Académique Documents
Professionnel Documents
Culture Documents
TRANSFORMATION FORMULAE
R.E. Deakin
Department of Mathematical and Geospatial Sciences, RMIT University
GPO Box 2476V, MELBOURNE VIC 3001, AUSTRALIA
email: rod.deakin@rmit.edu.au
April 2004
ABSTRACT
The standard and abridged Molodensky coordinate transformation formulae are sometimes used by
practitioners in the surveying and geodesy professions and are two of the standard models of
coordinate transformation widely used in Geographic Information System (GIS) software. The
formulae enable latitudes, longitudes and heights (, ,h ) related to one ellipsoid to be transformed to
, ,h related to another ellipsoid on the implicit assumptions that: (i) the X,Y,Z Cartesian axes of
both ellipsoids are parallel, (ii) the coordinate differences X , Y , Z between the origins of the two
reference ellipsoids are known and (iii) the defining geometric parameters of both reference ellipsoids
are known. The formulae accept , ,h as input variables and give changes , , h , thus the
standard and abridged Molodensky transformation formulae are known as curvilinear transformation
formulae. This paper provides a detailed derivation of the formulae together with a worked example
that may be useful to the interested practitioner.
INTRODUCTION
In Molodensky et al. (1962), the authors derive a set of differential equations for transforming
coordinates from one geodetic datum to another. Their equations (Molodensky et al., (I.3.2), p. 14),
linked changes in x, y, z Cartesian coordinates of a point with, (i) rotations x , y , z of the Cartesian
axes about some fixed point x 0 , y 0 , z 0 , (ii) "progressive translations" dx 0 , dy 0 , dz 0 of the ellipsoid origin
between x, y, z Cartesian axes, and changes in the ellipsoid parameters a and f with changes in
curvilinear coordinates , , h . Subsequent publications by other authors have described
"Molodensky's" transformation in terms different from the original. This confusion was addressed by
Soler (1976, p.2) who states:
" the differential equations published in the English translation of [Molodensky et al.,
1962] are equivalent to conventional conformal transformations. This dissipates the
confusion created recently by some authors [Badekas, 1969], Krakiwsky and Thomson,
1974], who credited [Molodensky et al., 1962] with a model they never wrote."
Z2
Z1
centre of ellipsoid 1
O1
Z
Y
ellipsoid 2
h
Q
l
rma
no
O2
X
H
Y1
Y2
id 2
l lipso
e
f
o
r
equato
X1
X2
the rotational axis of the ellipsoid at H. The distance PQ is the ellipsoidal height h and the distance
QH = is the radius of curvature in the prime vertical plane of the ellipsoid. , , h are latitude,
longitude and ellipsoidal height respectively, a is the semi-major axis of the ellipsoid and f is the
flattening of the ellipsoid. O1 is the centre of another ellipsoid ( a1, f1 ) and two assumed parallel
Cartesian coordinate axes are shown with origins at O1 and O2 . The Z 1 and Z 2 axes are assumed to
be parallel and are the rotational axes of the ellipsoids, the X1O1Y1 and X 2O2Y2 equatorial planes are
parallel and are the origin planes of latitude. The X1O1Z1 and X 2O2Z 2 planes are parallel and are the
origin planes of longitude. The origins O1 and O2 are related by the translations X , Y and Z .
The Cartesian coordinates of a point, related to the centre of an ellipsoid, are
X = ( + h ) cos cos
Y = ( + h ) cos sin
(1)
Z = ( (1 e 2 ) + h ) sin
is the radius of curvature of the ellipsoid in the prime vertical plane and e 2 is the square of the
(1 e
1/ 2
sin 2 )
e 2 = f (2 f )
(2a)
1 e 2 = (1 f )
The eccentricity e, the flattening f, the semi-major and semi-minor axes of the ellipsoid (a and b,
respectively) and , the radius of curvature of the ellipsoid in the meridian plane are defined by the
following equations
a 2 b2
a2
a b
f =
a
b = a (1 f )
e2 =
(2b)
a (1 e 2 )
3/2
(1 e 2 sin2 )
a cos cos
1/ 2
+ h cos cos
1/ 2
+ h cos sin
1/ 2
+ h sin
(1 f (2 f ) sin2 )
a cos sin
(1 f (2 f ) sin2 )
(3)
Z =
a (1 f ) sin
(1 f (2 f ) sin2 )
Using the theorem of the total differential (Sokolnikoff & Redheffer 1966), small changes in the X,Y,Z
coordinates can be linked to small changes in a, f , , and h .
X
X
X
X
X
a +
f +
+
+
h
a
f
h
Y
Y
Y
Y
Y
Y =
a +
f +
+
+
h
a
f
h
Z
Z
Z
Z
Z
Z =
a +
f +
+
+
h
a
f
X =
(4)
In equations (4), all -values are formed by subtracting ellipsoid 1 values from ellipsoid 2 values, e.g.,
a = a1 a2 . These equations are the basis of the standard Molodensky transformation.
The standard Molodensky transformation formulae are derived in the following manner.
The derivatives in (4) can be found from equations (3).
Derivatives
X Y Z
,
,
a a a
X
cos cos
cos cos
cos cos
=
=
=
1/ 2
1/ 2
2
2
2
a
a
(1 f (2 f ) sin )
(1 e sin )
(5a)
Y
cos sin
=
a
a
(5b)
2
(1 e 2 ) sin
(1 f ) sin
Z
=
=
a
a
a
(5c)
Similarly
Derivatives
X Y Z
,
,
f f f
3 / 2
X
1
= a cos cos (1 f (2 f ) sin 2 )
((2 2 f ) sin2 )
f
2
but =
(1 e 2 sin2 )
a (1 e 2 )
3/2
(1 e 2 sin2 )
a (1 f )
3/2
(1 e 2 sin2 )
hence
X
sin 2
cos cos
=
1 f
f
(6a)
Y
sin 2
cos sin
=
1 f
f
(6b)
Similarly
d u
=
Using the quotient rule of calculus:
dx v
du
dv
u
dx
dx
v2
1/ 2
(1 f (2 f ) sin2 )
Z
=
f
(1 f ) sin 2
1 f (2 f ) sin
2
1/ 2
1/ 2
1/ 2
2a sin (1 f ) (1 e 2 sin 2 )
+ a (1 e 2 ) sin (1 e 2 sin2 )
(1 f ) sin 2
1 e 2 sin 2
2a sin (1 f )
1/ 2
(1 e 2 sin2 )
a (1 e 2 ) sin 2 sin (1 f )
3/2
(1 e 2 sin2 )
= 2 sin (1 f ) + sin 2 (1 f )
giving
Z
= ( sin 2 2 ) sin (1 f )
f
Derivatives
(6c)
X Y Z
,
,
1/ 2
(1 e 2 sin2 )
X
=
h sin cos
=
a sin cos
1/ 2
(1 e 2 sin2 )
1/ 2
e 2 sin cos
1 e sin
(1 e 2 sin2 )
h sin cos
e 2 cos2
h sin cos
1 e 2 sin 2
e 2 cos2
= sin cos 1
h sin cos
1 e 2 sin 2
= sin cos + sin cos
{ }
is
2
2
2
1 e 2 sin 2 e 2 cos2 1 e (sin + cos )
1 e2
=
=
=
2
2
2
2
2
2
1 e sin
1 e sin
1 e sin
giving
X
= ( + h ) sin cos
(7a)
Y
= ( + h ) sin sin
(7b)
Similarly
1/ 2
(1 e 2 sin2 )
Z
=
+h cos
=
1/ 2
e 2 sin cos
1 e sin
a (1 e 2 ) cos 1 e 2 sin 2
a (1 e 2 ) cos 2
+
e sin 2 + h cos
1/
2
3/2
2
2
2
2
2
2
(1 e sin ) 1 e sin (1 e sin )
giving
Z
= ( + h ) cos
Derivatives
(7c)
X Y Z
,
,
1/ 2
a
cos sin h cos sin
=
1/ 2
2
(1 e sin2 )
giving
X
= ( + h ) cos sin
(8a)
Y
= ( + h ) cos cos
(8b)
Z
=0
(8c)
Similarly
Derivatives
X Y Z
,
,
h h h
X
= cos cos
h
(9a)
Y
= cos sin
h
(9b)
Z
= sin
h
(9c)
a
Y = B + J
f
Z
h
(10)
where
X
Y
B=
a
Z
X cos cos
a
f
Y cos sin
=
a
f
Z (1 f )2 sin
f
a
sin2
cos sin
1 f
( sin2 2 ) sin (1 f )
sin 2
cos cos
1 f
(11)
Y
J=
Y
= ( + h ) sin sin ( + h ) cos cos cos sin
h
0
sin
Z ( + h ) cos
(12)
= J Y + B
(13)
The inverse of J can be found by the method of cofactors and adjoints (Mikhail 1973, pp. 442-5).
a
11 a12 a13
a 31 a 32 a 33
A1 =
adj A
where adj A is the adjoint matrix and A is the determinant of A, a scalar quantity. Each element
aij of A has a minor mij and a cofactor cij . The minor of each element is the determinant of the
elements of A remaining after row i and column j are deleted, eg, m11 = a22a 33 a23a 32 ,
m22 = a11a 33 a13a 31 and m32 = a11a23 a13a21 . The cofactors cij = (1)i + j mij form a matrix C whose
a c
j =1
ij ij
The inverse J1 =
J1
adj J
J
CT
J
sin cos
+h
sin
=
( + h ) cos
cos cos
sin sin
+h
cos
( + h ) cos
cos sin
cos
+h
sin
(14)
+h
sin
=
( + h ) cos
h
cos cos
sin sin
+h
cos
( + h ) cos
cos sin
cos cos
sin 2 cos cos
cos
a +
f X
a
1 f
+h
cos sin
sin cos sin
0
a +
f Y
1 f
a
2
sin (1 f ) sin
sin
sin
1
f
Z
(
)
(
)
(15)
1
{X sin cos Y sin sin + Z cos
+h
e 2 sin cos
a
a
sin cos
+
e 2 sin 2 + 2 (1 e 2 )) f
1 f
{ }
(16)
manner
sin cos
(e 2 sin2 + + 2 (1 e 2 ))
1 f
sin cos
=
( (1 e 2 sin2 ) + + 2 (1 e 2 ))
1 f
3rd term =
but
1 e2
=
giving (1 e 2 sin 2 ) = (1 e 2 ) hence
1 e 2 sin 2
3rd term =
sin cos
+ (1 f )
+ (1 e 2 )) = sin cos
(
1 f
1 f
1
{X sin cos Y sin sin + Z cos
+h
e 2 sin cos
a
a
+ sin cos
+ (1 f ) f
1 f
(17)
Multiplying the right-hand-side of (15) and simplifying gives and expression for
=
1
(X sin + Y cos )
( + h ) cos
(18)
Multiplying the right-hand-side of (15) and simplifying gives and expression for h
h = X cos cos + Y cos sin + Z sin
e 2 sin 2
a
+ +
a
sin2 cos2
+
( sin2 2 ) sin 2 (1 f ) f
1 f
{ }
manner
2nd term =
and since =
a
1/ 2
(1 e 2 sin2 )
(1 e 2 sin2 )
a
then 1 e 2 sin 2 =
2nd term =
a2
and
2
(19)
{ }
manner
3rd term =
sin 2
( cos2 sin2 (1 f )2 + 2 (1 f )2 )
1 f
sin 2
( cos2 sin2 + e 2 sin2 + 2 (1 e 2 ))
1 f
sin 2
( (1 e 2 sin2 ) + 2 (1 e 2 ))
1 f
sin 2
( (1 e 2 ) + 2 (1 e 2 ))
1 f
sin 2
(1 e 2 )
1 f
Now, since 1 e 2 = (1 f )
a
a + (1 f ) sin 2 f
(20)
Equations (17), (18) and (20) are the standard Molodensky transformation formulae
1
{X sin cos Y sin sin + Z cos
+h
e 2 sin cos
a
a
+ sin cos
+ (1 f ) f
1 f
+
(21)
1
(X sin + Y cos )
( + h ) cos
a
a + (1 f ) sin 2 f
sin cos
+h
sin
=
( + h ) cos
h
cos cos
sin sin
+h
cos
( + h ) cos
cos
+h
e 2 sin cos
a ( + h )
cos sin
sin
10
X
sin cos
+ (1 f )
+ h 1 f
Y
Z
0
a
2
(1 f ) sin
f
(22)
Noting that b = a (1 f ) ,
b
a
1
an alternative presentation of the standard
= (1 f ) , =
a
b
(1 f )
1
{X sin cos Y sin sin + Z cos
RM + h
+
RN e 2 sin cos
a
a
b
+ sin cos RM + RN f
b
a
=
(23)
1
(X sin + Y cos )
(RN + h ) cos
a
b
a + RN sin 2 f
RN
a
(1 e
1/ 2
sin 2 )
(1 + x ) = 1 + nx +
n (n 1) 2 n (n 1)(n 2) 3
x +
x +"
2!
3!
giving
1/ 2
(1 e 2 sin2 )
2
1 1 3
1
= 1 + (e 2 sin 2 ) + (e 2 sin 2 ) + "
2
2 2 2
1
3
= 1 + e 2 sin 2 + e 4 sin 4 + "
2
8
11
(24a)
1/ 2
(1 e 2 sin2 )
=1+
2
1
3
(2 f f 2 ) sin2 + (2 f f 2 ) sin 4 + "
2
8
The flattening f is a small quantity ( f 0.003) and f 2 is exceedingly small ( f 2 0.00001) ; hence, in
the equation above, ignoring terms containing f 2 , f 3 , f 4 , etc
1/ 2
(1 e 2 sin2 )
1 + f sin 2
(24b)
1 + f sin 2
a
(24c)
Other approximations may be derived from (24c), again ignoring terms f 2 , f 3 , f 4 , etc
e 2
1
e 2 (1 + f sin 2 )
2a
2
1
= (2 f f 2 )(1 + f sin 2 )
2
1
= (2 f + 2 f 2 sin 2 f 2 f 3 sin 2 )
2
f
(24d)
(1 f ) a (1 f )(1 + f sin 2 )
= a af + af sin 2 af 2 sin 2
(24e)
a af + af sin
a (1 e 2 )
3/2
(1 e 2 sin2 )
(25a)
(1 e 2 sin2 )
2
1 3 5
3
= 1 + (e 2 sin 2 ) + (e 2 sin2 ) + "
2
2 2 2
3 2
15
= 1 + e sin 2 + e 4 sin 4 + "
2
8
(1 e 2 sin2 )
= 1+
2
3
15
(2 f f 2 ) sin2 + (2 f f 2 ) sin 4 + "
2
8
(1 e 2 sin2 )
1 + 3 f sin 2
(25b)
1 + 3 f sin2
a (1 e 2 )
(25c)
12
(1 f )
a (1 f )(1 + 3 f sin 2 )
= a + 3af sin 2 af 3af 2 sin 2
(25d)
a af + 3af sin 2
Approximations given in equations (24) and (25) are used in the following derivation of the abridged
Molodensky transformation formulae
Equation for :
The 2nd and 3rd lines of the equation for in (21) can be combined as
e 2
(1 f )
a +
f
2 sin cos
+
2
2 (1 f )
2a
{ }
(26)
e 2
a f a
2a
(27a)
Using the approximations in (24e) and (25d) the second term in the braces
{ }
as
(1 f )
1
2
2
= (a af + 2af sin 2 ) f
(27b)
f is a small quantity ( f 0.003) and f = f1 f2 , the difference in ellipsoid flattening, will be very
small ( f 8 108 ) and products f f 0 . Hence, (27b) can be written as
(1 f )
+
f a f
2 (1 f )
2
(27c)
Substituting (27a) and (27c) into equation (26) and noting that 2 sin cos = sin 2 we have an
approximation
e 2
(1 f )
+
2 sin cos
a +
f f a + a f sin 2
2
2 (1 f )
2a
Using this approximation and ignoring the ellipsoidal height h, we may write the equation for as
=
1
{ X sin cos Y sin sin + Z cos + f a + a f sin 2}
13
(28)
Equation for :
The equation for in (21) is modified by omitting the parameter h giving
=
1
(X sin + Y cos )
cos
(29)
Equation for h :
In the equation for h in (21) the last two terms can be written as
a
A = a + (1 f ) sin 2 f
(30a)
1 f sin 2
(30b)
(30c)
(30d)
(31)
Equations (28), (29) and (31) are the abridged Molodensky transformation formulae
1
{X sin cos Y sin sin + Z cos + f a + a f sin 2}
1
(X sin + Y cos )
cos
(32)
14
REFERENCES
Badekas, J., 1969. Investigations Related to the Establishment of a World Geodetic System, Report
124, Department of Geodetic Science, Ohio State University, Columbus, Ohio, USA.
ICSM, 2003. Geocentric Datum of Australia Technical Manual - Version 2.2, Intergovernmental
Committee on Surveying & Mapping (ICSM), February 2002, available online at:
www.icsm.gov.au/icsm/gda/gdatm/index.html
15
Sokolnikoff, I.S. and Redheffer, R.M., 1966. Mathematics of Physics and Engineering, 2dn edn,
International Student Edition, McGraw-Hill, Inc., London.
Soler, T., 1976. On Differential Transformations between Cartesian and Curvilinear (Geodetic)
Coordinates, Report No. 236, Department of Geodetic Science, Ohio State University,
Columbus, Ohio, USA.
16
APPENDIX 1
TEST VALUES FOR MOLODENSKY TRANSFORMATION
Transformation from Australian Geodetic Datum 1966 (AGD66) to World Geodetic System 1984
(WGS84).
AGD66 Geodetic coordinates:
a = 6378160 m
X = ( + h ) cos cos
= 4131857.9379 m
Y = ( + h ) cos sin
= +2896741.9218 m
f = 1 298.25
Z = ( (1 e 2 ) + h ) sin = 3887971.3157 m
e 2 = f (2 f )
a
=
1/ 2
2
(1 e sin2 )
= 6.69454185459 103
= 6386195.1797 m
X
X
X
= Y
+ Y
Y
Z
WGS 84 Z AGD 66 Z
X
134
= Y
+ 48
AGD 66 +149
X = 4131991.9379 m
Y = +2896693.9218 m
Z = 3887822.3157 m
a = 6378137 m
tan =
f = 1 298.257223563
Z + ve 2 sin
r
Y
tan =
X
r
h=
cos
r = X 2 +Y 2
17
(1 e
1/ 2
sin 2 )
e 2 = f (2 f ) = 6.69437999014 103
n
Iteration n
6386171.9561 m
n +1
37 48 00.0000
37D 47 54.5522
.4079
37D 47 54.5294
37D 47 54.5294
.4056
37D 47 54.5293
37D 47 54.5293
.4056
37D 47 54.5293
= 37D 47 54.5293
= +144D 58 04.7508
h = 46.382 m
18
37 47 54.5522
D
APPENDIX 2
STANDARD MOLODENSKY TRANSFORMATION
EXAMPLE
Australian Geodetic Datum 1966 (AGD66)
TO
World Geodetic System 1984 (WGS84)
Formulae:
=
1
{X sin cos Y sin sin + Z cos
+h
e 2 sin cos
a
a
+ sin cos
+ (1 f ) f
1 f
1
(X sin + Y cos )
( + h ) cos
a
a + (1 f ) sin 2 f
a = 6378160 m
f = 1 298.25
a = 6378137 m
f = 1 298.257223563
X = 134 m
Y = 48 m
Z = +149 m
a = 23 m
f = 8.120449 108
a
1/ 2
(1 e sin2 )
a (1 e 2 )
=
3/2
(1 e 2 sin2 )
2
19
= 6386195.179722 m
= 6359435.481976 m
1
{67.249167380 16.888371512 + 117.733096844
+h
+0.074662477
+ 0.501242257}
168.669797446
+h
1
{76.923088948 + 39.303274197}
( + h ) cos
20
= 46.378 m
APPENDIX 3
ABRIDGED MOLODENSKY TRANSFORMATION
EXAMPLE
Australian Geodetic Datum 1966 (AGD66)
TO
World Geodetic System 1984 (WGS84)
Formulae:
=
1
{X sin cos Y sin sin + Z cos + f a + a f sin 2}
1
(X sin + Y cos )
cos
a = 6378160 m
f = 1 298.25
a = 6378137 m
f = 1 298.257223563
X = 134 m
Y = 48 m
Z = +149 m
a = 23 m
f = 8.120449 108
a
1/ 2
(1 e sin2 )
a (1 e 2 )
=
3/2
(1 e 2 sin2 )
= 6386195.179722 m
= 6359435.481976 m
1
{67.249167380 16.888371512 + 117.733096844
168.670249809
21
1
{76.923088948 + 39.303274197}
cos
22
= 46.378 m