CONFORMAL MAP

PROJECTIONS IN
GEODESY

E. J. KRAKIWSKY

September 1973

TECHNICAL REPORT
LECTURE NOTES
NO.
37217
CONFORMAL MAP PROJECTIONS
IN GEODESY

Edward J. Krakiwsky

Department of Geodesy and Geomatics Engineering

University of New Brunswick
P.O. Box 4400
Fredericton, N.B.
Canada
E3B 5A3

September 1973
Latest reprinting January 1998
 PREFACE

In order to make our extensive series of lecture notes more readily available, we have
scanned the old master copies and produced electronic versions in Portable Document
Format. The quality of the images varies depending on the quality of the originals. The
images have not been converted to searchable text.
 PREFACE

The purpose of these notes is to give the theory and use of

some common conformal map projections. The approach used is straight-

forward. It begins with the basics of differential geometry and

conformal mapping. Then, given the conditions for a particular map

projection, the particular conformal mapping equations are derived.

This is a self-contained analytical approach.

The author is aware of attempts by at least one mathematician

[Wray, 1973] to unify all classes of map projections such that the same

set of mapping equations can be used to obtain all or many of the well-

known projections, simply by assigning appropriate v~lues to certain

parameters in some generalized equations. This is somewhat of a

simplified explanation of what actually happens, but it at least

illustrates the generality of this contemporary approach. As of the

date of writing these notes, the details (necessary for instruction at

the undergraduate level) were not yet worked out and thus no use was

made of it.

There is yet another attempt to obtain a generalized set of

equations for conformal map projections alone. As explained in Section

9, this approach consists of one set of mapping equations which is

capable of producing the well-known conformal projections · Mercator,

Transverse Mercator, Lambert Conformal Conic, Stereographic and even

other unnamed projections. This is achieved simply by assigning specific

values for certain constants in a generalized set of equations. This

approach is not readily usable since the expressions have not been

i
developed to a sufficient degree of accuracy, and thus are not satis-

factory for the practising surveyor. Only a brief description of this

approach is given in these notes.

More on the approach used herein. Complex arithmetic is

exploited at every opportunity. Series expansions are avoided at

occasions when the closed form exists. The reason being that computer

centres nowad~s have routines to evaluate natural logs, exponentiation,

etc. Derivations are given to show the origin and important steps in

the development of the main equations. Lengthy and detailed derivations

are omitted from the text and reference made to an appropriate source or

an appendix added.

These notes have been w:~i tten under the assumption that the

reader has knowledge of differential and integral calculus, complex

arithmetic, ellipsoid geometry, and some knowledge of computer programm-

ing.

These notes have evolved from the author's lecture notes over

the past few years. Two main sources are acknowledged at the outset as

being the starting point for these notes - Thomas' "Conformal Projections

in Geodesy and Cartography", U.S. Geodetic Survey Special Publication

No. 251, and Dr. Richard H. Rapp's lecture notes on "Advanced Map Pro-

jections", Department of Geodetic Science, The Ohio State University,

Columbus. Other sources used for important details are referenced within

the text.

The author wishes to acknowledge the contribution made by the

Surveying Engineering undergraduate class of 1974-75 to improving the

notes by finding typographical errors. Messrs. Donald B. Thomson and

M. Nassar, graduate students, are acknowledged for their constructive

criticism of the notes, Mr. Thomson is particularly acknowledged for

the preparation of the appendices.

E • Krakiwsky
December 9, 1974

The author wishes to acknowledge Kresho Frankich for his

detailed critique of the December, 1974 version of this work. His

involvement has lead to the improvement of section 2.4 and Chapters

7, 8 and 10. Dr. Donald B. Thomson helped to clarify aspects of the

stereographic projection. Robin Steeves extended various formulae to

meet a higher degree of accuracy. Ms. Wendlynn Wells is thanked for

her editorial assistance in preparing this version.

E. Krakiwsk:y
July 18, 1977

iii
 TABLE OF CONTENTS
Page

PREFACE i

TABLE OF CONTENTS iv

GENERAL 1

SECTION I - CONFORMAL MAPPING . 2

1. INTRODUCTION TO CONFORMAL MAPPING • . 3

2. REVIEW OF COMPLEX VARIABLES 5

2.1 Rules of Complex Arithmetic 5
2.2 Complex Plane .... . ..
..... 7
2.3 Complex Function . . . . . • . 8
2.4 Differentiation of a Complex Function 9
2.5 Special Complex Functions .••.. 14
3. REVIEW OF DIFFERENTIAL GEOMETRY 16
3.1 Parametric Equations 16
3.2 Gaussian Fundamental Quantities 18
3.3 Angle Between Parametric Curves 21

1+ • CONFORMAL PROJECTIONS IN GENERAL 24

4.1 Isometric Plane • . . . . 24
4.2 Scale Factor Definition 30
4.3 Condition for Conformality 32
4.4 Scale Factor Evaluation 36
4.5 Geometry of Projected Curves 38
4.6 Meridian Convergence 40
5. MERCATOR PROJECTION 44
5.1 Direct Problem .. 44
5.2 Scale Factor .... ... 45
5.3 Meridian Convergence .... ... 46
5.4 Appearance of the Mercator Projection .... 46
5.5 The Loxodrome .... .... 48
5.6 Inverse Problem .... .. 49
6. TRANSVERSE MERCATOR PROJECTION • • • . 51
6 .1 Direct Problem . . . . . • • • . 51
6.2 Meridian Convergence in Terms of Geodetic Coordinates 59
6.3 Scale Factor in Terms of Geodetic Coordinates 60
6.4 Appearance of the Transverse Mercator Projection . 61
6.5 Inverse Problem . • . . . . . . . • . • . . . . . 62
6.6 Meridian Convergence in Terms of Map Coordinates . 69
6.7 Scale Factor in Terms of Map Coordinates . • . • . • 70
6.8 Universal Transverse Mercator (UTM) . . . . 71
6.9 Transverse Mercator in Three Degree Zones • • • • • • 77

iv
TABLE OF CONTENTS (cont'd)
Page

7. LAMBERT CONFORMAL CONIC PROJECTION ..... 81

7.1 Direct Problem ..... 81
7.2 Scale Factor ........... 87
7.3 Meridian Convergence 87
.....
7.4 Two Standard Parallel LCC ..... 88
7.5 One Standard Parallel LCC 89
7.6 Modification to Coordinate System .... .... 91
7-7 Inverse Problem .... .... 92

8. STEREOGRAPHIC PROJECTION 94
8.1 Polar Stereographic Projection • • • • 94
8.2 The Stereographic Projection System in New Brunswick 96
9. A GENERALIZED SET OF CONFORMAL MAPPING EQUATIONS 97
9.1 Introduction . • . . • . . • • . . • • . • • "". 97
9.2 Generalized Formulae for Conformal Projections • 98
9.3 Application of the General Scale Formula • • 98
9. 4 summary . . . . . . . . . . . . . . . . . . • 106

SECTION II - COMPUTATIONS ON A CONFORMAL

MAP PROJECTION PLANE 109
10. INTRODUCTION TO COMPUTATIONS ON A CONFORMAL
MAP PROJECTION PLANE 110

11. REDUCTION OF OBSERVATIONS 113

11.1 Reduction of Angles 113
11.2 Reduction of Azimuths 114
11.3 Reduction of Distances 116
12. MATHEMATICAL MODELS FOR COMPUTATION OF POSITIONS 117
12.1 Intersection • • • . . • . • • • • • • • • • 117
12.2 Traverse with Astronomic Azimuth Orientation 119
12.3 Traverse with Grid Azimuth Orientation . . • • 120
12.4 Adjustment of a Network on a Conformal Map Plane 120
12.5 Gyro-Theodolite Calibration Using Map Coordinates • 121

13. GENERAL FORMULAE FOR REDUCTION TO THE MAP PROJECTION PLANE. • 122
13.1 Curvature of the Projected Geodesic ••.•.• 123
13.2 Parametric Equations of the Projected Geodesic 124
13.3 Difference in Length Between Projected Geodesic
and Chord . . . . . . . . . • . • . . . • . 125
13.4 Difference in Length Between the Projected Geodesic
and the Geodesic . . . . . • . . . . . . . . . . . 126
13.5 The Angle Between the Projected Geodesic and the
Chord . • . . • . . . . .. . . • . . . • . . . . 127

v
TABLE OF CONTENTS (cont'd)
Page

14. SPECIFIC FORMULAE FOR REDUCTION TO VARIOUS MAP

PROJECTION PLANES . . . . . . . . . • . . • 129
14.1 Reduction Formulae for the Transverse Mercator 129
14 .1.1 Curvature . . • • . . . • . • . 130
14.1.2 Arc to Chord Evaluation .•.•• 130
14.1.3 Line Scale . . . . . • • . • . . 133
14.2 Reduction Formulae for the New Brunswick
Stereographic Projection . • • . . . . 135
14.2.1 Arc to Chord Reduction . . • . • 135
14.2.2 Geodesic to Chord Length Reduction 136

REFERENCES 137

APPENDIX I: EVALUATION OF MERIDIAN ARC LENGTH 138

APPENDIX II: EVALUATION OF FOOTPOINT LATITUDE •. 141

APPENDIX III: ALTERNATIVE DERIVATION OF LAMBERT CONFORMAL

CONICAL PROJECTION . • • . • • • . • •

vi
 GENERAL

Conformal map projections are the class of projections in

which angles on the surface to be mapped are preserved, tnat is, corr-

esponding angles on the map plane and the surface are equal. We will

restrict the surface to be mapped to be the ellipsoid as defined in

geodesy. The map plane is regarded as a flat - two dimensional surface.

The two corresponding sets of coordinates for any point are the geodetic

latitude cp and longitude A., and the mapping coordinates x andy.

There are two main problem areas in conformal map projections.

The first area is concerned with the transformation (mapping) of the

geodetic latitude (cp) and longitude (A.) into a pair of mapping coor-

dinates x andy, and vice versa, that is the mapping of x andy into

¢ and A.. The second problem area involves the computations of geodetic

positions on the conformal mapping plane from observed quantities that

have been appropriately projected onto the map plane. Because of the

clear identification of these two tasks, the notes have been structured

into two sections. Section I is called Conformal Mapping, while

Section II is called Computations on a Mapping Plane.

1
 2

SECTION I

CONFORMAL MAPPING
 3

1. INTRODUCTION TO CONFORMAL MAPPING

It was stated in the general introduction that the problem at

hand is the determination of the conformal transformation equations for

mapping~' A into x, y, and x, y into~' A. Employing complex arith-

metic notation, the problem is symbolized as

1-1
and

1-2

where r 1 and r 2 are the analytic functions to be determined, and i = ;-:I,

The first equation describes the direct problem and the second the

inverse problem.

The solution of these problems requires the use of same

elementary complex algebra which is reviewed in Section 2. Also, same

differential geometry is needed to characterize differential elements on

the ellipsoid and plane surfaces - this material is reviewed in Section 3.

The principles of conformal mapping theory needed for our problemsare

briefly given in Section 4. Sections 5 through 8 respectively treat the

Mercator, Transverse Mercator, Lambert Conformal Conic and Stereographic

conformal projections. A brief description of the implementation of some

of these projections in practise as official coordinate systems is

given. The generalized set of mapping equations mentioned in the

preface is briefly described in Section 9.

 5

2• REVIEW OF COMPLEX VARIABLES

Complex algebra is used liberally in the development of the

various conformal mapping equations. The purpose of this section is to

briefly review, thus summarize, those parts of complex algebra needed

herein. The formal and complete treatment can be found in many math-

ematical texts.

Reviewed in this section are: rules of complex arithmetic;

the complex plane; the complex function; differentiation of a complex

function and the derivation of the Cauchy-Riemann equations; special

complex mapping functions.

2.1 Rules of Complex Arithmetic

For a complex number

z =a + ib; i = 1(:15 2-l

where a is the real part and b the imaginary part, the following rules

and definitions apply:

 6

1. The sum of two complex numbers is a complex number, that is,

2-2

2. Addition is associative (grouping is immaterial), that is,

2-3

3. Every complex number has a unique negative, that is

z = a + ib
and -z = -a - ib 2-4
4. Addition is co'1liilutative (order is immaterial), that is

2-5

5. Multiplication is defined as

2-6

6. Multiplication is associative, that is

7. Every non-zero complex number z has a unique inverse, that is

z -1 =-
1 1 <a-ib) = a-ib
z = (a+ib) (a-ib) a2+b2

a b
= 2 b2
i
2 b2
a+ a+

8. The complex conjugate of z is

-z = a - ib. 2-9
 7

2.2 Complex Plane

The rectangular form of a complex number is

z = x1 + iy1 . 2-10

The geometric interpretation is a vector in two dimensional space, with

the real part (x) as the abscissa and the imaginary part (y) as the

ordinate (Figure 2-1)

y

1
y
1

0 X

Figure 2-1. He<:'l!ini<:uJ '1.r ?0rm

The polar form of a complex number is (Figure 2-2)

2-11

since x1 =r1 cose 1

2-12
y1 = r1 sine 1

Figure 2-2 Polar Form

 The argument of z is denoted as e1 , and defined by

-1
el = tan 2-13

The modulus of z 1 is denoted as r 1 , and defined by

2 2 l/2
rl = lzll = lxl + iylj = (xl + yl) 2-14

The x-y plane is called the complex plane.

2.3 Complex Function

w is said to be a function of z, that is

w= f(z) , 2-15

if there exists one or two values of the ·complex variable w for every

value of the complex variable z. In this arrangement, z is the independent

variable and w the dependent variable.

An ex&~ple of a single valued function is

2
w = z 2-16

and of a many valued (twcin this example) function i~

l/2
w 0: z 2-17

In mapping the ellipsoid onto a plane, we are interested only in single

valued functions.

In general

w = f(z) =u + iv ,

= f(x+iy) = u(x, y) + iv(x, y). 2-18

The corresponding conjugate is

w= f(z) = u(x, y) - iv(x, y) . 2-19

 9

2.4 Differentiation of a Complex Function

This section first defines the derivative of a complex function, that

is f'(z), and then the very useful and important Cauchy-Riemann equations.

Consider a set of complex numbers {z , z1 , ••• , z }, where z

· o n o
and z are particular values. It is said that f(z) is differentiable
n
at z if
0

f( z ) - f( z )
limit n o
z -+ z 2-20
z - z
n o n o
or

limit f(z + ~z) - f(z )

0 0 exists. 2-21
~z -+ 0 ~z

The basic equation for defining the derivative f'(z) follows

from above after recognizing that from

Z =X + iy

~z =~ + i!J.y 2-22

and

f(z + ~z) = u(x + ~. y + !J.y) + iv(x + ~. y + ~y) . 2-23

The basic equation-is

f(z + ~z)- f(z)

f'(z) = l i m i t . - - - - - - - -
~z -+ 0 ~z

limit {u(x + rue, y + ~y.) - u(x, y) +

-~z -+ 0 ~ + i~y

+ 1. v ( x + t.x , y + t.y ) - v (x , y ) } 4
2-2
rue + i~y

and is independent of the approat:h tuken~ But, for the'· approach

y = constant, that is, ~y = 0, and ·l'J.z = t:lx., . t'he result is

 10

f' ( z) = limit
u~x + b.x2 ;d - u{x 2 :t1 +
b.x
b.x -+ 0

+ limit i
v{x + b.x2 J::) - v{xz :tl 2-25
b.x
b.x -+ 0

au av
f'(z) = -ax+ i
ax
2-26

For the approach ~z -+ 0 along x = constant, that is b.x = 0,

~z = i!::.y • The result is,:

~ u(x, y + ~y) - u(x, y) +

f'(z) = !::.ylimit
-+ 0 i ~y

limit v(x, y + ~y) - v(x 2 y)

+ 2-27
!::.y -+ 0 ~y

1 au av
f' ( z) =--.-+- 2-28
i ay ay

The Cauchy-Riemann equations are obtained by equating the two

definitions of the derivative, equations 2-28 and 2-26, namely

2-29

-i au + av = au + i av 2-30
ay ay ax ax

Equating real and imaginary parts yields the Cauchy-Riemann equations

au av
-=
ax ay
2-31
av au
a;z =-ay
 ll

A function of complex number z which is single valued and

differentiable at every point of its domain, is said to be regular. The

necessary condition for the function f(z) to be differentiable at the

point z is that the four partial derivatives u , u , v , v must exist

X y X y
and satisfy the Cauchy-Riemann equations.

Since the partial derivatives of u and v are connected by the

Cauchy-Riemann equations (2-31) and assuming that these derivatives exist

and satisfy the relation u

xy
= uyx' v
xy = vyx' it follows by partial

differentiation that

and 2-32

Thus both u and v satisfy Laplace's equation

2 2 .
V2u =a u + ~ =0
ax 2 ay 2
2-33

By separating any regular function of z, i.e. a function with

existing derivatives, into its real and imaginary parts, we obtain immed-

iately two solutions of Laplace's equation.

From the equations (2-31) we can also have the following

relationship

2-34

whose geometrical interpretation is that the families of curves in the

x, y plane, corresponding to constant values of u and v, intersect at

right angles at all their points.

Let us now take an analytic function f which maps the domain

D of the independent complex variable w =u + i v into the domain of D'

 12

of the dependent complex variable z =x + i y. If two arbitrary curves

in the domain D intersect at the point P(u , v ) at an angle

0 0
e, then the

mapping is called isogonal if the corres.ponding curves intersect in the

domain D' also at the angle e. If the sense of the rotation as well as

the magnitude of the angle is preserved, the mapping is said to be

conformal.

To prove this statement let us take in the w plane an

arbitrary point P(w ) and two continuous curves passing through the
0

point and let tangents to these curves at the point P mak~ angles a 1 and

a 2 with the real axis of the coordinate system. Suppose we select on

each curve additional points close to P and at the distance r from P.

Then
i8
=r e l
2-35
w2 - w0 =r ei 8 2

and as r approaches zero, 81 and 82 become the angles a1 and a 2 •

w-plane

The point P(wa·) is transformed from the domain D by the

analytic function into P'(z ) of the domain D~. The projected curves are
0

denoted by ci and c2.

 13

z-plane

Then
ih
zl - z
0
= pl e
2-36
- z = p2 ei<P2
z2 0

By the definition of derivatives we have

lim zl - z
0 lim
z
2
- z
0
w -+w - w
= f' ( z0 ) and
w -+w - w
= f I ( Z
0
)' or
l 0 wl 0 2 0 w2 0

p e ih p
e i<P2
lim
1
ie 1 =k .e ia.
and lim
2
=k . e ia. 2-37
r e r e i82

It is obvious from the above relations that

and 2-38

or the angles between the corresponding curves in the conformal mapping

remain unchanged.

At the same time it is important to notice that

pl . p2
1 im -
r
= l1m -
r
= k = If' ( wo ) I 2-39

is the scale factor and has the same value in all directions through the

same point but it varies from point to point. This is an extremely

important property of conformal mapping.

 Other expressions related to the derivative of a complex

function are needed. The first of these is

dw • dw = (du + idv)(du- idv)

2 2
= du + dv

= f'(x + iy) f'(x iy)(dx + idy)(dx- idy)

= f'(z) f'(z)(dx 2 + dy 2 ) 2-4o

and realizing that

au au
f'(z) = f' (x - iy) = - - i av = av + i (ly
ax ax ay
and that
2 2
f' ( z) f' ( z) = (au) + ( av) 2-41
ax ax
2 2
= ( ay
av) + (au)
ay
2-42

2 2
dw • dw = ·[ (au) + ( av) ] ( dx2 + dy2)
ax ax
2-43

2.5 Special Complex Functions

The following identities are useful in simplifying conformal

mapping equations. The first identity is a result of De Moivre's Theorem

which gives a relationship between multiples and powers of trigonometric

functions. The theorem is briefly stated. For

Z =X + iy

=r (cos e + i sin e)

the product of two complex numbers is

z1 • z 2 = r 1 (cose 1 + i sin8 1 ) r 2 (cose 2 + i sine 2 )

= r 1 r 2 [cose 1 cose 2 - sine 1 sin8 2 + i(cose 1 sine 2 + sin8 1 cose 2 )J

= r 1 r 2 [cos(e1 + e 2 ) + i sin(e 1 +'e 2 )] 2-44

 1 ')

and an extrapolation ton complex numbers for the case r 1 =r2 - ... = rn '
and el = e2 = ... =en' yields

zn = rn[cos ne + i sin ne] = [r(cose + i sin6)]n

and finally
j cos ne + i sin ne = [cose + i sine]n .j 2-45

The second useful formulae is Euler's which relates exponential

and trigonometric functions. It is employed in the derivation of the

Lrunbert conformal conic mapping equations. Begin by considering

e 2 3 en
e =1 + e +!_+!_+
2! 3!
+-
n!
2-46

and substituting ie for e, the above becomes

2 ( ie )n
e
ie
= l + ie + (ie)
2!
+ +
n!
... 2-47

and after grouping real and imaginary parts

ie e2 e4 e6 e3 e5
e = (l--+1.7-"""T:"I
2! 4! Dl
+ ... ) + i(e - -
3,, +-,-
5.
... )
which is of course
e ie = cos e + 1
. .
s1n e 2-48

The above is known as Euler's formula. Substituting -8 for e in the above

yields
e -ie = cos 1
e - i sin e . 2-49

Together the above two formulae yield two more useful formulae, namely
( ie
cose = 21 e + e -ie) 2-50

sine = 12 i ( e ie - e -ie)
2-51

The last set of formulae are the sin and cos of complex numbers.

Substituting z for e in 2-50 and 2-51, yields

cos z = 21 (
e iz + e-iz) 2-52

sin z = 21 i (
e
iz - e -iz) 2-53

Further
iy
e
2-54
= eX ( cos y + 1' si n y) •
 l()

3. REVIEW OF DIFFERENTIAL GEOMETRY

The purpose of this section is to review and write in our

notation some well-known results of differential geometrr. No attempt

has been made to be exhaustive, in fact, only those aspects needed

herein are given. We treat: parametric equations for a surface; the

Gaussian fundamental quantities; angles between parametric curves.

3.1 Parametric Eauations

Let us nm-r discuss hmr to ;:-,athematically describe the surface

to be mapped.

The general equation describing any surface* is

F(X, Y, Z) =0 , 3-l

where

X = X(¢, A)

Y=Y(¢,A) 3-2

Z=Z(¢,A)

are known as the parametric eguations. X, Y and Z are understood to be

*X, Y, Z used as coordinates on sur:f'ace to be mapped.; x, y used as

as coordinates on map plane.
 17

Cartesian coordinates while ~ and A are regarded as curvilinear coordin-

ates on the particular surface in question.

For example, the equation of a spherical surface is

F(X, Y, Z) • X2 + Y2 + Z2 - R2 =0 3-3

The corresponding parametric equations are

X = X(~, A) = R cos~ COSA

y = Y(~, A) = R cos~ sinA 3-3a
z = Z{~) = R sincp
where R is the radius, and ~ and A the spherical latitude and longitude,

respectively.

The equation of an ellipsoid surface is

F = x 2 + y2 + z2 - N2 = 0 . 3-4
The parametric equations are

X = N coscp COSA
y = N cos~ sinA 3-4a
2
z = N(l-e ) sin~

where if> and A are the geodetic latitude and longitude respectively, and

= ______a______~~
2
N = ------~a~------------~- 3-4b
1 2
(a2 cos 2 ~ + b 2 sin 2 ~) /
1 2 2 /~
(1-e sin •)

is the radius of curvature of the ellipsoid in the ~rime vertical plane,

a and b are the semi-major and semi-minor axes, respectively, and

Parametric equations allow us to describe, in a mathematical

way, certain curves on a surface. For example when 4> is equal to some

constant c 1 , then the parametric equations become <.

X= X(c 1 , A)
y = Y(cl, A)
3-5
z = z ( c, , A)
 18

which can be imagined to describe some sort of parallel of latitude.

Analogously a meridian arc is described by

X= X ($, c 2 )

y =Y ($, c2) 3-6

Z =Z ($, c 2 )

The equation of an arbitrary curve is denoted by S($, A.), which is in

essence an expression of some functional relationship between $ and A..

3.2 Gaussian Fundamental Quantities

The Gaussian fundamental quantities are a means of describing

the geometrical properties of the surface to be mapped. The derivation

:i.s as follows.

Consider an arbitrary curve on any surface (Figure 3-1) with

P(X, Y, Z)
any surface
z

~--curve S = S(cjl, A.)

X
Flgure 3-l. 3urf~::tce Element
a point P(X, Y, Z) on it and a second point Q(X+~X, Y+~Y, Z+~Z) located

infinitesimally close to P. Let ~L be the chord length between P and Q

and 6.S the corresponding length on the surface. We can then write that

3-7

and
 19

As the point Q approaches P, ~L approaches ~S and we get

1
= (d.X) 2 (~) 2 + ( d.Z) 2
3-9
d'S + dg dS

which can be interpreted as the direction cosines of a line on the

surface.

From the above we can get an expression for the.. change in arc

length as a function of changes in the Cartesian coordinates, namely

d tf = d + d:l- + d i- 3-10

Recall that

X= X(cp, >.),

.y = Y(cp, >..) ' 3-11

Z = Z( <P, /,) ,

and thus the differentials in 3-10 are

ax ax
dX =~ d4> + ai dA ,

3-12

dz = ~
acp d"'"' + ~
:n '
d/\ •

Squaring 3-12 and substituting into 3-10 yields an expression for the

changes in arc length on a surface as a fUnction of changes in curvi-

linear coordinates, the result is

3-13
 20

where
2 2
ay az
E = (ax)
o4> + (~) + (~) ' 3-14

ax ax
F =~a>:+ ~a>:+ ~ai'
ay ay az az

ax 2 ay 2 2
G = (a>;) + (a>;) + (~)
a). .
'

E, F, and G are known as the Gaussian fundamental quantities for a

.,
surface. Their specific use is in Sections 4.2 and 4.3, which respect-

ively deal with the definition of the scale factor and condition for

conformality.

For the ellipsoid (from 3-4a and 4-14)

E = M2
F =0 3-15
2 2
G = N cos ¢

where
2
M = a(l-e )
. 2,._)3/2
(1-e 2 Sl.n "'

is the radius of curvature of the meridian of the ellipsoid. The differ-

ential arc length squared is

3-16

and a useful form is

3-17
 21

3.3 Angle Between Parametric Curves

In this section we treat the description of·the angle between

curves in terms of the Gaussie.n fund8lllentli..l !qutintiti~s::·: ' IA ':.p~ticular .

. .............

the angl~ between meridians and.parallels is described.

To do this we need the evaluation of the Cartesian total

differentials (equation 3-12) for ~ = c1 and >. = C0 (Figure

'-
3-2).

• c2

Figure 3-2. Angle Between a Merid:i.an and a Parallel

The result is tabulated in Table 3-1.

Also needed is the evaluation of the total differential of the

length of a curve S (equation 3-13} For ~ = c1 and d~ = 0,

as~ = IG) d>. 3-139.

and for >. = c2 , d>. = 0 ,

3-13b-
 22

~ = cl ). = C2
dX ax.d:\ .... a.x. d~
dA ·if·
~' ' ·:~' ,'!', ' •.

2_1 dA aY
d;Y (l). ~ d~

az az
d:Z aT dA ~del>

Table 3-1. Evaluation of Cartesian Total Differentials

The angle 6 between the meridian and parallel is the angle

between the two tangent vectors T1 and T2 (Figure 3-2), and is computed

from

where J, K, and L are direction cosines of the two vectors. The direction

cosines are (using 3-13a.;:aild; 3•-,_lb')

'

Kl = dY
dS4>
= .£I. .
dA
-r:::-Gl
Yti
3-18a
 23

dX .. ax 1
J2 = dsT = ~ • 7E '

3-18b

· az.az 1
12 = dSr = a., .• 7E .

After substitution of 3-18a and 3-18b into 3-18, we get

ax ax ay aY az az 1
cos e = [ax ~ + 33: ~ + 33: ~] - 3-19
lEG
and noting that the term in brackets is the Gaussian fundamental.quantity

F, the final result is achieved, that is

e =lEG"
F
·I 3-29

The above equation is valid for any surface (e.g. sphere,

ellipsoid, plane). It defines the angle between a meridian and a

parallel. It can be used to determine the value of an angle or to help

enforce a certain value for e in the way of a condition.

Applying it to the ellipsoid simply tells us that 6 • 90°,

since F = O, E # 0, and G # 0 (equation 3-15).

 24

4. CONFORMAL PROJECTIONS IN GENERAL

This section describes the general problem of conformal mapping

of the ellipsoid surface onto the plane. The expressions developed are

applicable to all conformal map projections alike. Discussed immediately

below are the isometric plane, scale factor definition, condition for

conformality, scale factor evaluation, geometry of projected curves,

and meridian convergence.

4.1 Isometric Plane

We are concerned with three surfaces in the subject of conformal

map projections - the ellipsoid surface which is to be mapped, the map

plane surface which is the surface where the mapping takes place, and

the isometric plane surface which is the intermediate plane surface

-
through which all,the conformal map projections are derived.

The isometric plane is mathematically described by first

considering the expression (equation 3-17) for the total differential

of the length of a curve on the ellipsoid surface.

Recall

4-1

By defining the total differential of a new quantity q as

dq =M
N sec ~ d~ 4-2

4-1 cru1 be written as

 25

4-3

By inspection one can see that the term in brackets in 4-3 resembles a
distance element squared on some sort of plane surface. This plane

surface is called the isometric plane. The two coordinate values on

"
this plane are the geodetic longitude. ;l. and isometric latitude q.

An expression for the isometric latitude q, in terms of the

geodetic latitude ~, is obtained by integrating 4-2, that is

q ~ M
f dq =! N sec ~ d~ , 4-4
0 0

where

2
M= a(l-e ) 4-5
. 2,~,)3/2
(l -e 2s1n '~'

N = ----~a~----~
2 . 2,~,)1/2
4-6
(l -e s1n 'I'

Continuing the integration we get

q
! dq
0

~ dp
= ! cos ~
4-8
0

The first integral yields

!~ dm
~
__
in [tan(r + ~
2 )]
1T "'
~ 4-9
cos~ q
0

where in denotes the natural logarithm.

 26

To evaluate the second integral let

sin ~ = p, cos ~ dcp = dp , 4-10

then

e2 f _d=p;;.._...,.. =~ 1 I 1 + ep]
2 2 2 n 1 - ep
1-e p
. : e/2
1 + e sln
= ln [ !] 4-11
1 - e sin <P

Combining the results of the first two integrals gives

TI ~ 1 + e sin : e/ 2
q = ln [tan(4 + 2)]- ln [.;::.l----=e;......=.;sl::.;:;.n;:.....L.¢J '

TI ~ (...;:.1 - e sin p e/ 2 ]
q = ln [tan(,- + .:t..
4 2 ) 1 --"--=;;..._r...;;;)
+ e sin '~'
4-12

where e 2 = (a. -b
'Thus we see that there is a value of q for each value of ~. Listed in

Ta:Jle 4-l are pairs of values for the isometric and geodetic latitudes

from¢ equals 0° to 89°. Note that the value of q is less than ¢up to

about <P = 11°, and is greater beyond about <P = 12°; At cp = 89°, q is

several times greater.

The geometric interpretation of the situation on the isometric

plane is that meridians are parallel with constant spacing, while the

spacings of the parallels vary. In fact, we witness for example, large

spa~ings between the northern or southern parallels (Figure ~~-1',.

There is the task of computing the geodetic latj +,ude <P from

the isometric latitude q. Since 4-12 cannot be inverted, that is, ¢

expressed explicitly as a function of q, some other method is needed.

We give below an iterative method of conversion of the isometric latitude

 2'7

Geodetic Latitude (~) Isometric Latitude (q)

l 0.9934132219
10 9.9851128986
11 10.9956288708
12 12.0096232035
20 20.2888725073
30 31.2726570656
4o 43.4668126053
50 57.6161578380
60 75-1262119163
70 99.0738773214
75 115.804191.6752
80 139.2112650896
85 179.0306399306
86 191.8283989049
87 208.3211451844
88 231.5595670367
89 271.2781638574

Table 1. Isometric and Geodetic Latitudes (units-degrees)

Everest Ellipsoid ClYt ='30D~80l't

·a= 6377276.345
 28

Figure 4-1.
Plot of Meridians and Parallels on the Isometric
Plane.

into geodetic latitude. The iteration method employed is that of

Newton-Raphson [ Conte and Boor 197_2; Wells- 1971].

A summary of the method follows. Given the non-linear

equation f(x) = 0, the steps for the solution are:

(1) Select an approximate solution p ;

. 0

(2)Evaluate the iterative improved solutions p from

n

p
n
= g(p
n- 1
); n = 1, 2, ••• , 4-13

where
( f(x)
gx)=x-f'(x) 4-14

( 3) Stop the iteration when I pn - pn_1 1 < e:, where e: is a

predetermined accuracy limit ( say 10-6 for single precision on the

IBM 360, or 10-12 for double precision);

 29

(4) The solution to f(x) = 0 is then x = p , and is accurate

n
to within + e::.

The application of the above to the isometric latitude q

(equation 4-12) follows. Recall

~ 1 . ~ e/2
q = ln Itan. <% + 2)( 1 ~ : :~: ~) ] 4-15

and

q =~ fln (l+sin ~) - ln(l-sin ~) + e ln(l - ~ sin ~) -

- e ln (1 + e sin ~)] . 4-16

Our function f(x) that equals zero is

f(~) =0 '

r(¢) = ~
c
[ln(l+sin ¢)- ln(l-sin ¢)+e ln(l-e sin ¢)-e ln(l+e sin ¢)1-

- q = 0 ' 1 4-17
~--------------------------------------------~-------J

and
I
f(¢) = 4-18

(1) For an initial approximation (¢ 0 ), assume a spherical case

in l1-l), that is e = 0. Then

1T ¢0
q = ln [tan(4 + ~ )] , 4-19

aml inverting the equation;

¢
0
=2 tan-l [exp (q)] TT
2
4-20

(2) Evaluate the iterative improved solutions ¢n from

¢n =g (¢n-l)' n = 1, 2~ ... 4-21

 30

where in general

g(¢) = "''I'-
f(p)
f'(¢} 4-22

namely

(3) Stop iteration when I(~~,\11

n- 1
< £.
( For £ = 10 -12 radians,

eon vergence is achieved within about three iterations for 0° < ~ < 89°).

h.? Scale Factor Definition

The scale factor describes, at each point on the map projection,

the amount of distortion in length. This distortion is of course due

to maintaining conformality and fulfilling other conditions prescribed

for the projection. It should not be confused with the "line scale"

(discussed in Section II), which is concerned with the scale distortion

over a finite length of line.

We now focus our attention on two surfaces - the ellipsoid

and the !!lap plane surfaces. It was :proved in Section 3.3 that the

meridians and :parallels on the ellipsoid surface are :perpendicular.

Since the projection is conformal, these two curves are also perpendicular

on the map plane. In mathematical language, the Gaussian fundamental

quantity F equals zero. It follows then from 3-13 that the square of

the differential change in the length of a curve on the ellipsoid

sur race j s
 31

4-24

and on the map surface*the corresponding equation is

4-25

The scale factor k is defined through the ratio

4-26

We now define the scale factor for the isometric and ellipsoid

surfaces with the aid of the following two parametric equations:

and 4-27

Note the replacement of 4> by q in thP :para:~etric eCJ.uations which

formerly related the map coordinates x and y with the ellipsoid

coordinates 4> and X.

The linear distance element squared on the map plane in terms

of q and A is (see equation 3-13)

2 2
ds 2 =e dq + 2f dq dX + qdX 2 4-28

where
2 2
e = (ax) + (b::) 4-29
aq aq

f = (ax) (ax) + (~) (2z) 4-30

aq ax aq ()A '
2 2
g (ax) + (b::) 4-31
= dA dA

* Note the use of lower case x, y for the coordinates on the map
plane.
 If angles on the ellipsoid are to be preserved on the map plane, then

f = 0, and

4-32

The scale factor is now defined through the ratio

2
ds 2 e dg2 + gd:A 2
k2 = --2 = 4-33
N cos <P(dq + d:A2) '
2 2 2
dS
1

where the denominator is the square of the differential change in the

length of a curve on the ellipsoid surface in terms of the isometric

latitude (see equation 4-3). Note the usage of the isometric latitude -

first on the mapping surface (numerator 4-33) and secondly on the

ellipsoid surface (denominator 4-33).

l~ . 3 Condition for Conformali ty

In this section we derive the equation from which the

definiticn of conformality is obtained. This condition is then rep-

resented by the Cauchy-RiemB.!ln equations, in terms of the map coor-

dinates x and y, and the isometric latitude q and geodetic longitude :>...

From Figure 4-2,

_ N cos cp
tan a - M d <P d:A , 4-34

where the newly introduced quantity a is the geodetic azimuth of the

diagonal of the differentially small figure. Recall the definition of

 33

equator
~+d::..

Figure 4-2. Different~al Area on the Ellipsoid S~rface

dq (equation 4-2)

d _ M d¢ _
q - N cos¢ 4-35

Substituting 4-35 into 4-34 yields

d>..
tan a ·-
dq 4-36

or

d/. = tana dq • 4-37

Now substituting 4-37 into the definition of the scale factor (equation 4-33),

we get an expression for the scale factor as a function of the direction,

namely
2 . 2
e· cos a + g s1n a
2 2
N cos ¢

We know, however that for conformal projections the scale factor

cannot be a function of the direction if angles are to be preserved.

 jll

Therefore, to force 4-38 to satisfy this condition, we choose e equal

to g in l1-38. Thus

k2 = _....;e:;;.__ = _ _.g..___ 4-38a

2 2 2 2
N cos ¢ N cos ¢

Now we can describe conformal projections in terms of Gaussian fundamental

quantities, namely

f= 0 4-39

and
e=g 4-40

Recall, that the first was a result of requiring the meridians and

parallels to intersect at 90° on the map plane (see Section 3.3).

~he Cauchy-Riemann equations can be derived by application of

4-3SI and 4-l,o along with equations 4-29, 4-30, and 4-31, which respectively

cld'i :we, f and g in terms of x, y, q and A. Firstly

f =0 '

4-41

and

'dx ax
2z_
- -
ag at-
en - 4-42
'}.x
aq

Secondly
e =g
2
+ (~) 4~43
aq
 35

Substituting 4-42 in 4-43, yields

( ax)2 4-44
aA +

4-45

These are only two possible cases for which the above equation is satis-

fied, namely
2
(ax) + (ll.)2 =0 4-46
aq aq
and
2
(ax) 2 = (Ez) 4-47
n aq

From 4-47, we get

~
at. = -+ .£z
Clq
. 4-48

Taking the positive root and substituting into 4-41 (f = 0), results in

ax
aq = - 2z
a>. 4-49

The above equation, along with the positive root of 4-48, are known as
the Cauchy-Riemann equations, which we label as set number one:

ax _ _£z
a>. - Clq

4-50
 36

By choosing the negative root of 4-48, we arrive at the alter-

native set of Cauchy-Riemann equations, namely

ax 2L
ai = - aq
4-51

As stated earlier these equations represent the conformality

condition and are used to _either help derive the mapping function f 1 or f2 in

x + iy = fl (A + iq) • 4-52
and

A. + iq = f2 (x + iy) , 4-53
or corroborate whether these functions, as determined by other means,

result in a conformal transformation.

L1. l+ Scale Factor Evaluation

We now can evaluate the general expression for the scale

factor {equation 4-26). Recall

2
k2 ds2 dx 2 + d y 2
4-54
= dS 2 =
l

Also, recall from complex algebra (equation 2-40) that the aistance

squared is given by

4-55
(general-note.tion). In map proje~~tion-notation and for the map plane,

the distar:ce element squared is

 37

'l'he scale factor expression, in terms of derivatives of complex

functions, is

k2- !'(~+iq) f'(~-ig) (dq 2 + d.\ 2 )

- 2 2 2 2 4-57
N cos ~ (dq + d.\ )

= f'(A+ig) f'(~-ig)
4-58
2 2
N cos ¢

'rhe above expression is simplified by recalling that (equation 2-33)

r'(:z.) r'(z) 4-59

(general-notation). In map 11ro,jection-notation

z = .\ + iq

thcre-!'(.1re

:f'(A+iq) f'(.\-iq) 4-60

4-61

Substituting 4·-60 and 4-61 into l1-58, yields the final result for the

scale factor:

k - N cos
4-62
~

;((Jx)2 + (~)2
aq aq
k ·- 4-63
N cos ¢
 38

Note the two forms. For example, the first is used to evaluate the

scale factor for the Transverse Mercator (since x = x( A.)), while the second

is used for the Lambert Conformal Conic (since x = x(q)).

4.5 Geometry of Projected Curves

At this stage in the notes, it is necessary to describe the

geometry of projected curves. Specifically, we show the configuration of

meridians, parallels, and geodesics on the map plane, and their relation-

ships to the map grid system. Also,numerous terms are defined for later

use.

Consider a map plane, as depicted in Figure 4-3, upon which two

points have been mapped (transformed) from the ellipsoid surface. The

parallel and meridian passing through point 1, along with the geodesic

from points 1 to 2, have also been mapped. Shown on the map are

tangents to the latter two curves.

Three coordinate systems are shown: the map coordinate system

(grid system) denoted by x and y; the translated map system denoted by

x' and y'; and the geodesic system denoted by ~ and 11· The latter

system is orientated with the abscissa axis tangent to the projected geodesic.

The ordinate axis of the grid system makes an angle of y with the tangent

to the projected meridian. This angle is called the meridian convergence.

The geodetic azimuth a of the projected geodesic is the clock-

wise angle from the tangent to the meridian, to the tangent to the projected

geodesic. The grid azimuth T of the projected geodesic is the clockwise

angle from the ordinate axis to the tangent to the projec~ed geodesic.
 proje~ted meridian

y to projF>cted meridian

prnjP~ted geodesic

tangent to
p~·c.j ected
geodesic

x, y- map plane coordinates (grid coordinates)

a - geodetic azimuth of projected geodesic

T - grid azimuth of projected geodesic

t - grid azimuth of chord

y -- meridian convergence

S - length of projected geodesic on the map plane

d - chord length

Figure 4--3. Geometry of Projected Curves.

 4o

The grid azimuth t is the clockwise angle from the crdinate axis to the

straight line (chord) connecting the two points. The chord length d is

the straight line distance between points l and 2. The projected length

S is the length of the geodesic on the map plane.

4.6 Meridian Convergence

In this section we derive a general expression for the meridian

..,
convergence in terms of general functions of the isometric latitude q

and geodetic longitude A.

From Figure 4-3, we define the angle.... /3 between the. x • and l;

abscissae axes (Figure 4-4). We can write

r----·· ... - ·-------· · - - - - - - - - - - - - - - - - - - - - - - - - - ,

l
I
y (parallel toy)
I 2 .
I

projected geodesic

tan /3 = .9z.
dx
4-64

The equation for the projected geodesic in terms of q and A is

f(q, A) =0 , 4-65
and the parametric equations are

x=x(q, A.),
4-66
y=y(q, A.).

The total differentials are

ax
dx = - dq + - d:A
ax
aq . en 4-67

and

dy =~ dq + %f dA . 4-68

Substituting the above in 4-64, yi~lds

'q_d
- aq q + 'EL
a:A d:A
tan S
- ax d ax 4-69
aq q + aJ: d:A

and changing its form gives

2z. + 2z. ~
tan S
_ a9 a» dg 4-70
ax +ax- dA.
- - -
aq n dq
Recall from 4-36, that
d:A
-=tan a·
dq
4-71

Substituting 4-71 into 4-70, yields

2z. av tan a
tan S
= ag + 3>:" 4-72
ax + ax tan a
aq a;..
and

EL _ax_ tan 13
aq - a9
tan a -- ..;;._;;j,_....:;_;;.___ _
'q_ ax 4-73
3A - a5: tan 13
The above two equations are rather basic and need further development

before an expression for the meridian convergence can be obtained.

Consider the case (Figure 4-5)

y =8 4-74
and
4-75
y

projected ~
parallel

pr
m

~----------------------------•X

Figure 4-5. Meridi~n Co~vergence

Rewriting Ll-72 to handle tan 90° == oo, we get

1 2::r+2J£..
tan a Clq ClA 4-76
tan 8 =
1 .£2s.. + Clx
tan a Clq ClA

Substituting 4-74 and 4-75 in 4-76, yields

4-77
tan y =

'I'he Cauchy-Riemann equations

2:;L _ dX
Clq - a>.

q_ = ·- ax
31. <lq '

allow us to write
 ax
tan y -- - .£z
ag • 4-78
aq

Both of the above expressions are used to obtain the value of the

meridian convergence for specific map projections. For example, the

latter E used for the Lambert Conformal Conic, while the former is used

in the Transverse Mercator.

 41~

5. MERCATOR PROJECTION

In about 1550, Mercator created this projection imperically as

a result of attempting to have the loxodrome on the globe appear as a

straight line on the map. About !10 years later, Wright gave the math-

ematical description of the projection.

The requirements for the Mercator projection are:

( 1) 'rhe scale is true along the equator;

(2) The origin for the ordinate y is at the equator.

In this section we treat the direct problem, the scale factor,

the meridian convergence, the appearance of the Mercator projection,

the loxodrome, and the inverse problem.

).l Direct Problem

The direct problem is the determination of the map coordinates

x andy from the geodetic coordinates ¢ and A. In other words, we are

to determine the mapping function f in

X + iy = f(A + iq) . 5-l

Notice~ is replaced by q~ as will be the case for all projections,since

the isometric plane is employed as an intermediate mapping plane.

The first requirement i~plies that

x = aA , 5-2

where a is the semi-major axis of the ellipsoid. The equator portion of

the ellipsoid surf~;tce is mapped at iir.s true size.

The Beccmd requirement impli1ts :a: particulai' correspondence

between three quantities at the equator, namely

 ¢ = q_ = y = 0 . 5-3
Under th·cse conditions our complex mapping function becomes

I x + iy = a( A + iq) . I 5-4

'T'rk above mapping function is conformal since the Cauchy-Riemann

equations are satisfied, namely

dX Ex _
"'3'I = 3q - a '

'lilll

dX
3q = - lJ.x
J>.
=0

'rhis complex function can be programmed using complex arith-

weti c routines, or explicitly separated into real and imaginary parts to

x = a>. ,
5-5
y = aq .

:iote that the units of x and y are those of a, since the units of >. and

q :u·e cadians.

).~ 0cale Factor

Recall one of the general equations (equation 4-62) for

C\''.i.lu:~'.ing the scale factor:

k = H C'OS
5-6

:·Joe p,no!· ial de:r.i. ve.ti ves an· eval.uateC. by ,·escrt ing to the para:netric

·.>qwtt ions of the Jvjercator rroj er';~ ion \ equa.t ion 5-5). The rese1J t is
 46

ax ~-
ai =a; a>. - o . 5-7

The scale factor becomes

k =_.;;;;a_ _
5-8
N cos 4>

The scale distortion is zero on the equator and increases with latitude.

At 4> = 90° the distortion is infinite which simply means the pole is

represented as a line like a parallel.

5.3 Meridian Convergence

Recall the general expression (equation 4-77) for evaluating

the meridian convergence:

~
tan y
a.>.
=a.x 5-9
a.A

For the Mercator projection

0
tan y =- = 0 '
a

and

That is, the grid ordinate axis coincides with the tangent to the

projected meridian.

5.4 Appearance of the Mercator Projection

The appearance of the Mercator projection is deduced from the

several factors listed below (see Figure 5-l).

 47

Appearance:
X = aA

Y = aq @ ~= 90 q = 00

40° ------~--~--~---+---+---4----~--+---~---+--~--~--

scale maintained along Equator

-10° ------~--~---4----~--+---~---+--~r---+----r---+--~--

-20° -------+---+----~--4----+--~~--+----+--~----4----+--~--

-30° ------~--~--~--~---+---4----~--+---+----+---+---+--

-40° ------+---+---+---+---4---4---~---r---r---+--~---+--

Figure 5-l. Appearance of the Mercator Projection

 48

(1) y = 0;
(2) k = 1 along equator;

(3) meridians and parallels intersect at 90°;

(4) meridians are equally spaced;
(5) the spacings of parallels increase as one goes away from
the equator.

5.5 The Loxodrame

Let us now show that a loxodrame or rumb•line is a straight line

on the Mercator projector. Recall that a loxodrome is a curve on the

surface of the earth (more strictly on the ellipsoid surface) that meets

the meridians at the same angle that is constant azimuth (Figure 5-2).

pole

~oxodrome

N cos~ dA

Figure 5-2. Loxodrome

From the figure

tan a = N cos p 5-10

M

and

M
dA. = tan a N sec ~ d~ • 5-11

Integrating the above, yields ·

A - A
0
= tan a q
or
A = q tan a + >.. 5-12
0

which is the basic equation for the loxodrame on the surface of ellipsoid

in terms of q. Given the mapping equations

q=Z.
a
5-13
A. = .!.a

and substituting these into 5-12, yields the equation of the loxodrome

on the map surface, namely

y tan a + a>..
0
5-14

This is nothing else but the equation of a straight line, where tan a is

the slope and a>.. is the x-intercept.

0

5.6Inverse Problem

The inverse problem is to compute ~ and A given x and y. The

steps for achieving this are:

(1) Compute the longitude ). from

~ 5-15
~
 50

(2) Compute the isometric latitude q from

Iq=~ ;I 5-16

(3) Compute the value of the geodetic latitude ~ corresponding

to q by the iterative method described in S:!ction 4.1 .

 51

6. TRANSVERSE MERCATOR PROJECTION

This very widely used conformal map projection was invented by

Johann Heinrich Lambert in 1772. His development was based on elementary

considerations (probably a geometric approach). Some 50 years later,

Gauss gave an analytical derivation for the projection, showing that it

was a special case of the conformal mapping of one surface onto another.

Then L. Kruger, in 1912, completed the development of the Transverse

Mercator projection by developing the formulae further in order that they

would be suitable for numerical calculations (similar to those of 6-25

and 6-26). What a beautiful example of creating- formalizing-

implementing, all three processes taking over a century.

The requirements for the Transverse Mercator projection are:

( l) 'l'he scale is true along the central meridian;

( 2) The origin of the ordinate y is at the equator;

( 3) The origin of the abscissa X is at central meridian.

Treated in this section are: the direct problem; the scale

factor in terms of geodetic coordinates; the meridian convergence in

terms of geodetic coordinates; the appearance of the Transverse Mercator

projection; the inverse problem; the scale factor in terms of map

coordinates; the meridian convergence in terms of map coordinates; the

Universal Transverse Mercator Projection; and the Transverse Mercator

in three degree zones.

6.1 Direct Problem

Given the general complex mapping function
X + iy = f(A + iq) , ( 6-1)
the problem is to determine the specific mapping function f which fulfills

the requirements stated above.

 52

The third requirement implies that for the central meridian

X =. A = 0
. . 6-2
thus for the central meridian (from equation 6-1)

iy = f(iq)

y = f(q) 6-3
On the other hand, the first requirement implies that the ordinate value

y on the projection be equivalent to the length of the meridian Sci> on the

ellipsoid (Figure 6-1).

ellipsoid surface

Figure 6-1. Lengtit of Meridian on Ellipsoid

The length of the meridian, from the equator to the point in question, is

given by the integral

4>
y = s 4> = ! Mdcj> 6-4
0

Appendix I contains the solution of this integral.

 53

Further,

«P
y = ! N cos «jl dq , 6-5
0

y = f(q) '

since, through the definition of the isometric latitude,

Md«jl = N cos «jl dq • 6-6

We thus have been able to determine the function f in 6-1~ but only for

points on the meridian.

The next step is to generalize the approach for points away

from the central meridian, but still close - say within 3° or so. For

these points, the abscissa~ value will· be non-zero and the ordiute will

not be equal.to the length of the .Dlet";l:dianfrom the. equator to the point.
To help solve this problem, we approximate the mapping function

X + iy = f(A + iq)

by a Taylor series, choosing some arbitrary point with coordinates A

0

and q as the point of expansion (Figure 6-2). First,

0

Central
meridian ~point in question

I
lllq
. I
po~nt of 1
expansion e---_J
( "o' qo)

equator

Figure 6-2. Arbitrary Point of Expansion

 54

X + iy = f_(A. + iq)

= flA +A>. + i(q +l.\q)]

o· 0

= fl(A 0 +iq0 ) + ( .l.\A+il.\q)]

= f(z 0 +l.\z) = f(z) . 6-7

Then, approximating the complex function f( z) with a Taylor series

r1~z. )(l1z) 2 fti~z )(l1z) 3
f ( z) = f ( z 0) + ri ( z 0) l.\z + 2~ + __3_!.;:;o_ _ _ +. 6: 8
or

x+iy = r ( A0 +iq0 ) + f I( A +iq )( l1A+il.\q

, ) + fii('A +i q ){_AA+iAq) 2 6-9
. 0 0 . 0 0 2!

The above equations are made practical by choosing a different point of

expansion. This time the point is chosen to have coordinates q of the

point in question and to be on the central meridian (Figure 6-3). This

implies

Central
meridian

( 0, q) (A' q)
~point of expansion •
point in question

----~------------------------------~equator

Figure 6-3. Transverse Mercator Point of Expansion

 55

). 0 = 0. '
6-10
D.q = 0 '

and
n
':1.
= q0 + 6q = q0 '
6-11
A =A 0
+ 6A = D-A .
•

Substituting 6-11 into 6- .9, yields

x+iy = f(iq) + ri(iq)A + -2-1- - + --3-~--··+. · ·

6-12

To evaluate the function and its derivatives~·we-retlift~~el:.1tJae fact

that the length of the meridian is (6-5)

~
s~ = r(q) = ! N cos ~ dq • 6-13
0

The function itself is

cp
f(iq) = if(q) = i f N coscj> dq • 6-14
0

The first derivative is

= df{z) = df( z) . .s!q - idf( 9) • .s!q

6-15
dz dq dz - · dq dz

Since

z = iq 6-16

and
 ~::::1:..=-i 6-17
dz i

6-18

6-19

The second derivative is

= .9:_ {df(q)} ~
dq dq dz

6-20

The higher derivatives, are derived in an c.nalogous manner,

that is

fiii(.lCJ ) :::: - fiii (g)

'
rrv (iq) = if'IV ( q) 6-21
'
rv (iq) ::: rv ( q) '

Substituting the evaluation of the derivatives into 6-12,

y:i.elds
,________________________..., ,

x+iy = f(~+iq) = f(iq) + f'(q)~ +

6-22

--------------------------~
 57

All that we have achieved up to this stage is to replace the complex

derivatives with real ones (compare 6-22 with 6-12).

After separating into real and imaginary parts

X = + ••• '

6-22a

x and y can be solved for in the above after specifying the values of A

(longitude positive east of' point re·lative to· the centraJ:\~ridi~~ i.~·.

~ = ACM - A • t) and q (computed from <fl). · The only problem left- to be

p01n
solved is the evaluation of·f(iq), f 1 (iq), etc.

We have to evaluate the real derivatives still further to get

an expression that is suitable for computations.

The first derivative is (from 6-5)

dS Jf( q) }
fl(q) =~
dq
= dq
= N cos <fl •

The second derivative is

~ dN
f .. tq) = [dcfl cos cfl - N sin cj>] !!t
dq
t

where, from the definition of the isometric latitude,

!!t
dq
= !M cos "'"' '
 58

dN
dcj> = (N-M) tan <I> , 6-23

so
f"(q) = -N2 sin 241 • 6-24

The higher derivatives are computed in Thomas [1952]. Substituting these

into 6-22a yields

-NX =

+ 4n 6 - 64n 4t 2 - 24n 6t 2 )

+ ~ 7 cos 7 <1> {61- 479t 2 + 179t 4 - t 6 ) 6-25

5040

i = ~ + ~ 2 sin <1> cos <1> +~sin <1> cos 3 cj> {5 - t 2 + 9n 2 + 4n 4 )

A6 . 5 2 4 2 2 2
+ 720 s1n <1> cos cj>(6l - 58t + t + 270n - 330t n

+ 445n 4 + 324n 6 - 68on 4t 2 + 88n 8 - 6oon 6t 2 - 192n 8 t 2 )

AB • 7 2 4 6
+ 40320 s1n <1> cos cj>(l385 - 3llt + 543t - t ) , 6-26

where the longitude A is expressed in radians, and

<I>
s 41 = l Mdljl , (from Appendix I)

t = tan <1> , 6-27

2 2 2
n = (e') cos <1> 6-28
2 2
(e')2=a -b 6-29
b2

The above mapping equations yield x andy values accurate to 0.001 metres

for A=± 3°. This completes the direct problem.

 59

6.2 Meridian Convergence in Terms of Geodetic Coordinates

Recall the general equation (equation 4-J7) defining the

meridian convergence

a
a..>.
tan y = - - 6-30
a.x
;n

If we differentiate equations 6-25 and 6-26 with respect to A, and

substitute into 6-30, we get {Thomas 1952, p. 97]

4 4
+ A cos P (2 + 4t 2 + 2t 4 + 15n 2 + 35n 4 -
15

6-31

This expression is developed by a series expansion of the arc tan, that is

-1
y =tan {R.H.S.} in the above. The result is [U.T.M., 1958]

2 2 . 4 4
y =l . t [ 1 + A cos
s1n ( ") ~ ( 1 + 3n 2 + 2n 4) + l cos p (2- t 2)] , 6-32
3 4
p 15(p")
where y and A. are in radians; p" = cosec 1". The above expression has

been truncated and can be applied out to 3° from the central meridian

with an expected accuracy of O'! 01 in y. We note that the meridian

convergence increases as we go away from the central meridian.

 60

6.3 Scale Factor in Terms of Geodetic C~ordinates

Recall the general expression for evaluating the scale factor

on a conformal projection (equation 4-62)

/c.( Clx)2 . (2z)2

()),. + aA
k = --~------~--
N cos cjl

We first change the form of the above to be a function of tan y by

squaring 6-30, namely

2
tan y =

and

6-34

Substituting 6-34 into 6-33, yields

~~ h + tan2 y
k = ~-----------
N cos cjl 6-35

Since tan 2 y is small, we can expand the square root term into a series,

and after evaluating the partial derivative, the result is

6-36
6 6
+ A cos p (61 - 148t 2 + 16t 4 ) •
720

A somewhat truncated expression is

 61

6-37

where the eighth significant digl.t may be in error by one or two units.

Note the scale factor increases as one's longitude relative to the

central meridian increases;; for a constant latitude. Along the same

meridian an increase in latitude causes a decrease in the scale factor.

6. 4 Appearance of the Transverse Mercator Pro,jection

The following facts govern the appearance of the Transverse

Central
merid.i. an

Pole

equator

Figure ~-4. Appearf.l.nce uf the 'I'rsnsverse Mercator

 62

(1) k =1 along the central meridian, thus the two poles

are plotted at ordinate values equivalent to the length of the meridian

between the pole and the equator;

(2) for points off the central meridian the ordinate value y'

is greater than the ordinate value y, a point on the same parallel of

latitude but on the central meridian;

(3) y increases with longitude away from central meridian;

(4) meridians and parallels intersect at 90°.

6.5 Inverse Problem

The direct problem was stated in equation form as

X + iy = f(A + iq) 6-38

We write the inverse problem analogously as

A + iq = F(x + iy) 6-39

where F is the ampping function to be determined and is different from

f o f the direct problem.

We begin by approximating F(x + iy) in 6-39 by a Taylor series.

rhe point of expansion will, at first, be chosen to be some ·~bitrary

point on the map plane with coordinates x and y (Figure 6-5).

0 0
 63

y
parallel of latitude
Transverse Mercator
Point of expansion c
(x, y) point in question
( 0' y) -------------- I ~
I
I
I
(xo, yo) I
----------'
I
arbitrary point of expansion

--t-------------------------------------------~~ X

Figure 6-5. Arbitrary and Transverse Mercator Points of

Expansion (Inverse Problem)

Thus

= F(x 0 +iy0 ) +F 1 (x +iy )(~x+i~y) +

0 0

F1 fx +iy )
+ __;::.0_ _..;::.0_
2!
(~+i~y) 2 + •.• 6-40
... ,.

I FEl(z )(~zj 2 Jil:1 1(z )(~z) 3

== F(z ) + F ( z 0 )(~z) + ~! + --....;~;;..1--'-·- + •••
0

6-41
Now we choose a more practical point of expansion for our Taylor series -

this time one with coordinates (o, y) (Figure 6-5).

x
0
== 0 implies x = x
0
+ ~ =~ and (A -· ~A) ; 6-h2

y0 =y implies ~y =0 and (~q = 0) • 6-43

'l'aking into account the above implications in 6-l~O, we get
 64

'+' = F(. ) + FI(. )x + iU{iy)x2 + Jil!IIl{iy)x3 + ••• 6-44

A. ~q · ~y ~y 2!. 31

Note that on the central meridian

X= 0

iq = F{iy) • 6-45
The geometrical situation associated with the above equation

is depicted in Figure 6-·6. Shown are two parallels of latitude - the

parallel through the point in question with latitude 4>. , ~d the parallel

through the ordinate value y on the central meridian. The latitude of

the latter parallel is 4>1 .. -~d!~is~_c.~!~~?-~~e:-,:_f~t?~~~l,iliatittide"l'.~n: :-... >~

Thomas [1952]. Corresponding to 4> 1 is the footpoint isometric latitude

ql · Accordingly, 6-45 is written as

iq1 = F(iy) . 6-46

As a first step in making the mapping equations (6-44) practical

for numerical computations, we evaluate F(iy) and its derivatives, leav-

ing off the subscript l for the time being.

central
meridian through y of point in
y

Foot point Parallel of point in question

latitude
4>1
I
latitude----- I
ly
~ s~ {..__-- I
I
I
I
I
___, __ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ L_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _~

X X

Figure 6-6. Footpoint Latitude.

 b)

The first derivative is

FI(i )
y
= dF(iy)
d ( iy)
= d(ig)
d ( iy)
=• l. ~
dn •
dSq
6-47
· dSq d( iy) '

where in general

iy = iS4> , 6-48

and in particular

6-49

Then the last derivative in the above becomes

dS
q 1
=- 6-50
i
i dS
q
If we define

~=Iq 6-51
dS
q
then the result is

Fr(iy) = qr ·I 6-52
The second derivative is

dS<). Ill
=q i 6-53
d(iS )
q

6-54

The higher derivatives can be derived in ari analogous fashion,

they are:
III
FIII( iy) - - q
FIV (iy) = iqIV 6-55
Fv (iy) = qv

(sequence repeats)
 66

Replacing the subscripts for the footpoint latitude in the

function and its derivatives, 6-44 becomes

.II2 III 3 i IV 4
~ql X ql X ql X
21 3! + 4! + ••• , 6-56

and separating into real and imaginary parts the above becomes

+ .. • • , 6-57

+ ••. • • 6-58

Note the last equation in the above gives the difference, 6q = q-q1 ,

between the two isometric latitudes. This is used below in deriving the

final expression for the geodetic latitude.

Derivatives of complex functions in 6-44 have been replaced by

real derivatives as shown above. Now we evaluate the real derivatives.

Recall

S
q
= J N cos~ dq • 6-59
0

The first derivative is

q I _ .!!g_ _ _1____
- dS - N cos<jl 6-60
q

The second derivative is

II d2 t
q = ~=
2 2
6-61
dS N cos~
q

The t.hird derivative is

d3 1
q
III
= .9:......9.. = (1+2t 2+n 2 ) . 6-62
dS 3 N3 cos<jl
q
 6(

The higher derivatives have been worked out in Thomas

[1952, p. 101].

Substituting the derivatives in 6-57 and 6-58 along with the

above expressions results in the following expression for the longitude

A:

A = sec

6-63

where the subscript 1 denotes that the functions be evaluated using the

footpoint latitude ~l (Appendix II), and·t 1 , n1 are given by 6-2'7 and 6-29.
The accuracy of the above formula is plus or minus 0.00001 arcsecond for

A less than 3° from the central meridian.

The derivation of the corresponding expression for the latitude

~ is more involved. It is necessary to obtain an expression for ~~ = ~-~ 1

as a function of the difference ~q = q-q1 (6-58). Then the solution for

the latitude is
6-64
where ~l is the footpoint latitude corresponding to a meridian of length

y. The determination of ~l given y is documented in Appendix II.

We now expand ~~ by a Taylor series in ~q,

2 2 3
d~l ~ql d ~1 ~ 3 d ~1
~·- cp
1
= g(~q) =~q--+----+~-+
dql 2! d 2 3! d 3
... 6-65
ql ql

Noting that, from the definition of the isometric latitude,

 68

6-66

The second derivative is

2 N
cos cpl ( 2 - 3 ~)

The higher derivatives are given in Thomas [1952, p. 102]. Substituting

the derivatives of q (6-60, 6-61, 6-62) into 6-58, and the result for Aq

into 6-65 along with the derivatives of ¢1 (6-66, 6-67) into 6-65, the

final equation for determining the latitude is obtained:

t x8
+ 1 7 (1385 + 3633t~ + 4095t~ + 1575t~) 6-68
40320~N 1

where
t1 = tan cp 1

2
n1 = (e') 2 ,~,
cos '1'1
2

(e')2 = (a2- b2)/b2

The accuracy of the formulae is ± 0~00001 within 3° of the central

meridian.
 69

6.6 Meridian Convergence in Terms of Map Coordinates

In the inverse problem the map coordinates x and y are given.

The aim of this section is to express the meridian convergence as a

function of x only.

From equations 4-77 and 4-78,

!x. ax
tan y a" ·-:~ .'!9..._ .9x. 6-69
= ax = 1··'2z. = dx
at.. aq
for q = constant. Knowing that in general

q =q (x, y)

and that the total differential for the case when q is constant is

dq = !9. dx + .£.9... dy
ax ay
=0 ' 6-70

and

~=-!9.;!9.
dx ax - ay
Thus
h h
tan y =~ =
ax ax
- ----= - aA. 6-71
dx .£.9... '
ay ax
since from the Cauchy-Rieman equations (equation 4-50)

h-~ 6-72
ay - ax

The derivatives of 6-71 are obtained by differentiating 6-57

(A.= A.(x)) and 6-58 (q = q(x)). The result is

 70

.!9...
ax
II x 3 IV x 5 VI
xq1 -6q1 +-q
120 1
- ... 6-73
tan y = - - - = I 2 III ·x4 v
aA
ox ql- 2
x
ql + 24 ql - ...
Substituting the expressions for the partial derivatives, from 6-60

through 6-62, into the above, yields [Thomas, 1950, p. 104]

(~ )3
tl tl 4
tan y = - x -- (1 - nl2 - 2n1 ) +
Nl 3 1

t 5
+ _1:. (.!-) (2 + 2n 12 + 9n 14 + 6t 1
2 2 +
n1
6
20n 1 + 6-74
15 N1

y may be computed knowing x and the footpoint latitude ~l corresponding

to y.

6.7 Scale Factor in Terms of Map Coordinates

It is often convenient to compute the scale factor as a

function of the map coordinates. You will witness this is Section II

when we deal with computations on a conformal map plane.

From equation 6-35, the reciprocal of the scale factor is

1 oA 2 l/ 2
k= N cos ~ ax I (l+tan y) • 6-75

After computing the part~al derivative from 6-57 (A = A(x)), expressing

the square root term in a series, and expanding N cos ~ in a Taylor

series with the point of expansion being the footpoint latitude ~1 , the

equation for the reciprocal of the scale factor becomes [Thomas 1952, p. 105]
 71

2 4 6 2 4 2 2
2 5 + 6n - 3n 1 - 4n + 24t 1 n1 + 24t 1 n1 4 6
1+111 2 (!.....) + 1
-kl =l --- 1 (!.....) 61 (X )
2 Nl 24 Nl - 72o 'N 1
6-76
The reciprocal of both sides yields the expression for the scale factor,

namely

1 +

6-77

6.8 Universal Transverse Mercator (UTM)

The Universal Transverse Mercator projection is based completely

on the transverse.mercator projection. In this section we give the

specifications for the UTM, and the equations for the direct and inverse

problems.

Its specifications are as follows [UTM 1958]:

(l) transverse mercator in zones 6° wide;

(2) reference ellipsoid - Clark 1866 in North America, with other

ellipsoids used elsewhere;

(3) longitude origin - the central meridian;

(4) latitude origin - the equator;

(5) unit -metre;

(6) false northing (y): 0 metres for northern hemisphere; 10,000,000

metres for southern hemisphere;

(7) false easting (x): 500,000 metres;

(8) scale factor at the central meridian: 0.9996;

(9) zone numbering: beginning with 1 for zone between l8v 0 W and 174° W

meridians and increasing to 60 for the zone bounded by meridians

174°E and 180° E (Figure 6-7).

 72

Zone Zone Zone

l 2 60
A
v v equator

Figure 6-7. UTM Zones

(See Figure 6-8 for the zones for North America).

(10) The latitude limits for the system are 80°N and 80°S due to the many

zones·' that would he involved when v.:;l'k.ing :within an area ef any

appreciable extent.

Scale Factor on the UTM

By choosing a scale factor of 0.9996 on the central meridian

we expect from equation 6-37

). 2 2
k = l +~cos~+ ••• ' 6-78

that the scale factor increases as we go away from the central meridian,

reaches a value of l, and then increases beyond a value of l (Figure 6-9).

li

73

A R c I c
0 c E

Yigure 6-8. UTM

~~nes rur North America

c F 1

E A
 74
---------- Central meridian

= 0.9996

220 km

k =1 k =1

Isoscale curve

Figure 6-9. Scale Factor on the UTM

At what distance away from the central meridian, say along

the equator, is the scale factor equal to 1? To answer this question

we first write 6-78 as

2
k = k0 [1 + A2 cos
2
p]
'

where k 0 is the scale factor on the central meridian (for the UTM

k0 = 0.9996). If 6k is the scale difference between that at the

central meridian k 0 &hd k, then

 '75

we get

6k 1/2
I k- J 6-80
cos 4> 0

For our case, Ak =1 - 0.999 6 = 1 I 2500 = 4 x 10-4 • Therefore at

Along the equator, at a distance of about 220,000 metres, the

scale factor is equal to unity.

Mapping Equations for the UTM

Let us trace through the equations of the transverse mercator

and see how they are affected by the stipulation of k = 0.9996 on the
0

central meridian. First,the meridian distance is

4> 4>
y = k 0 s <P = k 0 JM •.. dcf> = k 0 /N coscj> dq = k 0 f(q).6-81
0 0

We see that the function f(q) is modified by k , and thus all derivatives
0

will be also modified, namely

k 6-82
0
 76

Then the mapping equations for the direct problem for the UTM in terms

of the Trans'i'erse Marcator equation is

X X
{ } =k 0
{ } 6-83
y UTM y TM

The meridian convergence for the. UTM is the same as the TM since in

_ 2z
t any-en I ax 6-84
at..

the k 0 cancels. Therefore

6-85

The scale factor for the UTM becomes

6-86

For the inverse problem we wish to use the TM expressions but with
xUTM
UTM coordinates. This means ~!1 = k , so in the expression for A we
0
write
3
~M I ~ III v
A=--q ---q ql + ••• 6-87
k0 1 31 k3 1
0

Thus we see that the derivatives need appropriate division when UTM

coordinates are used in the expression. The same holds for the equation

for the latitude.

Tables have been prepared for the computation of the direct

and inverse problems. The interested reader is referred to [UTM 1957].

 77

6.9 Transverse Mercator in Three Degree Zones

We have seen that the U.T.M. is the transverse mercator

projection in 6° zones. In Canada the transverse mercator in 3° zones

is also being used. For example, in Ontario [Dept. of Highway 1968]

Canadian numbered zones 8 through 17 are used as a basis for the Ontario

Coordinate System (Figure 6-10). In fact, the boundaries of these 3°

zones are slightly modified (see Figure 6-11).

Since the zone width is chosen to be only 3°, and a scale

factor of k
0
= 0.9999 assigned to the central meridian, we can expect

a smaller scale error throughout the zone. In fact, the error is

l/10,000 on the central meridian as compared to 1/2,500 for the UTM.

We can compute the distance away from the central meridian where the

scale factor is l. This can be done by simply employing equation 6-80,

where, in this case, ~k = 1-0.9999 = l/10,000. Therefore at

cp = 0° , A = 0° . 81,

cp = 50°, A =l 0 • 26.

At :>. = l 0 30 ', and cp = 42?712, k = l. 000085. (see figure 6-12).

The transverse mercator equations can be employed here as well

by simply using the scale faetor k = 0.'9999.

0
 78

PROJECTED
M. T.l\1. ZONES

3" WIDE

\ t

I
\
I \
~--~--~-------~--+----r--
4(1"

8
1 I I 10
ll
79

ONTARIO
COORDINATE SYSTEM.

3' M.T.M. ZONES

&
CENTRAL MERIDIANS
 CROSS SECTION OF A 3" ZONE
I
I
I

Zl
0
co <'

t:l..-
~~ ·
Approx.
20 Miles
~,-

1
Approx. 56 Miles ~~~ I
I
Approx. 56 Miles • '4
:
Approx. ___J E-
20 Miles li
:31 j_ _,/J
..
~ r·~------
I "30'
:~-
I" 30'
l~
~~
<•I
2\ a:;
E-
z
I
~
u I
Figure t-6:;1~
I
 81

7. LAMBERT CONFORMAL CONIC PROJECTION

Lambert developed his conformal conic preojection in 1772 -

the same year in which he created the Transverse Mercator projection.

The Conformal Conic is used worldwide.

The requirements of this projection are:

(1) parallels are to be parts of concentric circles;

(2) meridians are to be radii of concentric circles.

Treated in this section are: the direct problem; the scale

factor; the meridian convergence; the one and two standard parallel

cases; and the inverse problem.

7.1 Direct Problem

As before we state the direct problem as

X + iy = f(A + iq) ,

where the complex mapping function f is to be determined. f is

determined by first separating into two functions r1 and f 2 • These two

functions are determined by the help of the Cauchy-Rieman equations.

The following equation of a circle is a mathematical char-

acterization of the first requirement for circular parallels:

 82

.o ( q )
X2 + y2 = •l\.-fl 7-1

where K is the constant for the radius and is to be solved for, ·f1 (q} is

a function of the isometric latitude q and is to be determined, and x and

y are the map coordinateScorresponding to the coordinate system shown

in Figure 7 .1.

y ar~ straight lines

paralJe1s are circles

central
Pole x meridian

Figure 7-l. Parallels and Meridians on the LCV

The second requirement for straight line meridians is given by

y = f 2 0.)x + 0. 7-2
where f 2 (~) is the second function to be determined.

Substituting 7-2 into 7-1 yields

X 2 + f 2 2 ( ). ) X2 =r f 1 ( q) •

From the above, the following parametric equations can be written:

x = x(A,q)
K[fl (q) ]1/2
X = 7-3
[1 + f 2p,) ]l/2 ,
2
and

7-4

We now employ theCauc~Ri~man. equations to help us solve for the t 1 (q)

and f 2 (;\).

Recall from equation 4-50

ax =·.£.l. . ax
aq aA ' aA
= 2z.
aq 7-5

The above partial derivatives of 7-3 and 7-4 are (note variable is left

off function from this point forward):

K fi
ax _....;1=---=--------,,..... 7-6
aq = 2(r1)1/2(1+f22)1/2
-K(f ) 1 / 2 f f I
dX 1 2 2
~= (1 + f 2)3/2
2

2z. - 1. 7-8
aq - 2

7-9

From the first set of Cauchy-Rieman equations we get

f I 2f I
1 2
-- = __..;;::;...__ = 2i ' 7-10
f1 1 + f 2
2
where i is a new constant to be determined. So from 7-10, we can write

two equations:
 84

fr
l
fl = -2 R. 7-ll

fr
2
-=-- = .R.. 7-12
l+f 2
2

We shall see later that the constant i is associated with the longitude

and is a function of the latitude ~.

From the second set of Cauchy-Rieman equations we get
f I f I
2 1 l
:::: 2 f = ... .f,.; 7-13
l

where i is the same constant as for the first set. The two equations

arising from the above are:

f I
-1- = -2R., 7-14

f I
2
::;: 2. 7-15

Up to this point we have determined two equations from which we shall

be able to compute expressions for f 1 (q) and f 2 (t..) in terms of the constant

K and £.

'ro begin with, let us determim f 1 (q) from 7-14 by integration,

namely

= -2tdq 7-16

dfl
! - - = -211,! dq' 7-17
fl
J.l'l f 1 + c2 = -2Y.q + c1 , 7-18
 NJ. f 1 = -2tq + c3 , 7-19

where c3 is the combined constant of integration. c3 is evaluated by

considering that

.l.n f1 = 0, 7-20

when q = 0, 7-21

thus c3 =0 . 7-22

We have then

ln f1 = -2R.q, 7-23

f
1
= e -21q 7-24

The second function f 2 (A) is determined from 7-15 by integration,

namely
1
f df2 = It~ 7-25
1 + (f )2
2

-1
tan f2 = lA ·. , 7-26

f2 = tan (H). 7-27

Introducing the expressions for f 1 and f 2 into the parametric

equations given by ·}3 and 7-4, yields

X= K (e-2R.q)1/2
7-28
[l + tan 2 (1A)] 1 /~

x = K e-iq cosR.A 7-29

y = K e-iq sinR.A 7-30
From the above mapping equations, we can compute the radius of

any parallel from:

 86

2
r = X2 + y
2
'
r
2
= K2 e -2R.q
. 2 2
(cos R.A + sin R.>..) '

lr = K e-
R. q '
I 7-31

The mapping equations can be now rewritten as

x = r cosR.A. 7-32

y =r sinR..>.. 7-33

where r and £A are shown in Figure 7-2.

y (east

meridian

central
---l.,....,..meridian
X
Pole

Figure 7-2. Polar Coordinates for LCC

Figure 7-2. Polar Coordinates for LCC

The complex mapping function f is finally determined by com-

bining 7-29 and 7-30. He get

X + iy = f(A + iq)

= Ke-lq (cosR.A. + i sin R.A.) 7-34

and by using Euler's equation ( 2-l~O) ,

1\: -lq e iQ.A
+ iy -- .e
X

R. ( -q+iA)
X + iy = Ke

[ X. + iy = ..v e H(A.+iq) . 7-35

 87

Note the above complex mapping equation ~s in closed form. This expression

can be programmed using complex arithmetic once the constan~K andi are

defined. K and 1 are defined below for the one and two standard parallel

cases.

7.2 Scale Factor

The general expression for the scale factor is (from equation 4-63).

For the LCC

ax -iq
aq =- K£ . e cos £/.. 7-37

7-38
and
Kle-tq
k = N cos <I> 7-39

The scale factor can be computed for any point on the projection

with known geodetic coordinates, again, once the constants K and :t defined.

7.3 Meridian Convergence

One of the gerer~ expre~sions for the meridian convergence

derived earlier (equation 4-77) was

tan y 7-40

F·or the LCC

 88

~-
dA - R. r sin R. A' 7-41

ax = ..i r cos R. A' 7-42

dA

thus

tan y = tan R. A

y = R. A 7-43

7.4 Two Standard Parallel LCC

In this section we develop the mapping equations for the two

standard parallel LCC.

If we stipulate the scale factor to be equal to one along two

parallels, then we can write (using 7-39)

K £ e-~1 K£. e--~2

N1 cos~ 1 = N~o~ 2
= .R. 7-44

e -.(ql Nl cos~ 1
= 7-45
e-lq2
N2 cos~~
where the subscript quantities pertain to the two parallels.

Taking natural log of each side of the above equation yields

which defines the constant t:

ln N1 - ln N2 + ln cos~ 1 - ln cos~ 2
R. = 7-47
q2 - ql
 From 7-44, the following expnssion·defines the constant K:

N1 cos~ 1 N2 cos~ 2
K = = --=---.::::.. 7-48
~.e-_R.ql R.e--R.q2
...
The above two expressions for the constants K and ~ complete

the description of the direct problem - recall the complex mapping equation

(7-35)
X + iy =K ei.R,( :\+iq)
7-49

7.5 One Standard Parallel LCC

The one standard paralled LCC has as its basis the stipulation

that the scale factor be equal to one along only one parallel of latitude

¢0 . We can then write

k =l = 7-50

-R.q
N0 cos~
0
=R. Ke o 7-51

But from 7-31, the radius of any parallel is given by

r = K e-iq , 7-52
so we can write an expression for the radius of the parallel with latitude

¢ , namely
0

r
0
= K e -R.qo . 7-53
'(-51 then becomes
N0 cos¢ 0 = £r 0 . 7-54
 90

This special radius also has a geometric interpretation as illustrated in

Figure 7-3.

Figure T-3: Geometric Interpretation of Radius of Standard Parallel

Another expression for standard radius is

r
0
= N0 cot ¢
0
• 7-55

'I'he constant 1 can be determined by substituting 7-55 into 7-51+,

N0 cos¢ 0 = £ N0 cotcp 0

< = sin~ 0 j 7-56

The expression for the constant K becomes (from 7-31)

I K = N0 cotcp 0 eqosincpo I 7-57

'I'he above two expressions along with the complex mapping f'untion

[X + iy = K e"- ~ +iq) •I 7-58

constitute the equations for the direct problem for the LCC one standard

parallel projection.
 91

7 ·6 Modification to Coordinate System

The coordinate system exhibited in Figure 7~2 is not convenient

in practise. We make two modifications: interchange the x andy, and

change the location of the origin to decrease the magnitude of one of the

coordinate values.

We had from 7-32 and 7-33

x =r cos R. A 7-59
y =r sin R. A

Now we interchange the coordinates which results in

x =r sin 1 A

y =r cos R. A 7-60
The equations which reflect a change in the origin are

(Figure 7-4)

x =r sin R. A. ,
7-61
y = r0 - r cos R. A.,

where r is usually selected such that it is slightly larger than the

0

I
I
I origin parallel - close to area
I
I
r I
0 I
/
/
I /
_..........,.
--
I ,"'

Figure 7-4: Modification to Original LCC Coordinate System

 92

Two ways in which to select r is to select some cj> below the area
0 0

in question and compute

r (q
0
from cj>
0
), 7-62
0

or

r 0 = N0 cot<P
. 0
7-63

7.7 Inverse Problem

In the inverse problem the map coordinates x a~d y are given along

with the constants r 0 ,t., and. K. Required are the geodetic coordinates

<P and A.

r
Ci

Figure 7-5. Inve!"8e ~lem

From Figure 7-5, we can get the ·expression for the longitude by writing

X
tane = r -y
7-64
0

and
e =u 7-65

~ 7-66
~
 93

The latitude is obtained in the following steps. First

compute r from
r - y
r = ......;;.o_ __
7-67
cos e
We know that
r = K e-R.q 7-68
so

ln r = ln K -5/, .q 7-69
and thus
q = ln K - ln r 7-70
5/,

Once the isometric latitude is determined,~ may be determined by using

the Newton - Raphson approach described earlier.

 8. STEREOGRAPHIC PROJECTION

In this section we first derive the expressions for polar

stereographic projection and its implimented system - the Universal

Stereographic Projection. Secondly~ we give references pertaining to

the stereographic projection as implemented in New Brunswick.

8.1 Polar Stereographic Projection

The polar stereographic projection is a limiting case of the

LCC. Meridians are straight lines and parallels are concentric circles.

In this section we deduce expressions for the direct problem~ scale

factor, meridian convergence, and inverse problem from the expressions

already derived for the LCC.

The constant t for the one standard parallel is (7-56)

t = sin ~0 8-1

For the pole, ~

0
= 90, and t =1 The expression for the scale

factor then becomes (7-39)

.k = 8-2

and

K =k N cos ~ eq 8-3
We have seen that q increases without limit as ~ approaches
TI
2 , so the above expression breaks down at the pole. We transform the

expression as follows:
 95

K =k N cos ~ eq =k N cos ~ tan (~ + ~) ( 1 - e sin f) e/2

4 2 1 + e sin + 8-4
We make the trigonometric substitution
!) p
'IT
·tan ( 4 + 2 = -1cos
;;.:-;.=_s~"-·n-~ 8-5

2
K = k N cos p (1 -e sin p )e/2
8-6
1 -· sin ~ 1 + e sin ~

=k a (1 + sin p) (1 - e sin p)e/2 8-7

. 2 ~)1/2
(1 - e 2 s~n 1 + e sin.~
"'
At. the pole ~ = 90° and k = 1, so

K = 8-8

8-9

The expression for the scale factor is

k = 2a 2 (1-e)e/2 e -q 8-10
b N cos ¢ l+e

The meridian convergence is

y = R,).. = ).. 8-11

The compleS mapping equation for the direct problem is

I x + ~y
- = lt e ii.(~+iq) ;I 8-12

where t and IC are definei above. The inverse problem is solved a.s··ror the

LCC.

The Polar stereographic system is applied in practice as the

universal stereographic projection (U.P.S .). It has the following char-

acteristics [UPS 19 58 ] :

1) units - metres

2) ellipsoid - international;
 3) pole coordinates: x = 2,000,000; y = 2,000,000;
4) kpole = 0.994 .
The UPS was designed to be used in conjunction with the UTM to

cover the entire globe: UTM to B0° latitude and USP from B0° to 90°.

B.2 The Stereographic Projection System in New Brunswick

The above is not a complete treatment of stereographic pro-

jection. For a more detailed coverage, see the following two Department

of Surveying Engineering, University of New Brunswick Technical Reports:

(1) Frankich, K. (1977). A Study in Conformal Mapping. Department of

Surveying Engineering, Technical Report No. 4~, University of

New Brunswick, Fredericton.

(2) Thomson, D.B., M.P. Mepham, R.R. Steeves (1977). The Stereographic

Double Projection. Department of Surveying Engineering,

Technical Report No. 1+6, University of New Brunswick, Fredericton.

 97

9. A GENERALIZED SET OF CONFORMAL MAPPING EQ0ATIONS

9.1 Introduction

In the previous sections, we indicated Which conformal map

projection is best suited for areas of a particular shape such that the

scale error would be a minimum. For example:

(a) North-south extent, Transverse Mercator,

(b) East-west extent, Lambert Conformal Conic,

A legitimate question to pose is: What map projection is to

be employed when an area is located obliquely to the north-south or

east-west directions. An oblique projection? This is one possible

answer, but how about areas which are arc-shaped, e.g. circular,

hyperbolic? A possible answer to the above questions is given below in

connection with the analysis of a generalized set of formulae for the

conformal mapping of the ellipsoid onto the plane.

 98

9. 2 Generalized Formula.e for Conformal Projections

The generalized set of formulae were formulated after. a careful

analysis of the equations belonging to the various particular conformal

map projections .[~$.k::l.:'li.sky· 1967]. Given below is the generalized set of

formulas in our notation where C* is a constant, R is the radius of curva-

ture of the ellipsoid at the central parallel, S~ the length of the

meridian, at,td· the length of a parallel given by Sp = ).N cps 4>. 9-1

3 s 2
X = (1 + p) S + (C* - 6 + ••• 9-2
C* ~ p 6C* 2R

y = 84> + + ••• 9-3

C* R2
The corresponding general scale factor equation is

6s cp2 + (c* - 6) s 2p
k =1 .. -2 9-4
2 C* R

~e main result is the above equation for the scale factor.

The above set of formulas are good for maps at a scale of 1:50,000. In

order to insure an accuracy of 1 to 4 meters in x andy, the area is

limited to 450 km in latitude and longitude. More accur~cy in x andy

can possibly be achieved simply by including more terms in the expansion.

9.3 Application of the General Scale Formula

Specific map projections along with their varying shaped isoscale

curves are derived by simply stipulating particular values for the constant

C* in 9-4. An interesting analogy exists in physics where the path of

 99

motion of a particle in a central force field is determined to be an

ellipse, hyperbola, etc. by simply stipulating a certain value of a

constant in the solution of the differential equation of motion.

A wise selection of the constant C* makes it possible to obtain

projections whose isoscale curves have the shape approaching that of the

area to be mapped.

It should be noted that 9-4 is to be used only to investigate

possible conformal projections; the actual computation of"' the scale should

be performed with extended formulas giving more accuracy.

lrunediately below, seven possible conformal projections are..

discussed as special cases of the general, among which will appear the

four basic (Mercator, T.M., L.C.C., Polar Stereographic) conformal

projections described earlier in these notes.

Case C* = oo (Transverse Mercator):

The scale factor equation becomes

6 s 2 C* S 2 6s 2
k =1 + ~ + E E
2 C* R 2 2 C* ~ 2 C* ~

9-5
2
which is identical to the truncated form for the Transverse Mercator

derived earlier. Similarly, expressions for x and y are obtained by

substituting C* = oo in 9-2 and 9-3:

 100

X= 8
p ...' 9-6

t s 2
9-7
Y = 8~ + 2N p + • • •

FOr the remaining cases, the x-y equations will not be given.

The main point of interest lies in the shape of the isoscale

curves. From 9-5, the longitude from the central meridian of all points

having a scale of 1 + 6k is

12 6k l/ 2 .
"Ao 'o
=p--,~,k-
( ) 9-8
'I' cos 0

1
Basically, the curve has the shape curve which changes slowly
cos cp
near the equator and rapidly near the_ pQ]..~s (_EJgure 9-1 }_ _!

~----------------------------------------------------------------~

Pole
isoscale curves

equator

isoscale curves

Pole

Figure 9-1: Isoscale Curves for Transverse Mercator C* = oo

 101

Conclusion. The above figure illustrates that an area great in

north-south extent but not too close to the poles may enjoy constant scale

error throughout its length and further the scale error ~ be made

minimal if east-west zones are devised like in the U.T.M. It has already

been mentioned that zones near the poles may be made as wide as 30° simply

because of the shape of the isoscale curves in this area (e.g., at~= 85°
and k = 0.9996, k = 1 at A = 18°).
0

Case 12 < C* < ~. C* = 18:

The equation for the scale error becomes

s/ + 2 sP 2
k =1 + 9-9
6 R 2
or
=k - 1 . 9-10

According to quadratic surfaces in solid analytical geometry,

9-11

is the trace of an ellipsoid in the x-y plane with semi-maJor and mino~

axes o"r a and b, respectively. In our problem the elliptical trace

has parameters

a = 9-12
b=I3R. 9-13

Thus the isoscale curves are ellipses elongated approximately two times

more along the meridians than along the parallels (Figure 9-2).
 102

____ ________
.;._ ""'···------------~

central
merid.ian

isoscale curves

k =1 or k 0

Figure 9-2: Isoscale Curves for 12 < C* < oo

Conclusion. According to 9-9, k =1 at one point only and from

that general point the scale error increases in a specified manner. In

order to have a nearly constant and minimal scale error in an elliptically

shaped area,a value less than unity could be chosen for the central point.

Such a selection could be based on a least square adjustment.

Case C* = 12 (stereographic):

The scale error equation becomes

s 2 s 2
K=l+l + _F_;.
9-14
- 2' -2
R R
3 3

and clearly the traces on the mapping plane are circles c~ntered at a

central point within the area to be mapped (Figure 9-·3).

 103

---- --z_central point

ot area

k =1 or k
0

Figure 9-3: Stereographic Projection, C* = 12

The Polar Stereographic Projection is obtained by choosing

the central point as the pole. Further, the U.&.P. is obtained by choosing

k =0.994 at the pole.

0

Conclusion. The stereographic projection or U.S.P. seems well

suited for small circular areas, which is undoubtedly the reason for using

it in the polar areas.

Case 6 < C* < 12 3 C* = 9.

The scale error equation becomes
s 2 s 2
k =1 + l + E 9-15
3R 2 6 'R 2

This case is simply Case C* = 19, rotated 90°.

 104

Conclusion. This case of conformal map projection should be

used for small areas elliptically elongated in the longitudinal direction.

Case C* =6 (Lambert Conformal Conic)

The scale error equation becomes

9-16

The isoscale curves are functions solely of the meridian distance from

the standard parallel and coincide with the parallels of ~atitude. This

is obviously the case in the Lambert Conformal Conic Projection.

Conclusion. Since the scale can be kept constant along a

given parallel, the use is obvious in any area of great longitudinal extent.

In order to increase the latitudinal coverage, two standard parallels are

introduced with scale factors different from unity.

Case C*:: 6, C* = 3

The scale error equation is

s 2
K = 1 +- 9-17
'R2
The unequal denominators and the minus sign in the above equation signify

that the traces of the isoscale curves on the map plane are hyperbolic

curves whose vertical axis coincides with the central meridian (Figure 9-4).
The Transverse Mercator case which is similar to this one can

be obtained from Equation 9-17 by omitting the second term and changing

the sign of the third, however without mathematical justification.

 105

central
meridian
isoscale.curves

arced-
shaped area

Figuxe 9-l+: Isoscale Cuxves Case C* = 3

~------------------------------------------------------------------------~
Conclusion. The above figure depicts the odd shape of the isoscale

curves. A stipulation of unity for the scale factor at the central point

in this case would be a bad choice since there is a rapid change of the

scale factor from this point. A better choice would be some value less

than tmi ty such that a curve away from the central point and which passes

through the centre of an arced-shaped area would have a value of k =1

(Figure 9-lt) .

Case C* < 02 C* = -3
s 2 s 2
K = 1 + E l 9-18
2-2 -2
-R R
3
 106

The isoscale curves in this case are hyperbolic with the main axis coin-

ciding with the central parallel.

This case slightly resembles the Mercator. Equation 9-18 could

be revamped to fit the Mercator Projection by omitting th~ 2nd last term

(making the scale factor dependent on latitude only), thus

1 + <1>2 9-19

as compared to

k = a =1 + a - N cos ; _ 1 +1 - cos p 9-20

N cos 4> N cos 4> - cos 4>

These two equations are not completely equivalent.

Conclusion. This projection ~d be used for areas which are

arced-shaped and extend slightly obliquely to the E-W direction, i.e., East-

East-North (EEN).

9.4 Summary

By studying the gerneral equation for the ~cale factor, it was

demonstrated how projections with quite distinctly different shaped isoscale

curves could be derived. It was also stated which case or shape of isoscale

curve may be best suited for an area of particular shape. Table 9-1 serves
as a summary of the foregoing.

In general, it may be stated that when mapping small areas, pro-

jections with closely knit isoscale curves should be used, thus achieving

minimal scale error within the area with a minimum number of zones.

Medium and large sized areas of' varying shape may be mapped with projections
 107

possessing more sweeping isoscale curves like hyperbolas. When the

entire earth ·is to be mapped, the projection with the most sweeping

isoscale curves used in conjunction with zones is the most satisfactory,

namely the Transverse Mercator.

 Table 9-l: Su[gested Conformal Projection

Area Description Case Projection Name Shape of Isoscale Curves

+'
N- S C* = "' Transverse Mercator Oblong along central meridian
s::(!)
+' E- W C* = 6 Lambert Conformal Conic Corresponding to parallels
~
@ C* =-3? Mercator Straight lines (infinitely large
.,..;
'Cl
hyperbolas)
~
(!)
}NNE and other C* = 3 ? Hyperbolas with central meridian
0
Oblique quadrants as main axis
+' or
(!)
bO
Arced EEN and other c* = -3 ? Hyperbolas with central parallel
;... quadrants as main axis
41l
...::!
I-'
0
CD

+'
s:: Circular (e.g. polar areas) C* = 12 Stereographic Concentric circles
(!)
+'
~ Elliptical N - S C* = 18 ? Ellipses extended along meridian
.-'1
=9 Ellipses extended along parallel
~
Elliptical E - W C* ?

-----·-·- -- -- ---------- -----

Note: Based on the assumption that minimal error is desired within area.
 109

SECTION II

COMPUTATIONS ON A CONFORMAL MAl)

PRO,JECTION PLANE
 110

10. INTRODUCTION TO COMPUTATIONS ON A CONFORMAL MAP PROJECTION PLANE

In Section I we treated the problem of transformation of

geodetic coordinates (~, A) on the ellipsoid to map coordinates (x, y)

on the conformal map plane, and vice versa. The first objective of

this section is to show how geodetic position computations, such as

the direct and inverse problem, can be made in terms cf map coordinates.

The advantage of this approach is that the mathematical models can be

characterized by plane geometry. The disadvantage is that observations,

such as distances, directions, astronomic or gyro-theodolite azimuths,

must be reduced to the conformal map projection plane. The development

of the formulae for the reductions is the second ob,jective of Section II.

It is worthwhile to place the map projection method of geodetic

position computation in context by outlining at least two other methods

of computation. First, recall that computations in three-dimensions

required that observations remain in space, as observed, with correction

made only for the effects of the gravity field and the atmosphere. No

surface such as an ellipsoid, sphere, or plane was eA~licitly employed,

thus no reductions of the observations was necessary. Mathematical models

were characterized by three-dimensional geometry. Coordinates of points

were expressed in terms of three dimensional coordinates (x, y, z, or ~.

 111

A, h). Also, recall that computations on the surface of the ellipsoid

required that observations be reduced to the surface of the ellipsoid by

first taking into account the effects of the gravity field and the atmos-

phere, and then the effect of the height of target and geodesic-normal

section separation. Mathematical models were characterized in terms of

ellipsoid surface geometry. Coordinates for points were given in terms

of horizontal geodetic coordinates($, A).

Why compute geodetic positions on a conformal projection pl~ne

instead of, for example, on an equiareal or an equidistan~ map projection

plane? The answer to this question is straightforward. Conformal pro-

jections preserve angles and therefore measured angles can be directly

used for computational purposes. However, sometimes these angles need to

be corrected in order to transform them into the corresponding plane angles

required for plane computations (see equation 11-1). The scale factor at

a point in conformal mapping is a function of the position of the point

only.

k = k($, A) 10-1

In other words, all distances in a relatively small area must be multiplied

by one and the same scale factor. However, for longer lines some average

scale factor should be used.

In equiareal and equidistant projections, the scale factor is

also a function of the azimuth at a point, i.e.

k = k($, A, a) 10-2

which implies that the angles are not preserved.

The third important advantage of conformal projections is the

fact that with a priori defined relative accuracy of linear scale factor

conformal maps will cover the largest domain compared to all other
 112

conceivable map projections. That means with conformal maps we shall

have a minimum number of zones or projection systems.

In the chapters to follow, we discuss: the problem of reduction

of observations onto the map projection plane (section 11); the various

mathematical models for computation on a map projection plane (section

12); the general formulae for reductions (section 13); the reduction

formulae for specific projections (section 14)~

 113

11. REDUCTION OF OBSERVATIONS

In this section we treat the problem of the reduction of

distance, direction, and azimuth observations onto the conformal map

plane. First we describe why these reductions are needed.

During the conformal mapping process, lengths of ellipsoid

geodesics are altered. The directional characteristics of these

geodesics on the map plane is dictated by the· e0n:fcmn~tj:"~Ondt'tfcijl~

. " . .,.,. .. · ,. ....
that

is to say, geodesics are projected onto the map plane in a manner that

preserves the angle which a given pair of geodesics define.

11.1 Reduction of Angles

Let us examine two specific angles on the map plane. The

first is the angle between two projected geodesics at a point to two

other points, and the second is the angle between the two chords at a

point to two other points.

 ll4

II

(T-t)l2

(T-t)l3

Figure 11-1. Angles.

Because of conformality the ellipsoid angle is equivalent to

the angle between the two projected geodesics (a2 13 ) . The grid azimuth

of the projected geodesic is denoted by T12 , while the grid azimuth of

the chord is denoted by t 12 • The difference between these two azimuths

is known as the "T minus t" correction. A similar situation exists

between points 1 and 3. The angle between the two g~odesics is related

to the grid angle by the equation

1•213 = 8~13 + (T-t)12 + (T-t)l3 ' 11-1

a 213 is the angle needed in the computation of positions of unknown

points. In general (T-t) 12 does not equal (T-t) 13 , and ~rther the

formulae for computing the (T-t) correction is different for each

conformal map projection.

 115

11.2 Reduction of Azimuths

There are three different azimuths to contend with. Two have

already been defined (T and t). The third is the angle b~tween the

tangent to the projected meridian and the tangent to the projected

geodesic and is denoted as a 12 (Figure ll-2). The following relation-

ship is valid

11-2

where y is the meridian convergence. Further

11-3

projected geodesic
projected
meridian

Figure ll-2. Azimuths

The above expression gives the value of the grid azimuth on the plane in

terms of the "observed" ellipsoid azimuth. t 12 is needed in computations

 ll6

on the plane. Expressions for y have already been derived. What remains

to be done is the derivation of expressions for (T-t} for the various

projections.

11.3 Reduction of Distances

Three types of distances are involved: the "observed" geodesic

distance S*; the length of the projected geodesic distance S; the chord

length d (Figure ll-3). The chord distance is

11-4

where (S*-8) 12 is the correction due to the length distortion of the

geodesic, while (S-d12is due to the difference in length between the

projected geodesic and the chord. These two corrections are usually

combined into one as we will see later when it is de~ived for the various

map projections.

Figure ll-3. Distance

 117

12. MATHEMATICAL MODELS FOR COMPUTATION OF POSITIONS

Let us turn away from the conformal mapping problem pursue to

the problem of computing on the mapping plane. The reason for doing

this is purely stategical. We use this application-oriented section as

motivation, which will undoubtedly help us endure the rather tedious but

interesting, development of the reduction formulae.

The mathematical models are not new to· a student at this level.

The intent is to recapituate the models and show where and when the

various reductions are necessary. For an exhaustive treatment of

models for the computations on a plane, see for example, Faig [19721.

We treat the problems of: intersection, traverse with astro-

nomic azimuth orientation; traverse with grid azimuth orientation;

adjustment of a network on a plane; and calibration of a gyro-theodolite

using conformal map projection coordinates.

12.1 Intersection

The intersection problem is depicted in Figure 12-1 along with

the projected geodesics. The knowns are the map coordinates of the two
 118

points 1 and 2; the unknowns are the map coordinates of the third point

3; the observed quantities are the two angles at 1 and 2.

t
1

1
Figure 12-l. Intersection and Projected Geodesics

The solution for the two unknown coordinates x 3 ,a.nd y 3 is made

from the two equations:

tan t 13 =

= ~3 - x2
y3 - y2

and
 119

Note t 12 and t 21 are computed from known coordinates; ~ and a 2 observed

(ellipsoid values); while the (T-t) correction needs to be computed

(expressions to be derived in Section 13).

12.2 Traverse with Astronomic Azimuth Orientation

The traverse problem along with the projected m~ridians and

geodesics is depicted in Figure 12-2. The knowns are the~ap coordinates

of a starting point 1. The observed quantities are the astronomic

azimuth of the geodesic ( 1-2) ~ the angles and the distances • 'r!W-U'nk.nowris
are the coordinates of points 2 to n.

1
Figure 12-2. Traverse

The coordinates of point 2 (e.g. x 2 ) can be computed from

sin t 12 =
where

and
 120

dl2 = 8 ~2 - (S* - S)l2 - (S - d)l2 '

Again we see that (T-t) correction is necessary along with the

meridian convergence. Distances also need reduction.

12.3 Traverse with Grid Azimuth Orientation

The solution to this problem is similar to the previous case,

with one change, that is

tl2 = tl + al + (T-t)l- (T-t)l2 .

Note an observed angle at 1 is necessary. This requires a computation of

a (T-t) correction for the backsight. All other corrections are

identical to that explained immediately before.

Ll. l1 Ad,justment of a Network on a Conformal Map Plane

The unknowns in the adjustment of a network on a conformal

map plane are the usual unknown coordinates (possibly some nuisance

parameters such as an orientation unknown at each station). The

knowns are the map coordinates of the stations to be fixed. The

observed quantities are directions, distances and azimuths. These

need to be reduced to the map plane as expressed by 11-1, 11-3, 11-4.

The adjustment then proceeds as usual.

 121

12.5 Gyro-Theodolite Calibration Using Map Coordinates

The basic idea is to compare the azimuth observed by the gyro

with the azimuth computed from map coordinates. They are, of course,

two different kinds of azimuths, thus one must be "reduced" to the

other. What reductions are necessary? All necessary concepts needed

to answer this question have already been dealt with - thus this

problem is left as an exercise.

Assume that a gyro-azimuth is equivalent to an astra-azimuth.

 122

13. GENERAL FORMULAE FOR REDUCTION TO THE MAP PROJECTION PLANE

The general formulae for reduction of distances and directions

to the conformal map projection plane are discussed in this section.

These general formulae are valid for most conformal projections. The

purpose of this discussion is to give the reader an appreciation for the

fundamentals underlying the computational versions of these formulae given

in Section 14 for some specific projections. By understanding the fun-

damentals, the users of the computational formulae can better understand

their application and limitations.

Discussed immediately below are:

(1) the radius of curvature of a projected geodesic;

(2) the parametric equations of the projected geodesic;

(3) the difference in length between the projected geodesic and the chord;

(4) the difference in length between the geodesic and the projected

geodesic ;

(5) the angle between the projected geodesic and the chord.
 123

Just the main steps of the derivations are given. More details

can be f01md in 'rhomas [1952].

13.1 Curvature of the Projected Geodesic

By studying the curvature of the projected geodesic we gain

insight into the geometry of these curves on the map plane. From calculus

we know that the curvature is defined as

13-1

where H is the radius of curvature, and

x = x(s)
13-2
y = y(s)
are functions of the variable s, which is the length of the projected

geodesic. From Figure 1+-4,

I dx
X = - = cos (3 ' YI : : : 2x.
ds -
- sin (3
ds
2 13-3
II d X (3 d(3_
X = sin 6 cos
ds 2 ds

Substituting 13-3 into 13-l yields

2 dS . 2(3 dS
cos s - + s1n
ds
cJ
1
= - =
R - 2
ds
. 2(3)3/2
(cos (3 + Slll
13-4
= dB
ds

'J'homas derives an expression for the curvature as a function of

the c>cale factor k and coordinate n (figure 4-4). Namely

 124

dB= 1 ok . ok
0 = -
ds
- ( - S1n 6 -
k ax ay cos a ) '
13-5
l (ak h ok ~) = 1 ok
= + ay
k ax on an k an

Note in the above all quantities are evaluated. at the point where d is

desired.

13.2 Parametric Equations of the Projected Geodesic

The projected geodesic is described by two parametric equations

of the form

n=n(s),
13-6
~=~(s).,

Approximating the above equations with a Maclaurin series, with point 1 as

the point of expansion, we get (Figure 13-1)

projected
geodesic

Figure 13-1: Coordinate Systems for

Parametric Equations
 1?5

= s
dt; ( 0) s2 d 2 ~~o) 3 3
+ ~ d ~~0~ + ...
ds + 21 2 3! ds3 '
ds
13-5
2 2 3 3
n :::: s dn(o) +s-
ds 2!
d n(o)
2
+ E._ £_nl£) +
3! ds 3
....
ds
From 13-l

13-6

By f:Lrst evaluating the derivatives in 13-5 and then replacing them with

curvatures (13-6). we get

3 4 5
s a s 4
2 I
·-- (4a 0 a0 I I ... '
c•

~ ·- s -6" a
0
- -
8
a
0 0
- 120 + 3(a I)2
0
a
0
) -
13-7
2 I
6a a ) + ••• ,
0 0

where
I dO II
a -· ds ' 0

are eval.uated from

13-8

and where the subscript "o" means evaluated at the point of expansion.

13.3 Difference in Length Between Projected Geodesic and Chord

Since we compute with chords, we need to determine the difference

in length between the projected geodesic and chord. From Figure 13-1, we

can write and expression for the chord length d in terms of the coordinates

~ and n, (13-7), namely

 126

d2 = n2 + 1;;2 13-9
4
= s2 - s

13-10

d = (1;;2 + n2)1/2
2
s 2 I s2 II I 2 4 l/ 2
= s { 1 - 12 [a 0 +a 0 a0 s + 30 ( 9a 0 a0 + 8 (a 0 ) -a 0 ) + .• ]} •
13-11

Since the second term in brackets is small the expression to the 1/2 power

may be expanded. The result gives the difference

I
a a
0 0
s - d 24

13-12

If we expand about the mid-point of s instead of at one end, then

3 2
s 02
s-d = 24 ( ..... ) , 13-13

where o 2 is the radius of curvature at the midpoint of the projected

geodesic. The value of this difference is usually small.

13.4 Difference in Length.Between the Projected Geodesic and the Geodesic

Combined with the difference s-d, the difference in length

between the projected geodesic and the geodesic (s), that is 8- sallows

one to reduce geodesic lengths on the ~ipsoid to chord lengths on the

projection.

The basic relationship used in this context follows from the

definition of the scale factor, namely

ds =k dS . 13-14
 127

The distance on the map projection (projected geodesic) is then

s =r k ds 13-15

Thomas derives an expression for the above which involves a series develop-

ment of the scale factor with point of expansion being one end of the line.

The result is

13-16

I dk
where k = --
ds
. The second and higher terms account for the change of scale

away from the point of expression. The equation for the difference is

s - s

53
6 + .•. ' 13-17

where 1/k is the reciprocal for the scale factor.

Analogous expressions to 3-16 and 3-17 for a midpoint-point of

expansion are

13-18
and
l
s - s = ( - - l) s + +.•••• 13-19
kM

13.5 The Angle Between the Projected Geodesic and the Chord

In conformal projections, the azimuths of the geodesic and

projected geodesics are identical, but not so for distances as we have just

seen. For this reason we need concern ourselves only with the one angle -

that between the chord and projected geodesic.

 128

The angle we seek is defined by (Figure 13-l) e =T - t.

In terms of the coordinates ~ and n

tane = .!1E;; " 13-20

 14. SPECIFIC FORMULAE FOR REDUCTION TO VARIOUS MAP PROJECTION

PLANES

In Section 13 we derived the general formulae for the reduction

of ellipsoidal distances and directions to the map projection plane.

These formulae are expressed in terms of the curvature of the projected

geodesic and the scale factor. In this section we derive reduction

formulae for the Transverse Mercator projection by evaluating the curva-

ture and scale factor. We also state the reduction formulae of the

Stereographic Projection in New Brunswick.

Reduction formulae are not given for the Mercator or Universal

Stereographic Projections as geodetic computations of positions are not

usually performed on these projections. The reasons being is that the

USP is used in the polar regions while the Mercator is used mainly for

navigation.

14.1 Reduction Formulae for the Transverse Mercator

Evaluated immediately below is the curvature of the projected

geodesic. Then the reduction formulae for directions and distances are

derived.
 130

14.1.1 Curvature

Recall the expression for the curvature (13-5)

a =~(~sin e - Clk cos e) . 14-1

k ax ay
For the TM~ has the form

14-2

where, for example

1 + 2
nl
c = ----:- 14-3
1 2N 2
1
We recognize that k depends mainly on x and only slightly on y because

y is only implicitly•involved in n1 .and Ni through ~l the foot point latitude

Equation 14-1 can then be approximated by

1 Clk . Q
a= k Clx Sl.n iJ, 14-4

For further use we need the following expressions which are

derived in Thomas [1952]

I 3k
k = ax cos (3 14-5

II 2
a k cos 2 e - 2
k = -- ka 14-6
ax 2
1
1 I d(k)
(k) = "dX cos (3 14-7

2
(~)II= EL_ (-l) cos2i3 14-8
k dx2 k

14.1.2 Arc to Chord Evaluation

Recall the general formulae giving the angle between the arc of

the projected geodesic and the chord (13-21)

 131

2
s
e = T - t = 2 o0 + 6s o0
I
+ ••• 14-8

We first evaluate o0 for the TM. For 1/k in 14-4 we get from (14-2 and

6-77 ).

-1k =• 1 --12 X
2 14-9
Nl
and
2
k :: 1 +l X
14-10
2 2
Nl
Also

dk X
14-11
dx =N 2
1
Sub'stituting 14-9 and 14-11 into 14-4 yields

X
0 = (1 - l -L) sin ~ · 14-12
. 2 N 2 N 2
1 1
The curvature at one end of the line (say point 1) can then be evaluated

as
xl xl
0
0
= (1 --12 -)
N2 N2
sin al . 14-13
1 1
y 2

~-----------------------------------------~
X

Figure 14-1 Chord and Arc

 132

We now wish to approximate x in 14-12 with (Figure 14-1)

X::: x1 + S COS a 14-14

where s · d at any point between 1 and 2. Substituting 14-14 in 14-12

yields

(x1 + s cos 8)2 x1 + s cos 8

0 = [1 --12 N2
] (
N2
) sin 8
1 1

(x1 + s cos 8) sin B (x1 + s cos 8 ) 3 sin 8 cos 13

1
=
N2
-- 2
+
1 N1 14-15

The other quantity reading evaluation is the change of the

I
curvature o in 14-18. From 14-15
0
2 2
I do cos 8 sin a- l (xl + s cos 6) sin 8 cos 8
(J =-
ds
= 2
14-16
N2 N4
1 1
I
Thus a0 at point l is obtained from above by replacing B by 81 (Figure

14-1).

The expression for T - t is obtained by substitution of 14-16

(with S = 81 ) and 14-13 into 14-8.

The result is
3
s x1 sin 81 1 x1
8 =T - t =2 N 2 - -:::-4
2 N
1 1

2 sin sl cos 81 3 2
+ ~ (---~:------= + •.• 14-17
6 N 2 - 2 xl
1

s x1 sin8 1 2
---=--=+~ + . • • 14-18
- 2 N 2 6
1
 133

Eliminating 81 from the above by (Figure 14-1)

s
14-19

s
we get

a=T-t= 14-20

where N1 can be approximated by the Gaussian mean radius at point 1.

The above equation is of sufficient accuracy for most work. It

can be compared to the. following formulae of Bamford [1962]

(y2- yl} (x2 + 2xl)

T - t = 14-21
6R 2
m

where

R =IMN,
m
M and N evaluated at (~ 1 + ~ 2 )/ 2. The accuracy of this formulae is

claimed to be O". 02 for a line 100 km in length but within 3° of the

central meridian.

The cd>rreet application of the correction is made by realizing

that projected geodesics are concave towards the central meridian (see

Figure 6-4 ) .

14.1.3 Line Scale

The line scale is the average scale over the line. It allows

us to get the difference between the lengths of the geodesic (S) and the

projected geodesic (s). Recall the general expression (13-17)

 134

2 3
S - s = {_!_-
k
0
1) s + (~I .L + (1)II .L +
k 0 2 k 0 6 ...' 14-22

where the derivatives are defined by 14-7 and 14-8. We now evaluate

14-23

14-24

14-25

xl
(j =- sin 131, 14-26
0 2
Nl
y2 - y 1
sin 131 = 14-27
d

14-28
d
Substitution of 14-24 to 14-28 in 14-7 and 14-8 yields

1 I -xl x2 - xl
{k)o =N 2 d 14-29
1 2
{J:.) I I = _ {x2 - x1 ) + 14-30
k o N 2d2
1
Substitution of 14-29, 14-30, 14-23, in 14-22, along with the approximation

that d = s, we get

s=d [ 1

14-31
 135

-sd = [1 - -
1
6N2
2
( x + x1 x 2 + x 2 ) ]
1
2
14-32
1

where N1 can be approximated by R = IIMli. The formula is of sufficient

accuracy for most work.

The following more accurate formula for the line scale (k) is

given in Bamford [1962]:

~--------~----------~
22
X X
k= 1 + u 2 (1 + _u_) 14-33
6R 36R2
m m

where

14-34
R
m = IMN
M and N evaluated at ($ 1 + $2 )/2. The accuracy of the above formula is
6
about 0 .l/10 for lines of about 150 km in length and within 3° of the

central meridian. Thus d = ks, without computing s (see 14-31).

14.2 Reduction Formulae for the New Brunswick Stereographic Projection

The reduction formulae for the New Brunswick stereographic

projection are given here without derivation. The full derivation is

given in Thomson et al. [1977].

14.2.1 Arc to chord (T-t) reduction

The arc to chord correction is given by [Thomson et al., 1977];

_1 xjy. - x.yj
(T-t) .. = tan [ - ~ ~ 2] h

~J x.xj + y.yj + (k R)
l. l. 0

where, xi, y i, x j, y j are the map coordinates of points Pi and P j; k 0 is

the scale factor at the origin and

 136

evaluated at ~ , the latitude of the origin.

0

14.2.2 Geodesic to Chord Length Reduction

The line scale factor mij used to obtain the plane length d

from the geodesic lengthS between two points Pi and Pj is given by

[Thomson et al., 1977],

-
1 s 1 1 4 1
= - = - (- + - + -)
m.j
l.
d 6 m.l. m
m mj

where, m., m and mj are the point scale factors at P., the mid-point of
l. m l.

the line (Pm), and Pj respectively. The point scale factor at Pi, for

example, is given by[Thomson et al., 1977],

k
0
= k0
 137

REFERENCES

Bomford, A.G. (1962). Arc to Chord and Finite Distance Scale Factor
Formulae, Transverse Mercator. Empire Survey Review #125,
P. 318.

Bomford, G. (1971). Geodesy. Third Edition, London: Oxford Press.

Conte, S.D. and Carl de Boor (1972). Elementary Numerical Analysis.

Toronto: McGraw-Hill.

Faig, Wolfgang (1972). Advanced Surveying I. Department of Surveying

Engineering Lecture Notes No. 26, University of New Brunswick,
Fredericton.

Krakiwsky, Edward J. (1967). A Discussion Concerning Ge~ralized Formulae

for the Conformal Projection of the Ellipsoid on the Plane with
an Emphasis on IsoscSle Curves. Unpublished paper, Advanced
Map Projection course at the Ohio State University, Columbus.

McLaughlin, A. (1971). Personal communication to the author.

Thomas, Paul D. (1952). Conformal Projections in Geodesy and Cartography.

U.S. Geodetic Survey Special Publication No. 251.

Thomson, D.B., M.P. Mepham, R.R. Steeves (1977). The Stereographic Double
Projection. Department of Surveying Engineering Technical
Report No. 46, University of New Brunswick, Fredericton.

UPS (1958). Universal Polar Stereographic Grid Tables for Latitudes

79°30' - 90" (International Spheroid). Department of the
Army Technical Manual TM5-241-9, Washington, D.C.

UTM (1958). Universal Transverse Mercator Grod. Department of the Army

Technical Manual TM5-241-8, Washington, D.C.

Wells, David E. (1971). A Method for Iterative Conversion of Isometric

Latitude into Geodetic Latitude. Unpublished paper, Department
of Surveying Engineering, University of New Brunswick, Fredericton.

Wray, Thomas (1973). On the Formalization of Map Projections. An

invited lecture to the Department of Surveying Engineering,
University of New Brunswick, Fredericton. (Mr. Wray is a
mathematician with the Surveys and Mapping Branch, 615 Booth
Street, Ottawa).
 138

APPENDIX I: EVALUATION OF MERIDIAN ARC LENGTH :..

~ = constant

Meridian Arc

The length of a differential :part of a meridian arc S is

given by

dS = Md¢1 , (1)
where
M = a(1-e 2 )
(1-e 2 . 2,~,)3/2
s~n "'
is the corresponding radius of curvature. For the total arc, S, we can

write

(2)

In the case of the TM projection, we are interested in the

length of the meridian arc from the equator (<P 1 = 0) to our point of

interest (q, 2 = q,). Then we write

2
= 1 q, a( 1-e ) d<P (2a)
0 (1-e 2 . 2,~,)3/2
s~n "'
 139

The expression for M can be expanded in a series and written

as
2
M _ a(l-e ) = a ( 1-e 2)(l + 23 e 2 s~n
. 2,~. lS 4 . 4,~. 3S
"'~ e a1n '~'+it e
6 . 6,~.
s~n
)
"'+ ••••
. 2,~.)3/2
-(1-e 2 s~n "' (3)
2
We treat a(l-e ) as a constant outside the integration and obtain

S
2
= a(l-e) <P 3
! 0 [1~
2.2
e s1n <P +
12.
8 e s~n
4.4
<P
12,6
+ 16 e sin
6
<P(t)·· Jd<P.
Splitting this into parts yields

... ] .
( 5)

Evaluating each integral separately

I!d<t> =<~>I!= <t> , (5a)

f<P(l e2sin2<P)d<P
0 2
= l2 e2[i- sin 2p
2 4
]<P ( 5b)
0

2~ . 2~
= 43 e "' - 83 e
2
s1n "' ,

f cJ>(l2 4 . 4~)d~ _ 12. 4r3~ sin 24> + sin 4p] <P

(5c)
o 8 e s~n "' "' - 8 e ~ - 4 3?.
0

Thus,

S = a(l-e 2 )[<P + t e 2cp- i e 2 sin 24> + ~ e 4<P- ~ e 4sin 24> + ;;6e 4sin 44> + •• ]
( 6)
and

S
3 2
= a {cJ>+t e <P - 83 e
2 .
s~n
2 "',~. + ~ e 4 <P - ~
lS 4 . 2 ,~. _12 4,~. . 4~
s1n "' + 256 e "'s1n "' -
64
(7)
2 3 4 3 4 . ~ 6 lS 6 . _12 6 . )
-e 4>- 4 e <P + 8 e s~n 24> - 64 e <1> + ~ sl.n 24> - 256 e sl.n 4<jl + • • • •

Collecting terms for cp, sin 2cj>, sin 4¢, etc.

 45 4 3 4 45 6
S-a((¢+ 43 e 2rp- e
2
4> + 64 e ¢- 4 e 4>- 64 e Q>+ ••• )

+ ( 15 4 . 4
256 e s:m 4> -
15 .• 44> + . . . ) ) •
"2'"§6 e 6s1n (Ta)

Including more terms in our series expansion results in the

following formula for the meridian arc length,

(8)

where
l 2 3 l~ 5 6 1"(5 8
A0 - 1 - 4 e - 64 e - 256 e - 16384 e (9)

A2 = i (e
2 1 l.~
+4e + 128 e - l+096 e )
15 6 455 8 (10)

15 4 3 6 77 8 (11)
A4 = "2'"§6 ( e + 4 e - 128 e )

3'5 6 _ hl e8) (12)

A6 = 3072 (e 32 -

A8 - - - 315 e8 (13)
131072

'l'his formula for the meridian arc length is accurate to less than 0.001. m.
 APPENDIX II. EVALUATION OF FOOTPOINT LATITUDE

When dealing with the T.M. conformal projeetknwe remember

that the inverse formulae for ~A and ~~ are given in terms of the

"foot point" latitude, ~1 . Thus, we need an expression for determining

~l from y.

Remembering that the length of a meridian arc is given by

S = !~0 Md~

and that along the central meridian,

y = S = !~0 Md~ ,
then our footpoint latitude will be given by the meridian arc length "y"

along the central meridian.

To solve this problem, we use the Newton-Raphson iterative

procedure (see pp. 26-28). In brief, we require the following three

steps:

(1) Obtain an initial appropriate value for ~ (.!. ) •

'~'o '

(2) Evaluate the iterative improved solution from

~n = ~n-1 - ~~¢~~) ; (14)

(3) Continue the iterative procedure until

~~n- ~n-11 < £ '

where E = lo- 12 radians (which is approxim,ately 2 x ia-7.· arcsec).

Specifically, we have (from 8)

 142

Then

and

f'(~) = a(A0 - 2A 2cos 2~ + 4A4 cos 4' - 6A6 cos 6' + 8A8cos 8') , (16)

where A0 , A2 , A4 , A6 , A8 are given by (9), (10), (11), (12), (13) res-

pectively. For an initial approximation, we use a spherical approximation,

that is e = 0, and (8) becomes

Y = a' (A0 = 1; Ai = 0, i = 2, 4, 6, 8, .•. ) (17)

Then

~
o
= y_a (18)

Using ~ , we get first approximations for f( ~) and f' ( 4>) (equations (15)
0

and (16) respectively). For n =l

f(q, )
q,1 = <Po- f'(q, )
0
(19)
0

and
-12
E:l = jq,l <Po I £1 < 10 rad.? ( 20)

Continuing, we then get n =2 and

f(q,l)
~2 = q,l- f'(<P )
1

and
-12
£2 = I<~>2 - ~1l ; c2 < 10 rad.

Thus, we compute the footpoint latitude, and using this we can

complete the inverse problem for the Transverse Mercator projection

(e.g. solve for 6A, 6~, thence <f> and A).

 143

APPENDIX III
ALTERNATIVE DERIVATION OF LAMBERT CONFORMAL CONICAL PROJECTION

In the polar coordinate system, the linear element ds is com-

puted by the well-known formula

ds2 = dp2 + P2 do2 = p2[(~)2 + do2] (1)
p

It is obvious from the above expression that o and p are not

isometric coordinates. The isometric coordinates must generally have

the following form

( 2)

However, by a simple substitution

- = £e.
- dp ( 3)
p

we can transform into an isometric system, where new variable p is directly

determined by the integration of the above equation

- p = in p - in K (4)

or
p =K • e-p ( 5)
 Now, using the obtained isometric coordinates we can set the

fundamental expression of conformal mapping

p + i 0 = f{g + i 1) (6)
where

l = A0 - A (7)

Conical projections are defined as projections in which meridians

are transformed into straight lines intersecting at one point and parallels

become concentric circles with the centre in the intersection point of

meridians. The angles between meridians are proportional~to the

corresponding angles on the ellipsoid.

Mathematically translated, these conditions mean

p + i 6 = K(p + i £), P =K p , o =K £ (8)

Substituting the derived value for p from (5) we have
-Kq
p =K • e 6 =K • £ (9)
or for rectangular coordinates

x =p sin 6 = K e -Kq sin K 1 ,

(10)
y = p cos o= K e -Kq cos K l .