Chapter 3

Coordinate and Reference Systems

This chapter describes coordinate systems and references frames, giving in-
formation about the various usages of these within the MPG.

3.1 Positional Conventions

A coordinate system is a method of locating points in a given space of given
dimensions by means of numerical quantities specified with respect to a ref-
erence. These quantities are the coordinates of the point. Each coordinate
system corresponds to a particular way of expressing the location of a point
with respect to the given reference. Each set of coordinates corresponds to
only one point in that coordinate system. In general, each position in space
has three degrees of freedom, and is therefore represented by three coordi-
nates. Coupled with a position in time measured from a reference epoch, the
4-tuple defines the spacetime coordinates of the point.

Although any specific coordinate system is useful for numerical calculations

in a given space, the space itself is considered to exist independently of the
particular choice of coordinate system.

The spatial coordinates of a point relative to the origin define a position vec-
tor. The time rate of change of that point relative to the coordinate axes de-
fines the velocity vector of the point. The 6-tuple containing pos ition and
velocity vector components is called the state vector of the point. State vec-
tors are normally assumed to be in rectangular component form, and this is
the SPICE standard.

The use of coordinate systems permits geometrical objects, their relation-

ships, and their motions to be described mathematically. The 3-dimensional

1
2 Explanatory Supplement to Metric Prediction Generation

coordinate systems used predominately within the MPG are the rectangular,
(Cartesian), spherical (latitude-longitude-radius), and cylindrical systems.
Two dimensional systems include rectangular, polar coordinates, azi-
muth/elevation, hour-angle/declination, right-ascension/declination, XY, and
X’Y’. These will be discussed more fully later in this chapter.

A reference frame, also called a coordinate reference frame, is defined by (1)

a designated location, called the origin , (2) a designated orientation, called
the reference plane, and (3) a designated direction in the reference plane,
from which all other locations and directions in the plane are reckoned, (4) a
designated metric for distance measurement, and (5) a designated metric for
time measurement. The reference plane itself is defined by its distance from
the origin and a vector normal to the plane. The reference frame may also be
associated with an epoch that defines the time origin in the frame.

Frames are classified into two major categories, inertial and non-inertial
frames. An inertial frame is one in which Newton’s laws (especially the first)
are hypothesized to hold; a body at rest remains at rest, and a body in uniform
motion retains that motion. The term celestial frame is also used to denote an
inertial frame. A non-inertial frame is one in which Newton’s laws are not
assumed to hold. Examples of non-inertial frames are accelerating frames,
rotating frames, and frames subject to gravitational fields or spacetime curva-
ture.

Due to gravitation, which permeates the universe, there are no true physical
inertial frames, but approximations within regions of space are possible.
Such frames exhibit no perceptible short term rotation with respect to the star
background, but certain adjustments in distances (light times) and apparent
positions must be made due to relativistic effects.

3.1.1 Rectangular Coordinates

The concept of a rectangular coordinate system is attributed to the French
philosopher and mathematician Rene Descartes in 1637. In this system, a set
of mutually orthogonal axes are defined at a point of origin, from which the
components of a given point are measured as orthogonal projections of the
point onto each of the axes.

For three dimensional space, there are three such axes and three coordinate
components per point. Such points are commonly designated as (x, y, z).
Coordinate and Reference Systems 3

If the three orthogonal unit vectors defining the axes of a rectangular system
have a designated numbering, and are such that u1 × u 2 = u3 , then the system
is said to be right-handed.

When applied to reference frames, the three axes are (1) the designated direc-
tion in the reference plane, (2) a direction in the reference plane that is a 90º
rotation in a counter clockwise direction about the normal vector, and (3) the
normal direction.

If ( x , y, z) represents a vector based at the origin, then the angles

(θ x ,θ y ,θ z ) between the vector and each of the coordinate axes are called the
direction angles, and are given by

 x 
θ x = arccos  2 
 x + y2 + z 2 
 
 y 
θ y = arccos   (3-1)
 x2 + y2 + z 2 
 
 z 
θ z = arccos  
 x2 + y2 + z 2 
 
The cosines of these angles are referred to as direction cosines and satisfy the
relation

cos 2 θ x + cos 2 θy + cos 2 θz = 1 (3-2)

A unit length vector based at the origin is called a pointing vector. The Car-
tesian coordinates of a pointing vector are thus (cos θ x , cos θy , cos θz ) . For
this reason, the term direction cosines is often used to designate a pointing
vector in rectangular coordinates.

3.1.2 Spherical Coordinates

Points specified as spherical coordinates ( r , λ , ϕ ) also called spherical polar
coordinates, are located by the distance r from the origin; the longitude angle
λ in the reference plane measured counter clockwise from the reference di-
rection about the normal vector; and the latitude angle f measured from the
origin and reference plane, positive toward the normal. In many cases the
polar angle (90o − ϕ ) , or complementary latitude (colatitude), is used.
4 Explanatory Supplement to Metric Prediction Generation

Sometimes the angle is measured in the opposite direction and termed the co-
longitude. It is also sometimes termed east longitude or west longitude to
distinguish the direction of measure. The symbols used for longitude and
latitude are often different in different contexts. Usage will clarify the con-
ventions appearing in this document. Care should be exercised when apply-
ing a SPICE coordinate transformation utility to make sure that the input and
output arguments match the assumed definitions of the parameters.

3.1.3 Cylindrical Coordinates

Points in cylindrical coordinates ( r ,λ , z ) are located by the length r of the
projection of the point on the reference plane, the longitude λ measured from
the reference direction counterclockwise about the normal, and the distance
z of the point from the reference plane, positive in the direction of the nor-
mal. Points are commonly designated by ( r ,λ , z ) , although symbols and
order of components tend to change according to the application. The con-
text of usage will clarify the conventions used in this document.

Cylindrical coordinates are commonly used within the DSN to define station
locations, where they use the convention (CU, CV, LO), denoting, respec-
tively, the distance from Earth’s polar 1 axis, the signed distance above the
equator, and the geocentric longitude.

3.1.4 Coordinate Transformations

The MPG finds it necessary and convenient to represent objects in space
sometimes in several different coordinate systems according to the context of
the application. The basic transformations are (1) simple coordinate transla-
tions in which the axes represent the same frame, such as in rectangular-to-
latitudinal conversion; (2) rotations, in which coordinates of one frame are
translated into those of another frame in a way that preserves angles and dis-
tances, such as the transformation from the J2000 inertial frame to a topocen-
tric frame at a deep space station; and (3) spacetime transformations, in
which the relativistic relationships between space and time are accounted for,
such as in the transformations between the solar system barycentric frame
and local topocentric frames. Such relativistic effects were discussed in a
previous chapter and are not further elaborated here.

1
In NSS MP documentation, Earth’s polar axis is referred to as the “spin axis”.
Coordinate and Reference Systems 5

Fortunately, the SPICE toolkit and NAIF utilities contain all the utilities
needed to make such transformations, without the need to resort to the under-
lying mathematical formulas. There are geometrical translations among rec-
tangular, spherical, and cylindrical coordinates, such as RECLAT, RECSPH,
RECCYL, LATREC, LATCYL, LATSPH, SPHREC, SPHCYL, and SPHLAT.
Here, of course, REC refers to rectangular, SPH to spherical, CYL to cylindr i-
cal, and LAT to latitudinal systems. The LAT utilities apply to spherical co-
ordinates in latitude and longitude, whereas the SPH utilities apply to spheri-
cal coordinates in colatitude and longitude. The initial mnemonic refers to
the system being transformed and the latter one to the target system.

SPICE also has utilities GEOREC and RECGEO for translating between geo-
detic and rectangular coordinates, treated later in this chapter.

Rotations from one recognized frame to another are provided in SPICE by the
functions SXFORM and PXFORM, which return the rotation matrices required
to transform vectors in one frame into vectors in another. The former is used
for translation of state vectors, and the latter, for position vectors. The state
vector between an observer and a given target in a given frame of reference
and at a given ephemeris time is provided by the SPICE SPKEZ2 function,
which not only applies the geometric transformations, but also makes the
special relativity correction (stellar aberration) and accounts for the relative
time differences between observer and target (planetary aberration).

Explicit coordinate rotations, as at times are needed in the MPG, are made
using SPICE rotation routines. One method of specifying a coordinate rota-
tion that is particularly useful is through the use of Euler angles. The set of
coordinates in one coordinate system may be uniquely transformed into those
of another system by the application of three successive rotation angles about
specified axes.

If the (rectangular) axes are designated 1, 2, 3, then the notation [θ ]n signifies

the matrix that rotates transforms one coordinate system by an angle ? about
axis n. Applying this matrix to a vector yields the vector’s representation
relative to the rotated coordinate system. The right-handed rule applies, so a
counter-clockwise rotation is in the positive direction. The SPICE function
ROTATE generates this matrix.

2
In generating predicted pointing angles, only the position vector of the state vector returned
by SPKEZ is used. Aberration corrections applied by SPKEZ are under user control.
6 Explanatory Supplement to Metric Prediction Generation

The composition of rotations [θ1 ]a [θ2 ]b [θ3 ]c is sometimes referred to as an a-

b-c rotation with respect to the given Euler angles. For example,
[θ1 ]3[θ2 ]1[θ3 ]3 is a “3-1-3” rotation. The individual rotations are applied to a
vector in the order c-a-b. Given the Euler angles and axes, the resulting
composite rotation matrix may be found using the SPICE function EUL2M.

But while it is true that all rotation matrices can be decomposed into a set of
Euler angles, this decomposition may not be unique. Given a rotation matrix
and the set of axis designations, in which the second Euler angle axis differs
from the first and third, the SPICE function M2EUL does produce a valid set
of Euler angles. See the source code commentary for the method used in this
case.

The NAIF high precision time conversion routines HPTx2y, where x and y
specify the particular type of translation desired between TA(I) or U(TC) and
E(phemeris) time, handle transformations between UTC, local atomic time,
and solar system barycentric time.

General relativistic effects on light time of flight between target and observer
are made in the NAIF RLTIME utility.

Hence, given the SPICE and NAIF programming environments, there is little
need for MPG routines to implement their own mathematical equations of
coordinate transformation. The interested reader may consult the SPICE and
NAIF source code commentary and other information for further details on
the conversion formulas, usage, restrictions, and limitations.

3.1.5 Frame Classifications

There are many celestial coordinate frames that are used to specify pos itions
of objects. These are commonly named according to their origins or points of
observation. The principal celestial frames are designated as

1. Topocentric : having its origin located at an observer’s location,

on the surface of Earth or other body.
2. Geocentric : having its origin at the Earth center.
3. Body-centric or planetocentric: having its origin at the center of a
designated celestial body. There are corresponding designations
for individual planets.
4. Selenocentric : having its origin at the center of the Moon.
Coordinate and Reference Systems 7

5. Heliocentric : having its origin at the center of the Sun.

6. Barycentric : having its origin at the center of mass of a system,
such as the solar system, or a system comprised by a planetary
and its satellites.
The principal celestial reference planes are designated as

1. Horizon: the plane normal to the local vertical (apparent direction

of gravity) with origin at an observer.
2. Local meridian: the plane with origin at an observer and contain-
ing the local vertical and Earth’s axis of rotation.
3. Planet meridian: the local meridian of a designated planet.
4. Celestial equator: the plane with origin at the geocenter and nor-
mal to the axis of rotation.
5. Planet equator: the plane with origin a planet center normal to its
axis of rotation.
6. Ecliptic: the mean plane of Earth’s orbit around the Sun.
7. Orbital plane: the plane of an orbit of a body around another.
8. Invariable plane or Laplacian plane: the plane with origin at a
system’s barycenter and normal to the axis of angular momentum
of the system.
The principal non-inertial frames of interest are

1. Body-fixed frame: the rotating frame with its origin at the center
of mass of a designated body and its reference plane being the
equatorial plane. The designated direction is toward an arbitrarily
defined prime meridian. The spherical coordinate angles are lon-
gitude and latitude.
2. Crustal-fixed frame: the body fixed frame whose axis of rotation
is the instantaneous axis of rotation, which is the axis of figure, or
axis of maximum moment of inertia.
3. Topocentric frame: the rotating frame with its origin at a point on
a body’s surface and its reference plane perpendicular to the local
vertical (i.e., the horizon). The designated direction is north. The
spherical coordinate angles are azimuth and elevation3 .

3
The preferred term in astronomical usage is altitude. However, the term elevation is used for
8 Explanatory Supplement to Metric Prediction Generation

Body-fixed and topocentric frames rotate with respect to inertial frames ac-
cording to the dynamical effects of diurnal and orbital motions.

3.2 Celestial Reference Systems

Reference systems are defined so as to provide consistent bases for observa-
tion and computation. Such systems specify the reference frame at a given
epoch, called the standard epoch, together with all the necessary formulas,
procedures, and constants required to transform from the reference epoch to
any other date. The motions of objects in the solar system offer a number of
possible bases for reference frames, the standard being the Earth equator and
ecliptic. The choice of epoch and whether the planes are mean or dynamical
offer further alternatives.

A celestial reference frame, however defined, involves a process of generat-

ing ephemerides in the chosen frame based on the equations of motion in that
frame and determining the necessary transformations that make the computed
ephemerides consistent with observations. This chapter does not elaborate on
this process, but does discuss a few of the standard frames the MPG must
handle.

The following material is meant to clarify the relationships among the various
inertial frames used within the MPG. Since celestial reference systems are
defined in terms of terrestrial features, it is instructive to elaborate on the par-
ticular characteristics of influence.

Terrestrial reference frames will then be more fully discussed in a subsequent

section.

3.2.1 The International Celestial Reference System

Since antiquity, celestial and terrestrial coordinate systems have been tied to
Earth’s orientation in space, but not always in a coordinated way. The Inter-
national Earth Rotation Service (IERS) was founded on 1 January 1988 as a
replacement for the Earth-rotation section of the Bureau International de
l'Heure (BIH) and the International Polar Motion Service (IPMS), with the
goal of providing an interdisciplinary service to maintain the key connections
between astronomy, geodesy, and geophysics. In particular, it was to coordi-
nate the defin itions of the IERS Celestial Reference Frame (ICRF) and IERS

historic reasons.
Coordinate and Reference Systems 9

Terrestrial Reference Frame (ITRF), and to oversee the generation and distri-
bution of Earth orientation parameter (EOP) data.

The ICRF is a celestial coordinate system, updated yearly by the IERS, that is
based on a set of VLBI-determined coordinates of compact extragalactic ra-
dio sources. The ITRF is a terrestrial coordinate system, updated yearly by
the IERS, implicitly defined by a standard set of station-location coordinates.
The directions of its axes are continuous with those of the previously used
BIH Terrestrial System.

The basic spatial reference system for the MPG and SPICE is the J2000 sys-
tem. This is an inertial reference frame specified by the orientation of Earth's
mean equator and equinox at the J2000 epoch, which is Greenwich noon on
January 1, 2000 Barycentric Dynamical Time (JD 2451545.0). The termi-
nology J2000 is used to name both the fundamental inertial frame and its des-
ignated epoch. To avoid confusion, usage here will attempt to distinguish the
context in which the term is used.

The ICRF epoch is that of J2000, its origin is the solar system barycenter, its
reference plane is that of the mean equator at epoch, and its reference direc-
tion is the equinox, as determined by IERS VLBI sources. IERS conventions
equate it with the frame of JPL Development Ephemeris DE-403, and SPICE
documentation equates the frame of DE-403 with J2000. It is also designated
as the basis of the fundamental catalog of stellar objects, FK5. Thus, the cur-
rent ICRF, J2000, FK5, and the frames of the DE-4xx series of fundamental
ephemerides, all refer to the same inertial frame of reference.

3.2.2 Equatorial and Ecliptic Frames

The intersection of the plane of Earth’s equator and the ecliptic is a line that
defines the directions of equinoxes. In equatorial and ecliptic frames, the
direction of the vernal equinox at a particular epoch is chosen as the origin of
longitude. It is often denoted in the literature and almanacs by the symbol L.
In ancient times the equinox was located in the constellation Aries4 , and the
equinox was thus termed “the first point of Aries.” The symbol, depicting a
ram’s head, became the sta ndard notation of the equinox in the astronomy
literature. Both celestial and ecliptic frames use this equinox as their desig-
nated direction in their respective reference planes.

4
The autumnal equinox is called “the first point of Libra” because it was located in that con-
stellation in ancient times.
10 Explanatory Supplement to Metric Prediction Generation

But, whereas the ecliptic plane may be determined from observations of Earth
in its orbit, determination of the equator depends on observations of Earth’s
axis of rotation.

Right ascension, or RA, is the angular distance of a celestial body or point on

the celestial sphere, measured eastward from the vernal equinox along the
celestial equator to the hour circle of the body or point. It is the longit udinal
coordinate in the celestial (equatorial) frame.

Right ascension is commonly measured in hours, minutes, and seconds, 24

hours being equivalent to 360º. This convention makes it necessary to distin-
guish right ascension angular measure from regular angular units so as to
avoid confusion. One hour in right ascension is 15 degrees of arc; one min-
ute in right ascension is 15 minutes of arc, or ¼ degree of arc; and one second
in right ascension is 15 seconds of arc.

Hour angle , or HA, is a topocentric measure of the angle, reckoned westward,

from the local meridian to the hour circle passing through a given point on
the celestial sphere. Hour angle, like right ascension, is commonly measured
in hours, minutes, and seconds of time, where 24 hours corresponds to 360
degrees of arc.

The relationship between hour angle and right ascension is one between ter-
restrial timekeeping and cele stial coordinates. Objects rise in the east, transit
the local meridian, and set in the west. Hour angle is essentially the length of
time between the object and the meridian, positive westward. Thus an object
setting due west is at an hour angle of 6 hours, while an object rising due east
is at an hour angle of -6 hours.

In contradistinction, right ascension (RA) increases to the east from the vernal
equinox. The right ascension of a point on the meridian increases continu-
ously with time. Therefore, if the right ascension of an object now on the
meridian is RA, then an object just now rising due east has right ascension
RA + 6 hours, and an object just now setting due west has right ascension RA
– 6 hours. Since Local Sidereal Time (LST) is the right ascension of a point
on the meridia n, the relationship between RA, HA, and LST is

HA = LST − RA (3-3)
If a target’s location at a specified time is specified in right ascension, then it
can be converted to local hour angle at that time by computing the local side-
real time and applying the formula above.
Coordinate and Reference Systems 11

Declination, or d, or DEC, is measured from the equator, positive in the

northerly direction, along an hour circle or circle of right ascension to the
point. Declination is commonly measured in degrees, minutes, and seconds
of arc.

Longitude in the ecliptic frame is termed ecliptic longitude and also celestial
longitude (the latter sometimes being confusing, since it does not refer to the
celestial frame). It is measured in radians or degrees.

Latitude in the celestial frame, measured from the equator, is termed declina-
tion and measured in radians or degrees, positive northward. Ecliptic lati-
tude, measured from the ecliptic, is also termed celestial latitude (again caus-
ing some confusion). It is also measured in radians or degrees.

The vernal equinox is at the ascending node of the ecliptic on the equator. It
is the direction at which the Sun crosses the equator from north to south. The
angle between the two planes is known as the obliquity of the ecliptic. The
equator and ecliptic are moving because of perturbing forces on the rotation
and orbital motion of Earth. For this reason, both the location of the equinox
and the value of obliquity change over time. The vernal equinox currently
resides in the constellation Pisces.

In order to define an inertial frame based on the equinox, it is necessary to

specify the standard epoch of the reference equinox position. Moreover, it is
necessary to state the method by which the reference equinox is reckoned.
The usual specification is the mean equator and equinox at the standard ep-
och, which ignores small variations of short period in the motions of the ce-
lestial equator and includes only precession effects.

The positions of the equator and ecliptic for a standard epoch are not ob-
served directly, but are calculated from catalogued positions and motions of
stars and other celestia l objects that act as reference points in the sky. A
number of such frames have been defined5 , such as those designated as
J2000, FK5, and the DE-4xx series of ephemerides, all of which refer to the
same inertial frame, and those related to B1950, such as FK4 and the frames
of the DE-1xx series of ephemerides, which all slightly differ from one an-
other. Others include the ECLIPB1950, ECLIPJ2000, and GALACTIC
frames.

5
Oddly enough, the ICRF is not one of the frames explicitly listed in SPICE system documen-
tation. However, it is currently the same as J2000.
12 Explanatory Supplement to Metric Prediction Generation

Fortunately, as a standard, JPL ephemerides each contain identification of its

frame of reference. The SPICE library recognizes both inertial and body-
fixed frame identifiers. Its utilities automatically and transparently translate
from the various ephemeris frames to J2000, so that the generation of predic-
tions thereafter proceeds smoothly. The PXFORM and SXFORM routines re-
turn the position and state transformation matrices, respectively, for transla t-
ing between any two recognized frames at a given ephemeris time. The
SPKEZ routines produce state vectors in any requested inertial or body-fixed
frame in a seamless manner.

Transformations between rectangular coordinates and right ascension, decli-

nation, and range coordinates may be made using the RECLAT and LATREC
utilities.

3.2.3 Earth Precession, Nutation, and Polar Motion

The normal to the equatorial plane is the true celestial pole, as defined by the
axis of Earth’s rotation. As viewed in the background of stars and extraga-
lactic radio sources at a given instant of time, its direction is not fixed in
space. Its motion is due to gravitational forces, mainly from the Sun and
Moon, acting on the nonuniform distribution of mass within Earth. Mathe-
matically, this motion is composed of three effects: precession, nutation, and
polar motion.

The term precession refers to the slow, periodic conical motion of the cele s-
tial pole about the ecliptic pole. The component of precession caused by the
Sun and Moon acting on Earth’s equatorial bulge is called lunisolar preces-
sion, and the component caused by the action of the planets is called plane-
tary precession. The sum of the two effects is called general precession.
Precession causes the equinox to move backward along the equator at a rate
of about 50 arcseconds per year. Earth’s rotational axis will return thus to its
present location in about 26,000 years.

While precessing, Earth’s axis wobbles slightly due to shorter term effects of
the Moon’s gravity on Earth’s equatorial bulge. This effect, called nutation,
is a short period motion of the true pole about the mean pole with amplitude
of about 9 arcseconds and a variety of periods up to 18.6 years.

Whereas the effects of precession determine the location of the mean equinox
of date , the added effects of nutation determine the location of the true equi-
nox of date . The difference in right ascension between the true and mean
Coordinate and Reference Systems 13

equinoxes of date is called the equation of the equinoxes, and is the difference
between true and mean sidereal time.

The combination of precession and nutation define a reference axis called the
Celestial Ephemeris Pole (CEP). It is normal to the true equator and is the
axis about which the diurnal rotation of Earth is applied in the transformation
between celestial and terrestrial frames. However, it does not coincide with
instantaneous axis of rotation, which is the Earth’s axis of figure, or axis of
maximum moment of inertia.

The difference between the axis of figure and CEP is known as polar motion.
This motion is affected by unpredictable geophysical forces on the deform-
able Earth, and is determined from observations of stars, radio sources, the
Moon, and Earth satellites, using Very Long Baseline Interferometry (VLBI),
laser ranging, and other applicable techniques. The intersection of the axis of
rotation with the Earth’s surface appears to wander in a quasi-circular motion
around the CEP with maximum amplitude of about 0.3 arcseconds, corre-
sponding to a motion of about 9m on the surface and having principal periods
of about 365 and 428 days.

The obliquity of the ecliptic is known to oscillate between about 22.0º - 24.6º,
with a period of around 41,000 years. Its current value is about 23º26’.

Theories and formulas for precession and nutation, including changes in

obliquity, may be found in the Explanatory Supplement to the Astronomical
Almanac. They are implemented in the Navigation Ancillary Information
Facility (NAIF) Precision Earth Model, discussed under Terrestrial Reference
Systems, below.

Polar motion is unmodeled by SPICE and the MPG; that is, there are no
mathematical models or ephemerides that characterize the motio n. Rather, its
effects are entered into the MPG by means of an ephemeris called the Preci-
sion Earth Model (PEM) which is generated by NAIF special MPG routines
that make use of Earth Orientation Parameter (EOP) files and Universal Time
and Polar Motion (UTPM6 ) files received by the Service Preparation Subsys-
tem (SPS) from the Kalman Earth Orientation Software service, which proc-
esses the EOP data supplied by the IERS. The reader is referred to the refer-

6
Also known as the UTPM STOIC file format, created by the Standby Timing Operations In
Contingencies program. T his format contains information derivable from the EOP and is more
limited in extent. The NSS was the predominant user of STOIC files.
14 Explanatory Supplement to Metric Prediction Generation

enced DSN Interface Agreement for further details. The PEM is an imple-
mentation of ITRF93, discussed later.

Since polar motion is quasi random and cannot be accurately predicted very
far into the future, the MPG requires frequent updates of EOP files in order to
generate accurate products. Even with accurate current EOP files, it is not
possible to guarantee that high accuracy predictions can extend into the more
distant future. MPG predictions are accurate within the accuracy and validity
of its inputs, but may not echo reality in cases where inputs do not represent
reality.

3.2.4 Other Inertial Frames

3.2.4.1 Ephemeris-Defined Frames

Fundamental ephemerides of the positions and velocities of solar system bod-
ies are computed using the basic dynamical equations of motion and fit to
observational data. They are the bases for computing apparent ephemerides,
representational ephemerides, phenomena, orbital elements, and stability
characteristics of entities of interest.

The gravitational model used includes all the known relevant forces (i.e.,
those producing an observable or measurable effect) acting upon and within
the solar system, with relativistic effects approximated to terms includ-
ing 1/ c 2 . The method is further elaborated in the ESAA.

The positions and velocities of objects in the ephemerides are referenced with
respect to inertial space, since the equations of motion are defined with re-
spect to inertial space. Associated with the ephemerides is the set of astro-
nomical constants used in the creation of the ephemerides. These “constants”
are “solved-for” parameters, as they must provide a best-fit of the
ephemerides to the observational data. The constants are associated directly
with the ephemerides and are considered to be an integral part of them.

Since the dynamical equinox is a position defined by motions of the solar

system bodies, it is possible to determine its location using the ephemeris it-
self. The alignment of the ephemeris frame with the J2000 equinox of the
ephemerides then becomes a relatively straightforward, but iterative task.
Since the dynamical equinox depends on the set of observational data used to
correct the ephemeris, the equinox found using an ephemerides based on con-
Coordinate and Reference Systems 15

stants derived from observations of solar system objects is not precisely

equivalent to one based on observations of stars and radio-source catalogs

For this reason, the reference frames of JPL development ephemerides may
vary somewhat from the standard reference frames used in prediction genera-
tion (B1950 for the NSS MP, or J2000 for the MPG), depending on the ob-
servational data applied. The SPICE required reading document on frames
identifies the reference frames associated with JPL development ephemerides
and other ephemeris-derived inertial frames, and also provides the rotation
matrices required to transform states in the ephemeris frames to J2000.

Fortunately for the MPG, SPICE automatically translates recognized ephem-

eris frames to J2000, so these are largely transparent to MPG applications.

3.2.4.2 Star Catalog Frames

The fifth fundamental star catalog, denoted FK5, is a listing of the mean posi-
tion and proper motions at equinox and epoch J2000 for a number of funda-
mental stars to magnitude of about 9.5. A previous version, FK4, which was
referenced to B1950, was in effect prior to the adoption of the FK5 frame. It
differed from B1950 by a rotation of 0.525 arc seconds about the polar axis.

Although referenced to the J2000 frame, FK5 positions differ slightly in opti-
cal and radio source observations by about 1 milliarcsecond. FK5 was used
primarily before the introduction of the ICRS, which on January 1, 1998 su-
perseded it.

3.2.4.3 Time and Standard Epochs

The NSS MP bases its computations on the standard B1950 frame, which
corresponds to the equinox and mean equator of the Besselian epoch JD
2433282.423. A Besselian epoch, named after the German mathematician
and astronomer Friedrich Bessel (1784 – 1846), is an epoch that is based on a
tropical year of 365.242198781 days, measured at the point where Sun’s lon-
gitude is exactly 280°.

Since 1984, however, Besselian epochs have been superseded by Julian ep-
ochs, the current standard epoch being J2000. Besselian and Julian epochs
are related according to:

JulianEpoch - 2415020.31352
BesselEpoch = 1900 + (3-4)
365.242198781
16 Explanatory Supplement to Metric Prediction Generation

Even though the MPG bases its computations on the standard J2000 frame,
nevertheless, since many older star catalogs are still useful, those having
B1950 coordinates cannot just be shelved. However, these older positions
must be corrected for precession and nutation to yield J2000 positions in or-
der that they be used by the MPG. Even though the applicable Interface
Agreement shows a field for epoch of observation (see References, below), it
specifically states that all radio source RA/DEC coordinates are given in the
ICRF frame.

Future versions of prediction generators will probably base their computa-

tions on standard Julian epochs that are spaced 50 years apart. In this way,
the unit of time, viz., seconds past the reference epoch, will stay within preci-
sion bounds of the data type used in programming.

3.3 Terrestrial Reference Systems

The ITRF is an Earth-crust-fixed frame that is geocentric, with its center of
mass being that of the whole Earth, including oceans and atmosphere. Its
scale is that of a local Earth frame, in the meaning of a relativistic theory of
gravitation. Its initial orientation was taken to be the same as that of the BIH
Terrestrial Frame of 1984.0, and its evolution in time is such that there is no
residual global rotation with regards to the crust.

The method of realization of the ITRF and the transformation formulas be-
tween ICRF and ITRF are detailed in the IERS conventions (see references),
and do not appear here. However, the Navigation Ancillary Information Fa-
cility (NAIF) utility I2TERR generates transformation matrices from speci-
fied inertial frames to the ITRF. It also has a utility WRTPEM, which writes
the Precision Earth Model ephemeris used by the MPG. The SPICE system
now also contains a built-in implantation of the ITRF. By defining DSS loca-
tions relative to this frame, SPICE utilities are then able to develop DSS
frames that the MPG can use to generate accurate topocentric predictions.

The particular terrestrial frame implemented is ITRF93, the particular version

of the ITRF published in 1993. Time and polar motion data input to the
NAIF model are consistent with this frame definition.

The SPICE system also implements a terrestrial model labele d IAU_EARTH,

which does not have the accuracy required for MPG pointing and Doppler
predictions. The IAU models are discussed in a later section.
Coordinate and Reference Systems 17

3.3.1 Terrestrial Coordinate Systems

Earth is not spherical, is not flat, and is not uniform in makeup. In fact, it is a
bumpy, somewhat elastic irregular glob of nonuniformly distributed matter,
whose surface features change (slowly) over time. Most terrestrial coordinate
systems for measuring positions near the Earth’s surface use, as their refer-
ence, an abstract surface near that of Earth. There are several common ab-
stract representations of the reference surface, such as the geoid and various
ellipsoids. This section discusses the fundamentals of the various terrestrial
coordinate systems that are used to define the coordinates of an object on or
near Earth’s surface.

3.3.1.1 Pole and Prime Meridian Coordinates

Points on Earth’s surface can be expressed in rectangular coordinates (x, y,
z), in which the z-axis is the rotational axis, the x-axis is in the prime merid-
ian, and the y-axis is 90º eastward from the x-axis. The origin is the center of
Earth. Other coordinate systems that translate to these values are also used,
such as geocentric latitude, geocentric longitude, and distance from the geo-
center, and such as geodetic latitude, geodetic longitude, and height above a
reference surface.

The locations of deep-space stations are contained in the MPG Station Pa-
rameter File, and expressed in cylindrical coordinates, (CU, CV, LO), denot-
ing, respectively, the distance from Earth’s polar axis, the signed distance
above the equator, and the geocentric east longitude. The SPICE function
CYLREC can be used to convert the cylindrical coordinates to rectangular
ones, if needed7 .

3.3.1.2 Geocentric, Geodetic, and Astronomical Coordinates

It is usual practice in some fields to specify locations relative to some useful
surface, as opposed to an origin and axes. Cartography, geodesy, and naviga-
tion, for example, deal with coordinates derived from Earth’s gravitational
field. Points are specified by a location on the reference surface and a height
above it.

Due its non-uniformity, an ellipsoid is often used as the first approximation to

a surface of constant gravitational potential. However, a best-fitting triaxial

7
Besides prediction generators, the MPG also has several stand-alone utilities. One of these,
GPS_Translate, implements a number of such coordinate transformations.
18 Explanatory Supplement to Metric Prediction Generation

ellipsoid does not appear to have significant benefit over an ellipsoid of rota-
tion, but does substantially complicate computations. Therefore, the oblate
spheroid (a rotationally symmetric ellipsoid having its polar axis shorter than
its equatorial diameter) is most often used because it does model the flatten-
ing of Earth at its poles and yet is still a fairly simple mathematical surface.
Such ellipsoids are termed local or terrestrial, depending on whether they
approximate a given localized area or the entire Earth.

Another abstract surface that proves useful when instruments sensitive to the
local gravitational gradient are involved is the geopotential surface, or geop8 .
The geop defines a surface of equal gravity potential, including the effects of
both gravitational and centrifugal forces. The gravity potential is everywhere
perpendicular to the geop.

One geop of particular importance is the geoid , which is a geop whose poten-
tial is that of Earth’s mean sea level. The geoid itself lies close to the refer-
ence ellipse, but differs by an amount called the geoid undulation (N), which
is the height of the geoid above the ellipsoid.

An observer at a given point on Earth’s surface lies on a geop above the ge-
oid. The height of the observer above the ellipse is the ellipsoidal height (h),
and the height of the observer above the geoid is called mean sea level height
or orthometric height (H). These are related approximately by

N =h −H (3-5)
This is only approximate because H is actually measured along the curved
normal or plumb line between the observer and geoid.

An observer’s longitude and latitude can be measured in a number of ways.

If measured with respect to the geocenter, they are said to be geocentric . If
measured with respect to a reference ellipsoid, they are said to be geodetic .
And, if measured with respect to the observer’s local vertical, they are said to
be astrometric. The reader should not confuse the term geodetic here as hav-
ing to do with the geoid.

Geocentric latitude is the angle as viewed at the center of the reference ellip-
soid between the ellipsoid equator and the observer, positive in the northerly
direction. Geodetic latitude is the angle between the ellipsoid equator and the

8
The pronunciation of this term is not found in literature available to the authors. The pre-
sumption is that, since it refers to Earth, it would be pronounced GEO-P. However, the word
geoid does not follow this rule, so the pronunciation could well be GEE-OP.
Coordinate and Reference Systems 19

normal to the ellipsoid at the observer, also positive northward. Astronomi-

cal latitude is the angle between the Earth’s equator and the local vertical.
The reader will note that the line of the local vertical does not necessarily
pass through the Earth’s center or even intersect the polar axis.

The angle between the local vertical and normal to the ellipsoid at the ob-
server is known as the deflection of the vertical. This angle is sometimes ex-
pressed as the combination of a meridian increment and an orthogonal incre-
ment along the meridian, positive northward. It is unused within SPICE and
the MPG.

Geocentric longitude is the angle between the reference (prime) meridian and
the observer’s meridian. Geodetic longitude is the same as geocentric longi-
tude, under the usual assumption that the reference axes and prime meridians
are chosen to be the same. The astronomical longitude is somewhat more
complicated; it is the angle between the prime meridian plane and the plane
perpendicular to the equator that includes the observer and observer zenith
vector.

If a and b are, respectively, the semimajor and semiminor axes of the refer-
ence ellipsoid, the flattening factor f is defined as

a −b
f = (1.6)
a
The height associated with geocentric measure is the distance between the
observer and geocenter, the height associated with geodetic measure is the
ellipsoidal height, and the height associated with astronomical measure is the
orthometric height.

A geodetic frame of reference is referred to as a datum. Geodetic datums

define the size and shape of Earth and the origin and orientation of the coor-
dinate systems used to map it. Hundreds of different datums have been used
to frame position descriptions since the first estimates of Earth's size were
made by Aristotle. Datums have evolved from those describing a spherical
Earth to ellipsoidal models derived from years of satellite measurements.

One such datum is the World Geodetic System of 1984 (WGS84) , last re-
vised in 2004) that is used by the Global Positioning Satellite (GPS) system.
The coordinate system is essentially that of the ITRF and the reference ellip-
soid is one having
20 Explanatory Supplement to Metric Prediction Generation

a = 6378.137km
(3-7)
1/ f = 298.257223563
It is usual to find the flattening factor in geodetic models expressed as an in-
verse, because the resulting figure has an integer part that, to practitioners,
appears to be a more meaningful representation. The flattening value is actu-
ally

f = 0.0033528106647475 (3-8)
Datums also specify an Earth Gravitational Model (EGM), which is typically
a spherical harmonic series of the gravitational potential from which a geoid
undulation model is derived. The EGM96 series, for example, was of order
and degree 360. The MPG and SPICE do not use gravity models, and hence
do not compute heights relative to the geoid. Heights above the reference
ellipsoid, however, are used.

The ESAA contains formulas for geocentric -geodetic transformations. These

are not particularly enlightening and are omitted here. SPICE contains rou-
tines GEOREC and RECGEO that perform these functions in an equivalent, but
seemingly more elegant manner. The interested reader is referred to the
commentary in the source code for further details.

The conversion from a given geodetic longitude, geodetic latitude, and height
output (? f , h) of a GPS (WGS84) receiver to geocentric coordinates (x, y,
z) is found directly using GEOREC along with the Earth radius and flattening
given above. The conversion of this point (x, y, z) into cylindrical coordi-
nates is made using RECCYL, which outputs the values (R, LONG, Z), which,
in the parameters of the Station Parameter File, are (CU, LO, CV).

3.3.2 Topocentric Coordinates

The SPICE convention for topocentric coordinates is the north-west-zenith
frame, and the components are denoted (n, w, z). The MPG observes topo-
centric rectangular coordinates as north-west-zenith components, as well.
The SPICE frame transformation utilities, such as SXFORM and PXFORM,
and state vector functions such as SPKEZ, all observe this convention. These
utilities rely on geodetic coordinates of tracking stations provided in the
Topocentric Reference Frames file and Station Ephemeris File. Both these
files are generated from the user-accessible Station Parameter file.
Coordinate and Reference Systems 21

The PX6 prediction data type lists direction cosines in the order (z, e, n). For
this reason, the Network Support Subsystem Metric Prediction (NSS MP)
application used this convention internally. This notation forms a right-
handed coordinate system, but is not a rotation of the SPICE and MPG frame.
The generation of PX6 direction cosines in the MPG requires the transposi-
tion

(cos θ N ,cos θW ,cos θ Z ) → (cos θ Z , − cos θW ,cos θN ) (3-9)

to provide the correct output format.

It should be noted that, while the NSS MP used the (z, e, n) frame conven-
tion, the JPL Orbit Determination Program applies a north-east-zenith con-
vention for topocentric pointing vectors. The (n, e, z) frame, however, is a
left-handed coordinate system. Right-handed coordinate conventions incor-
porating east are (e, n, z), (n, z, e), and (z, e, n).

3.3.2.1 Antenna Coordinate Systems

DSN DSS antennas are categorized depending on their construction. The
principal differences among the various types of antennas are aperture, wave
optics, and mount. The coordinate system of an antenna is defined by the
orientations of its axes of motion and measures of rotation about each axis.
The mounts are classified by the type of coordinate system most natural to
the motion axes. There are hour-angle/declination (also called equatorial or
HA-DEC), azimuth/elevation, and XY and X’Y’ mounts. Transformations of
pointing angles in the coordinate systems used by these mounts are discussed
below. These transformations do not include the effects of atmospheric re-
fraction and mount-specific range corrections , which are discussed in other
chapters of this Supplement.

A keyhole is an area in the sky where an antenna cannot track a target be-
cause the required slew rate along one of its axes of motion would be too
high, or because mechanical limitations prevent the antenna from pointing in
that direction. Keyholes impose restrictions on an antenna’s ability to serve
its function. As will be seen, each mount type has differing keyhole con-
straints.

Another constraint on antennas is restricted pointing areas. These are some-

times not physical limitations, but ones imposed by safety or security. Hor i-
zon masks define the limits of reception from a target, while high-power
transmitter masks define areas in which it is safe to radiate to a distant target.
22 Explanatory Supplement to Metric Prediction Generation

3.3.2.2 Hour Angle/Declination Coordinates

Antennas with so-called equatorial or HA-DEC mounts have one axis parallel
to Earth’s polar axis and the other in a latitudinal direction that slews above
and below the equator. With proper calibration and the application of the
ITRS model, antenna coordinates may be related to those deriving from the
true pole, prime meridian, and equinox of date. Elongations relative to the
polar axis are thus hour angle , and those relative to the equatorial axis are
declination.

Some of the older DSS antennas were equatorial mounts. To track an inter-
planetary spacecraft, the antenna would point to the spacecraft's known coor-
dinates, and then, for the rest of the tracking period, it would simply rotate in
hour angle about the tilted axis as the Earth turns.

While being a fine type of mount for a small instrument, it proved to be very
unsuitable for large, heavy structures because the tilted polar bearing has to
sustain large asymmetric loads. These loads include not only the whole re-
flector dish and its associated wavefront focusing apparatus, but also an HA
counterweight heavy enough to balance the antenna, the DEC bearing, and its
DEC counterweight. Also the structure has to be designed specifically for its
location, since the polar bearing's angle depends on the station's latitude.

The DSN built such antennas at Goldstone, Madrid, and Canberra, but these
are no longer in service. The Goldstone site has been converted to a radio
telescope dedicated to educational use. The discussion here is included for
historical purposes, because hour angle and declination coordinates are still
used in pointing optical instruments, and because right ascension and declina-
tion still appear in star catalogs.

If the geocentric position of the DSS antenna, designated here as DSS, is

( xDSS , y DSS , z DSS ) and that of a target in space is ( xtgt , ytgt , ztgt ) , then the
declination and hour angle are given by
Coordinate and Reference Systems 23

p = ( xtgt − xDSS , ytgt − yDSS , ztgt − zDSS )

p
u=
| p|
λtgt = arctan( ytgt , xtgt ) (3-10)
λDSS = arctan( y DSS , xDSS )
δ = arcsin( uz )
HA = λDSS − λtgt

where λ denotes the longitude of the subscripted object. The arctangent

function appearing here computes the quadrant-corrected arctan( y / x) . The
longitude of the station, of course, is fixed, and only needs to be computed
once. The body-fixed direction to the target, however, changes continually,
and must be reevaluated for each computation of hour angle and declination.

The inverse formulas are

ux = cos(δ )cos( λDSS − HA)

u y = cos(δ )sin( λDSS − HA) (3-11)
uz = sin(δ )
Conversion of a station location and pointing vector in rectangular coordi-
nates to hour angle and declination may use the SPICE RECLAT function,
which returns the elevation as the declination argument and the hour angle as
the difference in longitude between the DSS and target. The inverse conver-
sion may use LATREC, observing the same conventions.

For a HA-DEC antenna, the keyhole is large, in the northern hemisphere, and
centered near the North Star. To track through that area the antenna would
have to whip around prohib itively fast in hour angle.

3.3.2.3 Azimuth-Elevation Coordinates

Antennas with so-called azimuth-elevation or AZ-EL mounts have one axis
along the local vertical so that they turn in the horizontal plane. The other
axis lifts toward zenith. This design permits mechanical loads to be symmet-
ric, resulting in less cumbersome, less expensive hardware that is easier to
maintain. Most of the DSN antennas are of this type.

Such mounts locate a point in the sky by elevation (el) in degrees above the
horizon, and azimuth (az) in degrees clockwise (eastward) from true north.
24 Explanatory Supplement to Metric Prediction Generation

Given a topocentric unit vector (n, w, z) in the SPICE north-west-zenith coor-

dinate convention, azimuth and elevation computations follow the familiar
formulas

az = arctan( −w, n )
(3-12)
el = arcsin( z )
The arctangent function appearing here computes the quadrant-corrected
arctan( −w / n) . The inverse formulas are

n = cos(az )cos(el )
w = − sin( az )cos( el ) (3-13)
z = sin( el )
Conversion of a pointing vector in rectangular coordinates to azimuth and
elevation may use the SPICE RECLAT function, which returns the elevation
as the latitude argument and the negative of the azimuth as the longitude.
The inverse conversion may use LATREC, observing the same conventions.

Such mounts have keyholes at zenith and cannot generally dip below a min i-
mum elevation limit. If a spacecraft were to pass directly overhead, the AZ-
EL antenna would rise in elevation until it reached its straight-up maximum
near 90°. But then the antenna would have to whip around rapidly in azimuth
as the spacecraft is first on the one side of the antenna, and then, a moment
later, is on the other. When the antenna slew rate is not fast enough to track
through this region, there will be an interruption in tracking until the link can
be reacquired on the other side.

Certain AZ-EL antennas9 require long cables that connect the dish electronics
to the ground electronics. As the antenna tracks its target, the cable between
the pedestal and alidade wraps around until its limit in that direction, clock-
wise or counterclockwise, is reached. The antennas are designed so that the
entire traverse from counter-clockwise limit to clockwise limit is greater than
360°. Cable wrap limitations restrict the time-on-target that can be achieved
for some missions. For this reason, the MPG antenna pointing predictions
include cable -wrap status (pointing) and warnings (view period events).

9
Antennas with beam waveguide technology avoid the use of such cables by the addition of
five precision radio frequency mirrors that reflect signals along a beam waveguide tube from
the vertex of the antenna to the equipment in the antenna pedestal. However, there remains a
power cable connecting the alidade and pedestal to which the wrap limits apply.
Coordinate and Reference Systems 25

3.3.2.4 X-Y and X’-Y’ Coordinates

Two other mounting schemes for antennas are the X-Y and X’-Y’ orienta-
tions. Like AZ-EL, and HA-DEC antennas, these mounts also have two per-
pendicular axes. They are mechanically similar to the HA-DEC antenna, but
have their "p olar" axes oriented horizontally, and not necessarily aligned to a
cardinal direction.

An antenna having a so-called X-Y mount has one of its rotational axes fixed
and oriented horizontally toward north. The angle X is then measured from
zenith positive eastward. The other axis rotates perpendic ularly to the first,
so that the angle Y is measured as the northern elongation. Such mounts
have keyholes at the northern and southern horizons. The conversion from
(n, w, z) to (X, Y) is

n = sin(Y )
w = − sin( X )cos(Y ) (3-14)
z = cos( X )cos(Y )
and the inverse mapping is

X = arctan( − w / z )
(3-15)
Y = arcsin(n )
These formulas are of the same form as those for azimuth and elevation under
the mapping n®z, w®w, z®n, Y®el, X®az.

An antenna having a so-called X’-Y’ mount has one of its rotational axes
fixed and oriented horizontally toward east. The angle X’ is then measured
from zenith positive southward. The other axis rotates perpendic ularly to the
first, so that the angle Y’ is measured as the eastern elongation. Such mounts
have keyholes at the eastern and western horizons. The conversion from (n,
w, z) to (X’, Y’) is

n = − sin( X ')cos(Y ')

w = − sin(Y ') (3-16)
z = cos( X ')cos(Y ')
and the inverse mapping is

X ' = arctan( −n / z )
(3-17)
Y ' = − arcsin( w)
26 Explanatory Supplement to Metric Prediction Generation

Again, these formulas are of the same form as those for azimuth and eleva-
tion under the mapping w®z, n®w, z®n, Y’®–el, X’®az.

These two antenna types have the advantage over HA-DEC and AZ-EL an-
tennas in that they can rotate freely in any direction from the upward-looking
zenith central position without any cable wrap-up issues. Another advantage
is the location of keyholes. These antennas are oriented so that its keyholes
are either at the northern and southern horizons, or at the eastern and western
horizons. This leaves the whole sky open for tracking spacecraft without re-
quiring high angular rates around either axis. Such mounts were first built for
tracking Earth-orbiting spacecraft, which may require high angular rates and
overhead passes.

The DSN is currently equipped with three X’-Y’, 26-m aperture antennas
(DSS-16, DSS-46, DSS-66), used primarily in tracking Earth orbiters, which
usually have inclinations that avoid the east and west keyholes.

Of course, the 26-m antennas can also be used with interplanetary spacecraft.
Such spacecraft do not typically pass overhead, but rather stay near the eclip-
tic plane in most cases, and may occasionally pass through a keyhole at rise
or set. However, since their apertures are smaller than most other DSN sta-
tions, they are not deemed effective for most interplanetary craft.

3.4 Planetary Reference Systems and Frames

At regular intervals the IAU revises and publishes tables giving the directions
of the north poles of rotation and the prime meridians of the planets, satel-
lites, and asteroids. It also publishes revised tables giving their sizes and
shapes, along with a summary of changes since the previous report.

These planetary body-fixed coordinate frames are defined relative to their

mean axes of rotation and various definitions of longitude, depending on the
body. The longitude systems of most of those bodies with observable rigid
surfaces have been defined by references to a surface feature, such as a crater.
These are generally simplified models in which the models of rotation, pre-
cession, and nutation are less complex than that found in the ITRF. The
ICRF is the reference coordinate frame, with variable quantities expressed in
units of days (86400 SI seconds) or Julian centuries (36525 days).

The SPICE system incorporates the IAU tables in defining reference frames
for these objects with respect to the J2000 inertial coordinate system. The
Coordinate and Reference Systems 27

names of these frames are of the form IAU_object. For example,

IAU_EARTH is the IAU Earth frame.

The MPG uses IAU frames, together with IAU estimates of body radii, in
computing occultation and eclipse predictions, where bodies are treated as
triaxial ellipsoids.

3.5 DSN and NAIF Frame Identification

The MPG recognizes two frame identification designations. The DSN ID
system is used in user interfaces, while the NAIF ID system is used within
the MPF and SPICE. The two bear similarities, but are also very dissimilar in
many aspects. The two are different partly for historical reasons and for con-
sistency with NSS identifiers.

Correspondences between the two are defined in the

DSN_NAIF_Conv.txt file used by the MPG.

Basically, in the DSN ID system, planet barycenters are designated by single

digits 0-9. Sun is 10, Earth is 300, Moon is 301, spacecraft are negative and
deep space stations are positive integers less than 100.

SPICE files and subroutines internally refer to ephemeris objects, reference

frames, and instruments by integer codes. The codes and conventions may be
found in the SPICE required reading document NAIF_IDS. Spacecraft iden-
tifiers are negative and are usually assigned by a NASA control authority.
This authority, at times, is forced into reusing some IDs, which can affect the
way the SPICE system handles these codes.

As for NAIF numeric identifiers, the smallest posit ive codes are reserved for
the solar system barycenter (0) and Sun (10). The nine integers in between
refer to planet system barycenters in order of distance from Sun. Integers of
the form PNN, where P is one of the planet indices and NN is the number of a
satellite, refer to planets and satellites; a planet center is considered to be the
99th satellite of its barycenter. Thus, 399 refers to the geocenter.

Spacecraft are assigned negative integer codes. The code assigned to an in-
terplanetary spacecraft is normally the negative of the code assigned to the
same spacecraft by the DSN, as determined by the NASA control authority at
GSFC.
28 Explanatory Supplement to Metric Prediction Generation

Earth orbiters not having a DSN identification code are assigned numbers
equal to -100000-US_Space_Command_code.

The NAIF identification codes for tracking stations and landed spacecraft are
of the form P99NNN, where P is the planet number, and NNN is the station or
target number.

NAIF also has identification techniques for spacecraft instruments, comets,

and asteroids.

SPICE subroutines permit textual names to be assigned to these integer

codes. More than one name may be assigned to any particular numeric ID,
and these names may be used to refer to that ID. However, if one asks for the
name of a given ID, only the last-entered assignment is retrieved.

As the SPICE system has expanded, so has the number of objects requiring
identifying codes. Many of these objects do not fit neatly into the schemes
originally envisioned as needing ID codes. Needless to say, the particular set
of ID codes recognized by SPICE now shows the wear that results from an
expanding system.

The definition of frames corresponding to NAIF IDs is explained in the

FRAMES required reading. Basically, there are a number of built-in frames,
augmented by the means to define other frames, given the information by
which to do so.

The built-in inertial frames include the Earth mean equator and equinox of
J2000 (‘J2000’), the mean ecliptic and equinox of J2000
(‘ECLIPJ2000’), the Galactic System II frame (‘GALACTIC’), and the
Mars mean equator and IAU vector of J2000 frame (‘MARSIAU’). The
built-in body-fixed frames include those based on IAU rotation models, such
as ‘IAU_EARTH’ and ‘IAU_MARS’, as well as the high-precision Earth
rotation model ‘ITRF93’.

SPICE also provides the means to define other reference frames using pa-
rameters and mathematical models defined in text or binary files.

DSS frames are defined by a combination of antenna location information

found in the station parameters file and parameters in a frames kernel for
generating the rotation matrix from the Earth model (ITRF93) to topocentric
coordinates.
Coordinate and Reference Systems 29

The interested reader is directed to the SPICE required reading files for fur-
ther information.
30 Explanatory Supplement to Metric Prediction Generation

References
Acton, C. H., et al., “Reference Frames”, SPICE Required Reading, Naviga-
tion Ancillary Information Facility, Jet Propulsion Laboratory, Pasadena, CA,
2005 (rev).

Acton, C. H., et al., “Rotations”, SPICE Required Reading, Navigation Ancil-

lary Information Facility, Jet Propulsion Laboratory, Pasadena, CA, 2005
(rev).

Acton, C. H., et al., “Most Useful SPICELIB Subroutines”, SPICE Required

Reading, Navigation Ancillary Information Facility, Jet Propulsion Labora-
tory, Pasadena, CA, 2005 (rev).

Acton, C. H., et al., “NAIF Integer ID Codes”, SPICE Required Reading,

Navigation Ancillary Information Facility, Jet Propulsion Laboratory, Pasa-
dena, CA, 2005 (rev).

Connally, Michael, “VLBI System Interfaces, Radio Source Position Catalog

Interface”, Deep Space Mission System Subsystem Software Interface Speci-
fication 810-13 VLB-13-4, Jet Propulsion Laboratory, Pasadena, CA, Janu-
ary, 2006.

McCarthy, Dennis D., ed., IERS Standards (1992), IERS Technical Note 13,
Observatoire de Paris, Paris; July, 1992.

McCarthy, Dennis D., ed., IERS Conventions (1996), IERS Technical Note
21, U. S. Naval Observatory; July, 1996.

Seidelman, P. Kenneth, ed., Explanatory Supplement to the Astronomical

Almanac, University Science Books, Mill Valley, California.

Seidelman, P. Kenneth (chair), “Report of the IAU/IAG Working Group on

Cartographic Coordinates and Rotational Elements of the Pla nets and Satel-
lites: 2003”, Celestial Mechanics and Dynamical Astronomy, Volume 91,
Numbers 3-4, March 2005, pp. 203-215(13).

Walther, J. Y., “Kalman Earth Orientation Filter (KEO) and Metric Predic-
tions Generation (MPG) Interface”, Deep Space Mission System Subsystem
Software Interface Specification 810-16 3010-KEO-MPG, Jet Propulsion
Laboratory, Pasadena, CA, July, 2001.