Geodesia Geometrica MIT
Geodesia Geometrica MIT
Geodesia Geometrica MIT
N
GEOMETRIC
Gre
en
wi
ch
GEODESY
b
PART A Q
a
H
equator
RMIT University
Geospatial Science
FORWARD
These notes are an introduction to ellipsoidal geometry related to geodesy.
Many
computations in geodesy are concerned with the position of points on the Earth's surface
and direction and distance between points. The Earth's surface (the terrestrial surface) is
highly irregular and unsuitable for any mathematical computations, instead a reference
surface, known as an ellipsoid a surface of revolution created by rotating an ellipse about
its minor axis is adopted and points on the Earth's terrestrial surface are projected onto
the ellipsoid, via a normal to the ellipsoid.
projected points on the ellipsoidal reference surface; hence there is a need to understand
the geometry of the ellipsoid.
These notes are intended for undergraduate students studying courses in surveying,
geodesy and map projections. The derivations of equations given herein are detailed, and
in some cases elementary, but they do convey the vital connection between geodesy and
the mathematics taught to undergraduate students.
The information in the notes is drawn from a number of sources; in particular we have
followed closely upon the works of G. B. Lauf, Geodesy and Map Projections and R. H.
Rapp, Geometric Geodesy, and also 'Geodesy' a set of notes produced by the New South
Wales Department of Technical and Further Education (Tafe).
RMIT University
Geospatial Science
TABLE OF CONTENTS
page
1.
1.1
THE ELLIPSE
1.1.1
1.1.2
10
1.1.3
10
1.1.4
10
1.1.5
11
1.1.6
12
x and y in terms of
13
15
15
16
1.1.7
Curvature
17
1.1.8
Radius of curvature
18
1.1.9
Centre of curvature
20
1.2
21
24
1.2.1
24
1.2.2
28
1.2.3
32
33
35
ii
RMIT University
1.2.4
1.3
Geospatial Science
Normal curvature
37
Meuisner's Theorem
38
40
42
Surfaces of Revolution
44
Euler's equation
47
THE ELLIPSOID
48
1.3.1
52
55
56
56
57
58
58
58
59
59
Meridian distance
60
61
Binomial series
61
Quadrant distance
65
1.3.2
66
69
70
71
73
75
Newton-Raphson Iteration
75
1.3.3
78
1.3.4
81
1.3.5
Volume of ellipsoid
82
1.3.6
83
83
iii
RMIT University
83
84
84
84
85
1.3.8
86
1.3.9
87
1.3.7
2.
3.
Geospatial Science
89
2.1
91
2.2
92
2.2.1
Successive Substitution
95
2.2.2
Newton-Raphson Iteration
96
2.2.3
Bowring's method
97
2.2.4
98
2.2.5
Paul's method
102
MATLAB FUNCTIONS
106
3.1
ELLIPSOID CONSTANTS
106
ellipsoid_1.m
106
MERIDIAN DISTANCE
109
mdist.m
110
latitude.m
112
latitude2.m
114
117
118
Cart2Geo.m
118
120
Geo2Cart_Substitution.m
120
Geo2Cart_Newton.m
123
Geo2Cart_Bowring.m
126
Geo2Cart_Lin.m
129
Geo2Cart_Paul.m
132
3.2
3.3
3.4
iv
RMIT University
Geospatial Science
radii.m
4.
135
REFERENCES
138
RMIT University
Geospatial Science
ch
Gre
en
wi
b
a
h
Q
H
equ ator
a
y
RMIT University
Geospatial Science
Figure 1 show a schematic view of the reference ellipsoid upon which meridians (curves of
constant longitude ) and parallels (curves of constant latitude ) form an orthogonal
network of reference curves on the surface.
coordinated via a normal to the ellipsoid passing through P. This normal intersects the
surface at Q which has coordinates of , and P is at a height h = QP above the ellipsoid
surface. We say that P has geodetic coordinates, , ,h . P also has Cartesian coordinates
x,y,z; but more about these coordinate systems later. The important thing at this stage is
that the ellipsoid is a surface of revolution created by rotating an ellipse about its minor
axis, where this minor axis is assumed to be either the Earth's rotational axis, or a line in
space close to the Earth's rotational axis. Meridians of longitude are curves created by
intersecting the ellipsoid with a plane containing the minor axis and these curves are
ellipses; as are all curves on the ellipsoid created by intersecting planes. Note here that
parallels of latitude (including the equator) are circles; since the intersecting plane is
perpendicular to the rotational axis, and circles are just special cases of ellipses. Clearly,
an understanding of the ellipse is important in ellipsoidal geometry and thus geometric
geodesy.
The ellipse, parabola and hyperbola, as sections of a cone, were first studied by Menaechmus (circa 380 BC
- 320 BC), the Greek mathematician who tutored Alexander the Great. Euclid of Alexandra (circa 325 BC 265 BC) investigated the ellipse in his treatise on geometry: The Elements. Apollonius of Perga (circa 262
BC - 190 BC) in his famous book Conics introduced the terms parabola, ellipse and hyperbola and Pappus of
Alexandra (circa 290 - 350) introduced the concept of focus and directrix in his studies of projective
geometry.
apex
nappes
elem
ent
gen
e
rato
r
Geospatial Science
axis
RMIT University
The conic sections are the curves created by the intersections of a plane with one or two
nappes of the cone.
hyperbola
parabola
ellipse
circle
Depending on the angle between the axis of the cone and the plane, the conic sections
RMIT University
Geospatial Science
three ways:
Pk
axis
(a)
r'
major
2b
F'
P
3 P
2
P1
axis
minor
(1)
axis and d = OF = OF
2d
is the focal
2a
equation of the ellipse. From Figure 4 and equation (1) we may write
(x + d ) + y 2 + (x d ) + y 2 = 2a
2
= 4a 2 4xd
and
d
2
(x d ) + y 2 = a x
a
x
+ y2 = a2 d 2
a 2
2
(2)
Now, from Figure 4, when Pk on the ellipse is also on the minor axis, r = r = a and from
a right-angled triangle we obtain
RMIT University
Geospatial Science
a 2 = b2 + d 2;
b 2 = a 2 d 2;
d 2 = a 2 b2
(3)
Substituting the second of equations (3) into equation (2) and simplifying gives the
Cartesian equation of the ellipse
x 2 y2
+
=1
a 2 b2
(b)
and
are
x 2 + y2 = a2
If auxiliary circles
x 2 + y 2 = b2
(4)
drawn
on
angles
and
x P2
yP2
+
=1
a2
b2
and x P2 = a 2
a2 2
yP
b2
(5)
Now the x- coordinates of A and P are the same and so the right-hand sides of equations
(5) may be equated, giving
RMIT University
Geospatial Science
a2 2
yP
b2
a 2 yA2 = a 2
This leads to the relationship
yP =
b
yA
a
(6)
Hence, we may say that the y-coordinate of the ellipse, for an arbitrary x-coordinate, is b a
times the y-coordinate for the circle of radius a at the same value of x.
Now, we can use equation (6) and Figure 5 to write the following equations
xP = xA
x A = a cos
and
yA = a sin
yP =
b
yA
a
x = a cos
(7)
y = b sin
Similarly, considering points B and P; using equation (4) and the equation for the
auxiliary circle of radius b we may write
x B2 + yB2 = b 2
and
x P2
yP2
+
=1
a2
b2
and yP2 = b 2
b2 2
xP
a2
(8)
Now the y- coordinates of B and P are the same and so the right-hand sides of equations
(8) may be equated, giving
b 2 x B2 = b 2
b2 2
xP
a2
a
xB
b
(9)
RMIT University
Geospatial Science
And we may say that the x-coordinate of the ellipse, for an arbitrary y-coordinate, is a b
times the x-coordinate for the circle of radius b at the same value of y.
Using equation (9) and Figure 5 we have
x B = b cos
yB = b sin
a
xB
b
y P = yB
xP =
and
and
= sin2 . Then using the trigonometric
a2
b2
x 2 y2
identity sin2 + cos2 = 1 we obtain the Cartesian equation of the ellipse: 2 + 2 = 1 .
a
b
these can be re-arranged as
(c)
An ellipse may be defined as the locus of a point P that moves so that its distance
from a fixed point F, called the focus, bears a constant ratio, that is less than unity, to its
distance from a fixed line known as the directrix, i.e.,
PF
=e
PN
(10)
F'
latus
E'
x
E
L
p
G'
2a
directrix
rectum
D'
RMIT University
Geospatial Science
From Figure 6 and definition (c), the following relationships may be obtained
FE
=e
EL
FE
=e
E L
and
a
e
(11)
Also
FE FE = e (E L EL )
EE 2 (FE ) = e (EE )
EE (1 e ) = 2 (FE )
2a (1 e ) = 2 (FE )
hence
FE = a (1 e )
(12)
(13)
In Figure 6, the line GG', perpendicular to the major axis and passing through the focus F
is known as the latus rectum2 and l = FG is the semi latus rectum.
Using equations (11) and (13), the perpendicular distance from G to the directrix DD' is
a
l
OL OF = ae , and employing definition (c) gives a
= e and the semi latus
e
ae
e
rectum of the ellipse is
l = a (1 e 2 )
(14)
Latus rectum means "side erected" and the length of the latus rectum was used by the ancient Greek
RMIT University
Geospatial Science
p=
(15)
Also, in Figure 6, let PF = r , be the angle between PF and the x-axis; then
PF
r
PN = p r cos . Using definition (c),
= e , hence = p r cos , which can be rePN
e
arranged to give a polar equation of an ellipse (with respect to the focus F)
r=
ep
1 + e cos
(16)
r=
a (1 e 2 )
(17)
1 + e cos
l
1 + e cos
(18)
x = r cos
y = r sin
P
x
x 2 = r 2 cos2
and
y 2 = r 2 sin2
r=
ab
a sin + b 2 cos2
2
(19)
(20)
RMIT University
Geospatial Science
OF
a
e=
(21)
a 2 b2
. The more
a
familiar way that eccentricity e is defined in geodesy is by its squared-value e 2 as
From Figure 2 and equations (3) OF = d = a 2 b 2
and e =
a 2 b2
b2
=
1
a2
a2
e2 =
(22)
e 2 =
a 2 b2
a2
=
1
b2
b2
(23)
f =
a b
b
= 1
a
a
(24)
a2
b
(25)
a 2 b2
a 2 + b2
(26)
a b
a +b
(27)
c=
m=
n=
Note: c is the polar radius of the ellipsoid and m is sometimes called the 3rd-eccentricity squared. A 2ndflattening is defined as f = (a b ) b with a 3rd-flattening as f = n = (a b )
(a + b ) .
10
RMIT University
Geospatial Science
a = b 1 + e 2 =
1 n
= c
= c (1 f )
1 + n
1 e
b
(28)
c
= c (1 e 2 )
1 + e 2
(29)
b
1
e
1n
1m
= (1 f ) = 1 e 2 =
= =
=
2
1+m
a
e 1 + n
1 + e
(30)
b = a (1 f ) = a 1 e 2 =
f = 1 1 e2 = 1
1 + e
1
1 + e
2n
1+n
(31)
e 2
4n
2m
= f (2 f ) =
2 =
2
1 + e
(1 + n )
1+m
e2 =
(32)
1 e 2 = (1 f )
e 2 =
(33)
f (2 f )
e2
4n
2m
=
2 =
2 =
2
1 e
(1 n )
1m
(1 f )
(34)
(1 e 2 )(1 + e 2 ) = 1
(35)
m=
f (2 f )
2n
2 =
1 + n2
1 + (1 f )
(36)
n=
f
1 1 e2
1 + e 2 1
=
=
2f
1 + 1 e2
1 + e 2 + 1
(37)
F'
a
b
a(1-e 2)
ae
O
a(1-e)
2a
11
RMIT University
Geospatial Science
tangent to ellipse
A
Q
P
x
a
90
+
a
E'
nor
ma
l
H
Q'
auxiliary circle
x 2 + y 2 = a2
In Figure 9, the angles , and are known as latitudes and are respectively, angles
between the major axis of the ellipse and (i) the normal to the ellipse at P, (ii) a normal to
the auxiliary circle at A, and (iii) the radial OP. The x,y Cartesian coordinates of P can
be expressed as functions of and relationships between , and established. These
functions can then be used to define distances PH, PD and OH in terms of the ellipse
parameters a and e 2 . These will be useful in later sections of these notes.
12
RMIT University
Geospatial Science
x and y in terms of
Differentiating equation (4) with respect to x gives
2x 2y dy
+ 2
=0
a2
b dx
and re-arranging gives
dy
b2 x
= 2
dx
a y
Now by definition,
dy
is the gradient of the tangent to the ellipse, and from Figure 9
dx
dy
b2 x
= tan (90 + ) = cot = 2
dx
a y
y =
and
x=
b2
x tan
a2
a2 y
b 2 tan
(38)
(39)
(40)
Substituting equation (39) into the Cartesian equation for the ellipse (4) gives
x 2 b2 2
+ x tan2 = 1
a2 a4
x 2
b 2 sin2
1 + 2
=1
a 2
a cos2
Now, from equation (30)
b2
= 1 e 2 hence
a2
2
x 2
2 sin
1
+
1
e
(
)
=1
a 2
cos2
=1
a 2
cos2
x 2 1 e 2 sin2
=1
a 2 cos2
giving
x=
a cos
(1 e 2 sin2 )
12
(41)
13
RMIT University
Geospatial Science
Similarly, substituting equation (40) into the Cartesian equation for the ellipse (4) gives
a2 y2
y2
+
=1
b 4 tan2 b 2
y 2
a 2 cos2
1 + 2
=1
b 2
b sin2
Now, from equation (30)
a2
1
=
hence
2
b
1 e2
y 2
cos2
=1
1
+
2
2
2
b
(1 e ) sin
y 2 sin2 e 2 sin2 + cos2
=1
b 2
(1 e 2 ) sin2
y 2 1 e 2 sin2
=1
b 2 (1 e 2 ) sin2
giving
y=
b 1 e 2 sin
1 e 2 sin2
12
(42)
Equations (41) and (42) may be conveniently expressed as another set of parametric
equations for the ellipse
a
cos
W
b 1 e2
y=
sin
W
W 2 = 1 e 2 sin2
x =
(43)
W2 = 1
14
RMIT University
Geospatial Science
1 + e 2 cos2
.
Putting V 2 = 1 + e 2 cos2 and using equation (30) gives
1 + e 2
V2
b2 2
W2 =
=
V . Using these relationships gives another set of parametric equations
1 + e 2
a2
for the ellipse
and W 2 =
a 2 cos
c
= cos
bV
V
b
y = sin
V
a2
c=
b
2
V = 1 + e 2 cos2
x =
(44)
a2
b
b2 2 b 2
V2
2
2
W = 2 V = V = V (1 e ) =
a
c
1 + e 2
2
V2 =
a2 2 c 2
W2
W
=
W
=
= W 2 (1 + e 2 )
2
2
b
b
1 e
(45)
(46)
(47)
PH =
x
a
a
c
=
=
=
1
2
cos (1 e 2 sin2 )
W V
(48)
(49)
(50)
a 2
c
e sin = e 2 sin
W
V
(51)
DH = PH PD =
Length OH along minor axis
OH = DH sin =
15
RMIT University
Geospatial Science
dy
= b cos
d
then
dy
dy d
b
=
= cot
dx
d dx
a
Now by definition,
(52)
dy
is the gradient of the tangent to the ellipse, and
dx
dy
= tan (90 + ) = cot
dx
(53)
b
tan = 1 e 2 tan = (1 f ) tan
a
(54)
tan =
giving relationships between and
tan = (1 e 2 ) tan =
b2
2
tan = (1 f ) tan
2
a
(55)
And with equations (54) and (55) the relationships between and are
tan = 1 e 2 tan =
b
tan = (1 f ) tan
a
(56)
16
RMIT University
Geospatial Science
1.1.7 Curvature
To calculate distances on ellipses (and ellipsoids) we need to know something about the
curvature of the ellipse. Curvature at a point on an ellipse can be determined from general
relationships applicable to any curve.
y = y (x)
curv
e
s 0
d
=
s
ds
P1
tangent
(57)
d 2y
d
d ds
= sec2
= sec2
2
dx
dx
ds dx
P2
(58)
d
and from the elemental triangle
ds
ds
dx
dy
we obtain
ds
1
d 2y
=
= sec . Substituting these results into equation (58) gives
= sec3 and
2
dx
cos
dx
a re-arrangement gives the curvature as
d 2y
2
= dx3
sec
(59)
The denominator of equation (59) can be simplified by using the trigonometric identity
sec2 = 1 + tan2 ; so sec = (1 + tan2 )
12
2
2
dy
dy
dy
= tan , then = tan2 , thus sec3 = 1 + . This result for sec3 can
dx
dx
dx
32
17
RMIT University
Geospatial Science
d 2y
y
dx 2
=
=
3
2
2
2
dy
1 + (y )
1
+
dx
where
32
(60)
dy
d 2y
= y and
= y
dx
dx 2
y
y = y (x)
(61)
C(u,v)
vy
no
r
m
al
xu
e
circl
P
(x,y)
ta
ng
e
nt
curv
e
dy
cos
= cot =
dx
sin
(62)
d 2y
d 1 d
1 d
=
=
2
dx
d tan dx
sin 2 dx
(63)
y =
and the 2nd-derivative is
y =
The derivative
d
can be obtained from equation (41) where
dx
x=
a cos
(1 e 2 sin2 )
12
18
RMIT University
Geospatial Science
d u
=
and using the quotient rule for differentiation:
dx v
du
dv
u
dx
dx gives
v2
a sin
{1 e (sin
2
(1 e sin )
a (1 e 2 ) sin
=
32
(1 e 2 sin2 )
2
32
12
) 2e
sin cos
+ cos2 )}
hence
2
2
d (1 e sin )
=
dx
a (1 e 2 ) sin
32
(64)
32
(65)
23
23
23
cos2 a
= 1 +
sin2
23
23
(1 e 2 )
23
sin2
1 e 2 sin2
(1 e 2 )
23
1 e 2 sin2
a (1 e 2 )
(1 e
sin )
2
32
a (1 e 2 )
W
c
V3
(66)
Note that equations (45), (46) and (47) have been used in the simplification.
19
RMIT University
Geospatial Science
tangent: y y 0 = m (x x 0 )
normal:
1
(x x 0 )
m
dy
m = tan =
= y
dx
y y0 =
(67)
The centre of curvature C (u, v ) lies (i) on the normal passing through P (x , y ) and (ii) at a
distance from P measured towards the concave side of the curve y = y (x ) .
This leads to two equations:
v y =
(equation of normal)
1
(u x )
y
(68)
(1 + y 2 )
(Pythagoras)
(u
x )2 + (v y ) = =
2
y 2
(69)
(1 + y 2 )
(y )
(v y ) + (v y ) =
2
y 2
(1 + y 2 )
(v y ) {1 + y
2
}=
y 2
(1 + y 2 )
(v y ) =
2
y 2
and
v y =
1 + y 2
y
(70)
20
RMIT University
Geospatial Science
Note that when the curve is concave upward, y > 0 and since C lies above P then
v y > 0 and the proper sign in equation (70) is +. This is also the case when y < 0
and the curve is concave downward so
v y =
1 + y 2
y
(71)
(72)
Re-arranging equations (71) and (72) gives the equations for the coordinates (u, v ) of the
centre of curvature C as
u=x
y (1 + y 2 )
y
(73)
1 + y 2
v =y+
y
1.1.10
centres of curvature.
P3
evolute
C3
C1
P2
P1
C2
ellipse
auxiliary circle
21
RMIT University
Geospatial Science
x = a cos ;
y = b sin
dx
= a sin ;
d
dy
= b cos
d
and the chain-rule for differentiation gives the gradient of the tangent to the ellipse as
dy
dy d
b
=
=
dx
d dx
a tan
y =
or
b
a tan
(74)
The second-derivative is
d 2y
b
d
b
=
= 2
2
2
dx
a sin dx
a sin 3
or
y =
b
a sin 3
2
(75)
Substituting equations (74) and (75) into the equations for the centre of curvature (73)
gives
b cos b 3 cos3
y (1 + y 2 )
a sin a 3 sin 3
= x
u=x
2
3
a
sin
b cos b 3 cos3
a sin a 3 sin 3
u = x
2
3
a
sin
(a 2 sin 3 )
= x
a 3 sin 3
b
2
2
2
3
a sin cos + b cos
= x
22
RMIT University
Geospatial Science
then
au = (a 2 b 2 ) cos3
(76)
Similarly, substituting equations (74) and (75) into the equations for the centre of
curvature (73) gives
b 2 cos2
1
+
1 + y 2
a 2 sin2
v =y+
= y +
a 2 sin 3
b 2 cos2
a sin
v = y + 1 + 2
2
a sin
b
= y +
2
2
a sin
b
2
3
2
2
a sin + b cos sin
= y
2
3
2
2
by {a sin + b (1 sin ) sin }
=
b
bv = (a 2 b 2 ) sin 3
(77)
Using equations (76) and (77); and equations (22) and (23), a set of parametric equations
of the evolute of an ellipse are
a 2 b 2 3
x =
cos = ae 2 cos3
a
a 2 b 2 3
y =
sin = be 2 sin 3
b
(78)
23
RMIT University
Geospatial Science
normal sections of the ellipsoid and equations for mean radii of curvature.
Differential
geometry relies heavily on vector representation of curves and surfaces and the two vector
products; the dot (or scalar) product and the cross (or vector) product. Some familiarity
with these terms (and manipulations) and vector notation is assumed.
r (t ) = x (t ) i + y (t ) j + z (t ) k
curve
s
ce
(79)
r +r
C.
dx
dy
dz
ds
ds
dr
dr dr
= + + or
=
=
i
and s =
dt
dt
dt
dt
dt
dt
dt dt
2
dr dr
i dt
dt dt
a,b
are magnitudes or lengths of the vectors, is the angle between them and the dot product is
a scalar quantity equal to the projection of the length of a onto b. If a is orthogonal to b,
then a i b = 0 .]
24
RMIT University
Geospatial Science
Let Q, a small distance s along the curve from P, have a position vector r + r . Then
r
r = PQ and r s . Both when s is positive or negative
approximates to a unit
s
dr
vector in the direction of s increasing and
is a tangent vector of unit length denoted by
ds
t ; hence
t = dr = dx i + dy j + dz k
ds
ds
ds
ds
(80)
Since t is a unit vector then tit = 1 and differentiating with respect to s using the rule
dt
d
dt
dt
d
dv
du
dt
(uv ) = u
+v
gives
tit) = ti + ti = 2 ti = 0 . This leads to ti = 0
(
ds
ds
ds
ds
dx
dx
dx
ds
from which we deduce that
dt
is a vector orthogonal to t and write
ds
dt
,
= k = n
ds
>0
(81)
dt
is called the curvature vector k, and should not be confused with the unit vector in the
ds
is a unit vector called the principal normal vector, the
direction of the z-axis. n
1
curvature and
= is the radius of curvature. The circle through P, tangent to t with
dt
i = ; i.e., n
is the unit vector in
this radius is called the osculating circle. Also n
ds
the direction of k.
be a third unit vector defined by the vector cross product
Let b
= t n
(82)
a = a 1i + a 2 j + a 3 k
and
b = b1i + b2 j + b3 k , then
= p.
a b = a b sin p
a,b
are
screw rule, i.e., if a and b are in the plane of the head of a screw, then a clockwise rotation of
a to b through an angle would mean that the direction of p would be the same as the
direction of advance of a right-handed screw turned clockwise.
25
RMIT University
Geospatial Science
(+)
()
(+)
p = a b = a1
a2
b1
b2
b3
Note here that the mnemonics (+), () , (+) are an aid to the evaluation of the determinant.
The perpendicular vector p = p1i + p2 j + p3 k has scalar components p1 = (a 2b3 a 3b2 ) ,
2
2
2
3
dn
db
dt
dn
dn
) = n
+ t
n
+ t
= (t n
= n
= t
ds ds
ds
ds
ds
ds
then
dn
tidb = tit dn
= i(t t) = 0
ds ds
ds
db
idb = 0 so that db is
ib
= 1 it follows that b
is orthogonal to t . But from b
ds
ds
ds
.
orthogonal to b and so is in the plane containing t and n
so that
db
, and
is in the plane of t and n
ds
is orthogonal to t , it must be parallel to
db
. The direction of
as it
n
is opposite n
ds
db
must be to ensure the cross product
t
ds
. Hence
is in the direction of b
osculating plane
Since
db
,
= n
ds
>0
rectifying
plane
(83)
torsion, and
the radius of torsion. t , n
^t
^
n
r
j
normal plane
^
b
^
Figure 14: The tangent ^t, principal normal n
and binormal ^
b to a space curve
26
RMIT University
Geospatial Science
is
is the osculating plane, the plane containing n
and b
The plane containing t and n
is the rectifying plane. Figure 14
the normal plane and the plane containing t and b
shows these orthogonal unit vectors for a space curve.
t and the derivative with respect to s is
=b
Also n
d
dn
db
dt = n
n
t
t + b
= b
= ( b t) =
t + b
ds ds
ds
ds
(84)
Equations (81), (83) and (84) are known as the Frenet-Serret formulae.
dt
= n
ds
db
= n
ds
dn
t
= b
ds
(85)
dt ds
0 0 t
ds = 0 0 b
db
dn
ds 0 n
(86)
or in matrix notation
Frenet
derived these formulae in his doctoral thesis at the University of Toulouse; the latter part
of which was published as 'Sur quelques proprits des courbes double courbure', (some
properties of curves with double curvature) in the Journal de mathmatiques pures et
appliqus (Journal of pure and applied mathematics), Vol. 17, pp.437-447, 1852. Frenet
also explained their use in a paper titled 'Thormes sur les courbes gauches' (Theorems on
awkward curves) published in 1853.
same formulae in 'Sur quelques formules relatives la thorie des courbes double
courbure' (Some formulas relating to the theory of curves with double curvature) published
in the J. de Math. Vol. 16, pp.241-254, 1851 (DSB 1971).
27
RMIT University
Geospatial Science
dt
d dr
d 2r
t n
= t = t = t 2
ds
ds ds
ds
(87)
, so t n
and also, from equation (80), t = dr ;
=b
= b
Now, from equation (82), t n
ds
so equation (87) becomes
=
b
dr d 2 r
ds ds 2
(88)
= b
= ; so taking the magnitude of both sides
is a unit vector, then b
Now, since b
of equation (88) gives an expression for the curvature as
dr d 2 r
ds ds 2
(89)
latitude
replacing
^t
the
general parameter t,
r ( ) = x ( ) i + y ( ) j + z ( ) k
evolute
(90)
t and n
are the unit tangent vector and
unit normal vector respectively, both of
^
n
x
C
ellipse
auxiliary circle
is directed
parametric latitude and n
towards the centre of curvature C.
Figure 15:
28
RMIT University
Geospatial Science
dr
dr d
=
ds d ds
(91)
And
d 2r
d dr d dr d
= =
2
ds
ds ds ds d ds
dr d d d d dr
+
d ds ds ds ds ds
dr d 2 d d dr d
=
+
d ds 2
ds d d ds
=
dr d 2 d 2 r d
+ 2
d ds 2
d ds
(92)
Now, substituting equations (91) and (92) into equation (89) gives
dr d dr d 2 d 2 r d 2
=
+
d ds d ds 2
d 2 ds
2
dr d dr d 2 dr d d 2 r d
d ds d ds 2 d ds d 2 ds
3
dr d 2 r d
= 0 + 2
d d ds
dr d 2 r
d d 2
d
ds
(93)
d
can be determined as follows.
ds
From
t = dr = dr d = dx i + dy j + dz k d
ds d ds
d
d ds
d
Taking the dot product of the unit vector t with itself gives
tit = 1 = dx + dy + dz
d
d
d
d 2
ds
(94)
29
RMIT University
Geospatial Science
Substituting equation (94) into equation (93) gives the expression for curvature as
dr d 2 r
d d 2
dr
d
(95)
We can now use equation (95) to derive an equation for radius of curvature =
1
.
Parametric equations of the ellipse in the x-y plane are [see equations (7)]
x = x ( ) = a cos
y = y ( ) = b sin
z = z ( ) = 0
r = a cos i + b sin j + 0 k
The derivatives are
dr
= a sin i + b cos j + 0 k
d
d 2r
= a cos i b sin j + 0 k
d 2
and the cross product in equation (95) is
(+)
()
dr d 2 r
= a sin b cos
d d 2
a cos b sin
(+)
and
2
dr d 2 r
2 = 02 + 02 + (ab sin2 + ab cos2 ) = ab
d d
3
32
dr
= (a 2 sin2 + b 2 cos2 )
d
Substituting these results into equation (95) and taking the reciprocal gives
(a 2 sin2 + b 2 cos2 )
32
ab
(96)
30
RMIT University
Geospatial Science
The term a 2 sin2 + b 2 cos2 in equation (96) can be simplified in the following manner
a 2 sin2
a sin + b cos =
cos2 + b 2 cos2
2
cos
2
= cos2 (a 2 tan2 + b 2 )
(97)
Using equation (54) that gives the relationships been tan and tan we may write
a 2 tan2 = b 2 tan2
(98)
and from the parametric equations of an ellipse (7) and equations (43) we equate the xcoordinate, which leads to
cos2
1 e 2 sin2
cos2 =
(99)
cos2
(b 2 tan2 + b 2 )
1 e 2 sin2
cos2
=
b 2 (1 + tan2 ))
(
2
2
1 e sin
cos2
=
(b 2 sec2 )
2
2
1 e sin
=
b2
1 e 2 sin2
(100)
b2
1 e 2 sin2
(101)
b2
2
2
1 e sin
32
ab
b2
a
32
2
a (1 e 2 sin2 )
a (1 e 2 )
(1 e
sin )
2
32
a (1 e 2 )
W
c
V3
(102)
This is identical to equation (66) which was derived from classical methods.
31
RMIT University
Geospatial Science
rv
^
N
v = constant
ru
and v-curves.
The vectors
r
x
y
z
=
i+
j+
k
u
u
u
u
r
x
y
z
rv =
i+
j+
k
=
v
v
v
v
ru =
(104)
are both tangent vectors to the surface S and ru is tangential to the parametric curve
v = constant and rv is tangential to the parametric curve u = constant . ru and rv are
not unit vectors and they do not coincide in direction (except perhaps at an isolated point)
so that in general ru rv is not a null vector. Higher order derivatives are expressed as
ruu =
r 2 r
r 2 r
r
2r
=
,
r
=
=
,
r
=
=
, etc
vv
uv
u u u 2
v v v 2
v u u v
(105)
Using the Theorem of the Total Differential (Sokolnikoff & Redheffer 1966) we may write
dr =
r
r
du +
dv = ru du + rv dv
u
v
(106)
32
RMIT University
Geospatial Science
r
r
r
r
du +
dv du +
du +
dv dv
u u
v
v u
v
d 2r =
(107)
)(
= ru du + rv dv i ru du + rv dv
= E (du ) + 2F du dv + G (dv )
2
(108)
where
E = ru iru = ru
F = ru irv
G = rv irv = rv
(109)
2
r = r (u, v ) = r (u (t ), v (t )) r (t )
(110)
is the one-parameter position vector equation of a curve on the surface. The arc-length s
of this curve between t = t1 and t = t2 is given by
s=
=
=
t2
t1
t2
t2
t1
t1
dr
dt =
dt
t2
t1
du
dv
ru
+ rv
dt =
dt
dt
t2
t1
dr dr
i dt
dt dt
12
du
dv du
dv
ru
+ rv i ru
+ rv dt
dt
dt dt
dt
12
2
du 2
dv + G dv dt
E + 2F du
dt
dt
dt
dt
12
(111)
Also
2
du 2
du dv
dv
ds
dr
=
= E + 2F + G
dt
dt dt
dt
dt
dt
12
(112)
33
RMIT University
Geospatial Science
Since ru and rv are tangent vectors along the v = constant and u = constant parametric
is given by
curves on the surface, then a unit surface normal N
= ru rv
N
ru rv
(113)
N
N
du + N
dv
du +
dv = N
u
v
u
v
(114)
is orthogonal to N
. This can be proved by the following: (i) N
iN
= 1 and
Note that dN
iN
) = d (1) = 0 ; (ii) d (N
iN
) = N
idN
+ dN
iN
= 2dN
iN
which leads to
the differential d (N
iN
) = 2dN
iN
= 0 giving dN
iN
= 0 and dN
is orthogonal to N
.
(iii) d (N
An expression for the denominator of equation (113) can be developed using a formula for
vector dot and cross products: (a b)i(c d ) = (a ic) (bid ) (a id)(bic) giving
ru rv
= (ru rv )i(ru rv )
= (ru i ru )i(rv i rv ) (ru i rv )
(115)
= EG F 2
Defining a quantity J, that is a function of the First Fundamental Coefficients, as
J ru rv = EG F 2
(116)
as
we may express the unit surface normal N
= ru rv
N
J
(117)
ru rv = ru rv sin = EG sin
(118)
F = ru i rv = ru rv cos = EG cos
(119)
so that the angle between the tangent vectors to the parametric curves on the surface is
given by
cos =
F
EG
and
sin =
J
EG
(120)
34
RMIT University
Geospatial Science
We can see from this equation that if F is zero, then the parametric curves on the surface
S intersect at right angles, i.e., the parametric curves form an orthogonal network on the
surface.
If we consider an infinitesimally small quadrilateral on the surface S whose sides are
bounded by the curves u = const. , v = const. , u + du = const. and v + dv = const. then
the lengths of adjacent sides are dsu and dsv . These infinitesimal lengths are found from
equation (108) by setting dv = 0 and du = 0 respectively, giving
dsu = E du
and
dsv = G dv
(121)
dA = dsudsv sin
= EG sin du dv
= J du dv
(122)
)(
= r du + r dv i N
du + N
dv
SFF = dr idN
u
v
u
v
(du ) (r iN
+ r iN
)du dv + r iN
(dv )
= ru iN
u
u
v
v
u
v
v
2
= L (du ) + 2M du dv + N (dv )
2
(123)
where
L = ru iN
u
+ r iN
)
2M = (ru iN
v
v
u
(124)
N = rv iN
v
are the Second Fundamental Coefficients (SFC).
Alternative expressions for the Second Fundamental Form and the Second Fundamental
Coefficients can be obtained by the following.
= 0 and r iN
= 0 (from the definition of N
), then
Since ru iN
v
35
RMIT University
Geospatial Science
(i)
) = ( r iN
) = 0 ; i.e., r iN
+ r iN
= 0 and so r iN
= r iN
.
ru iN
(
u
u
u
uu
u
u
uu
u
u
(ii)
(iii)
(iv)
Hence
i r = ru rv i r
L =N
uu
uu
J
i r = ru rv i r
M =N
uv
uv
J
i r = ru rv i r
N =N
vv
vv
J
(125)
= d riN
(126)
(127)
Let P be a point on a surface with coordinates (u, v ) and Q a neighbouring point on the
surface with coordinates
(u + du, v + dv ) .
r (u, v ) = r (uP , vP ) + (u uP ) ru + (v vP ) rv
1
2
2
(u uP ) ruu + 2 (u uP )(v vP ) ruv + (v vP ) rvv
2!
+ higher order terms
+
}
(128)
36
RMIT University
Geospatial Science
1
2
2
ruu (du ) + 2ruv du dv + rvv (dv )
2
+ higher order terms
+
}
(129)
Dropping the subscript P, then using equation (127) and re-arranging gives
1
r (u + du, v + dv ) r (u, v ) = dr + d 2 r + higher order terms
(130)
2
is the projection of PQ onto the unit
Now PQ = r (u + du, v + dv ) r (u, v ) and PQ iN
surface normal, so using equation (130) we may write
= dr iN
+ 1 d 2 r iN
+ higher order terms
PQ iN
2
(131)
= 0 (since dr and N
are orthogonal)
Now, using equation (126) and noting that driN
equation (131) becomes
= 1 SFF + higher order terms
PQ iN
2
(132)
This shows that the Second Fundamental Form (SFF) is the principal part of twice the
so that SFF is the principal part of twice the perpendicular
projection of PQ onto N
distance from Q onto the tangent plane to the surface at P. It should be noted here that
as PQ 0 the higher order terms 0 .
In Figure 17, C is a curve on a surface S and P is
and n
are the
a point on the curve. t , b
^
b
P
^
N
z
is
the surface and the plane containing t and N
the normal section plane.
In general, the
^t
C
S
^
n
other.
37
RMIT University
Geospatial Science
At P on the curve C, the normal curvature vector k N is the projection of the curvature
)N
.
, so that k = ( k iN
vector k of C onto the surface unit normal N
N
The scalar
is given by
component N of k N in the direction of N
N = k iN
(133)
(134)
Using equation (133) and the first of the Frenet-Serret formulae (85)
= dt iN
= n
= cos
iN
N = k iN
ds
(135)
This is Meusnier's theorem3 that relates the normal curvature N with the curvature of
= 0 ; i.e., n
are (by convention) parallel
N
and N
a curve on a surface. When = 0 , n
and pointing in the same direction, and N = .
Since =
1
, Meunier's theorem can also be stated as:
Between the radius of the osculating circle of a plane section at P and the radius
= N cos
(136)
Meusnier's theorem is a fundamental theorem on the nature of surfaces, named in honour of the French
38
RMIT University
Geospatial Science
) = 0 . That is dt iN
+ tidN = 0 , hence dt iN
= tidN .
= 0 , and d (tiN
tiN
dt
dt
dt
dt
dt
Now, using equation (135) we may write
= dt iN
N = k iN
ds
So, by the chain rule
dt
dt
dt
dt dt
ti dN
dr ds i dN
dr dt i dN
dt
dt dt
N = iN =
iN =
iN =
=
=
2
ds
dt ds
ds dt
ds dt
ds dt
ds dt
du
dv du
dv
ru
dt + rv dt iNu dt + Nv dt
N =
2
ds
dt
and the numerator and denominator can be simplified using equations (123) and (112)
respectively and finally the normal curvature N becomes
du
dv
du dv
L + 2M
+ N
2
2
L (du ) + 2Mdudv + N (dv )
SFF
dt
dt dt
= dt
N = k iN
=
2
2
2
2 =
du
dv
FFF
du dv
E (du ) + 2Fdudv + G (dv )
E + 2F
+ G
dt
dt
dt dt
2
(137)
Dividing the FFF and SFF by (du ) and making the substitution
2
dv
du
(138)
L + 2M + N 2
E + 2F + G 2
(139)
=
gives the normal curvature as
N =
39
RMIT University
Geospatial Science
d u
=
differentiation
dx v
d N
= 0 , that is, using the quotient rule for
d
du
dv
u
dx
dx gives
v2
(E + 2F + G 2 )(2M + 2N ) (L + 2M + N 2 )(2F + 2G )
d N
=
=0
2 2
d
E
F
+
2
+
(
)
that simplifies to
(E + 2F + G2 )(M + N ) (L + 2M + N 2 ) (F + G ) = 0
Now since
(140)
E + 2F + G 2 = E + F + (F + G ) and
L + 2M + N 2 = L + M + (M + N )
E + F + (F + G ) = L + M + (M + N )
E + F
L + M
+ =
+
F + G
M + N
E + F
L + M
=
F + G
M + N
and extreme N satisfies
N =
E + F
L + M
=
F + G
M + N
(141)
or
N (F + G ) (M + N ) = 0
N (E + F ) (L + M ) = 0
that can be re-cast as
N F M + (NG N ) = 0
N E L + (N F M ) = 0
(142)
40
RMIT University
Geospatial Science
(M + N )(E + F ) (L + M )(F + G ) = 0
that can be expressed as a quadratic equation in
(FN GM ) 2 + (EN GL ) + EM FL = 0
(143)
Two values of are found, unless SFF vanishes or is proportional to FFF. These values
dv
of , or
are called the directions of principal curvature labelled 1 and 2 and the
du
normal curvatures in these directions are called the principal curvatures, and labelled 1
and 2 . These principal curvatures are the extreme values of the normal curvature N
and correspond with the two values of found from equation (143).
x 1 b b 2 4ac
The solutions for the quadratic equation ax + bx + c = 0 are x =
and
2
2a
2
x 1 and x 2 are real and unequal if a, b, c are real, and b 2 4ac > 0 . Also, x 1 + x 2 = b a
and x 1x 2 = c a . Using these relationships we have from equation (143)
(144)
1 + 2 =
12 =
EN GL
FN GM
EM FL
FN GM
(145)
(146)
N F M
E L
N
NG N 1 0
=
N F M 0
These homogeneous equations have non-trivial solutions for x if, and only if, the
determinant of the coefficient matrix A is zero (Sokolnikoff & Redheffer 1966). This leads
to a quadratic equation in N
N F M
NG N
N E L
N F M
41
RMIT University
Geospatial Science
whose solutions are N = 1 and N = 2 , the principal curvatures. Half the sum of the
solutions and the product of the solutions can be used to define two other curvatures:
(1 + 2 ) =
EN + GL 2FM
EN + GL 2FM
=
2
2J 2
2 (EG F )
1
2
LN M 2
LN M 2
12 =
=
EG F 2
J2
(147)
(148)
We will now show that the principal directions are orthogonal. Consider two curves C 1
and C 2 on a surface S with curvilinear coordinates u, v .
dr
r du r dv
=
+
ds
u ds v ds
and
r
r u r v
=
+
s
u s
v s
dr r
,
is zero then the two vectors are
ds s
orthogonal. Using equations (109) the dot-product is
r du r dv r u r v
dr r
i
=
+
+
i
u ds v ds u s
ds s
v s
r du r u
r du r v r dv r u r dv r v
=
i
+
i
+
i
+
i
u ds u s
u ds v s v ds u s
v ds v s
r r du u r r du v dv u r r dv v
= i
+ i
+
+ i
u u ds s u v ds s ds s v v ds s
du v dv u
du u
dv v
= (ru i ru )
+ (ru i rv )
+
+ (rv i rv )
ds s ds s
ds s
ds s
du v dv u
du u
dv v
=E
+ F
+
+ G
ds s
ds s ds s
ds s
du v dv u ds s
dv v ds s du u
= E + F
+
+G
ds s ds s du u
ds s du u ds s
v dv
dv v du u
= E + F + + G
u du
du u ds s
42
RMIT University
Geospatial Science
dr r
du u
i
= {E + F (1 + 2 ) + G 12 }
ds s
ds s
GL EN
EM FL du u
= E + F
+
G
FN GM
FN GM ds s
Hence the unit vectors of the curves C 1 and C 2 are orthogonal as are the directions of
principal curvatures 1 and 2 .
If, at a point on a surface, the normal curvature N is the same in every direction, then
such points are known as umbilical points. In the directions of the parametric curves u =
constant (du = 0) and v = constant (dv = 0), the normal curvatures are found from
equation (137) as
du =0 =
N
G
and
dv =0 =
L
E
du =dv =
L + 2M + N
E + 2F + G
Now, if the normal curvature is the same in every direction then we have two equations
L
N
=
E
G
L
L + 2M + N
=
E
E + 2F + G
(149)
2FL 2EM = EN GL .
N M
L
=
=
G F
E
(150)
43
RMIT University
Geospatial Science
revolution are: a sphere (a circle rotated about a diameter), a cone (excluding the base),
and a cylinder (excluding the ends).
The x,y,z Cartesian coordinates of a general surface of revolution having u,v curvilinear
coordinates can be expressed in the general form
x (u, v ) = g (u ) cos v
y (u, v ) = g (u ) sin v
(151)
z (u, v ) = h (u )
where g (u ), h (u ) are certain functions of u. A point on the surface of revolution has the
position vector
(152)
ru =
rv =
ruv =
rvu =
ruu =
rvv =
where g =
r (u, v )
u
r (u, v )
v
ru
v
rv
u
ru
u
rv
v
g cos v i + g sin v j + h k
= g sin v i + g cos v j + 0 k
= g sin v i + g cos v j + 0 k
(153)
= g sin v i + g cos v j + 0 k
= g cos v i + g sin v j + h k
= g cos v i g sin v j + 0 k
d
d
d
d
g (u ) ; g =
g (u ) and h =
h (u ) ; h =
h (u )
du
du
du
du
Using equations (109) and (153) gives the First Fundamental Coefficients as
44
RMIT University
Geospatial Science
E = ru i ru = g 2 + h 2
F = ru i rv = 0
(154)
G = rv i rv = g 2
In equations (154), F = 0 which indicates that the parametric curves on a surface of
revolution (the u-curves and v-curves) are orthogonal.
Equation (116) gives
g 2 + h 2
J ru rv = EG F 2 = g
(155)
N = ru rv = g cos v
g sin v
()
(+)
(156)
is
and the unit normal vector N
= N = N = N = gh cos v i gh sin v j + gg k
N
N
ru rv
J
J
J
J
(157)
Using equations (125) with (153) and (157) gives the Second Fundamental Coefficients as
i r = ru rv i r = g (g h g h ) = g h g h
L =N
uu
uu
J
J
g 2 + h 2
i r = ru rv i r = 0
M =N
uv
uv
J
2
gh
i r = ru rv i r = g h =
N =N
vv
vv
J
J
g 2 + h 2
(158)
and M = 0
(159)
Substituting these results into equation (137) gives the equation for normal curvature on a
general surface of revolution as
L (du ) + N (dv )
N =
2
2
E (du ) + G (dv )
2
(160)
The normal curvatures along the parametric curves u = constant (du = 0) and v =
constant (dv = 0) are denoted 1 and 2 respectively and
45
RMIT University
Geospatial Science
1 =
N
h
=
1
G
g (g 2 + h 2 )2
(161)
L
g h g h
2 = =
3
E
(g 2 + h 2 )2
Figure 18 shows two points P and Q on a surface S separated by a very small arc ds . The
parametric curves u, u + du and v, v + dv form a very small rectangle on the surface and
ds can be considered as the hypotenuse of a plane right-angled triangle and we may write
G dv
tan =
(162)
E du
where is azimuth; a positive clockwise angle measured from the v-curve (v = constant) .
dsv = E du
v + dv
ru
ds
u + du
P
rv
dsu = G dv
(163)
du
G
and using equation (162) we have =
and equation (163) can be written as
dv
E tan2
2
LG
E tan2 + N
=
N =
EG
+
G
E tan2
L N
+ tan2
E G
1 + tan2
46
RMIT University
Geospatial Science
1
sin2
2
and
tan
=
, we can write the normal
cos2
cos2
curvature on a surface of revolution as
L
N
N = cos2 + sin2 = 2 cos2 + 1 sin2
E
G
(164)
the function N must have optimum values, unless 1 = 2 . These optimum values
d N
are found by setting the derivative
to zero. From equation (164)
d
d N
= 2 (1 2 ) sin cos = (1 2 ) sin 2 = 0
d
and if 1 2 0 then the optimum values are where sin 2 = 0 , i.e., 2 = n or
= 21 n where n = 0,1,2, 3, ( = 0 , 90 ,180 , 270 , ) .
optimum curvatures (of normal sections) are in the directions of the parametric
curves on the surface of revolution.
(ii)
there may be a point (or points) on the surface of revolution where 1 = 2 in which
case the curvature would be constant for any normal section in any direction. Such
points are known as umbilic points. (For an ellipsoid representing the mathematical
shape of the earth, the minor axis of the ellipsoid is the earth's polar axis and the
north and south poles of the ellipsoid are umbilic points.)
With denoting the curvature of a normal section having the direction on the surface,
equation (164) becomes
= 2 cos2 + 1 sin2
(165)
12
1 cos + 2 sin2
2
(166)
47
RMIT University
Geospatial Science
This is Euler's equation4 (on the curvature of surfaces) and gives the radius of curvature of
a normal section in terms of the radii of curvature 1, 2 along the parametric curves u =
constant, v = constant respectively and the azimuth measured from the curve v =
constant.
where the parametric curves are orthogonal and in the directions of principal curvature.
Gre
en
wi
ch
b
a
H
equator
a
y
p
This equation on the curvature of surfaces is named in honour of the great Swiss mathematician Leonard
Euler (1707-1783) who, in a paper of 1760 titled Recherches sur la courbure de surfaces (Research on the
2 fg
curvature of surfaces), published the result r =
where f and g are extreme values of
f + g + ( f g ) cos 2
the radius of curvature r (Struik 1933).
cos + sin = 1 his equation can be expressed in the form given above.
2
48
RMIT University
Geospatial Science
reference planes; the equator and the Greenwich meridian. Longitudes are measured 0 to
180 (east positive, west negative) from the Greenwich meridian and latitudes are
measured 0 to 90 (north positive, south negative) from the equator.
The x,y,z
Cartesian coordinate system has an origin at O, the centre of the ellipsoid, and the z-axis
is the minor axis (axis of revolution). The xOz plane is the Greenwich meridian plane (the
origin of longitudes) and the xOy plane is the equatorial plane.
The positive x-axis passes through the intersection of the Greenwich meridian and the
equator, the positive y-axis is advanced 90 east along the equator and the positive z-axis
passes through the north pole of the ellipsoid. The Cartesian equation of the ellipsoid is
x 2 + y2 z 2
+ 2 =1
a2
b
(167)
(a < b )
(168)
no
rm
al
49
RMIT University
Geospatial Science
In Figure 20, is the latitude of P (the angle between the equator and the normal), C is
the centre of curvature and PC is the radius of curvature of the meridian ellipse at P. H is
the intersection of the normal at P and the z-axis (axis of revolution).
All parallels of latitude on the ellipsoid are circles created by intersecting the ellipsoid with
planes parallel to (or coincident with) the xOy equatorial plane.
Replacing z with a
constant C in equation (167) gives the equation for circular parallels of latitude
C2
x 2 + y 2 = a 2 1 2 = p 2
( 0 C b;
a >b
(169)
All other curves on the surface of the ellipsoid created by intersecting the ellipsoid with a
plane are ellipses. This can be demonstrated by using another set of coordinates x , y , z
that are obtained by a rotation of the x,y,z coordinates such that
x
x
y = R y
z
z
r r
11 12 r13
where R = r21 r22 r23
and
x r r r x
11 21 31
y = r r r y
12 22 32
Substituting into equation (167) gives the equation of the ellipsoid in x , y , z coordinates
2
2
2
2
2
2
2
2
2
1
+ 2 {r132 x 2 + r232 y 2 + r332 z 2 + 2r13r23x y + 2r13r33x z + 2r23r33y z } = 1
b
(170)
50
RMIT University
Geospatial Science
In equation (170) let z = C 1 where C 1 is a constant. The result will be the equation of a
curve created by intersecting an inclined plane with the ellipsoid, i.e.,
r112 + r122
r212 + r222
r132 2
r11r21 + r12r22 r13r23
r232 2
+ 2 x + 2
+ 2 x y +
+ 2 y
a 2
a 2
b
a2
b
b
+ {2C 1 (r11r31 + r12r32 + r13r33 )} x + {2C 1 (r21r31 + r22r32 + r23r33 )} y
= 1 C 12 {r312 + r322 + r332 }
(171)
(172)
where it can be shown that AB H 2 > 0 , hence it is the general Cartesian equation of an
ellipse that is offset from the coordinate origin and rotated with respect to the coordinate
axes (Grossman 1981). Equations of a similar form can be obtained for inclined planes
(a, b )
(ii)
(a, e 2 )
(iii) (a, f )
a, b, e 2 and f are ellipsoid parameters (or constants) and they have been defined (for the
ellipse) in earlier sections. Other parameters; c, e 2 and n are also useful in developments
to follow. They have also been defined in earlier sections as well as the interrelationship
between all these parameters [see equations (22) to (37)].
functions W and V are useful in the computation of radius of curvature and Cartesian
coordinates [see equations (43) to (47)] and the relationships between the normal at P and
the axes of the ellipse are the same as for P on an ellipsoid [see Figure 9 and equations
(48) to (51) ].
ellipsoid and their interrelationship is often used to simplify formula [see Figure 9 and
equations (54) to (56)]
51
RMIT University
Geospatial Science
(173)
where
g () = g =
c
cos
V
and
h () = h =
b
sin
V
(174)
f =
; e2 =
=
f
2
f
;
e
=
=
(
)
2
2
2
a
a
b
(1 f )
(175)
dV
e 2
= cos sin , the derivatives g , h , g , h are
d
V
d c
c
cos = 3 sin
d V
V
d b
c
sin = 3 cos
h =
V
d V
g =
(176)
1 + e 2
d c
c
3 sin = 3 cos 2 3
g =
d V
V 2
d c
c
3
3 cos = 3 sin 2 2
h =
V
d V
V
c
g h g h = 3
V
and
(177)
52
RMIT University
Geospatial Science
First Fundamental Coefficients E, F and G are found from equations (154), (174) and
(176) as
c
E = g 2 + h 2 = 3
V
F =0
(178)
G = g 2 = cos
V
J = EG F
(179)
In equations (178), F = 0 which indicates (as we should expect) that the -curves
(parallels of latitude) and -curves (meridians of longitude) are orthogonal.
Second Fundamental Coefficients L, M and N are found using equations (158), (174), (176)
and (177) as
L=
g h g h
c
= 3
2
2
V
g + h
M =0
N =
(180)
gh
g + h
2
c
cos2
V
Identical results for E,F,G and L,M,N can be obtained, with slightly less algebra, in the
following manner.
The position vector of P on the surface of the ellipsoid is
r = r (, ) = g cos i + g sin j + h k
c
c
b
cos cos i + cos sin j + sin k
V
V
V
and using equations (153), (174) and (176) the derivatives of the position vector are
r = g cos i + g sin j + h k
c
c
c
= 3 sin cos i 3 sin sin j + 3 cos k
V
V
V
r = g sin i + g cos j + 0 k
c
c
= cos sin i + cos cos j + 0 k
V
V
53
RMIT University
Geospatial Science
= r r
N
J
and by equation (156) the cross product r r is
r r = gh cos v i gh sin v j + gg k
2
2
2
c
c
c
2
2
Now using equations (109) the vectors above, the First Fundamental Coefficients E, F and
G are
E = r i r
2
2
2
c
c
c
= 3 sin2 cos2 + 3 sin2 sin2 + 3 cos2
V
V
V
2
c
= 3
V
F = r i r
c2
c2
= 4 sin sin cos cos 4 sin sin cos cos
V
V
=0
G = r i r
2
2
c
c
= cos2 sin2 + cos2 cos2
V
V
2
c
= cos
V
and using equations (124) the Second Fundamental Coefficients L, M and N are
54
RMIT University
Geospatial Science
L = r iN
c
c
c
= 3 sin2 cos2 + 3 sin2 sin2 + 3 cos2
V
V
V
c
= 3
V
+ r iN
)
2M = (r iN
c
c
= 3 sin cos sin cos + 3 sin cos sin cos
V
V
c
c
V
V
=0
N = r iN
c
c
= cos2 sin2 + cos2 cos2
V
V
c
2
= cos
V
These results are identical to those in equations (178) and (180)
The element of arc length ds on the surface of the ellipsoid is found from equations (108)
and (178) as
(ds ) = E (d ) + 2F d d + G (d )
2
c
c
2
2
= 3 (d ) + cos (d )
V
V
(181)
The element of arc length ds along the -curve (a meridian of longitude where
= constant and d = 0 ) and the element of arc length ds along the -curve (a
parallel of latitude where = constant and d = 0 ) are given by equations (121) as
c
d
V3
c
ds = G d = cos d
V
ds = E d =
(182)
The angle between the tangents to the meridian of longitude and parallel of latitude
curves are given implicitly by equations (120) as
cos =
F
=0
EG
and
sin =
J
=1
EG
(183)
55
RMIT University
Geospatial Science
and as we should expect, = 90 , since the meridians and parallels ( -curves and
-curves ) form an orthogonal net (since, from above, F = 0 ).
The area of an infinitesimally small rectangle on the surface of the ellipsoid bounded by
meridians and + d , and parallels and + d is given by equation (122) as
c
dA = J d d = 2 cos d d
V
2
(184)
c
c
2
2
(d ) + cos2 (d )
3
V
V
2
2
c
c
2
2
cos (d )
(d ) +
3
V
V
( )
(185)
N
L
and 2 = , and these are in the directions
G
E
of the parametric curves = constant (d = 0) and = constant (d = 0) respectively.
The optimum normal curvatures are 1 =
( )
( )
and
c
3
L
2 = = V
c
E
V3
1
c
V3
( ) ( )
2
(186)
and since radius of curvature is the reciprocal of the curvature , the optimum radii of
curvature of the principal normal sections of the ellipsoid are
1 =
c
V
and
2 =
c
V3
(187)
These optimum radii are the principal radii of curvature and are in the directions of the
parametric curves; parallels of latitude ( = constant) and meridians of longitude ( =
constant) respectively. These directions are the principal directions of the ellipsoid.
In general, at a point on an ellipsoid where 90 and 0 < cos 1 the quantity V > 1
c
c
and V 3 > V ; and since c = constant for any particular ellipsoid then
> 3 . Hence the
V V
principal radii of curvature at a point are:
c
, that is the largest radius of curvature of a normal section and this normal
V
section is in the direction of the parallel of latitude. In geodesy, this is known as the
1 =
56
RMIT University
Geospatial Science
c
, that is the smallest radius of curvature of a normal section and this normal
V3
section is in the direction of the meridian of longitude. In geodesy, this is known as
2 =
c
V
c
= 3 =
V
a
W
a (1 e 2 )
W3
r
W
^
N
a
equato r
pr
im
ve
l
ic a
rt
me r
i di
a
l
ma
n or
57
RMIT University
Geospatial Science
For umbilical points the following condition is obtained from equation (150)
G F
E
=
=
N M
L
( )
( )
c 2
c
cos2
0
V
= = Vc
c
2
0
cos
V
V
or
a2
b
(189)
c
c
b2
=
=
3
V3
(1 + e 2 )2 a
and
equator =
c
c
=
1 = a
V
(1 + e 2 )2
(190)
The radius of curvature of a normal section having azimuth is and from Euler's
equation [equation (166)]
=
sin + cos2
(191)
The mean radius of curvature m is the mean value of the radii of curvature for all values
of 0 2 in equation (191) where in general the mean value of a function f (x )
between x = a and x = b is
fmean
1
=
b a
f (x )dx
a
hence
2
1
2
=
d =
d
2
2 0
0 sin + cos2
since 2 = 4
2
dividing the numerator and denominator of the integrand by cos2 and taking
out
as a constant gives
58
RMIT University
Geospatial Science
m =
1 + tan2
sec2
then du = sec2
d and u 2 = tan2 , and with
2
m =
dx
1+x
1+u
du
= arctan x
2
2
[arctan u ]
0 =
0
2
m =
(192)
The Average and Gaussian curvatures are, using equations (147), (148), (186) and (188)
1 1
Average curvature = 12 (1 + 2 ) = 12 +
1
Gaussian curvature = 12 =
(193)
The radius of a parallel of latitude, a circle on the surface of the ellipsoid, is found from
Meusnier's theorem [equation (136)] as
rparallel = cos
(194)
Figure 22 illustrates the use of Meusnier's theorem to determine the radius of the parallel
of latitude. At P on the ellipsoid, the parallel of latitude (radius r) and the prime vertical
normal section (radius of curvature ) have a common tangent vector r , and the plane
containing the parallel of latitude and the prime vertical normal section plane make an
is the unit normal to the surface and the
angle of with each other. In Figure 22, N
distance PH = is the radius of curvature of the prime vertical section at P.
59
RMIT University
Geospatial Science
r is the unit tangent in the direction of the meridian normal section and r and r are
mer
idi
an
^
N
a
equator
r
r
N
paral lel
prime vertical
normal section
Figure 22: The prime vertical section and parallel of latitude
dm = d
(195)
a (1 e 2 )
3
(1 e 2 sin2 )2
a (1 e 2 )
W3
(196)
a2
b (1 + e 2 cos2 )
3
2
c
V3
(197)
Substituting equation (196) into equation (195) leads to series formula for the meridian
distance m as a function of latitude and powers of e 2 . Substituting equation (197) into
equation (195) leads to series formula for m as a function of and powers of an ellipsoid
a b
constant n =
.
a +b
60
RMIT University
Geospatial Science
Series formula involving powers of e 2 are more commonly found in the geodetic literature
but, series formula involving powers of n are more compact; and they are easier to
"reverse", i.e., given m as a function of latitude and powers of n develop a series formula
(by reversion of a series) that gives as a function m.
m=
a (1 e 2 )
W3
d = a (1 e
3
1
2
2
2 2
d
=
a
1
e
1
e
sin
d
(
)
(
)
W3
0
)
0
(198)
This is an elliptic integral of the second kind that cannot be evaluated directly; instead,
23
1
2
2
the integrand
=
1
e
sin
23
1
2
2
=
1
e
sin
(1 + x ) = Bn x n
(199)
n =0
An infinite series where n is a positive integer, is any real number and the binomial
coefficients Bn are given by
Bn =
(200)
The binomial series (199) is convergent when 1 < x < 1 . In equation (200) n! denotes nfactorial and n ! = n (n 1)(n 2)(n 3) 3 2 1 . zero-factorial is defined as 0 ! = 1 and
the binomial coefficient B0 = 1 .
In the case where is a positive integer, say k, the binomial series (199) can be expressed
as the finite sum
k
(1 + x ) = Bnk x n
k
(201)
n =0
61
RMIT University
Geospatial Science
Bnk =
k!
n ! (k n ) !
(202)
The binomial series is an important tool in geodesy where, as we shall see, it is used in the
development of various solutions to geodetic problems. The invention of the binomial series is
attributed to Isaac Newton, who, in a letter to the German mathematician Gottfried Leibniz in
June 1676, set out his theorem as:
The Extraction of Roots are much shortened by the Theorem
(P + PQ )m n = P
m n
m
n
AQ +
m n
2n
BQ +
m 2n
3n
CQ +
m 3n
4n
DQ + &c
(a)
where P + PQ stands for a Quantity whose Root or Power or whose Root of a Power is to be
found, P being the first Term of that quantity, Q being the remaining terms divided by the first
term, and m n the numerical Index of the power of P + PQ . This may be a Whole Number or
(so to speak) a Broken Number; a positive number or a negative one.
several cases to describe how P, Q, m and n are obtained and then defines A, B, C and D] In
this last case, if (a + b x )
3
Q =
2 3
3
to be taken to mean (P + PQ )2 3 in the Formula, then P = a ,
m n
work in the Quotient, I shall use A, B, C, D, &c. Thus A stands for the first term P ; B for
m
the second term
AQ ; and so on. The use of this Formula will become clear through
n
Examples. [The examples show the application of the formula in cases in which the exponents
are 1 2 , 1 5 , 1 3 , 4 3 , 5, 1, 3 5 ]
In a subsequent letter to Leibniz in October 1676, Newton explains in some detail how he made
his early discoveries, and discloses that his binomial rule was formulated twelve years earlier, in
1664, while he was an undergraduate at Cambridge University (Newman 1956).
Letting P = a, Q =
(a
and =
+ x ) = a + a
m
n
x +
( 1)
2!
x +
( 1)( 2)
3!
x +
(b)
Letting a = 1 in equation (b) will give the expanded form of equation (199).
Letting a = 1 and setting = k a positive integer in equation (b) will give the expanded form
of equation (201)
62
RMIT University
Geospatial Science
(1 + x )
= Bn 2 x n = B0 2 x 0 + B1 2 x 1 + B2 2 x 2 + B3 2 x 3 +
3
(203)
n =0
The binomial coefficients Bn 2 for the series (199) are given by equation (200) as
3
n=0
B0 2 = 1
n =1
B1 2 =
n=2
B2 2 =
n=3
B3 2 =
( 23 )
1!
( 23 )( 25 )
2!
3
2
=
35
24
( 23 )( 25 )( 72 )
3!
357
246
Inspecting the results above, we can see that the binomial coefficients Bn 2 form a sequence
3
3 35
357 3579
3 5 7 9 11
1, ,
,
,
,
,
2 24
246 2468
2 4 6 8 10
Using these coefficients gives (Baeschlin 1948, p.48; Jordan/Eggert/Kneissl 1958, p.75;
Rapp 1982, p.26)
3
1
3
35 4
357 6
= (1 e 2 sin2 ) 2 = 1 + e 2 sin2 +
e sin 4 +
e sin6
3
W
2
24
246
3579 8
3
9 11 10 10
+
e sin 8 +
e sin +
2468
2 4 6 8 10
(204)
To simplify this expression, and make the eventual integration easier, the powers of sin
can be expressed in terms of multiple angles using the standard form
sin2n =
1
22 n
2n (1)n
+
n 22n 1
2n
2n
cos 2n cos (2n 2) + cos (2n 4)
1
2
2n
2n
n
cos 2
cos (2n 6) + (1)
3
n 1
(205)
2n
Using equation (205) and the binomial coefficients Bn2n = computed using equation
n
(202) gives
sin 2 =
1 1
cos 2
2 2
63
RMIT University
Geospatial Science
sin 4 =
3 1
1
+ cos 4 cos 2
8 8
2
sin 6 =
5
1
3
15
cos 6 + cos 4
cos 2
16 32
16
32
sin 8 =
35
1
1
7
7
+
cos 8 cos 6 +
cos 4 cos 2
128 128
16
32
16
sin10 =
63
1
5
45
15
105
cos 10 +
cos 8
cos 6 +
cos 4
cos 2
256 512
256
512
64
256
(206)
Substituting equations (206) into equation (204) and arranging according to cos 2 ,
cos 4 , etc., we obtain (Baeschlin 1948, p.48; Jordan/Eggert/Kneissl 1958, p.75; Rapp
1982, p.27)
32
1
2
2
=
1
e
sin
A =1+
B=
C =
D=
E=
F =
3 2 45 4 175 6 11025 8
43659 10
e + e +
e +
e +
e
4
64
256
16384
65536
3 2 15 4 525 6
2205 8
72765 10
e + e +
e +
e +
e
4
16
512
2048
65536
15 4 105 6
2205 8
10395 10
e +
e +
e +
e
64
256
4096
16384
35 6
315 8 31185 10
e +
e +
e
512
2048
131072
315 8
3465 10
e +
e
16384
65536
693 10
e
131072
+
+
+
(208)
+
+
+
Substituting equation (207) into equation (198) gives the meridian distance as
m = a (1 e
cos ax dx =
0
sin ax
gives the
a
64
RMIT University
Geospatial Science
m = a (1 e 2 ) A
B
C
D
E
F
sin 2 + sin 4 sin 6 + sin 8 sin 10 +
2
4
6
8
10
(209)
where is in radians and the coefficients A, B, C, etc., are given by equations (208).
From equation (209), the quadrant distance Q5, the meridian distance from the equator to
the pole, is
Q = a (1 e 2 ) A ( 12 )
Equation (209) may be simplified by multiplying the coefficients by
(210)
(1 e 2 )
and
B
C
, A4 = (1 e 2 ) , etc., and
2
4
1
3
5 6
175 8
441 10
A0 = 1 e 2 e 4
e
e
e +
4
64
256
16384
65536
3
1
15 6
35 8
735 10
A2 = e 2 + e 4 +
e +
e +
e +
8
4
128
512
16384
15 4 3 6 35 8 105 10
e + e + e +
A4 =
e +
256
4
64
256
35 6 5 8 315 10
e + e +
A6 =
e +
3072
4
256
315 8 7 10
e + e +
A8 =
131072
4
693
A10 =
(e 10 + )
131072
(211)
(212)
The quadrant distance is the length of the meridian arc from the equator to the pole and the
ten-millionth part of this distance was originally intended to have defined the metre when that
unit was introduced.
Things (Adler 2002) has a detailed account of the measurement of the French Arc (an arc of
the meridian from Dunkerque, France to Barcelona, Spain and passing through Paris) by JohnBaptiste-Joseph Delambre and Pierre-Franois-Andr Mchain in 1792-9 during the French
Revolution. The analysis of their measurements enabled the computation of the dimensions of
the earth that lead to the definitive metre platinum bar of 1799.
65
RMIT University
Geospatial Science
(213)
where
1
3
5 6
B0 = 1 e 2 e 4
e
4
64
256
3
1
15 6
B2 = e 2 + e 4 +
e
8
4
128
15 4 3 6
e + e
B4 =
256
4
35 6
B6 =
e
3072
(214)
This is a contraction of equation (211) and the coefficients B0 , B2 , B4 and B6 exclude all
terms involving powers of the eccentricity greater than e 6
in the coefficients
A0 , A2 , A4 and A6 . Equations (213) and (214) are the same formula given in Lauf (1983,
p. 36, eq'n 3.55).
Meridian distance as a series expansion in powers of n
The German geodesist F.R. Helmert (1880) gave a series formula for meridian distance m
as a function of latitude and powers of an ellipsoid constant n that requires fewer terms
than the meridian distance formula involving powers of e 2 .
Using equations (195) and (197) the differentially small meridian distance dm is given by
dm =
c
d
V3
(215)
a b
f
=
a +b
2f
(216)
66
RMIT University
Geospatial Science
c=
a2
1 + n
= a
,
1 n
b
e2 =
4n
2 ,
(1 + n )
e 2 =
4n
2
(1 n )
(217)
(1 n ) + 4n cos2
2
(1 n )
2
V2 =
(218)
e i = cos i sin in
(i 2 = 1)
and
e = 2.718281828 is the base of the natural logarithms. e in Euler's identities should not
be confused with the eccentricity of the ellipsoid.
Adding Euler's identities gives
2 cos 2 = e
i 2
+e
i 2
2 cos = e i + e i
and replacing
with 2
gives
V2 =
1
in equation (215) can be developed as
V3
3
1
2 2
=
V
(
)
V3
3
2 2
= ((1 n )
) (1 + nei 2 ) (1 + ne i 2 )
32
23
= (1 n ) (1 + ne i 2 )
3
32
23
(1 + nei 2 )
(219)
Using equation (219) and the first of equations (217) in equation (215) gives
3
3
1 + n
3
(1 n ) (1 + ne i 2 ) 2 (1 + ne i 2 ) 2 d
dm = a
1 n
(220)
67
RMIT University
Geospatial Science
1 + n
3
2
(1 n ) = (1 + n )(1 n ) = (1 n ) (1 n 2 ) and equation (220) becomes (Lauf
Now
1 n
1983, p. 36, eq'n 3.57)
32
dm = a (1 n ) (1 n 2 ) (1 + ne i 2 )
32
(1 + nei 2 )
(221)
Using the binomial series as previously developed [see equation (204)] we may write
23
(1 + ne i 2 )
3
3 5 2 i 4 3 5 7 3 i 6
= 1 ne i 2 +
ne
ne
2
24
246
3 5 7 9 4 i 8 3 5 7 9 11 5 i 10
+
+
ne
ne
2468
2 4 6 8 10
and
23
(1 + ne i 2 )
3
3 5 2 i 4 3 5 7 3 i 6
= 1 ne i 2 +
ne
ne
2
24
246
3 5 7 9 4 i 8 3 5 7 9 11 5 i 10
+
ne
ne
+
2468
2 4 6 8 10
The product of these two series, after gathering terms, will be a series in terms
(e
i 2
+ e i 2 = 2 cos 2 ,
(e
i 4
+ e i 4 = 2 cos 4 ,
(e
i 6
simplifying gives
9 2 225 4
n +
n +
4
64
3
45 3 525 5
n +
n +
n +
2
16
128
15 2 105 4
8 n + 32 n +
35 3 945 5
n +
n +
16
256
315 4
n +
128
dm = a (1 n ) (1 n 2 ) 1 +
2 cos 2
+2 cos 4
2 cos 6
+2 cos 8
693 5
2 cos 10
n + + d
256
68
RMIT University
Geospatial Science
cos ax dx =
0
sin ax
gives the
a
where
9 2 225 4
n +
n +
4
64
3
45 3 525 5
a2 = n +
n +
n +
2
16
128
1 15
105 4
a 4 = n 2 +
n +
2 8
32
a0 = 1 +
(223)
1 35 3 945 5
n +
n +
3 16
256
1 315 4
a8 =
n +
4 128
1 693 5
a10 =
n +
5 256
a6 =
a (1 e 2 )
W3
and
(1 n )
c
c
rather than = 3 and dm = 3 d as above but the end result
(1 e ) =
2
(1 + n )
V
V
2
(Jordan/Eggert/Kneissl 1958, eq'n 38, p.83) is similar in form to equation (222) but
without the term a10 sin 10 and the coefficients exclude all terms involving powers of n
greater than n 4 . With these restrictions we give Helmert's formula as (Lauf 1983, p. 36,
eq'n 3.55)
(224)
69
RMIT University
Geospatial Science
where
9 2 225 4
n +
n +
4
64
3
45
n + n3 +
2
16
1 15 2 105 4
n +
n +
2 8
32
1 35 3
n +
3 16
1 315 4
n +
4 128
b0 = 1 +
b2 =
b4 =
b6 =
b8 =
(225)
1+n
(1 n ) (1 n 2 )
1+n
(1 n 2 )(1 n 2 )
1+n
m=
a
{c0 c2 sin 2 + c4 sin 4 c6 sin 6 + c8 sin 8 }
1+n
(226)
where
1 2
1 4
n +
n +
4
64
3
1
= n n 3
2
8
15
1
= n 2 n 4
16
4
35 3
=
(n )
48
315 4
=
(n )
512
c0 = 1 +
c2
c4
c6
c8
(227)
70
RMIT University
Geospatial Science
Equation (226) with expressions for the coefficients c 0 , c2 , c 4 etc., is, except for a slight
change in notation, the same as Rapp (1982, p. 30, eq'n 95) who cites Helmert (1880) and
is essentially the same as Baeschlin (1948, p. 50, eq'n 5.5) and Jordan/Eggert/Kneissl
(1958, p.83-2, eq'ns 38 and 42)
The reverse
9
225 4
Q = a (1 n ) (1 n 2 ) 1 + n 2 +
n +
2
4
64
(228)
Q
9
225 4
= a (1 n ) (1 n 2 ) 1 + n 2 +
n +
4
64
1
2
(229)
71
RMIT University
(ii)
Geospatial Science
m
G
(230)
45
15 2 105 4
n + n 3 +
n +
n +
1
sin 2 + 8
sin 4
16
32
= 2
1 + 9 n 2 + 225 n 4 +
2 1 + 9 n 2 + 225 n 4 +
4
64
4
64
35 3
315 4
n +
n +
1
1
sin 6 +
16
128
sin 8 (231)
3 1 + 9 n 2 + 225 n 4 +
4 1 + 9 n 2 + 225 n 4 +
4
64
4
64
Using a special case of the binomial series [equation (199) with = 1 ]
1
(1 + x )
= 1 x + x2 x3 + x4
9 2 225 4
9 2 225 4
1 + n 2 +
n + = 1 n +
n + + n +
n +
4
4
4
64
64
64
3
225 4
9
n 2 +
n + +
4
64
1 + n 2 +
n + = 1 n 2 + n 4
4
64
4
64
(232)
Substituting equation (232) into equation (231), multiplying the terms and simplifying
gives the equation for as (Lauf 1983, p. 37, eq'n 3.67)
m
9
15
3
15
= n n 3 sin 2 + n 2 n 4 sin 4
2
16
G
16
32
35
315 4
n 3 sin 6 +
n sin 8
48
512
(233)
72
RMIT University
Geospatial Science
If we require the value of corresponding to a particular value of , then the series (233)
needs to be reversed.
(234)
f (y ) = f (z ) + xF (z ) f (z ) +
x2 d
x3 d2
2
3
(z )} f (z ) +
{
F
{F (z )} f (z ) +
2! dz
3! dz
x n d n 1
n
{F (z )} f (z )
n ! dz n 1
(235)
In our case, comparing the variables in equations (233) and (234), z = , y = and
x = 1 , and if we choose f (y ) = y then f (z ) = z and f (z ) = 1 . So, in our case equation
(233) can be expressed as
= F ( )
(236)
( ) +
1 d
1 d2
1 d n 1
2
3
n
(
)
(
)
{F } +
{F } + +
{F ( )}
2
n 1
2 d
6 d
n ! d
(237)
9
15
3
15
F ( ) = n n 3 sin 2 n 2 n 4 sin 4
2
16
16
32
35
315 4
n sin 8
+ n 3 sin 6
48
512
9
15
3
15
F ( ) = n n 3 sin 2 n 2 n 4 sin 4
2
16
16
32
35
315 4
+ n 3 sin 6
n sin 8
48
512
(238)
73
RMIT University
Geospatial Science
Squaring F ( ) gives
45
27
9
2
{F ( )} = n 2 n 4 + sin 2 2 n 3 sin 2 sin 4
4
16
16
35
225 4
+ n 4 sin 2 sin 6 +
n sin2 4
16
256
and expressing powers and products of trigonometric functions as multiple angles using
sin2 A = 12 12 cos 2A
and
sin A sin B =
1
2
{cos (A B ) cos (A + B )}
simplification
207 4
31
9
45
9
2
n n 3 + cos 2 n 2 n 4 + cos 4
{F ( )} = n 2
8
512
32
8
16
45
785 4
+ n 3 cos 6
n + cos 8 +
512
32
Differentiating with respect to and then dividing by 2 gives the 3rd term in equation
(237) as
45
1 d
31
9
2
{F ( )} = n 3 sin 2 + n 2 n 4 sin 4
32
2 d
8
135 3
785 4
n sin 6 +
n sin 8 +
32
128
(239)
Using similar methods the 4th and 5th terms in equation (237) are
1 d2
27
135 4
3
n sin 4
{F ( )} = n 3 sin 2 +
2
6 d
16
16
81
135 4
+ n 3 sin 6
n sin 8 +
16
1 d3
27
27
4
{F ( )} = n 4 sin 4 + n 4 sin 8 +
3
4
24 d
(240)
(241)
Substituting equations (238) to (241) into equation (237) and simplifying gives an equation
for as a function of and powers of n as (Lauf 1983, p. 38, eq'n 3.72)
27
55
3
21
= + n n 3 sin 2 + n 2 n 4 + sin 4
2
32
16
32
151 3
1097 4
+
n + sin 6 +
n sin 8
96
512
(242)
74
RMIT University
Geospatial Science
m
radians and G is given by equation (229). This very useful series now gives
G
a direct way of computing the latitude given a meridian distance.
where =
Newton-Raphson method for the real roots of the equation f ( ) = 0 given in the form of
an iterative equation
n +1 = n
f (n )
f (n )
(243)
where n denotes the n th iteration and f ( ) can be obtained from Helmert's formula
[equation (224)] as
(244)
d
{f ()} is given by
d
(245)
m
and the functions f (1 )
a
and f (1 ) evaluated from equations (244) and (245) using 1 . 2 ( for n = 2) is now
An initial value for (for n = 1 ) can be computed from 1 =
computed from equation (243) and this process repeated to obtain values 3 , 4 , . This
iterative process can be concluded when the difference between n +1 and n reaches an
acceptably small value.
Newton-Raphson iteration is a numerical technique used for finding approximations to
the roots of real valued functions and is attributed to Isaac Newton (1643-1727) and
Joseph Raphson (1648-1715). The technique evolved from investigations into methods
of solving cubic and higher-order equations that were of interest to mathematicians in
the 17th and 18th centuries. The great French algebraist and statesman Franois Vite
(1540-1603) presented methods for solving equations of second, third and fourth degree.
He knew the connection between the positive roots of equations and the coefficients of
the different powers of the unknown quantity and it is worth noting that the word
75
RMIT University
Geospatial Science
"coefficient" is actually due to Vite. Newton was familiar with Vite's work, and in
portions of unpublished notebooks (circa 1664) made extensive notes on Vite's method
of solving the equation x 3 + 30x = 14356197 and also demonstrated an iterative
technique that we would now call the "secant method".
f (x n ) f (x n 1 )
x n x n 1
x n +1 = x n f (x n )
(x 0 + p ) + 2 (x 0 + p ) 5 = 0
3
x 03 + 3x 02 p + 3x 0 p 2 + p 3 2x 0 + 2p 5 = 0
p 3 + 3x 0 p 2 + (3x 02 + 2) p = 5 2x 0 x 03
The second and high-order polynomial terms in p were discarded to calculate a
numerical approximation p0 from 3 (x 02 + 2) p0 = 5 2x 0 x 03 .
Now p = p0 + q (q
much smaller than p0 ) is substituted into the polynomial for p, giving a polynomial in
numerical terms, calculated at each stage, became insignificant. The final result was
the
initial
estimate
x0
plus
the
results
of
the
polynomial
computations
76
RMIT University
Geospatial Science
x =
c + bg g 3
3g 2 b
f (a ) = a 3 ba c .
The
following comments on Raphson's technique, recorded in the Journal Book of the Royal
Society are noteworthy.
30 July 1690: Mr Halley related that Mr Ralphson [sic] had Invented a
method of Solving all sorts of Aquations, and giving their Roots in Infinite
Series, which Converge apace, and that he had desired of him an Equation
of the fifth power to be proposed to him, to which he return'd Answers true
to Seven Figures in much less time than it could have been effected by the
Known methods of Vieta.
17 December 1690: Mr Ralphson's Book was this day produced by E
Halley, wherin he gives a Notable Improvement of ye method of Resolution
of all sorts of Equations Shewing, how to Extract their Roots by a General
Rule,which doubles the known figures of the Root known by each Operation,
So yt by repeating 3 or 4 times he finds them true to Numbers of 8 or 10
places.
It is interesting to note here that Raphson's technique is compared to that of Vite,
while Newton's method is not mentioned, although it had, by then, appeared in Wallis'
Algebra.
In the preface to his tract of 1690, Raphson refers to Newton's work but
states that his own method is not only, I believe, not of the same origin, but also,
certainly, not with the same development. The two methods were long regarded by
users as distinct, but the historian of mathematics, Florian Cajori writing in 1911
recommended the use of the appellation Newton-Raphson and this is now standard in
mathematical texts describing Raphson's method with the notation of calculus.
The information above is drawn from the articles; Thomas, D. J., 1990, 'Joseph
Raphson, F.R.S.', Notes and Records of the Royal Society of London, Vol. 44, No. 2,
(July 1990) pp. 151-167, and Tjalling, J., 1995, 'Historical development of the NewtonRaphson method', SIAM Review, Vol. 37, No. 4, pp. 531-551.
77
RMIT University
Geospatial Science
(246)
and integration gives the area of a rectangle bounded by meridians 1, 2 and parallels
1, 2 as
2 2
A=
cos d d
(247)
1 1
A=
cos d d
0
1
2
= 2 cos d
(248)
cos =
a 2 (1 e 2 )
W4
cos
2
1
1
2
2
=
=
1
e
sin
(
)
2
4
W
(1 e 2 sin2 )
and
(249)
(250)
(1 + x )
= 1 2x + 3x 2 4x 3 + 5x 4
we have
1
= 1 + 2e 2 sin2 + 3e 4 sin 4 + 4e 6 sin6 + 5e 8 sin 8 + 6e 10 sin10 +
4
W
and hence
78
RMIT University
Geospatial Science
A = 2a (1 e
2
cos x sinn x dx =
sinn +1 x
, term-by-term integration gives
n +1
2
2e 2
3e 4
4e 6
sin 3 +
sin5 +
sin7 +
A = 2a (1 e ) sin +
3
5
7
1
(251)
To simplify this expression the powers of sin can be expressed in terms of multiple
angles using the standard form
sin2n 1 =
n 1
(1)
22n 2
2n 1
2n 1
sin (2n 1)
sin (2n 3) +
1
2 sin (2n 5)
2n 1
2n 1
sin (2n 7 ) + (1)n 1
sin
(252)
n 1
3
2n 1
computed using
Using equation (252) and the binomial coefficients Bm2n 1 =
m
3
1
sin sin 3
4
4
sin5 =
5
5
1
sin sin 3 + sin 5
8
8
16
sin7 =
35
35
21
1
sin sin 3 +
sin 5 sin 7
64
64
64
64
2e 2 3
1
sin sin 3
A = 2a 2 (1 e 2 ) sin +
3 4
4
3e 4 5
5
1
5 8
16
16
2
4e 6 35
21
7
1
sin sin 3 +
+
sin
5
sin
7
7 64
64
64
64
1
79
RMIT University
Geospatial Science
(253)
(254)
x y
x + y
Using the trigonometric relationship sin x sin y = 2 sin
cos
and with the
2
2
mean latitude m and latitude difference as
m =
2 + 1
2
and = 2 1
(255)
A = 2a 2 (1 e 2 ) C 1 sin
cos m + C 3 sin 3
cos 3m
2
2
+C 5 sin 5
cos 5m + C 7 sin 7
cos 7m +
2
2
(256)
80
RMIT University
Geospatial Science
)
1
cos
(1 e 2 sin2 )
(257)
If the lower terminal of latitude is the equator (1 = 0) and the upper terminal is the pole
A = 4a (1 e
2
)
0
cos
(1 e 2 sin2 )
(258)
transformed to x = 0,1
2
A = 4a (1 e
2
)
0
dx
(1 e x )
2
2 2
= 4a 2 (1 e 2 ) I
(259)
I =
dx
(1 e x )
2 2 2
1
e dx
1
=
2
e 0 (1 e 2x 2 )
To evaluate the integral, let u = ex then du = e dx and with the terminals x = 0,1
transformed to u = 0, e the integral I becomes
e
1
du
I =
2
e 0 (1 u 2 )
Using the standard result
du
(1 u )
2 2
1
du
1 1 1 + u 1 u
1 1 + e 1 e
ln
+
=
ln
+
2
2
e 0 (1 u 2 )
e 4 1 u 2 1 u 0
4e 1 e 2e 1 e 2
e
I =
1 1 + u 1 u
ln
+
gives the integral I as
4 1 u 2 1 u 2
e
Substituting I into equation (259) and re-arranging gives the surface area of the ellipsoid
1 e 2 1 + e
A = 2a 1 +
ln
1 e
2e
2
(260)
81
RMIT University
Geospatial Science
N
r d
Gr
een
wi
ch
dz
b
dr
dz
element of volume dV
r
eq uator
y
d
r2 z2
+
= 1 and the volume of
a 2 b2
the ellipsoid is twice the volume of the hemi-ellipsoid or Vellipsoid = 2Vhemi where
Vhemi =
r2
2 z 2 (r )
r1
(r )
dV =
1 z1
r2
dz d r dr =
r1
z2 r
dz d r dr
z1 r
( )
( )
(i)
for z, from z 1 = 0 to z 2 =
(ii)
for , from 1 = 0 to 2 = 2
(iii)
for r, from r1 = 0 to r2 = a
dz = [z ]
z2
z1
= z 2 z1 =
z1
b
a2 r2
a
82
RMIT University
Geospatial Science
a 2 r 2 d = 2
b 2
a r2
a
Vhemi =
r1
b
b
2
a 2 x 2 r dr = 2
a
a
b
=
a
r (a 2 r 2 )2 dr
2r (a 2 r 2 )2 dr
b
=
a
u 2 du =
a2
b
a
2 3
2 2
ab
u2 =
3 a 2
3
0
4 2
ab
3
(261)
A sphere having a radius equal to the mean of the 3 semi-axes of the ellipsoid
Rm =
2a + b
3
(262)
x 2 y2 z 2
+
+
=1
a 2 b2 c2
and has three semi-axes a, b, and c. The term tri-axial could be used to distinguish
[Note here that in mathematics an ellipsoid is a surface defined by
x 2 y2 z 2
+
+
= 1]
a 2 a 2 b2
A sphere having the same surface area as the ellipsoid [see equation (260)]
4RA2 = 2a 2 1 +
ln
1 e
2e
giving
83
RMIT University
Geospatial Science
a 2
1 e 2 1 + e
R = 1 +
ln
1 e
2
2e
2
A
3.
(263)
A sphere having the same volume as the ellipsoid [see equation (261)]
volume of sphere = volume of ellipsoid
4
4
RV3 = a 2b
3
3
giving
RA3 = a 2b
4.
(264)
A sphere having the same quadrant distance as the ellipsoid [see equation (228)]
9
225 4
= a (1 n ) (1 n 2 ) 1 + n 2 +
n +
2
4
64
giving
9
225 4
RQ = a (1 n ) (1 n 2 ) 1 + n 2 +
n +
4
64
(265)
84
RMIT University
Geospatial Science
of the Earth's rotation. The geometric parameters e 2 and f of an ellipsoid (known as the
normal ellipsoid) can be derived from these defining parameters as well as the gravitational
potential of the ellipsoid and the value of gravity on the ellipsoid (known as normal
gravity).
Table 1 shows the geometric parameters of various ellipsoids.
Date
Name
a (metres)
1/f
1830
Airy
6377563.396
299.324964600
1830
Everest (India)
6377276.345
300.801700000
1880
Clarke
6378249.145
293.465000000
1924
International
6378388 (exact)
297.0 (exact)
1966
6378160 (exact)
298.25 (exact)
1967
6378160 (exact)
298.247167427
1980
6378137 (exact)
298.257222101
1984
6378137 (exact)
298.257223563
The Geodetic Reference System 1980 (GRS80), adopted by the XVII General Assembly of
the IUGG in Canberra, December 1979 is the current estimate with a = 6378137 m ,
GM = 3986005 108 m 3s2 , J 2 = 108263 108 and = 7292115 1011 rad s-1 (BG 1988).
The World Geodetic System 1984 (WGS84), the datum for the Global Positioning System
(GPS), is based on the GRS80, except that the dynamical form factor of the Earth is
expressed in a modified form, causing very small differences between derived parameters of
the GRS80 and WGS84 ellipsoids (NIMA 2000).
negligible for all practical purposes (e.g., a difference of 0.0001 m in the semi-minor axes).
The Geocentric Datum of Australia (GDA) uses the GRS80 ellipsoid as its reference
ellipsoid.
85
RMIT University
Geospatial Science
a = 6378137 metres
flattening:
f = 1 298.257222101
These two defining parameters and other computed constants for the GRS80 ellipsoid are:
a =
b = a (1 f )
a
1 f
c =
e 2 = f (2 f )
f (2 f )
= 6.739 496 775e-003
2
(1 f )
f = 1 298.257222101 = 3.352 810 681e-003
e 2 =
f
= 1.679 220 395e-003
2 f
quadrant distance Q = 10 001965.729 metres
n =
surface area
A = 5.10065622e+014 square-metres
volume
V = 1.08320732e+021 cubic-metres
e 4 = 4.481472 4
e-005
e 4 = 4.542 0817
e-005
e 6 = 3.000 07
e-007
e 6 = 3.06113
e-007
e 8 = 2.008
e-009
e 8 = 2.063
e-009
e 10 = 1.3
e-011
e 10 = 1.4
e-011
f 2 = 1.124133 9
e-005
n 2 = 2.819 7811
e-006
f 3 = 3.769 0
e-008
n 3 = 4.735 0
e-009
f 4 = 1.26
e-010
n 4 = 7.95
e-012
f 5 = 4.2
e-013
n 5 = 1.3
e-014
86
RMIT University
Geospatial Science
From the above, it is clear that the higher powers of n are much smaller than the higher
powers of the other constants, so that in general, a series involving powers of n will
converge more rapidly than a series involving the powers of other constants. See equations
for meridian distance for an example.
The radii of equivalent spheres are:
V
and the radii of curvature are:
prime vertical =
meridian
mean
= 6 386175.289 metres
c
= 6 359 422.962 metres
W
V3
a (1 f )
c
= =
= 2 = 6 372 785.088 metres
2
W
V
=
m
a
c
=
W V
a (1 e 2 )
3
The meridian distance from the equator to P is [using equation (224)] with the coefficients
b0 , b2 , etc.
87
RMIT University
Geospatial Science
9
225 4
b0 = 1 + n 2 +
n +
4
64
3
45
b2 = n + n 3 +
2
16
1 15 2 105 4
b4 = n +
n +
2 8
32
1 35
b6 = n 3 +
3 16
1 315 4
b8 =
n +
4 128
88
RMIT University
Geospatial Science
Gre
en
wi
ch
P
Q h
b
a
H D
equ ator
a
y
p
coordinates as P, but h = 0 . Q is the projection of P onto the surface via the normal.] P
also has Cartesian coordinates x,y,z. The origin of both coordinate systems is at the centre
O of the ellipsoid.
Figure 25 (a) shows the meridian ellipse (meridian normal section) in the meridian plane
zOp. P is in this plane and is connected to the ellipse via the normal. The normal cuts
the ellipse at Q, the equatorial plane at D and the z-axis at H. The centre of curvature of
the meridian ellipse at Q lies on the normal at C. The centre of curvature of the prime
vertical normal section (an ellipse in a plane perpendicular to the meridian plane) at Q lies
on the normal at H. P is the projection of P onto the equatorial plane of the ellipsoid.
89
RMIT University
Geospatial Science
Also, the auxiliary circle is shown with Q projected onto the circle via a normal to the paxis, and the parametric latitude and geocentric latitude of Q are shown.
Figure 25(b) shows the equatorial plane of the ellipsoid and P is the projection of P.
auxiliary circle
N
Q P
O
H
h
P p
o
at
equ
r
P
p
x
(a) P related to the meridian ellipse
a
c
= = radius of curvature in prime vertical section
W V
a
c
b
(1 e 2 ) = (1 e 2 ) = (1 e 2 ) =
W
V
V
a
c
OH = e 2 sin = e 2 sin = e 2 sin
W
V
a
c
DH = e 2 = e 2 = e 2
W
V
QD =
90
RMIT University
Geospatial Science
a (1 e 2 )
b
z =
+ h sin = + h sin
V
= ( + h ) cos cos
= ( + h ) cos sin
(266)
= { (1 e 2 ) + h } sin
0, + h 0, (1 e 2 ) + h 0 and 12 12 ,
<
where
a
c
=
W V
2
V = 1 + e 2 cos2
=
b = a (1 f )
e 2 = f (2 f )
b
V
2
W = 1 e 2 sin2
(1 e 2 ) =
a
1 f
f (2 f )
e 2 =
2
(1 f )
c =
(267)
91
RMIT University
Geospatial Science
(268)
where p is the perpendicular distance from the z-axis (the rotational axis)
p = x 2 + y2 0
(269)
y
x
and cos = . These relationships seem
p
p
y
preferable to evaluate than 6 tan = , since each function, sine and cosine, is
x
2 -periodic whereas tangent is -periodic .
Choosing sin =
y
and solving for gives
p
0, if x > 0, y = 0
thus (, ] as desired.
tan = y x is the usual formula for evaluating , but requires inspection of x to avoid division by zero,
and then inspection of the signs of x and y to determine the correct value <
since
12 < arctan y x < 12 . Note: the atan2 (y, x ) function, common to many computer languages, will return
< .
92
RMIT University
Geospatial Science
y
p
(270)
x
1,
if
<
0
x
Choosing to evaluate from cos =
and with similar reasoning to the development of
p
y
from sin =
leads to
p
= sgn (y ) arccos
x
p
(272)
(273)
(274)
or
using =
a
1 e 2 sin2
93
RMIT University
Geospatial Science
Using equations (270), (272), (273) and the last of equations (268), the transformation
x , y, z , , h is accomplished by
p tan = z + e 2 sin
p
h =
cos
(275)
This transformation is not straightforward. While can be readily computed from y and
p, or x and p the same cannot be said for , (and thus h) as there are no simple
relationships linking with x,y,z [see the second of equations (275) where functions of
are on both sides of the equation].
documented in the literature and they fall into two categories: iterative solutions and
direct solutions.
Direct solutions for involve the formation of quartic equations (an algebraic equation of
the 4th-degree) which are reduced to cubic equations having a single real root [e.g., Paul
1973, Ozone 1985, Lapaine 1990 and Vermeille 2002].
Iterative solutions are generally simpler than direct solutions and some do not require a 2nd
iteration if points are reasonably close to the ellipsoid 5, 000 m h 10, 000 m , and they
fall into two groups; (i) trigonometric [e.g., Bowring 1973, Borkowski 1989, Laskowski 1991
and Jones 2002] where formula involve trigonometric relationships and (ii) vector [e.g., Lin
and Wang 1995 and Pollard 2002] where formula are derived from vector relationships.
Iterative solutions are usually easier to program and generally require less evaluations of
square-roots, exponentiations and trigonometric functions than direct solutions.
These
94
RMIT University
Geospatial Science
namely
p tan = z + e 2 sin
(276)
This procedure is
repeated until the difference between successive LHS values n , n +1 reaches an acceptable
limit. Thus the iteration, providing certain conditions are met (see below), converges to a
solution for p tan and hence .
The starting value 0 is obtained from the relationship between geocentric and geodetic
z
2
latitude: (1 f ) tan 0 = tan
giving
p
tan 0 =
z
2
p (1 f )
(277)
95
RMIT University
Geospatial Science
So, provided that p > ce 2 the criteria for convergence should be satisfied. For the GRS80
ellipsoid ce 2 43130 m .
f (n )
f (n )
(278)
d
{ f ()} is given by
d
f () = e 2 cos + e 2 sin
Now =
(279)
d
p
d cos2
(280)
c
dV
e 2
and noting that V 2 = 1 + e 2 cos2 and
= cos sin then
V
d
V
d
c dV
c
= 2
= 3 e 2 cos sin
d
V d V
(281)
3
V
cos2
V
c
p
= 3 e 2 cos (V 2 + e 2 e 2 cos2 )
V
cos2
c
p
= 3 e 2 cos (V 2 + e 2 + 1 V 2 )
V
cos2
c
p
= 3 e 2 cos (1 + e 2 )
V
cos2
e 2
and equation (282) becomes
e2
c
p
f () = 3 e 2 cos
V
cos2
(282)
(283)
96
RMIT University
Geospatial Science
a2
c2
, and equation (282) becomes
=
b2
a2
c3 2
p
e cos
2 3
aV
cos2
(284)
A starting value 0 can be obtained from equation (277) and the iteration continued until
f (n )
f (n )
magnitude.
(285)
pC = ae 2 cos3
and the expression for tan in terms of the parametric latitude can be obtained from
Figure 26 as
z
N
b
a
O
ae 2 cos 3
H
be 2 sin3
P p
97
RMIT University
Geospatial Science
tan =
z + be 2 sin 3
p ae 2 cos3
(286)
Equation (286) in conjunction with tan = (1 f ) tan can be solved simultaneously for
, and if necessary , iteratively.
Similarly to Successive Substitution, a starting value 0 is obtained from the relationship
z
between geocentric and parametric latitude: (1 f ) tan 0 = tan giving
p
tan 0 =
z
p (1 f )
(287)
the
maximum error in , induced by using only a single iteration, is 0.000 000 030 .
For points in space where h > 10, 000 m a second (or perhaps even a third) iteration of
Bowring's equation may be required.
from
tan = (1 f ) tan and using this new value in the RHS of equation (286) to give an
improved vale of ; and so on until the difference between successive LHS values n , n +1
reaches an acceptable limit.
It should be noted here that for h = 26,000,000 m (A GPS satellite has an approximate
orbital height of 20,000,000 m) only two iterations are required for acceptable accuracy.
98
RMIT University
Geospatial Science
Recalling section 1.2.1 Differential Geometry of Space Curves, a curve C in space is defined
as the locus of the terminal points P of a position vector r (t ) defined by a single scalar
parameter t,
r (t ) = x (t ) i + y (t ) j + z (t ) k
(288)
and i, j, k are fixed unit Cartesian vectors in the directions of the x,y,z coordinate axes.
If this curve C is constrained to lie on a surface
= (x , y, z ) = constant
(289)
then the scalar components of the position vector (288) must satisfy equation (289); thus
(x (t ), y (t ), z (t )) = constant
(290)
As the parameter t varies, the terminal point of the vector sweeps out C on the surface
and if s is the arc length from some point on C, then s is a function of t and x,y,z are
functions of s. The unit tangent vector t of the curve C in the direction of increasing s is
given by
t = dr = dx i + dy j + dz k
ds
ds
ds
ds
(291)
and at a fixed point P on the surface (x , y, z ) = constant there are an infinite number of
curves C passing through P.
plane of the surface at P and the normal to this plane is the surface normal vector of at
P. Differentiating equation (290) with respect to s, we obtain via the chain rule
dx dy dz
+
+
=0
x ds
y ds
z ds
(292)
i+
j+
k
x
y
z
(293)
i+
j+
k
x
y
z
(294)
so that
=
(295)
99
RMIT University
Geospatial Science
Because this scalar product is zero, must be perpendicular to the tangent plane and so
in the direction of the surface normal. Lin & Wang use this result in the following way.
The Cartesian equation of the ellipsoid at Q is
xQ2
a2
yQ2
zQ2
a2
=1
b2
(296)
Using equation
(294) then
Q =
2xQ
a
i+
2yQ
a
2zQ
j+
b2
(297)
The vector QP is
QP = (x P xQ ) i + (yP yQ ) j + (z P zQ ) k
(298)
and QP (being normal to the surface at Q) is also a scalar (denoted by q) multiple of the
(299)
Equating the scalar components of vectors (298) and (299), and re-arranging gives
xQ
a
yQ
ax P
;
a + 2q
2
zQ
ayP
;
a + 2q
2
bz P
b + 2q
2
(300)
Squaring each equation above, then substituting into equation (296) gives
f (q ) =
a 2 pP2
(a 2 + 2q )
b 2z P2
=1
(b 2 + 2q )
(301)
and q can be obtained by using Newton-Raphson iteration for the real roots of the
equation f (q ) = 0 given in the form of an iterative equation
qn +1 = qn
f (qn )
f (qn )
(302)
a 2 pP2
(a 2 + 2q )
b 2z P2
(b 2 + 2q )
f (qn )
(303)
100
RMIT University
Geospatial Science
a 2 p 2
b 2z 2
f (qn ) = 4 2 P 3 + 2 P 3
(a + 2q )
(b + 2q )
(304)
q0 =
ab (a 2z P2 + b 2 pP2 ) 2 a 2b 2 (a 2z P2 + b 2 pP2 )
(305)
2 (a 4z P2 + b 4 pP2 )
(306)
b 2z
zQ = 2 P
b + 2q
and the latitude and height h obtained from
tan =
zQ
(1 f ) pQ
h =
(pP pQ )
zQ
(1 e 2 ) pQ
(307)
+ (z P zQ )
It should be noted here that for all Earth-bound points 5, 000 m h 10, 000 m Lin &
Wang's method requires only a single iteration for acceptable accuracy.
For points in space where h > 10, 000 m a second (or perhaps even a third) iteration may
be required and for h = 26,000,000 m (A GPS satellite has an approximate orbital height
of 20,000,000 m) two iterations are required for acceptable accuracy.
101
RMIT University
Geospatial Science
(308)
a
a
=
1
W
(1 e 2 sin2 )2
a
=
1
(cos2 + sin2 e 2 sin2 )2
a
=
1
(cos2 + (1 e 2 ) sin2 )2
ae 2 sin
(cos
+ (1 e 2 ) sin2 )
1
2
ae 2 tan
(1 + (1 e ) tan )
2
1
2
(309)
p 2 a 2e 4
2pz
z2
2
p 2 tan 4 2pz tan 3 + z 2 +
tan
tan
+
=0
1 e 2
1 e2
1 e2
(310)
p 2 a 2e 4
=
1 e2
(311)
Letting
2zp 2
z 2p2
p
tan
+
=0
1 e2
1 e2
(312)
102
RMIT University
Geospatial Science
Equation (312) is a quartic equation in tan , for which there are direct solutions for the
four possible values of tan ; which could all be real, or all complex, or some real and the
others complex. It turns out that this equation can be expressed in terms of a subsidiary
value which itself a function of the roots of a cubic equation having a single real root.
Hence by means of appropriate substitutions we are able to find the single real value of
tan from equation (312). The method of solving quartic equations, by reduction to a
cubic whose solution is known, was first developed in 1540 by Lodivico Ferrari (1522-1565)
who resided in Bologna, Papal States (now Italy), the technique set out below is a
modification of his general method.
p tan = +
Let
z
2
(313)
where is a subsidiary variable; then expressions for p 2 tan2 , p 3 tan 3 and p 4 tan 4
can be substituted into equation (312) to give, after some algebra, a quartic equation in ,
but with no 3 -term.
z 2 2
z 2
2
+ z + z = 0
4 16
2
4
where
p 2 + a 2e 4
1 e2
(314)
(315)
(316)
where t1, t2 , t3 are the roots of the cubic equation t 3 + a1t 2 + a2t + a 3 = 0 , the following
relationships can be established, noting that A = t1 + t2 + t3
2 = (t1 + t2 + t3 ) + 2 t1t2 + 2 t1t3 + 2 t2t3
= A + 2 t1t2 + 2 t1t3 + 2 t2t3
(
2A (
)
A) = 4A
(317)
103
RMIT University
Geospatial Science
(318)
(319)
Comparing the coefficients of 2 and , and the constant terms in equations (314) and
(319) gives the following relationships
z2
t1 + t2 + t3 =
4
2
2 2
z
t1t2t3 =
64
2 z 2
z2
az
t1 +
4
2
4 t1
(320)
Substituting expressions for a1, a2 , a 3 into the general cubic equation gives
z 2 2 2 z 2
2z 2
t + t +
=0
t
2
16
4
8
64
(321)
+ z 2
z2
t =
u +
6
12 6
(322)
A further substitution
(323)
1
2
{ q+
3
q2 1 + 3 q q2 1
(324)
104
RMIT University
Geospatial Science
where
q = 1+
27z 2 ( 2 2 )
2 ( + z 2 )
(325)
Thus having x,y,z (hence p) for a point related to an ellipsoid, tan is obtained by
computing the following variables in order: from equation (315), from equation (311),
q from equation (325), u1 from equation (324), t1 from equation (322), from equation
(320) noting that all square-roots in equation (320) have the same sign as z, and finally
tan from equation (313).
105
RMIT University
Geospatial Science
3 MATLAB FUNCTIONS
MATLAB (an acronym for MATrix LABoratory) is a powerful computer program
designed to perform scientific calculations. It was originally designed to perform matrix
mathematics but has evolved into a flexible computing system capable of solving almost
any technical problem.
MATLAB language) that looks like the C computer language, and an extensive suite of
technical functions. MATLAB functions can be easily constructed using the MATLAB
editor and executed from the MATLAB command window.
f = 1 flat . The function is designed to be run from the MATLAB command window with
output from the function printed in the MATLAB command window.
MATLAB function ellipsoid_1.m
function ellipsoid_1(a,flat)
%
% ellipsoid_1(a,flat) Function takes semi-major axis 'a' of ellipsoid
%
and reciprocal of flattening 'flat' and computes some geometric
%
constants of the ellipsoid. Note that the flattening f = 1/flat.
%
e.g., ellipsoid_1(6378137,298.257222101) will return the constants of
%
the GRS80 ellipsoid
%==========================================================================
% Function: ellipsoid_1
%
% Useage:
ellipsoid_1(a,flat);
%
% Author:
% Rod Deakin,
% School of Mathematical and Geospatial Sciences, RMIT University,
% GPO Box 2476V, MELBOURNE VIC 3001
% AUSTRALIA
% email: [email protected]
%
% Date:
% Version 1.0 31 January 2008
%
% Remarks:
% Function takes semi-major axis 'a' of ellipsoid and reciprocal of
% flattening 'flat' and computes some geometric constants of the
% ellipsoid. Note that the flattening f = 1/flat.
106
RMIT University
Geospatial Science
%
% References:
% [1] Deakin, R.E. and Hunter, M.N., 2008, 'Geometric Geodesy', School of
%
Mathematical and Geospatial Sciences, RMIT University, February
%
2008.
%
% Variables:
% A
- surface area of ellipsoid
% a
- semi-major axis of ellipsoid
% b
- semi-minor axis of ellipsoid
% c
- polar radius of curvature
% e
- 1st eccentricity of ellipsoid
% e2
- 1st eccentricity squared
% e4,e6,
% e8,e10
- powers of e2
% ep2
- 2nd eccentricity squared
% ep4,ep6,
% ep8,ep10 - powers of ep2
% f
- flattening of ellipsoid
% f2,f3,f4,
% f5
- powers of flattening
% flat
- reciprocal of flattening f = 1/flat
% n
- ellipsoid constant
% n2,n3,n4,
% n5
- powers of n
% Q
- Quadrant distance of ellipsoid
% Ra
- radius of sphere having same surface area as ellipsoid
% Rm
- radius of sphere having mean of ellipsoid axes
% Rv
- radius of sphere having same volume as ellipsoid
% Rq
- radius of sphere having same quadrant distance as ellipsoid
% V
- volume of ellipsoid
%=========================================================================
% compute flattening f and powers of f
f = 1/flat;
f2 = f*f;
f3 = f2*f;
f4 = f3*f;
f5 = f4*f;
% compute semi-minor axis b and polar radius of curvature c
b = a*(1-f);
c = a/(1-f);
% compute eccentricity squared e2, eccentricity e and powers of e2
e2 = f*(2-f);
e
= sqrt(e2);
e4 = e2*e2;
e6 = e4*e2;
e8 = e6*e2;
e10 = e8*e2;
% compute 2nd eccentricity squared (e_primed squared) and powers of ep2
ep2 = e2/(1-e2);
ep4 = ep2*ep2;
ep6 = ep4*ep2;
ep8 = ep6*ep2;
ep10 = ep8*ep2;
% compute n
n = f/(2-f);
n2 = n*n;
n3 = n2*n;
n4 = n3*n;
n5 = n4*n;
107
RMIT University
Geospatial Science
108
RMIT University
Geospatial Science
Ellipsoid Constants:
===================
semi-major axis
semi-minor axis
polar radius of curvature
eccentricity squared
2nd eccentricity squared
flattening
denominator of flattening
quadrant distance
surface area
volume
radii of equivalent spheres
mean radius
area
volume
quadrant
powers of constants
e2 = 6.694380023e-003
e4 = 4.481472389e-005
e6 = 3.000067923e-007
e8 = 2.008359477e-009
e10 = 1.344472156e-011
f
f2
f3
f4
f5
=
=
=
=
=
3.352810681e-003
1.124133946e-005
3.769008303e-008
1.263677129e-010
4.236870177e-013
a
b
c
e2
ep2
f
flat
n
Q
A
V
=
=
=
=
=
=
=
=
=
=
=
6378137.000 metres
6356752.314 metres
6399593.626 metres
6.694380023e-003
6.739496775e-003
3.352810681e-003
298.257222101
1.679220395e-003
10001965.729 metres
5.10065622e+014 square-metres
1.08320732e+021 cubic-metres
Rm
Ra
Rv
Rq
=
=
=
=
ep2
ep4
ep6
ep8
ep10
=
=
=
=
=
6.739496775e-003
4.542081679e-005
3.061134483e-007
2.063050598e-009
1.390392285e-011
n
n2
n3
n4
n5
=
=
=
=
=
1.679220395e-003
2.819781134e-006
4.735033988e-009
7.951165642e-012
1.335175951e-014
6371008.771
6371007.181
6371000.790
6367449.146
metre
metres
metres
metres
>>
109
RMIT University
Geospatial Science
Helmert's formula, and is designed to be run from the MATLAB command window with
output from the function printed in the MATLAB command window.
The third function, latitude2.m, computes the latitude given the meridian distance m
and the ellipsoid parameters a (semi-major axis of ellipsoid), flat (the denominator of the
flattening f ). The function uses the Newton-Raphson iterative scheme to compute the
latitude from Helmert's formula [equation (224)] and is designed to be run from the
MATLAB command window with output from the function printed in the MATLAB
command window.
110
RMIT University
Geospatial Science
111
RMIT University
Geospatial Science
112
RMIT University
Geospatial Science
%
formulae (see function mdist.m). The reverse operation, given a
%
meridian distance on a defined ellipsoid to calculate the latitude,
%
can be achieved by series formulae published in THE AUSTRALIAN GEODETIC
%
DATUM Technical Manual Special Publication 10, National Mapping Council
%
of Australia, 1986 (section 4.4, page 24-25). The development of these
%
formulae are given in Lauf, G.B., 1983, GEODESY AND MAP PROJECTIONS,
%
Tafe Publications, Vic., pp.35-38.
%
This function is generally used to compute the "footpoint latitude"
%
which is the latitude for which the meridian distance is equal to the
%
y-coordinate divided by the central meridian scale factor, i.e.,
%
latitude for m_dist = y/k0.
%-------------------------------------------------------------------------% degree to radian conversion factor
d2r = 180/pi;
% calculate flatteninf f and ellipsoid constant n and powers of n
f = 1/flat;
n = f/(2.0-f);
n2 = n*n;
n3 = n2*n;
n4 = n3*n;
% calculate the mean length an arc of one radian on the meridian
g = a*(1-n)*(1-n2)*(1+9/4*n2+225/64*n4);
% calculate sigma (s) and powers of sigma
s = mdist/g;
s2 = 2.0*s;
s4 = 4.0*s;
s6 = 6.0*s;
s8 = 8.0*s;
% calculate the latitude (in radians)
lat = s + (3*n/2 - 27/32*n3)*sin(s2)...
+ (21/16*n2 - 55/32*n4)*sin(s4)...
+ (151/96*n3)*sin(s6)...
+ (1097/512*n4)*sin(s8);
% convert latitude to degrees
lat = lat*d2r;
% print result to screen
fprintf('\n a = %12.4f',a);
fprintf('\n f = 1/%13.9f',flat);
[D,M,S] = DMS(lat);
if D == 0 && lat < 0
fprintf('\nLatitude =
-0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLatitude
= %4d %2d %9.6f (D M S)',D,M,S);
end
fprintf('\nMeridian dist = %15.6f',mdist);
fprintf('\n\n');
113
RMIT University
Geospatial Science
114
RMIT University
Geospatial Science
%
% Variables:
%
a
- semi-major axis of spheroid
%
b0,b1,b2,... coefficients in Helmert's formula
%
corrn
- correction term in Newton-Raphson iteration
%
count
- iteration number
%
d2r
- degree to radian conversion factor 57.29577951...
%
F
- a function of latitude (Helmert's formula)
%
Fdash
- the derivative of F
%
f
- flattening of ellipsoid
%
flat
- denominator of flattening f = 1/flat
%
lat
- latitude
%
mdist
- meridian distance
%
n
- eta, n = f/(2-f)
%
n2,n4, - powers of eta
%
% Remarks:
%
For an ellipsoid defined by semi-major axis (a) and flattening (f) the
%
meridian distance (mdist) can be computed by series expansion
%
formulae (see function mdist.m). The reverse operation, given a
%
meridian distance on a defined ellipsoid to calculate the latitude,
%
can be achieved by using Newton's Iterative scheme.
%-------------------------------------------------------------------------% degree to radian conversion factor
d2r = 180/pi;
% calculate flattening f and ellipsoid constant n and powers of n
f = 1/flat;
n = f/(2.0-f);
n2 = n*n;
n3 = n2*n;
n4 = n3*n;
% coefficients in Helmert's series expansion for meridian distance
b0 = 1+(9/4)*n2+(225/64)*n4;
b2 = (3/2)*n+(45/16)*n3;
b4 = (1/2)*((15/8)*n2+(105/32)*n4);
b6 = (1/3)*((35/16)*n3);
b8 = (1/4)*((315/128)*n4);
% set the first approximation of the latitude and then Newton's iterative
% scheme where F is the function of latitude and Fdash is the derivative of
% the function F
lat
= mdist/a;
corrn = 1;
count = 0;
while (abs(corrn)>1e-10)
F = a*(1-n)*(1-n2)*(b0*lat...
- b2*sin(2*lat)...
+ b4*sin(4*lat)...
- b6*sin(6*lat)...
+ b8*sin(8*lat)) - mdist;
Fdash = a*(1-n)*(1-n2)*(b0...
- 2*b2*cos(2*lat)...
+ 4*b4*cos(4*lat)...
- 6*b6*cos(6*lat)...
+ 8*b8*cos(8*lat));
corrn = -F/Fdash;
lat = lat + corrn;
count = count+1;
end
% convert latitude to degrees
lat = lat*d2r;
115
RMIT University
Geospatial Science
116
RMIT University
Geospatial Science
MATLAB functions mdist.m, latitude.m and latitude2.m call functions DMS.m and
dms2deg.m to convert decimal degrees to degrees, minutes and seconds (for printing) and
ddd.mmss format to decimal degrees. These functions are shown below.
function DecDeg=dms2deg(DMS)
% DMS2DEG
% Function to convert from DDD.MMSS format to decimal degrees
x = abs(DMS);
D = fix(x);
x = (x-D)*100;
M = fix(x);
S = (x-M)*100;
DecDeg = D + M/60 + S/3600;
if(DMS<0)
DecDeg = -DecDeg;
end
return
117
RMIT University
Geospatial Science
118
RMIT University
Geospatial Science
% References:
% [1] Deakin, R.E. and Hunter, M.N., 2008, GEOMETRIC GEODESY - PART A,
%
School of Mathematical and Geospatial Sciences, RMIT University,
%
Melbourne, AUSTRALIA, March 2008.
% [2] Gerdan, G.P. & Deakin, R.E., 1999, 'Transforming Cartesian
%
coordinates X,Y,Z to geogrpahical coordinates phi,lambda,h', The
%
Australian Surveyor, Vol. 44, No. 1, pp. 55-63, June 1999.
%-------------------------------------------------------------------------% calculate flattening f and ellipsoid constant e2
f
= 1/flat;
e2 = f*(2-f);
% compute radii of curvature for the latitude
[rm,rp] = radii(a,flat,lat);
%
p
X
Y
Z
d2r = 180/pi;
a = 6378137;
flat = 298.257222101;
lat = -50/d2r;
lon = -150/d2r;
h = 10000;
[X,Y,Z] = Geo2Cart(a,flat,lat,lon,h);
[X,Y,Z]'
ans =
-3563081.36230554
-2057145.98367164
-4870449.48202417
>>
119
RMIT University
Geospatial Science
Cart2Geo_Lin.m is an iterative solution using Lin and Wang's method and the fifth
function: Cart2Geo_Paul.m is a direct solution using Paul's method.
In the the output from each of these functions shown below, x,y,z Cartesian coordinates
are first computed for a given , ,h and ellipsoid using Geo2Cart.m and then these x,y,z
coordinates are used in the function to compute , ,h .
120
RMIT University
Geospatial Science
% Purpose:
%
Function Cart2Geo_Substitution() will compute latitude, longitude
%
(both in radians) and height of a point related to an ellipsoid
%
defined by semi-major axis (a) and denominator of flattening (flat)
%
given Cartesian coordinates X,Y,Z.
%
% Variables:
%
a
- semi-major axis of ellipsoid
%
count
- integer counter for number of iterations
%
corrn
- correction to approximate value
%
d2r
- degree to radian conversion factor = 57.29577951...
%
e2
- 1st eccentricity squared
%
f
- flattening of ellipsoid
%
flat
- denominator of flattening f = 1/flat
%
h
- height above ellipsoid
%
lat
- latitude (radians)
%
lon
- longitude (radians)
%
p
- perpendicular distance from minor-axis of ellipsoid
%
rm
- radius of curvature of meridian section of ellipsoid
%
rp
- radius of curvature of prime vertical section of ellipsoid
%
% Remarks:
%
This function uses Successive Substitution, see Refences [1] and [2].
%
% References:
% [1] Gerdan, G.P. & Deakin, R.E., 1999, 'Transforming Cartesian
%
coordinates X,Y,Z to geogrpahical coordinates phi,lambda,h', The
%
Australian Surveyor, Vol. 44, No. 1, pp. 55-63, June 1999.
% [2] Deakin, R.E. and Hunter, M.N., 2008, GEOMETRIC GEODESY - PART A,
%
School of Mathematical and Geospatial Sciences, RMIT University,
%
Melbourne, AUSTRALIA, March 2008.
%-------------------------------------------------------------------------% Set degree to radian conversion factor
d2r = 180/pi;
% calculate flattening f and ellipsoid constant e2
f
= 1/flat;
e2 = f*(2-f);
% compute 1st approximation of geodetic latitude for the Simple Iteration
p
= sqrt(X*X + Y*Y);
lat = atan(Z/(p*(1-e2)));
corrn = 1;
count = 0;
while (abs(corrn)>1e-10)
%
Compute radii of curvature
[rm,rp] = radii(a,flat,lat);
%
Compute new approximation of latitude
new_lat = atan((Z+rp*e2*sin(lat))/p);
corrn = lat-new_lat;
count = count+1;
lat = new_lat;
end;
% compute radii of curvature for the latitude
[rm,rp] = radii(a,flat,lat);
% compute longitude and height
lon = atan2(Y,X);
h
= p/cos(lat) - rp;
% Print results to screen
fprintf('\n\nCartesian to Geographic - Simple Iteration');
fprintf('\n==========================================');
121
RMIT University
Geospatial Science
fprintf('\nEllipsoid:');
fprintf('\nsemi-major axis a = %13.3f',a);
fprintf('\nflattening
f = 1/%13.9f',flat);
fprintf('\nCartesian coordinates:');
fprintf('\nX = %13.3f',X);
fprintf('\nY = %13.3f',Y);
fprintf('\nZ = %13.3f',Z);
fprintf('\nGeodetic coordinates:');
[D,M,S] = DMS(lat*d2r);
if D == 0 && lat < 0
fprintf('\nLatitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLatitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
[D,M,S] = DMS(lon*d2r);
if D == 0 && lon < 0
fprintf('\nLongitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLongitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
fprintf('\nHeight
= %13.3f',h);
fprintf('\nIterations
= %3d',count);
fprintf('\n\n');
d2r = 180/pi;
a = 6378137;
flat = 298.257222101;
lat = -50/d2r;
lon = -150/d2r;
h = 10000;
[X,Y,Z] = Geo2Cart(a,flat,lat,lon,h);
[X,Y,Z]'
ans =
-3563081.36230554
-2057145.98367164
-4870449.48202417
>> [lat,lon,h] = Cart2Geo_Substitution(a,flat,X,Y,Z);
122
RMIT University
Geospatial Science
Ellipsoid:
semi-major axis a =
6378137.000
flattening
f = 1/298.257222101
Cartesian coordinates:
X = -3563081.362
Y = -2057145.984
Z = -4870449.482
Geodetic coordinates:
Latitude = -50 0 0.000000 (D M S)
Longitude = -150 0 0.000000 (D M S)
Height
=
10000.000
Iterations
=
3
>>
123
RMIT University
Geospatial Science
%
lon
- longitude (radians)
%
p
- perpendicular distance from minor-axis of ellipsoid
%
rm
- radius of curvature of meridian section of ellipsoid
%
rp
- radius of curvature of prime vertical section of ellipsoid
%
s
- sine(lat)
%
% Remarks:
%
This function uses Newton-Raphson Iteration, see Refences [1].
%
% References:
% [1] Deakin, R.E. and Hunter, M.N., 2008, GEOMETRIC GEODESY - PART A,
%
School of Mathematical and Geospatial Sciences, RMIT University,
%
Melbourne, AUSTRALIA, March 2008.
%-------------------------------------------------------------------------% Set degree to radian conversion factor
d2r = 180/pi;
% calculate flattening f and ellipsoid constant e2
f
= 1/flat;
e2 = f*(2-f);
ep2 = e2/(1-e2);
% compute 1st approximation of geodetic latitude for Newton_Raphson
% Iteration
p
= sqrt(X*X + Y*Y);
lat = atan(Z/(p*(1-e2)));
corrn = 1;
count = 0;
while (abs(corrn)>1e-10)
% Compute radii of curvature
[rm,rp] = radii(a,flat,lat);
s = sin(lat);
c = cos(lat);
%
Compute value of function and its derivative for approximate latitude
F = Z + rp*e2*s - p*s/c;
dF = rm*ep2*c - p/c/c;
corrn = F/dF;
new_lat = lat - corrn;
count = count+1;
lat = new_lat;
end;
% compute radii of curvature for the latitude
[rm,rp] = radii(a,flat,lat);
% compute longitude and height
lon = atan2(Y,X);
h
= p/cos(lat) - rp;
% Print results to screen
fprintf('\n\nCartesian to Geographic - Newton');
fprintf('\n================================');
fprintf('\nEllipsoid:');
fprintf('\nsemi-major axis a = %13.3f',a);
fprintf('\nflattening
f = 1/%13.9f',flat);
fprintf('\nCartesian coordinates:');
fprintf('\nX = %13.3f',X);
fprintf('\nY = %13.3f',Y);
fprintf('\nZ = %13.3f',Z);
fprintf('\nGeodetic coordinates:');
[D,M,S] = DMS(lat*d2r);
if D == 0 && lat < 0
fprintf('\nLatitude = -0 %2d %9.6f (D M S)',M,S);
124
RMIT University
Geospatial Science
else
fprintf('\nLatitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
[D,M,S] = DMS(lon*d2r);
if D == 0 && lon < 0
fprintf('\nLongitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLongitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
fprintf('\nHeight
= %13.3f',h);
fprintf('\nIterations
= %3d',count);
fprintf('\n\n');
d2r = 180/pi;
a = 6378137;
flat = 298.257222101;
lat = -50/d2r;
lon = -150/d2r;
h = 10000;
[X,Y,Z] = Geo2Cart(a,flat,lat,lon,h);
[X,Y,Z]'
ans =
-3563081.36230554
-2057145.98367164
-4870449.48202417
>> [lat,lon,h] = Cart2Geo_Newton(a,flat,X,Y,Z);
125
RMIT University
Geospatial Science
126
RMIT University
Geospatial Science
%
Melbourne, AUSTRALIA, March 2008.
%-------------------------------------------------------------------------% Set degree to radian conversion factor
d2r = 180/pi;
% calculate flattening f and ellipsoid constants e2, ep2 and b
f
= 1/flat;
e2 = f*(2-f);
ep2 = e2/(1-e2);
b
= a*(1-f);
% compute 1st approximation of parametric latitude psi
p
= sqrt(X*X + Y*Y);
psi = atan(Z/(p*(1-f)));
% compute latitude from Bowring's equation
s
= sin(psi);
s3 = s*s*s;
c
= cos(psi);
c3 = c*c*c;
lat = atan((Z+b*ep2*s3)/(p-a*e2*c3));
% compute radii of curvature for the latitude
[rm,rp] = radii(a,flat,lat);
% compute longitude and height
lon = atan2(Y,X);
h
= p/cos(lat) - rp;
% Print results to screen
fprintf('\n\nCartesian to Geographic - Bowring''s method');
fprintf('\n==========================================');
fprintf('\nEllipsoid:');
fprintf('\nsemi-major axis a = %13.3f',a);
fprintf('\nflattening
f = 1/%13.9f',flat);
fprintf('\nCartesian coordinates:');
fprintf('\nX = %13.3f',X);
fprintf('\nY = %13.3f',Y);
fprintf('\nZ = %13.3f',Z);
fprintf('\nGeodetic coordinates:');
[D,M,S] = DMS(lat*d2r);
if D == 0 && lat < 0
fprintf('\nLatitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLatitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
[D,M,S] = DMS(lon*d2r);
if D == 0 && lon < 0
fprintf('\nLongitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLongitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
fprintf('\nHeight
= %13.3f',h);
fprintf('\n\n');
127
RMIT University
Geospatial Science
d2r = 180/pi;
a = 6378137;
flat = 298.257222101;
lat = -50/d2r;
lon = -150/d2r;
h = 10000;
[X,Y,Z] = Geo2Cart(a,flat,lat,lon,h);
[X,Y,Z]'
ans =
-3563081.36230554
-2057145.98367164
-4870449.48202417
>> [lat,lon,h] = Cart2Geo_Bowring(a,flat,X,Y,Z);
128
RMIT University
Geospatial Science
129
RMIT University
Geospatial Science
130
RMIT University
Geospatial Science
if p+abs(Z)<pQ+abs(ZQ)
h = -h;
end;
% Print results to screen
fprintf('\n\nCartesian to Geographic - Lin & Wang');
fprintf('\n====================================');
fprintf('\nEllipsoid:');
fprintf('\nsemi-major axis a = %13.3f',a);
fprintf('\nflattening
f = 1/%13.9f',flat);
fprintf('\nCartesian coordinates:');
fprintf('\nX = %13.3f',X);
fprintf('\nY = %13.3f',Y);
fprintf('\nZ = %13.3f',Z);
fprintf('\nGeodetic coordinates:');
[D,M,S] = DMS(lat*d2r);
if D == 0 && lat < 0
fprintf('\nLatitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLatitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
[D,M,S] = DMS(lon*d2r);
if D == 0 && lon < 0
fprintf('\nLongitude = -0 %2d %9.6f (D M S)',M,S);
else
fprintf('\nLongitude = %4d %2d %9.6f (D M S)',D,M,S);
end;
fprintf('\nHeight
= %13.3f',h);
fprintf('\nIterations
= %3d',count);
fprintf('\n\n');
d2r = 180/pi;
a = 6378137;
flat = 298.257222101;
lat = -50/d2r;
lon = -150/d2r;
h = 10000;
[X,Y,Z] = Geo2Cart(a,flat,lat,lon,h);
[X,Y,Z]'
ans =
-3563081.36230554
-2057145.98367164
-4870449.48202417
131
RMIT University
Geospatial Science
132
RMIT University
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
p2
q
t
u
X,Y,Z
X2,Y2,Z2
zeta
-
Geospatial Science
p-squared
numeric term in cubic equation in u
single real-root of cubic equation in t
single real-root of cubic equation in u
Cartesian coordinates
powers of X,Y,Z coords
solution of quartic in terms of t
Remarks:
This function uses Pauls' direct method, see Refences [1] & [2].
References:
[1] Paul, M.K., 1973, 'A note on computation of geodetic coordinates grom
geocentric (cartesian) coordinates', Bulletin Geodesique, No. 108,
pp. 134-139.
[2] Deakin, R.E. and Hunter, M.N., 2008, GEOMETRIC GEODESY - PART A,
School of Mathematical and Geospatial Sciences, RMIT University,
Melbourne, AUSTRALIA, March 2008.
% compute zeta
root1 = sqrt(t);
if Z<0
root1 = -root1;
end;
root2 = sqrt(Z2/4 - beta/2 - t + alpha*Z/4/root1);
if Z<0
root2 = -root2;
end;
zeta = root1 + root2;
133
RMIT University
Geospatial Science
134
RMIT University
Geospatial Science
d2r = 180/pi;
a = 6378137;
flat = 298.257222101;
lat = -50/d2r;
lon = -150/d2r;
h = 10000;
[X,Y,Z] = Geo2Cart(a,flat,lat,lon,h);
[X,Y,Z]'
ans =
-3563081.36230554
-2057145.98367164
-4870449.48202417
135
RMIT University
Geospatial Science
MATLAB
functions
Geo2Cart.m
Cart2Geo_Substitution.m
Cart2Geo_Newton.m
136
RMIT University
Geospatial Science
%
% Remarks:
% Formulae are given in [1] (section 1.3.9, page 85) and in
% [2] (Chapter 2, p. 2-10) in a slightly different form.
%
% References:
% [1] Deakin, R.E. and Hunter, M.N., 2008, GEOMETRIC GEODESY PART A,
%
School of Mathematical and Geospatial Sciences, RMIT University,
%
Melbourne, AUSTRALIA, March 2008.
% [2] THE GEOCENTRIC DATUM OF AUSTRALIA TECHNICAL MANUAL, Version 2.2,
%
Intergovernmental Committee on Surveying and Mapping (ICSM),
%
February 2002 (www.anzlic.org.au/icsm/gdatum)
%-------------------------------------------------------------------------% compute flattening f, polar radius of curvature c and 2nd-eccentricity
% squared ep2
f
= 1/flat;
c
= a/(1-f);
e2 = f*(2-f);
ep2 = e2/(1-e2);
% calculate the square of the sine of the latitude
c1 = cos(lat);
c2 = c1*c1;
% compute latitude function V
V2 = 1+(ep2*c2);
V = sqrt(V2);
V3 = V2*V;
% compute radii of curvature
rm = c/V3;
rp = c/V;
137
RMIT University
Geospatial Science
4 REFERENCES
Adler, K., (2002), The Measure Of All Things, Little, Brown, London.
BG, (1988), The Geodesist's Handbook 1988, Bulletin Godsique, Vol. 62, No. 3.
Baeschlin, C.F., (1948), Lehrbuch Der Geodsie, Orell Fssli Verlag, Zurich.
Borkowski, K. M., (1989),
(1958),
Handbuch
der
Vermessungskunde,
Band
IV
138
RMIT University
Geospatial Science
York.
NIMA, (2000), Department of Defense World Geodetic System 1984, Technical Report
TR8350.2, 3rd edition, Amendment 1, 3 January 2000, National Imagery and
Mapping Agency.
Paul, M. K., (1973),
139
RMIT University
Geospatial Science
Struik, D.J., (1933), 'Outline of a history of differential geometry', Isis, Vol. 19, No.1, pp.
92-120, April 1933. [Isis is an official publication of the History of Science Society and has
been in print since 1912.
Division: http://www.journals.uchicargo.edu/]
140