GEODESY

Lecture Notes

2008.

Budapest University of Technology and Economics

Compiled by Gy. Tóth from W. Torge: Geodesy, 2nd Edition

 Contents

1. Introduction
1.1. Definition and Classification of Geodesy
1.2. The Problem of Geodesy
1.3. Historical Development of Geodesy
1.3.1. The Spherical Earth Model
1.3.2. The Ellipsoidal Earth Model
1.3.3. Arc Measurements
1.3.4. The Geoid and the Ellipsoid

2. The Gravity Field of the Earth

2.1. Components of the Gravity Field
2.1.1. Gravitation. Gravitational Potential
2.1.2. Gravitational Potential of a Spherically Symmetric Earth
2.1.3. Properties of the Gravitational Potential
2.1.4. Centrifugal Acceleration, Centrifugal Potential
2.1.5. Gravity Acceleration, Gravity Potential
2.2. Level Surfaces and Plumb Lines
2.2.1. Definition and Properties of Level Surfaces
2.2.2. Analytical Representation of Level Surfaces
2.2.3. Curvature of the Level Surfaces
2.2.4. Curvature of Plumb Lines
2.2.5. Gravity Gradient
2.3. Spherical Harmonic Expansion of the Gravitational Potential
2.3.1. Expansion of the Reciprocal Distance
2.3.2. Expansion of the Gravitational Potential
2.3.3. The Geometrical Meaning of the Surface Harmonics
2.3.4. Physical Meaning of the Lower Degree Harmonic Coefficients
2.4. Temporal Variations of the Gravity Field
2.4.1. Tidal Acceleration, Tidal Potential
2.4.2. Earth Tides
2.4.3. Other Temporal Variations of the Gravity Field

3. Geodetic Reference Systems

3.1. Global Spatial Cartesian System, Polar Motion
3.2. Coordinate Systems in the Earth’s Gravity Field
3.2.1. Global Astronomic System
3.2.2. Local Astronomic Systems, Computations in the Earth’s Gravity Field
3.2.3. Local Astronomic and Global Cartesian System
3.3. The Geoid as Reference Surface for Heights
3.3.1. Definition of the Geoid
3.3.2. Geopotential Number and Orthometric Height
3.3.3. Mean Sea Level
3.4. Ellipsoidal Reference Systems
3.4.1. Geometric Parameters and Coordinate Systems of the Rotational Ellipsoid
3.4.2. Curvature of the Rotational Ellipsoid
3.4.3. Spatial Ellipsoidal Coordinate System
3.5. The Normal Gravity Field
3.5.1. The Normal Figure of the Earth, Level Ellipsoid
3.5.2. The Normal Gravity Field of the Level Ellipsoid

2
 3.5.3. Series Expansions in the Normal Gravity Field
3.5.4. The Triaxial Ellipsoid
3.5.5. Geodetic Reference Systems
3.5.6. Normal Geographic Coordinates, Normal Heights

4. Methods of Measurement in Geodesy

4.1. Astronomic Measurements
4.1.1. Coordinate Systems of Spherical Astronomy
4.1.2. Variation of Stellar Coordinates, Star Catalogues
4.1.3. Time Systems
4.1.4. Observational Instruments
4.1.5. Methods to Determine Astronomic Positions, Azimuth, and Time
4.2. Gravity Measurements
4.2.1. Absolute Gravity Measurements
4.2.2. Relative Gravity Measurements
4.2.3. Gravity Measurements on the Ocean and in the Air
4.2.4. Gravity Reference Systems
4.2.5. Determination of the Gravity Gradient
4.2.6. The Measurement of Earth Tides
4.3. Terrestrial Geodetic Measurements
4.3.1. Atmospheric Refraction
4.3.2. Horizontal Angle Measurements
4.3.3. Distance Measurements
4.3.4. Zenith Angle Measurements, Trigonometrically Determined Heights
4.3.5. Leveling
4.4. Satellite Observations
4.4.1. The Unperturbed Motion of a Satellite
4.4.2. The Perturbed Motion of the Satellite
4.4.3. Artificial Earth Satellites, Time Measuring Systems
4.4.4. Direction Measurements, Early Distance Measurements
4.4.5. Laser Distance Measurements
4.4.6. Doppler Frequency Shift Measurements
4.4.7. Global Positioning System
4.4.8. Very Long Baseline Interferometry
4.4.9. Satellite Altimetry

5. Methods of Evaluation, Global Geodesy

5.1. Astrogeodetic Methods
5.1.1. Deflection of the Vertical, Geoid Undulation, Height Anomaly
5.1.2. Three-Dimensional Computations, Geodetic Datum
5.1.3. Geometric-astronomic Leveling
5.1.4. Positional Systems
5.1.5. Height Systems
5.1.6. Astrogeodetic Determination of the Geoid and Quasigeoid
5.1.7. Orientation of Astrogeodetic Systems, Best Fitting Ellipsoids
5.2. Gravimetric Methods
5.2.1. Geodetic Boundary-Value Problem, Disturbing Potential, Gravity Anomaly
5.2.2. Linearization of the Geodetic Boundary-Value Problem
5.2.3. Solution of the Geodetic Boundary-Value Problem for the Geoid
5.2.4. Gravity Reductions, Cogeoid
5.2.5. Solution of the Geodetic Boundary·Va1ue Problem for the Surface of the Earth
5.2.6. Physical Surface of the Earth and External Gravity Field

3
 5.2.7. Prediction of Gravity Anomalies
5.2.8. Potential, Mass and Position of the Gravimetric Reference Ellipsoid
5.3. Methods of Satellite Geodesy
5.3.1. Observation Equations
5.3.2. Geometric Method
5.3.3. Dynamic Method, Harmonic Coefficients
5.3.4. Orbital Method, Absolute and Relative Positioning
5.3.5. Analysis of Satellite Altimetry
5.4. Combined Methods of Evaluation
5.4.1. Astrogravimetric Leveling
5.4.2. Remove - Restore Methods in Gravity Field Determination
5.4.3. Least Squares Collocation
5.4.4. Earth Models (Gravity Field and Geocentric Coordinates)
5.4.5. Centering and Strengthening Astrogeodetic Systems
5.4.6. Optimal Earth Parameters
5.5. Structure and Dynamics of the Earth’s Body
5.5.1. The Radial Structure of the Earth
5.5.2. The Earth as a Body in Equilibrium
5.5.3. Interpretation of the Gravity Field, the Crust of the Earth, and the Upper Mantle
5.5.4. Isostasy
5.5.5. Geodesy and Geodynamics
5.6. Lunar and Planetary Geodesy
5.6.1. Lunar Geodesy (Selenodesy)
5.6.2. Planetary Geodesy

6. Geodetic Networks
6.1. Horizontal Control Networks
6.1.1. Design, Monumentation, Observations
6.1.2. Computations
6.1.3. Geodesics on the Rotational Ellipsoid
6.1.4. Solution of Ellipsoidal Triangles
6.1.5. Direct and Inverse Geodetic Problems
6.1.6. National Networks: Examples
6.1.7. Network Unification and Large-Scale Control
6.2. Vertical Control Networks
6.2.1. Design, Monumentation, Observations
6.2.2. Computations
6.2.3. National Networks: Examples
6.2.4. Network Unification and Large·Sca1e Control
6.3. Gravity Networks
6.3.1. Design, Monumentation, Observations, and Computations
6.3.2. National Networks: Examples
6.4. References
6.5. Index

The references cited in the text that are marked with A are found in section A of the bibliography – text books,
manuals, and symposia proceeding. References without special marking belong to the second section B –
individual publications – of the bibliography.

4
1. Introduction

1.1. Definition and Classification of Geodesy

According to the classical definition of F. R. HELMERT (A1880), geodesy (γη = earth, δαίω =
I divide) is the “science of the measurement and mapping of the earth’s surface.” This
definition has to this day retained its validity; it includes the determination of the earth‘s
external gravity field, as well as the surface of the ocean floor. With this definition, which has
to be extended to include temporal variations of the earth and its gravity field, geodesy may
be included in the geosciences, and also in the engineering sciences, e.g. NAT. ACAD.
SCIENCES (1978).
Triggered by the development of space exploration, geodesy turned in collaboration with
other sciences toward the determination of the surfaces of other celestial bodies (moon, other
planets). The corresponding disciplines are called selenodesy and planetary geodesy (BILLS
and SYNNOT 1987).
Geodesy may be divided into the areas of global geodesy, national geodetic surveys, and
plane surveying. Global geodesy is responsible for the determination of the figure of the earth
and of the external gravity field. A geodetic survey establishes the fundamentals for the
determination of the surface and gravity field of a country. This is realized by coordinates and
gravity values of a sufficiently large number of control points, arranged in geodetic and
gravimetric networks. In this fundamental work, curvature and gravity field of the earth must
be considered. In plane surveying (topographic surveying, cadastral surveying, engineering
surveying), the details of the terrain are obtained. As a reference surface for horizontal
positioning the ellipsoid is used in geodetic surveying. In plane surveying, the horizontal
plane is generally sufficient.
There is close interaction between global geodesy, geodetic surveying and plane surveying. The geodetic survey
adopts the parameters determined by measurements of earth, and its own results are available to those who
measure the earth. The plane surveys, in turn, are generally tied to the control points of the geodetic surveys and
serve then particularly in the development of national map series and in the formation of real estate cadastres.
Measurement and evaluation methods are largely identical in global geodesy and national geodetic surveys.
Particularly space methods (satellite geodesy) enter more and more into regional and even local surveys. This
also implies more detailed gravity field determination on regional and local scale.

With the corresponding classifications in the realms of the English and French languages, the
concept of “geodesy” (la géodésie, “höhere Geodäsie” after Helmert) is to be referred only to
global geodesy and geodetic surveying. The concept of “surveying” (la topométrie,
Vermessungskunde or “niedere Geodäsie” after Helmert) shall encompass plane surveying.
In this volume, geodesy is treated only in the more restricted sense as explained above. An
introduction to plane surveying is given by KAHMEN and FAIG (A1988).

1.2. The Problem of Geodesy

The problem of geodesy, generated from and partially supplementing Helmert’s definition,
may be described comprehensively as follows (DRAHEIM 1971, FISCHER 1975):
“The problem of geodesy is to determine the figure and the external gravity field of the earth
and of other celestial bodies as functions of time; as well as, to determine the mean earth
ellipsoid from parameters observed on and exterior to the earth‘s surface.”

1
This geodetic boundary value problem incorporates a geometric (figure of the earth) and a
physical (gravity field) formulation of the problem; both are closely related.
By the figure of the earth we mean the physical and the mathematical surface of the earth.
The physical surface of the earth is the border between the solid or fluid masses and the
atmosphere. Recently, the ocean floor has also been included in the formulation of the
geodetic problem, being the bounding surface between the solid terrestrial body and the
oceanic water masses. The extension of the problem to the oceans is designated marine
geodesy (MOURAD) 1977, SEEBER 1975). The irregular surface of the solid earth (continents
and ocean floor) is incapable of being represented by a simple mathematical relation; it is
therefore described point wise by the use of coordinates of the control points. On the other
hand, the ocean surfaces (70% of the earth‘s surface) possess a simpler principle of
formation. Under certain assumptions, they form a part of a level (equipotential) surface
(surface of constant gravity potential) of the earth‘s gravity field. We may think of this
surface as being extended under the continents and then identify it as the mathematical figure
of the earth (HELMERT A1880/1884). J. B. LISTING (1873) designates this level surface as
geoid.
C. F. Gauss had already referred to this surface: “What we call the surface of the earth in the geometrical sense
is nothing more than that surface which intersects everywhere the direction of gravity at right angles, and part of
which coincides with the surface of the oceans.” (C. F. Gauss: “Bestimmung des Breitenunterschiedes zwischen
den Sternwarten von Göttingen und Altona,” Göttingen 1828), see also MORITZ (1977).
The majority ofthe observed parameters used in geodesy refers to the earth‘s external gravity
field, whose study thereby becomes a concern of geodesy. The upper limit of space that is of
interest is governed by the geodetic usage of artificial satellites and space probes, as well as
the earth’s moon. The physical aspect of the problem of geodesy follows from the
consideration of the earth‘s surface and the geoid as bounding surfaces in the earth‘s gravity
field, The external gravity field may be described by the infinite number of level surfaces
extending completely or partially exterior to the earth‘s surface.

Fig.1.1 Earth`s surface and

reference surfaces

Reference systems are introduced in order to describe the motion of the earth in space
(celestial system), and surface geometry and gravity field of the earth (terrestrial system). For
global geodesy, the use of three-dimensional Cartesian coordinates in Euclidian space is
adequate. ln geodetic surveying, a reference surface is introduced in order to distinguish
curvilinear surface coordinates and heights. Because of its simple equation, a rotational
ellipsoid flattened at the poles is better suited as such a reference surface than the geoid,
which is determined by the uneven distribution of the earth‘s masses. Particular significance
is given to the mean earth ellipsoid, which is the optimal ellipsoid approximating the geoid.
Because of its physical meaning, the geoid is well suited as reference surface for heights. Fig.
1.1 shows the mutual arrangement of the surfaces to be determined in geodesy.
The body of the earth and its gravity field are subject to temporal variations of secular,
periodic, and abrupt nature, which can occur globally, regionally, and locally. The geodetic
measurement and evaluation techniques today have advanced to the extent that they can detect
a part of this change. Should average conditions be ascertained, observations must be
corrected for these changes. With the detection of a part of the variations, geodesy also

2
contributes to the investigation of the dynamics of the terrestrial body. The figure of the earth
and the external gravity field are accordingly conceived as time dependent variables. This
leads to the ideas of “four-dimensional geodesy”(ANGUS-LEPPAN 1973, MATHER 1973).

1.3. Historical Development of Geodesy

The formulation of the problem of geodesy expressed in [1.2] first developed in the course of the nineteenth
century. However, the question of the figure of the earth had already been raised in antiquity. After the sphere
first served as a model for the earth, the oblate rotational ellipsoid as figure ofthe earth asserted itself in the first
half ofthe eighteenth century; cf. FISCHER ( 1975), BIALAS (A1982), LEVALLOIS (A1988).

1.3.1. The Spherical Earth Model

Various opinions on the form of the earth prevailed in the past; e.g. the notion of an earth disk
encircled by Oceanus (Homer’s Illiad, ~800 B.C., Thales of Milet, ~600 B.C.). Pythagoras ( ~
580-500 B.C.) and his school, as well as Aristotle (384-322 B.C.), among others, expressed
themselves for the spherical shape.
The founder of scientific geodesy is Eratosthenes (276-195 B.C.) of Alexandria, who under
the assumption of a spherical earth deduced from measurements a radius for the earth
(SCHWARZ 1975). The principle of the arc measurement method developed by him was still
applied in modern ages: From geodetic measurement, the length ∆G of a meridian arc is
determined; astronomical observations furnish the associated central angle γ (Fig. 1.2). The
radius of the earth is then given by
∆G
R= . (1.1)
γ

Fig.1.2 Arc measurement of

Eratosthenes

Erathosthenes found that at the time of the summer solstice, the rays of the sun descended vertically into a well
in Syene (Assuan, today); whereas in Alexandria, roughly on the same meridian, they formed an angle with the
direction of the plumb line. From the length of the shadow of a vertical staff (“gnomon”) produced in a
hemispherical shell (“skaphe”), he determined this angle as 1/50 of a complete circle, i.e. γ = 7°12’. He
estimated the distance from Syene to Alexandria to be 5000 stadia as taken from Egyptian cadastre maps which
are based on the information of “bematists” (step counters). With the length of an Egyptian stadium as 157.5 m,
we obtain an earth radius of 6267 km. This value departs from the radius of a mean spherical earth (6371 km) by
–2%. A subsequent determination in antiquity is attributed to Posidonius (135-51 B.C.); using the meridian arc
from Alexandria to Rhodes, he arrived at a radius of the earth deviating by –11%.
During the middle ages in Europe, the question of the figure of the earth was not pursued further. An arc
measurement handed down by the Arabs was carried out (~827 A.D.) by the caliph of al-Mámûn, northwest of

3
Bagdad (+10% deviation). At the beginning of the modern ages, the French physician Fernel in 1525 observed
on the meridian through Paris the geographical latitudes of Paris and Amiens using a quadrant; he computed the
distance from the number of rotations of a wagon wheel (+0.l% deviation).
The remaining arc measurements based on the notion of a spherical earth are characterized by
fundamental advances in instrumentation technology (1611, Kepler telescope) and
methodology. After the initial application of triangulation by Gemma Frisius (1508-1555) in
the Netherlands. and by Tycho Brahe (1546-1601) in Denmark, the Dutchman Willebrord
Snellius (1580-1626) conducted the first triangulation to determine the figure of the earth,
HAASBROEK (1968).
In 1615 with the triangulation applied by Snellius to the arc measurement between Bergen op Zoom and
Alkmaar (Holland), the hitherto inaccurate estimate or direct measurement of the length of arc was replaced by a
procedure of high precision. This method served into the twentieth century for are measurements and for the
formation of principal control networks. For Snellius, the deviation with respect to the mean earth radius
amounts to –3.4%.

Fig.1.3 Determination of the

central angle from zenith angles

Through the initiative of the Academy of Sciences, founded in Paris 1666, France in the seventeenth and
eighteenth centuries assumed the leading role in geodesy. The French abbe J. Picard in 1669/70 carried out an
arc measurement on the meridian through Paris between Malvoisine and Amiens with the aid of a triangulation
network; he was the first to use a telescope with cross hairs. The value obtained by him for the radius of the earth
(deviation of +0.01%,) aided Newton in the verification of the law of gravitation which he had formulated in
1665/66.
Another solution of the determination of the central angle, different in principle, namely by using reciprocal
zenith angles, found application in 1645 by the Italians Grimaldi and Riccioli (Fig. 1.3). The angle may be
computed from the zenith angles z1 and z2 observed at P1 and P2 according to

γ = z1 + z2 – π . (1.2)

This procedure does not yield satisfactory results due to the insufliciently accurate determination of the curvature
of light rays (refraction anomalies).

1.3.2. The Ellipsoidal Earth Model

In the sixteenth and seventeenth centuries, new observations and ideas from astronomy and physics decisively
influenced the perception of the figure of the earth and its position in space. N. Copernicus (1473-1543) achieved
the transition from the geocentric universe of Ptolemy to a heliocentric system (1543: “De revolutionibus orbium
coelestium”), which Aristarchus of Samos (~320-250 B.C.) had already postulated. J. Kepler (1571-1630)
discovered the laws of planetary motion (1609: “Astronomia nova ...”, 1619: “Harmonices mundi”), and Galileo
Galilei (1564-1642) developed modern mechanics (law of falling bodies, law of pendulum motion).
In 1666, the astronomer J. D. Cassini observed the flattening of the poles of Jupiter. The
astronomer J. Richer in 1672 discovered on the occasion of an expedition to Cayenne to
determine martian parallaxes, that he must shorten a one-second pendulum which had been
regulated in Paris, in order to regain oscillations of one second. From this observation and on
the basis of the law of pendulum motion, one can infer an increase in gravity from the equator
to the poles. Building on these and on their own works, Isaac Newton (1643-1727) and
Christian Huygens (1629-1695) developed earth models flattened at the poles and founded on
principles of physics.

4
Newton (1687: “Philosophiae naturalis prineipia mathematica”) obtained a rotational ellipsoid as an equilibrium
figure for a homogeneous. fluid, rotating earth based on the validity of the law of universal gravitation. The
flattening

f = (a – b) / a. (1.3)

(f for “flattening”, a = semimajor axis, b = semiminor axis) in this case amounts to l/230. At the same time,
Newton postulated an increase in gravity acceleration from the equator to the poles proportional to sin2 ϕ (ϕ =
geographical latitude). Huygens ( 1690: “Discours de la Cause de la Pesanteur”) shifts the source of the earth’s
attractive forces to the center of the earth and develops a rotationally symmetric equilibrium surface which
possesses a meridian curve of fourth order with f = 1/576.
For a geometric verification of the ellipsoidal earth model, one has employed arc
measurements at various latitudes. Namely, the length of a one-degree arc (meridian arc for a
difference of 1° in latitude) in the case of flattened poles increases poleward from the equator.
The ellipsoidal parameters a, b or a, f can be computed from two arc measurements [1.3.3].
An evaluation of the existing older arc measurements (Snellius, Picard, among others) led to an earth model
elongated at the poles. The same result was obtained by La Hire, J. D. and J. Cassini (1683-1718) who extended
the arc of Picard north to Dunkirk and south to Collioure (latitude difference of 8°20′). The computations from
two arc segments yielded a "negative" flattening off = –1/95, which can be attributed particularly to
measurement errors of the astronomic latitudes. The intense dispute between the supporters of Newton and those
of Cassini over the figure of the earth was resolved by two further are measurements sponsored by the French
Academy of Sciences.
Maupertuis and Clairaut, among others, participated in the expedition to Lapland (1736/37);
the results of this arc measurement (average latitude 66°20′, latitude interval 57′.5) confirmed
the polar flattening. In combination with the are measurement on the meridian through Paris,
revised by Cassini de Thury and La Caille, 1739/40, the result was f = 1/183. On a second
expedition (1735-1744) to Peru (regions of today`s Ecuador), an arc of average latitude 1°31′
S and 3°07' amplitude was de·termined by Godin, Bouguer, and La Condamine. Combination
with the Lapland arc led to f = 1/210. The flattening of the earth at the poles was thereby
demonstrated by geodetic measurements.
A synthesis between the physical and geodetic substantiations of the ellipsoidal shape of the
earth was finally achieved by A.-C. Clairaut (1713-1765) with the theorem (1743) named for
him, which permits the computation ofthe flattening from two gravity measurements at
different latitudes [3.5.2]. The practical application of this “gravimetric method” suffered
until the twentieth century from the lack of accurate and well distributed gravity
measurements and from the problem of reducing these data to the earth ellipsoid.

1.3.3. Arc Measurements

After the rotational ellipsoid had asserted itself as a model for the earth, numerous arc
measurements were conducted until the middle of the nineteenth century to determine the
dimension of this earth ellipsoid. The arc length was invariably obtained by triangulation We
distinguish between arc measurements along an ellipsoidal meridian (latitude arc
measurement), along a parallel (longitude arc measurement), and arc measurements oblique to
the meridian.
For the computations in a latitude arc measurement (Fig. l.4), the angles ∆ϕ = ϕ2 – ϕ1 , ∆ϕ′ = ϕ′2 – ϕ′1 are
formed from the observed geographic latitudes ϕ1, ϕ2, ϕ′1, ϕ′2. The corresponding meridian arcs ∆G and ∆G' are
obtained from triangulation networks. For short arcs one can replace the meridian ellipse by the osculating circle
having the meridian radius of curvature M = M(ϕ) evaluated at the mean latitude ϕ = ½(ϕ1 + ϕ2), where M is
also a function of the ellipsoidal parameters a, f [3.4.2]. From ∆G = M ∆ϕ and ∆G' = M′ ∆ϕ′ , a and f may be
determined. The larger the latitude interval ϕ′ – ϕ, the more accurate is the computed flattening; whereas, the
accuracy of a depends in particular on the lengths of the meridian arcs.

5
 Fig.1.4 Latitude arc measurement

Particular significance was attained by the measurement commissioned by the French

National Assembly and carried out by Delambre and Méchain on the meridian through Paris
between Barcelona and Dunkirk (1792-1798); it was supposed to serve for the definition of
the meter as a natural unit of length. In combination with the Peruvian arc measurement, this
yielded an ellipsoidal flattening of f = 1/334.
Of the numerous are measurements carried out in the nineteenth and twentieth centuries, which were largely the
foundations of geodetic surveys, we mention here only the older, historically important arcs of Gauss (arc
measurement between Göttingen and Altona, 1821- 1825, adjustment according to the least squares method) and
of Bessel and Baeyer (arc measurement oblique to the meridian in East Prussia, 1831-1838). References to more
recent and to some extent still currently significant works are made in the treatment of astrogeodetic methods
[5.1.4].

1.3.4. The Geoid and the Ellipsoid

As P.-S. Laplace (1802), C. F. Gauss (1828), F. W Bessel (1837), and others had already
recognized, the assumption of an ellipsoidal earth model is not tenable under sufficiently high
observational accuracy. Namely, one can no longer ignore the deviation (deflection of the
vertical) of the physical plumb line, to which the measurements refer, from the ellipsoidal
normal. By an adjustment of several arc measurements for the determination of the ellipsoidal
parameters a and f , contradictions arise which exceed by far the observational accuracy.
An initial adjustment of arc measurements was carried out in 1806 by A. M. Legendre in his treatise "Sur la
méthode des moindres carrées". C. F. Gauss was the first to adjust a triangulation network (in and around
Brunswick, 1803-1807) by the method of least squares (GERARDY 1977).
Despite these discrepancies, numerous adjustments were undertaken until the mid-nineteenth
century to determine the dimensions of the ellipsoid, whereby the deflections of the vertical,
being physically caused, and hence, having systematic characteristics were treated as random
observational errors. With the definition of geodesy [1.1] and the introduction of the geoid
[1.2], F. R. Helmert made a transition to the current concept of the figure of the earth. Here,
the deflections of the vertical are taken into account in the computation of the ellipsoidal
parameters. The three-dimensional concept of geodesy was also introduced in that time
(BRUNS 1878).
Friedrich Robert Helmert (1843-1917), one of the most distinguished geodesists of modern times. was professor
of geodesy at the technical university at Aachen and director of the Prussian Geodetic Institute in Potsdam and of
the central office of the "Internationale Erdmessung". Through his work, geodesy has experienced decisive
impulses, which until today have their effect. In his fundamental monograph (A1880/1884) Helmert established
geodesy as a proper science (WOLF 1970).
The determination of the geoid was for about 70 years (1880-1950) a major goal of geodesy.
Its importance diminished after 1945 with the development of methods for the direct
derivation of the physical surface of the earth; however, its determination still remains an

6
essential problem of geodesy. In fact, the significance of the geoid has again increased with
the establishment of three-dimensional continental and global systems [5.1.2], as well as with
the requirements of marine geodesy [3.3.3].

2. The Gravity Field of the Earth

The significance of the external gravity field of the earth in geodesy may be described
comprehensively as follows:
1. The external gravity field is the reference system for the overwhelming part of the
measured quantities in geodesy. This field must be known in order to reduce the quantities
into geometrically defined systems [5.1].
2. If the distribution of gravity values on the surface of the earth is known, then in
combination with other geodetic measurements, the shape of this surface may be
determined [5.2].
3. The most important reference surface for height measurements, the geoid [3.3], as an
idealized ocean surface is a level surface of the gravity field.
4. The analysis of the external gravity field yields information on the structure and
characteristics of the interior of the earth. In making the corresponding gravity field
parameters available, geodesy becomes an auxiliary science of geophysics [5.5].

2.1. Components of the Gravity Field

A body rotating with the earth experiences the gravitational forces of the earth and of other
celestial bodies, as well as the centrifugal force due to the earth`s rotation. The resultant force
is the force of gravity. It is a function of position, but also undergoes temporal variations.
For the geodetic use of earth satellites, one should note that a satellite does not partake in the
rotation of the earth. Hence, only gravitation acts in this case.
The unit of acceleration in the SI-system (Systeme Intemational d’Unités), MARKOWITZ (1973), is ms-2. The
acceleration of gravity can be measured with an accuracy of 10-7 to 10-8 ms-2; the deviations of the terrestrial
gravity field from a "normal earth", in general, remain less than 2 × 10-3 ms-2. Therefore, the sciences of geodesy
and geophysics have until recently adopted the more suitable units
mGal = 10-5 ms-2, µGal = 10-8 ms-2 = 10 nms-2.
They are derived from the unit “Gal” (after Galilei) = cm s-2 used in the cgs-system. In the sequel we shall
mainly use the units
µ ms-2 = 10-6 ms-2 , nms-2 = 10-9 ms-2.

2.1.1. Gravitation, Gravitational Potential

According to Newton’s law of gravitation (1687), two point masses m1 and m2 attract each
other with the gravitational force (attractive force)
m1m2 l
K = −G , (2.1)
l2 l

7
 Fig.2.1 Gravitation b

(G = gravitational constant, l = distance between point masses), where K and l point in

opposing directions. The unit mass situated at the attracted point P (Fig. 2.1) in the
gravitational field experiences a gravitational acceleration (henceforth, also termed
“gravitation”)
ml
b = −G , (2.2)
l2 l
due to the mass element at the attracting point P'. b lies on the line joining P and P′ and is
directed toward P′; l may be represented by the position vectors r and r', e.g. in the Cartesian
x,y,z-system:
  x  x' 
    
 l = r − r' , r =  y  , r' =  y' .
 z  z'  (2.3)
    
l = l = ( x − x' ) + ( y − y' ) + ( z − z' )2 .
2 2

The value of the gravitational constant is (Committee on Data for Science and Technology –
CODATA – system of physical constants 1986)
G = (6.67259 ± 0.00085) × 10-11 m3 kg-1 s-2.
The first experimental determination of G was carried out in 1798 by Cavendish, who used the torsion balance.
The goal ol current work is to increase the relative accuracy of G to better than 1 × 10-4. This includes
investigations into a dependency of G on material. external influences, as well as distance and direction. Until
today, these investigations have not rendered signilicant results (GILLIES 1987).
A body such as the earth, composed of an inlinite number of mass elements, induces a
gravitation on the unit mass at P which is computed by summing the individual accclerations
(2.2) vectorially. The computations are simplified, if one changes from the vector field to a
scalar field.
Because the gravitational field is irrotarional,
curl b = 0 , (2.4)
b may be represented as the gradient of a potential (see e.g. KELLOGG A1929, SIGL A1985):
b = grad V . (2.5)

8
With lim V = 0 , we introduce the gravitational potential
l→∞

Gm
V = V (l ) = (2.6)
l
as a positive quantity (as it is customary in geodesy). The value of the potential at the point P
in the gravitational field indicates the work that must be done by the gravitation in order to
move the unit mass from infinity (V = 0) to P.
The acceleration potential has the dimension of work per unit mass and it has the unit m2s-2.
From the mass element m, we turn to the earth with continuously distributed elements
dm = ρ dv , (2.7)
where ρ = ρ(r' ) is the density and dv: is the volume element. According to the superposition
principle, the gravitational potential of the earth is given according to (2.6) by
dm ρ
V = V (r ) = G ∫∫∫ = G ∫∫∫ dv . (2.8)
earth
l earth
l

In the computation of V, one therefore has to assume that the density function ρ = ρ(r' ) is
completely known; in fact however, it is better known only for parts of the upper crust of the
earth.

2.1.2. Gravitational Potential of a Spherically Symmetric Earth

To a first approximation, the earth can be viewed as a sphere with a centrally symmetric
density structure. For the computation of the gravitational potential. we introduce the
spherical coordinates r,ϑ, λ (Fig. 2.2) which are also required later. With the usual orientation
of this system with respect to a global x,y,z-system (the ϑ = 0 axis coincides with the z-axis
which lies along the spin axis, the λ = 0 axis coincides with the x-axis which lies in the
meridian plane of Greenwich). we have the following relationship
 x  sin ϑ cos λ 
   
r =  y  = r  sin ϑ sin λ  . (2.9)
z  cos ϑ 
   
ϑ is then the spherical polar distance, λ is the geographical longitude.

Fig.2.2 Spherical coordinates r,ϑ, λ

9
ln the subsequent derivation, the spherical coordinate system is oriented such that the ϑ = 0
axis coincides with the line joining O and P (Fig. 2.3).
The potential of a homogeneous spherical shell of infinitesimal thickness dr′ and density ρ
and having a radius r′ is given in analogy to (2.8) by
df
V = G ρ dr ' ∫∫ ,
f
l

where the integration is over the surface of the shell f and where
df = r ' 2 sin ϑ' dϑ' dλ '
is the surface element. A distinction is made in the integration as to whether the attracted
point is exterior or interior to the spherical shell.
For an attracted point lying in the exterior, the potential is given by
r'2 dm'
Ve ' = 4πGρ dr ' = G , (2.10)
r r
here
dm' = 4πρr '2 dr '
represents the mass of the spherical shell. The potential of the spherical earth composed of
concentric homogeneous shells is
dm' GM
Ve = G ∫∫∫ = . (2.11)
earth
r r
Hence, it is equal to the potential of the entire mass M of the earth concentrated at the center
of mass.

Fig.2.3 Gravitational potential of

a centrally symmetric sphere

With GM = 398.6 × 1012 m3 s-2 and the radius of the earth R = 6371 km, the value of the potential at the surface
of the earth (r = R) amounts to V = 6.26 × 107 m2 s-2, and the gravitation is b = 9.82 ms-2.

10
For a point in the interior, we obtain for the potential of the spherical shell:
dm '
Vi ' = 4πGρ r ' dr ' = G . (2.12)
r'
V' here is constant; therefore, the gravitation is zero. The potential inside an earth constructed
of shells is composed of the contribution (2.10) due to the masses interior to the sphere
r = const., as well as the contribution (2.12) from the spherical shell having thickness R – r:
4πG
r R
Vi =
r 0∫ ρ r ' 2 dr ' + 4πG ∫ ρ r ' dr ' .
r
(2.13)

For a homogeneous spherical earth (ρ = const.) we have

4π  r2 
Vi = Gρr 2 + 2πGρ ( R 2 − r 2 ) = 2πGρ R 2 −  . (2.14)
3  3
2.1.3. Properties of the Gravitational Potential

We investigate the properties of the potential function V and its lirst and second derivatives.
If the attracted point P lies exterior to the attracting region, that is, the "physical body of the
earth", then l ≥ 0 always. Here we neglect the mass of the atmosphere MAtm ≈ 10-6 M. The
potential V (2.8) and the first derivatives (in view of (2.3))
∂V x − x'
= Vx = −G ∫∫∫ 3 dm etc. (2.15)
∂x earth
l
as well as the second derivatives
∂ 2V 1 ( x − x' ) 2
∂x 2
= V xx = −G ∫∫∫
earth
l 3
dm + 3G ∫∫∫
earth
l 5
dm etc. (2.16)

are single-valued, finite, and continuous functions in the entire exterior space. They all vanish
at infinity.
The unit of the second derivative of the potential is s-2. In view of the magnitude of the second derivative and the
measuring accuracy (10-8 to 10-9 s-2), the units commonly used are 10-9 s-2 = 1 E(Eötvös) =l µms-2/km = 0.1
mGal/km.

∂2 ∂2 ∂2
Applying the Laplacian operator ∆ ≡ 2 + 2 + 2 (here in Cartesian coordinates) to V
∂x ∂y ∂z
yields the partial differential equation of second order which describes the gravitational field.
With (2.16) we obtain Laplace’s differential equation
∆V = 0 . (2.17)
Continuous functions having continuous first and second derivatives and which fulfill (2.17)
are called harmonic functions or potential functions.
If the attracted point lies inside the body of the earth, then the case l = 0 is possible. This
requires special attention because of the discontinuity of 1/l.
To this end. we consider P enclosed by a sphere K (center at P0, radius p); p is chosen sufficiently small, so that
p = const. inside this sphere (Fig. 2.4). The potential is composed of the contributions due to the masses lying
interior and exterior to the sphere.

11
From (2.8) and (2.14) and using

R = p, R = q = ( x − x0 ) 2 + ( y − y 0 ) 2 + ( z − z0 ) 2

Fig.2.4 Gravitational potential

inside the earth

we find

dm  2 q2 
V =G ∫∫∫ l
earth − K
+ 2 πGρ p −  .
 3
In the limit p → 0, q → 0, agreement is obtained with the expression for the exterior potential (2.8).
Differentiation of V yields
x − x' 4π
V x = −G ∫∫∫
earth − K
l 3
dm − Gρ( x − x0 ) etc.
3
As q → 0, hence also x – x0 = y – y0 = z – z0 → 0, so that we again obtain agreement with the exterior case
(2.15). The second derivatives are given by
1 ( x − x' ) 2 4π
Vxx = −G ∫∫∫
earth − K l
3
dm + 3G ∫∫∫
earth − K l 5
dm −
3
Gρ etc. (2.18)

The gravitational potential and its first derivatives are thus single-valued, finite, and
continuous in the interior as well, The second derivatives, according to (2.18), exhibit
discontinuities for sudden changes in density. For the interior of the earth, Poisson’s
differential equation is
∆V = −4πGρ , (2.19)
which follows from (2.18).

2.1.4. Centrifugal Acceleration, Centrifugal Potential

The centrifugal force arises as a result of the rotation of the earth about its axis. We assume
here a rotation of constant angular velocity ω about the rotational axis, which is further
assumed to be fixed with respect to the earth. The centrifugal acceleration
z = ω2p (2.20)
acting on a unit mass is directed outward perpendicularly to the spin axis (Fig. 2.5).
The angular velocity
ω = 2π: 86164.10 s = 7.292 115 × 10-5 rad s-1
is known with high accuracy from astronomy [4.1.3].

12
If the z-axis of an earth-fixed x,y,z-system coincides with the axis of rotation, then we have
 x
 
p =  y, p = p = x2 + y2 .
0
 
With
z = grad Φ (2.21)
we introduce the centrifugal potential
ω2 2
Φ = Φ( p) = p . (2.22)
2

Fig.2.5 Gravitation b, centrifugal

acceleration z, gravity acceleration g

Differentiating twice and applying the Laplacian operator yields

∆Φ = 2ω 2 . (2.23)
The analytic function Φ, as opposed to V (2.17), is therefore not harmonic.
For points on the equator of the earth, the centrifugal potential has a value of Φ = 1.1 × 105 m2 s-2 and the
centrifugal acceleration is z = |z| = 0.03 ms-2 (≈0.3% of gravitation). At the poles Φ = 0, z = 0.

2.1.5. Gravity Acceleration, Gravity Potential

The gravity acceleration, or gravity g (Latin: gravitas) is the resultant of gravitation b and
centifugal acceleration z (Fig. 2.5):
g =b+z. (2.24)
The direction of g is known as the direction of the plumb line; the magnitude g = |g| is called
the gravity intensity (often just gravity). Using (2.8) and (2.22), the gravity potential W of the
earth becomes
ρ ω2 2
W = W (r ) = V + Φ = G ∫∫∫ dv + p . (2.25)
earth
l 2
The gravity acceleration is given by
g = grad W . (2.26)
W and its first derivatives are single-valued, finite, and continuous as a consequence of the
characteristics of V and Φ, with the exception of the uninteresting cases r → ∞ (then also

13
Φ → ∞ ) and g = 0 (direction of the plumb line is not unique). The second derivatives posses
discontinuities at abrupt density variations.
In geodesy. the most important surface of discontinuity is the physical surface of the earth with a jump in density
from ρ = 0.0013 g cm-3 (density of air) to ρ = 2.7 g cm-3 (mean density of the upper crust of the earth).
From (2.19) and (2.23). we obtain the generalized Poisson differential equation
∆W = −4πGρ + 2ω2 . (2.27)
In the exterior space (ρ = 0, neglecting the density of the air), it turns into the generalized
Laplace diferential equation
∆W = 2ω 2 . (2.28)
Because of the flattening at the poles and the centrifugal acceleration, g varies on the surface of an earth ellipsoid
between 9.78 ms-2 (equator) and 9.83 ms-2 (pole).

2.2. Level Surfaces and Plumb Lines

2.2.1. Definition and Properties of Level Surfaces

The surfaces of constant gravity potential

W = W (r ) = const. . (2.29)
are designated as equipotential, level, or geopotential surfaces (geops) of gravity. As a result
of an elemental displacement ds, the potential difference of the differentially separated level
surfaces (Fig. 2.6) is given, in view of (2.26), by
dW = g ⋅ ds = g ds cos(g, ds) . (2.30)
Therefore, the derivative of the gravity potential in a certain direction is equal to the
projection of the gravity along this direction. If ds is taken along the level surface W = Wp ,
then it follows from dW = 0 that the gravity g is perpendicular to W = Wp. The level surfaces
are intersected perpendicularly by the plumb lines; the tangent to the plumb line has already
been defined in [2.1.5] as the direction of the plumb line. If ds is directed along the outer
surface normal n, then because cos(g, n) = –1, the following important differential
relationship exists:
dW = − g dn . (2.31)

Fig.2.6 Level surfaces and plumb

lines

It provides the link between the potential difference (a physical quantity) and the difference in
height (a geometric quantity) of neighboring level surfaces.
Since only the projection of ds along the plumb line enters in (2.30), dW is independent of the path. Hence, no
work is done for a displacement along a level surface W = const.: the level surfaces are equilibrium surfaces.

14
 Fig.2.7 Level (equipotential)
surfaces and plumb lines near
the earth

If g varies on a level surface, then according to (2.31) the distance dn to a neighboring level
surface must also change. Therefore, the level surfaces are not parallel and the plumb lines are
space curves. As a consequence ofan increase of 0.05 ms-2 in gravity from the equator to the
poles, the level surfaces of the earth converge toward the poles (Fig. 2.7).
The relative decrease ofthe distance between two level surfaces near the earth from the equator to the pole is on
the order of 5 × 10-3. Two level surfaces which are 100.0 m apart at the equator have a distance of only 99.5 m
between them at the poles.
The level surfaces inside the earth and in its more immediate exterior space are closed, spheroidal (resembling a
sphere) surfaces. As an outer limit in the realm of the definition of gravity, one may consider that level surface
for which the gravitation and centrifugal accelerations in the equatorial plane cancel each other. The equatorial
radius of this surface has a value of 42200 km.
The concept of a level surface was introduced by MacLaurin (1742); whereas, Clairaut (1743) thoroughly
discussed “Level Surfaces and Plumb Lines” as a whole. Bruns (1878) put forth the determination of the exterior
level surfaces in their entirety as the fundamental problem of geodesy.

2.2.2. Analytical Representation of Level Surfaces

From the properties of the potential function W = W(r) [2.1.5], it follows that the level
surfaces extending completely in the exterior space are analytical surfaces; that is, they have
no salient or singular points. Level surfaces extending partially or completely inside the earth
exhibit discontinuities in the second derivatives at density irregularities. These surfaces can
thus be constructed only from pieces of different analytical surfaces. For a continuous passage
from one surface segment to another, the curvature [2.2.3] changes discontinuously with the
second derivatives. The analytical parts of the level surfaces can be expanded in Taylor series.
We introduce a local Cartesian system at the attracted point P (Fig. 2.8). The z-axis coincides
with the direction of the plumb line and points toward the zenith; the x(north) and y(east) axes
span the (horizontal) plane tangent to the level surface at P. This astronomically oriented
reference system (left-handed system) is also known as the local astronomic system, cf.
[3.2.2].
In the neighborhood of P, W(x, y, z) is developed into a series:
1
W = W P + Wx x + W y y + W z z + (W xx x 2 + W yy y 2 + Wzz z 2 ) + Wxy xy + W xz xz + W yz yz + ⋯ .
2
Here, Wx, Wy, etc. represent the partial derivatives at P. If the calculation point is on the level
surface, then neglecting terms of third and higher order, and because

15
 Fig.2.8 Local cartesian
coordinate system

W = WP , Wx = W y = 0 , Wz = − g
it follows that the equation of this surface is
1
− gz + (W xx x 2 + W yy y 2 ) + Wxy xy + ⋯ = 0 . (2.32)
2

2.2.3. Curvature of the Level Surfaces

In order to derive the curvature of the level surface at the point P, we introduce the vertical
plane defined by the surface normal (direction of the plumb line) at P and a second point. Its
intersection with the level surface forms a plane curve, the normal section. The direction of
the normal section is given by the angle measured in the horizontal plane between the x-axis
and the normal section. This angle is called the astronomic azimuth A (Fig. 2.8). The
curvature of the normal section (normal curvature) can be determined for small distances s
from the geometrical relations of Fig. 2.8:
1 2z
=− 2 . (2.33)
RA s
Here RA is the radius of curvature. Introducing the plane polar coordinates
x = s cos A , y = s sin A
and substituting (2.32) leads to
1 1
= − (Wxx cos 2 A + 2Wxy sin A cos A + W yy sin 2 A) . (2.34)
RA g
The normal curvature assumes is extreme values in the mutually perpendicular directions of
principal curvature. From a consideration of extremas, we find that for their azimuths A1, and
π
A2 = A1 ± ,
2
Wxy
tan 2 A1, 2 = 2 . (2.35)
Wxx − W yy
The curvatures in the x- and y-directions (A = 0 and A = 180°) are given by

16
 1 W 1 W yy
= − xx , =− . (2.36)
Rx g Ry g
where Rx, Ry are the corresponding radii curvature. (2.34) to (2.36) reveal the connection
between the curvature of the level surfaces and the second derivatives of the gravity potential.

2.2.4. Curvature of Plumb Lines

We start with the curvature vector of the plumb line

 x′′   cos A 
   
r ′′ =  y ′′  = κ  sin A  , (2.37)
 z ′′   0 
   
which lies along the principal normal to the plumb line. Here " denotes the second derivative
with respect to the arc length of the plumb line, κ is the total curvature and A is the azimuth of
the principal normal in the horizontal plane. We obtain the components of r" by
differentiating the gravity vector
 x′   Wx 
   
g = g  y′  =  W y  ,
 − z′   − W 
   z 

and considering that x' = y' = 0, z' = –1:

 W xz 
1  
r ′′ = −  W yz  .
g  0 
 
Substituting this into (2.37) yields the curvature of the plumb line
Wxz W yz
κ=− =− , (2.38)
g cos A g sin A
and the azimuth of its principal normal
W yz
A = arctan . (2.39)
Wxz
The curvatures of the projections of the plumb line on the xz- and yz-planes are given by
Wxz W yz
κx = − , κy = − (2.40)
g g

respectively, where κ = κ 2x + κ 2y . (2,38) to (2.40) show that the curvature of the plumb lines
also depends on the second derivatives of the gravity potential.

2.2.5. Gravity Gradient

The gravity vector expressed in the local astronomic system, cf. [3.2.2], is given by

17
 g T = (grad W ) T = (W x , W y , − W z ) , (2.41a)
see also [2.2,4]. Differentiation yields the gravity gradient tensor (Eötvös tensor)
 Wxx Wxy − Wxz 
 
grad g = grad (grad W ) =  W yx W yy − W yz  . (2.41b)
 −W − Wzy − Wzz 
 zx

Since the gravity field is irrotational, we have

W xy = W yx , Wxz = Wzx , W yz = W zy . (2.41c)
Taking Poisson`s differential equation (2.27) into account, we recognize that (2.41b) only
contains five mutually independent elements. They are closely related to the curvature of the
level surfaces and the plumb lines.

Fig.2.9 Horizontal gradient

of gravity

The gradient of gravity

 ∂g / ∂x   − W xz 
   
grad g =  ∂g / ∂y  =  − W yz  (2.41d)
 ∂g / ∂z   W 
   zz 
describes the variation of gravity in the horizontal plane and in the vertical. The horizontal
gradient (Fig. 2.9) formed by the components ∂g / ∂x and ∂g / ∂y points in the direction of the
maximum increase in gravity in the horizontal plane.
The vertical component ∂g / ∂z = −Wzz describes the variation of gravity with respect to
height. If we combine the generalized Poisson differential equation (2.27) with the mean
curvature

1 1 1  1
H* =  + =− (Wxx + W yy ) , (2.42)
2  Rx R y  2g

computed from (2.36), then we obtain the relation found by BRUNS (1878) between the
vertical component of the gravity gradient and the curvature of level surfaces:
∂g
= −2 gH * +4πGρ − 2ω 2 . (2.43)
∂z

2.3. Spherical Harmonic Expansion of the Gravitational Potential

Because the density function ρ = ρ(r′) is not known well enough, the gravitational potential of
the earth V = V(r) cannot be computed using (2,8). However, as a solution of Laplace`s

18
differential equation (2.17), a series expansion of V is possible which is convergent in the
space exterior to the earth (eg. HOBSON A1931, SIGL A1985).

2.3.1. Expansion of the Reciprocal Distance

Applying the law of cosines in Fig. 2.1, we obtain

− 12
1  r  
2
1 r
= (r 2 + r ' 2 −2rr ' cos ψ ) 2 = 1 +   − 2 cos ψ 
−1
(2.44)
l r   r '  r' 

for the reciprocal distance, appearing in (2.8), between the attracted point P and the attracting
point P'. Here, ψ is the central angle between P and P'.
lf 1/l is expanded in a series converging for r' < r, and if the terms are arranged according to
increasing powers of r'/r, then it follows that
l
1 1 ∞  r' 
= ∑   Pl (cos ψ ) . (2.45)
l r l =0  r 
The Pl (cos ψ ) represent polynomials of lth degree in cos ψ. They are known as Legendre
polynomials (zonal harmonic functions) and they are computed for the argument t = cos ψ by
means of
1 dl 2
Pl (t ) = × (t − 1) l . (2.46)
2 l × l! dt l

Fig.2.10 Spherical distance

and spherical coordinates

If we interpret ψ to be the spherical distance on a unit sphere between the attracted point and
the attracting point, then according to Fig. 2.10 and using the spherical coordinates introduced
in [2.l.2], the following relationship exists:
cosψ = cosϑ cosϑ ′ + sin ϑ sin ϑ ′ cos(λ − λ ′) .
The corresponding decomposition of Pl (cos ψ ) leads to
Pl (cosψ ) = Pl (cosϑ ) Pl (cosϑ ′)
l
(l − m)! (2.47)
+2∑ ( Plm (cosϑ ) cos mλ Plm (cosϑ ′) cos mλ ′ + Plm (cosϑ ) sin mλ Plm (cosϑ ′) sin mλ ′).
m=1 (l + m)!

19
Here, the Pl (t ) are again the Legendre polynomials with argument t = cos ϑ or t = cos ϑ ′ .
The associated Legendre functions of the first kind Plm (t ) (l = degree, m = order) are obtained
by differentiating Pl (t ) m times with respect to t:
m
dm
Plm (t ) = (1 − t 2 ) 2 × Pl (t ) . (2.48)
dt m
We complete the expansion of 1/l by substituting (2.47) into (2.45).
The functions
Plm (cosϑ ) cos mλ , Plm (cosϑ ) sin mλ (2.49)
depending on ϑ and λ are known as (Laplace’s) surface harmonics. They characterize the
behavior of a function on a unit sphere.

2.3.2. Expansion of the Gravitational Potential

After substituting the spherical harmonic expansion (2.45), (2.47) of 1/l , we obtain for the
gravitational potential (2.8)
G ∞ l (l − m)! 1 
V= ∑∑ k ×
r l =0 m=0 (l + m)! r l
 Plm (cos ϑ ) cos mλ ∫∫∫ r ′l Plm (cos ϑ ′) cos mλ ′ dm

 earth
(2.50)
 1 for m = 0
+ Plm (cos ϑ ) sin mλ ∫∫∫ r ′l Plm (cos ϑ ′) sin mλ ′ dm  , k = .
earth  2 for m ≠ 0
For l = 0, the integration yields the potential of the earth's mass M concentrated at the center
of mass (2.11). We extract this term, introduce the semimajor axis of the earth ellipsoid as a
constant, and denote the integrals of the mass by Clm, Slm (harmonic coefficients). The
gravitational potential expanded in spherical harmonics is then written as
 ∞ l

1 + ∑∑  a  (Clm cos mλ + Slm sin mλ ) Plm (cosϑ )  ,
l
GM
V= (2.51)
r  l =1 m=0  r  
 
where

1 r′ 
l

∫∫∫
Cl 0 = Cl =   Pl (cosϑ ′) dm ,
earth 
M a


Clm  2 (l − m)!  r ′ 
l
cos mλ ′ 
 = × ∫∫∫   Plm (cosϑ ′) 
 Slm  M (l + m)! earth  a 
dm
 sin mλ ′ 
 (2.52)


for m ≠ 0 . 


Particularly in satellite geodesy, the coefficients

J l = −Cl , J lm = −Clm , K lm = − Slm . (2.53)
are generally used.
The fully normalized harmonics Plm (cosϑ ) are also employed frequently. These may be
computed from the conventional harmonics (2.46), (2.48) according to

20
 (l − m)! 1 for m = 0
Plm (cos ϑ ) = k (2l + 1) Plm (cos ϑ ) , k= .
(l + m)! 2 for m ≠ 0
In addition to the orthogonality relations for the surface harmonics, we now also have
2
1  cos mλ  
4π ∫∫σ  Plm  sin mλ   dσ = 1
over the unit sphere σ. Correspondingly, the harmonic coeflicients Clm , S lm of the expansion
analogous to (2.51) of the gravitational potential are given by
Clm  (l + m)! Clm  1 for m = 0
 =  , k= . (2.52a)
 Slm  k (2l + 1)(l − m)!  Slm  2 for m ≠ 0
The expansion (2.50) converges outside a sphere of radius r = a which just encloses the terrestrial body. As an
approximation, such a representation may also be applied on the surface of the earth. It should be noted that the
harmonic expansion is in no case valid for the interior of the masses, since the interior is governed by Poisson’s
differential equation (2.19).

2.3.3. The Geometrical Meaning of the Surface Harmonics

We consider here the behavior of the surface harmonics (2.49) which appear in the harm.onic
expansion of the gravitational potential.
The condition m = 0 yields the special case of the Legendre polynomials Pl (cos ψ ) . Because
of their independence of the geographical logitude λ, they divide the surface of the sphere into
zones in which they have alternately positive and negative signs: zonal harmonics. These
harmonics possses l real zeros in the interval 0 ≤ ϑ ≤ π, so that for even l, the sphere is
divided symmetrically with respect to the equator ϑ = 90°; and the case for odd l results in an
asymetric division (Fig. 2.11). The Pl (cos ψ ) up to l = 3 and computed according to (2.46)
are given as follows:
3 1 5 3
P0 = 1 , P1 = cosϑ , P2 = cos 2 ϑ − , P3 = cos 3 ϑ − cos ϑ . (2.54)
2 2 2 2

Fig.2.11. Zonal
harmonics

The Plm (cosϑ ) (m ≠ 0) have (l – m) zeros in the interval 0 < ϑ < π. Because of the
multiplication by cos mλ or sin mλ, the surface harmonics are longitude dependent, furnishing

21
2m zeros in the interval 0 ≤ λ ≤ 2π: tesseral harmonics (Fig. 2.12). For Plm (cosϑ ) up to
l = m = 3, (2.48) yields
P1, 1 = sin ϑ , P2, 1 = 3 sin ϑ cos ϑ , P2, 2 = 3 sin 2 ϑ

 15 3
P3, 1 = sin ϑ  cos 2 ϑ −  , P3, 2 = 15 sin 2 ϑ cos ϑ , (2.55)
2 2
P3, 3 = 15 sin 3 ϑ .

Fig.2.12. Tesseral
harmonics

Finally. for m = l , the dependence on ϑ disappears, and the sphere is divided into sectors of
alternating signs: sectorial harmonics (Fig. 2.13).

Fig.2.13. Sectorial harmonics

The amplitudes of the terms given by the surface harmonics in the gravitational potential expansion are
determined by the harmonic coefficients (2.52). For example, the series has only zonal harmonics for an earth
rotationally symmetric with respect to the ϑ = 0 axis; the Clm, Slm (m ≠ 0) must all vanish. For a mass
distribution which is symmetric with respect to the equator, the zonal harmonic coeilicients with odd l must be
absent.
Surnmarizing we state that the spherical harmonic expansion of the gravitational potential
represents a spectral decomposition of the gravitational field. The field is separated into
structures of wave-length 360°/l, corresponding to a spatial resolution of 180°/l. Contrary to
the integral over all masses of the earth (2.8), the potential is now described by the elements
of the series, with the harmonic coeflicients as specific mass integrals. From well distributed
observations of the potential or functionals of the potential, these coellicients (l < ∞) can be
determined [5.2.3], [5.3.3.], [5.4.4].

22
 2.3.4. Physical Meaning of the Lower Degree Harmonic Coefficients

Some of the mass integrals of lower degree have a simple physical interpretation. To see this,
we substitute the harmonic functions Pl, Plm, for l = 1, 2 and m = 0, 1, 2 from (2.54) and (2.55)
into (2.52) and subsequently transform the spherical coordinates into Cartesian coordinates
using (2.9). This yields
1 1 1
C1 =
aM ∫∫∫ z′dm ,
earth
C1, 1 =
aM ∫∫∫ x′dm ,
earth
S1, 1 =
aM ∫∫∫ y′dm .
earth
(2.56)

The integrals divided by M are the coordinates of the center of mass of the earth. If we place
the origin of the coordinate system at the center of mass. as it is commonly done, then this
gives
C1 = C1, 1 = S1, 1 = 0 . (2.57)
For l = 2, we obtain
1  2 x′ 2 + y′ 2  1
C2 = 2
a M ∫∫∫
earth
 z ′ −
 2
 dm ,

C2 , 1 =
a M2 ∫∫∫ x′z′ dm ,
earth

∫∫∫ (x′ )
1 1
S 2, 1 = ∫∫∫ y′z′ dm , C2 , 2 = − y′ 2 dm ,
2
2
(2.58)
a M earth
4a 2 M earth

1
S 2, 2 =
2a 2 M ∫∫∫ x′y′ dm .
earth

These expressions contain the moments of inertia

(
A = ∫∫∫ y′2 + z ′ 2 dm , ) (
B = ∫∫∫ x′2 + z ′2 dm ,) (
C = ∫∫∫ x′ 2 + y′ 2 dm )
and the products of inertia
D = ∫∫∫ y′z ′ dm , E = ∫∫∫ x′z ′ dm , F = ∫∫∫ x′y ′ dm

with respect to the coordinate axes. The z-axis contains the mean axis of rotation which, if
polar motion [3.1] is neglected, coincides with one of the principal axes of inertia (maximum
moment of inertia). Whence, D = E = 0 and C becomes a principal moment of inertia. If we
let the x-axis coincide with one of the two principal axes of inertia in the plane of the equator,
then A and B become principal moments of inertia with respect to these equatorial axes;
furthermore F = 0. Since the locations of the principal axes of inertia in the equator are
unknown, the usual reckoning of the geographic longitude λ from the Greenwich meridian is
retained.
Using the moments of inertia A, B, C and the product of inertia F, the harmonic coefficients
may be expressed as follows:
1  A+ B  B− A F
C2 = 2 −C, C2, 1 = S 2, 1 = 0 , C2, 2 = , S 2, 2 = . (2.59)
a M 2  4a 2 M 2a 2 M
J 2 = −C 2 is also known as the dynamic form factor.
The flattening of the earth at the poles represents the greatest deviation from spherical symmetry. This is evident
from the numerical value for C2 which is three orders of magnitude larger than the values of the successive
coefficients. C2, 2 and S2, 2 characterize the deviation of the terrestrial mass distribution from rotational symmetry
(ellipticity of the equator) [3.5.4], [5.3.3].

23
 2.4. Temporal Variations of the Gravity Field

Gravity changes with time may be divided into gravimetric earth tides and changes of earth
rotation, on one hand, and temporal variations caused by terrestrial mass displacements, on
the other.
Time dependent tidal accclerations are caused by the lunar and solar gravitational forces
acting on different parts of the rotating earth, in combination with the effects of the
revolutions of the moon about the earth and the earth about the sun. The accelerations produce
variations in the terrestrial gravity field on the order of 10-7 g. A comprehensive description is
given by MELCHIOR (A1983). Other variations of the gravity field in time generally are at
least one order of magnitude smaller than the tidal effects.

2.4.1. Tidal Acceleration, Tidal Potential

For a rigid earth, the tidal potential may be determined from the law of gravitation and the
orbital elements of the moon and the sun (“theoretical tides”), see e.g. BARTELS (A1957). The
computations are carried out separately for the earth-moon and earth-sun systems; the results
are subsequently added. The moon and the sun in this case may be regarded as point masses.
We introduce a coordinate system whose origin is at the earth’s center of mass, moving with
the earth in space, but not rotating (revolution without rotation). All points of the earth
describe the same rotational motion in this system, with monthly (moon) or yearly (sun)
periods. Hence, the centrifugal acceleration acts equally at all points. At the earth’s center of
mass S, it is compensated by the gravitation b0 of the moon and sun, respectively (equilibrium
in the respective systems). At the other points of the earth, tidal accelerations bt arise being
the difference between the gravitational accelerations b and b0 (Fig. 2.l4):
bt = b − b0 . (2.60)
We make the transition from the acceleration to the tidal potential Vt:
b t = grad Vt = grad (V − V0 ) . (2.61)
V is obtained from (2.6) as the gravitational potential of a point mass, in which, for the case of
the moon, m is to be replaced by the mass of the moon MM. For the determination of V0, we
introduce a rectangular coordinate system with its origin at S, and whose z-axis coincides with
the line joining S and M. The potential of the homogeneous b0-field is then b 0 × z .

Fig.2.14. Lunar tidal

acceleration

With
GM M
b0 = and z = r cosψ M
rM2

24
and incorporating the constant GMM /rM (then the tidal potential as S becomes Vt = 0), one
obtains
 1 1 r cosψ M 
Vt = GM M  − −  . (2.62)
 lM rM rM2 
Since for points on the earth’s surface (r = R) we have the ratio r/rM = 1/60 (for the sun the
corresponding ratio is 1/23600), the reciprocal distance
1 −1
= (r 2 + rM2 − 2rrM cosψ M ) 2
lM
may be expanded, corresponding to (2.45). into spherical harmonics, with only degrees l ≥ 2
being effective. For l = 2 and with Doodson’s tidal constant
3 r2
GM (r ) = GM M 3 (2.63)
4 rM

( rM = mean distance to the moon) and cos 2 ψ M = 12 (cos 2ψ M + 1) , we get

3
r  
  cos 2ψ M +  .
1
Vt = GM (r ) M (2.64)
 rM   3
A corresponding formula holds for the tidal potential due to the sun. For r = R the tidal
constants for the moon and the sun are
GM (R) = 2.628 m2 s-2, GS (R) = 1.208 m2 s-2.
Hence, the solar tides amount to 46.2% of the lunar tides.
For stationary earth-moon and earth-sun systems, the level surfaces of the earth would experience a deformation
that is constant with respect to time. Freely moving masses of water covering the entire earth would assume the
form of one of these level surfaces: equilibrium tide. The variations in height of the level surfaces due to the tidal
potential may be computed using (2.31). For the moon (sun), this gives an increase of 0.36 (0.16) m at ψ = 0°,
180° on the earth's surface; at ψ = 90°, 270°, there is a decrease of 0.18 (0.08) m.
Differentiating (2.64) furnishes the tidal acceleration. The radial component (positive
outward) for a mean distance to the moon ( rM = rM ) is found to be
∂Vt 2  1
br = = GM (r )  cos 2ψ M +  . (2.65)
∂r r  3
The tangential component (positive in the direction toward the moon) is
∂Vt 2
bψ = − = GM (r ) sin 2ψ M . (2.66)
r∂ψ r

The time dependent change in gravity –br (r deviates only slightly from the direction of the plumb line) thus
varies in the case of the moon (sun) between –1.1(–0.5) µms-2 and +0.5(+0.3) µms-2 . The direction of the plumb
line fluctuates on the order of bψ /g, hence about ±0.″017(±0.″008).
Formula (2.62) establishes the dependence of the tidal potential on the position of the moon
(rM, ψM). The variation of this field in time is more easily recognized if we change to an earth-
fixed coordinate system. that is, one that rotates with the earth. According to Fig. 2.15, we
have

25
 Fig.2.15. Astronomic triangle

cosψ M = sin ϕ sin δ M + cosϕ cos δ M cos t M

( ϕ = geocentric latitude of the attracted point P, δ M = declination and tM = Θ – αM = hour
angle of the moon, αM (right ascension) and δ M are obtained for the sidereal time Θ from
astronomic almanacs [4.1.2]). Substituting this into (2.64) yields
3
r   1  1 
Vt = GM (r ) M   3 − sin 2 ϕ  − sin 2 δ M 
 rM   3  3 
(2.67)
+ sin 2ϕ sin 2δ M cos t M + cos ϕ cos δ M cos 2t M
2 2

for the tidal potential of the moon (a corresponding equation may be obtained for the sun).
In (2.67), the quantities rM , δM , tM vary with various periods. The first term, which is in- dependent of the
earth‘s rotation, exhibits long periodic oscillations (14 days; and 0.5 years for the sun). Its nonperiodic part
causes a permanent deformation of the level surfaces (they are lowered at the pole by about 0.20 m and raised at
the equator by about 0.10 m). The second term oscillates with diurnal periods; the third with semidiurnal periods.
In the present (1990) state of the art of tidal observations. the spherical harmonic expansion (2.67) must include
terms up to l = 4 for the moon, and l = 3 for the sun.
For a comparison of the observed and theoretical (rigid earth) tides, DOODSON (1921)
decomposed the tidal potential of the moon and sun into a sum of harmonic oscillations
(partial tides or waves), which are determined by their frequencies, amplitudes, and phases.
The expansion of CARTWRIGHT and TAYLER/EDDEN (1971/1973) contains 505 waves. The
most important ones are the semidiumal waves M2 (moon), S2 (sun), N2 (eccentricity of the
lunar orbit) and the diumal waves O1 (moon), P1 (sun). K1 (lunisolar declination). More
recent developments include more than 1000 tidal waves and enable the gravity tides to be
computed to ±0.1 nms-2 and better (e.g.TAMURA 1987).

2.4.2. Earth Tides

The partially elastic body of the earth is deformed by the tides; tides of the solid earth (earth
tides); see e.g. TOMASCHEK (A1957), MELCHIOR (1974), ZÜRN and WILHELM (1984). For a
spherically symmetric, non-rotating elastic body, the earth’s tides are described by the theory
of Love.
Under the influence of the tidal potential Vt (2.64), the earth is deformed in a radial direction
(Fig. 2.16) by a fraction of the variation ∆rt in the level surfaces
Vt
∆rel = h∆rt = h . (2.68)
g

26
 Fig.2.16. Tides of the solid
earth

As a consequence of the new distribution of mass, the level surfaces of the equilibrium tide
W + Vt = const. are deformed; the additional potential is proportional to Vt :
Vd = kVt . (2.69)
The deformations of the earth in the horizontal direction are proportional to the horizontal
tidal accclerations (x = north, y = east):
l ∂Vt l ∂Vt
∆xel = , ∆yel = . (2.70)
g ∂ϕ g cosϕ ∂λ
In (2.68) to (2.70), the (dimensionless) Love parameters h, k, l (l is also called Shida number)
enter as proportionality factors. They depend on the density and rigidity in the solid earth. For
a homogeneous spherical earth, k = 0.6h = 2l.
For a point on the earth’s surface, the gravity potential is altered by the amounts Vt , Vd due to
the tides, and as a result of the change in potential due to ∆rel :
Vel = Vt + Vd − g ⋅ ∆rel = Vt (1 + k − h) . (2.71)
By differentiating with respect to r, we obtain the radial component of the tidal acceleration.
Under the assumption that Vd can be represented by a spherical harmonic of second degree, it
is found, using (2.65), that
∂Vel  3 
br ( el ) = = 1 − k + h  br . (2.72)
∂r  2 
The tangential component is given by
∂Vel
bψ (el ) = − = (1 + k − h) bψ . (2.73)
r∂ψ
Therefore, due to the flexibility of the earth`s surface, the observed deviation of the vertical
bψ (el) /g is smaller than the theoretical one.
The amplitude factors of the principal waves may be obtained by (2.72)
3
δ = 1− k + h (2.74)
2
using a gravimeter, and by (2.73)
γ = 1+ k − h (2.75)
using a tiltmeter. Finally, extensometers deliver linear combinations of h and l. Together with
these amplitude factors, one can determine the accompanying phase shift between the
observed and the theoretical tides [4.2.6].

27
 2.4.3. Other Temporal Variations of the Gravity Field

Besides the tides, the terrestrial gravity field is affected by a number of additional time
dependent processes. Changes in gravitation are caused by the shifting of mass in the
atmosphere, in the oceans, on the solid surface of the earth, and in its interior. These changes
can occur in various forms (abrupt, periodic or quasi-periodic, secular). Their effect at the
earth’s surface can be local, regional, or global, with amplitudes seldom exceeding 10-8 g (eg.
TORGE A1989). Consequently, research and modeling in this field is still at its beginning. This
refers also to a possible secular change of the gravitational constant. The causes of these
processes, their variation in time, and how they are recorded will be taken up in the treatment
of geodynamics [5.5.5]. The variations of the centrifugal acceleration caused both by polar
motion [3.1] and by the changes in the earth’s rotational velocity [4.1.3] could only be
recognized in long-term gravity recordings of high accuracy.

28