Earth aerial images of
Tucson, Arizona.
Final weather and astronomical boundary condition checks are made.

Photometers are installed and initiated; GPS lock
and data logging are confirmed.
Aircraft pre-flight checks are completed.
The flight is launched and communication is
maintained with local air traffic control.
The survey path is flown typically for about 3
hours.
A simultaneous ground mobile sky brightness
measurement path is driven using a SBM Model
4 mounted on a car.
Upon completion of the flight pattern the aircraft
is returned to its home base and the photometers
are shut down.
Data files are all downloaded and storage folders
are created in the Sky Brightness Data Archive
(SBDA).
Figure 5. Sky brightness data over the NW quadrant of

Tucson, Arizona plotted on a Google Earth aerial image.
This plot is constructed with 15 levels of intensity reso2
lution represented as mag/arcsec . The light yellow is
the bright end of the scale.
Data are pre-processed.

Data are available for processing and analysis.
Flights are repeated as needed to achieve the desired spatial density of observations.
6. Project Status
During 2012, STEM initiated a comprehensive,
community wide, light at night mapping program for
the Tucson, Arizona area. In this section we show
examples of some of the flight test data collected in
anticipation of the official start of LAN MAP 1. The
flight test data were collected over the northwestern
most quadrant of the LAN MAP 1 survey area.
These data were collected with a LANM Model
5 photometer, using both zenith and nadir looking
channels. The data were optically constrained by the
silicon detector sensitivity curve as modified by an
IRC-20 infrared cut filter as shown in Figure 2.
The test flights were for patterns flown at altitudes of 5,000, 6,500, and 8,000 ft MSL; the data
Figure 6. The same flight as Figure 5, plotted for the

down-looking data.
To understand these figures one must appreciate

that the isophotal contours indicate the brightness of
the sky as seen by an observer looking up from each
143
geographical location of interest on the aerial photograph.

Even casual comparison of these up- and downlooking plots indicates localized bright areas on the
ground that clearly contribute to anomalously bright
areas in the sky. It is a simple matter to use the display software to zoom through the graphics overlay
at those points to identify the specific contributing
light sources.
In 2012 February, the first of the formal LAN
MAP 1 data collection flights was conducted. The
ground path for this flight is shown in Figure 7. The
path data have been truncated to include only that
portion of the flight conducted in the LAN MAP 1
survey area. The parallel flight legs on this flight
were spaced three miles apart, and the intervening
gaps will be filled by later flights. Slight deviations in
the path are a consequence of Air Traffic Control
vectoring in the vicinity of Tucson International Airport and Davis Monthan AFB to accommodate a simultaneous flight test program of the Boeing 787
Dreamliner.
Figure 7. The ground track of the first of the LAN MAP 1

data flights, plotted on a Google Earth satellite image.
Quick look representations of these initial data

can be seen in Figures 8 and 9. In Figure 8 we show
an isophotal map of the sky brightness data as seen
over a substantial portion of the Tucson Basin. Figure
9 shows the same region, but looking down at the
lights on the ground. These data are much more
sparsely distributed that what we expect for the completed survey, but are still quite suggestive of the
lower frequency patterns of sky brightness over Tucson.
In the case of Figures 8 and 9, the data are
binned at 15 levels. This is done to provide a more
realistic picture of the smoothed appearance of both
sky brightness and ground lights. Because the survey
grid is only partially filled on this first flight, binning
the data to provide more contour intervals introduces
144
numerous spurious features that can be very misleading in interpreting the images.
Figure 8. Sky brightness distribution over the Tucson

Basin as seen from the flight of Figure 7.
Figure 9. Ground lighting distribution as seen from the

flight of Figure 7.
7. Conclusion
The tools and protocols developed by STEM to
enable low-altitude wide-area-survey automatic photometry of both the ground and the sky are offering a
powerful new way to look at issues of light trespass.
These protocols can be applied to ground, air, and
sea-going vehicles, and offer community planners,
developers, and managers of light sensitive facilities
a unique means of evaluating the impact and evolution of light at night.
Even a preliminary look at the LAN MAP 1 data
demonstrates areas of concern for local astronomical
facilities. Contrary to conventional wisdom on the
population expansion of Tucson and its suburbs, the
brightest expansion of the Tucson light dome is to the
northeast, where it clearly poses a threat to the University of Arizona observatories on Mt. Bigelow and
Mt. Lemmon. Conversely, observatories on Mt.
Wrightson, to the south of Tucson, remain in a fairly
dark region. Sky brightness over Mt. Wrightness is
quite likely to evolve for the worse as development
expands down the I-19 corridor, and leaps over the
Santa Rita Mountains to the southeast.
It will be of interest to watch annual changes in
LAN MAP 1 data to the west of Tucson as well.
There, development has for several years progressed
west of the Tucson Mountains, and is steadily filling
Avra Valley and points south and west, posing significant threats to Kitt Peak. Kitt Peak has almost
certainly seen the end of large telescope installation,
but preservation of existing facilities will remain a
priority for at least a few more years.
8. Acknowledgements
The authors would like to thank Scott Fouts for
some of the piloting duties and Ed Tuohy for mechanical assistance with one of the airplanes used in
the data collection. We gratefully acknowledge an
anonymous corporate donor for very generous grant
support to undertake this program.
9. References
Craine, E. R., Craine, B. L., Craine, E. M. (2011).
The Sky Brightness Data Archive (SBDA). in Proceedings for the 30th Annual Symposium on Telescope
science (Warner et al., eds). pp 45-52. Society for
Craine, P. R., Craine, E. R. (2011). Airborne Light
at Night (LAN) Photometry Protocols. STEM
TechRep-11-1203. STEM Laboratory, Tucson, AZ.
NMEA (2011). www.nmea.org
SIRF (2005). NMEA Reference Manual SiRF Tech,
Inc., San Jose, CA.
145
146
Observing Double Stars Genet et al.
Observing Double Stars

Russell M. Genet
California Polytechnic State University and Cuesta College
4995 Santa Margarita Lake Road, Santa Margarita, CA 93453
RussMGenet@aol.com
B.J. Fulton, Federica B. Bianco, John Martinez
Las Cumbres Observatory Global Telescope
6740B Cortona Drive, Ste. 102, Goleta, CA 93177
fbianco@lcogt.net, bjfulton@lcogt.net, jmartinez@lcogt.net
John Baxter, Mark Brewer, Joseph Carro, Sarah Collins, Chris Estrada,
Jolyon Johnson, Akash Salam, Vera Wallen, Naomi Warren
Cuesta Community College, Highway 1, San Luis Obispo, CA 93403
Jrbax@aol.com, mrk.a.brewer@gmail.com, jcarro@charter.net, sjcollins94@gmail.com,
pr1est0112@hotmail.com, jolyonjohnson@gmail.com, akashumm@sbcglobal.net,
jvwallen@charter.net, renoman599@gmail.com
Thomas C. Smith
Dark Ridge Observatory, 701 NM Highway 24, Weed, NM 88354
tcsmith@darkridgeobservatory.org
James D. Armstrong
University of Hawaii, Institute for Astronomy, and Las Cumbres Observatory Global Telescope
34 Ohiaku Street, Makawao, Maui, Hawaii 96768
jd@ifa.hawaii.edu
Steve McGaughey
Haleakala Amateur Astronomers, 5 Kamauhalii Way, #3-A, Wailuku, Maui, HI 96793
stevemcgee2@yahoo.com
John Pye
University of Hawaii Maui College, 310 W. Kaahamanu Ave., Kahului, HI 96732
jpye@hawaii.edu
Kakkala Mohanan, Rebecca Church
University of Hawaii, Leeward Community College, 96-045 Ala Ike St., Pearl City, HI 96782
mohanastrogglcc@gmail.com, rjchurch@hawaii.edu
Abstract
Double stars have been systematically observed since William Herschel initiated his program in 1779. In 1803 he
reported that, to his surprise, many of the systems he had been observing for a quarter century were gravitationally bound binary stars. In 1830 the first binary orbital solution was obtained, leading eventually to the determination of stellar masses. Double star observations have been a prolific field, with observations and discoveries
often made by students and amateursroutinely published in a number of specialized journals such as the Journal of Double Star Observations. All published double star observations from Herschels to the present have
been incorporated in the Washington Double Star Catalog. In addition to reviewing the history of visual double
stars, we discuss four observational technologies and illustrate these with our own observational results from
both California and Hawaii on telescopes ranging from small SCTs to the 2-meter Faulkes Telescope North on
Haleakala. Two of these technologies are visual observations aimed primarily at published hands-on student
science education, and CCD observations of both bright and very faint doubles. The other two are recent technologies that have launched a double star renaissance. These are lucky imaging and speckle interferometry,
both of which can use electron-multiplying CCD cameras to allow short (30 ms or less) exposures that are read
147

out at high speed with very low noise. Analysis of thousands of high speed exposures allows normal seeing limitations to be overcome so very close doubles can be accurately measured.
1. Introduction
2. Early Double Star Observations
Observing double stars has engaged generations

of student, amateur, and professional astronomers for
the better part of three centuries. Over 100,000 double stars have been discovered. As each year goes by,
new doubles are discovered and the observations of
already known doubles continue to accumulate.
A wide range of telescopes, instruments, and
techniques are currently used to measure the separations and position angles of double stars. Telescopes
range from a 3-inch Tasco refractor equipped with a
$150 astrometric eyepiece observing relatively wide
optical doubles (Grisham et al., 2008) to a 10-meter
Keck reflector with adaptive optics/laser guide star
observing very close late-M binaries (Dupuy et al.,
2012). Instruments include visual eyepieces and filar
micrometers, regular CCD cameras, and fast, low
noise EMCCD frame transfer cameras. Advanced
techniques include blind or other deconvolution
strategies, lucky imaging, and speckle interferometry.
Although beyond the scope of this paper, the absolute cutting edge with respect to close separation
(and hence short binary periods) is amplitude interferometry arrays such as the Center for High Angular
Resolution Array (CHARA) on Mt. Wilson (McAlister 1999). Six 1-meter telescopes in a Y configuration interferometer provide milli-arcsecond resolution allowing binaries with orbital periods of just a
day to be observed. In this paper we will, after a brief
history of double star observations, describe two
spreadsheets we have developed to aid us in selecting
doubles for observation. We will then describe both
visual and CCD observations we make with students
that have both scientific and educational goals. Finally, we will describe two advanced techniques we
are exploringlucky imaging and speckle interferometryto aid us in observing close double stars.
Our observations were made on 8- to 32-inch telescopes in California, and 9-inch to 2- meter telescopes in Hawaii.
Most of the coauthors of this paper will be giving talks at the Maui International Double Star Conference in February 2013. In many cases they will be
meeting each other for the first time, and will also be
meeting other double star observers from around the
planet. For details on this conference see
www.AltAzInitiative.org.
Galileo thought that the sun was the center of the

universe and stars were just points of light on a
nearby sphere. For those, contrary to Galileo, who
thought stars were distant suns, he suggested they
should look for and observe close double starsone
bright and one faint that happened, by chance, to appear close together along the line-of-sight from Earth.
If the bright star was nearby and the faint one far
away, then the nearby bright star should shift its position with respect to the faint star over a six month
period due to the rotation of the Earth around the sun
(parallax).
Following up on Galileos suggestion, William
Herschel began measurements of the separation of
position angles and separation of double stars in earnest in 1779. Three years later he reported his observations of 269 binaries to the Royal Society
(Herschel 1782). Continued observations led, a couple of decades later, to the somewhat unexpected
conclusion by Herschel (1803) that many double stars
were gravitationally bound binaries. Interestingly,
John Michel (1767) had earlier reported to the Royal
Society that, based on a statistical analysis of a few
bright visual double stars, he had concluded they
occurred at a higher frequency than one would expect
due to chance alignment alone and hence were
probably truly gravitationally bound stars. It is
probably a good thing that Herschel was unaware of
Michels paper or he might not have initiated his
search for and a measurement of double stars, as his
goal was determining parallaxes.
148
Figure 1. William Herschels 20-foot telescope was relocated to Table Mountain near Cape Town, South Africa,
where his son John observed double stars in the southern skies.
In the mid 1800s, while he was director of the

Dorpat Observatory in what was then Livonia, Friedrich Georg Wilhelm von Struve made micrometric
measurements of 2,714 double stars from 1824 to
1837 and published these in his work Stellarum duplicium et multiplicium mensurae micrometricae.
F.G.F. von Struves son, Otto Wilhelm von Struve,
(1897-1963), continued his fathers work using a 15inch refractor at the Pulkovo Observatory near St.
Petersburg, cataloging several thousand double stars.
Another notable double star observer was Robert
Innes at the Meteorological Observatory in Johannesburg, South Africa. He compiled in 1927 The
Southern Double Star Catalog consisting of 1,600
entries. Robert Aitken, with the help of collaborator
W.J. Hussey, published in 1932 the New General
Catalogue of Double Stars within 120 of the North
Pole consisting of 17,180 doubles. The Lick Observatory at Mount Hamilton near San Jose, CA, published the 1963 Index Catalog of Visual Double
Stars, containing 64,250 objects covering the entire
sky. In 1965 it was transferred to the United States
Naval Observatory as the Washington Double Star
Catalog (WDS) which now contains over 100,000
double stars.
Observations of the double star Zeta Ursae Majoris were indirectly used by Savary (1827) to obtain
an orbital solution, while a solution directly using all
the observations was provided shortly thereafter by
John Herschel. Orbital solutions, when combined
with stellar parallaxes and other data, led to the determination of stellar masses, a key parameter influencing stellar evolution. Parallaxes were hard to
come by until the Hipparcos satellite accurately determined a large number of parallaxes (Perryman
2011). The Sixth Orbital Catalog provides orbits for
2,187 binaries (as of 1 January 2012).
For more detailed historical expositions, please
see Aitken (1918), Herrmann (1973), Couteau
(1978), Argyle (1986), Tanguay (2003), and Hoskin
(2011).
3. Two Spreadsheets
Students in Genets astronomical research seminar (described below in the next section) have, for
many years, used Haas (2006) catalog of 2,100 doubles to pick out relatively bright, well-separated double stars for their first projects. The Haas catalog was
convenient for students to use as it listed doubles by
constellation, only provided key information, and just
included relatively bright doubles observable with
smaller telescopes. However, one had to manually
page through the catalog to select observational candidates. Furthermore, some vital information, such as
SAO and other identifiers, number of past observations, etc., was not included. Increasingly sophisticated students would prefer to use their laptops to
help them sort and select potential doubles for observation, while SAO or other identifiers are handy for
entry into todays computerized telescopes.
Joseph Carro has prepared a spreadsheet catalog
of 3,950 double stars that is ideal for student use.
Carro incorporated information from SIMBADthe
astronomical data base operated at the CDS in Strasbourg, Franceinto a series of Microsoft Excel
spread sheets that facilitate sorting and selecting doubles for study. The spreadsheets are arranged by constellation, and include star names, common identifiers, and details on each double star. Summary tables
of all 3,950 stars are sorted in various ways as a convenience for the user. A diagram showing the major
stars of each constellation is provided. Using the star
name as the key, a hyperlink to SIMBAD allows additional information, such as parallax, to be conveniently obtained. Carros catalog can be accessed at
www.AltAzInitiative.org, or you can contact him at
jcarro@charter.net. Carro is expanding his catalog to
include additional double stars and further information.
To select doubles for advanced projects, such as
multiple stars across many constellations, faint double stars, etc., it is helpful to have a way to choose
them quickly and easily. Tom Smith created a master
catalog using a relational database engine and Structured Query Language (SQL) with the Washington
Double Star Catalog (WDS) as the foundational catalog. Several other databases of stars were incorporated and relationally linked to the engine, including
the Sixth Catalog of Orbits of Visual Double Stars
(6th Orbital Catalog) and the Hipparcos 2 Catalog
(Hip2). Many of the desired data fields needed in our
studies were not contained directly in any of the
original catalogs and were synthesized using various
mathematical and astrophysical algorithms. In many
cases, the units for some of the data were different
across the various catalogs, so new fields were created with uniform units. Smith routinely generates
Microsoft Excel worksheets that are convenient for
selecting doubles for observation, data manipulations,
charting, and generating population statistics.
Genet has sorted and reduced the initially large
spreadsheets provided by Smith to short, compact
lists of doubles for specific observational programs.
This process typically involves removing columns
not of interest, narrowing the columns so the relevant
data appears all together on the screen, selecting a
range of RA and dec of interest, and sorting and selecting on such parameters as separation, delta magnitude, parallax, proper motion, spectral type, etc.
Lists have been prepared for high proper motion
149
pairs, faint pairs that include an M star, pairs with a

white dwarf, etc.
Our research is incorporating photometry as well
as the astrometry of our target double star systems.
The new AAVSO Photometric All Sky Survey
(APASS) catalog has secondary photometric standards in every field of view. This will facilitate transformations to the standard photometric systems.
Please see http://www.aavso.org/apass for details.
Figure 2. The twin 8-inch APASS-North astrographs at

Tom Smiths Dark Ridge Observatory in Weed, New
Mexico.
4. Visual Observations and Student

Education
Several different groups of students, educators,
and amateur astronomers are working together to
provide middle school, high school, and community
college students a hands-on research experience. One
group, the California Double Star Association, is
centered on a research seminar offered by Genet at
Cuesta College. This seminar consists, primarily, of
students at Arroyo Grande High School. Another
group, the newly formed Maui Double Star Association, meets at the University of Hawaiis Institute for
Astronomy in Maui, and provides a research opportunity for both middle school and high school students.
In both groups, students, using a Celestron or
Meade ruler/protractor astrometric eyepiece, measure
the separations and position angles of visual double
stars, typically following the procedures outlined by
Argyle (2003). Students observe with telescopes supplied and operated by local amateur astronomers. The
students analyze their data and write up brief papers
describing their observations, analysis, and results.
The papers are given extensive internal and external
reviews, and then appropriately rewritten. Students
present their results at scientific conferences and their
final papers are submitted to the Journal of Double
150
Star Observations for consideration by its editor,

Kent Clark. The entire process is overseen by professional educators and astronomers. See Genet et al.
(2010) for details.
The intent of the seminar is not to recruit future
astronomers, although that is certainly happening,
given the enthusiasm level of some of the students.
The purpose is to give the students an opportunity to
conduct original science research, thus discovering
how real science is done. Understanding the nature of
scientific research is useful for many different careers, as well as exercising responsible citizenship in
todays complex world.
Double star measurements were first made at
Cuesta College on Californias central coast in the
Fall of 2007 (Grisham et al., 2008, Johnson and
Genet 2007, and Johnson et al., 2008). The following
year the seminar was moved to Cuesta Colleges
South Campus at Arroyo Grande High School
(AGHS), which has been the home base for this
seminar for the past four years. Published papers,
with AGHS students primarily as coauthors, include
Alvarez et al., 2009, Brewer 2011, Estrada et al.,
2010, Estrada et al., 2011, Grisham et al., 2008,
Johnson et al., 2008, Johnson and Genet, 2007, and
Marble et al., 2008. The 2011 class was the largest to
date, with an enrollment of 17 students all high
school students divided into three teams. Each team
published a paper (Baxter et al., 2012; Fluitt et al.,
2012; Warren et al., 2011), and one team published a
second paper (Collins et al., 2011). Students may
take the seminar up to three times for college credit.
Students in their second or third year are team leaders
for the students who are enrolled for the first time. A
student research seminar was also held at Coast Union High School, in nearby Cambria (Dowdy et al.,
2009).
Figure 3. Cuesta College 2011 research seminar students pose in front of Nimbus II, a 22-inch Dobsonian
telescope built by Reed and Chris Estrada, used to observe double stars with an astrometric eyepiece.
The seminar class is conducted as a graduatelevel seminar. Several student groups are formed.
The groups are given a general directive to collect
position angle and separation data on a visual double
star pair. Specific double star systems are selected by
the students. Each group is responsible for observations, data analysis, literature research, and a written
paper. The various groups are the initial editors of
each others papers. Student research is valuable in
its own right as it adds to the knowledge base of double stars. It is also contributes to student academic
careers. It is rare indeed for a student to be a published science researcher before leaving high school,
a point recognized with respect to college admissions
and awarding of scholarships. Most of the seminars
graduates have been accepted to their first choice
college and have also received scholarships based on
their published research.
Having a local supportive community of educators and amateur astronomers has been vital. They
have donated their time to instruct and demonstrate,
to edit student work, and generally be science role
models. Also vital are mentorsoften amateur astronomers who are professional educators in fields
other than astronomy. As Joseph Carro points out, the
primary task of a mentor is to organize information
so that a coherent and meaningful learning session
can be achieved.
Based now on many years of double star seminars, the instructor and advisors note several factors
that contribute to the success of published student
research:
Students can and should complete scientific research projects in a single semester. (Once the
semester is over, students rarely complete projects.)
Double star projects are more amenable to this
short time frame than most others.
Requiring the publication of research is vital, as
are outside reviews prior to publication.
It helps to have a mix of experienced and inexperienced observers.
Student research teams can exercise their diverse
talents and experience by allocating their skills to
the various aspects of scientific research.
Based on the success of the early Cuesta College
seminars, Genet initiated a student summer workshop
in 2008 at University of Oregons Pine Mountain
Observatory east of Bend, Oregon. Although student
contact hours remained the same, the summer workshop differed from the fall seminar by being concen-
trated into a single long weekend. During the four

consecutive summers of 2008-2011, students and
science teachers from several high schools and undergraduate colleges in Washington, Oregon, Idaho,
and California participated in and successfully published double star research projects. Double star research is complex enough to challenge students without being so extensive as to require months of observation and analysis. This makes their study appropriate for students engaged in a several-day, hands-on
research workshop.
At each workshop, students were broken up into
research teams led by an experienced double star
observer. Most attendees had never engaged in quantitative research before, so they received training in
this area on the first day of the workshop prior to
taking measurements that night. This was followed
by daytime data analysis and paper writing and further observations, as required, at night.
Figure 4. Although they used smaller telescopes for

their double star observations, students and instructors
pose in front of the 32-inch telescope at the University
of Oregons Pine Mountain Observatory 2009 Student
Research Workshop.
A total of eight papers have been published in

the Journal of Double Star Observations from these
Pine Mountain Observatory workshops. Most papers
report on the separation and position angle of a double star such as Iota Botes (Schrader et al., 2010),
Beta Lyrae (Carro et al., 2012), 61 Cygni (Baxter et
al., 2011), and Delta Botis (Estrada et al., 2012).
However, Frey et al. (2011) and Alduenda et al.
(2012) reported observations of multiple star systems.
Another team measured the double star STTA 123
AB and, in determining its components to be bound
by gravity, estimated a rough orbit (Brashear et al.,
2012).
Another group, The Maui Double Star Association, consists of middle and high school students, and
amateur and professional astronomers living on the
island of Maui in Hawaii. Genet, who spends each
winter in Hawaii, helped found and is a part-time
member of the group. The students, who attend dif-
151
ferent schools, meet at the University of Hawaiis

Institute of Astronomy (IfA). IfA astronomers J. D.
Armstrong and Joe Ritter, and Haleakala Amateur
Astronomers member Steve McGaughey lead the
group.
For their visual observations they are using Steve
McGaugheys Celestron C-9.25 at the Haleakala
Amateur Astronomers observatory on the summit.
McGaughey and Genet have developed procedures
for student visual observations at the summit
(McGaughey and Genet, 2012).
LCCs Kilohoku Hale observatory is equipped

with an Optical Guidance Systems 0.5-meter f/8.2
telescope and an Apogee Alta U6 CCD camera which
was used for observations of the multiple star system
Beta 321 in Lepus (Church et al., 2012).
Figure 7. The Leeward Community College observatory

is not far from Pearl Harbor.
Figure 5. Maui Double Star Association: (back row) Eric
Rohzinski, J.D. Armstrong, McKayla Wandell, Audreanna Leatualii, and Stephen McGaughey, and (front
row), Noah Rohzinski, Aaron Rohzinski, and Isaac Sato.
Figure 8. Left to right: Rebecca Church (student), Kakkala Mohanan (LCC instructor), and Russ Genet
(visitng). The telescope is a 0.5 meter Optical Guidance
Systems Cassegrain.
Figure 6. Stephen McGaughey and his Celestron C-.925
at the Haleakala Amateur Astronomers observing location at the 10,000 foot summit of Haleakala.
5. CCD Observations and Student

Education
A program of student education via CCD observations of double stars was initiated this past winter
at the University of Hawaiis Leeward Community
College (LCC) on Oahu. LCC is one of the seven
community college campuses of the University of
Hawaiis system, with some 6,200 students. It is located to the west of Honolulu over looking Pearl
Harbor.
152
The Maui Double Star Association (MDSA) is

fortunate in that 30% of the time on the Las Cumbres
Observatory Global Telescope (LCOGT) 2-meter
Faulkes Telescope North (FTN) has been allocated to
the University of Hawaii for student education and
public outreach. Thus the MDSA is in a good position to pursue advanced double star projects using
this telescope. FTN can be conveniently operated in a
Real Time Interface (RTI) mode from the University
of Hawaiis Institute for Astronomy Maui headquarters at the base of Haleakala. Although initially
clouded out, the weather was clear on the night of
March 31st when J. D. Armstrong, Cindy Krach,
Auddreyanna Leatualii, Steve McGaughey, Aron
Rohzinski, Noah Rohsinski, and McKayla Wandell,

were at the controls at IfA, while Russ Genet participated via speakerphone.
Internet. A paper reporting these observations is being prepared.
6. Lucky Imaging
High spatial resolution is vital to obtaining accurate measurements of close visual double stars. The
atmosphere compromises image resolution, limiting
the minimum separation of the systems that can be
studied. However, when short exposureson the
order of 0.1Hz or lessare collected, one is essentially imaging instantaneous atmosphere configurations. In lucky imaging (Law et al., 2006), targets are
observed at >10Hz cadence, and only the images
least affected by the atmospheretypically only 1%
in the R band, less in bluer bandsare selected.
Those lucky exposures, once stacked, can reconstruct a highly spatially resolved image, often diffraction limited, of a sky region of one to a few tens of
arc seconds, the nominal size of an isoplanatic atmosphere patch. Note that the majority of the data is
not used in the image reconstruction, thus this technique is extremely time consuming. Thus observations must be planned for 10 to 100 times the conventional image exposure required to achieve the desired
signal-to-noise ratio.
Figure 9. Steve McGaughey, J.D. Armstrong, Russ
Genet, and John Pye stand in front of the 2-meter Faulkes Telescope North on the summit of Haleakala. An inperson observational session scheduled for that night
were clouded out but was successfully completed remotely at a later date.
Brian Mason at the U.S. Naval Observatory had

suggested that observations of late M-star doubles,
doubles with a white dwarf, and high common proper
motion pairsespecially the faint pairs only observed a single time by Luyten (1997) could be usefully observed by the 2-meter FTN. Smith and Genet
worked together to develop three observational
listsone for each of these areas. Rafael Caballero, a
member of Agrupacion Astronomical Hubble in
Spain, has suggested the observation of faint common proper motion pairs (Caballero, 2012). He developed a new list of high common proper motion
pairs for us that contained a mix of known pairs (in
the WDS) and potential new double stars that could,
if observationally confirmed, be added to the WDS as
new discoveries. Genet narrowed the list down to 9
Rafa systems (with stars from 14th to 19th magnitude) of which 7 were observed in the allotted two
hour observing window. Maui students operated the
telescope under the supervision of Armstrong and
McGaughey, while Genet participated via speakerphone, sending Aladin Previewer images via the
Figure 10. B.J. Futon, Russ Genet, and John Martinez at

the Las Cumbres Observatory Global Telescopes BOS
32-inch telescope. With Federico Bianco, they used the
EMCCD Andor Luca-R camera on this telescope for
lucky imaging and speckle interferometry observations.
Lucky imaging was used as early as the 1920s

for planetary observations. However, the typical
brightness of close double stars and multiple star systems rendered them too faint to be imaged at the necessary high cadence. The advent of high sensitivity,
low noise cameras over the last decade, particularly
EMCCDs, has unlocked the potential of this technique. Although the cost of EMCCD cameras began
in the $40,000 range, new cameras can now be pur-
153
chased for less than $8,000, and some used cameras

have been purchased for as little as $2,500.
On a 2-meter telescope, such as Faulks Telescope North (FTN), lucky imaging is feasible in presence of a 12th magnitude or brighter guide star within
10 or a few tens of arc seconds of the target and, provided that the pixel size is sufficiently small, the optical limit of the telescope is routinely achieved by
selecting the luckiest few percent images. Furthermore, frame transfer CCD and EMCCD cameras,
which can download one image while recording another, maximize the imaging duty cycle, reducing
total time requirements. When a guide star possibly
the target itself is used, the lucky exposures are
selected on the basis of the Strehl ratio, a measure of
the sharpness of the star image. The images are then
aligned on the brightest pixel, nominally taken as the
peak of the diffraction or optics limited point-spreadfunction (PSF), although more sophisticated methods
that exploit cross correlation or wavelets can provide
a more accurate reconstruction of the image.
Figure 11. A selection of images of the double star Zeta

Bootis from data collected at the 32-inch BOS telescope
at ~30Hz cadence, and a reconstructed "lucky" image.
In the figure, panel (a) is a selection of images of

the double star Zeta Bootis. The data was collected at
BOS at ~30Hz cadence (each exposure was
~0.034sec) by an Andor Luca-R EMCCD camera
with an SDSS r' filter and a x2 Barlow. The three
sharpest, as well as the three least sharp images in the
series are shown on the top and bottom respectively.
Panel (b) at the top is a stack of the first 200 frames
in the series, equivalent to a single 3.4 second exposure. Panel (b) at the bottom is the reconstructed
lucky image a stack of the sharpest 200 frames (a
1% selection) for a total exposure of 3.4 seconds.
Frames were selected and aligned using the brightest
pixel in the star image. The ghost star appearing at
the roughly 2 oclock position in the lower right
panel is an artifact of the shift-and-add these technique used to create the lucky image, known as tripling. The pair, which at the time of the observation
had a 0.45 arc second separation, is clearly resolved.
154
7. Speckle Interferometry
When observed at a sufficiently high cadence,
star images reveal the presence of speckles, resulting from the atmosphere producing multiple sharp
images of the target. In the case of a single star, these
speckles will be randomly distributed inside the
Gaussian seeing envelope on the detector. However,
in the case of a binary, there will be a characteristic
separation and position angle that the speckles will
follow. These duplicated images of the star are randomly distributed, and simply summing the images
or a long exposure will blur out this high-frequency
information, leaving only a Gaussian seeing envelope.
It was first suggested by Labeyrie (1970) that
Fourier analysis of speckles could allow the position
angle and separations of close binaries to be measured with great precision. This technique is now
known as speckle interferometry. Harold McAlister
and his associates at George State University pioneered speckle interferometry, while Brian Mason
and associates at the US Naval Observatory, Elliott
Horch at Southern Connecticut State University, and
others have launched sizeable double star speckle
interferometry observing programs. Horch (2006)
reviewed the history and current status of speckle
interferometry.
Since Fourier decomposition is blind to phase
shifts, the location of each of the duplicated double
star images on the detector is lost, but any correlations between speckles in different locations remains.
By summing the 2-dimensional Fourier power spectrum of the high-cadence images, one can discern any
patterns in the speckles that are common throughout
the dataset.
If the target is an unresolved binary, the resulting
power spectrum will form a set of fringes. The separation of these fringes is inversely related to the projected sky separation, thus making it possible to
measure the separation of binaries that are closely
spaced, down to the diffraction limit (or pixel scale)
of the telescope.
This method of speckle interferometry creates
180 degree degeneracy in the position angle that must
then be resolved using other imaging techniques
(such as lucky imaging) or more advanced interferometry methods (e.g. bispectrum image reconstruction, Lohmann et al., 1983). In most cases we use a
combination of speckle interferometry and lucky imaging to measure the position angle, separation, and
flux ratio of each binary pair.
At Genets Orion Observatory, an Andor Luca-S

EMCCD frame transfer camera is being used for
lucky imaging and speckle interferometry of double
stars, as well as high speed photometry of X-ray binaries and other variables stars. The Andor Luca
cameras are lightweight and are easy to transport,
install on large aperture telescopes as a guest observer to achieve higher resolution, or use in portable
field operations to observe asteroid or TransNeptunian Object occultations.
8. Conclusions
Figure 12. Results for the close visual binary HD 23985.

See text for explanation.
Figure 13. The Orion Observatorys Andor Luca-S camera and 10-inch telescope equipped with a Sidereal
Technology control system.
We observed HD 23985 (WDS 03503+2535AB)

with LCOGTs BOS 32-inch telescope. An x2 Barlow and I filter were used in conjunction with the
Andor Luca-R EMCCD camera. The components of
the binary HD 23985 are reported in the WDS as
magnitude 5.75 and 6.52. Referring to the figure,
upper left is a single 34 ms exposure of HD23985,
also known as a speckle. Upper right is a 10 second
exposure of HD23985. Lower left is a 1%
Lucky image of HD23985. The separation of the two
components is 0.325. Lower right is a speckle interferogram of HD 23985 created by summing the 2-D
Fourier power spectrum of all 5,000 34 ms images.
Measuring the separation and position angles of

double stars is cooperative activity that has engaged
many astronomers students, amateurs, and professionals for well over two centuries. Observations
are welcome for publication in the Journal of Double
Star Observations and similar journals. Once published, the observations are incorporated in the Washington Double Star Catalog along side those by William Herschel and all the other contributors.
Visual double observation projects using a low
cost astrometric eyepiece on small telescopes have
proven to work well for student seminars and summer camps where published results are a requirement.
While somewhat more complex, CCD double star
observations provide students with an opportunity to
learn advanced techniques, and with larger telescopes, such as the 2-meter Faulkes Telescope North,
to observe unusually faint doubles.
Finally, the advent of relatively affordable
EMCCD frame transfer cameras is extending lucky
imaging and speckle interferometry to fainter magnitudes, allowing atmospheric degradation to be circumvented and the resolution limits of telescopes to
be achieved.
9. Acknowledgements
We thank the Las Cumbres Observatory Global
Telescope for observational time on their 0.8-meter
telescope at Santa Ynez and 2.0-meter telescope on
Haleakala. This research was supported in part by
NASA through the American Astronomical Society's
Small Research Grant Program which funded the
Andor Luca-S camera at the Orion Observatory. This
research made use of the Washington Double Star
and other catalogs provided by the U.S. Naval observatory. We thank CDS in Strasbourg, France for use
of Aladan, the SIMBAD database, and access to
various catalogs. Finally, we thank Richard Berry for
supplying the lithograph from John Herschels report
from the Cape.
155
10. References
Aitken, Robert. (1918). The Binary Stars. Reprinted
2012. BiblioLife. Charleston, SC.
Alduenda, C., Hendrix, A., Hernandez-Frey, N., Key,
G., King, P., Chamberlain, R., Frey, T. (2012).
Separation and position angle measurements of
double star STFA 46 and triple star STF 1843.
JDSO 8, 48.
Argyle, B. (ed.). (2003). Observing and Measuring
Visual Double Stars. Chapter 12. Springer. New
York.
Argyle, B. (2004) Observing and Measuring Visual
Double Stars. Springer-Verlag. London, UK.
Argyle, R. (1986). Webb Society Deep-Sky Observers Handbook, Vol 1: Double Stars. (second
edition). Enslow. Hillside, NJ.
Alvarez, P., Fishbein, A., Hyland, M., Kight, C., Lopez, H., Navarro, T., Rosas, C., Schachter, A., Summers, M., Weise, E., Hoffman, M., Mires, R., Johnson, J., Genet, R., White, R. (2009). A comparison
of the astrometric precision and accuracy of double
star observations with two telescopes. JDSO 5, 60.
Baxter, J., Collins, S., Berlin, K., Jordano, C.,
Waymire, T., Cardoza, C., Shore, D., Carro, J.,
Genet, R. (2012). Observations of the triple star
system Phi Cassiopeia. JDSO, submitted.
Baxter, A., Johnson, J., Genet, R., Estrada, C., Medley, D. (2011). Comparison of two methods of determining the position angle of the visual double star
61 Cygni. JDSO 7, 212.
Brashear, N., Camama, A., Drake, M., Smith, M.,
Johnson, J., Arnold, D., Chamberlain, R. (2012).
Observations, analysis, and orbital calculation of the
visual double star STTA 123 AB. JDSO 8, 122.
Brewer, M. (2011). Astrometric measurements of
the double star STFA 43AB. JDSO 7, 107.
Caballero, R. (2012). 351 new common propermotion pairs from the Sloan Digital Sky Survey.
JDSO 8, 58.
Carro, J., Chamberlain, R., Schuler, M., Varney, T.,
Ewing, R., Genet, R. (2012). Measurements of Beta
Lyrae at the Pine Mountain Observatory Summer
Workshop 2011. JDSO 8, 150.
156
Collins, S., Jananne, K., Berlin, C., Cardoza, C., Jordano, T., Waymire, D., Shore, Baxter, R., Johnson,
R., Carro, J., Genet, R. (2012). Visual astrometry
observations of the binary star Beta Lyrae. JDSO 8,
153.
Couteau, P. (1978). Observing Visual Double Stars.
English translation 1981. MIT Press. Cambridge MA.
Chruch, R., Mohanan, K., Genet, R. (2012). CCD
Astrometry of the Multiple Star System Beta 321 in
Lepus. In preparation.
Dowdy, M., Heaston, N., Mathieson, C., Powell, B.,
Vega, E., Hodes, B., Hoffman, M., Frey, T., Johnson,
J., Arnold, D., Genet, R., Hodges, L. (2009). Astrometric measurements of the neglected visual double
star ARY 54 AC. JDSO 5, 74.
Dupuy, T., Liu, M. Ireland, M. (2012). Testing theory with dynamical masses and orbits of ultracool
Binaries. ASP Conference Series, in publication.
Estrada, C., Gupta, A., Kohli, M., Lund, A., Stout,
A., Daglen, B., Daglen, J. (2012). Astrometric
measurements of the visual double star Delta Botis.
JDSO 8, 6.
Estrada, C., Magana, S., Salam, A., Van Artsdalen,
A., Baxter, J., Brewer, M., Carro, J., Genet, R., Graf,
M., Herman, D. (2011). Astrometric measurements
of the visual double star Epsilon Lyrae. JDSO 7,
195.
Estrada, C., Johnson, J., Weise, E., Fisher, J., Howard, T., Salam, A., Almich, C., Kessinger, D., Cavanillas, S., Matakovich, T., Maly, K., Wallen, V.,
Genet, R. (2010). Astrometric measurements of the
visual double star H 5 12A. JDSO 6, 230.
Frey, T. G., Bensel, H., Bensel, R., Muller, F., Gasik,
R., Ruppe, M., Scimeca, D., Johnson, J., Medley, D.
(2009). Visual measurements of the neglected double star ARY 52 at the Pine Mountain Observatory
summer research workshop 2009. JDSO 5, 212.
Fluitt, J., Heath, E., Johnson, B., Ortiz, G., Charles,
H., Estrada, R., Genet, R. (2012). Observations of
the binary star Nu Draco. JDSO, submitted.
Frey, T., Achildiyev, I., Alduenda, C., Bridgman, R.,
Chamberlain, R., Hendrix, A. (2011). Visual measurements of the binary star S 654. JDSO 7, 45.
Frey, T., Achildiyev, I., Alduenda, C., Bridgman, R.,
Chamberlain, R., Hendrix, A. (2011). Visual meas-
urements of the multiple star STT 269 AB-C and

ARN 8 AB-D. JDSO 7, 50.
Genet, R., Johnson, J., Wallen, V. (2010). Onesemester astronomical research seminars. in Small
Telescopes and Astronomical Research. Collins
Foundation Press. Santa Margarita, CA.
Grisham, D., Johnson, J., Genet, R., Arnold, D.
(2008). Double star measurements with a three-inch
Tasco telescope. JDSO 4, 10.
Haas, S. (2006). Double Stars for Small Telescopes.
Sky Publishing. Cambridge, MA.
Herrmann, D. (1973). The History of Astronomy from
Herschel to Hertzsprung. English translation 1984.
Cambridge University Press. Cambridge MA.
Herschel, W. (1782). Catalogue of double stars.
Philosophical Transactions of the Royal Society 72,
112-162.
Herschel, W. (1803). Account of the changes that
have happened, during the last twenty-five years, in
the relative situation of double stars. Philosophical
Transactions of the Royal Society 93, 339.
Horch, E. (2006). The Status of Speckle Imaging in
Binary Star Research. RevMexAA 25, 79-82.
Luyten, W. J. (1997). LDS catalog: Doubles with

common proper motion (Luyten 1940-1987). On line
catalog in SIMBAD based on Publ. Astr. Obs. Univ.
Minnesota.
Marble, S., Gonzalez, C., Cameron, C., Johandes, J.,
Chapman, B., Fishbein, S., Johnson, J., White, R.,
Genet, R. (2008). High school observations of the
visual double star 3 Pegasi. JDSO 4, 24.
McAlister, H. (1985). High angular resolution
measurements of stellar properties. in Annual Review of Astronomy and Astrophysics 23, 59.
McAlister, H. A. (1999). The CHARA array on Mt.
Wilson: an overview. in Optical and IR Interferometry from Ground and Space. ASP Conference Series
194.
McGaughey, S., Genet, R. (2012). Observation of
STF 2686 from Haleakala. JDSO, submitted.
Michell, J. (1769). An inquiry into the probable parallax, and magnitude of the fixed stars, from the
quantity of light which they afford us, and the particular circumstances of their situation. Philosophical Transactions of the Royal Society 57, 234-264.
Law, N. (2006). Lucky Imaging: Diffraction-Limited
Astronomy from the Ground in the Visible. Cambridge University thesis.
Hoskin, M. (2011). Discoverers of the Universe: William and Caroline Herschel. Princeton University
Press. Princeton, NJ.
Perryman, M. (2010). The Making of Historys

Greatest Star Map. Springer. New York.
Johnson, J., Genet, R. (2007). Measurements of the

visual double star STF 2079. JDSO 3, 147.
Savary, F. (1830). Connaissance des Tems pour Pan.

Bachelier. Paris, France.
Johnson, J., Frey, T., Rhoades, S., Carlisle, J., Alers,

G., Genet, R., Atkins, Z., Nasser, M. (2008). Student
group measurements of visual double stars. JDSO 4,
52.
Schrader, A. B., Walker-LaFollette, A. M., Johnson,

J. M., Medley, D. J., Genet, R. M. (2010). Measurements of visual double stars with a NexStar 6 SE
at the Pine Mountain Observatory Summer Research
Workshop 2009. JDSO 6, 225-229.
Labeyrie, A. (1970). Attainment of diffraction limited resolution in large telescopes by Fourier analyzing speckle patterns in star images. Astron. Astrophys. 6, 85.
Warren, N., Wilson, B., Estrada, C., Crisafi, K.,

King, J., Jones, S., Salam, A., Warren, G., Collins, S.,
Genet, R. (2012). Astrometric observations of Delta
Cepheus. JDSO 8, 140.
Law, N., Mackav, C., Baldwin, J. (2006). Lucky

imaging: high angular resolution imaging in the visible from the ground. Astron. Astrophys. 446, 739.
Lohmann, A. W., Weigelt, G., Wirnitzer, B. (1983).
Speckle masking in astronomy: triple correlation
theory and applications. Applied Optics 22, 4028.
157
158
Image Processing with Python/PyRAF - Green
Amateur Image Pipeline Processing

using Python plus PyRAF
Wayne Green
Boulder Astronomy and Space Society
7131 Oriole Lane
Longmont, CO 80503
dxwayne@gmail.com
Abstract
A template pipeline spanning observing planning to publishing is offered as a basis for establishing a long term
observing program. The data reduction pipeline encapsulates all policy and procedures, providing an accountable framework for data analysis and a teaching framework for IRAF. This paper introduces the technical details
of a complete pipeline processing environment using Python, PyRAF and a few other languages. The pipeline
encapsulates all processing decisions within an auditable framework. The framework quickly handles the heavy
lifting of image processing. It also serves as an excellent teaching environment for astronomical data management and IRAF reduction decisions.
1. Introduction
The project's goal is to assist the amateur astronomy community with little computing experience
to develop a highly scripted solution to match their
observing environment to their reporting needs. Python is a platform-agnostic environment growing in
popularity with professional astronomers. It provides
a framework to allow the Image Reduction and
Analysis Facility (IRAF) to be combined with powerful libraries for mathematical analysis, astronomical
computation and publication.
Python is a mature language with widespread use
in systems administration, academia, science, engineering, business and industry. It has a rich suite of
libraries, including SciPY for scientific mathematics,
matplotlib for producing publication ready graphics,
and specialized astronomy libraries including the
IAU Standards of Fundamental Astronomy (SOFA)
(SOFA Board 2010), USNO Novas C and Novas-Py
(novas module) (Barron et al 2011). SOFA manages
time and earth-pointing modeling to high precision.
Novas allows using the JPL DE-405 solar-system
model and provides other astrometric capabilities.
The complexity of IRAF is both its strength and
its weakness. IRAF was started in 1981, extending
best practices from prior years and becoming a serious service in 1984 (Tody, 1993). Today IRAF is a
large collection of algorithms and techniques used for
a consistent analysis of astronomical data. IRAF is
large (approximately 40,700 files with approximately
3,450 help files) making it difficult to learn and easy
to make mistakes.
IRAF is the primary core for undergraduate and

graduate courses in observing and instrumentation.
IRAF is one of the gold standards of professional
astronomy. It provides analysis for both photometry
and spectroscopy.
The pipeline described here is the result of observing sessions for the novae 2011dh and 2011aj.
The main goal of the campaign was to ring out issues
with the instrumentation, data capture and the analysis environment at the Sommers-Bausch Observatory
(SBO).1 Lots of images and very repetitive analysis
activities led to developing this pipeline. Some serious defects were uncovered, as discussed below.
A pipeline represents a closed framework of accountability where symbols (elements of instrument
configuration and the analysis process) are managed
by a framework (a documented archive). The pipeline is closed in the sense that all original data and
germane intermediate results are retained. It is accountable since steps of the process may be audited
and any errors identified, understood and corrected.
Thus all aspects of the system are available through
configuration management; no questions about the
veracity of results should remain unanswerable.
2. Pipeline Overview
Planning an observation develops several of the
files used later in the process. The rawdata directory
is created and includes a special 'etc' directory storing
the star catalog's .tsv file and a reference image.
The catalog and image may be augmented with ds9
(SAOImage) regions to locate the camera's field-of1
University of Colorado, Boulder, CO USA
159
view relative to comparison stars, to show known bad

areas of a chip, or to set the slit orientation for spectroscopy. A special observations.txt file contains
FITS header entries (Pence, 2010), encoded as one
keyword:value:comment per line, and is applied by
the fittofits script. Anything else, like the time allocation request document, may be added as well. The
project's etc directory is the place where cooked
zero, dark and normalized flats are placed.
The FIT files are converted from the Amateur Astronomical Instrument Processing Package (AIP)
format (Longmire, 2003) to match the pipeline FITS
keyword policies (NRAO, 2011). Verbose file names
are edited down to a manageable size. The fitserial
script prepends a small serial number to all files in
the order specified by observation time FITS header
keywords. Special FITS keywords are added that
describe the filters, their serial numbers, the RA/Dec
of the target(s), spectrograph settings and site and
instrument data. This is mostly handled by the fit2fits
script. Special header values for equipments manual
settings are made with the fitsrewrite script on a file
by file basis.
The results may be checked with a fitsls and fitsgrep program. The HEASARC program fitsverify
(see NASA HEASARC) may be used as well.
The pipeline = policy+algorithms+data and the
fit2fits/fitsrewrite scripts are designed to align local
conventions to IRAF conventions. The fit2fits script
copies all files to a special attic directory prior to
making changes. This retains the original files in a
handy location for reference.
This pre-analysis directory is frozen and an archive copy is made.
2.3 Analysis Directory and Analysis
Figure 1. Overview of the pipeline process.
It is of value to be able to move the etc directory contents across sites and share material in a way
that is insensitive to the platforms and software releases encountered. Develop on Linux at the office,
move to a Windows/Sun observatory, and share with
Windows/Mac/Linux users.
2.1 Raw Data

Observations are made and the raw telescope image files are copied into the rawdata directory. At this
time, the directory is frozen (made read-only) and an
archive copy is made. The rawdata directory is copied to a preanalysis directory before any changes are
made.
2.2 Preanalysis Directory

The purpose of this directory is to support cleaning of
file names and augmenting them with proper header
information. This makes them ready for processing.
160
The pre-analysis directory is then copied to the

analysis directory and work begins in earnest. At this
point if a misstep is made in the process the effort of
cleaning and augmenting files is not lost. Simply recopy the pre-analysis directory and start again. The
scripts run quickly and little time is lost.
The fitsort program is executed first. It reads the
header keywords and builds a directory substructure.
The subdirectories hold images by type, filter,
binning and exposure time. In each created subdirectory a special l.l file is created that contains the pertinent file names. This file is a batch file or at file
used as @l.l in IRAF.
The pipeline script traverses the directory and
cooks all the zeros, darks, flats, comparison files, etc.
It copies these files to the project's etc directory. It
then applies these files to all the science images.
At this point all the data are inspected for the
kind of gross defects that crop up with hot pixels
and atmospheric artifacts being the most dangerous.
The core analysis process is now performed.
This varies with the target and observing program.
Photometry proceeds along one path while spectrographic analysis follows its own path. Other examples include high-cadence observations of CV or Be
stars and cases where targets overlap nebula or galaxy structure.
2.4 Photometric Analysis

The next phase uses daofind/daophot or sextractor (Linux) to extract instrumental magnitudes. The
.tsv catalog files are used to acquire reference magnitudes for stars. At this stage the point-spread function (PSF) for images becomes critical.
Additional finishing steps are performed as
needed.
3. Developing Pipeline Steps

The 2011dh campaign's data analysis started by
hand in the usual way, with an eye open to ways to
code the steps into scripts. The scripts were desired
as a framework for other projects especially those
involving undergraduate research assistants.
One key aspect about IRAF is its unlearn
command. This cleans out all hidden sticky parameters from the system. Analysis was done by taking
the epar quiz and filling in all the parameters. The
epar command opens into hidden subordinate parameter files shared across common stages of analysis where defaults are not announced. For example
average vs. median combining with the imcombine
command. Once acceptable results were obtained and
the parameter files were deemed correct the lpar
command was used to list the parameters for a script.
This information was cut and pasted into the script
and edited to meet specific needs.
In this way scripts for creating reduced zero,
dark and flat files were made. It was during zero file
production that we noticed structure in the zero files.
This was in the form of noisy pixels that appeared
hot, but not always in the same place. Thus pixelmasks would not work. Dark images suffered the
same effects.
Gaussian smoothing function to the raw image. This

removes structure as well as hot and cold pixels.
The results are shown in figures 2 and 3. There is
a small corner-to-corner bias structure in the comlpar zerocombine
input = "@l.l"
(output = "zmaster")
(combine = "average")
(reject = "minmax")
(ccdtype = "zero")
(process = no)
(delete = no)
(clobber = no)
(scale = "none")
(statsec = "")
(nlow = 0)
(nhigh = 1)
(nkeep = 1)
(mclip = yes)
(lsigma = 3.0)
(hsigma = 3.0)
(rdnoise = "0.")
(gain = "1")
(snoise = "0.")
(pclip = -0.5)
(blank = 0.0)
(mode = "al")
zerocombine_kw = {
'input'
: iname,
'output' : rawname,
'combine' : "average",
'reject' : "none",
'ccdtype' : "zero",
'process' : iraf.no,
'delete' : iraf.no,
'clobber' : iraf.no,
'scale'
: "none",
'statsec' : "",
'nlow'
: 0,
'nhigh'
: 1,
'nkeep'
: 1,
'mclip'
: iraf.no,
'lsigma' : 3.0,
'hsigma' : 3.0,
'rdnoise' : "0.",
'gain'
: gain,
'snoise' : "0.",
'pclip'
: -0.5,
'blank'
: 0.0,
'mode'
: "a"
}
Table I. Convert IRAF lpar to Python Dictionary
bined zero file. The smoothing of outlier pixels in the

overall image removed artifacts from later processing
steps.
During processing, the special feature from
CL/ECL that prepends a pattern to a file name is retained in Pythons iraf module. Thus IRAFs command:
imarith @l.l-cooked_masterzero.fits z_//@l.l
3.1 Writing Basic IRAF Oriented Scripts

As an example, the cooking of image zero reference files is explored. Once PyRAF is started, the
IRAF lpar zerocombine command produces the listing in column 1 of table 1 (lpars comments deleted).
Each lpar line is edited to produce a Python dictionary shown in column 2.
Within the masterzeros script the source line:
subtracts the cooked_masterzero.fits from each file

in l.l and prepends a z_ to each new file. When
dark subtracting:
imarith z_//@l.l-cooked_darkmaster.fits d_//@l.l
followed by
imarith d_//@l.l / masterflat.fits f_//@l.l
iraf.zerocombine(**zerocombine_keywords)
passes the dictionary to the zerocombine function.

That is all that is needed to cause Pythons iraf module to process all the files listed in the l.l file to produce a single file called raw_1x1_zeromaster.fits.
Another
dictionary
is
passed
to
iraf.imsurfit(**surfit_keywords) applying the
produces a set of f_ files that are normalized and

ready for later processing. In a similar policy: c_
represents imcombined files; a_ means imaligned
files, etc. While the commands are shown in command-line form, each is scripted to identical behavior within the Python code. The prepending feature
only requires the one file, l.l, that holds the base file
161
names to guide the various steps of the pipeline. This

alleviates the need for numerous files (or lists) with
multiple variations of the file names.
[log(alpha/beta) ln(3)] / d. This simple formula is

easily applied to two combined FITS images.
Two programs were written in Python. The first
applies this equation to the arrays of the values in the
Hydrogen alpha and beta images. The second makes
a surface plot of these data using matplotlib.
Figure 4 is the Hydrogen alpha image; Figure 5
is the Hydrogen beta image; Figure 6 is the ratio;
Figure 7 is the surface plot of the results. The plot in
Figure 7 reveals the column density of the nebula in a
very visceral way.
Figure 2. Raw zero image.

Figure 4. Hydrogen alpha image.
Figure 5. Hydrogen beta image.
Figure 3. Zero image cooked with imsurfit.
3.2 A More Complex Script

In a more sophisticated application: Green et al.
(2009) applied line strengths from Osterbrock (1989)
to images of the Bubble Nebula (NGC 7635), giving
the ratio of Hydrogen alpha to Hydrogen beta based
on d = ratio of the relative line strengths.
The ratio d = 1 (4.861 x 10-5/6.563 x 10-5) is
computed and used to calculate column density as
Figure 6. Column density ratio of images.
162
6. References
Barron, E. G, Kaplan, G. H., Bangert, J., Bartlett, J.
L., Puatua, W., Harris, W., Barrettt, P. (2011) Naval
Observatory Vector Astrometry Software (NOVAS)
Version, 3.1, Introducing a Python Edition. Bull.
AAS 43, 2011.
http://aa.usno.navy.mil/software/novas/novas_py/
novaspy_intro.php
IAU SOFA Board. IAU SOFA software collection.
Issue 2010-12-01. http://www.iausofa.org
Figure 7. Surface plot of ratios.
4. Issues Uncovered Using the

Framework
During the 2011dh campaign, issues with the
camera's chip and very dirty filters pushed analysis of
the images to the breaking point that of losing confidence in reductions. Several times, sticky IRAF
parameters led to erroneous results. The pipeline was
created and refined to support auditing work needed
to understand the problems with the manual process,
equipment and images.
The IRAF program imsurfit was discovered and
applied to the zeros and darks. The functionality was
easily added to the pipeline, removing the problems
we encountered.
An outstanding issue is with the goose-footed
PSF for star images with SBO's mis-collimated
Boller and Chivens 24-inch. The PSF does not match
any pre-packaged software. Special handling of the
telescope's PSF was needed.
The third issue involved the filthy filters great
for teaching but lousy for science. New filters are
ready for use, together with new filter parameters for
IRAF.
5. Conclusions
The pipeline provides a fully accountable way of
managing data and processing for observations. IRAF
provides a rich kit of resources for dealing with
problems and delving into image issues. The Python
environment and special Python libraries allow developing and sharing programs between computing
operating systems. The scripting forces each parameter to a known state and provides a more consistent
session-to-session analysis framework. The solutions
to newly identified problems can be easily integrated
into the framework by using rich features of the existing IRAF code.
Green, W., Stiner, J., Davis, M. (2009) Study of

NGC 7635. Unpublished.
Longmire, M. (2003). SBFITSEXT Version 1.0 A
Set of FITS Standard Extensions for Amateur Astronomical Processing Software Packages. Santa Barbara Instruments Group, Santa Barbara, CA.
FITSVerify. FITS file verification tool. NASA
Goodard Space Flight Center.
http://heasarc.gsfc.nasa.gov/docs/software/ftools/
fitsverify/
IRAF (2011). National Radio Astronomy Observatory.
National Radio Astronomy Observatory.
http://iraf.noao.edu/
Osterbrock, D. (1989). Astrophysics of Gaseous
Nebulae and Active Galactic Nuclei. (Table 4.1).
University Science Books, Mill Vally, CA.
Pence, W. D., Chiappetti, L., Page, C. G., Shaw, R.
A., Stobie, E.. Definition of the Flexible Image
Transport System (FITS), version 3.0. (2010).
Astron. Astrophys. 524, A42.
Tody, D. (1993). IRAF in the Nineties. in Astronomical Data Analysis Software and Systems II (R. J.
Hanisch, R. J. V. Brissenden, J. Barnes, eds.). A.S.P.
Conference Series 52, 173.
163
164
High Speed Photometry Vander Haagen
High Time Resolution Astronomy

or High Speed Photometry
Gary A. Vander Haagen
Stonegate Observatory
825 Stonegate Rd, Ann Arbor, MI 48103
garyvh2@gmail.com
Abstract
High Time Resolution Astronomy (HTRA) or High Speed Photometry offers opportunity to investigate phenomena that take place at timescales too fast for standard CCD imaging. Phenomena that occur within fractionalseconds to sub-millisecond timescales have been largely overlooked until recently due to technology `limitations
and/or research priorities in other areas. Discussed within the 1-second to millisecond range are flares, oscillations, quasi-periodic oscillations (QPOs), and lunar occultations. A Silicon Photomultiplier (SPM) detector with
associated pulse amplifier and high-speed data acquisition system were used to capture stellar intensities at
rates up to 1000 samples/sec on stars to 8 V-mag. Results on YY Gem flares, exploratory searches for oscillations on AW Uma, BM Ori, and X Per, and lunar occultations are discussed along with data reductions techniques. Opportunity for future study is also discussed.
1. Introduction
Pulsars
Stellar
Physics
Flares
Oscillations
its
Fundamental
Astronomy
s
Tran
Solar
System
ns
ar
L unc ultatio
Oc
Our fascination with solar the system, star systems and their evolution, asteroid tracking, rotation,
sizes, and composition as a few examples have led us
to fainter and fainter objects, larger optics, and longer
exposures. With longer exposures, events that take
place in the fractional-second to millisecond timescale are averaged and lost.
Interest in this area opens up a whole new range
of possible explorations. High speed photometry or
high time resolution astronomy (HTRA) can fill an
area of scientific exploration that complements conventional CCD imaging and offers unique insight into
relatively unexplored areas. Richichi (2010) presents
a good overview of how HTRA augments astronomical research. Figure 1 shows research areas versus
timescale and the regions of study. Pulsars, oscillations, quasi-periodic oscillations (QPOs), and occultations push the millisecond and beyond range. These
represent significant equipment challenges requiring
very fast and sensitive photo detectors and larger
aperture optics for 8-9 magnitude and beyond. In
addition, with data rates in the kilohertz range there
are challenges in coping with the storage and analysis
of the data. With intensity sampling at 1000 samples/sec a two-hour data collection run generates 7.2
million data points for storage and subsequent analysis.
For the slower oscillations, transits, flares, and
flicker events, data rates in the 10 to 100 Hz range
still present the same equipment challenges but offer
opportunity for the study of fainter objects. The previously cited Stellar Oscillation and Occultation arti-
cle presents numerous examples for study in these

areas.
10-2
10-1
Event - Seconds
Figure 1. Research opportunities at various timescales,
facsimile reproduced by permission from A. Richichi.
The trade-off between sampling rates and S/N is

shown in Figure 2 for sampling rates of 10, 100, and
1000 samples/sec. This graph shows ideal conditions
using a clear filter with minimal scintillation noise.
The telescope has a 17-inch aperture and a silicon
photomultiplier detector (SPM) with an internal gain
in excess of 106. At 1000 samples/sec the measurement limit is approximately 8th magnitude.
165
2. The Optical Train for High Speed

Photometry
Figure 2. S/N versus magnitude for 3-sampling rates

with clear filter and low scintillation noise
When pursuing fast events bandwidth (BW) is

limited by our sampling rate. To avoid alaising
during analysis, the sampling rate (fs) must exceed
the Nyquist rate (fn),
fs > fn,
where fn = 2BW
Or, for a BW of 1000 Hz the sampling rate must
be greater than 2000 Hz. Literature places a number
of the oscillations events in the mHz-1900 Hz region
for x-ray binaries (XB) (Warner, 2008). Targeting of
the 1900 Hz oscillation dictates a minimum sampling
rates of 3800 samples/sec thereby limiting objects to
approximately 6th magnitude or brighter for 14-17inch scopes. For slightly less demanding lunar
occultation studies a resolution of 1-3 milliseconds is
required for diffraction fringe capture. This places
sampling rates in the 300 to 1000 samples/sec range.
These rates and the slightly slower flares and flickers
can push the upper limit to as high as 10th magnitude
under ideal conditions.
Target
Type
X Per
AW UMa
BM Ori
YY Gem
RZ Cas
Moon
Occultations
HMXB Pulsar
CV
EB, X-ray source
EB, X-ray source
EB, X-ray source
Single & double
stars
V Mag
6.7
6.9
8
9.8
6.2
<8.0
Table I. HTRA study targets
With an practical upper limit of 9th magnitude

the initial candidates for fast event studies are shown
in Table I.
166
Key components of the optical system include a

telescope with 10 or greater aperture, a fast response
detector system with internal gain, data acquisition
system and analysis software.
The detector represents the biggest challenge for
HTRA. While there are several CCD cameras available with internal gain (electron multiplication) they
are expensive and do not provide the S/N of photomultiplier type sensors. However, they do provide a
mechanism for obtaining a reference star and using
more conventional data reduction techniques. The
discussion presented in Vander Haagen (2011) provides and overview of numerous sensor types and
their performance. The characteristics that make the
Silicon Photomultiplier (SPM) most attractive are
high internal gain (106+), good noise performance,
wide bandwidth and insensitivity to overloading,
troublesome with conventional vacuum photomultipliers.
The SPM used was manufactured by sensL
(2012). Figure 3 shows both the unpackaged sensor
along with it fully enclosed and cooled and attached
to a nose piece with filter and f-stop. The output of
the SPM is millivolt level pulses at rates up 10 MHz
depending on the equivalent photon count per
second, one photon per count. These low voltage
pulses are fed to a wideband pulse amplifier to
provide a logic level signal for the data acquisition
system. The amplifier provides a voltage gain of
2500 at a bandwidth of 10 MHz necessary for
undistorted photon pulse counts. This is a higher
performance version of the previously used amplifier.
The new amplifier provides about double the BW and
greater stability to temperature and input-output
loading changes. A schematic of the circuit may be
requested from the author.
The initial optical train used had a field of view
of 80 arcseconds. The configuration used a video
camera for alignment and relied on accurate tracking
from the mount for extented periods. Without
guiding, tracking errors were excessive for multiplehour data collection runs. The optical train was
modified to improve ease of initial alignment and
provide continuous autoguiding. The modified
optical train is shown in Figure 4.
mately 80% of the light is reflected 900, passing

through the optical filter, f-stop, and onto the SPM.
The CCD camera for initial set up and autoguiding
uses the 5% transmitted light. A software controlled
servo motor positions the pellicle for either beam
splitting mode or 100% pass through for normal CCD
camera operation. The optical train has demonstrated
easy initial alignment and guiding for the long data
collection runs.
The f-stop most generally used was 750-micron
diameter providing a field of 57-arcseconds. A
smaller 400-micron diameter f-stop was required for
lunar occultation work.
The full optical train is shown in Figure 5
mounted on the 17-inch PlaneWave scope.
Figure 3. Top: SPM module. Bottom: SPM module packaged with cooling fan, nosepiece, f-stop and filter. A
cooling fan is required to keep the thermoelectric cooler
operating at 30 C below ambient.
Pulse
Amp
Data Acquisition
System
SPM
F-Stop
Camera
Control
Filter
80%
5%
Flip Pellicle
Beam Splitter
CCD Camera
Figure 4. Optical train with flip pellicle beam splitter,

SPM module, and CCD camera for alignment and guiding
The light enters and strikes a 5-micron pellicle

beam splitter, National Photocolor (2012). Approxi-
Figure 5. Optical train showing SPM and CCD camera

attached to software controlled pellicle beam splitter
3. High Speed Data Acquisition and

Analysis
The basic data collection sequence will operate
without a reference star. Since we are operating with
a single pixel camera there is no option for a reference without adding a second complete optical train
with associated equipment and alignment issues. In
addition, the two data streams are generally correlated in real time. Such a setup is described in Uthas
(2005) and is generally not necessary for the higher
frequency data particularly where data is at a frequency substantially higher than the scintillation
noise. An alternative calibration option will be discussed in Section 7.
The data acquisition system used was from,
Measurement Computing Corporation (2012), a
DaqBoard/1000 series data acquisition board. The
PCI board takes one PC slot with the data inputted
through a standalone connection board. Important
feature of the data acquisition system include:
167
1.
Easy connection and use with PC at lowest

system cost
2.
Bandwidth capability to at least 8x106-photon

integration counts per second.
3.
Ability to totalize counts over any period; microseconds to seconds
4.
Ability to trigger data collection on a count,

count rate or analog event and employ a caching technique for pre-trigger data capture
5.
Integrated DaqView software for real time

data storage and review of events
The integrated DaqView software met all the

data acquisition requirements but did not include a
suite of digital signal processing (DSP) subroutines
for Fourier transforms, power spectral density, filtering, smoothing, etc. Sigview 2.3, a DSP package by
SignalLab (2012) had the required capability. Sigview has an intuitive structure and easily imports
ANSI data from the MC data acquisition system. This
software provides capability for all the filtering
modes such as low pass, high pass, band pass, fast
Fourier transforms (FFT), spectrograms, statistical
analysis, amplitude and frequency measurements,
adding and parsing of data runs, etc. This capability
helps to reduce the dependency on a secondary reference star, to some extent, as will be described later.
4. Exploratory Look at Rapid Events in

Eclipsing and X-ray Binaries
The primary interest of this study is the stellar
physics events as shown in Figure 1. The focus will
be on cataclysmic variables (CVs) particularly X-Ray
binaries (XB) exhibiting QPO or suspected optical
oscillations. Current models suggest magnetically
controlled accretion onto a rapidly rotating equatorial
belt of gas. These energetic events may generate RF,
X-Rays, and occasionally modulate emitted light in
the form of oscillations or QPOs with signatures
varying widely, Uthas (2005) and Warner (2008).
Flares are well known events for our Sun and
other stars and CVs with strong magnetic fields at
their surfaces. An energetic flare can produce luminosity increases of many orders of magnitude on time
scales of seconds to days. Flickering is a smaller outburst of a few tenths of a magnitude over a much
shorter time, generally seconds to minutes.
Be it oscillations, flares, or flickering events they
occur rapidly such that detailed high-speed data is
necessary for capture. These data are used to develop
or validate models for the stellar systems. For the
oscillations, target frequencies range from 40-2000
168
Hz placing these frequencies generally beyond the

worst of the scintillation noise. Targeted will be those
stars or star systems with suspected or verified events
and at data rates necessary to capture the oscillation.
The limiting factor will be target brightness necessary
to support the fast exposures. The data will be captured at data rates as high as possible in segments of
106 data points for each file. The data acquisition
system supports continuous data collection but limits
the file size for analysis purposes. The prime tool for
identifying the oscillations will be the Fourier transform power spectral density (PSD) of the amplitudetime data and the Spectrogram or a time sequence
PSD showing where in time the frequency occurred.
For flares and flickering events the time domain
data will be reviewed directly or bandpass filtering
and/or resampling of the data will be used to reduce
the high frequency variations. The same segmenting
of data will limit individual files to 106 points.
Data collection objectives will be exploratory in
nature:
1.
Confirming the range of data gathering of both

magnitude and BW.
2.
Learning to separate noise from stellar information focusing on XB with known oscillations in the X-Ray spectrum and suspect in
the optical.
3.
Refining equipment and procedures for improve results.
4.
Firming up a plan for next steps including

working with a researcher.
4.1 AW UMa
AW UMa is a cataclysmic variable; HMXB targeted because of its general classification and previous interest by other researchers. The first run of AW
UMa was at 200 sample/sec and the initial look at the
data generated considerable excitement. Figure 6
shows an expanded section of the 3600-second run.
Figure 6. AW Uma data at 200 samples/sec, 7-second

section noting periodic pulses
These periodic pulses were found later to originate from an open ground on a coax connection.
Considerable attention is necessary to address noise
across the full spectrum prior to any data collection.
Flipping the pellicle to the pass through position
blocks the sensor and helps to confirm proper noise
performance. This step was accidentally omitted prior
to this run. Three additional 1-hour runs were completed on a second night. No high frequency components were detected.
Another X Per run occurred on 2010 March 15.

10 segments of 1 million points each. A composite of
the first 4 segments is shown in Figure 8.
The data encompasses a short eclipse that seems
unreported in the literature. The cloud sensing system
showed no change in the sky condition during the
period. Note also what appear to be small flares during the eclipse. A PSD of the flare section is shown
in Figure 8.
4.2 X Per
X Per is a HMXB, Pulsar, and gamma-ray source
and suitable classification for high frequency activity.
The 6.7 V magnitude allowed for data collection at
1000 samples/sec. All runs were with Astrodon Johnson-Cousins blue filters. Run #1 on 2012 March 7
was 6 segments of 1 million data points each. The
files are seamless from one to the next so no data is
lost. Run #2 on 2012 March 9 was 8-segments of 1million points each. A PSD of a typical run is shown
in Figure 7.
Figure 9. X Per PSD of flare section
Figure 9 shows no identifiable high frequency

components. However, on run #6 there was an identifiable peak at 383 Hz as shown in Figure 10, a potential QPO.
Figure 7. X Per PSD data at 1000 samples/sec, typical

spectra with no high frequency components to 500 Hz
Figure 10. X Per PSD data at 1000 samples/sec, run #6; a

potential QPO at 383.85 Hz
Figure 8. X Per data at 1000 samples/sec, composite of 4

segments. The smooth data were resampled at 1sample/sec
A very high amplitude peak was detected during

segment 5. With the mean count running at approximately 100, just after 306 seconds on segment 5 the
counts peaked over 3200. Figure 11 shows this event.
The signal profile does not match a flash tube typical
of emergency vehicles. No clouds were present for
possible lightening flash. A possible explanation is a
169
stray emergency vehicle rotating light flash from a

nearby road. These data will be investigated further.
Figure 11. X Per data for segment 5 showing very high

peaking near 306 seconds
Figure 12b. BM Ori PSD data. Region around 167.6 Hz.
4.3 BM Ori
BM Ori is an eclipsing binary and X-Ray source
with good sky position for long runs. The EB was
studied two nights 2012 March 17/20 at 1000 samples/sec. On the first run PSD analysis showed peaking of data at 167 Hz on three of the 1-million data
point segments. Figure 12 shows both the full PSD
and a section focused on 168.6 Hz. The same peak
occurred in all the other segments of this run. On
2102 March 20 another run was completed with 8
segments of 1 million points each. The data looked
similar with the 167 Hz components showing in all 8
segments. Figure 13 shows the PSD and a spectrogram of segment 8.
With so little data available at these data rates it
is difficult to validate and exciting to see what is captured on each nights runs.
Figure 13a. BM Ori PSD and Spectrogram. Region

around 166 Hz.
Figure 12a. BM Ori PSD data. Full 500 Hz BW.
Figure 13b. Spectrogram showing the distribution of the

166 Hz oscillation over 1000 seconds. The spectrograms brighter spots indicate higher concentration of
the oscillation. Note small frequency change over the
1000-second period.
170
4.4 YY Gem
As a clouded-in participant of the YY Gem
photometric team, this star generated considerable
interest for future study of its active flares. YY Gem
is an M dwarf eclipsing binary and known flare star.
With its 9.8 V magnitude, data was collected at 200
samples/sec giving a BW of 100 Hz. Data were collected for a 2-hour and 6-hour run on two nights.
Figure 14 shows the full 3600 seconds of data for
2012 March 13.
With data compressed it looks noisier than
would be expected. However, with an average count
of 600+ the S/N was 15 and acceptable. The PSD of
the same data is shown in Figure 15.
The FFT shows strong components below 5 Hz
and reaching background levels by 25 Hz. There
were two low level components at 65 and 98 Hz at
very low power density.
Figure 14. YY Gem data for 2012-02-13, blue filter, 200

samples/sec, with 3600-second trace
Figure 15. YY Gem PSD data for 2012-02-13, blue filter,

200 samples/sec, BW = 100Hz
longer data runs through multiple eclipses (0.814

days primary) to increase the chase of flare and
flicker capture.
5. Lunar Occultations: a High Accuracy

Measurement Tool
The Moon has held all those studying the heavens for thousands of years in great fascination. However, it has only been in the last 60 years that the
Moons predictable and known apparent angular rate
of motion (0.4/sec) and accurately known distance
has been augmented by a well-understood occultation
process, Nather et al. (1970a,b). With good physical
data, timing to 1-millisecond, and a detailed model of
the diffraction pattern created by the lunar edge, occulted object angular diameters and double star separations can be calculated with high accuracy
(Richichi, 2005).
The limitation of this tool is that the lunar occultations cover about 10% of the sky area over its 18.6year cycle. Furthermore, many occultations occur
during the day, are reappearances and require accurate off axes guiding for the hour or more the star is
behind the Moon, and the light scattered from the
illuminated portion has to be sufficiently low to yield
adequate S/N. And, then there are still sky conditions
and the instrumentation. The author recommends
reading a good primer Lunar Occultations (Blow,
1983), that sets the stage well. Combining optical
Interferometry with Lunar Occultations (Richichi,
2004), is also very informative. With all these difficulties lunar occultations still present an exciting
challenge and area for accurate measurement of star
diameters and separations using HTRA techniques.
The lunar occultation sequence can be seen in
Figure 17 where a moving diffraction pattern is generated on the earths surface by the star prior to the
geometric shadow. The pattern and shadow is moving at approximately 750 meters/sec dependent upon
the limb contact angle. To illustrate, a simple diffraction pattern F(w) for a monochromatic point-source
the intensity is given by
Figure 16. YY Gem flicker event
Figure 16 shows a flicker event with a total duration of less than 10 seconds at approximately a magnitude brightening at the peak. This star warrants
where S(w) and C(w) are the Fresnel sine and cosine
integrals.
This shows that the diffraction intensity is a
function of both the wavelength of observation and
the lunar distance D. The Fresnel integrals can be
calculated with a MathCad type program or using
Keisan (2012). The results are shown in Figure 18.
171

Earths
Surface
D
~384,000 km
Diffraction
Profile
25% Io for
Pt Source
Geometric
Shadow
Moving
~750m/s
Diffraction
Pattern
Io
Moving
~750m/s
Figure 17. Lunar occultation of star showing moving

diffraction pattern on earths surface
detector and filter response, integration time, Moons

physical parameters, etc. The stars diameter is an
additional parameter to be varied to find the best chisquared fit of the diffraction pattern to the data. A
further refinement of the least squares model came
with the composed algorithm (CAL), Richichi
(1989), which incorporates an iterative algorithm to
select the most likely brightness profile of the
source thereby yielding correct solutions over a much
wider range of noise levels. Researchers have reported angular diameters to 0.5% uncertainty using
the CAL approach (Richichi, 2005). In my brief contact with researchers these models are not available
for general use and require collaboration with the
researcher. With good models and millisecond data, a
measurement range in angular diameters/separation
from 2 to 300 milli-arcsecond (MAS) is possible
(Richichi, 2012).
As a simple example of how the fringe analysis
process works using only one element of the diffraction profile, Richmond (2005), presents a simulation
of a uniform circular disk on the fringe pattern. Figure 19 estimates the angular size of a disk based on
the
height
ratio
of
the
first
fringe,
Ifringe peak/Io. For a point source the ratio is 1.31. For a
20 MAS diameter the ratio is 1.08.
Figure 18. Single and multiple colors diffraction patterns

formed on earths surface, distance in meters from the
geometric shadow
The solid line shows a monochromatic pattern

for source at 430 nm. Clearly defined interference
fringes exist past 41 meters from the geometric
shadow. With the fringes moving at 750 m/s the
fringe rate is 0.75 m/millisecond or the first fringe
passes in about 12 milliseconds. Since our stars are
anything but monochromatic we average three monochromatic wavelengths, to keep the math simple, and
get the dotted interference pattern. Note the significant blurring of the fringes. It only gets worse as we
integrate all the fringes over the BW of a typical narrow band filter and increase the diameter of the
source to something finite. A good set of pictorials
illustrating the effects of wavelength, BW, and star
diameter is presented by Richmond (2005).
Nather and Evans (1970) developed one of the
first diffraction models of the occultation event leading to more than two decades of refinements. A least
squares fit to the data is made using model parameters such as star brightness profile, sky background,
172
Figure 19. Angular size of a disk using only the ratio of

first fringe height to steady state Io
As seen this relies on only one characteristic of

the datas profile and provides limited utility.
The video occultation measurement software
LiMovie (Miyashita, 2008), offers a graphing and
diffraction matching routine. To date it has not been
possible to use non-video data as a source for this
program.
Two excellent references for lunar occultation
predictions include: The International Occultation
and Timing Association, IOTA (2012) and the Dutch
Occultation Association with their Lunar Occultation
Workbench, DOA (2012). The Lunar Occultation
Workbench (LOW) was found particularly user
friendly and included many helpful cross-references.
Initial occultation data collection using a 750micron f-stop proved unsuccessful for a 7.05 V-mag
star due to excessive dark limb illumination of the
field. A second 150-micron f-stop, 11 arcseconds
field, proved nearly impossible to align and focus.
Use of the 150-micron diameter pinhole required a
precision focuser in the SPM path, which could not
be accommodated due to insufficient optical path
length. A compromise 400-micron f-stop, 30
arcseconds field, gave acceptable S/N for lunar
phases to around 50% and 6+ magnitude. Further
work is necessary to determine the range of illumination and star magnitudes for the two smaller pinholes.
Occultation of XZ-5588, 53 Tauri, SAO-76548,
a 5.5 V-mag star is shown in Figure 20.
SPM system parameters. The model found a point

source assumption provided a good data fit with chisquared of 0.97. The SPM system parameters are
being reviewed and the model rerun to confirm the
stars angular diameter.
6. Future Study Opportunities

Target
Type
V Mag.
X Per
HMXB Pulsar
6.7
AW UMa
CV
6.9
CygX-1
HMXB
8.9
YY Gem
EB, X-ray source
9.8
Vela X-1
XB
6.9
RZ Cas
EB, X-ray source
6.2
Lunar
Occultations
Single & double

star detection
and measurement
< 8.0
Table II. Candidates for future HTRA study
Figure 20. Occultation of XZ-5588 at 1000 samples/sec

using a blue J-C filter exhibiting 3-diffraction fringes
Figure 21. Occultation of XZ-5588 with point source diffraction pattern shown (dotted) as determined by the
CAL model, first trial.
Using the first fringe ratio of 410/310 counts =

1.32, Figure 19 would place the star angular diameter
at a point source or too small to be measured. A large
amount of uncertainty results in this calculation due
to the curves irregularity and simplicity of the model.
Referring to Fig. 21, a CAL diffraction model
was run on XZ-5588 data with an estimate of the
Table II summarizes several good candidates for

future study. The X-Ray binaries and cataclysmic
variables are candidates that exhibit short or even
extended periods of rapid variability in both the RF
and the optical spectrum. Star systems such as Cygnus X-1 have exhibited modulation at frequencies in
the 100 to 1300 Hz frequency range. This remains an
opportunity for work in late summer and early fall
when at its highest altitude. Vela X-1, HD77581, is
another candidate for lower latitude study. Detection
of these signals are challenging due to their low intensity, high frequency, and unpredictable occurrence. A study by Uthas (2005) and summary by
Warner (2008) give some insight into this very interesting study group.
Flaring stars such as YY Gem are exciting study
opportunities and allow for higher magnitude targets
with sampling rates as low as 10 samples/sec.
The website ephemerides by Kreiner (2004), are
excellent for identifying EB targets and timing as is
the AAVSO (2012) for observing campaigns, descriptions of star types and their characteristics, classifications, outbreaks, data, charts, etc.
Occultations are a source of major interest at the
higher sampling rates. The International Occultation
and Timing Association, IOTA (2012) provides targets, technical assistance, and a depository for data.
7. Discussion and Conclusions

The SPM proved a capable detector for photon
counting, high-speed photometric study. The use of
the flip pellicle and capability for continuous guiding
made set up easy and long term data acquisition pos-
173
sible without intervention. A setup modification using a helical focuser in the SPMs optical path would
improve focusing for the smaller f-stop apertures.
The focuser would also greatly reduce the wobble
and misalignment that accompanies the sliding-tube
focusing technique used. Field stops of 750, 400, and
150-micron proved a reasonable choice. Modification
of the optical train for quick change and registration
of the f-stops would significantly reduce down time.
The absence of a reference star is problematic
when lower frequency phenomena are investigated,
e.g., flares, eclipses. One solution is use of the intensity data from the guide star as a reference. These
data could be easily stored on one of the extra channels of the data acquisition system and used to eliminate the slower variability from atmospheric transmission changes occurring over periods of seconds.
Considering the 5-study objectives:
1.
Confirm the range of data gathering of both

magnitude and BW.
Sampling rates to 1000 samples/sec were possible to 8th magnitude providing the filter used
matched the spectral characteristics of the target. A magnitude dimmer is possible without a
filter. For brighter targets sampling rates to
10,000 samples/sec are possible.
2.
Learn to separate noise from stellar information focusing on XB with known oscillations
in the X-Ray spectrum and suspect in the
optical.
Stellar oscillations were found on X Per at
383.86 Hz, and on BM Ori at 166-7 Hz, possible QPOs. No literature confirmation was
found for these frequencies. Use of digital filtering and FFT processing easily separates
these signals from the lower frequency scintillation noise. Attention must be given at all
times for elimination of system noise from
ground loops, other electronic equipment, and
optical sources visible to the sensor.
3.
Refine equipment and procedures for improve

results.
Changes to the optical path for better initial
alignment and tracking were successful.
4.
Confirm the capability for data collection on

lunar occultation events and processing of
data.
The lunar occultation data for XZ-5588 was
174
analyzed with the CAL model and found to be

an unresolvable point source at high confidence.
5.
Firm up plans for next steps including working

with a researcher .
Identification of a researcher for collaboration
on HTRA in high frequency oscillations and\or
flaring is underway. Additional data collection
on lunar occultations will also be pursued.
8. Acknowledgements
The author thanks Larry Owings for his development of the control software for the flip pellicle
beam splitter. His speed and thoroughness on the
project was greatly appreciated. Thanks and appreciation to Andrea Richichi of the European Southern
Observatory and National Astronomical Research
Institute of Thailand for his use of Figure 1 and CAL
model data reduction for occultation of XZ-5588.
9. References
AAVSO (2012). American Association of Variable
Star Observers. http://www.aavso.org/
Blow, G. L. (1983). Lunar Occultations. Solar System Photometry Handbook. (R. M. Genet, ed.) pp. 91 to 9-25.
DOA (2012). The Dutch Occultation Association.
Lunar Occultation Workbench, LOW Software v.
4.1. http://www.doa-site.nl/
IOTA (2012). International Occultation and Timing
Association.
http://lunar-occultations.com/iota/iotandx.htm
Keisan (2012). High accuracy scientific calculations.
http://keisan.casio.com/has10/SpecExec.cgi?
id=system/2006/1180573477
Kreiner, J. M. (2004). Up-to date linear elements of
close binaries. Acta Astronomica 54, 207-210.
http://www.as.up.krakow.pl/ephem/
Measurement Computing Corporation (2012). IOTech
Data
Acquisition
Equipment.
http://www.mccdaq.com/products/
daqboard_series.aspx
Menke, J., Vander Haagen, G. A. (2010). High
speed photometry detection and analysis techniques.
The Alt-Az Initiative. (R. M. Genet, J. M. Johnson, V.

Wallen, eds.) pp. 443-469.
Miyashita, K. (2008). LiMovie software for video
analysis of occultation events. http://www005.upp.sonet.ne.jp/k_miyash/occ02/limovie_en.html
Nather, R. E., Evans, D. S. (1970a). Astron. J. 75,
575-582.
Vander Haagen, G. A. (2008). Techniques for the

study of high frequency optical phenomena. in Proceedings for the 27th Annual Conference of the Society for Astronomical Sciences. (Warner et al., eds.)
pp 115-122. Society for Astronomical Sciences, Rancho Cucamonga, CA.
Nather, R. E., McCants, M. M., (1970b). Photoelectric Measurement of Lunar Occultations IV. Data
Analysis. Astron. J. 75, 963-968.
Vander Haagen, G.A. (2011). The Silicon Photomultiplier for High Speed Photometry. in Proceedings for the 30th Annual Conference of the Society for Astronomical Sciences. (Warner et al., eds.)
pp 87-96. Society for Astronomical Sciences, Rancho
Cucamonga, CA.
National Photocolor, Inc. (2012). Manufacturers of

pellicle mirrors and beam splitters.
http://www.nationalphotocolor.com/specs.html
Warner, B., Woudt, A. (2008). QPOs in CVs: An

executive summary. Dept. of Astronomy, University
of Cape Town, South Africa.
Richichi, A. (1989). Model Independent Retrieval of

Brightness Profiles from Lunar Occultation Light
Curves in the Near Infrared Domain. Astron. Astrophys. 226, 366-372.
Richichi, A. (2004). Combining Optical Interferometry with Lunar Occultations. ASP Conference
Series 318, 148-156.
Richichi, A., Roccatagliata, V. (2005). Aldebarans
angular diameter: How well do we know it? Astron.
Astrophys. 433, 305-312.
Richichi, A. (2010). Stellar Oscillations and Occultations. High Time Resolution Astrophysics IV.
Agios Nikolaos, Crete, Greece, May 5, 2010.
Richichi, A. (2012). Personal communications.
Richmond, M. (2005). Diffraction effects during lunar occultations.
http://spiff.rit.edu/richmond/occult/bessel/bessel.html
sensL Technologies Ltd. (2012). Manufacturers of
SPMs.
http://sensl.com/products/silicon-photomultipliers/
SignalLab (2012). Sigview 2.3 software for DSP applications. http://www.sigview.com/index.htm
Soydugan, E., Soydugan, F., Demircan, O., Ibanoglu,
C. (2006). A catalogue of close binaries located in
the Scuti region of the Cepheid instability strip.
Not. R. Astron. Soc. 370, 2013-2014.
Uthas, H. (2005). High-Speed Astrophysics: Rapid
Events Near compact Binaries. Lund Observatory,
Sweden.
175
176
Photon Counting - Stanton
Photon Counting One More Time

Richard H. Stanton
JPL Retired
43665 Oak Leaf Trail
Three Rivers, CA 93271
rhstanton@gmail.com
Abstract
Photon counting has been around for more than 60 years, and has been available to amateurs for
most of that time. In most cases single photons are detected using photomultiplier tubes, old technology that became available after the Second World War. But over the last couple of decades the
perfection of CCD devices has given amateurs the ability to perform accurate photometry with modest
telescopes. Is there any reason to still count photons? This paper discusses some of the strengths
of current photon counting technology, particularly relating to the search for fast optical transients.
Technology advances in counters and photomultiplier modules are briefly mentioned. Illustrative data
are presented including FFT analysis of bright star photometry and a technique for finding optical
pulses in a large file of noisy data. This latter technique is shown to enable the discovery of a possible
optical flare on the polar variable AM Her.
1. Introduction
An ideal photon counting system has at least four
properties making it attractive for astronomical photometry:
1.
Given 100% quantum efficiency and zero

noise, one can do no better than count the arriving photons vs. time
2.
Since all information is contained in short duration photon pulses (~30ns), very high-speed
photometry is possible, without any increase
in measurement noise, simply by sampling
these counts at a high rate.
3.
These count samples start out as digital data

that can be binned to give several views of the
same data with different time resolutions. Subsequent analyses can draw directly on powerful digital processing techniques, such as filtering and FFT analysis, to bring out additional information.
4.
Being digital in nature, photon counting is inherently linear over a large brightness range,
typically on the order of ten stellar magnitudes.
Of course real devices have neither 100% quantum efficiency nor zero noise. But modern detectors
can come close to this ideal. CCDs have pushed
quantum efficiency beyond the 70% level while photomultiplier tubes often achieve respectable levels of
25 to 30%. The very low noise readout electronics in

many CCD cameras make them ideal for a wide
range of astronomical applications, including photometry (Howell, 2000). But most CCD systems are
not configured for fast photometry, or for counting
individual photons. Video camera systems can do an
excellent job of recording high-speed events such as
asteroid occultations (Menke, 2007), but they usually
operate at fixed bandwidth and also dont count single photons. Finally, newer devices such as the Silicon Photomultiplier described by Vander Haagen
(2011), have interesting potential, but are apparently
not yet demonstrating performance competitive with
readily available commercial photomultiplier tubes
(PMTs).
PMTs have long boasted the ability to count individual photons with essentially zero noise. Now the
availability of modern tube modules with integrated
electronics means that amateurs no longer have to
deal with high voltage power supplies and high-speed
amplifier/discriminator electronics. Simply plug in a
5V supply and you are ready to start counting photons!
This paper focuses some aspects using PMTs for
astronomical photometry, a few of the interesting
objects one can observe and example techniques for
analyzing large files of count data. Finally, a brute
force but effective way to search for pulses in noisy
data is presented, along with an unanticipated success story.
177
2. Early Years
When the author started counting photons in the
mid-seventies there were many obstacles to overcome. One needed a PMT specifically designed for
photon counting, a high-voltage power supply, an
amplifier / discriminator to generate countable pulses,
a high-speed counter and a computer to store the
data. Since personal computers were just arriving on
the scene, some assembly-level coding was needed to
make it all work. The author (Stanton, 1983) has described some of this early complexity. With luck and
a lot of soldering and programming, one might be
able to gather and store 10,000 points of photon data
in a single run. Although this system was complex
and strictly limited in the data capacity, it did demonstrate some of the strengths of photon counting. The
next four figures illustrate the power of this early
system for low noise photometry.
Figure 1. The polar variable AM Her in an active state on

June 14, 1983. Points are phonon counts per 2s, no filter
with sky background subtracted.
Figure 1 shows approximately one hour of the

count rate for the variable AM Her in an active state.
All data in this paper were taken using a 16-inch Cassegrain telescope built in the 1960s by Cave Optical
Co. For polar variables stars like AM Her, material
falling from the secondary star to its white-dwarf
companion is channeled by an intense magnetic field
to a small spot on the dwarfs surface. The obvious
flickering seen here is believed to be due to changes
in the rate of material flowing in this process.
A second example from this early system is the
light curve of the cataclysmic variable HT Cas, captured during one of its rare outbursts (Figure 2). This
light curve shows the deep eclipses that occur every
orbit when the systems bright accretion disk is
eclipsed by the darker secondary star. Also seen are
the transient superhumps that move through the
light-curve with a period somewhat longer than the
systems 106 minute orbit.
178
Figure 2. Variability of cataclysmic variable HT during

outburst on 1985 January 20. V-filter with background
counts subtracted. Note motion of the bright superhumps relative to the eclipses.
These days light curves like these can be readily

duplicated using CCD imaging. The greater quantum
efficiency of these devices, the relative ease of recording repeated frames of data, and the ability to
simultaneously measure an entire field of comparison
stars, give CCDs a clear edge for observing faint stars
at this time resolution.
The next two figures illustrate a regime where
the advantage still belongs to PMTs. As mentioned
above, one of the advantages of a photon counting
system is its ability to acquire data at high time resolution without introducing extra noise into the signal.
Figure 3 illustrates this with count data gathered with
millisecond resolution during a lunar occultation of
75 Tau. Note that the main Fresnel diffraction peak is
clearly visible at this resolution.
Figure 3. Lunar occultation of 75 Tau on 1979 March 5.
Figure 4 illustrates the serendipitous discovery

of the close binary Hu 1080 made the same night as
the moon passed through the Hyades. This binary is
not well known to amateurs since the separation of its
6.6 and 7.3 magnitude components varies in the

range 0.04 to 0.4 over its 40 year orbital period. It
was apparently near its widest separation when this
occultation occurred.
ula pulsar. But it also has the potential for finding

currently unknown fast signals that professional
astronomers simply dont have the time or inclination
to look for. Jocelyn Bell Burnell, discoverer of pulsars 45 years ago, more recently (2004) put it this
way:
we have neglected, incidentally, timevariability. The X-ray astronomers are alert
to it, the radio-pulsar people are alert to it;
gamma-ray astronomers are too, but only
now are the optical astronomers starting to
study things that go bump in the night. Its a
fascinating and under-researched topic.
4. Data Analysis
4.1 Analysis Approach
Figure 4. Lunar occultation of binary star Hu 1080 on
1979 March 5. Note that both components exhibit
brightening just before disappearing due to Fresnel
diffraction.
3.
Hardware
Modern technology has made photon counting

much easier to implement:
1.
Photomultiplier, high-voltage supply and amplifier /discriminator electronics are now

available encapsulated in a small, light weight
package. One literally inputs 5V and gets photon pulses out with these modules. The author
currently uses a Hamamatsu H8259-01 which
gives a broad spectral response, QE peak of
~25%, a pulse-pair resolution less than 30 ns.
2.
High speed, multiple channel counters are now

available as relatively inexpensive PC boards.
These boards feature DMA data transfer to the
host computer, software-controlled functionality, and the ability to resolve pulse pairs separated by as little as 15 ns (see for example, National Instruments PCI-6602 Counter Timer
board).
3.
In conjunction with these counters, modern

PCs can gather and store files of hundreds of
millions (even billions) of data points per run.
And the processing power of PCs enables efficient analysis of these huge files.
What can all this capability do for the amateur?

It gives him a tool for going after known high-speed
optical phenomena, such as lunar and asteroid occultations and more exotic targets such as the Crab Neb-
In addition to hardware for generating time series

of photon data, there are powerful analysis tools
available that can help make sense out of the vast
data files these systems generate. Two of these tools,
digital filtering and FFT analysis, are illustrated here.
Except where noted, all data in the remainder of this
paper were gathered using the same set of timing
parameters to give a consistent starting point for
analysis. These parameters, a 3-s sample time and
300-second observation duration, result in files of
100 million data points each. While the resulting files
dramatically over-sample any expected target brightness variation, they provide ample data for testing
various data processing techniques. Of course one
can select almost any desired time resolution by simply combining the 3s samples into longer sample
bins.
4.2 Digital Filtering

Figure 5 illustrates a small segment of data taken
of the bright star (21 Lyr) and binned to give 96-s
samples. The average photon rate for this star was
approximately 3.5x106 counts/s. Both dark count
(typically 10/s) and sky background (around 5000/s)
are negligible in his case. As can be seen in Figure 5,
two types of noise dominate this data. The first is the
large amplitude, low frequency variation due to atmospheric seeing. The second, high frequency pointto-point variation, is due to statistical shot noise.
Shot noise is always present in photon count data
since it is inherent in the statistical nature of the incoming light itself.
179
4.3 Frequency Spectrum Analysis

These two sources of noise also leave clear signatures in the frequency spectrum of the data. Figure
7 gives the power spectrum of the original
21 Lyr data, calculated using an FFT program available on the Internet (Ooura, 1998). In this case the
original 3-s samples are combined into samples 768
s long. Note that while each FFT spectrum uses only
2048 of these samples, the final curve averages about
380 of these individual spectra. This technique
greatly reduces the noise content in the final spectrum.
Figure 5. A small section of count data for the bright star
21 Lyr, binned to 96 s per sample.
The seeing signature can be easily eliminated

with a high-pass digital filter. Finite Impulse Response (FIR) digital filters can be designed as highpass, low-pass or band-pass using readily available
software programs (Rabiner and Gold, 1975). Typically these filters can achieve a high level of attenuation for stop-band frequencies, while having a negligible effect in the pass-band.
Figure 6 shows the result of passing the raw data
(Figure 5) through a high-pass digital filter cutting
off frequencies below about 300 Hz. While this filter
removes all trace of seeing from Figure 5, it also
eliminates any low-frequency information that was
also in the signal. Note that, with the removal of seeing noise, shot noise alone is responsible for the significant scatter still seen in Figure 6. The mean data
value is now zero since the lowest frequency (zero
Hz) has also been removed.
Figure 6. Data from Figure 5 after passing through a

high-pass digital filter.
180
Figure 7. Power spectrum of the 21 Lyr data, after

binning to 768 s per sample.
The spectrum in Figure 7 can be understood in

terms of the two noise sources discussed above. The
increasing power level at low frequencies is clearly
due to atmospheric turbulence. As frequency increases, the atmosphere makes less of a contribution
until it has essentially no effect on the data above 200
Hz. The interesting wiggles in this curve are almost
always present, perhaps due to the interaction between seeing cell size and the telescope aperture and
obscuration.
In the right half of Figure 7, the detection floor
is determined by the shot noise in the signal. This
noise is sometimes referred to as white noise since
it makes an equal power contribution to all frequencies. Any signal that produces a power level significantly below this floor will not be detected. However, a small signal that rises above this level can be
found, even though it is totally invisible in the original time-series data.
Such a case is shown in Figure 8. This figure
again uses the 21 Lyr input data, this time binned to
192-s samples to show higher frequencies. This plot
dramatically illustrates that the atmospheres influence is totally absent for higher frequencies. We
would get the same spectrum above 200 Hz even if

the data were taken on the moon!
Figure 8. Power spectrum of the 21 Lyr data, after

binning to 192-s samples to show higher frequencies
(note that the vertical scale has also been expanded
relative to Figure 7).
Pulse duration is also a key parameter. A long

pulse might not be detected with a short sample time.
Conversely, a short pulse smeared out over a long
sample period might also escape detection.
Finally, the amplitude of a pulse centered exactly
on the boundary between two samples will be cut in
half relative to a more fortuitous positioning. The
result could be that a perfectly viable pulse passes
undetected.
The algorithm described graphically in Figure 9
can address each of these concerns. Starting with a
nominal file of 108 samples, the data are searched for
the largest count value. This value and position of
this maximum is saved, along with related statistical
data (mean and variance). Next, samples in the original file are combined, two at a time, to provide an
input file for the next iteration. In order to increase
the chance of detecting a pulse split between two
samples in this new file, a second binned file is computed, staggered by one input sample.
The resonance seen near 800 Hz was a total surprise. Could this be a message sent by an advanced
civilization orbiting 21 Lyr? Perhaps some optical
tone sent to attract our attention? Alas, not this time.
After looking at the data from other stars brighter and
fainter than 21 Lyr it became clear that this was some
type of instrument artifact that only affects data from
the very brightest stars. Whether this artifact arises
from the PMT module, the counter module, or some
defect in the setup, it has not been seen in data from
stars 6th magnitude or fainter.
4.4 Searching for Pulses

While the power spectrum can reveal the presence of a low-level repeating waveform, other approaches must be used to find isolated pulses in noisy
data. The author developed a program to search for
pulses, assuming that nothing is known a priori about
the pulse. The program uses an intuitive approach,
rather than one based on mathematical theory. The
following unknown parameters must be considered in
the search:
Pulse amplitude
Pulse duration
Position relative to discreet samples
The amplitude of the pulse must be large enough
to be detected above the noise. It is assumed that an
isolated pulse that is smaller than the peak noise
pulses will not be found. If it is a repeating pulse it
may be detected in the power spectrum, but this
search assumes that only single, isolated pulse exists.
Figure 9. Program flowchart for finding peak values in a

data file.
181
Each of these two new files is then treated as a

new input file, and the process of identifying the peak
repeated for both files. The result of this process for
through 13 iterations of a faint test star is illustrated
in Figure 10.9 The peak value for each iteration is
plotted against the number of original samples
summed (Bin Ratio). Since each iteration doubles the
length of the count sample, the largest sample size
corresponds to 213 = 8192 samples of the original 3s
data added together. Note that, except for the initial
point, there are two points for each bin ratio, corresponding to the two staggers mentioned above. In this
case, for many of the bin ratios, the measured peaks
for the two staggers fall on top of each other.
Figure 10. Peak values found vs binning ratio for file of

th
highly attenuated sunlight (equivalent to a 15 magnitude star).
The second curve in Figure 10 plots the ratio

standard deviation divided by the square root of the
mean value. For purely random shot noise this value
should be unity. If seeing or source variability add to
the measured standard deviation, this ratio rises
above 1.0. Some instrument effects (e.g. deadtime)
can drive it below unity. It is plotted here as a check
on data randomness.
As one might expect, the peak value increases
as the sample length increases. But since the count
rate is very low (0.0028 per 3-s sample) it doesnt
increase dramatically even after 8192 samples are
binned together. Note that the number of samples
drops proportionally as longer samples are created
from the same input data. With fewer samples, one
expects the probability of a truly large false signal
(noise pulse) to decrease.
But what does statistics tell us to expect? We
know that the Poisson distribution should apply for
counting random events, even when the mean value
is very small. For this distribution, the probability P
9
This star was actually attenuated sunlight to reduce seeing

effects for this special test case.
182
of finding a value of k counts, given a mean count

value of is given by
P(k, ) = ke- /k!
(1)
Given N data points in a run, the number of these

points expected with exactly k counts per sample is
simply
n(N, k, ) = N P(k, )
(2)
One can use the probability of having a single

pulse in the data (1/N), and the mean count value ,
to calculate the peak value k, that is likely to occur,
on average, once in that pass. Although k is an integer in Poissons equation it is possible to construct a
continuous curve by interpolation. If this is done correctly, one would expect measured peaks in a set of
random count values to all fall close to this curve. If
a measured peak value for one or more bin levels
falls well above the curve it should be looked at as a
possible optical pulse. Of course for longer pulses,
atmospheric seeing may also play a role.
Figure 11 shows the results of this calculation for
the test case previously described. The solid curve
shows the expected peak value vs. bin ratio, agreeing
quite well with the data except for bin levels less than
10. This agreement suggests that this automated approach to searching huge files of photon count data
can reliably locate pulses that are larger than expected statistically. The dashed curve in the Figure 10
shows the peak size predicted the Gaussian distribution. As expected, this distribution does not work
well for small mean count values but approaches
Poisson for larger values.
Figure 11. Predictions of Poisson and Gaussian distribution compared to data in Figure 10.
Why are the peak values for bin levels less than
10 significantly higher than predicted? One might
argue that were only talking about samples that are
one or two photons larger than predicted. Is this

really important? Perhaps not, but this effect is seen
for virtually every faint star case measured to date.
No good explanation has yet been found for this deviation.
Setting aside these very small pulses, one can get
a sense of the power of this process at separating real
pulses from noise by looking at some numbers. Consider the bin level 1024 in Figure 11, corresponding
to 1024*.002838 = 2.906 counts per sample. The
number of samples at this level is 108/1024 = 97,656.
Using these values the computer estimated a value of
k ~ 13 as the count value that is likely to occur only
once in this data set (solid curve in Figure 11). Equation (2) gives the following expected number of occurrences for this case:
Pulse size (k)
12
13
18
Expected # of pulses (n)

4.0
0.9
0.00018
ticularly for pulses corresponding to a bin ratio

around 1000 (Figure 12). The plot indicates that the
pulse(s) found are about twice as large as noise peaks
predicted by Poisson (or Gauss), far larger than anything expected statistically.
When the time series data for this bin ratio is examined, two very large pulses are found (Figure 13).
Unlike the flickering usually associated with this
variable (Figure 1) these appear to be short, intense
flares corresponding to a brightness increase of at
least two magnitudes with durations of roughly 36
and 66 ms. Are these pulses real? There is little
doubt that they are caused by actual optical pulses,
not noise or some system artifact. But are they associated with AM Her? Could they just be light from a
passing airplane or distant lightning? At least one
other instance of a strong optical flare from this star
has been reported (Cherepashchuk et al., 1996), but
its not certain that this flare was anything like those
shown in Figure 13.
Evidently pulses significantly larger than predicted

by the Poisson curve are statistically very unlikely to
be due to random noise, at least above bin level ~10.
5. AM Her Pulses
Figure 13. Pulses responsible for the non-random signature in Figure 13. These are roughly twice the size of the
largest pulse expected at this resolution, a huge difference.
Figure 12. Pulse search results for a file of AM Her data.

Divergence from Poisson curve at bin rations above 100
indicates presence of peaks far larger than those expected for random noise.
Over the last few months, the analysis tools described in Section 4 were applied to photon data from
a variety of bright and faint stars. For the most part
the pulse search results were consistent with Poisson
statistics of random noise, except for very low bin
ratios, as discussed above. This included a number of
observations of the AM Her. A notable exception
occurred for one observation made on 2011 Sept. 4.
The pulse size plot for this observation shows a
significant deviation from Poisson expectations, par-
More observations are needed! But this event

emphasizes a very real problem when searching for
isolated pulses or other transient events. How does
one separate noise such as airplanes, satellites,
lightning, illuminated birds, etc., from a real signal?
Having two widely separated telescopes that simultaneously observe the same target could eliminate most
artifacts. The author is currently implementing another approach, monitoring the light level reflected
off of the photometer diaphragm. Any pulse that appears in this reflected light is clearly not associated
with the target star.
6. Conclusion
There is much that can be done by amateurs using photon counting. Very capable equipment is now
183
available at moderate prices. Combined with a midsize telescope such equipment can observe many interesting, rapidly varying objects. The possibility
certainly exists that an appropriately equipped observer will someday discover some previously unknown class of fast astronomical phenomena.
.
7. References
Burnell, J. B. (2004). Pliers, Pulsars, and Extreme
Physics, in Radiations of Sigma Pi Sigma 10, pp 510.
Cherepashchuk, A, et al. (1996). Highly Evolved
Close Binary Stars: Catalog. Gordon and Breach
Publishers, p. 291.
Howell, S. B. (2000). Handbook of CCD Astronomy.
Chp. 5. Cambridge University Press, Cambridge.
Menke, J. (2009). The Addictive Properties of Occultations. in 28th Annual Conference of the Society
for Astronomical Sciences (Warner et al., eds.). pp. 917. Society for Astronomical Sciences, Rancho
Cucamonga, CA.
Ooura, T., (1998). Freeware program fft4g.c
Rabiner, L., Gold, B. (1975). Theory and Application
of Digital Signal Processing. Chp. 3. Prentice-Hall.
New York.
Stanton, R. H., (1983). Design and Use of a Computer-Based Photon Counting System in Advances
in
Photoelectric
PhotometryVolume
1
(R.C.Wolpert, R. M. Genet, eds.). pp. 87-129.
Vander Haagen, G. A. (2011), The Silicon Photomultiplier for High Speed Photometry. in 30th
Annual Conference of the Society for Astronomical
Sciences (Warner et al., eds.) pp. 87-96. Society for
184
Tracking Bolides Kaye et al.
Tracking Bolides, 3D Visualization and Data

Thomas G. Kaye
Foundation for Scientific Advancement
7023 Alhambra Dr., Sierra Vista, AZ 85650
tomkaye@u.washington.edu
Robert Crawford
Tucson AZ
Mark Bowling
Tucson AZ
John Kalas
Tucson AZ
Abstract
For the past decade progress has been made in using consumer-grade low light video cameras to watch the
night skies for meteor trails. Sophisticated software can now monitor the video stream in real time and parse out
the events of interest from a continuous recording. Sandia Laboratories has funded over 100 of such cameras
spread throughout the country. Their primary purpose has been to record large fireballs. The fireball data is primarily used to compare with space based observatories that are watching for nuclear explosions on Earth. The
Desert Fireball Array in southern Arizona has deployed three of these cameras in overlapping visual fields. The
mutual detection of events allows for three-dimensional reconstruction of the entry angle, speed, brightness and
spectroscopy. The stated goal of the project is to use the tracking information to locate and recover a meteorite
that was tracked through the atmosphere. The acquired data would allow direct comparison of atmospheric ionization characteristics to ground based laboratory analysis using simulated ablation of the same meteorite. To
achieve this goal, high precision calculations and final analysis using 3D projections in Google Earth will be described.
1. Introduction
Meteors are the least expensive sample return
missions available on Earth. Known meteor showers
have been a subject of intensive studies especially in
Europe (Berezhnoy, 2010; Borovicka, 1994). The
sporadic meteors however present a challenge for
research due to the random nature of their arrival and
difficulty of recovery. While many large random bolides have been captured on home video, comparatively few have been recovered and almost none with
reliable reentry data. Most data available from previous studies consists of velocity and brightness information (Borovika, 2007). Spectroscopic data is
available for showers (Madiedo, 2011) but almost
non-existent for sporadics. The upper atmosphere
where bolides begin ionization is molecularly extremely thin and is at a higher vacuum than can be
achieved on Earth. This leads to unusual ionization
species that normally recombine instantly in a lower
vacuum plasma environment but in the upper atmosphere, live long enough to display forbidden lines
in their spectra.
Velocity upon reentry while consistent in meteor

showers varies widely with sporadics and has not
been well characterized. Velocity and entry angle
combine to influence the survival potential of a meteor through the burn phase. Certain orbit trajectories
can closely match Earths orbital velocity and are
very favorable for meteor survival. These bolides
exhibit slow apparent velocity and deep penetration
before dark flight.
For the past three years the group known as the
Desert Fireball Array has been running three cameras
on a triangular 100 kilometer baseline that covers the
south eastern corner of Arizona.
2. Materials and Methods

The initial Sandia Labs camera is pictured in
Figure 1. The housing uses standard PVC fittings
with an acrylic dome. Self-contained in the housing is
the camera, power supply, heating unit and temperature control module. It is powered by 120 VAC and
feeds an analog video signal via the coax cable to a
commercial computer video card at video frame
rates.
185
The software is proprietary to Sandia and scans

each series of 5-10 frames looking for brightness
changes above a certain threshold. It is continuously
recording and holds approximately the last 30 seconds in a memory buffer at all times. Once an event
is triggered, the buffered recording allows the entire
event to be saved with no clipping due to detection
lag. The resulting data file displays a subframe of the
video with the target centered in the field. A keystroke will provide a light curve for the duration of
the burn.
shoulder of the third camera to maintain a 360 degree mutual detection zone as shown in Figure 3.
Fig. 2 Second generation camera housing with 100 degree field of view.
Fig. 3 The three cameras overlap the field of view within

the 100 kilometer baseline. Dual camera detections are
possible out to several hundred kilometers.
Fig. 1. Sandia video camera housing. System incorporates PVC fittings for waterproofing with internal heater
and power supply.
Figure 2 shows the second generation housing

developed by the Fireball Array team. It is based on
an inexpensive waterproof electrical enclosure that
has been modified with an optically flat side window.
The heater and onboard power supply have been dispensed with in order to save space and it is now powered by an external supply. The original camera utilized a 360-degree fisheye lens for horizon-to-horizon
coverage. In practicality, bolides close to the horizon
are typically many hundreds of kilometers away and
do not produce useful data. The horizon in view also
creates multiple false triggers from cars etc. Experimentation showed that three cameras using lenses
with a 120 degree field-of-view, in a triangular deployment, provided a higher resolution image. In this
arrangement, each pair of cameras looks over the
186
Fig. 4 Automatically generated masking in UFO Capture

in dark blue. This reduced the number of false hits by
approximately 95%. Light blue is a captured meteor trail.
The Sandia software has now been replaced with

UFO Capture (Sonotaco.com) that was developed in
Japan. The new software features a self generating

mask that blocks areas of twinkling lights such as
stars or streetlights that continually vary (Fig. 4).
This very powerful feature eclipses other software
and drastically reduces the number of false detections, which can easily fill a hard drive in one night.
Astrometric calibrations are handled by matching a
star field to the mask. A least-squares calibration between the field and the mask determines RA and
DEC for each pixel. In Japan, UFO Capture is used
primarily with narrow field camera lenses (25~40
degree field of view). It remains to be determined
whether the camera model used in the astrometric
calibration can represent the greater distortions present in fisheye lenses.
Fig. 5 Reconstructed path through the atmosphere in

Google Earth. The position data was extracted for each
frame.
Combining the astrometric position information

from two or more cameras allows for a three dimensional reconstruction of the bolides path through the
atmosphere. In order to determine a potential impact
zone, Google Earth data files were created from the
multi-camera reentry data (Fig. 5).
The 3D rotation in Google Earth allows rotation
in order to sight down the barrel of the bolides
trajectory and define a spot on the ground as the
maximum, or farthest point of impact for the search
ellipse (Fig 6.) Additional rotation allows for the end
of the burn phase to be viewed normal to ground,
thereby determining the minimum, or nearest position
for the ellipse discounting wind.
The light curve for the bolide is extrapolated
from each frame of the video stream. The light curve
is superimposed three dimensionally on top of the
trajectory again using Google Earth. Penetration into
the atmosphere is then visually correlated with
brightness (Fig. 7).
Fig. 6 Using Google Earth to calculate the search area

for a fall. Upper image is looking down the barrel of
the bolide path in order to determine the furthest extent
of the strewn field. Lower image is a side view of the
trajectory in order to determine the minimum starting
point for the ellipse of the strewn field.
Fig. 7 White vertical lines above the red reentry path

indicated the relative brightness during the burn phase.
In this configuration the light curve can be visually correlated to the penetration into the atmosphere.
187
Proprietary software has been created in Matlab

to triangulate a bolides position in three dimensional
space on a frame-by-frame basis. For events seen by
two or more stations, the video recordings are aligned
in time to sub-frame accuracy (<1/30th second) by
correlating the light curves. The bolides pixel position in each video frame is turned into a line-of-sight
vector based on the astrometric calibration of each
cameras orientation on the sky. The lines of sight
from two or more stations are then intersected to find
the estimated location of the bolide that approaches
all lines of sight most closely at that instant in time.
The triangulation takes account of the geographic
location of each station, the rotation of Earth, and the
location of Earth in its orbit around the Sun to compute the instantaneous position of the bolide in Cartesian XYZ coordinate systems, both geocentric and
heliocentric. Frame-to-frame differences in the positions are used to compute the bolides velocity. With
careful astrometric calibration of each cameras orientation on the sky, scientifically useful data on bolide dynamics can be extracted from the video
frames.
Spectroscopy is accomplished through a standalone system that incorporates a CCD camera with a
low F-number lens in front of a transmission grating.
The camera is set to continuously expose 30-second
images through the night. This systems visual field
overlaps the center of the three detection cameras. A
detection from one of the other systems is time
stamped and allows the proper frame to be extracted
from the nights sequence of frames. Figure 8 shows
a reentry and its associated spectra. Each horizontal
row of pixels produces a spectrum so the entire evolution of the ionization and ablation process can be
analyzed.
Fig. 8 Spectrum of bright bolide.
3. Discussion
Data on meteor showers is available for virtually
all of the major events. Efforts in the 1980s and
1990s centered around film and continuously recorded video, which were cumbersome and lacked
188
sensitivity. Recent interest in sporadics has been

helped by the marriage of computer recorded video
and low light video cameras bringing new technology
to bear on the problem. The original systems brought
out for meteor detection were either proprietary or
Linux-based, which made wide-scale deployment
difficult (ASGARD). The second generation of software described here, makes video integration much
easier and the detection algorithms are much more
robust.
This projects first goal is to acquire as much
data as possible including vector, velocity, brightness
and spectroscopic information. This data has been
available before but never has the actual meteorite
that generated the data been available for analysis.
Using the 3D reconstruction, a search ellipse becomes apparent and a local team of searchers is
standing by to fulfill the second goal, which is to
acquire the meteor that generated the data.
Once in hand, laboratory analysis of the material
can be directly compared to reentry data to ground
truth the observations. Laboratory work can identify
limits of detection, anomalies in the spectra, temperature and ablation characteristics which can all be
compared to the atmospheric data. This will allow
refinement of the characteristics of bolide reentry
and, with further processing, the reentry vector can
be extrapolated into orbital parameters as is currently
being done by NASA to determine the original
source of the bolide (NASA).
4. Conclusions
The technology for full-time automated detection
of meteors has finally come of age. The combination
of inexpensive low light cameras and high performance real-time analysis of the video stream, creates a
powerful tool for bolide research. Visualization tools
such as Google Earth are bringing the data to life in
simple but effective ways and allowing an unprecedented look into all aspects of Earth crossing bolides.
The instrument infrastructure is now in place and all
that is necessary is the cooperation of a random bolide to enter the atmosphere in the right area. In the
last several years at least three large bolides have
fallen inside the detection area. One of these was
recovered and is now known as the Whetstone fall
(Wikimedia.org). Unfortunately, technical problems
had cameras off-line during these events and a mutual detection was not recorded. The technical difficulties have now been overcome and multiple cameras are scanning the skies on a regular basis just
waiting for the next big event.
5. References
ASGARD, All Sky and Guided Automatic Realtime
Detection. Meteor detection software.
meteor.uwo.ca/~weryk/asgard
Berezhnoy, A. A. (2010. Formation of molecules in
bright meteors. Icarus 210, 150157.
Borovika, J., Spurn, P. (2007). Atmospheric deceleration and light curves of Draconid meteors and
implications for the structure of cometary dust.
Astron. Astrophys. 473, 661-672.
Borovicka, J. (1994). Two components in meteor
spectra. Plan. Space Sci. 42, 145-150.
Madiedo. J. M. (2011). Emission Spectra of Geminid Fireballs from 3200 Phaeton: Preferential Depletion of Volatile Phases. IAU Symposium.
NASA, All Sky Fireball Network,
http://fireballs.ndc.nasa.gov
Sonotaco.com, UFO Capture Software.
Trigo-Rodriguez, J. M., ed. (2008). Advances in Meteoroid and Meteor Science, Springer Pub. 564 pgs.
Wikimedia.org, Whetstone Mountains Meteorite,
http://commons.wikimedia.org/wiki/File:Whetstone_
Mountains_meteorite.jpg
189
190
Asteroid Occultation Techniques - Hoot
Tools and Techniques for Measuring

Asteroid Occultations with DSLR and CCD Cameras
John E. Hoot
SSC Observatory
615 S. El Camino Real
San Clemente, CA 92672
Observatory@ssccorp.com
Abstract
Currently most asteroid occultations are measured with video equipment. This technique is limited to stars bright
enough to be measured at 30 frames per second and limits participation to observers that have portable low-light
video and time standard tagging equipment. This paper presents new observation tools and analysis methods
that allow the larger community of astroimagers to make precise occultation measurements with tracking telescopes and DSLR or CCD integrating cameras.
1. Introduction
Occultation events occur when one body, typically non-luminous, passes directly between an observer and a more distant luminous body, completely
blocking the distant bodys radiation. Occultations
are a sub-class of transit events. This special alignment of the observer, distant object, typically a star,
and the occulting body provides an opportunity to
determine the shape of the occulting body. The most
typical forms of occultation measurements are performed on asteroids passing in front of stars and lunar
limb features passing in front of stars.
For the case of asteroids, if we know the orbit of
the asteroid with sufficient precision to predict the
transit, we also know its distance to the observers and
its apparent velocity in the plane perpendicular to the
observers line of sight. With this information, simple
algebra and multiple observers it is possible to accurately render the cross sectional profile of the asteroid.
With the advances in precision stellar and asteroid astrometry in combination with low cost computing power, it is now possible to predict, with reasonable accuracy, many more occultation events than
ever before. For lack of sufficient observers, many of
these events are under recorded.
Currently, most occultation observers are diligent shadow chasers. They are typically equipped
with telescopes, low light video cameras, a time standard capable of imprinting accurate time codes onto
the video, and a video recording system. With this
equipment they do fantastic work. This type of
equipment is pretty specialized and requires a commitment to this type of observation that not many
observers are going to make. At the same time tele-
scopes, DSLR cameras, CCD cameras and computers

proliferate.
Drift scan imaging is a known technique for using this most common type of astroimaging of
equipment to measure occultation events. The
method allows integrating cameras to make accurate
occultation observations. Historically, the precision
of this method was limited due to time jitter in the
observation. This situation has changed with the proliferation of smart-phones, GPSs, computers and
wireless networks.
This paper presents an overview of how to make
drift scan occultation measurements using modern
timing and geodesy technology. It also presents an
overview of currently available free computing tools
to assist in making these measurements. Finally, it
introduces a set of freely distributed tools developed
by the author to facilitate occultation planning, observing and data analysis.
We are near a tipping point in occultation astronomy. The correct equipment is in enough hands.
Enough events are being predicated with sufficient
accuracy that observers may wait for events to come
to them. The tools are freely available. The time has
come when drift scan occultation observation now
belongs in every astroimagers repertoire. And perhaps soon it will become the predominant method for
occultation observation.
2. Drift Scan Fundamentals

Traditional video occultation measurement
methods point a telescope at the star to be occulted
and record the stars image 30 frames or 60 fields per
second. To measure an occultation one examines the
video frame by frame measuring the stars brightness.
191
The brightness of the star is plotted against the time

coded frames to look for the time and duration of the
stars dimming or complete disappearance. This is a
mechanization of the visual method of calling out the
events seen through the eyepiece, recording the observations on audio tape, while a time signal plays in
the background.
Drift scanning uses a different method. One
starts to make a traditional long exposure image, but
at a precise moment prior to the predicted occultation
the user disables, or retards, the clock drive of the
telescope. This results in stars being trailed. In some
variations of the method, the drive is re-enabled. By
measuring the star trails flux verses position in the
image, the precise time and duration of the occultation can be determined.
drive speed, you have complete control over the drift

rate. This allows you trade off timing accuracy for
improved signal to noise ratios by lowering your drift
speed.
These two advantages provide the ability to
measure events, albeit to lower resolution, than are
simply inaccessible to video, for a given aperture of
telescope.
The methods greatest weaknesses is that time
jitter in the triggering of the drift tends to be larger
than in video measurements. Even if the power to the
telescope drive is interrupted, flywheeling of the motors and energy stored in the scope power supply
capacitors will cause the drifting to begin some time
later than when it is triggered. In the case where the
drive servos are commanded to stop or retard, the
delay in the servo motor loop will cause the drifting
to be delayed by the servo latency.
4. Drift Scan Planning
Figure 1. Drift Scan Occultation Observation Example
Figure 1 shows an example of how a drift scan

observation appears in an image. In order to determine the precise time and duration of the occultation,
the observer needs to know the time that the drifting
began, and the rate of drift. Provided the observer
knows the rate of drift in pixels/second, reduction
becomes a straight forward measurement and calculation.
The first step in making an occultation observation is to locate events proximate to you. IOTA, the
International Occultation Timing Association, has a
wonderful web site that will show you predicted
events in your vicinity. I am particularly fond of
Steve Prestons Worldwide Occultation Prediction
Page. It provides a concise summary of upcoming
events with a notation of the broad geographical regions from which the event is expected to be visible.
If an event is going to be visible in your broad geographical region, you can then link to a more detailed
map of the shadows path across the Earth. If this
looks like something accessible to you, you can then
download a summary of the occultation event and the
visible path with 1 sigma error margins.
3. Drift Scanning vs. Video

Drift scanning has both disadvantages and advantages when contrasted with video measurements.
The greatest strength of the drift scan method is that
integrating cameras, particularly in the case of CCD
cameras, have better signal to noise ratios than video
cameras. Video signals rarely approach 10 bits signal
to noise ratios. DSLR cameras typically have 12 to
14 bits S/N ratios, and cooled CCDs can have 14 to
16 bits S/N ratios. Additionally, since you read out
your imager only once, you have only one unit of
readout noise per pixel, as opposed to 60 per second.
This latter advantage is only significant if your drift
rate is less than 60 pixels/second. The other strength
of this method is that by varying your plate scale or
192
Figure 2. Occultation Event Summaries
Once you have located an event, you need to

pick a site from which it will be observed. If you are
fortunate, the site may be your own back yard, or
your observatory site. More likely you will need to
travel some distance, so prior planning is key. To
facilitate this effort the author has developed a utility
that converts occultation event summaries to .kml
files that load as overlays to Google Earths application.

The application is titled Occultation Maps.
When run from a PC it presents a screen as shown in
Figure 3.
Once you have loaded an event, the header
summary of the event shall appear in the dialogs
main window. Assuming you already have Google
Earth on your computer, simply click on the Convert and View button. Shortly, the globe will be
displayed with the shadow path and 1 sigma error
estimate imposed over it
Figure 5. Shadow Path In Google Earth
Figure 3. Google Earth Map Generator
Figure 6. Site Selection
5. Making The Observation
Figure 4. Loading An Event
Once you have the path in Google Earth, it is

easy to use the interactive controls and overlays to
locate an observing site that provides you with convenient access and infrastructure. For the event pictured in Figure 5, it was fortunate that the path of
expected visibility passed directly over the Mount
Wilson Observatory.
With the permission of the facility superintendent, I was able to set up in their 16-inch dome to
attempt to view this event.
Since you should have selected your opportunity

well in advance, best practice is to setup and configure your gear a week or so ahead of time and rehearse. This will prevent nasty surprises, such as
overcrowded star fields, poor signal to noise ratios,
faulty cables, missing software, over filled disk
drives, etc.
The keys to making a good observation are being
on target, getting a good exposure, accurate timing,
and precise geodesy. All of these can be practiced in
your backyard BEFORE you go, should travel be
required. Once you can do it perfectly in your yard,
you can pack everything that is there and know you
will have all the tools required when you get to your
site. Even if the site is your own yard, you will have
worked out all the kinks when you are not under the
gun.
193
a function of your plate scale and the declination of

the target star.
The formula to convert pixels to time for Drive
Off observations is:
Figure 7. Cell phone GPS/Compass
First, work out the exact location from which

you will observe. This is easy if you are already using
Google Earth. Just put a pushpin on your site and
read out the latitude and longitude. Alternatively, if
you must change locations at the last minute, you can
retroactively locate the site in Google Earth, or often
smart cell phones now contain GPSs that can give
your exact location.
Next, select your drift method. The two primary
choices are Drive Off or Drive Guided. The
Drive Off method simply removes power from
your telescope and lets the Earths rotation move the
stars across your image frame. This is absolutely the
simplest method. It can be as simple as:
1.
Centering Your Star
2.
Starting Your Exposure
3.
Punching off the power switch on your telescope as you listen to WWV on a shortwave
radio.
4.
Waiting for your exposure to finish.
A more complicated method may use a computer

triggered switch or relay to interrupt the power to
your scope. This approach takes human reaction
times out of the equation and does not impart motion
to the scope. Regardless of the method used to perform a Drive Off observation, the drift rate will be
194
While this method is easy, it does mean that you

will need to employ focal reducers and/or binning of
your images to adjust the plate scale to fit the brightness and duration of your event.
The other observing method, that I call Guided
Drift, requires you to have a telescope with either an
autoguider port, computer commanded tracking rates
or slewing. In this method you command the telescopes servo system to alter its drive rate by initiating
a guiding command, or altering the drive rate for a
fixed amount of time, starting before the event and
ending after the event. This method requires a computer with a connection to your telescope, but has the
advantage that you can precisely tailor the drift rate
to suit your instruments sensitivity. Additionally,
data reduction does not require any astrometry to
solve plate scales on your images.
Figure 8. Drift Scan Trigger Screen Shot
The author has developed a PC application that

controls this process. Once you have entered your
event parameters, the software alerts you as to when
to start your exposure. After that, all the key timing
events are triggered automatically. The application is
titled Drift Scan Trigger. Figure 8 shows a screen
shot of the application.
To use the drift scan trigger program with telescopes that accept guiding and tracking commands
directly from their serial ports, simply connect your
PC to you telescope. The command sequences to start
and stop drifting can be entered in the Start Command and Stop Command windows.
In the example in Figure 8, the program is configured to command a generic Meade telescope to
apply east guiding for the drift duration. Since my
Meade telescope hand controller lets me specify a
guide rate from 1 to 100 percent of sidereal rates, I
can pre-configure my scope with a tracking rate appropriate to my target. A guide speed setting of 100%
effectively halts the drive system on my scope.
If your scope lacks the ability to serially command the drive system, it may still be equipped with
an auto guiding port. Typically, these are presented
as modular phone connectors where the grounding
individual pins cause the scope to slowly guide east
and west in RA and north and south in DEC. In these
systems, you can build a simple serial port controlled
switch to interface with the scopes guider port. A
schematic for a simple control is shown in Figure 9.
end of the drift sequence. This will cause you to have

two bright images of your target star in your exposure
with a thread of light between recording the target
stars flux during the drift interval.
There is a Drift Test button that allows you to
simulate the event instantly. This allows you to frame
your target. Start your exposure and click on Drift
Test. The program will command your telescope
through the same sequence of commands that will be
sent during the event. It allows you to inspect your
result and if necessary adjust the position of the target
star in the frame or modify your drift rate or plate
scale to optimally capture the event.
When your occultation event time nears, click
the Start button. The status line will count down
the time until the start of the exposure. Starting 6
seconds before the exposure, the computer will start
beeping at one pulse per second to prepare you to
start the exposure. As the program counts to 0, start
your exposure. From here on, all the critical actions
will be triggered under computer control.
6. Precise Computer Time

The key to success with drift scan method is having the computer clock maintain as precise a time as
possible. If you computer has a broadband Internet
connection, keeping precise time is easy. There are a
number of excellent programs that will synchronize
your computers clock with the National Institute of
Standards and Technologys atomic clock. My personal favorite is Dimension 4. Once loaded, it runs in
the background, waking up at programmable intervals, pinging the Atomic clock and updating your
PCs clock.
Figure 9: Guider Input Interface
Finally, you can modify the circuit in Figure 9 to

activate a relay interrupting power to your telescope.
If you are using the guider interface or a relay,
you need to check the relay box in the program to
change the way the serial port is commanded. Otherwise, uncheck the relay box if you are serially commanding your telescope.
To use the Drift Scan Trigger program, synchronize your computers time with a standard time signal
(see section 6). Next determine how long you want to
drift. Your drift time should be longer than the expected occultation duration plus 2-sigma of the expected time and an additional margin for trigger.
Configure the start time to be at least 15 second
prior to the start of the drift timing, and be sure your
total exposure time leaves some additional time at the
Figure 10. Dimension 4 Screen Shot
195
One of its excellent features is keeping a running

history of its corrections so you can get an estimate
of the drift rate and the statistical accuracy you may
expect from your observations.
Figure 11. Time Correction History.
Figure 11 indicates that my machine tends to

stay within 10 msecs of standard time as long as I
maintain an Internet connection. Left on its own, it
drifts at a rate of 180 msec in 3 hours, or 60 msec per
hour.
This tool is even useful in the field. As long as
your observing location has 3G or 4G cell phone service, you can tether your phone to your computer, run
Dimension 4 and have time accurate to 10 msec.
shortly before observing and have characterized the

clock drift with Dimension 4 previously, you can still
get pretty accurate results.
NEMA GPS signals are the final option. An inexpensive USB GPS module runs about $25 from
various Internet vendors. It looks to the computer just
like a serial port connected to a GPS receiver.
Several years ago the author developed a utility
to capture locations information and synchronize
computer clocks from GPS serial data streams.
One of the well known problems with
GPS/NEAM data streams is that they show a delay
between the correct time and the output of the time
string that is device dependent. However it is the authors experience that while these delays differ from
device to device. For a given device the delays are
relatively constant. For that reason, the GPS Geodesy
program includes a Delay parameter field. It contains a static correction to be applied to the GPS time
standard prior to its being applied to the PC clock.
Figure 13. GPS Geodesy Screen Shot
Figure 12. USB GPS Receiver
In the event you do not have network connectivity, there are two other methods that are very reliable. First is WWV and WWVB, the NIST radio stations that transmit time signals on 2.5MHz, 5MHz,
10 MHz and 15MHz. When using these sources, I
turn on only the computer and radio and keep them as
far apart as I can and still hear the radio. My experience is that these signals can be difficult to use near
computers, telescopes and cameras that all emit radio
interference. But if you use WWV to sync your clock
196
Figure 14. GPS-NEMA Jitter
Figure 14 shows the results of 60 trials of applying GPS serial time corrections verses Dimension 4
network time standards. It demonstrated that GPS
serial data streams provide accurate timing to 50
milliseconds, 1/20th of a second accuracy from virtually any location.
To characterize a new device, simply connect
yourself to the Internet and put the GPS where it can
get enough satellites to get a fix. Often you can obtain a fix in a wood frame structure. Then bring up
both Dimension 4 and GPS Geodesy on your system.
Set the GPS delay to 0 and alternate between synchronizing to network time and GPS time. About 50
or so tries will get you a statistically useful sample
population. Now go to Dimension 4s history window
and export the history file as a CSV file and use your
favorite spreadsheet to analyze the results. Take the
simple mean and plug it into the GPS Geodesy programs delay parameter and you are ready to go into
the field.
7. Data Reduction
The ultimate product of your observation will be
the duration and start time of the occultation at your
observing location, or lack thereof. Data reduction is
the process of extracting that information from your
drift scan image.
The steps in data reduction of a drift scan CCD
image start out as in any other CCD image, by calibrating for bias, dark current and potentially flat
fielding the optics.
To facilitate the easy extraction of the event data,
the next step is to rotate the image such that the star
trails are precisely horizontal. The tool used to do this
should be flux conserving, so as not to corrupt the
photometric information in your image.
Next you want to extract the brightness history
versus position in your image. Most astronomical
image processing tools provide a method for this
process.
Figure 15. Virtual GPS
The final time tool, developed by the author, is

titled virtual WWV Clone. It uses the PCs clock and
the audio system in a computer, to create a fake
WWV audio stream. It assumes the computer has
been synchronized with some time standard. The PC
itself has some delay inherent in the multimedia system, so the best way to calibrate a computer is to set
the computer time via your preferred method and
then tune WWV in on a good radio. You can then
adjust your correction factor so that your computers
time signals sync with WWVs. The human sense of
rhythm is accurate to better than 10 milliseconds. So,
when it sounds good, it is good.
With a PC and a WWV clone, you can make
drift scan observations by simply watching the PCs
clock. Open the shutter on your camera and pull the
plug on the right WWV time tick.
Regardless of the drift scan exposure method
you employ, the duration of the occultation will be
measured very precisely. The primary error will be
the time displacement of the observation related to
the jitter in the drift triggering mechanism.
Figure 16. Line Profile Extraction
Figure 16 shows how to accomplish this in

Maxim/DL. Using the cursor, box the target stars
trail including both the start and end points. From
within the View menu, select Line Profile.
Finally, within the line profile dialog box, select
Horizontal box.
At this point you have a graph of the sum of the
columns within the box plotted against the row
pixels. To maximize signal to noise ratios you want
to box the star trail while encapsulating as few of the
background pixels as possible.
197
Cyanogen Systems. Maxim/DL Image Reduction and

Analysis Software
http://www.cyanogen.com/
Google. Google Earth Download
http://www.google.com/earth/index.html
Meade Instrument. Planetarium
Processing: AutostarSuite 5.5
http://meade.com/software-manuals
and
Image
NOAO. IRAF- Imaage Reduction and Analysis

Facility
http://iraf.noao.edu/
http://acs.pha.jhu.edu/~shy/x-iraf-windows/
Figure 17. Scanalyzer Screen Shot
Similar graphs can be extracted by IRAF and

IRIS. Both are free image analysis software.
At this point there a couple of different ways to
go. Easiest in Maxim/DL is to move the over the
graph, reading out values. Locate the center of the
star before and after the drift. The drfit time divided
by the difference in pixels between the two star
images is your SecondsPerPixel coefficient. Now
measure the pixels between start of drift and the
midpoint of the occultation ingress. Next measure the
duration of the occultation in pixels. Multiplying both
values by the SecondsPerPixel coefficient to get the
timing of the events.
Alternatively, you can export the data to a
.CSV file and perform similar computations in
your favorite spreadsheet.
Finally, John Broughton has developed an easy
interactive drift scan analysis tool called Scanalyzer
that he distributes for free from his website. It takes
exported profile drift scan text or .CSV files as input
and provides an interactive means to calculate the
occultation timing and the error bars associated with
the observation. It also has advanced features
including time corrections for non linear plate scale
distortions.
8. Tool Links
John Broughton. Drift scan tools: Scanalyzer &
ScanTracker
http://www.asteroidoccultation.com/observations/
DriftScan/Index.htm
Christian Buil. Iris CCD and DSLR image processing
and analysis software
http://www.astrosurf.com/buil/us/iris/iris.htm
198
Steve Preston. Worldwide Occultation Prediction Site

http://www.asteroidoccultation.com/
Space Telescope Science Intitite. STSCI Data
Analysis System - IRAF Extensions
http://www.stsci.edu/institute/software_hardware/stsdas
SSC. Occultation Tools: GPS Geodesy, Asteroid Occultation Maps, Drift Scan Trigger
http://www.ssccorp.com/Observatory
Thinking Man's Software. Dimension 4 - Network
Time Synchronizer
http://www.thinkman.com/dimension4/
Do not forget to report your observations. There
is strength in your numbers, but only if you share
them. For full instructions on reporting your observations see:
http://www.asteroidoccultation.com/observations/
9. Conclusion
The odds are that if you are reading this article,
you have all the necessary equipment and expertise to
observe an occultation. Give it a try. You may not
always see the occultation, but that is where the asteroid isnt and that fact is worth reporting. And you
may find that measuring an asteroids cross section is
exciting. You and a handful of collaborates can solve
the cross section of an asteroid with observations that
span a handful of seconds. Shape modelers spend a
lot more time for their results!
10. Acknowledgements
The author would like to thank the Mount Wilson Institute for the use of its facilities while developing and testing the tools referenced in this paper.
11. References
Brown, M. E., Rajuillo, C. A. (2004). Direct Measurementsof the Sice of the Large Kuiper BeltObjecct
(50000) Quaoar. Astron. J. 127, 2413-2417
Buie, M. W., Millis, R. I., et al. (1993). CCD Camera Occultation System. Bull. Am. Astron. Soc. 25,
1115.
Chen, W. P., et al. (2003). Fast CCD Photometryin
the Taiwan American Occultation Survey. Baltic
Astrononomy 12, 568-573.
Cooray, A. (2003). Kuiper Belt Object Sizes from
Occultation Observations. Astrophys. J. 589, L97L100.
Dunham, E. W., Baron, R. I, et al. (1985). A High
Speed Dual-CCD Imaging Photometer. PASP 97,
1196-1204.
Fors, O, Nunez, J. Richichi, A. (2001). CCD Driftscan Imaging Lunar Occultations. Astron. Astrophys. 378, 1100-1106.
Massey, P. (1997). A Users Guide toCCD Reductions with IRAF. IRAF Documentation.
Millis, P., Dunham, D. W. (1989). Precise Measurement of Asteroid Sizes and shapes from Occultations. in Asteroids II (Shipley et al., eds.). pp. 148170. University of Arizona Press. Tucson.
Nuggent, R. (ed.) (2007). Chasing The Shadow: The
IOTA Occultation Observation Guide. International
Occultation Timing Association.
Olkin, C. B., et al. (1996). Astronomy of Singlecord Occultations: Application to the 1993 Triton
Event. PASP 108, 202-210.
199
200
Papers without Presentation and Posters
201
202
VARMiNT - Beaver
A Virtual Astronomical Research Machine in No Time

(VARMiNT)
John Beaver
University of Wisconsin Fox Valley
1478 Midway Rd.
Menasha, WI 54952
john.beaver@uwc.edu
Abstract
We present early results of using virtual machine software to help make astronomical research computing accessible to a wider range of individuals. Our Virtual Astronomical Research Machine in No Time (VARMiNT) is
an Ubuntu Linux virtual machine with free, open-source software already installed and configured (and in many
cases documented). The purpose of VARMiNT is to provide a ready-to-go astronomical research computing environment that can be freely shared between researchers, or between amateur and professional, teacher and
student, etc., and to circumvent the often-difficult task of configuring a suitable computing environment from
scratch. Thus we hope that VARMiNT will make it easier for individuals to engage in research computing even if
they have no ready access to the facilities of a research institution. We describe our current version of VARMiNT
and some of the ways it is being used at the University of Wisconsin Fox Valley, a two-year teaching campus
of the University of Wisconsin System, as a means to enhance student independent study research projects and
to facilitate collaborations with researchers at other locations. We also outline some future plans and prospects.
1. Introduction
Astronomical research is very much tied to computing. For a professor or graduate student at a research institution, there is an existing shared computing infrastructure, with networked computers already
configured for a variety of research tasks, and a local
community of support to assist beginners. But for
amateur astronomers or faculty and students at small
colleges, the lack of an existing research computing
environment can be a serious obstacle to conducting
astronomical research.
The actual physical computing power necessary
for many research tasks is now easily available; even
a low-cost netbook may have enough power for some
data reduction and analysis tasks. The difficult part is
the initial setup and configuration of the software and
necessary dependencies, along with the tutorials and
documentation necessary for those beginning in research computing. The lack of a common infrastructure also makes it difficult for people to help each
other.
Toward addressing this need, we have developed
our Virtual Astronomical Research Machine in No
Time (VARMiNT). The purpose of VARMiNT is to
provide a ready-to-go research computing environment, mostly for astronomy, that can be freely shared
between researchers, or between professional and
amateur, teacher and student, etc.
VARMiNT is a fully-configured Ubuntu Linux
virtual machine (VM) that can be booted and run
from any computer that has VMware Player installed.

VMware Player is proprietary software from
VMware (www.vmware.com), but it is available for
free (with registration) for MS Windows and Linux.
We have attempted to create VARMiNT with an operating system (OS) and installed software that are
free and open-source, mostly under the Gnu Public
License (Kuhn et al., 2008).
Since individual copies of VARMiNT act as independent computers, they can be customized for
particular uses. Thus a particular copy of VARMiNT
may be modified for a particular combination of project and user, with appropriate documentation and
data installed for that particular project, customized
to fit with the experience of the intended user. In this
way a given VM can spawn clones that evolve over
time according to local needs.
We hope that VARMiNT can be a useful tool for
extending astronomical research computing to individuals who currently lack the resources to set up
such an environment themselves. We also hope
VARMiNT can facilitate (and thus encourage) new
collaborations that bring research in astronomy to
individuals not affiliated with research institutions.
Here we describe some of the properties and possibilities of sharable virtual machines and our VARMiNT in particular.
203
VARMiNT - Beaver
2. Virtualization Software
VARMiNT is set up as a VMware Virtual Machine. It can be run directly with several VMware
products, including the free (with registration)
VMware Player, which is available for most MS
Windows and Linux host operating systems.
VMware was chosen because their virtual machines are portable and can be self contained within
one directory. Thus VARMiNT can be shared simply
by copying it to, for example, an external USB hard
drive. The copy can then be moved freely back and
forth between host computers, even across platforms.
Simply shut VARMiNT down on one host, disconnect the drive, reconnect it to another host, and boot
it back up.
The other popular choice for virtualization software is VirtualBox (www.virtualbox.org). VirtualBox
has the advantage that it is open-source GPL software, and it is available for Mac OSX as well as MS
Windows and Linux. VirtualBox virtual machines,
however, are not truly portable; they must be created
new on each host computer. It is fairly easy however,
to create a VirtualBox virtual machine from a
VMware VM. One can simply create a new VirtualBox VM, but instead of creating a new virtual disk,
simply use the disk from a VMware VM. At present,
this is the only way we know to run VARMiNT on a
Mac OSX host.
3. VARMiNT 1.0 Operating System and

Installed Software
VARMiNT 1.0 is a VMware VM, designed to
run directly on VMware Player or VMware Workstation. The OS is Ubuntu 11.04, 32 bit, and it can be
run on both 32 and 64-bit host machines. VARMiNT
is configured with a single 40-GB capacity virtual
disk that is also the boot disk (additional virtual disks
can easily be added). The entire VM is contained
within one folder and sub-folders. The current size of
the virtual disk is about 12 GB.
Since VMware support of graphics acceleration
is uneven, the VARMiNT's desktop manager (Gnome
Classic) is configured to run without effects. It can be
reconfigured to run with effects if your virtualization
software can handle 3D graphics acceleration. We
have had some success in running VARMiNT with
3D acceleration under VirtualBox and with the most
recent version of VMware Player.
Many useful computing tools and software packages are already installed and configured in VARMiNT, including the basic utilities that would be present on any Linux system. The list below gives a
204
brief description of some of the most important packages.

The Ubuntu Software Center, which gives easy
categorized access to a central repository of free
software. This combined with the Synaptic
Package Manager allows the user to easily (and
safely) install software and manage dependencies.
The LibreOffice suite (an open-source fork of
Open Office)
A full Latex environment, including the excellent
Texmaker source code editor, and the Jabref bibliography manager
GIMP, an open-source alternative to Photoshop.
The astronomical image viewers Aladin, DS9
and ImageJ. All three of these include some basic image processing capabilities.
IRAF, v2.15.1a, 32-bit, with the STSDAS and
TABLES external packages. This is the venerable NOAO Interactive Reduction and Analysis
Facility for all types of astronomical image processing.
STScI Pyraf and Pyfits. These allow one to integrate Python programming with IRAF or to access FITS data directly from Python.
STScI Specview, a free utility for display and
simple analysis of spectra.
GDL, the Gnu Data Language, a free and opensource alternative to IDL.
CLOUDY, the photoionization modeling code of
Gary Ferland
A Python/Pylab/Matplotlib programming environment, with the Ipython interface. The R programming language is also included as an alternative.
Scilab and Octave, open-source (and mostly
compatible) clones of Matlab; and WxMaxima,
an open-source alternative to Mathematica.
The popular open-source graphics utility
Gnuplot, as well as SciDavis, a useful data management facility combined with an objectoriented plotting package.
VARMiNT - Beaver
4. Modes of Use
One reason for using a virtual machine is to use a
different operating system from the one that is
natively booted on your host computer. For example,
one might want to use a Linux virtual machine in
order to gain access to Linux programs on a MS
Windows computer.
But even if the virtual machine uses the same OS
as the host computer, there still may be advantages to
the VM approach. In particular, a virtual machine can
be kept lean as it can be configured around a particular subset of possible uses. The VM need not perform all possible computing tasks, and so it can be
kept small (and thus easily backed up), uncluttered,
and customized for the project at hand.
Thus one may find it useful to do most of one's
work on a choice of a few different customized virtual machines, each booted when needed on a host
computer that can also be kept lean and uncluttered.
If the host computer has enough resources, both
VMWare and VirtualBox allow the user to have multiple VM's booted simultaneously on the same host.
Another possible advantage, with VMWare
VM's in particular, is portability. Thus one can, for
example, easily move a project, contained within a
VM stored on a usb hard drive, back and forth between computers at home and work. This portability
also makes it easy to share projects between users.
5. License Issues
We have made every attempt to include only
GPL, or GPL-compatible, open-source software in
VARMiNT. The Ubuntu OS was configured without
the installation of proprietary hardware drivers. This
is possible because the virtualization software (e.g.,
VMWare or Virtual Box) used to boot the VM creates its own virtual hardware drivers. A new user of
VARMiNT has the option to install free software
(called Guest Additions for Virtual Box or
Vmware Tools for VMWare) to better integrate the
guest VM and the host OS. These proprietary (but
free) add-ons then allow for such features as copy
and paste or click and drag between the guest VM
and the host OS.
Information on the licenses for installed software
on VARMiNT is contained within a text file, prominently linked by a launcher on the initial desktop.
Since these are all GPL, or GPL-like, license agreements, then this should satisfy the license agreements
when VARMiNT is copied and shared between users.
For a particular use, one may want to install
software that while free, requires the active acceptance of a license agreement in order for it to be in-
stalled (we believe there is no such software on the

current version of VARMiNT). In that case, copying
and sharing the VM may violate that agreement, as
the new user has not actively accepted the license
agreement.
But even with this restriction, the VM approach
still provides advantages for potential users to share
this software (perhaps it is a necessary piece of software for a desired research collaboration), even if it
cannot be installed on the VM before sharing. The
usual method for sharing such software between
potential users is to provide a set of detailed instructions explaining step-by-step how to install the software. The problem is that the instructions often fail in
practice, because of different hardware and software
configurations, different versions of software appearing on a given download server, etc.
In the case of a VM, the installation instructions
can be included, and they can be specific to the initial
state of that particular VM. And so the instructions
can be tested beforehand to be sure they work on that
particular shared VM. In fact the entire installation
process can be tested, and then the software uninstalled before sharing. Since the active acceptance of
a license agreement usually occurs at the installation
stage, while there are often no restrictions on
downloading the necessary installer files, this can be
done ahead of time to be sure the tested version is the
actual version installed by the new user.
In principal, the entire installation process could
be scripted. And so when a new user of a recently
copied and shared VM first boots the machine, a
script runs that carries out the installation process
from previously downloaded installer files, prompting the new user to accept the license agreements as
they appear. A second script could even be included
to undo this process and reset the VM to its initial
state for further sharing.
6. Examples of Uses of VARMiNT

VARMiNT and its predecessors have been used
heavily at the University of Wisconsin Fox Valley
(UWFV), a two-year campus of the University of
Wisconsin system, with the primary mission of transfer of students to one of the four-year campuses.
UWFV has no institutional research computing
infrastructure, let alone one specifically geared toward astronomy. Faculty office computers are not
suitable for research, in large part because the individual researcher has no control over the installation
of software (let alone the OS) on their computer. And
so VARMiNT was developed by the author partly to
address his own research computing needs, and re-
205
VARMiNT - Beaver
cent work by the author (Beaver and Conger, 2012;

Beaver et al., 2012; Beaver and Robert, 2011) has
been carried out almost entirely on VMs similar to
VARMiNT.
A project currently under development to perform spectrophotometry of point sources at extremely
low cost (Beaver and Conger, 2012) uses VARMiNT
heavily, as this allows diverse users to carry out the
rather complex data reduction and calibration procedures within a commonly shared computing environment. Since the point of this project is to bring
stellar spectrophotometry to those who would otherwise have no access to the technique, VARMiNT is a
necessary part of the project.
Here are a few other scenarios that come to
mind:
A researcher establishes collaborations with individuals who lack the institutional resources to
set up their own research computing environment, simply by sending USB hard drives containing VMs back and forth through the mail.
A graduate student, as part of her own dissertation research, creates and maintains a VM environment and then trains a couple of local amateur astronomers to use it in order to help with
some of the labor-intensive aspects of her research. The astronomy department thereby
strengthens relationships with the amateur astronomers, the amateurs get a unique experience
and their names included on published work, and
the graduate student gains valuable experience as
a teacher and research team leader. Several years
later another graduate student does the same, but
this time the now-experienced amateurs help
train the graduate student.
An instructor at a small two-year college designs
VMs specifically for student independent study
projects in astronomy. The computing environment is custom made for the particular project
and configured to a point appropriate to the level
of the students and the fact that they will only be
around for two years at most. The student keeps
the machine on a USB hard drive connected to
their home computer, but disconnects it and
brings it to the instructor's office when needing
help. A copy of the machine is later modified by
the instructor so it will be appropriate for a different student project.
7. Future Prospects
VARMiNT 1.0 is designed to provide a rather
generic environment for astronomical research, in
206
particular for the reduction of astronomical image

data and spectra. We are currently customizing versions of VARMiNT specifically for our ongoing projects in Stromgren CCD Photometry and extremely
low-cost point-source spectrophotometry. In particular, we hope to make these VMs more user friendly
for amateur astronomers or for beginning students
through further work on the documentation and tutorials, and by writing appropriate scripts.
We hope to see VARMiNT and similar VMs foster collaborations that perhaps would not have taken
place otherwise due to the stumbling block of sharing
computing facilities with individuals not associated
with a research institution. One could envision the
construction of a central repository of VM's geared
toward different types of astronomical computing, in
order to facilitate such access.
Much further work would be needed in order to
realize the widespread adoption of VMs for research
collaborations, particularly to ensure compatibility
with existing institutional research programs while
still maintaining free and open-source VMs that can
be shared without running afoul of license agreements. As an example, many institutions rely heavily
on the proprietary IDL. VARMiNT 1.0 is equipped
with GDL, an open-source alternative that, although
incomplete, is mostly compatible with IDL. This incompleteness means much work would need to be
done in order to get the GDL environment on VARMiNT ready for a particular project of active research, as many common publicly-available procedures would likely need to be modified somewhat.
8. Conclusions
Virtual Machines provide an elegant solution to
the problem of bringing astronomical research to a
greater diversity of individuals. Our VARMiNT can
provide a ready starting point for those who wish to
set up an astronomical research computing environment that is customized for a particular combination
of project and user. For inquiries about how to obtain
a copy of VARMiNT, or for suggestions for future
versions or desires to collaborate, please contact the
author.
9. Acknowledgements
Dale Thibideau, Network Administrator at the
University of Wisconsin Fox Valley, first introduced
me to virtualization software, and his continuing help
and insight have been invaluable. Chuck Conger,
(UW Fox Valley) and Dan Phiel (UW Oshkosh) have
both tested early versions of VARMiNT and offered
helpful suggestions.
VARMiNT - Beaver
This work was supported in part by the following

grants:
UW Fox Valley Foundation Grant (2010)
UW Fox Valley Professional Development
Grants (2010, 2012)
UW Colleges Department of Computer Science,
Engineering and Physics Professional Development Grants (2010, 2012)
NSF ROA Supplement, AST 1135760 (2011)

VMware, VMware Player and VMware Workstation
are registered trademarks of VMware, Inc.
10. References
Beaver, J. E., Conger, C. (2012). Extremely Low
Cost Point-Source Spectrophotometry (ELCPSS). in
Proceedings for the 31st Annual Symposium on Telescope Science (Warner et al., eds.). pp. 113-120. Society for Astronomical Sciences, Rancho Cucamonga, CA.
Beaver, J. E., Piehl, D., Kaltcheva, N., Briley, M.
(202). Strmgren-H photometry of the galactic cluster M 11. In preparation.
Beaver, J. E, Robert, D. M. (2011) A CCD Spectrograph for One Dollar. in Earth and Space Science:
Making Connections in Education and Public Outreach. ASP Conference Series 433, 425.
Kuhn, B. M., Williamson, A, Sandler, K. M. (2008).
A Practical Guide to GPL Compliance. Software
Freedom Law Center.
www.softwarefreedom.org (retrieved 12 April 2012).
207
VARMiNT - Beaver
208
The Confusing Case of 16666 Liroma Buchheim et al.
The Confusing Case of 16666 Liroma

Robert K. Buchheim
Altimira Observatory (G76)
Coto de Caza, CA 92679 USA
Bob@RKBuchheim.org
John Ruthroff
H60
Carmel, IN USA
John@theastroimager.com
Abstract
In late 2011, asteroid 16666 Liroma made a very favorable apparition with unusually bright magnitude and passing through near-zero solar phase angle. This opportunity prompted a project to determine the rotation period,
lightcurve shape, and phase curve parameters (H, G) of this asteroid. The rotation period P = 112 5 h that we
find, with amplitude m 0.45 0.05 mag (peak-to-peak) seems to be well-justified. The lightcurve gives a fair fit
to almost all of the data gathered on 21 nights over an period of 2 months. However, three nights at the largest
phase angle (at the end of the observed apparition) fall significantly off the lightcurve. An attempt to fit the lightcurve to a two-period model gives a somewhat better fit, but still leaves substantial deviations from a model lightcurve, thus presenting an unresolved mystery.
1. Introduction
The 2011-2012 apparition of 16666 Liroma presented an unusually favorable set of circumstances: a
close approach to Earth (making the asteroid bright
enough for photometric study with small telescopes),
and a presentation at nearly zero solar phase angle
(offering the prospect of monitoring the asteroids
opposition surge, and determining its phase curve).
The authors independently began observations of this
object and then collaborated by merging our data sets
and jointly analyzing (and struggling over) the results.
2. Data, Reduction, and Analysis

Differential photometry was collected on different nights at our two observatories, using similar
equipment (both telescopes were 0.3m SchmidtCassegrain telescopes operating at f/6, with CCD
imagers). RKB used multiple filters (typically either
V-R or R-C, depending on the night), while JR used
only V-band. Images were reduced in the standard
way with flat, dark, and bias frames, and censored
to eliminate any images that were adversely impacted
by cosmic rays, aircraft lights, bad tracking, etc. Differential photometry was done with MPO Canopus.
Because it turned out that we had a hard time
finding a period with a plausible lightcurve, one of us
(RKB) undertook to calibrate the magnitudes of all
comparison stars from all nights onto a local R-band
system; the idea being that this would facilitate

bridging the asteroid magnitude across widelyseparated fields of view. All comp stars were selected
to have similar color indices of about [V-R] 0.45
0.1, using the MPO Comp Star Selector; so there is
little color difference between the comp stars and the
asteroid. It seemed simplest to leave the comp-star
data on the local Altimira observatory R-band
photometric system to avoid any errors or offsets
introduced by transformation to the standard system.
We note that this is quite close to the standard system
(the color terms in Altimira Observatory transforms
are small).
Three clear and stable nights were devoted to
calibrating the comparison stars, by imaging one or
two Landolt fields and the comp-stars. The Landolt
fields were imaged both near culmination and lower
in the west to determine the atmospheric extinction in
V-, R- and C-bands. The observed instrumental magnitude of each comp star was extrapolated to its exoatmospheric value by:
R0= R kRX
C0= C kCX
where R and C are the observed instrumental magnitudes [i.e. 2.5log(ADU/t)], X is the air mass of the
observation, and kR, kC, are the atmospheric extinction parameters (in mag/air-mass).
The values of R0, C0 are, of course, sort of inconvenient (of order 6 to 10). In order to translate
them to more conventional values, one comp star
209
was selected as the anchor star, and all other comp

stars from all nights were assigned magnitudes anchored to this star, as:
interval of time plotted in Figure 1, the solar phase

angle changed from 2 deg through zero, to
+6 deg. Figure 1 is plotted as reduced-magnitude,
using an assumed G = 0.15.
Rcomp = Ranchor [R0,anchor - R0,comp]

and a similar equation for C band.
The zero-point of the C-band magnitudes was selected so that C R for solar-color stars (not surprising, since there is not much color term between C and
standard-R in the Altimira Observatory system).
JRs data was taken only in V-band. His fields
and comp stars were imaged in Altimira V- and Rbands on one of the calibration nights. This provided the comp star magnitudes on the Altimira R
system; and since the asteroid and the comps are
nearly solar-color, it is reasonable to assume that
mag = VastVcomp can be treated as being equal to
mag = RastRcomp. (A few nights with both V and Rband data were examined, to confirm that there is no
evidence of the asteroid changing color as it rotates).
With this approach, all of the asteroid sessions
were put onto a single, consistent local photometric
R-band system.
3. A Differential-Photometry Lightcurve
With all comp stars on the Altimira R system,
the asteroid differential magnitudes were placed on a
consistent baseline
The resulting raw lightcurve for 2011 late November and early December is shown in Figure 1.
This shows a sequence of alternating upward
and downward slopes that suggests a period of
about 4 days.
Figure 1: Raw lightcurve of 16666 Liroma (in terms of

reduced magnitude (using G = 0.15), for UT 2011 Nov
22 through 2011 Dec 03
The solar phase angle was minimum ( 0.97

deg) on 2011 Nov 25 (JD 2455890.5). Over the
210
Figure 2: First estimate of phased lightcurve of 16666

Liroma, based on 9 nights near opposition. The inset
shows the results of the MPO Canopus period-search
algorithm.
The MPO Canopus period-search routine was

used on these 9 sessions, and found a plausible fit at
P 110 h, as shown in Figure 2. Phasing the observations to this period results in a double-peaked lightcurve, although this set of nights did not capture the
maximum brightness of the lightcurve.
Figure 3: Phased lightcurve of 16666 Liroma, using

differential photometry from 16 nights over 1 month,
showing best-fit lightcurve period P = 111.43. Solar
phase angles range from = 1.96 to = +21 degrees.
Inset shows the results of the MPO Canopus periodsearch algorithm.
Adding data gathered from 7 additional nights

over the following three weeks resulted in only a
slight change in the estimated period, and the lightcurve shape can be reasonably fit by a 4th order Fou-
rier curve, as shown in Figure 3. Some of the photometric data is noisier than we would normally like,
because the asteroid was getting pretty faint as its
distance from Earth increased.
This lightcurve has no arbitrary delta-comps, is
a plausible double-peak shape, and gives a reasonably good fit to the data (RMS = 4.2, in units of 0.01
mag). The only poorly-justified parameter hidden
within it is that it uses the default phase-curve parameter of G = 0.15 to compensate for the changing
solar phase angle.
We also tried several numerical experiments by
applying more-or-less arbitrary delta-comp adjustments (up to 0.3 mag) to groups of nightly data. The
net result of these experiments was that the best-fit
lightcurve periods ranged from 109 to 115 h, so we
conclude that the best-estimate lightcurve period is P
= 112 5 h.
4. Phase Curve Determination

The lightcurve shape in Figure 3 (described by a
4th-order Fourier curve) was used with the data of
asteroid R-brightness vs. time to determine the asteroids phase curve. The method invented by Harris
(1989) and applied by Buchheim (2010) was used. In
this method, each nights data (asteroid magnitude vs.
time) is lined up with the appropriate portion of the
lightcurve shape, and the implied rotational brightness maximum for that rotational cycle is found by
extrapolating along the lightcurve shape. An example
of this procedure is illustrated in Figure 4.
Figure 5:
Liroma.
Phase curve (Rmax vs. ) derived for 16666
Taken at face value, this gives an unusually low

slope parameter G = 0.03 0.03.
The Harris et al. (1989) method to determine G
becomes iterative at this stage the lightcurve shape
is re-determined based on the found value of G,
and then the extrapolated maximum brightness for
each observation night is re-calculated. The iteration
continues until mutually-consistent lightcurve and
phase curve are achieved. As Harris et al. (1989)
point out, this process normally converges within just
one or two iterations.
As it turned out for this project, re-plotting the
lightcurve using this value of G = 0.03 did not make
a significant change in the derived period or the
shape of the lightcurve, but it did result in a very
slightly poorer RMS fit to the data.
5. Discordant Results Late in the

Apparition
Figure 4: Example of the implementation of the procedure from Harris, et al (1989) to extrapolate observed
data to determine the max brightness of the rotational
phase curve on the indicated rotation. The solar phase
curve is formed by plotting Rmax vs.
The values of Rmax found in this way are then

plotted against the solar phase angle at the time of
observation, to create the phase curve. The result is
shown in Figure 5.
Additional data was gathered in the hope of filling in the gaps in the rotational lightcurve, and extending the coverage of the phase curve to larger solar phase angles. Alas, rather than helping, this additional data is a source of confusion. Altimira Observatory gathered R- and C-band data on UT 2011 Dec
28, 2012 Jan 18, and 2012 Jan 19 (at solar phase angles of 22 to 30 degrees).
When this data is plotted on the previouslyfound lightcurve with P = 111.43 h, it falls badly off
the curve, and at least in the case of 2012 Jan 18 and
2012 Jan 19 no arbitrary delta-comp would bring
the data into alignment with the lightcurve. Worse, a
period search that includes this data results in a complete garble, with no plausible lightcurve found at
any period between 40 hr to 250 h.
211

2011-12-28
2012-01-18
2012-01-19
Figure 6: Lightcurve data, showing the nights late in

the apparition, at phase angles 22 to 30 degrees
that dont fit the lightcurve formed by the other nights
What is going on with these nights?

One might suspect that there is a problem with
the comp-star magnitudes (i.e. adjusting the deltacomp might bring these nights into line). There are
a couple of factors that argue against this suspicion.
First, nights 2012 Jan 18 and 2012 Jan 19 used the
same comp stars: adjusting them up or down together
is not going to improve the overall situation. In addition, the calibration of the comp stars for all three
nights was done in a single observing session, with
the comp star fields for 2012 Jan 18/19 and 2011 Dec
28 imaged only about 30 minutes apart. There were
no apparent changes in atmospheric conditions during this time interval, and the images of Landolt
standard field SA-95 taken both before and after the
images of the comp-star fields showed no anomalies.
Other comp stars that were imaged and calibrated on
that same night, within an hour of the images of these
problem fields, show no discrepancies. Finally, the
comp stars for 2012 Jan 18 and 19 were re-imaged on
the next night, and these images gave essentially
identical exo-atmospheric instrumental magnitudes.
cal'd on 2012-01-26
FOV night C1
C2
C3
C4
C5
2012-01-18 -7.625032 -6.586032 -7.195032 -6.923032
cal'd on 2012-01-27
FOV night C1
C2
C3
C4
C5
2012-01-18 -7.588731 -6.544731 -7.203731 -6.886231
delta 0.036301 0.041301 -0.008699 0.036801
Considering that these two independent calibrations yielded the same comp-star magnitudes (0.02
mag), it is hard to argue that they are discrepant. By
extension, the calibration of the comp stars for other
nights (done in the same way, with the same instrument) is given some additional credibility.
It is possible that the shape of the lightcurve
changes at large solar phase angles ( 22 deg to
30 deg). Shadowing effects can certainly cause significant changes in the shape of a lightcurve from
opposition to large solar phase angle. We attempted
to test this notion by examining alternate lightcurve
212
shapes, or slightly different periods, using only the

half-dozen nights at the end of the apparition (i.e.
large solar phase angles). This did not provide any
useful clues.
Could the asteroid have undergone some sort of
cometary activity? There is no evidence of any coma
or emission on the images. The asteroid appears as a
normal star-like point spread function. Plus, although
it is easy to imagine how a comet-like activity might
make the asteroid appear brighter than expected (e.g.
2011 Dec 28), it is more difficult to imagine that dust
or gas emission could make it appear fainter than
expected (e.g. 2012 Jan 19).
6. Tumbling?
Liriopes relatively long primary period (assuming P 112 h to be at least approximately correct)
makes it a good candidate to be a non-principal axis
rotator (i.e. a tumbler).
We are not qualified to assess whether these
anomalous data points might be related to complex
(rotation + precession) motion of the asteroid. However, as a first step to investigate this possibility, a
two-period search was conducted with the utility in
MPO Canopus. The results were not spectacularly
successful, but were intriguing. The best combination
of primary and secondary periods was:
P1 = 107.44 h (rotation period?)
P2 = 198.80 h (precession period?)
Figure 7 shows the brightness vs. time [H(0,1,1)]
predicted by this two-period model. In the top panel
the two-period brightness model is compared to the
simpler single-period (P1 = 107.44 h) model. This
illustrates the complexity of the resulting brightness
vs. time profile that is caused by introducing the second period, compared with a single-period model.
The bottom panel of Figure 7 compares the observed
asteroid brightness with the prediction of the twoperiod model. (Observed data points are the measured magnitude adjusted to zero-solar-phase angle,
assuming a slope parameter of G = 0.03). The early
portion of this shows that the two-period model is a
nice fit to the data a noticeably better fit than is
achieved using just the P1 curve. Late in the curve,
the distinction between the single-period and the twoperiod predictions is not as clear, but again the twoperiod model seems to fit the data nicely. The complexity of the two-period brightness curve might explain why it was so difficult to find a single-period
model that could fit all of the data, even near opposition.
t, hr (t0= JD 2455887.50)
-500
-400
-300
-200
-100
100
200
300
400
300
400
13
13.1
13.2
13.3
13.4
13.5
-500
two-period model is
better fit than P1 only
13.6
-400
13.7
f(P1,t)
13.8
Two-Period Model (P1, P2)
13.9
14
-300
-200
-100
100
200
13
13.1
13.2
13.3
13.4
13.5
13.6
data still falls

far from model
13.7
Two-Period Model (P1, P2)
13.8
observed
13.9
14
Figure 7: A two-period lightcurve model offers an intriguing (but not completely successful) explanation of the
complex brightness changes of 16666 Liroma
Unfortunately, around t = 0 (UT 2011 Dec 08),

the observed points lie significantly above the predicted curve. Worse, very late in the apparition (not
shown in Figure 6), the final two nights of data (UT
2012 Jan 18 and 19) are substantially fainter than the
prediction of a G = 0.03 phase curve. So, the twoperiod model doesnt completely resolve the confusing case of this asteroids lightcurve.
Phase Curves. in Proceedings for the 29th Annual

Symposium on Telescope Science (Warner et al.,
eds.) p. 101. Society for Astronomical Sciences, Rancho Cucamonga, CA.
Harris, A.W., et al. (1989). Phase Relations of High
Albedo Asteroids. The unusual opposition brightening of 44 Nysa and 64 Angelina. Icarus 81, 365.
7. Availability of Data
All of the differential photometry data from this
project has been uploaded to the ALCDEF database,
so it can be examined and used by anyone who is
interested in re-analyzing it, or combining it with
other data sets.
8. References
Buchheim, R. K. (2010). Methods and Lessons
Learned Determining the H-G Parameters of Asteroid
213
214
Light at Night Student Project Craine et al.
Student Project and Curriculum Based on Light at Night

Data Collection
Erin M. Craine
4681 N. Cerritos Dr. Tucson, AZ 85745
ecraine@dakotacom.net
Jennifer C. DeBenedetti
Catalina Foothills School District
9566 N. Marsh Wren Pl. Tucson, AZ 85742
jennifercraine@yahoo.com
Abstract
There is a growing movement in the educational field to promote science, technology, engineering and math
studies, stemming from a concern about waning understanding and interest among K-12 students in these topics. STEM Laboratory, Inc. (STEM) has developed a Sky Brightness Meter (SBM) that can be used with ease yet
produces complex information relating to light at night monitoring. STEM sees the SBM and its corresponding
data archive as a means to involve students in projects that relate to scientific method exploration, makes science more accessible, and encourages a life long appreciation and understanding of scientific endeavors. In this
paper we present an example of a project template that could be used by students studying effects of artificial
light on sky brightness. STEM has developed several outreach lessons aligned with the National Common Core
Curriculum, Systems Thinking concepts and local standards to be implemented in classrooms or independent
youth organizations.
1. Introduction
This paper is intended to serve as a general guide
for students contemplating conducting and writing
their own report using light at night data collected
with the STEM Sky Brightness Meter or from the
STEM Sky Brightness Data Archive (SBDA), c.f.
Craine et al. (2011). It provides a basic overview of
the methodology and procedures for collecting and
reducing data, and gives an example of the sort of
information and product students could potentially
develop.
Additionally, this paper provides some sample
classroom applications, relating to light and light at
night, that teachers could apply and adapt to their
grade level. It is by no means a complete representation of all potential projects, but it does give an idea
of the types of lessons possible at a variety of grade
and age levels.
2. Methodology
The first step in this type of research project is to
determine the site to be monitored. Such a selected
site should be of some scientific interest, for example, due to environmental and ecological reasons or
new development going in. The site can be a small
sampling to be monitored multiple times, can be a
comparison of multiple sites, or may be a large area

monitored once to identify lighting fluctuations
within that zone. It is up to the student to decide what
sort of site and monitoring schedule best suits their
own research goals.
Alternatively, the student may choose to mine
the STEM Sky Brightness Database Archive (SBDA)
to utilize and analyze already collected light at night
data. This provides a good option if the student does
not have the means to collect their own original data.
Once a site has been selected for monitoring, a
detailed route should be established prior to the
commencement of data collection. This ensures consistency if the site is to be measured multiple times,
minimizes confusion during the actual collection, and
maximizes the use at the other end by guaranteeing a
logical and thorough data set. Table I is an example
of part of a pre-planned, written route that we followed when collecting data around Salpointe High
School in Tucson, Arizona. For additional ease, a
Google map with the exact route highlighted may be
used.
The Sky Brightness Meter (SBM) is designed to
be mounted at either a fixed location or on a moving
vehicle, depending on the specific goals of the project. In this example, to collect data we mounted the
SBM on the roof of an automobile using a plethora of
215
electricians tape. The photocell unit should be

pointed at the zenith to give sky brightness overhead.
1
E. on Blacklidge to Campbell
S. on Campbell to Grant
W. on Grant to Park
N. on Park to Adelaide
E. on Adelaide to Campbell
S. on Campbell to Spring
W. on Spring to Fremont
N. on Fremont to Mitchell
W. on Mitchell to Via Encantadora
10
N. on Via Encantadora to Glenn
11
E. on Glenn to Martin
12
S. on Martin to Spring
13
W. on Spring to Highland
14
N. on Highland to Copper
15
W. on Copper to Mountain
Table I. A portion of the Salpointe High School planned

route.
To commence collection it is simply a matter of

powering on the unit and confirming that satellite
locks have been achieved to trigger GPS signals to
the unit. Data acquisition and logging are completely
automatic, and it is only necessary to position the unit
on the vehicle. Ideally, data should be collected after
local astronomical twilight and should be completed
before moonrise. For consistency, it is desirable to
collect data on clear, cloudless nights. At the end of
the data collection session the data must be
downloaded as a comma delimited text file for import
into MS Excel for data manipulation.
3. Data
The original document produced from the data
collection session is a comma delimited text file
(.csv) of the raw data information. This file should be
saved in its original state; any subsequent data manipulations should be done in a new Excel (.xls or
.xlsx) file. The information downloaded from the
SBM in which we are primarily interested includes
the date, time, longitude, latitude, and magnitude per
square arc second for each data point.
In order to extrapolate more useful information
from the raw data it is necessary to conduct some
data manipulation. In the new Excel working document added information includes the longitude and
216
latitude in decimal form, and elapsed time for the

entire data collection session.
From this additional data we can now create a
variety of graphs that serve as useful visual representations for analysis of that nights data collection session. Plotting the decimal longitude and latitude of
each data point gives us a visual image of our exact
route for each session (see Figure 1).
Figure 1. Ground track of the Salpointe High School

route
With this graph we have multiple options for

data comparisons and interpretations. For example,
we could overlay this route with another of the same
location on a different night to compare exactly
where on the route data were collected. We could
isolate certain areas of interest, and look at data from
that specific leg of the route.
4. Comparison and Analysis

There is a wide range of possibilities for papers
and topics derived from data collected with the sky
brightness meter. Monitoring can be done to look at
sky brightness for seasonal cycles or as a function of
weather and air pollution. Observations can be made
over a lengthy period of time, to monitor long term
changes in light output as a function of population or
area growth or to determine efficacy of lighting ordinances. Key sites, due to their ecological importance,
can be observed to analyze lighting effects on environment and animal behavior. These are just a few
examples of potential studies and papers that can be
pursued by students.
Here we present one example using data collected around Salpointe High School in Tucson, AZ.
We compare the sky brightness levels of the area
collected over a period of two, non-sequential nights;
one night with the football stadium lights on and the
other night with those lights off. We export the data
extracted from the Excel file into the downloaded

EarthPlot program. In this program, we set our parameters for creating an isophotal map of the area.
We select contour as the plot type, change the
opacity to approximately 45 percent, and select the
number of contour levels to be plotted. We then export this .kml file to be mapped on Google Earth.
Looking at the different isophotal maps for each
night (Figures 2 and 3), we can visualize the brightness levels depending on location and environmental
factors.
magnitude of the change. In addition, we can see

changes in local traffic patterns indicative of increased activity near the stadium and reduced traffic
in some side streets.
5. Tools and Equipment

In order to collect light at night data, analyze the
findings and make a report, we used the following
materials:
STEM Lab Sky Brightness Meter (SBM)
Vehicle for mobile collection
Computer
Microsoft Excel software
Earth Plot software
GoogleEarth software
6. In The Classroom
Figure 2. A 7-Level isophotal map showing Salpointe

High School with stadium lights off
The previous segment of this paper described an

example project that could be pursued by an older
student on an independent study basis. The remainder
of the paper proposes similar projects based on the
SBDA and other materials put together by STEM, to
be undertaken by a range of grade levels under the
guidance of the teacher or by extracurricular instructors.
6.1 Education Standards

The goals in the K-12 curriculum are rooted in
the Scientific Inquiry Process and the history of science as a human endeavor. While there are a variety
of content specific goals for each grade level, varying
from state to state, these two concepts are the underlying directive.
6.1.1. Concept I
Figure 3. A 7-Level isophotal map showing Salpointe

High School with stadium lights on
In this instance, we can easily see the brightness

levels recorded on the night with the ball field lights
on are much higher than the night with the lights off.
This may perhaps be an obvious and expected occurrence but it does provide useful concrete data to support the assumption. Because there is a calibration
scale of brightness, it is possible to calculate the
The Scientific Inquiry Process includes the following performance objectives:

Observations, questioning, and hypothesizing
Investigating and modeling
Analysis and conclusion
Communication
217
6.1.2. Concept II
History of Science as a Human Endeavor focuses
on change over time, advancements individuals have
made in science and mathematics, and ultimately the
technology and engineering their discoveries enable.
6.2 Educational Applications

Projects and lessons at the elementary grade
school level will be less broad in scope, and focus
more on specific and directed tasks and skill sets.
Teachers will be provided with targeted sky brightness data in an Excel spreadsheet, containing information on date, time, and brightness levels. Students
will be asked to create a bar graph in the excel program, charting brightness levels according to data
and/or time.
If the computer technology or time for instruction is not available, students may create the bar
graph on standard graph paper or poster sized paper
to deliver as a possible presentation. Students should
be able to make observations and compare the data
and bar graphs (using their previously learned understanding of light behavior) to make inferences or
draw conclusions about why the same area would
have such differences in night sky brightness.
These observations should be recorded in a data
table and the conclusions written in paragraph form,
under a self-generated topic sentence to be supported
by students individual observations and bar graph.
At the middle school level instructors will be
provided with data files for date, time, latitude, longitude, and sky brightness levels. Instructors will need
to download Google Earth and Earth Plot for students
to create a sky brightness isophotal map. Students
will use this to calculate the areas with the most and
least saturation. Students will then create a scatter
plot using a random sampling of data points from
these two areas. This can be created with an Excel
program, standard graph paper, or poster sized paper,
for a possible presentation.
Because this is a blend of teacher provided information and student analysis, this lesson would
serve as a beginning of the year project to commence
understanding of the inquiry process while linking to
some of the Earth, Solar System and Energy performance objectives.
At the high school level students or student
groups may isolate a variable to analyze sky brightness as a function of that variable. Students may collect their own data using the SBM (as described earlier in this paper), compare different data sets from
the SBDA, or use a combination of the two. Students
may question and hypothesize: sky brightness as a
218
function of weather, air pollution, seasonal cycles,

building/development codes, or protected lands. Students will independently gather data, model the data
in appropriate graph form for analysis, and draw conclusions based on their observations of the data. Students will then present their findings in the form of a
written report, and an oral classroom presentation.
Any grade level may examine a timeline of significant individuals advancing math, science and
technology. Students will observe we are approaching a new frontier in scientific endeavor in this field,
in which they can actively participate. They may be
the next Galileo or Newton in an upcoming age of
research and discovery; students may create a resume
of their scientific exploits and add their own biography to the timeline at the end of they year.
7. Conclusion
The Sky Brightness Meter and Sky Brightness
Data Archive are innovative technologies and tools
which encourage student participation and interest in
the field of science. The STEM Educational Outreach
Program recognizes the importance of engaging
learners of all ages and abilities. The SBDA brings
real scientific data directly to students, sparking an
enthusiasm at an early age. The SBM allows students
of higher levels to target and direct their own projects
and collect data accordingly.
The Methodology, Data and Comparison and
Analysis sections of this paper serve as an introductory guide to potential student-driven scientific inquiry projects. The In the Classroom section provides
classroom ready SBM and SBDA based projects
linked to science standards. These suggested projects
represent a portion of the opportunities STEM provides for students and educators.
8. References
Arizona Department of Education
http://www.azed.gov/standards-practices/commonstandards/
Common Core Curriculum Standards
http://www.corestandards.org/the-standards
Craine, Eric R., Craine, Erin M., Craine, Brian L.
(2011). The Sky Brightness Data Archive (SBDA)
in Procceedings for the 30th Annual Symposium on
Telescope Science (Warner et al., eds.). pp 45-52.
Society for Astronomical Sciences, Rancho Cucamonga, CA.
Using an iPad in Astronomical Education Gill et al.
Enhancing the Educational Astronomical Experience of

Non-Science Majors with the use of an iPad and Telescope
Robert M. Gill, Michael J. Burin
California State University, San Marcos, Physics Department
rgill@csusm.edu, mburin@csusm.edu
Abstract
General Education (GE) classes are designed to broaden the understanding of all college and university students in areas outside their major interest. However, most GE classes are lecture type and do not facilitate
hands on experimental or observational activities related to the specific subject matter. Utilizing several astronomy application programs (apps), currently available for the iPad and iPhone, in conjunction with a small inexpensive telescope allows students unique hands on experiences to explore and observe astronomical objects
and concepts independently outside of class. These activities enhance the students overall GE experience in a
unique way not possible prior to the development of this technology.
1. Introduction
Small telescopes have become reasonably priced
and readily available. The iPad has gained popularity
through the thousands of applications (apps) that
have been written for almost any subject imaginable.
We have taken advantage of an inexpensive telescope, the Celestron Firstscope (Figure 1) and several of the astronomy related apps available for the
iPad (Figure 2) to enhance college and university
students General Education (GE) experiences allowing them the unique opportunity to observe astronomical objects independently outside of the classroom with little or no background knowledge of the
operation of a telescope or the location of objects in
the sky.
Ten iPads and telescopes were purchased by the
Physics Department and are available for the students
to check out from Instructional Technology (IT) and
the library Reference Desk. Specific astronomy apps
have been loaded onto the iPads for our activities.
Since the iPads also allow access to the students
email and internet, IT personnel clean and reinstall all
apps on the iPad prior to each checkout to ensure the
integrity and security of the devices. The students can
check them out for three day periods.
It is assumed that the students have never used a

telescope or iPad previously. The activities begin
with a familiarization of the equipment. This is quite
detailed and takes them through the activity step by
step. Safety is always a concern; the students are reminded never to look at the Sun.
The telescope is set up in the daylight using a
terrestrial object like a distant tree, building, lamp
post, etc. This allows the student to become familiar
with locating objects, changing eyepieces and focusing.
2. The Activities
With the availability of iPads and inexpensive
telescopes it has become possible to develop selfguided activities. They can be used with the telescope
alone, with the iPad alone or with the telescope and
iPad together (the preferred situation) depending on
the availability of the equipment at checkout.
Figure 1. Celestron Firstscope telescope.
219
Figure 2. The iPhone and iPad.
The students also familiarize themselves with the

iPad prior to using it with the telescope. Playing
with the apps is encouraged.
The first activity is to observe the moon. The
telescope is set up and aimed at the moon. The iPad
app Moon Globe is opened and a detailed map of
the moon is available (Figure 3). The students can
move around the iPad screen and zoom in on the region of the moon that they are looking at in the telescope.
The students must locate several objects, e.g.,
mare, craters, etc. and record their observations. They
may make sketches or take a photo through the telescope eyepiece if they wish. Photography is explained in the familiarization.
The next activity uses a planetarium program,
GoSkyWatch (Figure 4) that displays the night sky.
However, the iPad utilizes the GPS and accelerometer orientation system to give you a correct view of
what you are looking at when you point the iPad in a
particular direction in the sky. The app is powerful in
that it includes many databases of information and
can be used for reference and determining what you
are looking at or looking for.
This is the most useful of the apps since one
need only point the iPad at a location in the sky and
that portion of the sky is displayed on the screen.
Stars, constellations and other objects are also
identified and labeled on the screen.
The image on the screen can be zoomed in or
out to assist in the identification of objects (Figure 5).
The iPad may also be used during the daytime. The
suns location and zodiacal constellation can be obtained being careful to block the sun with the iPad.
Other activities with this app include the identification of bright stars, finding your zodiacal constellation, finding and identifying planets and the identifi-
220
cation and location of reference lines like the celestial

equator, ecliptic and meridian.
Access to any of the apps databases yields additional information on other astronomical objects
(Figure 6). This concludes our GoSkyWatch app
activity. The activities allow additional explorations
for the student to discover other interesting aspects of
the night sky on their own. These explorations should
of course, be recorded and included in their reports.
The remaining activities are associated with what
we call specialty apps. Most include reference material which we felt could augment the hands on experience. The students must at least open each of
these apps and answer a few questions about them.
The Messier List app (Figure 7) contains a
complete list of the Messier M objects and can be
used in conjunction with the GoSkyWatch app.
Figure 3. Moon Globe app and moon map.
Figure 6. Moon data.
Figure 4. GoSky Watch app and sky map.
Figure 7. Messier List.
Figure 8. M object photo and description.
Figure 5. Zoomed-in screen.
The app includes technical information about

each M object and includes a photograph of the
221
object. This is very helpful in observing the object

with the telescope (Figure 8).
Are the planets visible? If Jupiter is in the night
sky, what moons are visible? It is convenient to know
whats up in the sky on the night of your observations. An excellent reference source for that information is the SkyWeek app (Figure 9), presented by
Sky & Telescope. It presents a detailed look at todays happenings which might include star maps and
photo images (Figure 10).
Figure 11. 3D Sun app.
Figure 9. Sky Week app.
Figure 10. Sky Week maps and photos.
3D Sun is a very useful app (Figure 11). It

digitally reconsructs images downloaded from
NASAs STEREO satelites to show details about
the curent status of the sun. Images of the sun in
different wavelengths are also presented (Figure 12)
and the students are asked to look at each one and
compare them. Evidence of sunspots and flare
activity can be observed. There is also a gallery of
images that highlight past solar activity. This allows
the students another insight into astronomical
methods and techniques (Figure 13).
222
Figure 12. The Sun in different wavelengths.
image appears every day with a detailed explanation.

Within the explanation are hyperlinks to terms and
other related information. All the old images are
archived and accessable as well. We not only get
pretty pictures but a good educational reference.
3. Results
Figure 13. Gallery of solar images.
Figure 14, APOD app image.
Last but not least, the APOD app (Figure 14)

Astronomy Picture of the Day was added for bit of
fun. This app is presented by NASA and displays
amazing images from all areas in astronomy. A new
The activities were completed by approximately

one-third of the astronomy class and were well received. Questionnaires were completed by each student attempting the activities (39 participated in these
activities) and have been analyzed in an attempt to
validate the project. Keep in mind that these students
are non-science majors and have little to no background in astronomy and observational methods or
techniques.
The following questions were asked of the students:
1.
How likely are you to use a small telescope

again if you have the opportunity? Why or
why not?
2.
If so, will you bring an iPad or iPhone along

(or similar device) if you have one? Why or
why not?
3.
There are many variants of astronomy apps out

there, and the list is growing.
4.
Please comment on any that you have used (if

any) that were not included in these observational exercises. Is there one that is similar to
the ones that we use here that you prefer? If
so, why? You may comment on any significant
differences and make recommendations (if you
have any).
79% of the students indicated that they would

definitely like to use a telescope again. 21% indicated
that they would not; however, 63% of those, noted
problems with locating objects with the small telescope and would have liked the use a larger one. The
Celestron Firstscope is very small and does not have
setting circles, an adequate finder scope or a go-to
system for positioning. 8% of the students said they
used another telescope (belonging to a relative or
friend). Interestingly, 8% indicated that they were
going to buy a telescope for future observations.
92% of the students indicated that they liked using the iPad with the activities. They enjoyed pointing it at the sky to locate objects. 8% indicated that it
was not helpful. It was found that a majority of those
who disliked the iPad activities had problems with
the telescope.
223
The students were asked if they would look at

other astronomical apps in the future if they had access to an iPad. 41% said yes and 41% said no they
wouldnt but didnt give a reason why. 18% didnt
respond on this question. It is possible that the response to this question may have resulted from the
students misunderstanding what was really asked.
6. References
4. Conclusion
3D Sun
Dr. Tony Phillips, LLC
http://itunes.apple.com/us/app/3dsun/id347089078?mt=8
These telescope and iPad activities were considered successful after reviewing the initial semester
trial data. The majority of the students liked the activities and found them useful. They particularly enjoyed pointing the iPad at the sky and identifying
what it was pointed at. Problems operating the small
telescope contributed to some students comments.
Minor modifications are planned for the activities instructions and the questionnaire to eliminate
possible confusion in the future. We may be able to
eliminate some of the problems the students had
with the telescope. Purchasing more iPads and telescopes would also increase the availability of the
equipment for more student participation.
It is often difficult to stimulate a GE students interest in subject matter that they may not have an
interest in. It was refreshing to see the non-science
major student response to the use of these technologies, especially those few who stated they were going
to buy their own telescopes.
We also realized that regular structured astronomy laboratory class exercises could benefit from the
integration of the iPad telescope applications.
The reasonable cost of the iPad apps makes it
easy to afford. Most of the apps used in the activities
were free. The remaining apps cost is reasonable,
between $5 and $40. Additional apps may be utilized
in the future.
Integrating astronomy apps into teaching activities should increase drastically in the further. To assist faculty with the many possibilities for student
activities, catalogs of astronomy apps have already
been published, Fraknoi.
A copy of the activities is available upon request
from the authors.
5. Acknowledgements
We would like to thank the Physics Department
for funding the purchase of the telescopes and the
iPads and Colin McDonell, Instructional Technology,
for his preparation and maintenance of the iPads and
their apps.
224
Franknoi, A. (2011). Astronomy Apps for Mobile

Devices, a First Catalog. Astronomy Education Review 10, 010302-1, 10.3847/AER2011036
6.1 Apps / Equipment
Apple iPad
http://store.apple.com/us/browse/home/shop_iPad/fa
mily/iPad?afid=p219|GOUS&cid=AOS-US-KWGBOPIS
Apple ITunes
www.apple.com/iTunes
Astronomy Picture of the Day
Concentric Sky
http://itunes.apple.com/us/app/astronomy-picture-ofthe-day/id304006512?mt=8
Celestron Firstlight Telescope
Celestron, LLC. , 2835 Columbia St. Torrance, CA
GoSkyWatch Planetarium,
GoSoftWorks
http://itunes.apple.com/us/app/goskywatchplanetarium-for/id364209241?mt=8
Messier List
Scot Spencer
http://itunes.apple.com/us/app/messierlist/id364899443?mt=8
Moon Globe
Midnight Martian
http://itunes.apple.com/us/app/moonglobe/id333180321?mt=8
S&T SkyWeek
Sky & Telescope Media, LLC
http://itunes.apple.com/us/app/s-tskyweek/id398252674?mt=8
Rotation Period of the Sun - Gill
The Rotational Period of the Sun Using the Doppler Shift

of the Hydrogen alpha Spectral Line
Robert M. Gill
California State University, San Marcos, Physics Department
Merritt Observatory
5034 Hill Ranch Dr. Fallbrook, CA 92028
rgill@csusm.edu
Abstract
The fact that the sun rotates is obvious by observing the daily motion of sunspots. The overall sunspot movement to the west is a result of this solar rotation. However, solar rotation can also be determined by observing
the solar spectrum at the solar limbs. The absorption lines in the spectrum will display a Doppler shift since the
east limb is coming toward the observer and the west limb is moving away. The velocity of the limb, relative to
the observer, can be determined from these spectral line shifts. Knowing the solar radius, the rotational period
can be calculated.
1. Introduction
The observation of sunspots demonstrates solar
rotation. The overall sunspot movement to the west
is a result of this solar rotation. Galileo calculated
the suns rotational period in 1610 using these observations.
The utilization of high resolution spectroscopic
analysis employing small aperture telescopes (0.35m)
can also be used to measure the suns rotational period.
The rotational velocity can be calculated by using the observed Doppler Effect. Observations of the
Hydrogen absorption line at the limb shows a small
spectral shift to shorter wavelengths as the east limb
is coming toward the observer and to longer wavelengths as the west limb is moving away. Comparing
these wavelength differentials from the laboratory
standard of the H line, the rotational velocity can be
calculated. Knowing the solar radius and the rotational velocity, the rotational period can be calculated.
2. Observations, Measurements and

Calculations
Hydrogen absorption spectral lines are quite obvious in both low and high resolution images of the
solar spectra and can be measured quite easily using
available computer software programs (Figure 1).
High resolution spectra of the suns H
absorption line were obtained using the Merritt Observatory 0.35m Schmidt casagrain telescope and the
Merritt Spectrograph. Spectra were taken on the
eastern and western limb of the sun (Figure 2).
Figure 1. Low resolution solar spectra showing hydrogen alpha absorption lines (McMath-Pierce Solar Observatory).
Figure 2. High resolution H absorption line.
The spectra were profiled for analysis using

RSpec Real Time Spectroscopy software (Figure 3).
The H absorption line wavelength was determined
after the spectra was calibrated using a neon standard
(Figure 4).
225
Trot = 2Rs = 25.3 days

Vrot
Calculating the rotational period error using:
Trot published = 25.05 days
then
Error, Trot = 0.25 days
3. Results
The values calculated apply to the equatorial region of the Sun; the rotational angular velocity varies
across the surface of the Sun from the equator to the
poles. This is referred to as differential rotation.
At the equator the solar rotation period is 25.05
days. This is the sidereal rotation period (Figure 5),
and should not be confused with the synodic rotation
period of 26.91 days; which is the time for a fixed
feature on the Sun to rotate to the same apparent position as viewed from Earth. The synodic period is
longer because the Sun must rotate for a sidereal period plus an extra amount due to the orbital motion of
the Earth around the Sun.
Figure 3. Profiled spectra.
Figure 4. Neon calibration spectra.
From the profiled spectra above, H wavelength

values were obtained:
H eastern limb =
H western limb =
H lab standard =
6562.757 0.005
6562.837 0.005
6562.797 0.001
Calculating the rotational velocity using:

C = 2.99792458 x 108 ms-1
and
Vrot = - o C = 1999.9 ms-1
o
Calculating the rotational velocity error using:
Vrot published = 7.189 x 106 ms-1
then
Error, Vrot = 2.99 ms-1
Calculating the rotational period using:
Rs = 6.955108 m
and
226
Figure 5. Differential solar sidereal rotation.
4. Conclusion
Significant assumptions have been made when
measuring the solar spectral lines. Physical effects
such as temperature, pressure, elemental abundance,
density and surface gravity have been ignored. These
properties affect the width and the height as well as
the shape and position of the spectral lines.
Considering that these assumptions and constraints would drastically complicate this exercise,
they have been excluded. This technique yields a
satisfactory value for the rotational period of the sun
considering that the Doppler shifts obtained were at
the limits of the resolution of the Merritt Spectrograph. This technique and apparatus can be used as
in a simple laboratory exercise for physics and astronomy classes demonstrating the methods and
techniques utilized by astronomers.
5. Acknowledgements
I would like to thank Michael Burin, CSUSM
Physics Department, for his insight and willingness
to utilize this exercise in his astronomy laboratory
class.
6. References
Gill, R. M. (2011). Construction of an Inexpensive
High Resolution Littrow Spectrograph for Be Star H
Alpha Analysis. in Proceedings of the 30th Annual
Symposium on Telescope Science (Warner et al.,
eds.). Society for Astronomical Sciences, Rancho
Cucamonga, CA.
RSpec. Real time spectroscopy.
http://www.rspec-astro.com/
227
228
Spectrometer Construction - Menke
Fast Spectrometer Construction and Testing

John Menke
22500 Old Hundred Rd
john@menkescientific.com
Abstract
This paper describes the construction and operation of a medium resolution spectrometer used in the visual
wavelength range. It is homebuilt, but has built in guiding and calibration, is fully remote operable, and operates
at a resolution R=3000. It features a fast f3.5 system, which allows it to be used with a fast telescope (18 inch
f3.5) with no Barlow or other optical matching devices.
1. Background
Five years ago I completed an 18inch f3.5 Newtonian for use primarily in asteroid and stellar photometry as described in
http://menkescientific.com/18inchscope.pdf.
Soon after, I obtained an SBIG DSS7 low resolution R = 100 (5A) spectrometer. While the DSS7 is
capable of very good work, one of the difficulties of
using the DSS7 with the Newtonian was that it is an
f10 instrument, thus requiring use of a 2.7x Barlow.
After several years, I saw the benefits of an instrument with increased resolution. The goals for the
instrument design would be
Native operation at f3.5, ie., no Barlow needed.
Resolution R=2-3000 (approximately 2A)
High Stability to allow precise wavelength
measurements and tracking
Self Guiding in a closed loop
Full remote operation
Integral calibration source
Inexpensive
Reasonable weight and with good balance
Construction using basic machine shop practices
Operable with almost any CCD camera
2. Spectrometer Design
Figure 1.
The biggest design challenge was to accommodate the fast f-ratio. The problem is not so much the
optics, as it is that such a fast converging light beam
from the telescope allows very little space for optical
elements once it leaves the telescope and before it
enters the slit. While one could build a spectrometer
into the telescope, this would make instrument
changeover very cumbersome. After considering a
variety of designs, I settled on my own version of a
Littrow design (in which the light from the grating
reflects back at approximately the entering angle) that
appeared to meet the design goals.
Figure 1 shows the overall spectrometer layout

roughly to scale. The figure shows the optical elements and surfaces as heavy black.
229
2.1 Focusser
A key element in the design is the focuser itself.
The focuser is an early Clement design featuring a
highly constrained and stiff kinematic design that is
only 1inch high at minimum and 4inches high fully
extended. As manufactured, the focuser uses an external 4inch high post and screw to vary the focus
with virtually no backlash. While the mechanism
works extremely well, its placement interfered with
mounting a spectrometer. To correct this, I modified
the focuser drive by making a new adjusting screw
which projects into the telescope (rather than external), and indeed, I also mounted the remote controlled (RoboFocus) stepping motor inside (powered
off except when changing focus). The focuser surface
is now completely clear.
2.2 Guider
The telescope must be actively guided to keep
the star on the slit. The telescope is approximately 1
a-s/pixel, and the slit opening at 50u would be about
5.5 pixels wide. After examining alternatives, I chose
a partially reflecting pickoff mirror in the optical
path. This is a simple system to build and allows the
guider camera potentially to see a large field. I used a
8% reflection, 92% transmission pellicle (super thin
mirror) for the optical element. This is mounted on a
PVC carrier installed in the inlet 2inch tubing. The
guide camera is a Starlight Express Lodestar camera,
uncooled, no shutter, 8.5u pixels (closely matching
the 9u pixels of the imaging camera). In practice, it
has no trouble guiding on mag12 stars using 5 sec
exposures.
2.3 Calibrator
I also built a simple neon calibrator that uses the
rear of the pellicle mirror to reflect light into the slit.
The Ne-2 neon bulb is driven by a 12VDC to
120VAC converter (remote controlled). Its light
passes through several small diffusing sheets to create a reasonable area source, then through a lens that
focuses at about f3-f4 off the back of the pellicle into
the slit.
2.4 Slit
The slit is a 50u 3mm high air (not glass) slit
mounted in a PVC carrier in the inlet tube. The star
image from the telescope is approximately 2.5a-s
FWHM, which translates to about 25u. Thus, a 50u
slit allows virtually all of a reasonably well focused
star to enter the spectrometer, while allowing for

some pointing and guiding error.
2.5 Spectrometer Body

The spectrometer is built of 1/8in aluminum
round and square tubing. The elliptical flat reflecting
mirror for the incoming light from the slit is mounted
on a simple section of aluminum angle and is adjusted by hand to that it picks up all the incoming
light while sufficiently offset to allow the light from
the grating to miss it on the way to the camera.
The light from the slit diverges toward the collimating lens. This is a simple achromat of 55mm dia
and 170mm focal length mounted on a simple PVC
holder that can slide along the long axis of the spectrometer. Adjustment screws allow the lens carrier to
be tilted and turned approximately 3deg. The light
passes through the collimating lens as a parallel
beam. It then reflects off the grating (1200L/mm,
2inch square, blazed for 5000A), converges back
through the collimating lens, bypasses the elliptical
flat, and enters the main camera (which is on axis).
The camera (I use an ST1603 w/ST8 size sensor) is
mounted to the spectrometer body using a machined
block of PVC.
Returning to the grating, it is attached to 1/8in
aluminum plate (using a relatively soft 5 minute epoxy), which in turn is mounted on a piece of aluminum angle with three sets of push/pull screws. This
allows the user to tilt the grating and to tightly lock
the setting. The aluminum angle containing the grating is mounted on a home-cut sector of worm wheel
that is pivoted under the center front surface of the
grating. The worm wheel was made from in. aluminum plate held in a jig while a small lathe drove a
hand tap to cut the teeth. The wheel is turned by a
small stepper motor driving a length of -20
threaded rod acting as the worm. The system allows
easy shift of the wavelength band, is stable in a given
position, and has fairly small, easy to compensate
backlash. The section of worm wheel was trimmed to
allow the grating to be turned square to the light
beam (ie, as a simple reflector in Order Zero) as well
as to turn far enough to reach the beginning of second
order.
A serial controlled RoboFocus controller drives
the stepping motor to the desired center wavelength
for the spectral region of interest as predetermined
using the neon calibrator. The controller incorporates
anti-backlash logic so that all moves finish in the
same direction. The controller also controls remote
operation of the neon calibrator.
230
3. Setting Up the Spectrometer

Setting up the instrument requires establishing
the positions and orientations of the optical elements
appropriately. I used several simple devices in the
alignment and testing process including the neon
calibrator and other light sources, plus a variety of
cheap red lasers for ray tracing to assure proper component orientations.
Once a good spectrum is obtained, one can use
the neon calibrator to measure the calibration and
dispersion in A/pixels. For this, one has to identify a
series of lines unambiguously. This does take some
practice, using frequent reference to web sites, e.g.,
http://astro.u-trasbg.fr/~koppen/discharge/index.html
and various tables of emission lines. A sodium flame
provides the D lines at 5889 and 5895 which helps
verify operation of the spectrometer (this spectrometer will easily split these two lines).
Finally, one has a choice of alignment of the
spectrographwill the slit be E-W or N-S on the
sky? For several reasons, I prefer E-W so that variations in clock drive move the star along the slit, rather
than off the slit.
4. Initial Targeting in the Field

The problem of getting the target star onto the
slit the first time may appear intimidating because the
user cannot image the target star and slit at the same
time. However, the process is in fact quite easy if one
follows a step by step approach that is detailed in the
article on our website. Subsequent aiming and focus
verification typically takes 2-5 minutes at the beginning of an observing session.
5. Conclusion
This project has been fun and satisfying, and has
resulted in a fully usable instrument that can easily
track below mag12 and take good spectra to about
mag8. The cost was about $500 for materials, and
required about 150 hours of work (of course, a second one would be much faster to build and debug).
The spectrometer met or exceeded all the goals, and
will be used to supplement the DSS7 spectroscopy at
our observatory.
A much more detailed version of this paper with
photos is on www.menkescientific.com/JohnsPage.
231
232
ADILS Spectrograph - Green
ADILS: An Amateur Dual Imaging Littrow Spectrograph

Wayne Green
Boulder Astronomy and Space Society, Boulder, CO
dxwayne@gmail.com
233
Encoders for an Old Telescope - Green
High Accuracy RA/Dec Encoders for a ca 1890 GEM

Wayne Green
Boulder Astronomy and Space Society, Boulder, CO
dxwayne@gmail.com
234
A Single Beam Polarimeter - Horne
A Single Bearm Polarimeter

Jerry D. Horne
jdhorne@hotmail.com
235
Notes
236

SAS 2012 Proceedings

Uploaded by

Copyright:

Available Formats

SAS 2012 Proceedings

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SAS 2012 Proceedings

Uploaded by

Copyright:

Available Formats

Proceedings for the 31st Annual Conference of the Society

for Astronomical Sciences

Symposium on Telescope Science

May 22-24, 2012

Published by the Society for Astronomical Sciences, Inc.

PHOTOMETRY OF HUBBLES FIRST CEPHEID

IN THE ANDROMEDA GALAXY, M31

BK LYNCIS: THE OLDEST OLD NOVA?

OR: ARCHAEO-ASTRONOMY 101

JONATHAN KEMP, JOE PATTERSON, ENRIQUE DE MIGUEL, GEORGE ROBERTS,

PHOTOMETRIC MONITORING BY AMATEURS IN SUPPORT

BRUCE L. GARY, DR. LESLIE H. HEBB, JERROLD L. FOOTE,

THE LOWELL AMATEUR RESEARCH INITIATIVE

DEIDRE ANN HUNTER, JOHN MENKE, BRUCE KOEHN,

LUNAR METEOR IMPACT MONITORING

AND THE 2013 LADEE MISSION

FIRST ATTEMPTS AT ASTEROID SHAPE MODELING

DIURNAL PARALLAX DETERMINATIONS OF ASTEROID DISTANCES

ON THE MAXIMUM AMPLITUDE OF HARMONICS

PV CEPHEI AND GYULBUDAGHIANS VARIABLE NEBULA

WHY VISUAL OBSERVATIONS OF MU CEPHEI ARE IMPORTANT

ROAD (REMOTE OBSERVATORY ATACAMA DESERT):

ER URSAE MAJORIS: A DWARF NOVA WITH SURPRISES

ENRIQUE DE MIGUEL, JOE PATTERSON, JONATHAN KEMP,

SMALL TELESCOPE SPECTROSCOPY OF EPSILON AURIGAE

OBSERVATIONS USING A BESPOKE MEDIUM RESOLUTION FAST SPECTROGRAPH

HIGH RESOLUTION SPECTROSCOPY FOR THE AMATEUR:

EXTREMELY LOW-COST POINT-SOURCE SPECTROPHOTOMETRY (ELCPSS)

JOHN BEAVER, CHARLES CONGER

SPECTROSCOPIC ANALYSIS OF ALGOL DURING ECLIPSE CYCLE

A FRESH LOOK AT THE ALGOL-LIKE ECLIPSING BINARY, AO SER

KEVIN B. ALTON, ANDREJ PRA

E.R. CRAINE, B.L. CRAINE, P.R. CRAINE, E.M. CRAINE

OBSERVING DOUBLE STARS

RUSSELL M. GENET, B.J. FULTON, FEDERICA B. BIANCO, JOHN MARTINEZ,

AMATEUR IMAGE PIPELINE PROCESSING

USING PYTHON PLUS PYRAF

HIGH TIME RESOLUTION ASTRONOMY

OR HIGH SPEED PHOTOMETRY

GARY A. VANDER HAAGEN

PHOTON COUNTING ONE MORE TIME

TRACKING BOLIDES, 3D VISUALIZATION AND DATA

THOMAS G. KAYE, ROBERT CRAWFORD, MARK BOWLING, JOHN KALAS

TOOLS AND TECHNIQUES FOR MEASURING ASTEROID OCCULTATIONS

PAPERS WITHOUT PRESENTATION AND POSTERS

A VIRTUAL ASTRONOMICAL RESEARCH MACHINE IN NO TIME (VARMINT)

THE CONFUSING CASE OF 16666 LIROMA

ROBERT K. BUCHHEIM, JOHN RUTHROFF

STUDENT PROJECT AND CURRICULUM BASED ON

ERIN M. CRAINE, JENNIFER C. DEBENEDETTI

ENHANCING THE EDUCATIONAL ASTRONOMICAL EXPERIENCE

ROBERT M. GILL, MICHAEL J. BURIN

THE ROTATIONAL PERIOD OF THE SUN

FAST SPECTROMETER CONSTRUCTION AND TESTING

ADILS: AN AMATEUR DUAL IMAGING LITTROW SPECTROGRAPH

HIGH ACCURACY RA/DEC ENCODERS FOR A CA 1890 GEM

A SINGLE BEARM POLARIMETER

Apogee Instruments, Inc.

Sky Publishing Corporation

Santa Barbara Instruments Group