Reference Systems in Satellite Geodesy

R. Rummel & T. Peters

Institut fr Astronomische und Physikalische Geodsie

Mnchen 2001

1
 Reference Systems in Satellite Geodesy

Reiner Rummel & Thomas Peters

Institut fr Astronomische und Physikalische Geodsie
Technische Universitt Mnchen

1. Introduction

This summer school deals with satellite navigation systems and their use in science
and application. Navigation is concerned with the guidance of vehicles along a
chosen path from A to B. Precondition to any navigation is knowledge of position and
change of position as a function of time. Thus, navigation requires position
determination in real time. It has to combine time keeping and fast positioning. We
completely exclude here inertial navigation, i.e. the determination of position changes
- while moving - from sensors such as odometers, accelerometers and gyroscopes
operating inside a vehicle. The motion of a body has to be determined relative to
some reference objects. With inertial navigation methods excluded here, positioning
requires direct visibility of these reference objects. Typical measurement elements
are ranges, range rates, angles, directions or changes in direction. In some local
applications terrestrial markers may serve as reference objects. More versatile
reference objects in the past, because of their general visibility, were sun, moon and
stars and are artificial satellites today. In principle, positions as a function of time can
be deduced directly from the measured elements and relative to the reference
objects without any use of a coordinate system. Coordinate systems are not an
intrinsic part of positioning and navigation. They are introduced into positioning and
navigation as a matter of convenience, elegance and for the purpose of creating
order. To a large extent their choice is arbitrary and, again, in many ways a matter of
convenience. The description of objects or events in space and time in a coordinate
system requires four coordinates, three identifiing the position in space, the fourth
providing time. In Newtonian mechanics the time coordinate is independent from the
three space coordinates and "absolute". This is not the case when applying special
and general relativity. Although part of the coordinate definition in space geodesy is
done in the framework of the theory of relativity it is considered beyond the scope of
this lecture.

In the course of the centuries, the following hierarchy of three levels of coordinate
systems - or more generally - reference systems turned out to be particularily
meaningful:

Space-fixed or inertial systems, in which the positions of stars are fixed or almost
fixed and in which the motion of artificial satellites can be formulated according to
the Newtonian laws of mechanics.
Earth-fixed systems, in which all terrestrial points can be expressed conveniently
as well as vehicles in motion on the earth's surface.
Local horizon systems, fixed to observatories or instruments and often oriented
horizontally with one axis pointing towards north.

These three levels of reference systems are complemented, when needed, by some
specialized ones such as orbit or spacecraft systems or regional terrestrial systems.

2
Equally important, also time keeping requires some generally adopted reference for
maintenance, comparison and transfer of time.

2. Hierarchy of Reference Systems

When talking about reference systems it is useful to distinguish between the three
concepts coordinate system, reference system and reference frame.

Coordinate systems are the central - mathematical - element of any geodetic

reference system. The choice of a coordinate system in three dimensions requires
the definition of its origin (three elements), the orientation of the axes (three
elements), and the scale. It is convenient and common practice to choose
orthonormal base vectors and the same scale along all three axes. It is again a
matter of convenience to tie a system of curvi-linear coordinates, such as spherical,
geographical or ellipsoidal coordinates to any such orthonormal system of base
vectors. The transformation between coordinates given in two systems consists of a
shift of origin (three degrees of freedom) and rotations of the base vectors (three
degrees of freedom). We leave aside here the comparison of scale. By definition,
time is dealt here as completely independent.

A reference system consists of the adopted coordinate system and, in addition, of a

set of constants, models and parameters, that are required in order to achieve a
certain degree of generality or idealisation. This additional set could be necessary,
for example, in order to deal with tectonic plate motion, tides and the earth's
response to tides, it could define the constants of a reference ellipsoid or the
parameters of a reference gravity field. Since positioning and navigation are global
activities, nowadays, it is important that the same set definitions is used everywhere.
Thus, internationally adopted conventions are necessary. The International Earth
Rotation Service (IERS), a joint service of the International Association of Geodesy
(IAG) and of the International Astronomical Union (IAU) prepares the definition of so-
called conventional reference systems and of their implemention, see e.g.
(MCCARTHY, 1996). The conventional international celestial reference system (ICRS)
is adopted by IAG and IAU, the conventional terrestrial reference system (CTRS) by
IAG done.

Finally, a reference frame contains all elements required for the materialization of a
reference system in real world. In the case of space fixed or celestial frames it is
essentially an adopted catalogue of celestial objects such as stars or quasars, in the
case of a terrestrial frame it is the catalogue of coordinates of terrestrial points
(stations, observatories) as well as of their velocities. The catalogues are chosen to
be consistent with the conventions of the corresponding reference system.

As already said, it makes sense to introduce a hierarchy of three levels of reference

systems: space fixed systems, earth fixed systems and local or horizontal systems.
They are discussed now in more detail.

2.1 Space-fixed reference systems (or celestial reference systems (CRS)):

They are an approximation of an inertial system. Their purpose is twofold. Celestial

objects such as stars or quasars take a fixed position on the celestial sphere. After
correction for proper motion their direction can be expressed by two (fixed)
3
coordinates (kinematic part of a CRS). The motion of satellites and of the sun, moon
and planets can be described according to the laws of Newtonian mechanics (no
apparent forces). This may be denoted the dynamic component of the definition of a
CRS.

Definition of the coordinate system:

As origin either the barycentre of the solar system is chosen or the mass centre of
the earth. For our further discussion we only consider the geocentric definition. This
choice implies that the coordinate triad is accelerated. Its acceleration is mainly
caused by the gravitational attraction of sun and moon on the earth. The system is
not truely inertial. The base vectors are denoted here ei=1, i = 1,2,3 ("i " for inertial)
with ei=1 and ei=2 defining the {x, y}-plane and ei=3 the z-axis. The {x, y}-plane of a
CRS could either be chosen to coincide with the plane of the ecliptic, i.e. the orbit
plane of the earth about the sun, or with the equator plane of the earth. Here the
latter definition is adopted. The x-axis, i.e. base vector of ei=1, points in the direction
of the vernal equinox. It is the line of intersection between ecliptic and equator plane.
The z-axis, base vector ei=3, points into the direction of the mean rotation axis of the
earth. The y-axis completes a right-handed system. In ei the direction to any
object/event is expressed by the two angles right ascension (angle in the equator
plane counted from ei=1) and declination (elevation angle counted from the equator
plane). Due to the torque excerted by sun, moon and planets on the oblate earth and
with its equator not in coincidence with the ecliptic plane, the rotating earth
undergoes a complicated gyroscopic motion in space. Its steady part is a precession,
i.e. a constant rotation of the vernal equinox in the ecliptic plane, with a period of
approximately 25800 years or 50."3 per year. Superimposed to this steady pre-
cession are shorter period nutations. They are periodic changes of the inclination
angle of 23 26' of the equator plane with respect to the ecliptic. The main period is
thereby 18.6 years with an amplitude of 9."2.

As a consequence of the slow rotational motion of the adopted triad ei in space, one
has to distinguish the apparent or true system, i.e. the instantaneous orientation at a
chosen epoch, from a mean system, which is corrected for the periodic contribution
of the nutation. Finally, the definition of a conventional reference system, in which
celestial objects can be catalogued, requires the definition of a reference epoch.
Currently the adopted reference epoch is J2000.0, which is 12h January 1, 2000
Greenwich time.
In the past the realization of a CRS has been conducted solely by astronomical
methods. In recent years very long baseline interferometry (VLBI) and the ESA-
satellite HIPPARCOS provided a completely new set of catalogue information,
unprecedented in terms of the number of objects, their position accuracy and internal
consistency. Some basic information about the realization of the international
celestial reference frame (ICRF) is summarized in table 2.1.

Table 2.1: International Celestial Reference System (ICRS) and Frame (ICRF)

(according to IERS conventions)

origin: barycentre of solar system

geocentric origin derived from planetary motion, lunar motion and
from artificial satellites

4
orientation: in coincidence with ICRF at 1991.25
parallel to axes of FK5 ( 8mas)
mean equator at J2000.0
x-axis: mean vernal equinox at J2000.0

time: barycentric dynamical time (TDB)

directions: {, }

techniques: astronomy (FK5)

VLBI
HIPPARCOS satellite mission

realization: 212 defining quasars (VLBI)

608 quasar positions in total
144 000 star positions of HIPPARCOS catalogue

name: International Celestial Reference System (ICRS) and Frame (ICRF)

reference: (MCCARTHY, 1996)

2.2 Earth-fixed reference systems (or terrestrial reference systems (TRS)):

They serve the description of the position of points on the earth's surface or, in the
case of navigation, that of the motion of a vehicle on the earth's surface or close to it.
Also geophysical processes such as weather, temperature, magnetic or gravity field
are expressed in earth fixed systems. Finally, all our maps are based upon an earth
fixed reference system. During the past twenty years, due to the advance of space
techniques, precisions in positioning and navigation became so incredibly high that
the earth's surface cannot be considered anymore solid and fixed. Instead temporal
changes due to surface motions such as tectonic plate motions and deformations
due to tides or ocean and atmosphere loading have to be taken into account. This
complicates the definition and realisation of an earth fixed reference system severly.
On the other hand it implies that such a system can provide a framework for global
geophysical monitoning and consequently play a prominent role in earth system
studies.

Definition of the coordinate system:

The origin of the coordinate system is the geocentre. The geocentre is thereby
defined as the centre of mass of the earth including oceans and atmosphere. The
base vectors are denoted ee, e = 1,2,3 ("e" for earth-fixed). Again, the base vectors
ee=1 and ee=2 define the {x, y}-plane, ee=3 defines the z-axis. The {x, y}-plane
coincides with a conventional equatorial plane of the earth. The base vector ee=1 lies
by definition in the Greenwich meridian plane. The base vector ee=3 corresponds to
the mean position during 1900 to 1905 of the rotation axis of the earth. This terrestrial
pole is denoted conventional terrestrial pole (CTP) or IERS reference pole (IRP).
Finally, ee=2 completes the right-handed system. Its evolution in time will not change
its orientation relative to the crust, i.e. it will have no residual global rotation with
respect to the crust, (MCCARTHY, 1996).

The coordinate triad ee is accompanied by a mean earth ellipsoid with

5
 semi-major axis a = 6 378 137.0 m and
flattening f = 1 / 298.257 222 101.

It allows an easy conversion of cartesian into geographical coordinates.

The realization of the International Terrestrial Reference System (ITRS) is denoted

International Terrestrial Reference Frame (ITRF). It consists of an adopted global set
of cartesian station coordinates and velocities. Almost annually a new ITRF is
produced, based upon newest observations and identified as ITRF.yy, where the
numbers (yy) following the designation ITRF specify the last year whose data were
used for the realization of the frame. Currently the 2000 version, i.e. ITRF.00 is in
preparation. Transformation parameters between the annual realizations are
published in the IERS conventions, (MCCARTHY, 1996). Station coordinates at an
arbitrary epoch t are derived from

x(t ) = x(t o ) + v (t t o ) + x i (t )
i

with to the reference epoch, v the station velocities and x i coordinate corrections
due to various time variable effects, such as those listed in table 2.2.

Table 2.2: List of time variable effects, part of ITRS

- tectonic plate motions (angular velocities of 16 plates, model NNR-NUVEL 1, no-

net rotation)
- tides of the solid earth
- loading effects due to ocean loading
- atmospheric loading
- rotational deformation due to polar motion
- postglacial rebound
- instrument effects (antenna deformation, motion of antenna phase centres etc.)

There are several geodetic space techniques from which realizations of the earth
fixed reference frame are derived by so-called analysis centres. In a second step and
after a careful analysis of all individual solutions, one unique solution is computed
taking into account the results of the various centres as well as the various
techniques. Such combinations are based on the full variance-covariance error
matrices of the individual solutions.

Space techniques that are contributing to ITRS:

- very long baseline interferometry (VLBI)

(high precision and long term stability)

- satellite laser ranging (SLR)

(long term stability and geo-centricity)

- lunar laser ranging (LLR)

(geo-centricity, long term stability, relativistic effects)

6
- the French tracking system DORIS
(excellent global station distribution)

- Global Positioning System

(densest global network, short term stability, high precision).

We see that the various techniques are complementary to each other. Still, optimal
combination of all stations and all techniques remains a major challenge.
A summary of ITRS and ITRF is given in table 2.3. It should be added that there
exists a further earth fixed reference system: the World Geodetic System 1984
(WGS84). It is a system developed by the U.S. defence mapping agency, specifically
for the operational use of the GPS; ITRS can be seen as a refinement of WGS84.

Table 2.3: International Terrestrial Reference System (ITRS) and Frame (ITRF)

(according to IERS conventions)

origin: mass centre of the earth (including oceans and atmosphere)

scale: metre

orientation: in coincidence with BIH-System 1984 ( 3 mas)

evolution: no net rotation (NNR) with respect to crust

ellipsoid: Geodetic Reference System 1980 (GRS80)

a = 6 378 137.0 m
f = 1 / 298.257 222 101

directions: {, }, {, }, { B, L }

techniques: VLBI
SLR
LLR
GPS
DORIS

realization: coordinates x (to) and velocities v (to) at epoch to of a large number of

instrument locations of geodetic observatories equipped with one up to
six techniques in parallel

notation: ITRF.yy, z. B. ITRF.94

reference: (MCCARTHY, 1996)

2.3 Local horizontal reference systems (or topocentric systems): This class of
systems is associated with an instrument such as a GPS receiver, a VLBI telescope
or a camera. They are therefore topocentric, located in the origin or reference point of
the instrument and it is purpose of space positioning to determine the coordinates of

7
this reference point, either in ee or in ei. Local horizontal systems are introduced in
order to express the fixed or time variable pointing direction of the instrument to a
target point and in order to predict when and under what angles a target will rise or
fall. The orientation of base vectors of the local system can either be defined by the
local (level) horizontal plane, north direction and plumb line direction (zenith) or, in
ellipsoidal or spherical approximation, by the corresponding ellipsoidal or spherical
quantities.

Definition of the coordinate system:

The origin of the coordinate system is the instrument origin (topocentre). The base
vectors are denoted el, l = 1,2,3 ("l" for local). Base vectors el=1 and el=2 define the {x,
y}-plane, el=3 defines the z-axis. The {x, y}-plane coincides with the local horizon
(level surface). It is often approximated by an ellipsoidal or spherical reference
surface. The base vector el=1 points towards north. The z-axis points towards the
zenith (or normal of the ellipsoid or sphere). Base vector el=2 completes the left-
handed orthonormal triad; it points towards east.

The angle to an object in the horizontal plane counted from north (towards east) is
called azimuth A, the angle to an object from the zenith is denoted zenith distance
z, its complement to , the elevation angle above the horizontal plane is called
elevation angle b. Thus, when tracking a satellite at a station its changing direction
in the horizontal system is expressed by the angles {A, z}.

space-fixed
(inertial)

precession
nutation
-------------------
Rei length of day
polar motion

earth-fixed
(terrestrial)

from geo-centre
to topo-centre
Rle +
change in
orientation

local-horizon
(instrument)

Figure 2.1: Hierarchy of reference systems - from space-fixed via earth-fixed

to local horizontal

8
2.4 Special coordinate systems: In practice many additional coordinate systems
are applied. Only two examples are given here:

- Satellite orbit system eo, o = 1, 2, 3 ("o" for orbit):

It is a triad rigidly tied to the momentary orientation of the orbit plane of a satellite.
The orthonormal base vectors are

eo=1 pointing towards the satellite perigee (closest point), see figure 4.1
eo=3 perpendicular to the orbit plane
eo=2 completing a right handed system.

- Osculating orbit system es, s = 1, 2, 3 ("s" for satellite):

This triad is located at the centre of mass of the space craft with the following
orthonormal base vectors:
either
es=1 along track, in the direction of the velocity vector of the satellite,
es=2 perpendicular to the orbit plane (cross track),
es=3 completing a right-handed system (approximately radial)
or
es=3 radial, from the earth's centre of mass,
es=2 perpendicular to the orbit plane (cross track)
es=1 completing a right handed system (approximately along track)

3. Transformation between coordinate systems

Each of the three hierarchy levels of reference systems, discussed in the previous
chapter, takes an important role in positioning and navigation by space geodetic
methods, the space-fixed, the earth-fixed as well as the local horizontal system.
However, only if we know how to transform one into the other they become
operational. Under the assumption of equal scale along each axis and in each of the
coordinate systems considered here, transformation from one system into the other
consists of a shift of origin from one system to the other system followed by a rotation
between base vectors. (In reality it is all but trivial to warrant the same scale in all of
our coordinate systems. After all each instrument carries its own scale and it is
difficult to get all systems "calibrated".)

Let us consider the following situation, displayed in figure 3.1. The geocentre O, a
terrestrial point P and satellite position S form a basic triangle in three-dimensional
space; introducing the geocentric position vectors rP and rS and the topocentric
position vectors x S it can be expressed as:

rS = rP + x S . (3.1)

All positioning and navigation by satellites and all orbit determination rests on this
simple triangle condition between three fundamental vectors. However, behind this
simplicity some complications are well hidden. For, although these three vectors are
geometric objects invariant with respect to any chosen coordinate system, the com-
ponents forming these three vectors are not. Each of the three vectors is represented
in its own coordinate system.

9
(1) The geocentric orbit, i.e. vector rS , is given in the space fixed system e i .
(2) Terrestrial surface points, such as station or vehicle positions, i.e. vector rP , are
best represented in the earth fixed system e e .
S

measurements in xS orbit (satellite S) in

local horizon system space-fixed system
rS

stations P in rP
earth fixed system
mass
centre

Figure 3.1: Vector sum of geocentric point position vector, topocentric

measurement vector and geocentric satellite orbit vector, each one expressed
in a different reference system.

Since we assumed that the origin of both, e i and e e , is the geocentre, the two triads
can be brought into coincidence by rotation only. Three independent elements are
sufficient to perform this rotation. Thus, with the transformation Rei triad e i can be
rotated into the orientation e e and vice versa:

e e = Rei e i and e i = Rie e e . (3.2 a, b)

Here the summation convention over repeated indices is applied; Rei and Rie can be
represented by (3x3)-matrices containing the nine inner products (cosines) between
the two sets of orthonormal base vectors. For their matrix representations R ei and
R ie it holds:

R ei = R ie1 = R Tie .

The most common representation of rotations is that in terms of Eulerian

transformation angles, where e.g. R1 ( ) denotes a rotation about the 1-axis (x-axis)
by the angle in counter-clockwise direction.

The major contribution to Rei is the rotation of the earth about its spin axis with a
period of 24h; superimposed are precession, nutation, tiny variations in the steady
10
angular velocity and polar motion. Only with Rei (or Rie ) known, rS and rP can be
transformed from e i to e e and back.
(3) Measurements, and therefore x S too, are preferably represented in the local ho-
rizon system e l located at the instrument:
x S = x Sl e l
with time variable components x Sl , l = 1, 2, 3.

The triads e l and e e are connected to each other by

e l = Rle e e and e e = Rel e l . (3.3 a, b)

Since the y-axis of e l (direction east) lies in a plane parallel to the equator, i.e. the {x,
y}-plane of e e , only two independent angles are needed to rotate e e into e l . These
two angles are the astronomical latitude and longitude (or, as substitutes, the
geographical latitude B and longitude L or the spherical latitude and longitude
). The angles and define the plumb line direction at P in the coordinate
system e e . They can be regarded as fixed in time.

We take now a closer look into the transformations Rei and Rle .
Transformation between space-fixed and earth-fixed system: In principle, only
three Euler angles would suffice to express Rei. For phenomenological reasons it is
however common practice to separate Rei into four parts

precession P = R3 ( z A )R2 ( A )R3 ( A )

nutation N = R1 ( )R3 ( )R1 ( )
earth rotation R3 ( ) with = Greenwich apparent sidereal time (GAST)
polar motion W = R2 ( x p )R1 ( y p ) .

This means the transformation of the base vectors e i of the space fixed system to the
base vectors e e of the earth fixed system is composed of

e e = Rei e i
(3.4)
(
= W R3 ( ) N P ) e i
i
e

The equatorial precession parameters are (compare figure 3.2):

A = 2306."2181 t + 0."30188 t 2 + 0."017988 t 3

A = 2004."3109 t 0."42665 t 2 0."041833 t 3

z A = 2304."2181 t + 1."09468 t 2 + 0."018203 t 3

where the time is given in Julian centuries referred to J 2000.0 (= 2451545.0 ) :

t = (T J 2000.0) / 36525.0 . ( T is Julian Date)

11
The nutation part is more complicated. Several rather complex nutation models
exist, originally based on a model of a rigid earth, meanwhile taking its elasticity into
account. Corrections to these models are derived from VLBI and for short periodic
terms (< 20 days) improvements are expected from GPS.
The mean obliquity of the ecliptic is:

= 23 26' 21." 488 46."8150 t 0."000 59 t 2 + 0."001813 t 3

The correction angles and are expressed as series, see (MCCARTHY, 1996)
and compare figure 3.3. An approximation, accurate to about 11", is given in
(SEIDELMANN, 1992):

= 0.0048 sin (125.0 0.05295 d ) 0.0004 sin (200.9 + 1.97129 d )

= 0.0026 cos(125.0 0.05295 d ) + 0.0002 cos(200.9 + 1.97129 d )

where d is the number of day from J 2000.0 .

Figure 3.2: The precession angles A, zA and A

(from: SEIDELMANN, 1992, fig. 3.21.2)

Figure 3.3: The nutation angles + , and A

(from: SEIDELMANN, 1992, fig. 3.222.1)

12
The most important component of transformation Rei is R3 ( ) . It also defines the
interface between space-fixed and earth-fixed. The angle is Greenwich Apparent
(true) Sidereal Time (GAST) which is the hour angle expressed in units of time
between the Greenwich meridian at epoch and vernal equinox E. Consequently all
deviations of the earth's spin rate from a constant angular motion are contained in
. There is a long-term deceleration very likely caused by tidal friction. All additional
deviations are of the order of magnitude of a few milliseconds per day. Among others
these fluctuations are caused by the tides of sun and moon, by exchange of angular
momentum between atmosphere, oceans and solid earth and by long periodic effects
such as postglacial mass readjustment. They are denoted as changes in length of
day (LOD). See also figure 3.4.

Polar motion is defined as the movements of the earth's instantaneous spin axis
relative to ITRF, i.e. to e e . It is split into a x- and y-component in a coordinate system
fixed to the IERS reference pole (IRP) as a tangent plane. The x-axis points towards
the Greenwich meridian, the y-axis towards the meridian = 90 . Polar motion is
almost a circular motion superimposed by small fluctuations, compare figure 3.4. Its
most prominent component is the Chandler period (free rotation of an elastic earth
with a fluid core; the well-known Euler period would be the corresponding effect of
the free gyroscopic motion of a solid earth). The exact mechanisms leading to the
Chandler period are not yet completely understood. The Chandler motion is comple-
mented by annual variations, due to the interaction of the atmosphere with the solid
earth. Typical amplitudes are about 0."3 or about 8 m on the earth's surface. Both,
polar motion and changes in length of day cannot be represented very well by
models. They are derived from the various geodetic space techniques and represent
a contribution of space geodesy to earth sciences.

Figure 3.4: Variations in length of day between 1993 and 1999

13
 Figure 3.4: Secular polar wander between 1900 and 1992 and polar motion
between 7 July 1990 and 2 May 1993

Transformation between earth-fixed and local-horizontal: The shift of origin from

the geocentre to the topocentre is expressed by the position vector rP . The change in
orientation between e e and e l is derived from

e l = Rle e e
e

= Q1 R2 R3 ( ) e e (3.5)
2 l

Thereby it is Q1 the transformation from a left-handed to a right-handed system, in

matrix form

1 0 0

Q1 = 0 1 0 .
0 0 1

The complete transformation takes the form

sin cos sin sin cos

R le = sin cos 0 (3.6)
cos cos cos sin sin

14
with astronomical latitude and astronomical longitude . In the case of ellipsoidal
or spherical approximation {, } are replaced by the corresponding geographical or
spherical coordinates, {B, L} and { , } , respectively.

Transformation between cartesian and geographical coordinates: In practice, in

earth-fixed system e e , instead of cartesian coordinates x Pe often geographical
coordinates are used.
If we denote x Pe by {X , Y , Z } and geographical latitude, longitude and height above
the reference ellipsoid by {B, L, h} the following relations hold.
Forward Computation from {B , L, h} to {X , Y , Z }:

X = ( N '+ h ) cos B cos L

Y = ( N '+ h ) cos B sin L (3.7 a - c)

( )
Z = N 'e 2 N '+ h sin B

with normal curvature

a
N'=
1 e 2 sin 2 B

a2 b2 a2 b2
e2 = , e' 2 =
a2 b2

with a and b length of the semi-major and semi-minor axis of the reference ellipsoid.

Backward Computation from {X , Y , Z } to {B , L, h}:

(Method by Bowring)

tan L = Y / X (3.8 a - c)

auxiliary quantities:

p= X 2 +Y 2

Za
tan =
pb

Z + e' 2 sin 3
tan B =
p e 2 a cos 3

h = p cos B + Z sin B a 1 e 2 sin 2 B

15
4. Time

In the context of this lecture time is regarded as absolute and independent of space.
In reality time keeping, and satellite measurements in general, are so accurate
nowadays that adequate modelling requires the use of special and general theory of
relativity for practical reason.

Time keeping requires a periodic process, a counter (in order to count the number
of periods) and an origin where the counting starts. In addition, in order to be able to
keep the same time at different locations some means of transfer/transport of time
has to be available. There exist a number of natural "clocks" that produce very stable
periodic oscillations: the orbit of the earth about the sun, of the moon about the earth
and earth rotation. Their fundamental periods, year, month and day, are closely
related to natural processes such as seasons that affect our living conditions and
these periods define the basic structure of our life. From these fundamental periods
the basic long term counting structure has been deduced, our calendars (we use the
Gregorian calendar, adopted in 1582). In scientific work a continuous counting is
preferable to the complicated structure of counting with months or year of varying
length. For this purpose the Julian date (JD) has been invented with 36525 days per
century.

The adopted reference date is

J2000.0 = 2000 Jan 1.5 = January 1, 2000 at 12h

where it is

JD 2451 545.0.

For a long time the natural period day, and even more the revolution of the moon,
were superior in terms of stability to any artificial clock. Only with the advent of quarz
and atomic oscillators artificial clocks were created that meanwhile surpassed the
precision and stability of natural clocks. Our current definition of the unit of second is
based on the oscillation period of a caesium clock. In 1984 atomic time (Temps
Atomique International = TAI) has been introduced as official, internationally adopted
time. Its has a constant off-set of 32.s184 with respect to the terrestrial dynamic
time (TDT). The latter is derived from models of planetary motion and based on the
theory of relativity. TAI has a constant off-set of 19s with respect to GPS-time.

Civilian time is related to the rhythm of day and night, i.e. to the rise and fall of the
sun. Because of the complicated deviations of the apparent motion of the sun, some
model or mean solar motion has been conceived. It refers to the Greenwich meridian
and is denoted universal time (UT). From UT standard zonal times have been
deduced; in our case MEZ. Earth rotation - and therefore UT - exhibits a drift and
small irregular fluctuations (changes in LOD) with respect to TAI. In order to
circumvent this, a coordinated universal time (UTC) has been conceived, which on
the one hand is kept synchronous with respect to TAI and on the other hand, through
regular corrections (leap seconds), is kept within small bounds to follow the actual
angular rate of the earth. The actual and uncorrected universal time is denoted UT1.
It represents the actual phase angle of the rotating earth. The difference UT1-UTC is
provided in monthly tables (see (MCCARTHY, 1996)) and provided as coded message
in broadcasted time signals. If the difference UT1-UTC exceeds the size of 0.9s a
16
leap second is introduced. UT0 completes the system of universal times. It contains
all variations in rotation due to polar motion.

Finally, sidereal time is the angle of a terrestrial meridian (rotating with the earth)
with respect to vernal equinox. The most prominent types of sidereal time are

Greenwich Mean Sidereal Time (GMST)

and

Greenwich Apparent (or true) Sidereal Time (GAST) .

GMST is corrected for fluctuations caused by nutation. It is

= + n (4.1)
= + cos + 0."00264 sin + 0."000063 sin 2

and the mean node of the moon. Greenwich Apparent Sidereal Time is needed for
the transformation from earth-fixed to space-fixed. The complete chain of time
transformations is summarized in figure 4.1.

Since sidereal time is measured with respect to vernal equinox, i.e. the x-axis of ei,
while universal time is a solar time and counted with respect to the apparent pass of
the sun through the meridian at Greenwich the length of the year is different by one
day: tropical year

in solar days: 365.24220

in sidereal days: 366.24220.

This difference has to be accounted for when transforming UT1 to GMST. It holds

GMST = UT 1 + ( ) 12 h (4.2)

with the right ascension of the sun:

() = 12 h + (24110. s 54841 + 8640. s 812866 t + 0. s 093104 t 2 6. s 2 10 6 t 3 ) (4.3)

and

t = (T J 2000.0) / 36525.0

with T Julian Date at epoch and J2000.0 = JD2451545.0.

17
 TDB
Temps Dynamique
Barycentric

relativity theory

TDT
Temps Dynamique
Terrestre

-32.s184

TAI -19s T(GPS)

Temps Atomique GPS
International Time

-32s (currently)

MEZ UTC TJD JD

1h Coordinated Universal Julian
Central European
Time Time Date

tables with
corrections

UT1
Universal
Time 1

eq. (4.2)

GMST
Greenwich Mean
Sidereal Time

eq. (4.1)

GAST
Greenwich Apparent
Sidereal Time

Figure 4.1: Transformation between time systems (after: MLLER, 1999)

18
Tutorial - Representation of Satellite Orbits in Various Coordinate Systems

The purpose of this tutorial is to give an introduction into the representation of

satellite orbits in various, often used coordinate systems. A collection of related
formulas is given in table 5.1.
We consider the most simple type of satellite orbit, a Kepler ellipse. Its shape is
defined by a and b . The earth is assumed to be a perfect and homogeneous sphere
with its centre in one of the focal points of the ellipse. In the orbit coordinate system
e o the position vector of the satellite takes the form (compare figure 5.1):

rS cos a (cos E e)

rS = rS sin = a 1 e 2 sin E (5.1)

0 o 0 o

Since e o =3 is perpendicular to the orbit plane the z-coordinate of rS is zero. Both the
true anomaly and the eccentric anomaly E are functions of time, as is the radial
distance rs . Eq. (5.1) allows us to display the Kepler ellipse in the {x, y}-plane of e o ,
i.e. to show all satellite positions as a function of time.

a satellite

r
eo=2

E
eo=3
perigee
eo=1
geocentre
ae
Figure 5.1: Orbital ellipse

In the space-fixed triad e i the same position vector takes the form

rs cos cos

rs = rs cos sin , (5.2)
r sin
s i

with direction angles right ascension and declination .

19
Table 5.1: Collection of formulas

ellipse:

semi-major axis a
semi-minor axis b
a2 b2
first eccentricity e=
a2
a2 b2
second eccentricity e' =
b2

position in orbital plane (compare figure 5.1):

orbit period T
2 GM
mean motion n= = (mean angular velocity)
T a3
mean anomaly l = n (t t o )
eccentric anomaly E e sin E = l ( l counted from perigee)
1+ e E
true anomaly tan = tan or
2 1 e 2

1 e 2 sin E
tan =
cos E e
rS =
(
a 1 e2 )
1 + e cos

rS = a(1 e cos E )

gravitational constant mass of the earth: GM = 398 600.5 km/s

orbit elements of the two GPS-satellites shown in the figures (2 revolutions):

T 11h 58 m
= 0 and 60
= 0
i = 55
a = 26560 km
e = 0.01

20
Orbit and hierarchy of coordinate systems

In order to be able to display the Kepler ellipse in e i we have to know the

transformation from the orbit system to the space-fixed system, Rio . As long as the
earth is a sphere, the orbit ellipse remains fixed in space, no precession of the orbital
plane occurs. Thus only the three time invariant rotation angles from e o to e i are
needed. They are the ascending node , the inclination i of the orbit plane with
respect to the equator and the argument of perigee , compare figure 5.2:

e i = Rio e o (5.3)
= (R3 ( )R1 ( i )R3 ( )) e o
o
i

or in matrix form

cos cos sin cos i sin cos sin sin cos i cos sin sin i

R io = sin cos + cos cos i sin sin sin + cos cos i cos cos sin i (5.4)
sin i sin sin i cos cos i

satellite perigee

ei=3
geocentre ei=2
ei=1
vernal equinox
i
x y
ascending
node

satellite orbit
Figure 5.2: Orbital orientation in a space-fixed coordinate system

With Rio the orbit can now be displayed in e i ; a relation with right ascension and
declination ( , ) is established. The result is shown in figure 5.3 for GPS-type
satellite orbits. The orbit elements are given in table 5.1.

It is more interesting to connect the orbit triad via e i with the earth-fixed triad e e and
the local horizon system e l . If all small effects, such as precession, nutation,
changes in length of day, and polar motion are neglected, e.q. (3.4) becomes

21
 e e = Rei e i (5.5)
= (R3 ( )) e i
i
e

with matrix representation

cos sin 0

R ei = sin cos 0 (5.6)
0 0 1

where Greenwich apparent sideral time can be replaced by

= n(t t o ) .

Now we find for the components of rs in e e :

rSe =1 cos cos( )

e=2
rS = rS = rS Ri rS cos sin ( ) .
i e
(5.7)
r e =3 sin
S e

The result is shown in figure 5.4.

space fixed geocentric

= 0
= 60
20000
z [km]

20000

10000
0 20000
10000 0
20000
y [km] x [km]

Figure 5.3: The orbits of two GPS satellites in a space-fixed geocentric

coordinate system
earth fixed geocentric

= 0
= 60
20000
z [km]

20000

20000

0 20000
0
20000 20000
y [km]
x [km]

Figure 5.4: The orbits of two satellites in an earth-fixed geocentric coordinate

system
22
Finally, the same orbit is brought into the local horizontal triad e l . Since we are only
interested in the direction of the satellite, the shift of origin can be disregarded.
In terms of the orientation angles in e l , azimuth and zenith distance, it takes the
simple form:

sin z cos A

rS = rS sin z sin A (5.8)
cos z
l

and using Rie , eq. (3.6), applied to (5.7) we arrive at

cos sin sin cos cos

rS = rS cos sin . (5.9)
sin sin + cos cos cos
l

We replaced astronomical latitude and longitude, and , by the spherical ones,

and . The angle is the so-called hour angle

= + . (5.10)

Comparison of (5.8) and (5.9) yields A and z as a function of , and :

cos tan sin cos

cot A = ,
sin
(5.11)
cos z = sin sin + cos cos cos .

In figure 5.5 azimuth and zenith distance are shown in stereographic projection, with
the projection plane coinciding with the horizon and centered at station P . This type
of plot is denoted sky plot or visibility plot. It provides a convenient picture of the
rise of a satellite above the horizon and of its fall. It also shows under what elevation
and azimuth a satellite passes a station.

Figure 5.4, the orbit of a satellite in e e , is not very instructive. Instead, one often
computes from the cartesian coordinates, eq. (5.7), the ellipsoidal coordinates {B, L}
using eq. (3.8) or simple the spherical coordinates { , } . They can be shown as a
so-called ground track plot, i.e. as a projection of rs onto the sphere. Figure 5.6
gives the ground track of the two GPS satellites.

23
 visibility, Alpbach

60

= 0 30
= 60
00
330 30
30
300 60
60

270 90

240 120

210 150
180

Figure 5.5: Sky plot or visibility plot, centered at Alpbach ( = 4724' N,

= 1157' E), with traces of two GPS satellites

groundtrack
90

45
lat []

45
= 0
= 60
90
180 90 0 90 180
lon []

Figure 5.6: Ground tracks of two GPS satellites

24
 3D plot orbit shape plot

space-fixed fixed to
(inertial) , i, orbit plane

GAST

earth-fixed {x,y,z}
(terrestrial) ground track plot
{r,,}

local-horizon
(instrument)
A, z sky plot (visibility)

Figure 5.7: Hierarchy of reference systems and the associated graphical re-
presentations

5. Time

25
References

BROUWER, D. , G. M. CLEMENCE (1961) Celestial Mechanics, Academic Press, New

York.
[classical textbook on celestial mechanics]

COLOMBO, O. L. (1984) Altimetry, Orbits and Tides, NASA technical memorandum

86180, Goddard Space Flight Center, Greenbelt Md.

ILK, K. H. (1996) Reference Systems in Geodesy, lecture notes no. 5, 2nd Tropical
School of Geodesy, Bandung.

KAULA, W. M. (1966, 2000) Theory of Satellite Geodesy, Blaisdell, Dover.

[classical book on dynamic satellite geodesy]

MCCARTHY, D. D. (2000) IERS Conventions 2000, IERS technical note, IERS, Paris.
[description of all internationally adopted standards]

MLLER J. (1999) Zeitskalen, in: 3. DFG-Rundgesprch zum Thema Bezugssysteme

(Hrsg.: M. Schneider), Mitteilungen des BKG, no. 5, Frankfurt/Main.

ROTHACHER, M. (2000) Raumfeste und Erdfeste Referenzsysteme, in:

A. Bauch, M. Rothacher, R. Rummel: Bezugssysteme in Lage und Hhe,
IAPG/FESG, no. 7, Mnchen.

RUMMEL, R. (2000) Erdmessung, Teil 1, Vorlesungsmanuskript, IAPG, Mnchen.

SEIDELMANN, P. K. (ed.) (1992) Explanatory Supplement to the Astronomical

Almanac, University Science Books, Mill Valley, Ca.
[comprehensive handbook on global reference systems and time]

TORGE, W. (2001) Geodesy, 3. edition, de Gruyter, Berlin.

[modern introduction into geodesy]

26