Map Projections II
Map Projections II
Map Projections II
1 Introduction 1
1.1 What are map projections good for? . . . . . . . . . . . . . . 1
1.2 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Fundamentals 2
2.1 The Gaussian curvilinear parameters u and v . . . . . . . . . 2
2.2 The line element . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.3 Intersection angles of surface curves . . . . . . . . . . . . . . . 8
2.4 The infinitesimal area . . . . . . . . . . . . . . . . . . . . . . . 9
3 Theory of Distortions 11
3.1 The scale distortion . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 The distortion of areas . . . . . . . . . . . . . . . . . . . . . . 17
3.3 The distortion of the azimuth . . . . . . . . . . . . . . . . . . 18
3.4 The infinitesimal angular distortion . . . . . . . . . . . . . . . 19
3.5 The maximal distortion of the azimuth . . . . . . . . . . . . . 21
3.6 Ellipse of distortion . . . . . . . . . . . . . . . . . . . . . . . . 21
3.7 Projections with specific properties . . . . . . . . . . . . . . . 24
3.7.1 Equidistant projections . . . . . . . . . . . . . . . . . . 24
3.7.2 Equal area projections . . . . . . . . . . . . . . . . . . 27
3.7.3 Conformal projections . . . . . . . . . . . . . . . . . . 28
i
ii Script ”Map Projections”
6 Generalized projections 54
6.1 Pseudoconic projections . . . . . . . . . . . . . . . . . . . . . 54
6.2 Pseudocylindrical projections . . . . . . . . . . . . . . . . . . 59
6.2.1 Equal area with equidistant parallels . . . . . . . . . . 61
6.2.2 Equal area with elliptical meridians . . . . . . . . . . . 62
6.2.3 Equal area with elliptical meridians and pole lines . . . 65
6.2.4 With equidistant equator and equidistant central meridian 65
KIT, Geodetic Institute, Dr.-Ing. N. Rösch iii
A Exercises 93
C Formulary 107
C.1 Scale distortion . . . . . . . . . . . . . . . . . . . . . . . . . . 107
C.2 Area distortion . . . . . . . . . . . . . . . . . . . . . . . . . . 109
C.3 Distortion of angles . . . . . . . . . . . . . . . . . . . . . . . . 110
C.3.1 Distortion of the angle element . . . . . . . . . . . . . 111
C.3.2 Distortion of the azimuth . . . . . . . . . . . . . . . . 112
α, δ . . . . . . . . . . azimuthal coordinates
C . . . . . . . . . . . . integration constant
v
vi Script ”Map Projections”
Introduction
1.2 Classification
There are different possibilities to classify the map projections. The criteria
may be: extrinsic or intrinsic. The first aspect encompasses the nature, the
coincidence and the position of the projection surface. The latter refers to
the properties and the generation.
class varieties
extrinsic nature plane conical cylindrical
coincidence tangent secant polysuperficial
position normal transverse oblique
intrinsic properties equidistant eqaual area conformal
generation geometric semi-geometric conventional
Table 1.1: The classification of map projections
1
Chapter 2
Fundamentals
2
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 3
surface fulfilling the condition above are called regular . A surface consisting
of regular points is called analytical .
Two analytical surfaces can be mapped to each other, if there exists a rela-
tionship between the the parameters u , v and u, v.
∂u ∂v ∂u ∂v
− = 0 . (2.5)
∂u ∂v ∂v ∂u
Example: Exercise 1
Before starting with the calculations let’s have a look at formula A.1. This is the
parametric representation of a sphere, where as the parameter u can be treated
as the geographic latitude and the parameter v as the geographic longitude. Thus
the datum surface is the sphere. Hence there are other parametric representation
of the sphere possible.
This is obvious if we analyze the adjacent example. The in-
N terpretation of u̇ as an angle between the z-axis and position
P
vector to point P leads us to another parametric representa-
u. tion of the sphere:
Equator ⎧
⎨ x = R sin u̇ cos v
x(u̇, v) = y = R sin u̇ sin v (2.7)
⎩
z = R cos u̇
This kind of parametrization is equivalent to the latter. The
substitution of u = 2 − u̇ and the relationship (B.8) again delivers (A.1).
π
item a):
We notice in the problem definition that the two position vectors describe two
different surfaces. We already analyzed one surface and know that it is a sphere,
which is given as x(u, v). The second position vector is x (u , v ). Hence referring
to (2.3) a projection assumes a position vector like x (u, v). Exercise 1 already
encompasses the according relationship by (A.3). Applying this we get
⎧
⎨ x = R cos v
x (u, v) = y = R sin v .
⎩
z = R tan u
Further we have to test if equation (A.3) is fulfilled. This leads us to the below
Jacobian determinant:
∂u
∂u
∂u = cos2 u ∂v = 0 = R = 0 .
R
∂v = 0 ∂v cos2 u
∂u ∂v = 1
The determinant is different from zero. Hence if u = 90◦ than the fraction is not
defined. If we have a look at the third component of x(u, v) we recognize, that the
north pole can not be mapped. Still we get an unique solution for all other points.
item b):
The question about the geometric interpretation is equivalent to the question
about the image surface. As we already know the position vector of the image
surface we only need to analyze it. Therefore we simply took the projection in the
different planes.
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 5
and ⎧
⎨ x = 5019.629 km
◦ ◦
xM (38 , 56 ) = y = 3921.764 km
⎩
z = 9443.913 km .
item c):
6 Script ”Map Projections”
+y
M’
+x K’
Figure 2.4: The mapped curve in ground- and lateral view
by
s = (R · Δλ)2 + Δz 2 = 3949.952 km
as the scetch depicts.
Thus the shortest distance between datum and image surface is distorted by m =
s /s = 1.834 4
4
Note: The shortest distance on the image plane is not identical with the mapped
image of the great circle. Insofar the comparison between the two curves on the different
surfaces is not exact. We discuss this later on.
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 7
item e):
The directional derivatives are:
⎧
⎨ xϕ = −R sin ϕ cos λ
xϕ = yϕ = −R sin ϕ sin λ (2.10)
⎩
zϕ = R cos ϕ
and ⎧
⎨ xλ = −R cos ϕ sin λ
xλ = y = R cos ϕ cos λ (2.11)
⎩ λ
zλ = 0 .
Being regular (2.2) requires that the normal vector N of a point is not equal to
zero. N is ⎧
⎨ x = −R2 cos2 ϕ cos λ
N = xϕ × xλ = y = −R2 cos2 ϕ sin λ .
⎩ 2
z = −R sin ϕ cos ϕ
factoring out R2 we get
⎛ ⎞
cos2 ϕ cos λ
N = −R2 ⎝ cos2 ϕ sin λ ⎠ .
sin ϕ cos ϕ
Because
F = xϕ xλ = |xϕ | |xλ | cos τ = 0
=0 =0
the directional derivatives (which are vectors as well) are intersecting with τ = 90◦ ,
i. e. the parameter curves are orthogonal.
8 Script ”Map Projections”
T β ds δ |d x| = ds
δs |δ x| = δs .
d xδ x ( xu du + xv dv)( xu δu + xv δv)
cos γ = =
|d x||δ x| ds δs
xu xu duδu + xu xv duδv + xv xu dvδu + xv xv dvδv
=
dsδs
Eduδu + F (duδv + dvδu) + Gdvδv
= . (2.15)
dsδs
In the same way the azimuth β of an surface curve can be found. As depicted
in figure (2.5) the relationship between the azimuth β and the two directional
derivatives along the parametric lines (tangent vectors) in point P is given
by
d x xu
cos β =
|dx| | xu |
and can be simplified by replacing |dx| with
( xu du + xv dv) xu E du + F dv
cos β = √ = √ . (2.16)
ds E ds E
The intersection angle τ between the two parametric lines can be calculated
based on the inner product too. One gets (see figure (2.5))
xu xv F
cos τ = =√ √ . (2.17)
|xu | |xv | E G
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 9
In addition to the above derived equations also the sine and the cosine of the
relating intersection angles are shown:
√
EG − F 2 (duδv − dvδu)
sin γ = (arbitrary surface curves)(2.18)
√ ds δs
EG − F 2 dv
sin β = √ (azimuth (bearing)) (2.19)
Eds
√
EG − F 2
sin τ = √ √ (parameter curves) (2.20)
E G
dO = | xu du × xv dv| .
( xu × xv )2 = x2u x2v − ( xu xv )2 = EG − F 2
The integration for ϕ ∈ [0, π/2] and λ ∈ [0, π], should yield the area of the
hemisphere.
Since the order of the integration could be swapped, it is of no importance,
which of the parameters is the first. Taking the latitude we get
π/2
2
cos ϕ = R2 [sin ϕ]0 = R2
π/2
R
0
and next 2π
2
O=R dλ = R2 [λ]2π
0 = 2πR
2
,
0
which is in fact the surface of the hemisphere.
Chapter 3
Theory of Distortions
11
12 Script ”Map Projections”
and
x = 0
xλ = , (3.3)
y = R cos ϕ
based on this result the Gaussian quantities are
and the scale distortion along the meridian in point P(30◦ north, 45◦ east) is
Whereas the circle of latitude is projected in its original length (mλ = 1).
Next the scale distortion is examined, whereas mainly the relation with the
azimuth is of interest. The scale distortion can be written as
ds2 2 2
du du dv dv
m2 = = E + 2F + G .
ds2 ds ds ds ds
If we further regard the formulas found in section (2.3), in addition to (2.19)
and (2.16) we get
√ √
EG − F 2 dv dv E
sin β = √ ⇒ =√ sin β
E ds ds EG − F 2
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 13
as well as
Edu + F dv
cos β = √ combined with the relationship above
ds E
du cos β F sin β
= √ −√ √ ,
ds E E EG − F 2
which leads us to
E EF − F E
m2 = cos2 β + 2 √ sin β cos β
E E EG − F 2
EG − 2F F + GE E
+ − sin2 β . (3.4)
EG − F 2 E
Based on the above formula the following simplifications are introduced
E
A =
E
EF − F E
B = √
E EG − F 2
EG − 2F F + GE E
C = − ,
EG − F 2 E
leading to
m2 = A cos2 β + 2B sin β cos β + C sin2 β . (3.5)
This equation delivers, that the scale distortion depends not only on the
Gaussian quantities and thus on the parameters u and v but also on the
azimuth denoted as β.
m = f (u, v, β) . (3.6)
This leads us to the question if there exists a maximal and a minimal distor-
tion respectively. Therefore we set the differential quotient to zero.
dm
=0 . (3.7)
dβ
Thus we derivate the implicit function of equation (3.5) which yields
2mdm = −2A cos β sin βdβ + 2B cos2 βdβ − 2B sin2 βdβ + 2C sin β cos βdβ
and in combination with (B.4) as well as (B.5)) we get
dm
2m = −A sin 2β + 2B cos 2β + C sin 2β .
dβ
14 Script ”Map Projections”
The result of formula (3.9) and formula (3.10) comprises, that the principal
directions on the datum surface as well as on the projection surface are
perpendicular. Hence the principal directions are normally the only mutual
perpendicular directions.
In our next step we assume F ≡ F ≡= 0 which means, that the parameter
system of the datum surface as well as that of the image surface are per-
pendicular. In this case, the parameter system is also the system with the
principal directions.
The above mentioned aspects lead us to the question, which are the absolute
values of these distortions. Therefore we modify the two formulas (3.5) and
(3.8) which delivers
(A cos β + B sin β) cos β + (B cos β + C sin β) sin β = m2 as well as
−(A cos β + B sin β) sin β + (B cos β + C sin β) cos β = 0 .
If we multiply the first with cos β and the second with sin β and afterwards
subtract the second from the first we get
(A cos β + B sin β) cos2 β + (A cos β + B sin β) sin2 β = m2 cos β
or
A cos β + B sin β = m2 cos β . (3.11)
In our next step we multiply equation (3.5) and (3.8) with sin β and cos β
respectively. The result is
(A cos β + B sin β) sin β cos β + (B cos β + C sin β) sin2 β = m2 sin β
−(A cos β + B sin β) sin β cos β + (B cos β + C sin β) cos2 β = 0 .
Then we add the equations and obtain
B cos β + C sin β = m2 sin β . (3.12)
We transpose the formulas (3.11) and (3.12) and get
(m2 − A) cos β = B sin β and
(m2 − C) sin β = B cos β .
The multiplication of the equations delivers a quadratic expression of m.
This term does not depend on β. Is is
(m2 − A)(m2 − C) = B 2
16 Script ”Map Projections”
or
m4 − (A + C)m2 + AC − B 2 = 0 ,
from which we get the two solutions
A + C A + C 2
m21 = + − AC + B 2 und (3.13)
2 2
2 A+C A + C 2
m2 = − − AC + B 2 . (3.14)
2 2
The first solution is the square of the first principal direction and is abbre-
viated by a2 , whereas the square of the second principal direction is denoted
by b2 . This means that
From this we derive further quantities. The first is called medial distortion
U 2 and is defined as follows
2U 2 = a2 + b2 = A + C . (3.15)
V 2 = a2 b2 = AC − B 2
E EG − 2F F + GE E EF − F E 2
= − − √ ,
E EG − F 2 E E EG − F 2
for the area distortion respectively, which can be simplified to
E G − F 2
V2 = . (3.17)
EG − F 2
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 17
Equation
sin ω = sin(β − β) and cos ω = cos(β − β)
in combination with the addition theorem for the sine and the cosine (see
(B.1) as well as (B.2)) leads us to
The substitution of cos β and sin β with the term (3.25) and (3.26) delivers
an equation for the distortion of the azimuth the two formulas
b a b−a
sin ω = sin β cos β − sin β cos β = sin β cos β and
m m m
a b 1
cos ω = cos2 β + sin2 β = (a cos2 β + b sin2 β) .
m m m
For the tangent of the azimuth the same deductions can be performed. We
get tan(β − β) = tan ω in combination with (B.3)
tan β − tan β
tan ω =
1 + tan β tan β
dβ
=?
dβ
a2 cos2 βm + b2 sin2 βm = a b ,
which yields by substitution of the sine and the cosine and factoring out
cos2 βm (a2 − b2 ) = a b − b2
or
a b − b2 b a
cos2 βm = = ⇒ sin2 βm =
a2 − b2 a+b a+b
and thus
a
tan2 βm = . (3.30)
b
Based on
b
tan β = tan β
a
one gets
b
tan2 βm
= ⇒
tan βm tan βm =1
a
e. g.
π
βm + βm = .
2
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 21
(b − a) tan βm
tan ωm = .
a + b tan2 βm
By applying the specific equation for the direction with the maximum dis-
tortion for the infinitesimal angle (see 3.30) we get
(b − a) ab b−a
tan ωm = = √ . (3.31)
2a 2 ab
It is possible to deduce similar equations for the sine and the cosine of ωm
These are √
b−a 2 ab
sin ωm = and cos ωm = .
b+a b+a
The maximal distorted angle Ωm is due to symmetry
Ωm = 2 |ωm| ,
because the maximal distorted azimuth can be found in the first and in the
fourth quadrant.
Starting point for the next examinations is the question how the infinitesimal
circle with the radius ds is mapped on the image surface. Further F ≡ F ≡ 0
is assumed
The infinitesimal line element can be described on both the datum and the
image surface as
After the division by ds, which delivers the radius of the unit circle we get
for the image plane
dξ
ξ = = m cos β = a cos β as well as (3.38)
ds
dη
η = = m sin β = b sin β (3.39)
ds
and thus
ξ 2 + η 2 = m2
in combination with
ξ 2 η 2
+ 2 =1 .
a2 b
This is the equation of the so called distortion ellipse or Tissot’s Indika-
trix. The above formula delivers further, that the infinitesimal unit circle is
mapped in its affine image e. g. the ellipse.
At this point we can draw some conclusions. The three most important are
stated below.
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 23
1. The scale distortion m is the length of the radius vector of the distortion
ellipse. For two arbitrary direction which are perpendicular β and
β = β + π/2 the scale distortion m is
ξ = a cos β ; ξ = a cos(β + π/2) = −a sin β (3.40)
η = b sin β ; η = b sin(β + π/2) = b cos β (3.41)
(3.42)
squared and added delivers
ξ 2 + η 2 = m2 = a2 cos2 β + b2 sin2 β (3.43)
2 2 2 2 2 2 2
ξ +η = m = a sin β + b cos β (3.44)
and thus
m2 + m2 = a2 + b2 .
2 F − F E − E
m +2 m+ =0
G −G G −G
and the solution of the mixed quadratic equation delivers
F − F F − F 2 E − E
m1,2 = − ± − . (3.45)
G −G G − G G −G
In our next step we will have a look on the root, which can be rearranged by
finding the common denominator
F 2 − 2F F + F 2 (E − E)(G − G)
− =
(G − G)2 (G − G)2
d O = d O ,
is fulfilled, e. g. the infinitesimal area of the datum and the image surface
are equal. Further we can deduce
d O
= 1 = ab = V . (3.46)
dO
Combined with (3.19) we get
V 2 = 1 ⇒ E G − F 2 = E G − F 2
and
√ √
O= EG − F 2 du dv and O = E G − F 2 du dv ,
u v u v
which depicts the property equal area is not only fulfilled in an infinitesimal
environment but also in a finite.
Because of a b = 1 one of the principal distortions is always less than 1
whereas the other is greater than 1. As a consequence dealing with equal
area projections there always exists a direction with a scale distortion of 1.
This direction is given by (datum surface)
a2 − 1 a2 − 1
tan2 βl = = = a2
1 − b2 1 − a12
or
tan βl = ±a
and
b2 a2 − 1 b2 1 − b2
tan2 βl = = 1 = b2 ⇒ tan βl = ±b
a 1−b
2 2 b2
b2 1−b2
these directions βl and βl indicate also the maximal distortion of the azimuth.
Thus it is
tan βm = tan βl = ±a and tan βm = tan βl = ±b .
Normally the position on the sphere of an arbitrary point P are given in geo-
29
30 Script ”Map Projections”
graphic coordinates ϕ and λ, whereas the projection formulas use the Gaus-
sian parameters u and v. This makes sense because as already mentioned
the latter formulas depend not on the attitude of the central point however
they refer to it. This means that we have to find a relationship between the
Gaussian parameters and the geographic coordinates. This relationship is
not linear as figure 4.1 depicts.
As figure 4.1 depicts we need a polar triangle to find the above mentioned re-
lationship. To explain this in detail, we take a azimuthal projection as an ex-
ample. It is further assumed, that the central point is located at H(ϕH , λH ).
Furthermore we take an arbitrary point P (ϕP , λP ). Our next step leads us
to the spherical law of cosines
cos(90◦ −u) = cos(90◦ −ϕH ) cos(90◦ −ϕP )+sin(90◦ −ϕH ) sin(90◦ −ϕP ) cos(Δλ)
sin(v) sin(Δλ)
= bzw.
cos(ϕP ) cos(u)
sin(Δλ)
sin(v) = cos(ϕP ) .
cos(u)
As already outlined the central point is like a pole for the newly introduce
parameter system. Thus the definition of this point is crucial. Naturally
the central point depends on the projection surface which can be a cone a
cylinder or a plane. The latter is the easiest because the central point is the
coincidence of the plane and the sphere (if the plane is tangent). In the case
of the cylinder surface the central point is defined as the intersection of the
axis of the cylinder and the surface of the sphere. In almost the same way
we get the central point considering a conical projection. In this context it
is the intersection of the line between the apex of the cone and the center of
the sphere.
distance between two points in such a map. Thus the course angle e. g. for
navigation on a ship could be easily derived by a rule.
cBased on figure 4.2 we can conclude
dλ
tan β = cos ϕ
dϕ
or
R db
tan β
ds
dλ = dϕ
β cos ϕ
and by integration between P1 (ϕ1 , λ1 ) and
R cos b dl P2 (ϕ2 , λ2 ) and solving for tan β
λ2 − λ1
Figure 4.2: The differential tan β =
triangle ln tan π
4
+ ϕ22 − ln tan π4 + ϕ1
2
(4.2)
we get the course angle between the two points.
The distance along the loxodrome can be deduced (see fig. 4.2) from
Rdϕ Rdϕ
cos β = and thus ds =
ds cos β
or ϕ2
R R
s= dϕ = (ϕ2 − ϕ1 ) . (4.3)
ϕ1 cos β cos β
Exercise 11
Based on a map which is the result of a conformal projection of the ellipsoid of
revolution to the plane a captain derives the course angle of his passage from
San Juan (Puerto Rico) (ϕJ = 18.28◦ , λJ = 66.07◦) to Lisbon (ϕL = 38.43◦ ,
λL = 9.08◦ ). The course angle shall be the same for the whole journey.
a) Calculate the difference of the course angle, if the map is not derived
from a ellipsoid of revolution but from a sphere (radius = 6 370 km)
b) How is the difference of the length of the loxodrome between the ellip-
soid of revolution and the sphere? What is the length of the loxodrome
if we calculate with the course angle of the sphere?
c) And finally, what is the difference in the distance between the loxo-
drome and the shortest distance on the sphere?
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 33
General information:
or
π
ΦL 1 − e sin ΦL e2 π Φ 1 − e sin Φ 2e
J J
ΛL −ΛJ = tan βE ln tan + −ln tan +
4 2 1 + e sin ΦL 4 2 1 + e sin ΦJ
solved to tan βE delivers the course angle on the ellipsoid of revolution βE
ΛL − ΛJ
tan βE = e e2 .
ΦL 1−e sin ΦL 2 ΦJ 1−e sin ΦJ
ln tan π
4
+ 2 1+e sin ΦL
− ln tan π
4
+ 2 1+e sin ΦJ
(4.4)
If we apply the appropriate figures we get
βE = 68◦0490 .
34 Script ”Map Projections”
Δβ = βE − βK = 0◦ 1019 .
Item b)
The length of the loxodrome on the surface of the sphere can be solved
immediately by formula (4.3). The equation yields
sK = 5 992.894 km .
The differential triangle on the ellipsoid of revolution delivers the course angle
or azimuth respectively based on the Gaussian quantities
M(Φ)dΦ
cos β = .
ds
Paralleled to the sphere, we get an ordinary differential equation which leads
after the separation of the variables to
M(Φ)dΦ
ds =
cos β
which can be integrated in the next step. One gets
ΦL
1 a(1 − e2 ) ΦL 1
s= M(Φ) dΦ = 3 dΦ . (4.5)
cos β ΦJ cos β 2
ΦJ 1 − e sin Φ
2
This equation comprises an elliptic integral of the first kind and cannot be
expressed in terms of elementary functions. Thus we need another approach.
The first is the solution by numerical integration. One of this approaches is
Romberg’s method. Doing so we get the results depicted in table 4.1.2 .
Below a rough description of the algorithm:
(α − β) ∗ (f (α) + f (β))/2
2
Note: These results are only a first approximation!
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 35
• Stop the iteration if the difference of the last two linear combinations
is less than a threshold ε otherwise continue with the third step
The result for the length of the loxodrome sE from point J to point L com-
puted up to the fourth decimal place is
sE = 5 973.6382 km .
There are still other approaches to solve 4.5. One of these is the series
expansion. The mathematical textbooks dealing with integration and series
3
expansion delivers for (1 − x)− 2 the expression
3 3 15 2
(1 − x)− 2 = 1 + x+ x . . . (|x| < 1) .
2 8
In the above expansion the series was aborted after the third term. We will
see that three terms are enough to get a good approximation for the solution
of 4.5.
The substitution of
x = e2 sin2 Φ
yields for 4.5
a(1 − e2 ) ΦL 3 15
s= (1 + e2 sin2 Φ + e4 sin4 Φ + . . . ) dΦ or
cos β ΦJ 2 8
ΦL
a(1 − e2 ) 3 2 ΦL 2 15 4 ΦL 4
s= dΦ+ e sin Φ dΦ+ e sin Φ dΦ +. . . .
cos β ΦJ 2 ΦJ 8 ΦJ
Considering (B.19) and (B.21) we thus are able compute the length sE of the
loxodrome on the surface of the ellipsoid of revolution
!
a(1 − e2 ) 3 2 1 1
sE = · (ΦL − ΦJ ) + e (ΦL − ΦJ ) − (sin 2ΦL − sin 2ΦJ )
cos βE 2 2 4
15 4 3 1
+ e (ΦL − ΦJ ) − (sin 2ΦL − sin 2ΦJ ) (4.6)
8 8 4
!
1
+ (sin 4ΦL − sin 4ΦJ ) + . . . .
32
sE = 5 973.6382 km .
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 37
Compared with the reference the difference is less than some centimeters.
Furthermore the result reveals that the length of the loxodrome on the sphere
is almost the same as the length on the ellipsoid.
If we compute the length of the loxodrome applying the course angle of the
sphere, we get
sE = 5 947.396 km .
38 Script ”Map Projections”
90°
−
90°
Applying the addition theorem for the sine and the cosine yields
cos s = cos(90◦ −ϕJ ) cos(90◦ −ϕL )+sin(90◦ −ϕJ ) sin(90◦ −ϕL ) cos(λL −λJ ) .
Thus the length of the orthodrome so on the surface of the sphere with the
radius R = 6371 km equals
so = R arccos s = 5 905.364 km .
In this chapter only projections with orthogonal parameter systems are con-
sidered. That is to say that both systems – the system of the datum and
that of the image – have to be orthogonal. As a consequence the Gaussian
quantities F and F need to be zero. We first start with the analysis of the
cone as projection surface.
40
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 41
R2
r dr = sin δ dδ
n
which after integration leads to
R2 2R2
r dr = sin δ dδ or r 2 = − cos δ + C . (5.3)
n n
The above formula reveals the radius as a function of δ or more rigorous
r = r(δ). Thus we found a relationship between the parameter system of
42 Script ”Map Projections”
the sphere and the cone. But there are still two more variables or degrees of
freedom to define the properties of the projection. Below some examples.
Example 1:
Imposing one (coinciding) standard parallel yields n = cos δg and further
cos δg r(δg )
b(δg ) = 1 = .
R sin δg
cos2 δg (−2R2 + C)
=1 ,
R2 sin2 δg
resolved to C yields
2R2 1 + n2
r2 = − cos δ + R2
n n2
or
R√
r= 1 + n2 − 2n cos δ (5.4)
n
respectively. This formula shows , that with respect to r(δ = 0) = R 1−n
n
= 0
the principal point is not mapped as a single point.
By inserting (5.4) in (5.2) the principal directions which are the main axis of
the distortion ellipse are
√
nr 1 + n2 − 2n cos δ sin δ
b= = and (because of a = 1/b) a = √ .
R sin δ sin δ 1 + n2 − 2n cos δ
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 43
Example 2:
If we want to map the central point as a single point we impose (based on
formula (5.3)) r(δ = 0) = 0. Thus one gets
2R2
C=
n
for the integration constant. Introducing C in (5.3) delivers
2R2
r2 = (1 − cos δ) . (5.5)
n
The relationship 1 − cos δ = 2 sin2 ( 2δ ) (see (B.9)) yields together with (5.5)
2R δ
r = √ sin .
n 2
And the related principal distortions are (with respect to (B.4))
√
cos 2δ n
a= √ and b =
δ .
n cos 2
Hence the variable n can be chosen in such a way that another parallel circle
δ = δl is coinciding. Therefore we need
√ δ
l 2 δl
n = cos or n = cos .
2 2
which leads to
R sin δl1 R sin δl2
r(δl1 ) = as well as r(δl2 ) = .
n n
Inserting 5.4 into the above equation and squaring the result yields
2 R2 R2 sin2 δl1
− cos δl1 + C = and (5.6)
n n2
2 R2 R2 sin2 δl2
− cos δl2 + C = . (5.7)
n n2
Subtraction (5.7) from (5.6) delivers
and with sin2 δ = 1 − cos2 δ in combination with the third binomial theorem
the equation can be modified to
1
n = (cos δl2 + cos δl1 ) . (5.8)
2
R2 sin2 δl1 2 R2
C= + cos δl1
n2 n
and further obtain based on a common denominator
" #
2
sin δ 2n cos δ R2
C = R2 (sin2 δl1 + 2n cos δl1 ) .
l1 l1
+ = (5.9)
n2 n2 n2
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 45
R2
r2 = (sin2 δl1 + 2n cos δl1 − 2n cos δ) or explicitly
n2
R
r = sin2 δl1 + 2n(cos δl1 − cos δ) (5.10)
n
whereas the varialbe n refers to (5.8).
m = f (δ, α) .
m=a=b
has to be fulfilled. Taking the principal distortion of equation (5.2) one gets
after some deductions
1 dδ
dr = n .
r sin δ
In this equation the variables are already separated and thus could be inte-
grated. The integration delivers
δ
ln r = n ln tan + ln C ,
2
or
δ
ln r = ln C tann
2
which leads us to
δ
r = C tann . (5.11)
2
This formula encompasses the relationship of the radius and the according
parameter δ and thus r = r(δ) is solved. From this formula we can conclude
46 Script ”Map Projections”
R sin δg
C= δg
.
n tann 2
Example 2:
The alternative is to impose two coinciding standard parallels. If we do so,
we get for C the constraint (in combination with b(δ1 ) = b(δ2 ) = 1):
R sin δ1 R sin δ2
C= n δ1
= . (5.12)
n tan 2 n tann δ22
and
δ1 δ2
n ln tan + ln sin δ2 = n ln tan + ln sin δ1
2 2
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 47
which leads to
ln sin δ1 − ln sin δ2 ln sin δ1
sin δ2
n= = . (5.13)
ln tan δ21 − ln tan δ22
δ
tan 21
ln δ2
tan 2
The two variables C (see (5.12)) and n (see (5.13)) deliver in combination
with (5.11) the radius for a conformal conic projection with two coinciding
standard parallels
R sin δ1 δ
r= n δ1
tann .
n tan 2 2
This is a rather good example to prove, that this projection is equal area. Of
course its already shown by the principal distortions a and b, hence it could
also be explained by pure geometry. Looking at figure 5.1 (a) in combination
with equation (5.16) we realize that the result for x is within the interval
of −R/ sin δl ≤ x ≤ +R/ sin δl . On the other hand y (see fig. 5.1 (b)) and
equation (5.16)) is for all α ∈ [−π, π] → −Rπ sin δl ≤ y ≤ Rπ sin δl . The
area of this rectangle is 2 R / sin δl · 2 R π sin δl = 4 R2 π and thus the surface
of a sphere with radius R. So the equal area criterion is fulfilled for the
infinitesimal surrounding of an arbitrary point as well as for the finite area.
+x
X’
δ
+y
which leads to
dδ
dx = −c · R · .
sin δ
The integration delivers
δ
x = −cR ln tan +C and simplified
2
δ
x = cR ln cot +C .
2
As already explained the integration constant C fixes the y -axis. If we chose
C = 0 the parallel with b(δl ) = c/ sin δl = 1 coincides with the y -axis. Hence
there are also other definitions possible.
Finally the position vector x of a conformal cylinder projection is
x = cR ln cot 2δ
x = .
y = cRα
If c = 1 (i. e. one coinciding standard parallel) the position vector is
x = R ln cot 2δ
x = .
y = Rα
are applied for all for all azimuthal coordinates. Converted in Cartesian
coorinates
x = r(δ) cos α
x = (5.17)
y = r(δ) sin α
are derived. The derivative with respect to the parameters δ and α and the
dot product deliver the Gaussian quantities
dr 2
E = , F = 0 and G = r 2
dδ
form which the principal distortions
1 dr r(δ)
a= as well as b = . (5.18)
R dδ R sin δ
are obtained.
As shown by 5.17 the knowledge of only one parameter is sufficient. Thus
knowing the radius in 5.17 means knowing the equation of the projection.
This is deepened in the next sections.
r 2 (δ) = 0 , i. e.
C = 2R2 .
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 51
1 dr
a = and with
R dδ
dr R
Figure 5.2: Gnomonic pro- = whereas
dδ cos2 δ
jection 1 1
a= 2
as well as b= is derived.
cos δ cos δ
The special property of the projection is, that
the great circle on the sphere is mapped as a
straight line and thus it the shortest distance between two points in the plane
as well.
H r P’
R P
δ
As figure 5.3 depicts the radius r is given by r = R sin δ. The derivative with
respect to δ is dr/dδ = R cos δ and thus the first principal distortion
a = cos δ
V = a b = cos δ
Generalized projections
r = r(δ) and
ε = ε(α, δ) .
The above formulas show that the parallels are concentric. The verticals
however are mapped curvilinear. According to 5.1 the Cartesian coordinates
are obtained with
x = rg − r(δ) cos ε(α, δ) (mit rg = R tan δg )
x(δ, α) = . (6.1)
y = r(δ) sin ε(α, δ) .
54
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 55
and
xα ∂ε
= r(δ) sin ε(α, δ) ∂α
xα = (6.3)
yα ∂ε
= r(δ) cos ε(α, δ) ∂α
to
dr 2 dr ∂ε
E = xδ xδ = cos2 ε(α, δ) − 2 cos ε(α, δ)r(δ) sin ε(α, δ) +
dδ dδ ∂δ
∂ε 2
+r 2 (δ) sin2 ε(α, δ) +
∂δ
dr 2 dr ∂ε
+ sin2 ε(α, δ) + 2 cos ε(α, δ)r(δ) sin ε(α, δ) +
dδ dδ ∂δ
∂ε 2
+r 2 (δ) cos2 ε(α, δ)
∂δ
dr 2 ∂ε 2
= + r 2 (δ) (6.4)
dδ ∂δ
as well as
dr ∂ε ∂ε
F = xδ · xα = − cos ε(α, δ) + r(δ) sin ε(α, δ) r(δ) sin ε(α, δ)
dδ ∂δ ∂α
dr ∂ε ∂ε
+ sin ε(α, δ) + r(δ) cos ε(α, δ) r(δ) cos ε(α, δ)
dδ ∂α ∂α
∂ε ∂ε ∂ε ∂ε
= r 2 (δ) sin2 ε(α, δ) + r 2 (δ) cos2 ε(α, δ)
∂δ ∂α ∂δ ∂α
∂ε ∂ε
= r 2 (δ) (6.5)
∂δ ∂α
together with
∂ε 2 ∂ε 2
G = xα · xα = r 2 (δ) sin2 ε(α, δ) + r 2 (δ) cos2 ε(α, δ)
∂α ∂α
∂ε 2
= r 2 (δ) . (6.6)
∂α
56 Script ”Map Projections”
Thus the scale distortions h and k along the parameter curves are
$
%
√ % dr + r 2 ∂ε 2
E & dδ ∂δ 1 dr
h = √ = 2
= dδ
as well as (6.8)
E R R sin τ
$
% 2
√ % r 2 ∂ε
G & ∂α
∂ε
1 r ∂α
k = √ = = . (6.9)
G R2 sin2 δ R sin δ
sin τ dr ∂ε
r
V =a·b=h·k = dδ2 ∂α (6.10)
sin τ R sin δ
and the principal distortions are
√
a±b= h2 + k 2 ± 2V .
If we want to fulfill the equal area criterion based on the above formulas we
get
dr ∂ε
r
V = dδ2 ∂α = 1 (6.11)
R sin δ
and this expression can be simplified to
∂ε R2 sin δ
= dr
. (6.12)
∂α r dδ
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 57
Separating the variables of this differential equation combined with the inte-
gration yields
R2 sin δ
ε= dr
α+C . (6.13)
r dδ
A quite obvious choice for the integration constant is C = 0. As a conse-
quence for α = 0 also ε = 0 and thus the according vertical is a straight
line.
and
xα = 0
xα = .
yα = ∂y
∂α
From the above equations the Gaussian quantities are derived. These are
dx 2 ∂y 2
E = xδ · xδ = + (6.19)
dδ ∂δ
∂y ∂y
F = xδ · xα = · (6.20)
∂δ ∂α
∂y 2
G = xα · xα = (6.21)
∂α
As already discussed in section 5.2 the differential quotient dx /dδ < 0 is
also negative. Thus we need a negative sign. The intersection angle of the
60 Script ”Map Projections”
Replacing sin τ by 6.22 delivers after some deductions the formulas for the
principal distortions
1 dx ∂y 1 2 ∂y 2
a±b = ∓ + . (6.25)
R dδ ∂α sin δ ∂δ
y = R sin δ · α + C .
x = −R · δ + C . (6.27)
This is the radius of the circle depicted in figure 6.3 and according to the
sketch at the same time the semi minor axis of the ellipsoid denoted further
with b.
Imposing the equal area condition the area of the surface of the unit sphere
must be the same as the area of the ellipse. Thus
√
a · b · π = 4 · π and in combination with b = 2
can be derived. However we have to take into account, that no longer the
unit sphere, but the sphere with the radius R has to be mapped. Considering
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 63
B C
ψ
A D y
6.18 the first component of the position vector x = f (δ) is not yet solved,
because it depends on δ and not on ψ.
From geometry we derive
√
y = 2 2R · sin ψ .
The above expression in combination with the equation 6.29 obviously de-
scribe the shape of the ellipse. Considering α and the fact that α ∈ [−π, π]
we get √
2 2
y (ψ, α) = · R · α · sin ψ .
π
Thus we got the components of the position vector:
√
x = 2 · R · cos ψ
x (ψ, α) =
√
2 2 (6.30)
y = π
· R · α · sin ψ
As already mentioned the above formula delivers x (ψ, α), however this not
what we want. Hence we need a relationship between ψ and δ.
This is emphasized by 6.18. If we take the equal area criterion 6.23 into
account we get
1 dx ∂y
V = h · k sin τ = − · · =1 . (6.31)
R2 · sin δ dδ ∂α
64 Script ”Map Projections”
This equation encompasses the differential relationship between the two pa-
rameters ψ und δ. Separation of the variables leads to
π
sin2 ψdψ = · sin δdδ ,
4
and after integration we get
ψ 1 π
− sin(2ψ) + C = − cos δ . (6.32)
2 2 4
Imposing ψ = 0 if δ = 0 yields for the integration constant
π
C=− .
4
In combination with 6.32 one obtains after some transformations
π ψ 1
(1 − cos δ) = − sin(2ψ)
4 2 4
and together with
δ 1
sin2 = (1 − cos(δ)) (see (B.9))
2 2
finally after the multiplication with 4
δ
2ψ − sin(2ψ) = 2π sin2 (6.33)
2
is yielded. This equation is transcendent and can only be solved in an itera-
tive way. One of the best approaches to so is the Newton-method 1 .
Equation 6.33 delivers a relationship between ψ and δ. Thus the azimuthal
coordinates can be calculated and finally also the position vector on the image
plane. Further 6.18 is fulfilled.
1
Sometimes also called Newton-Raphson method
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 65
x2 + (y + ē)2 = r̄ 2 .
and
r̄ = ē + q .
It is
ē2 + p2 = r 2 = (ē + q)2 = ē2 + 2ēq + q 2 ,
from which we get the relationship
p2 = 2ēq + q 2 or p2 − q 2 = 2ēq .
R(π 2 − 4α2 )
ē = . (6.38)
8α
Thus it is
R(π 2 − 4α2 )
r̄ = ē + q = +R·α
8α
and summarized
R(π 2 + 4α2 )
r̄ = .
8α
With |α| = π/2 we get for the image of the meridian bordering the hemi-
sphere the realtionship
π π π
ē = 0 und r̄ =R· .
2 2 2
The angle φ is
p
tan φ =
ē
and applying the relationship 6.37 as well as 6.38 we get the equation
4 · πα
tan φ = . (6.39)
π2 − 4 · α2
In combination with (B.11) we get for tan φ/2
φ p
p 2·α
r̄
tan = ē = = . (6.40)
2 1+ r
r̄ + ē π
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 67
which leads to 2δ 2
sin τ =
1 − sin2 φ 1 − (6.47)
π
after some transformations. Inserted in equation (6.45) yields
R · π sin τ
y + ē =
2 · sin φ
and resolved to y , considering for ē the relationship (6.41), one finally obtains
R · π(sin τ − cos φ)
y = .
2 · sin φ
Thus we found the position vector for the image plane. Collecting all we get
x = R π
− δ
x = 2
R·π(sin τ −cos φ)
(6.48)
y = 2·sin φ
In this chapter different aspects of world maps are discussed. The first section
is dealing with projections encompassing specific properties. In the ensuing
sections quality aspects are emphasized. The term quality is meant in a nar-
rower sense of the lifelike representation of the globe. This actually presumes
mathematical tools for the decision-making process. Different approaches are
introduced and discussed.
According to a proposal introduced by Lee projections which are neither
conformal nor equal area are called aphylactic – some of them were already
presented before although the term ”‘aphylactic”’ was not mentioned. In
this script only two of this category are presented. Actually there are much
more such projections, e. g.:
• Trapezoidal
• Airy
• ...
69
70 Script ”Map Projections”
• ...
72 Script ”Map Projections”
Summarizing the above topics Robinson concluded further that the projec-
tion he was looking for can only be found by systematic “trial and error”.
Based on this idea he deduced a set of rather simple construction principles:
• The prime meridian is 0.5072 time as long as the equator of the pro-
jection.
x = f (ϕ)
y = f (ϕ, λ)
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 73
1. Approach (Beineke)
x = a · ϕ + b · s· | ϕ |c (7.1)
λ
y = (d + e · ϕ2 + f · ϕ4 + g · ϕ6 ) · (7.2)
π
with
1 for ϕ ≥ 0
s := sig (ϕ) =
−1 for ϕ < 0
and
2. Approach (Bretterbauer)
Bretterbauer suggested a solution using splines to approximate the images
of the meridians. For convenience a new parameter set with modified coeffi-
74 Script ”Map Projections”
'
18
xi =R qj |5j − ϕi | (7.6)
j=0
76 Script ”Map Projections”
with the coefficients for the parameters pj and qj as depicted in table 7.3
• Thus we get the three legs each representing a part of a great circle.
These legs now are taken as – with respect to the mapping scale – the
legs of a plane triangle.
• All other points on the datum surface are mapped to the plane in a
similar way. First the great circles of the datum surface connecting each
vertex of the triangle with the point to be mapped are determined.
• Afterwards the edge defined by the point and the vertex of the triangle
is mapped to the image surface too (also with respect to the scale).
As a result we don’t get a point but an area representing the mapped
point.
• The centroid of the above mentioned are could be treated as the rep-
resentation of the point on the datum surface. However, Chamberlin
didn’t define the algorithm for the final point determination.
1. Airy
1
2A = ((a − 1)2 + (b − 1)2 )
2
2. Airy-Kavrajski
1
2AK = (ln2 a + ln2 b)
2
3. Löbell a2 − b2 2
2L =
2ab
4. Fiorini a 2
2F = (a b − 1) + 2
−1
b
5. Jordan 2π
1
2J = (m − 1)2 dβ
2π 0
6. Jordan-Kavrajski 2π
1
2JK = ln m2 dβ
2π 0
7. Klingatsch
Pω ab − 1 + PV (a · b − 1)2
2Kl =
Pω + PV
Where Pω is the weight for the maximal distortion of the angle and PV
is the weight depending on the area distortion.
The local distortion quantities are all based on the ellipsoid of distortion.
Thus they simply represent the immediate vicinity of a specific point. Which
means they could be used to derive optimal projections with respect to spe-
cific points. This is what should be demonstrated in exercise 10.
Peters implemented this approach into a computer program and thus com-
pared the difference in length between 30 000 lines. Therefor he calculated
60 000 points randomly placed on the sphere connected each two points with
a Orthodrome and compared the length of the 30 000 great circles with their
projected images. From this comparison he derived a new quantity called Ep
which is defined as follows
1 ' |Si − Si |
n
EP = (7.7)
n i=1 |Si + Si |
Si
1'
n 1 − Si
= (7.8)
n i=1 1 + Si
Si
with
Si : length of the orthodrome (sphere)
Si : length of the straight line (image)
n : number of the lines .
82 Script ”Map Projections”
The quantities Ai and Ai denote the areas of the datum surface and the plane
respectively.
However, the shape of the projected circle has to be taken into account too.
Therefore the shape index according to Boyce & Clark has to be calculated
(see B.2). This is done by
n
'
r 100
ES = (16i · 100 −
j=1 rj n
i=1
In the above formula the quantity ri denotes the radius vector of the projected
circle on the plane. The result is called the shape index of the circle and
denotes the derivation of the resulting plane figure from an ideal circle. The
shape index for a circle is 0, whereas the shape index of all other geometric
forms are greater than 0.
The next step conforms to a normalization. To get rid of the unequal units
the quantities EA and ES are mapped on the interval [0,1]. This is expressed
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 83
by
1
EA,c = (KA − KAmin ) and
KAmax − KAmin
1
ES,c = (ES − ESmin ) in combination with
ESmax − ESmin
1 + EA
KA =
1 − EA
The above formulas however require minimal and maximal values for the
quantities KA and ES standing for the upper and the lower bound of the
interval. For the area distortion this can be achieved quite easy because
the minimal value is 1. Hence, the maximum distortion is difficult to find,
because there is no theoretical limit.
lower bounds:
ESm in = 0.0356
ESm ax = 0.1760
KAm in = 1
KAm ax = 1.8211
The pure addition of the two values ES,c and EA,c for any arbitrary projection
delivers its characteristic total distortion value E. Expressed as a formula
we get
E = EA,c + ES,c .
Capec also presented a method to describe the quality of small scale maps.
Unlike the hitherto mentioned approaches the ellipse of distortion is focused.
Based on this ellipse from which we easily can derive the angular as well as
the areal distortion (see section 3.3 and 3.2) Capek simply defined maximum
values for both. He called the combination of the two quantities the distortion
characterization Q. The limits for the deviations are more or less arbitrary,
however, Capek found out, that a maximum angular distortion of 40◦ and a
maximum distortion of the area of 1.5 are fairly good limits for the assessment
of cartographic projections.
In his analysis of different projections, Capek used the afore mentioned limits
as extreme values and evaluated the percentage of the projection surface
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 85
1' 2
k
εm = ε dF.
k i=1 i
Thus we sum up the local distortions of small areas which are treated as a
representative of the point and its vicinity.
Chapter 8
86
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 87
Together with μ2 = 1 we get as a result for the square of the line element
1
λ2 = .
μ2 (u, v)
x + iy = f (u + iv) .
Thereby the the function f (u + iv) is regular and analytic which means it is
continuous, can be differentiated and further the function fulfills the Cauchy-
Riemann differential equations.
For practical purposes these functions can be used to derive conformal pro-
jections. Therefore we only need to create complex quantities out of the
isometric parameters. This can be done by
and
z = x + iy (plane) or
z = u + iv (general surface) as well
z = f (w)
respectively.
88 Script ”Map Projections”
The isometric longitude remains unchanged. Figure 8.1 depicts the geo-
graphic as well as the isometric parameter system on the sphere. It clearly
shows that the isometric parameter system gets closer towards the poles.
Based on the isometric parameter system for the sphere derived in the section
above given by
w = q + il
and the isometric coordinates of the plane
z = x + iy .
The regular analytical function z = w maps the sphere to the plane. The
+60
+40
+20
−20
−40
−60
−80
In combination with
eiπ = cos π + i sin π = −1
we get
b π π b
r = k cot + = k tan − and
4 2 4 2
Φ = π−l .
Finally we derive
π b i(π−l)
z = 2 tan − e (again in accordance with the notation of Euler).
4 2
The transformation to Cartesian coordinates delivers
⎧
⎨ x = 2 tan π − b cos(π − l) = −2 tan π − b cos(l)
x = 4 2 4 2
⎩ y = 2 tan π − b sin(π − l) = 2 tan π4 − 2b sin(l)
4 2
(8.4)
Figure 8.3 depicts the graticule of the Stereographic projection.
x’
−135° 135°
y’
30°
−45° 45°
0°
n ∈ [0, 1] .
Further we need a rotation in oder to get the prime meridian to the south.
Thus it is
z = z̄ n eiπ(1−n) .
92 Script ”Map Projections”
x’
−180° 180°
y’
30°
−60° 60°
0°
π b π b
r̃ = k̄ cotn + = k̄ tann −
4 2 4 2
and
Φ̃ = π − nl.
Based on k and n further restrictions can be determined. The graticule in
figure 8.4 is depicted for n = 0.75 and b = 50◦ as a standard parallel.
Appendix A
Exercises
Exercise 1
The datum surface
⎧
⎨ x = R cos u cos v
x(u, v) = y = R cos u sin v (A.1)
⎩
z = R sin u
v = v
u = R tan u (A.3)
r = R
(R = 6370 km). Further two points are given: K (λK = 8◦ east, ϕK = 49◦
north) and M (λM = 38◦ east, ϕM = 56◦ north).
93
94 Script ”Map Projections”
Exercise 2
The formulas of the Mercator-Sanson projection are:
x = Rϕ
x(φ, λ) =
y = R λ cos ϕ .
Exercise 3
Find the mapping equations (in azimuthal coordinates) of an equal area conic
projection with one coinciding equidistant parallel.
c) Calculate the Euclidean distance A B between the two points (on the
image surface). Compare the shortest distance of the datum surface
with the shortest distance on the image surface.
96 Script ”Map Projections”
Exercise 4
b) Show the graticule for the two equidistant parallels ϕ1 = 40◦ north and
ϕ2 = 70◦ north, assuming the origin of the coordinate system λKN = 0◦ .
0◦ ≤ ϕ ≤ 90◦
−180◦ ≤ λ ≤ 180◦
c) Calculate the principal distortions for the two point P1 (ϕ1 = 40◦ north,
λ1 = 120◦ east) and P2 (ϕ2 = 60◦ north, λ2 = 120◦ west). Where are
the loci sharing the same properties?
Exercise 5
A conformal oblique conic projection with one coinciding parallel is pre-
sumed. The area between the parallels δ1 = 30◦ and δ2 = 70◦ has to be
mapped in such a way, that the distortions for δ1 and for δ2 are equal.
c) Calculate the intersection angle between the azimuth A = 50◦ and the
vertical in P on the image plane.
Exercise 6
Hint:
x 1 − cos x sin x dx x d tan x 1
tan = = , = ln tan , =
2 1 + cos x 1 + cos x sin x 2 dx cos2 x
Exercise 8
Exercise 9
Deduce the properties (equal area, conformal, equidistant, . . . ) of the below
depicted azimuthal projections:
H P’ H P’
P P
M M
H P’ H P’
P P
M M
Exercise 10
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 99
Exercise 11
Based on a map which is the result of a conformal projection of the ellipsoid of
revolution to the plane a captain derives the course angle of his passage from
San Juan (Puerto Rico) (ϕJ = 18.28◦ , λJ = 66.07◦) to Lisbon (ϕL = 38.43◦ ,
λL = 9.08◦ ). The course angle shall be the same for the whole journey.
a) Calculate the difference of the course angle, if the map is not derived
from a ellipsoid of revolution but from a sphere (radius = 6 370 km)
b) How is the difference of the length of the loxodrome between the ellip-
soid of revolution and the sphere? What is the length of the loxodrome
if we calculate with the course angle of the sphere?
c) And finally, what is the difference in the distance between the loxo-
drome and the shortest distance on the sphere?
100 Script ”Map Projections”
General information:
Exercise 12
Deduce:
The product of the distortion of the azimuth and the square of the scale
distortion is constant and equals the product of the principal distortions and
thus the distortion of area.
dβ 2
λ =V
dβ
Hint: Use the following equations:
b a
sin β = sin β cos β = cos β
λ λ
Exercise 13
Given is a map from South America in a transversal cylinder projection with
one coinciding equidistant parallel (see supplemantal sheet). Further on the
projection comprises the below properties:
• The maximal distorted azimuth and the amount (on the datum
and the image surface)
• The angle with the maximum distortion and the amount of the
distortion.
f) Depict the distortion ellipses for the origin of the coordinate system
and for point P .
Exercise 14
has been omitted
Exercise 15
Given is the trapezoidal projection of Eckert (known as Eckert II; see page
126):
2π
x = sign(φ) 2 − 4 − 3 sin |φ|
x (φ, λ) = 3 . (A.4)
y = √26π
λ
4 − 3 sin |φ|
• Outline the meridian for λ = ±45◦ . (Hint: The slope of the image of the
meridian is constant within the interval φ ∈ [0, π/2] and φ ∈ [0, −π/2]
respectively.)
Mathematical explanations
B.1 Trigonometry
Relationships for the sum and difference of arguments
103
104 Script ”Map Projections”
Spherical law of cosines : cos c = cos a cos b + sin a sin b cos (γB.12)
sin a sin b sin c
Spherical law of sines : = = (B.13)
sin α sin β sin γ
Spherical law of cotangents : cos b cos γ = sin b cot a − sin γ cot(B.14)
α
3 3 3·5 2 3·5·7 3
(1 − x)− 2 = 1 + x+ x + x . . . (|x| < 1) (B.15)
2 2·4 2·4·6
x3 x5
sin(x) = x − + −... (B.16)
3! 5!
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 105
x2 x4 x6
cos(x) = 1 − + − + ... (B.17)
2! 4! 6!
x x2 x3
ex = 1 + + + + ... (B.18)
1! 2! 3!
1 1
sin2 ax = x − sin 2ax (B.19)
2 4a
1 1
cos2 ax = x + sin 2ax (B.20)
2 4a
3 1 1
sin4 ax = x − sin 2ax + sin 4ax (B.21)
8 4a 32a
106 Script ”Map Projections”
Formulary
v + dv = const.
v = const.
v = const.
u + du = const. u + du = const.
E du
E’ du
’
ds
ds
β β’
u = const. u = const.
G dv G’dv
Definition:
ds line element on datum surface
m= =
ds line element on image surface
or
ds2 E du2 + 2 F dudv + G dv 2
m2 = =
ds2 Edu2 + 2 F dudv + Gdv 2
107
108 Script ”Map Projections”
u-Linie: v = const. ⇒ dv = 0
m2u = EE = h2
v-Linie: u = const. ⇒ du = 0
m2v = GG = k 2
⇒ m = m(u, v, β)
dm
Condition: dβ
=0
√
2(EF −F E ) EG−F 2
⇒ tan 2βh = 2F (EF −F E )+E(E −EG )
dm
Condition: dβ
=0
√
2(F E −EF ) E G −F 2
⇒ tan 2βh = 2F (F E −EF )+E (EG −GE )
⇒ βh 2 = βh 1 ± π
2
Assumption: F ≡ F ≡ 0
⇒ tan 2βh = 0 → βh1 = 0◦ and βh2 = 90◦
⇒ tan 2βh = 0 → βh 1 = 0◦ and βh 2 = 90◦
m2 = a2 · cos2 β + b2 · sin2 β
v + dv = const.
v + dv = const.
v = const.
v = const.
u + du = const. u + du = const.
E du
E’ du
dO dO’
τ τ’
u = const. u = const.
G dv G’dv
Definition:
sin τ
V =h·k·
sin τ
Area distortion as a function of a and b:
V =a·b
Theorem 3 The area distortion equals the product of the principal direc-
tions.
v = const.
v = const.
1. HVR
1. HVR
E du
E’ du
’
ds
ds
β dβ β’ dβ ’
F 0 F’ 0
u = const. u = const.
P(u,v) G dv 2. HVR P’(u,v) G’dv 2. HVR
Theorem 4 The product of the distortion of the azimuth and the square
of the scale distortion is constant and equals the product of the principal
distortions and thus the distortion of area.
v-Linie: m = b → dβ dβ
= a
b
>1
v−Linie
112 Script ”Map Projections”
dβ
dβ
=1⇒ m2 = a·b
datum surface : tan2 βm = ab
image surface : tan2 βm
= ab
conclusion : βm + βm = 90◦
d β’ d β’
=1 <1
u−Linie
dβ dβ
−βm βm
v−Linie
d β’
>1
dβ
dω dβ
Condition: =0⇒ =1
dβ dβ
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 113
Theorem 6 The distortion of the azimuth has its maximum where the angle
element is not distorted.
b−a
ωm = βm − βm oder sin ωm =
b+a
Maximal distorted angle (2 · |βm |):
The amount of the maximal distorted angle:
Ωm = 2 · ωm u−Linie
−βm βm
v−Linie
P(u,v)
Appendix D
Additional explanations
114
KIT, Geodetic Institute, Dr.-Ing. N. Rösch 115
80
60
40
80 80
60
5
40
10
60
20 20
40
10
20 20
20 20
10
40
5 20
20
60
10
40
60
80 80
40
60
80
129
130 Script ”Map Projections”
projection azimuthal
pseudoazimuthal, 59 conformal, 51
azimuthal, 26 equal area, 51
conformal, 98 conformal azimuthal proj., 51
equal area, 98 gnomonic projection, 52
equidistant, 26, 98 Robinson projection, 70
geometric interpretation, 98 Robinson, Arthur H. (1915–2004), 70
Bonne, 58 Romberg’s method, 34
conformal, 45
conic, 94 scalar product, see inner product
equal area, 94 scale distortion, 11, 111
oblique, 96 along the parameter curves, 109
one parallel, 96 arbitrary direction, 109
two parallels, 96 as as function of the azimuth, 13
azimuth, 108
cylinder, 93
extremes, 109
equal area, 65
parameter curve, 108
exercise, 4
series expansion, 36, 104
generalized, 62, 65
spherical trigonometry, 104
cylindrical
Stab-Werner-projection, 59
conformal, 48
Stereographic projection, 90
equal area, 27, 41, 61
surface
equidistant, 24, 26
analytical, 3
equator, 65
surface normal, 2, 7
meridians, 65
generalized, 54 Tissot’s Indikatrix, 21
Mercator, 89 Tissot, Nicolas Auguste (1824 - 1890),
Mollweide, 62 21
oblique, 29 triangle
orthogonal parameter system, 40 infinitesimal, 9
pseudocylindrical, 59 trigonometric funcions
equal area and elliptical merid- half argument, 104
ians, 62 trigonometric functions
equal area and equidistant par- for sum and difference, 103
allels, 61 general relationships, 103
Stab-Werner, 59 multiple arguments, 103
projection of Apian, 65 trimetric projection, 76