36 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 3, NO.

1, JANUARY 2006

A Novel ELF Radar for Major Oil Deposits

Jamesina J. Simpson, Student Member, IEEE, and Allen Taflove, Fellow, IEEE

Abstract—This letter proposes a novel extremely low frequency 3) good stability compared to higher frequency methods that
(ELF) radar for major oil deposits. Using our recently developed are more influenced by ionospheric disturbances.
whole-Earth electromagnetic wave propagation model based upon
the finite-difference time-domain method, we have determined that
Due to these particular properties of ELF waves, Velikhov
detection of the radial (vertical) component of the scattered -field et al. introduced the idea of simultaneously performing studies
provides a sensitive means to detect oil fields that are located within around the globe involving oil-bearing rock, seismic regions,
several kilometers of the Earth’s surface. As an example, we pro- ionospheric wave propagation, and other areas of geophysics
vide numerical simulations of ELF radar returns from a hypothet- using one powerful ELF source.
ical Alaskan oil field excited by a 20-Hz pulse emitted from the
former U.S. Navy site in Wisconsin. The proposed method would
The ELF sounding methodology described in [1] and [2],
potentially provide means to rapidly and inexpensively conduct however, involves EM field impedance measurements to deter-
aerial surveys of thousands of square kilometers for significant oil mine the resistivity of the underlying rock. This requires mea-
deposits. surement of both the tangential electric field and the tan-
Index Terms—Aerial electromagnetic surveying, Earth, electro- gential magnetic field near the Earth’s surface. In this
magnetic sounding, extremely low frequency (ELF), finite-differ- letter, we propose a novel ELF radar for major oil deposits that
ence time-domain (FDTD), oil field, SQUID magnetometer, U.S. requires detection of only the radial magnetic field, , as a
Navy Wisconsin Transmitter Facility (WTF). low-frequency pulse is radiated outward from a distant pow-
erful ELF antenna. Using rigorous, large-scale computational
I. INTRODUCTION solutions of Maxwell’s equations, we have determined that
exhibits an unexpected and very high degree of sensitivity to the

D EEP electromagnetic (EM) sounding of the Earth using

controlled sources has been employed for decades as a
means to determine the electrical properties of the Earth’s litho-
presence of deeply buried conductivity anomalies of the litho-
sphere, much more so than . We propose exploiting this
phenomenon to establish a means to rapidly and inexpensively
sphere. Previously, controlled sources such as power lines, elec-
conduct aerial surveys of thousands of square kilometers for sig-
trified railroads, and pulsed magnetohydrodynamic (MHD) gen-
nificant oil fields [3].
erators have been employed for such probing [1]. However, the
As an example of the capabilities of our proposed ELF
data obtained from these methods exhibits an unsatisfactory
radar, we report numerical simulations of ELF radar returns
wide variance [1]. Further, the above techniques can only be
from a hypothetical Alaskan oil field excited by a 20-Hz pulse
applied at considerable cost and effort, making their implemen-
emitted from the former U.S. Navy site in Wisconsin. For
tation infeasible in many regions of the world. As a result, new
this study, we use our recently developed whole-Earth EM
approaches to deep EM geophysical prospecting are desirable.
wave propagation model [4], [5] based upon the finite-differ-
If a superior technique applicable on a global scale were devel-
ence time-domain (FDTD) solution of Maxwell’s equations
oped, it could permit both rapid and inexpensive surveying of
[6]. This technique permits a direct three-dimensional (3-D)
the entire Earth’s crust. Such a method would be of particular
time-domain calculation of round-the-world ELF propagation
interest for locating anomalous conductivity structures such as
accounting for arbitrary horizontal as well as vertical geo-
mineral deposits and major oil fields.
metrical and electrical inhomogeneities/ anisotropies of the
Since the early 1990s, a promising new system for global re-
excitation, ionosphere, lithosphere, and oceans. The robustness
mote sensing has been under investigation [1], [2]. This tech-
of the FDTD method has in recent years caused it to become
nique involves the use of a powerful radio-transmitting antenna
a popular tool for analyzing EM wave propagation around the
in the extremely low frequency (ELF) (3 Hz to 3 kHz) range.
complete Earth-ionosphere waveguide [7]–[9].
Velikhov et al. [1] outlined the following advantages of em-
ploying ELF radio waves up to 100 Hz for EM sounding:
II. THREE-DIMENSIONAL LATITUDE-LONGITUDE FDTD GRID
1) sufficiently large skin depth for deep probing (100 m in
ocean and 10–15 km in crystalline shields); Fig. 1 illustrates the general layout of the FDTD space-lat-
2) low propagation attenuation, which could provide global tice as seen from the transverse magnetic (TM) plane at a con-
remote sensing of the Earth with a single source; stant radial coordinate [5]. The lattice is a logically Cartesian
-cell arrangement, where is a power of 2. We see
that the grid cells follow lines of constant latitude, constant,
Manuscript received April 19, 2005; revised June 26, 2005. This work was
supported in part by the National Computational Science Alliance under Grant where is the usual spherical angle measured from the north
DMS040006N. pole; and along lines of constant longitude, constant, where
The authors are with the Electrical Engineering and Computer Science De- is the usual spherical azimuthal angle measured from a spec-
partment, Northwestern University, Technological Institute, Evanston, IL 60208
USA (e-mail: j-simpson@northwestern.edu). ified prime meridian. In this manner, each TM plane of the grid
Digital Object Identifier 10.1109/LGRS.2005.856118 shown in Fig. 1 is comprised of isosceles trapezoidal cells away
1545-598X/$20.00 © 2006 IEEE
SIMPSON AND TAFLOVE: NOVEL ELF RADAR FOR MAJOR OIL DEPOSITS 37

excited by an ELF pulse radiated from a distant transmitter.

Specifically, we model the oil field as a lithosphere conduc-
tivity anomaly of lateral dimensions 28 39 km and thickness
1.25 km located in Alaska at 156 W, 69 N. The anomaly is as-
sumed to have a conductivity that is ten times lower than that
of the surrounding strata. By way of comparison, Alaska’s Ku-
paruk River oil field spans a lateral area of about 1000 km [10].
We use two separate FDTD models to calculate the potential
ELF radar signatures of all three -field components at a point
directly above the conductivity anomaly. Model A serves as
the reference (i.e., no conductivity anomaly present). Model B
includes the conductivity anomaly assumed to be located at a
median depth of 1.25 km.
For the remainder of the global lithosphere model, we uti-
lize topographic and bathymetric data from the NOAA-NGDC
“Global Relief CD-ROM.”1 These data are mapped onto the 3-D
Fig. 1. General layout of the 3-D FDTD lattice covering the complete
space lattices of both models with an assumed lateral resolution
Earth-sphere as seen in a TM plane at a constant radial coordinate. of 40 40 km at the equator. The complete FDTD grid extends
to a depth of 100 km into the lithosphere and to an altitude of
from the north and south poles [5], and isosceles triangular cells 100 km. The grid is terminated at both edges in the radial direc-
at the poles [5]. Similarly, each transverse electric (TE) plane tion by a perfect electric conductor (PEC). The PEC boundary
at a constant radial coordinate is comprised of isosceles trape- is adequate for this model since the skin depth within the litho-
zoidal cells away from the north and south poles, and a polygon sphere and reflection height of the ionosphere are each less than
cell at each pole. 100 km. Subgridding [6] is used to refine the radial grid reso-
We choose to have the same angular increment in latitude lution in the lithosphere near the Earth’s surface to 1.25 km, a
for each cell in the grid. Thus, the south–north span factor of four times finer than the nominal radial resolution of
of each trapezoidal or triangular grid cell is , 5 km used in the atmosphere and deeper within the lithosphere.
where is the radial distance from the center of the Earth. To Numerical experiments have shown that the subgridding intro-
maintain square or nearly square grid cells near the equator, we duces sufficiently low numerical error relative to the physical
select the baseline value of the angular increment in longitude phenomena being modeled. For the lithosphere, conductivity
, to equal . However, this causes the west–east span of values are assigned according to [11], depending upon the loca-
each cell to be a function of . This could tion of an E component (i.e., below an ocean or within a conti-
be troublesome for cells near the north and south poles where nent). For the atmosphere, both day- and nighttime exponential
and , respectively. There, the geometrical ec- conductivity profiles used in [12] are assumed, with midnight
centricity of each cell would occurring at GMT. The effective waveguide height of reflection
become quite large, and the numerical stability and efficiency for these profiles is roughly 48 km for day and 76 km for night
of the FDTD algorithm would be degraded. We mitigate this [12].
problem by merging pairs of adjacent cells of the TM plane in Velikhov et al. [1] and Bashkuev and Khaptanov [2] em-
the west–east direction, effectively halving the cell eccentricity ployed “Zevs,” an ELF antenna system located in the north
[5]. This process can be repeated several times as the grid ap- of the Kola Peninsula in Russia, for their impedance studies.
proaches a pole, allowing the user to specify a maximum allow- Aside from Zevs, the only known practical ELF communi-
able cell eccentricity. cation system is the facility operated by the U.S. Navy in
The wrap-around or joining of the east and west edges of upper Michigan and Wisconsin until September 2004 [13].
the lattice occurs along a specific line of constant longitude, or Our computational model assumes the Wisconsin Transmitting
meridian. This joining is, in effect, a periodic boundary condi- Facility (WTF) section of the Navy site located near Clam
tion applied at each -row of lattice cells, whether trapezoids or Lake, WI (90.9 W, 46.5 N) as the ELF source used to detect
triangles [5]. the lithosphere conductivity anomaly in Alaska (at a distance
Given the above assumptions, Ampere’s Law in integral of about 4.4 Mm). We assume nominal operation of this facility
form can be applied to develop an FDTD time-stepping relation with two orthogonal ground lines, each 22.5 km long and
for each electric field components of the grid [5]. Similarly, carrying a current of 300 A, one oriented in the north–south
Faraday’s Law in integral form can be applied to develop an direction and the other in the east–west direction. However,
FDTD time-stepping relation for the magnetic field compo- balancing between the skin depth in the lithosphere and the
nents of the grid [5]. radiation efficiency of WTF, our assumed carrier frequency
is 20 Hz, rather than the 75-Hz Navy frequency. We further
III. DETAILS OF THE FDTD EM SOUNDING STUDY AT ELF assume that the transmitted signal is a pulse that is generated by
double-sideband amplitude modulation of the 20-Hz carrier by
In this letter, our goal is to determine how the presence of a
major oil field influences the surface -field components when 1See http://www.ngdc.noaa.gov/mgg/fliers/93mgg01.html.
38 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 3, NO. 1, JANUARY 2006

Fig. 2. Comparison of the absolute value of the time waveforms for surface Fig. 3. Comparison of the absolute value of the time waveforms for surface
H at 156 W, 69 N in Alaska for a pulsed 20-Hz signal originating H at 156 W, 69 N in Alaska for a pulsed 20-Hz signal originating from WTF,

from WTF, normalized to the peak incident H : (a) Incident pulse and normalized to the peak incident H : (a) Incident pulse and (b) radar signature
(the difference between the calculated surface magnetic field component in
2 2
(b) radar signature (the difference between the calculated surface magnetic field
2 2
component in models A and B) for a 28 39 1:25 km conductivity anomaly models A and B) for a 28 39 1:25 km conductivity anomaly at a mean
at a mean depth of 1.25 km. The Matlab function “smooth” was applied to the depth of 1.25 km. The Matlab function “smooth” was applied to the data to
data to filter out high-frequency fluctuations arising from subtraction noise. filter out high-frequency fluctuations arising from subtraction noise.

a Gaussian envelope waveform of full-width at half-maximum

42.5 ms. Below the WTF, we model the Laurentian Plateau
(or Canadian Shield), a large region of low-conductivity rock
of 2.4E-4 S/m that extends northward from the Great Lakes
toward the Arctic Ocean and includes much of Canada and
Greenland.

IV. COMPUTATIONAL MODELING RESULTS

Using a decibel scale, Fig. 2 shows our calculated results for
the absolute value of the time-waveform of the surface at
observation point normalized to the incident . The fol-
lowing two cases are illustrated: the incident pulse (Model A, no
conductivity anomaly present [Fig. 2(a)] and a radar signature
(the difference between the calculated surface magnetic field
component in models A and B) for the km con- Fig. 4. Comparison of the absolute value of the radar signature time
ductivity anomaly at a median depth of 1.25 km [Fig. 2(b)]. We 2 2
waveforms of Figs. 2 and 3 for a 28 39 1:25 km conductivity anomaly
see that the radar signature of the conductivity anomaly is at a mean depth of 1.25 km at 156 W, 69 N in Alaska. Both are normalized
to the peak incident H for a pulsed 20-Hz signal originating from the WTF:
well below 80 dB relative to the incident pulse. Therefore, al- (a) H and (b) H . The Matlab function “smooth” was applied to the data to
though the incident pulse is detectable, given the experience of filter out high-frequency fluctuations arising from subtraction noise.
the Navy’s virtually worldwide communications with its deeply
submerged submarines, the presence of the conductive anomaly makes clear that the sensitivity of the impedance technique is
in the lithosphere introduces only a very small perturbation in considerably less than afforded by a measurement of .
the surface and is hence difficult to detect. Fig. 4 shows the (b) curves of Figs. 2 and 3, the calculated
Similar to Fig. 2, Fig. 3 shows the corresponding calculated time-waveforms of the surface and radar signatures, in
results for the absolute value of the time-waveform of the sur- a different form. Here, by normalizing both waveforms to the in-
face at observation point normalized to the incident . cident , we see that the presence of a 1.25-km median-depth
Here, we see that the radar signature of the conductivity conductivity anomaly generates and radar signatures
anomaly rises to 15 dB relative to the incident . Therefore, of comparable magnitudes. Therefore, considering the results of
the presence of the conductivity anomaly in the lithosphere in- Figs. 2 and 3, the perturbation produced by the anomaly relative
troduces a very large perturbation in at the surface, which to the background is over 65 dB larger than that for . We
greatly bolsters the possibility of detecting its presence over believe that this relatively enhanced radial magnetic field com-
. In fact, since is required for the impedance measure- ponent is due to looping currents within the lithosphere that cir-
ment technique reprted in [1] and [2], comparing Figs. 2 and 3 culate around the thin conductivity anomaly in the horizontal
SIMPSON AND TAFLOVE: NOVEL ELF RADAR FOR MAJOR OIL DEPOSITS 39

plane. We also find from analogous studies involving deeper Our ongoing work in this area includes developing a paral-
conductivity anomalies that measuring becomes increas- lelized alternative hexagonal/pentagonal 3-D geodesic FDTD
ingly superior for anomalies at greater depths. Furthermore, the grid with superior efficiency [15], [16].
outline (shape and size) of conductivity anomalies can be de-
termined. Finally, we find that these radar signatures are de-
tectable given the results of [2] in which was measured for REFERENCES
impedance variation studies at distances of 4 Mm from the ELF [1] E. P. Velikhov, A. A. Zhamaletdinov, A. N. Shevtsov, A. D. Tokarev,
source and for depths of 10–20 km. Y. M. Kononov, L. B. Pesin, G. M. Kadyshevich, M. I. Pertel, and A.
V. Veshchev, “Deep electromagnetic studies with the use of powerful
ELF radio installations,” Isvestiya, Phys. Solid Earth, vol. 34, no. 8, pp.
V. IMPLICATIONS FOR A POSSIBLE DETECTION SYSTEM 615–632, 1998.
Our FDTD calculations have shown that a powerful distant [2] Y. B. Bashkuev and V. B. Khaptanov, “Deep radio impedance sounding
of the crust using the electromagnetic field of a VLF radio installation,”
ELF antenna can be used to locate oil-bearing rock within sev- Izvestiya, Phys. Solid Earth, vol. 37, no. 2, pp. 157–165, 2001.
eral kilometers of the Earth’s surface by measuring the highly [3] J. J. Simpson and A. Taflove, “Whole Earth FDTD modeling of ELF
sensitive surface . Given these results, we propose that an electromagnetic sounding of oil deposits with the U.S. Navy’s ELF
transmitter system,” presented at the USNC/URSI National Radio
ELF source such as the WTF, used in conjunction with an air- Science Meeting, Washington, D.C., Jul. 2005.
borne magnetometer having sufficient sensitivity to , can be [4] , “Two-dimensional FDTD model of antipodal ELF propagation
and Schumann resonance of the Earth,” IEEE Antennas Wireless Prop-
used to rapidly and inexpensively detect oil fields over thou- agat. Lett., vol. 1, pp. 53–56, 2002.
sands of square kilometers. In fact, considering the very low [5] , “Three-dimensional FDTD modeling of impulsive ELF propaga-
propagation attenuation of EM waves at ELF [5], employing a tion about the Earth-sphere,” IEEE Trans. Antennas Propagat., vol. 52,
no. 2, pp. 443–451, Feb. 2004.
sufficiently powerful ELF antenna permits use of this sounding [6] A. Taflove and S. C. Hagness, Computational Electrodynamics: The Fi-
technique on a global scale. nite-Difference Time- Domain Method, 3rd ed. Norwood, MA: Artech
We propose the use of SQUID magnetometers for the aerial House, 2005.
surveys described in this letter because they: 1) measure the [7] Y. Ando, M. Hayakawa, A. V. Shvets, and A. P. Nickolaenko, “Finite
difference analyzes of Schumann resonance and reconstruction of
magnetic field directly; 2) offer higher sensitivity at frequencies lightning distribution,” Radio Sci., vol. 38, no. 6, p. 1103, 2003. DOI:
less than 150 Hz than other receiver technologies; and 3) pro- 10.1029/2002RS002752.
vide higher bandwidths [14]. Further, they have already been [8] A. Soriano, E. A. Navarro, D. L. Paul, J. A. Porti, J. A. Morente, and I. J.
Craddock, “Finite-difference time domain simulation of the Earth-iono-
successfully implemented in aerial surveys [14]. sphere resonant cavity: Schumann resonances,” IEEE Trans. Antennas
Propagat., vol. 53, no. 4, pp. 1535–1541, Apr. 2005.
VI. CONCLUSION AND ONGOING WORK [9] H. Yang and V. P. Pasko, “Three-dimensional finite difference time do-
main modeling of the Earth-ionosphere cavity resonances,” Geophys.
In this letter, we have proposed a novel ELF radar for major Res. Lett., vol. 32, no. L03114, 2005. DOI:10.1029/2004GL021343.
[10] U. S. Congress, Office Technol. Assessment, “Oil production in the
oil deposits. As an example of this new technology, we pro- Arctic National Wildlife Refuge: The technology and the Alaskan oil
vided numerical simulations of ELF radar returns from a hy- context,” U.S. Gov. Printing Office, Washington, DC, OTA-E-394, Feb.
pothetical Alaskan oil field excited by a 20-Hz pulse emitted 1989.
[11] J. Hermance, “Electrical conductivity of the crust and mantle,” in Global
from the former U.S. Navy site in Wisconsin. By employing Earth Physics: A Handbook of Physical Constants. Washington, DC:
our 3-D FDTD model of the entire Earth-ionosphere cavity, AGU, 1995.
we determined that the surface exhibits an unexpected and [12] P. Bannister, “The determination of representative ionospheric conduc-
tivity parameters for ELF propagation in the Earth-ionosphere wave-
very high degree of sensitivity to the presence of deeply buried guide,” Radio Sci., vol. 20, no. 4, pp. 977–984, 1985.
conductivity anomalies of the lithosphere, much more so than [13] J. Garza, “Navy pulls plug on embattled ELF site,” Milwaukie J. Sen-
. We therefore proposed that airborne SQUID magnetome- tinel, Sep. 17, 2004. [Online]. Available: www.jsonline.com.
[14] J. B. Lee, D. L. Dart, R. J. Turner, M. A. Downey, A. Maddever, G. Pan-
ters recording only , synchronized with the transmission of jkovic, C. P. Foley, K. E. Leslie, R. Binks, C. Lewis, and W. Murray,
an ELF pulse from a distant, powerful ELF antenna, may poten- “Airborne TEM surveying with a SQUID magnetometer sensor,” Geo-
tially provide the global locations of oil fields in a rapid and in- physics, vol. 67, no. 2, pp. 468–477, Mar.–Apr. 2002.
[15] J. J. Simpson and A. Taflove, “Efficient modeling of impulsive ELF an-
expensive manner. Although the case studied here is at 4.4 Mm tipodal propagation about the Earth sphere using an optimized two-di-
from the transmitter, we believe that detection at much larger mensional geodesic FDTD grid,” IEEE Antennas Wireless Propagat.
distances should be possible because at 20 Hz the additional Lett., vol. 3, pp. 215–218, 2004.
[16] , “3-D FDTD modeling of ULF/ELF propagation within the global
attenuation is only about 0.4 dB/4 Mm, which should be well Earth-ionosphere cavity using an optimized geodesic grid,” presented at
within the dynamic range of existing instruments. the IEEE AP-S Int. Symp., Washington, DC, Jul. 2005.