guide to geophysical equations
TRANSCRIPT
-
7/26/2019 Guide to Geophysical Equations
1/296
-
7/26/2019 Guide to Geophysical Equations
2/296
A Students Guide to Geophysical Equations
The advent of accessible student computing packages has meant that geophysics students
can now easily manipulate datasets and gain rst-hand modeling experience essential
in developing an intuitive understanding of the physics of the Earth. Yet to gain a more
in-depth understanding of the physical theory, and to be able to develop new models and
solutions, it is necessary to be able to derive the relevant equations from rst principles.
This compact, handy book lls a gap left by most modern geophysics textbooks,
which generally do not have space to derive all of the important formulae, showing the
intermediate steps. This guide presents full derivations for the classical equations of
gravitation, gravity, tides, Earth rotation, heat, geomagnetism, and foundational seismol-
ogy, illustrated with simple schematic diagrams. It supports students through the suc-
cessive steps and explains the logical sequence of a derivation facilitating self-study
and helping students to tackle homework exercises and prepare for exams.
w i l l i a m l o w r i e was born in Hawick, Scotland, and attended the University of
Edinburgh, where he graduated in 1960 with rst-class honors in physics. He achieved a
masters degree in geophysics at the University of Toronto and, in 1967, a doctorate at the
University of Pittsburgh. After two years in the research laboratory of Gulf Oil Company
he became a researcher at the Lamont-Doherty Geological Observatory of Columbia
University. In 1974 he was elected professor of geophysics at the ETH Zrich (Swiss
Federal Institute of Technology in Zurich), Switzerland, where he taught and researched
until retirement in 2004. His research in rock magnetism and paleomagnetism consisted
of deducing the Earths magnetic eld in the geological past from the magnetizations of
dated rocks. The results were applied to the solution of geologic-tectonic problems, and
to analysis of the polarity history of the geomagnetic eld. Professor Lowrie has authored
135 scientic articles and a second edition of his acclaimed 1997 textbookFundamentals
of Geophysicswas published in 2007. He has been President of the European Union of
Geosciences (19879) and Section President and Council member of the American
Geophysical Union (20002). He is a Fellow of the American Geophysical Union and
a Member of the Academia Europaea.
-
7/26/2019 Guide to Geophysical Equations
3/296
-
7/26/2019 Guide to Geophysical Equations
4/296
A Students Guide
to Geophysical Equations
W I L L I A M L O W R I E
Institute of Geophysics
Swiss Federal Institute of Technology
Zurich, Switzerland
-
7/26/2019 Guide to Geophysical Equations
5/296
c a m b r i d g e u n i v e r s i t y p r e s s
Cambridge, New York, Melbourne, Madrid, Cape Town,
Singapore, So Paulo, Delhi, Tokyo, Mexico City
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9781107005846
William Lowrie 2011
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2011
Printed in the United Kingdom at the University Press, Cambridge
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data
Lowrie, William, 1939
A students guide to geophysical equations / William Lowrie.
p. cm.
Includes bibliographical references and index.
ISBN 978-1-107-00584-6 (hardback)1. Geophysics Mathematics Handbooks, manuals, etc.
2. Physics Formulae Handbooks, manuals, etc. 3. Earth Handbooks, manuals, etc.
I. Title.
QC809.M37L69 2011
550.1051525dc22
2011007352
ISBN 978 1 107 00584 6 Hardback
ISBN 978 0 521 18377 2 Paperback
Cambridge University Press has no responsibility for the persistence or
accuracy of URLs for external or third-party internet websites referred to
in this publication, and does not guarantee that any content on such
websites is, or will remain, accurate or appropriate.
http://www.cambridge.org/http://www.cambridge.org/9781107005846http://www.cambridge.org/9781107005846http://www.cambridge.org/ -
7/26/2019 Guide to Geophysical Equations
6/296
This book is dedicated to Marcia
-
7/26/2019 Guide to Geophysical Equations
7/296
-
7/26/2019 Guide to Geophysical Equations
8/296
Contents
Preface pagexi
Acknowledgments xiii
1 Mathematical background 1
1.1 Cartesian and spherical coordinates 1
1.2 Complex numbers 1
1.3 Vector relationships 4
1.4 Matrices and tensors 8
1.5 Conservative force,eld, and potential 17
1.6 The divergence theorem (Gausss theorem) 18
1.7 The curl theorem (Stokestheorem) 20
1.8 Poissons equation 23
1.9 Laplaces equation 26
1.10 Power series 28
1.11 Leibnizs rule 32
1.12 Legendre polynomials 32
1.13 The Legendre differential equation 34
1.14 Rodriguesformula 41
1.15 Associated Legendre polynomials 43
1.16 Spherical harmonic functions 491.17 Fourier series, Fourier integrals, and Fourier transforms 52
Further reading 58
2 Gravitation 59
2.1 Gravitational acceleration and potential 59
2.2 Keplers laws of planetary motion 60
2.3 Gravitational acceleration and the potential of a solid
sphere 66
2.4 Laplaces equation in spherical polar coordinates 69
2.5 MacCullaghs formula for the gravitational potential 74
Further reading 85
vii
-
7/26/2019 Guide to Geophysical Equations
9/296
3 Gravity 86
3.1 The ellipticity of the Earths gure 86
3.2 The geopotential 88
3.3 The equipotential surface of gravity 91
3.4 Gravity on the reference spheroid 96
3.5 Geocentric and geographic latitude 102
3.6 The geoid 106
Further reading 115
4 The tides 116
4.1 Origin of the lunar tide-raising forces 116
4.2 Tidal potential of the Moon 119
4.3 Loves numbers and the tidal deformation 1244.4 Tidal friction and deceleration of terrestrial and lunar
rotations 130
Further reading 136
5 Earths rotation 137
5.1 Motion in a rotating coordinate system 138
5.2 The Coriolis and Etvs effects 140
5.3 Precession and forced nutation of Earths rotation axis 142
5.4 The free, Eulerian nutation of a rigid Earth 1555.5 The Chandler wobble 157
Further reading 169
6 Earths heat 170
6.1 Energy and entropy 171
6.2 Thermodynamic potentials and Maxwells relations 172
6.3 The melting-temperature gradient in the core 176
6.4 The adiabatic temperature gradient in the core 178
6.5 The Grneisen parameter 179
6.6 Heatow 182
Further reading 197
7 Geomagnetism 198
7.1 The dipole magnetic eld and potential 198
7.2 Potential of the geomagnetic eld 200
7.3 The Earths dipole magnetic eld 205
7.4 Secular variation 213
7.5 Power spectrum of the internal eld 214
7.6 The origin of the internal eld 217
Further reading 225
8 Foundations of seismology 227
8.1 Elastic deformation 227
viii Contents
-
7/26/2019 Guide to Geophysical Equations
10/296
8.2 Stress 228
8.3 Strain 233
8.4 Perfectly elastic stressstrain relationships 239
8.5 The seismic wave equation 244
8.6 Solutions of the wave equation 252
8.7 Three-dimensional propagation of plane P- and S-waves 254
Further reading 258
Appendix A Magnetic poles, the dipoleeld, and current loops 259
Appendix B Maxwells equations of electromagnetism 265
References 276
Index 278
Contents ix
-
7/26/2019 Guide to Geophysical Equations
11/296
-
7/26/2019 Guide to Geophysical Equations
12/296
Preface
This work was written as a supplementary text to help students understand the
mathematical steps in deriving important equations in classical geophysics. It is
not intended to be a primary textbook, nor is it intended to be an introduction to
modern research in any of the topics it covers. It originated in a set of handouts, a
kindof do-it-yourself manual, that accompanied a course I taught on theoretical
geophysics. The lecture aids were necessary for two reasons. First, my lectures
were given in German and there were no comprehensive up-to-date texts in the
language; the recommended texts were in English, so the students frequentlyneeded clarication. Secondly, it was often necessary to explain classical theory
in more detail than one nds in a multi-topic advanced textbook. To keep such a
book as succinct as possible, the intermediate steps in the mathematical derivation
of a formula must often be omitted. Sometimes the unassisted student cannotll
in the missing steps without individual tutorial assistance, which is usually in
short supply at most universities, especially at large institutions. To help my
students in these situations, the do-it-yourself text that accompanied my lec-
tures explained missing details in the derivations. This is the background against
which I prepared the present guide to geophysical equations, in the hope that it
might be helpful to other students at this level of study.
The classes that I taught to senior grades were largely related to potential
theory and primarily covered topics other than seismology, since this was the
domain of my colleagues and better taught by a true seismologist than by a
paleomagnetist! Theoretical seismology is a large topic that merits its own
treatment at an advanced level, and there are several textbooks of classical
and modern vintage that deal with this. However, a short chapter on the
relationship of stress, strain, and the propagation of seismic waves is included
here as an introduction to the topic.
Computer technology is an essential ingredient of progress in modern geo-
physics, but a well-trained aspiring geophysicist must be able to do more than
xi
-
7/26/2019 Guide to Geophysical Equations
13/296
apply advanced software packages. A fundamental mathematical understanding
is needed in order to formulate a geophysical problem, and numerical computa-
tional skills are needed to solve it. The techniques that enabled scientists to
understand much about the Earth in the pre-computer era also underlie much of
modern methodology. For this reason, a university training in geophysics still
requires the student to work through basic theory. This guide is intended as a
companion in that process.
Historically, most geophysicists came from the eld of physics, for which
geophysics was an applied science. They generally had a sound training in
mathematics. The modern geophysics student is more likely to have begun
studies in an Earth science discipline, the mathematical background might be
heavily oriented to the use of tailor-made packaged software, and some studentsmay be less able to handle advanced mathematical topics without help or
tutoring. To ll these needs, the opening chapter of this book provides a
summary of the mathematical background for topics handled in subsequent
chapters.
xii Preface
-
7/26/2019 Guide to Geophysical Equations
14/296
Acknowledgments
In writing this book I have beneted from the help and support of various
people. At an early stage, anonymous proposal reviewers gave me useful
suggestions, not all of which have been acted on, but all of which were
appreciated. Each chapter was read and checked by an obliging colleague.
I wish to thank Dave Chapman, Rob Coe, Ramon Egli, Chris Finlay, Valentin
Gischig, Klaus Holliger, Edi Kissling, Emile Klingel, Alexei Kuvshinov,
Germn Rubino, Rolf Sidler, and Doug Smylie for their corrections and sug-
gestions for improvement. The responsibility for any errors that escaped scru-tiny is, of course, mine. I am very grateful to Derrick Hasterok and Dave
Chapman for providing me with an unpublished gure from Derricks Ph.D.
thesis. Dr. Susan Francis, Senior Commissioning Editor at Cambridge
University Press, gave me constant support and friendly encouragement
throughout the many months of writing, for which I am sincerely grateful.
Above all, I thank my wife Marcia for her generous tolerance of the intrusion
of this project into our retirement activities.
xiii
-
7/26/2019 Guide to Geophysical Equations
15/296
-
7/26/2019 Guide to Geophysical Equations
16/296
1
Mathematical background
1.1 Cartesian and spherical coordinates
Two systems of orthogonal coordinates are used in this book, sometimes
interchangeably. Cartesian coordinates (x, y, z) are used for a system with
rectangular geometry, and spherical polar coordinates (r, , ) are used for
spherical geometry. The relationship between these reference systems is shown
inFig. 1.1(a). The convention used here for spherical geometry is dened as
follows: the radial distance from the origin of the coordinates is denoted r, the
polar angle (geographic equivalent: the co-latitude) lies between the radius
and thez-axis (geographic equivalent: Earths rotation axis), and the azimuthal
angle in the xy plane is measured from the x-axis (geographic equivalent:
longitude). Position on the surface of a sphere (constantr) is described by the
two anglesand. The Cartesian and spherical polar coordinates are linked as
illustrated inFig. 1.1(b)by the relationships
xr sin cos yr sin sin zr cos
(1:1)
1.2 Complex numbers
The numbers we most commonly use in daily life are realnumbers. Some of
them are also rationalnumbers. This means that they can be expressed as the
quotient of two integers, with the condition that the denominator of the quotient
must not equal zero. When the denominator is 1, the real number is an integer.
Thus 4, 4/5, 123/456 are all rational numbers. A real number can also be
irrational, which means it cannot be expressed as the quotient of two integers.
1
-
7/26/2019 Guide to Geophysical Equations
17/296
Familiar examples are , e (the base of natural logarithms), and some square
roots, such as2,3,5, etc. The irrational numbers are real numbers that do
not terminate or repeat when expressed as decimals.
In certain analyses, such as determining the roots of an equation, it is
necessary to nd the square root of a negative real number, e.g. (y2), where
yis real. The result is an imaginarynumber. The negative real number can be
written as (1)y2, and its square root is then(1)y. The quantity(1) is written
iand is known as the imaginary unit, so that(y2) becomes iy.
A complex number comprises a real part and an imaginary part. For example,
z = x+ iy, in whichx and y are both real numbers, is a complex number with a
real part xand animaginary part y. The composition of a complex number can
be illustrated graphically with the aid of thecomplex plane(Fig. 1.2). The real
part is plotted on the horizontal axis, and the imaginary part on the vertical axis.
The two independent parts are orthogonal on the plot and the complex numberz
r
z
xy
(a)
r
y= rsinsin
x= rsincos
z= rcos
rsin
(b)
Fig. 1.1. (a) Cartesian and spherical polar reference systems. (b) Relationshipsbetween the Cartesian and spherical polar coordinates.
+x
+y z= x+ iy
Realaxis
Imaginaryaxis
r
rcos
rsin
Fig. 1.2. Representation of a complex number on an Argand diagram.
2 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
18/296
is represented by their vector sum, dening a point on the plane. The distancer
of the point from the origin is given by
r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffix2 y2p (1:2)The line joining the point to the origin makes an angle with the real (x-)axis,
and so rhas real and imaginary components rcos and rsin , respectively. The
complex numberzcan be written in polar form as
zr cos isin (1:3)It is often useful to write a complex number in the exponential form introduced
by Leonhard Euler in the late eighteenth century. To illustrate this we make useof innite power series; this topic is described inSection 1.10. The exponential
function, exp(x), of a variablexcan be expressed as a power series as in (1.135).
On substitutingx = i, the power series becomes
exp i 1 i i 2
2! i
3
3! i
4
4! i
5
5! i
6
6!
1
i 2
2! i 4
4! i 6
6! i
i 3
3! i 5
5! 1
2
2!
4
4!
6
6!
i
3
3!
5
5!
(1:4)
Comparison with (1.135) shows that the rst bracketed expression on the
right is the power series for cos ; the second is the power series for sin .
Therefore
exp i cos isin (1:5)On inserting (1.5) into (1.3), the complex numberzcan be written in exponential
form as
zr exp i (1:6)The quantityris themodulusof the complex number and is itsphase.
Conversely, using (1.5) the cosine and sine functions can be dened as the
sum or difference of the complex exponentials exp(i) and exp(i):
cos exp i expi 2
sin exp i expi 2i
(1:7)
1.2 Complex numbers 3
-
7/26/2019 Guide to Geophysical Equations
19/296
1.3 Vector relationships
A scalar quantity is characterized only by its magnitude; a vector has bothmagnitude and direction; a unit vector has unit magnitude and the direction of
the quantity it represents. In this overview the unit vectors for Cartesian
coordinates (x, y, z) are written (ex, ey, ez); unit vectors in spherical polar
coordinates (r,,) are denoted (er,e,e). The unit vector normal to a surface
is simply denotedn.
1.3.1 Scalar and vector products
The scalar product of two vectors a and b is dened as the product of their
magnitudes and the cosine of the angle between the vectors:
a bab cos (1:8)If the vectors are orthogonal, the cosine of the angle is zero and
a b0 (1:9)The vector product of two vectors is another vector, whose direction is perpen-
dicular to both vectors, such that a right-handed rule is observed. The magnitudeof the vector product is the product of the individual vector magnitudes and the
sine of the angle between the vectors:
a bj j ab sin (1:10)Ifa and b are parallel, the sine of the angle between them is zero and
a b0 (1:11)
Applying these rules to the unit vectors (ex, ey, ez), which are normal to each
other and have unit magnitude, it follows that their scalar products are
ex ey ey ez ez ex0ex ex ey ey ez ez1
(1:12)
The vector products of the unit vectors are
ex
ey
ez
ey ez exez ex eyex ex ey ey ez ez0
(1:13)
4 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
20/296
A vectorawith components (ax,ay,az) is expressed in terms of the unit vectors
(ex,ey,ez) as
aaxex ayey azez (1:14)The scalar product of the vectorsaand bis found by applying the relationships
in (1.12):
a b axex ayey azez
bxex byey bzez
axbx ayby azbz (1:15)The vector product of the vectorsa and b is found by using (1.13):
a b axex ayey azez bxex byey bzez aybz azby
ex azbx axbz ey axby aybx
ez (1:16)
This result leads to a convenient way of evaluating the vector product of two
vectors, by writing their components as the elements of a determinant, as
follows:
a
b
ex ey ez
ax ay az
bx by bz (1:17)
The following relationships may be established, in a similar manner to the
above, for combinations of scalar and vector products of the vectors a,b, andc:
a b c b c a c a b (1:18)
a b c b c a c a b (1:19)
a
b
c
b c a
a b c
(1:20)
1.3.2 Vector differential operations
The vector differential operator is dened relative to Cartesian axes (x,y,z)
as
r ex x
ey y
ez z
(1:21)
The vector operator determines the gradient of a scalar function, which may
be understood as the rate of change of the function in the direction of each of the
reference axes. For example, the gradient of the scalar function with respect to
Cartesian axes is the vector
1.3 Vector relationships 5
-
7/26/2019 Guide to Geophysical Equations
21/296
r ex x
ey y
ez z
(1:22)
The vector operator can operate on either a scalar quantity or a vector. The
scalar product of with a vector is called thedivergenceof the vector. Applied
to the vectora it is equal to
r a ex x
ey y
ez z
axex ayey azez
axx
ayy
azz
(1:23)
If the vectora is dened as the gradient of a scalar potential , as in(1.22), we
can substitute potential gradients for the vector components (ax, ay, az). This
gives
r r x
x
y
y
z
z
(1:24)
By convention the scalar product ( ) on the left is written 2. The resulting
identity is very important in potential theory and is encountered frequently. In
Cartesian coordinates it is
r2 2
x2
2
y2
2
z2 (1:25)
The vector product of with a vector is called the curl of the vector. The
curl of the vector a may be obtained using a determinant similar to (1.17):
r aex ey ez
=x =y =zax ay az
(1:26)In expanded format, this becomes
r a azy
ayz
ex ax
z az
x
ey ay
x ax
y
ez (1:27)
The curl is sometimes called the rotation of a vector, because of its physical
interpretation (Box 1.1). Some commonly encountered divergence and curl
operations on combinations of the scalar quantity and the vectors a and b
are listed below:
r a r a r a (1:28)
6 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
22/296
r a b b r a a r b (1:29)
r a r a r a (1:30)
r a b ar b br a a r b b r a (1:31)
r r 0 (1:32)
Box 1.1. The curl of a vector
The curl of a vector at a given point is related to the circulation of the vector
about that point. This interpretation is best illustrated by an example, in
which a uid is rotating about a point with constant angular velocity .
At distancerfrom the point the linear velocity of the uidv is equal to r.
Taking the curl ofv, and applying the identity (1.31) with constant,
r v r w r wr r w r r (1)To evaluate the rst term on the right, we use rectangular coordinates (x,y,z):
wr r w ex
x ey
y ez
z
xexyey zez w ex ex ey ey ez ez
3w (2)The second term is
w r r x x
y y
z z
xexyey zez
xex
yey
zez
w (3)
Combining the results gives
r v2w (4)
w12r v (5)
Because of this relationship between the angular velocity and the linear
velocity of a
uid, the curl operation is often interpreted as the rotationofthe uid. When v = 0 everywhere, there is no rotation. A vector that
satises this condition is said to beirrotational.
1.3 Vector relationships 7
-
7/26/2019 Guide to Geophysical Equations
23/296
r r a 0 (1:33)
r r a r r a r2a (1:34)
It is a worthwhile exercise to establish these identities from basic principles,especially (1.19) and (1.31)(1.34), which will be used in later chapters.
1.4 Matrices and tensors
1.4.1 The rotation matrix
Consider two sets of orthogonal Cartesian coordinate axes (x,y,z) and (x0,y0,
z0) that are inclined to each other as in Fig. 1.3. Thex0-axis makes angles (1,1,
1) with each of the (x,y,z) axes in turn. Similar sets of angles (2,2,2) and
(3,3,3) are dened by the orientations of they0- andz0-axes, respectively, to
the (x,y,z) axes. Let the unit vectors along the (x,y,z) a n d (x0,y0,z0)axesbe(n1,
n2,n3) and (m1,m2,m3), respectively. The vectorr can be expressed in either
system, i.e.,r = r(x,y,z) =r(x0,y0,z0), or, in terms of the unit vectors,
rxn1yn2 zn3x0m1y0m2 z0m3 (1:35)We can write the scalar product (r m1) as
r m1xn1 m1yn2 m1 zn3 m1x0 (1:36)The scalar product (n1 m1) = cos 1=11denes11as thedirection cosineof
thex0-axis with respect to thex-axis (Box 1.2). Similarly, (n2 m1) = c o s1 = 12
z0z
x0x
y0
y
1
1
1
2
3
2
2
3
3
m
1 n1
m2
n2
m3n3
Fig. 1.3. Two sets of Cartesian coordinate axes, (x, y, z) and (x0, y0, z0), with
corresponding unit vectors (n1, n2, n3)and(m1, m2, m3), rotated relative to each other.
8 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
24/296
and (n3 m1) = cos 1= 13dene 12and 13as the direction cosines of the
x0-axis with respect to they- andz-axes, respectively. Thus, (1.36) is equivalent to
x011x 12y 13z (1:37)On treating they0- andz0-axes in the same way, we get their relationships to the
(x,y,z) axes:
y021x 22y 23zz031x 32y 33z
(1:38)
The three equations can be written as a single matrix equation
x0y0z0
24 35 11 12 1321 22 2331 32 33
24 35 xyz
24 35M xyz
24 35 (1:39)The coefcients nm(n = 1, 2, 3; m = 1, 2, 3) are the cosines of the interaxial
angles. By denition,12=21,23=32, and31=13, so the square matrixM
is symmetric. It transforms the components of the vector in the (x, y, z)
coordinate system to corresponding values in the (x0,y0,z0) coordinate system.
It is thus equivalent to a rotation of the reference axes.
Because of the orthogonality of the reference axes, useful relationships existbetween the direction cosines, as shown inBox 1.2. For example,
11 2 12 2 13 2 cos21 cos21 cos21 1
r2 x2 y2 z2 1
(1:40)
and
1121 1222 1323cos 1cos 2 cos1cos2 cos 1cos 20
(1:41)The last summation is zero because it is the cosine of the right angle between the
x0-axis and they0-axis.
These two results can be summarized as
X3k1
mknk 1; mn0; m6n
(1:42)
1.4.2 Eigenvalues and eigenvectors
The transpose of a matrix Xwith elements nm is a matrix with elements mn(i.e., the elements in the rows are interchanged with corresponding elements in
1.4 Matrices and tensors 9
-
7/26/2019 Guide to Geophysical Equations
25/296
Box 1.2. Direction cosines
The vectorr is inclined at angles,, and, respectively, to orthogonal
reference axes (x,y,z) with corresponding unit vectors (ex,ey,ez), as inFig.
B1.2. The vectorr can be written
rxexyey zez (1)
where (x,y,z) are the components ofrwith respect to these axes. The scalar
products ofr withex,ey, andezare
z
x
r
y
ex
ey
ez
Fig. B1.2. Angles,, and dene the tilt of a vectorr relative to orthogonal
reference axes (x, y, z), respectively. The unit vectors (ex, ey, ez) dene thecoordinate system.
r exxr cos r eyyr cosr ezzr cos
(2)
Therefore, the vectorr in (1) is equivalent to
r r cos ex r cos ey r cos ez (3)
The unit vectoru in the direction ofr has the same direction as r but its
magnitude is unity:
u rr cos ex cos ey cos ezlex mey nez (4)
where (l,m,n) are the cosines of the angles that the vectorr makes with
the reference axes, and are called the direction cosines ofr. They are
useful for describing the orientations of lines and vectors.
10 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
26/296
the columns). The transpose of a (3 1) column matrix is a (1 3) row matrix.
For example, ifXis a column matrix given by
Xx
yz
24 35 (1:43)
then its transpose is the row matrix XT, where
XT x y z (1:44)The matrix equation XTMX = K, where K is a constant, denes a quadric
surface:
XTMX x y z 11 12 1321 22 2331 32 33
24 35 xyz
24 35K (1:45)The symmetry of the matrix leads to the equation of this surface:
The scalar product of two unit vectors is the cosine of the angle they form.
Letu1and u2be unit vectors representing straight lines with direction
cosines (l1,m1,n1) and (l2,m2,n2), respectively, and letbe the anglebetween the vectors. The scalar product of the vectors is
u1 u2cos l1ex m1ey n1ez
l2ex m2ey n2ez
(5)
Therefore,
cos l1l2 m1m2 n1n2 (6)
The square of a unit vector is the scalar product of the vector with itself and isequal to 1:
u u r rr2
1 (7)
On writing the unit vectoru as in (4), and applying the orthogonality
conditions from (2), we nd that the sum of the squares of the direction
cosines of a line is unity:
lex mey nez lex mey nez l2 m2 n2 1 (8)
1.4 Matrices and tensors 11
-
7/26/2019 Guide to Geophysical Equations
27/296
fx;y; z 11x2 22y2 33z2 212xy 223yz 231zxK (1:46)When the coefcientsnmare all positive real numbers, the geometric expres-
sion of the quadratic equation is an ellipsoid. The normal direction n to the
surface of the ellipsoid at the point P(x,y,z) is the gradient of the surface. Using
the relationships between (x,y,z) and (x0,y0,z0) in(1.39) and the symmetry of
the rotation matrix,nm= mnforn m, the normal direction has components
f
x2 11x 12y 13z 2x0
f
y2 21x 22y 23z 2y0
f
z2 31x 32y 33z 2z0 (1:47)
and we write
nx;y; z rf ex fx
ey fx
ez fx
(1:48)
nx;y; z 2x0exy0ey z0ez 2rx0;y0; z0 (1:49)The normalnto the surface at P(x,y,z) in the original coordinates is parallel to
the vectorr at the point (x0,y0,z0) in the rotated coordinates (Fig. 1.4).
The transformation matrixMhas the effect of rotating the reference axes from
one orientation to another. A particular matrix exists that will cause the direc-
tions of the (x0, y0, z0) axes to coincide with the (x, y, z) axes. In this case the
normal to the surface of the ellipsoid is one of the three principal axes of the
x y
z P
tangentplane
(x0,y0,z0)
(x,y,z)r
n
er
Fig. 1.4. Location of a point (x, y, z) on an ellipsoid, where the normal n to the
surface is parallel to the radius vector at the point (x0,y0,z0).
12 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
28/296
-
7/26/2019 Guide to Geophysical Equations
29/296
1.4.3 Tensor notation
Equations describing vector relationships can become cumbersome when writ-
ten in full or symbolic form. Tensor notation provides a succinct alternative wayof writing the equations. Instead of the alphabetic indices used in theprevious
section, tensor notation uses numerical indices that allow summations to be
expressed in a compact form.
Let the Cartesian coordinates (x,y,z) be replaced by coordinates (x1,x2, x3)
and let the corresponding unit vectors be (e1, e2, e3). The vectora in (1.14)
becomes
aa1e1 a2e2 a3e3 Xi1;2;3
aiei (1:56)
A convention introduced by Einstein drops the summation sign and tacitly
assumes that repetition of an index implies summation over all values of the
index, in this case from 1 to 3. The vectora is then written explicitly
aaiei (1:57)Alternatively, the unit vectors can be implied and the expression ai is under-
stood to represent the vectora. Using the summation convention, (1.15) for the
scalar product of two vectorsa and b is
a ba1b1 a2b2 a3b3aibi (1:58)Suppose that two vectorsa and b are related, so that each component ofa is a
linear combination of the components ofb. The relationship can be expressed in
tensor notation as
aiTijbj (1:59)
The indicesiandjidentify components of the vectorsaandb; each index takeseach of the values 1, 2, and 3 in turn. The quantity Tij is a second-order (or
second-rank) tensor, representing the array of nine coefcients (i.e., 32). A
vector has three components (i.e., 31) a n d i s a rst-order tensor; a scalar property
has a single (i.e., 30) value, its magnitude, and is a zeroth-order tensor.
To write the cross product of two vectors we need to dene a new quantity,
the Levi-Civitapermutation tensorijk. It has the value +1 when a permutation
of the indices is even (i.e., 123 = 231 = 312= 1) and the value 1 when a
permutation of the indices is odd (i.e., 132 = 213 = 321 = 1). If any pair of
indices is equal,ijk= 0. This enables us to write the cross product of two vectors
in tensor notation. Letu be the cross product of vectors a and b:
u a b a2b3 a3b2 e1 a3b1 a1b3 e2 a1b2 a2b1 e3 (1:60)
14 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
30/296
In tensor notation this is written
uiijkajbk (1:61)This can be veried readily for each component ofu. For example,
u1123a2b3 132a3b2a2b3 a3b2 (1:62)The tensor equivalent to the unit matrix dened in (1.55) is known as
Kroneckers symbol,ij, or alternatively theKronecker delta.It has the values
ij 1; if ij0; if i6j
(1:63)
Kroneckers symbol is convenient for selecting a particular component of a
tensor equation. For example, (1.54) can be written in tensor form using the
Kronecker symbol:
ijij
xj0 (1:64)This represents the set of simultaneous equations in (1.50). Likewise, the
relationship between direction cosines in (1.42) simplies to
mknkmn (1:65)in which a summation over the repeated index is implied.
1.4.4 Rotation of coordinate axes
Letvkbe a vector related to the coordinates xlby the tensorTkl
vkTklxl (1:66)
A second set of coordinates xn is rotated relative to the axes xl so that the
direction cosines of the angles between corresponding axes are the elements of
the tensornl:
x0nnlxl (1:67)Let the same vector be related to the rotated coordinate axes xn by the tensor
Tkn:
v0kT0knx0n (1:68)vk and vk are the same vector, expressed relative to different sets of axes.
Therefore,
v0kknvnknTnlxl (1:69)
1.4 Matrices and tensors 15
-
7/26/2019 Guide to Geophysical Equations
31/296
-
7/26/2019 Guide to Geophysical Equations
32/296
r e1 x1
e2 x2
e3 x3
ei xi
(1:78)
Several shorthand forms of this equation are in common use; for example,
xi r i;ii (1:79)
Thedivergenceof the vectora is written in tensor notation as
r a a1x1
a2x2
a3x3
aixi
iai (1:80)
Thecurl(orrotation) of the vectora becomes
r a e1 a3x2
a2x3
e2 a1
x3 a3
x1
e3 a2
x1 a1
x2
(1:81)
r a iijkak
xjijkjak (1:82)
1.5 Conservative force,eld, and potential
If the work done in moving an object from one point to another against a force is
independent of the path between the points, the force is said to be conservative.
No work is done if the end-point of the motion is the same as the starting point;
this condition is called the closed-path test of whether a force is conservative. In
a real situation, energy may be lost, for example to heat or friction, but in an
ideal case the total energyEis constant. The workdWdone against the force F is
converted into a gain dEP in the potential energy of the displaced object. Thechange in the total energydEis zero:
dEdEP dW0 (1:83)The change in potential energy when a force with components (Fx, Fy, Fz)
parallel to the respective Cartesian coordinate axes (x,y,z) experiences elemen-
tary displacements (dx,dy,dz) is
dEP dW Fx dx Fy dy Fzdz (1:84)The value of a physical force may vary in the space around its source. For
example, gravitational and electrical forces decrease with distance from a
source mass or electrical charge, respectively. The region in which a physical
1.5 Conservative force,eld, and potential 17
-
7/26/2019 Guide to Geophysical Equations
33/296
-
7/26/2019 Guide to Geophysical Equations
34/296
-
7/26/2019 Guide to Geophysical Equations
35/296
-
7/26/2019 Guide to Geophysical Equations
36/296
-
7/26/2019 Guide to Geophysical Equations
37/296
-
7/26/2019 Guide to Geophysical Equations
38/296
-
7/26/2019 Guide to Geophysical Equations
39/296
-
7/26/2019 Guide to Geophysical Equations
40/296
-
7/26/2019 Guide to Geophysical Equations
41/296
is distributed in the volume with mean density , a volume integral can replace
the enclosed mass:
ZZZV
r aG dV 4GZZZ
V
dV (1:114)
ZZZV
r aG 4G dV0 (1:115)
For this to be generally true the integrand must be zero. Consequently,
r aG
4G (1:116)
The gravitational acceleration is the gradient of the gravitational potential UGas
in (1.88):
r rUG 4G (1:117)
r2UG4G (1:118)Equation (1.118) is known as Poissons equation, after Simon-Denis Poisson
(17811840), a French mathematician and physicist. It describes the gravita-
tional potential of a mass distribution at a point that is within the mass distri-
bution. For example, it may be used to compute the gravitational potential at a
point inside the Earth.
1.9 Laplaces equation
Another interesting case is the potential at a point outside the mass distribution.
Let S be a closed surface outside the point mass m. The radius vectorrfrom the
point mass mnow intersects the surface S at two points A and B, where it forms
angles1and2with the respective unit vectorsn1andn2normal to the surface
(Fig. 1.10). Leterbe a unit vector in the radial direction. Note that the outward
normaln1forms an obtuse angle with the radius vector at A. The gravitational
acceleration at A is a1and its ux through the surface area dS1is
dN1
a1 n1dS1
Gm
r2
1 r n1
dS1 (1:119)
dN1 Gmr21
cos 1 dS1Gm cos 1dS1
r21(1:120)
26 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
42/296
dN1Gm d (1:121)The gravitational acceleration at B is a2 and its ux through the surface area dS2 is
dN2 a2 n2dS2 Gm cos 2dS2r22
(1:122)
dN2 Gm d (1:123)The total contribution of both surfaces to the gravitational ux is
dNdN1 dN20 (1:124)Thus, the total ux of the gravitational acceleration aGthrough a surface S that
does not include the point massm is zero. By invoking the divergence theorem
we have for this situationZV
r aG dVZS
aG n dS0 (1:125)
For this result to be valid for any volume, the integrand must be zero:
r aG r rUG 0 (1:126)
r2UG 0 (1:127)Equation (1.127) is Laplaces equation, named after Pierre Simon, Marquis de
Laplace (17491827), a French mathematician and physicist. It describes the
gravitational potential at a point outside a mass distribution. For example, it is
applicable to the computation of the gravitational potential of the Earth at an
external point or on its surface.
In Cartesian coordinates, which are rectilinear, Laplaces equation has the
simple form
S
m a1
n2
d
dS1
1
2
dS2n1
a2A B
er
Fig. 1.10. Representation of the gravitational ux through a closed surface S that
does not enclose the source of the ux (the point massm).
1.9 Laplaces equation 27
-
7/26/2019 Guide to Geophysical Equations
43/296
-
7/26/2019 Guide to Geophysical Equations
44/296
f0 a0; dfdx
x0
a1
d2
fdx2
x0
2a2; d3
fdx3
x0
3 2 a3dnf
dxn
x0
nn 1n 2 . . . 3 2 1 ann!an
(1:133)
On inserting these values for the coefcients into (1.130) we get the power
series for(x):
f
x
f
0
x
df
dx
x0
x2
2!
d2f
dx
2 x0
x3
3!
d3f
dx
3 x0
xn
n!
dnf
dxn
x0
(1:134)
This is the MacLaurin series for(x) about the origin,x = 0. It was derived in the
eighteenth century by the Scottish mathematician Colin MacLaurin (1698
1746) as a special case of a Taylor series.
The MacLaurin series is a convenient way to derive series expressions for
several important functions. In particular,
sin xx x3
3! x
5
5! x
7
7! 1 n1 x
2n1
2n 1 !
cos x1 x2
2! x
4
4! x
6
6! 1 n1 x
2n2
2n 2 !
expx ex 1 x x2
2! x
3
3! x
n1
n 1 !
ln 1
x
loge 1
x
x
x2
2
x3
3
x4
4 1
n1 xn
n
(1:135)
1.10.2 Taylor series
We can write the power series in (1.134) for(x) centered on any new origin, for
examplex = x0. To do this we substitute (x x0) forx in the above derivation.
The power series becomes
fx fx0 x x0 dfdx xx0
x x0 2
2!
d2f
dx2 xx0 x x0
3
3!
d3f
dx3
xx0
x x0 n
n!
dnf
dxn
xx0
(1:136)
1.10 Power series 29
-
7/26/2019 Guide to Geophysical Equations
45/296
This is called a Taylor series, after an English mathematician, Brooks Taylor
(16851731), who described its properties in 1712.
The MacLaurin and Taylor series are both approximations to the function
(x). The remainder between the true function and its power series is a measure
of how well the function is expressed by the series.
1.10.3 Binomial series
Finite series
An important series is the expansion of the function (x) = (a + x)n. Ifn is a
positive integer, the expansion of(x) is a nite series, terminating after (n + 1)
terms. Evaluating the series for some low values ofn gives the following:
n0 : a x 0 1n1 : a x 1 a xn2 : a x 2 a2 2ax x2n3 : a x 3 a3 3a2x 3ax2 x3n4 : a x 4 a4 4a3x 6a2x2 4ax3 x4
(1:137)
The general expansion of(x) is therefore
a x n an nan1x n n 1 1 2 a
n2x2
n n 1 . . . n k 1 k!
ankxk xn(1:138)
The coefcient of the general kth term is equivalent to
n n 1 . . . n k 1 k!
n!
k! n k ! (1:139)
This is called thebinomial coefcient.
When the constanta is equal to 1 and n is a positive integer, we have the
useful series expansion
1 x nXnk0
n!
k! n k !xk (1:140)
Innite series
If the exponent in (1.140) is not a positive integer, the series does not terminate,
but is an innite series. The series for(x) = (1 +x)p, in which the exponentpis
not a positive integer, may be derived as a MacLaurin series:
30 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
46/296
-
7/26/2019 Guide to Geophysical Equations
47/296
1.11 Leibnizs rule
Assume thatu(x) and v(x) are differentiable functions ofx. The derivative oftheir product is
d
dx u x v x u x dv x
dx v x du x
dx (1:144)
If we dene the operatorD = d/dx, we obtain a shorthand form of this equation:
D uv u Dv v Du (1:145)We can differentiate the product (uv) a second time by parts,
D2 uv D D uv u D2v Du Dv Dv Du v D2uu D2v 2 Du Dv v D2u (1:146)
and, continuing in this way,
D3 uv u D3v 3 Du D2v 3 D2u Dv v D3uD4 uv u D4v 4 Du D3v
6 D2u
D2v
4 D2u
Dv v D4u
(1:147)
The coefcients in these equations are the binomial coefcients, as dened in
(1.139). Thus aftern differentiations we have
Dn uv Xnk0
n!
k! n k ! Dku
Dnkv
(1:148)
This relationship is known as Leibnizs rule, after Gottfried Wilhelm Leibniz
(16461716), who invented innitesimal calculus contemporaneously with
Isaac Newton (16421727); each evidently did so independently of the other.
1.12 Legendre polynomials
LetrandRbe the sides of a triangle that enclose an angle and letube the side
opposite this angle (Fig. 1.11). The angle and sides are related by the cosine rule
u2 r2 R2 2rR cos (1:149)Inverting this expression and taking the square root gives
1
u 1
R
"1 2
r
R
!cos
r
R
!2#1=2(1:150)
32 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
48/296
-
7/26/2019 Guide to Geophysical Equations
49/296
Equation (1.156) is known as the generating function for the polynomials Pn(x).
Using this result, and substituting h = r/R and x = cos , we nd that (1.151)
becomes
1
u 1
R
X1n0
r
R
!nPn cos (1:157)
The polynomialsPn(x) orPn(cos ) are called Legendre polynomials, after the
French mathematician Adrien-Marie Legendre (17521833). The dening
equation (1.157) is called the reciprocal-distance formula. An alternative for-
mulation is given inBox 1.4.
1.13 The Legendre differential equation
The Legendre polynomials satisfy an important second-order partial differential
equation, which is called the Legendre differential equation. To derive this
equation we will carry out a sequence of differentiations, starting with the
generating function in the form
1 2xh h2 1=2 (1:158)Differentiating this function once with respect toh gives
h x h 1 2xh h2 3=2 x h 3 (1:159)
Differentiating twice with respect to x gives
xh 1
2xh
h2 3=2 h
3
3 1
h
x
(1:160)
2
x2 3h2
x3h25
5 1
3h2
2
x2
(1:161)
Next we perform successive differentiations of the product (h) with respect to
h. The rst gives
h h h
h h x h 3 (1:162)
34 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
50/296
On repeating the differentiation, and taking (1.159) into account, we get
2
h2 h
h h x h 32
h x 2h 3
x h 3 3h x h 25 x 2h 3 2x 3h 3 3h x h 25 (1:163)
Now substitute for3, from (1.160), and 5, from (1.161), giving
Box 1.4. Alternative form of the reciprocal-distance formula
The sides and enclosed angle of the triangle in Fig. 1.11are related by the
cosine rule
u2 r2 R2 2rR cos (1)
Instead of takingRoutside the brackets as in (1.150), we can moveroutside
and write the expression foru as
1
u
1
r
1
2
R
r cos
R
r
2
" #1=2
(2)
Following the same treatment as inSection 1.12, but now withhR=randx = cos , we get
1
u1
r 1 2xh h2 1=2 1
r 1 t 1=2 (3)
wheret= 2xh h2. The function (1 t)1/2 is expanded as a binomial series,
which again gives an innite series inh, in which the coefcient ofhn
isPn(x). The dening equation is as before:
x; h 1 2xh h2 1=2 X1n0
hnPn x (4)
On substitutinghR=rand x = cos , we nd an alternative form for thegenerating equation for the Legendre polynomials:
1u1
r
X1n0
Rr
nPn cos (5)
1.13 The Legendre differential equation 35
-
7/26/2019 Guide to Geophysical Equations
51/296
2
h2 h 2x 3h 1
h
x
3h x h 2 1
3h2
2
x2
(1:164)
Multiply throughout byh:
h
2
h2 h 2x 3h
x
x h 2
2
x2
2x x
3h
x
x h 2
2
x2
(1:165)
The second term on the right can be replaced as follows, again using (1.160) and
(1.161):
3h
x3h23 1
2
2
x2
1 2xh h2 2x2
(1:166)
On substituting into (1.165) and gathering terms, we get
h 2
h2
h
x
h
2
1
2xh
h2 h i
2
x2 2x
x (1:167)
h
2
h2 h x2 1 2
x2
2x
x
(1:168)
The Legendre polynomialsPn(x) are dened in (1.156) as the coefcients ofhn
in the expansion of as a power series. On multiplying both sides of (1.156) by
h, we get
h X
1
n0 hn
1
Pnx (1:169)We differentiate this expression twice and multiply by h to get a result that can
be inserted on the left-hand side of (1.168):
h h
X1n0
n 1 hnPnx (1:170)
h
2
h2 h X
1
n0n n
1
hnP
nx
(1:171)
Using (1.156), we can now eliminate and convert (1.168) into a second-order
differential equation involving the Legendre polynomials Pn(x),
36 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
52/296
-
7/26/2019 Guide to Geophysical Equations
53/296
-
7/26/2019 Guide to Geophysical Equations
54/296
Thus
d
dx 1
x2 PlP0
nP0
l
Pn n n 1 l l 1 PlPn0 (1:186)Now integrate each term in this equation with respect to x over the range
1x 1. We get
1 x2 PlP0n P0lPn 1x1 n n 1 l l 1 Z1
x1PlPn dx0
(1:187)
The rst term is zero on evaluation of (1 x2) atx = 1; thus the second termmust also be zero. Forn l the condition for orthogonality of the Legendre
polynomials is Z 1x1
PnxPlxdx0 (1:188)
1.13.2 Normalization of the Legendre polynomials
A function is said to be normalized if the integral of the square of the function overits range is equal to 1. Thus we must evaluate the integral
R1x1 Pn x 2 dx. We
begin by recalling the generating function for the Legendre polynomials given in
(1.156), which we rewrite forPn(x) andPl(x) individually:X1n0
hnPn x 1 2xh h2 1=2
(1:189)
X1l0 h
l
Pl x 1 2xh h2 1=2 (1:190)
Multiplying these equations together gives
X1l0
X1n0
hnlPn x Pl x 1 2xh h2 1
(1:191)
Now let l= n and integrate both sides with respect to x, taking into account
(1.188):
X1n0
h2nZ1
x1Pn x 2 dx
Z1x1
dx
1 h2 2xh (1:192)
1.13 The Legendre differential equation 39
-
7/26/2019 Guide to Geophysical Equations
55/296
The right-hand side of this equation is a standard integration that results in a
natural logarithm:
Z dxa bx
1b
ln a bx (1:193)
The right-hand side of (1.192) therefore leads to
Z1x1
dx
1 h2 2xh 1
2h ln 1 h2 2xh
1
x1
1
2h ln 1 h2
2h ln 1 h2 2h (1:194)and
Z1x1
dx
1 h2 2xh 1
h 1
2ln 1 h 2 1
2ln 1 h 2
1h
ln 1 h ln 1 h (1:195)
Using the MacLaurin series for the natural logarithms as in (1.135), we get
ln 1 h h h2
2 h
3
3 h
4
4 1 n1h
n
n (1:196)
ln 1 h h h2
2 h
3
3 h
4
4 1 n1h
n
n (1:197)
Subtracting the second equation from the rst gives
Z1x1
dx
1 h2 2xh2
h h h
3
3 h
5
5
2
h
X1n0
h2n1
2n 1 (1:198)
Inserting this result into (1.192) gives
X1n0
h2nZ1
x1Pn x 2 dx 2
h
X1n0
h2n1
2n 1 (1:199)
X1n0
h2nZ1
x1Pn x 2 dx 2
2n 1
0@
1A0 (1:200)
40 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
56/296
-
7/26/2019 Guide to Geophysical Equations
57/296
x2 1 Dn2f 2x n 1 Dn1f 2 n 1n12
Dnf2nx Dn1f
2n n
1
Dnf
(1:207)
On gathering terms and bringing all to the left-hand side, we have
x2 1 Dn2f 2x Dn1f nn 1Dnf0 (1:208)Now we deney(x) such that
y x
Dnf
dn
dxn
x2
1 n (1:209)
and we have
x2 1 d2ydx2
2x dydx
nn 1y0 (1:210)
On comparing with (1.174), we see that this is the Legendre equation. The
Legendre polynomials must therefore be proportional to y(x), so we can write
Pnx cndn
dxn x
2
1 n (1:211)The quantitycnis a calibration constant. To determine cnwe rst write
dn
dxn x2 1 n dn
dxn x 1 n x 1 n (1:212)
then we apply Leibnizs rule to the product on the right-hand side of the
equation:
dndxn
x2 1 n Xnm0
n!m! n m !
dmdxm
x 1 ndnmdxnm
x 1 n (1:213)
The successive differentiations of (x 1)n give
d
dx x 1 n n x 1 n1
d2
dx2 x 1 n n n 1 x 1 n2
dn1dxn1
x 1 n n n 1 n 2 . . . 321 x 1 n! x 1 dn
dxn x 1 n n!
(1:214)
42 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
58/296
-
7/26/2019 Guide to Geophysical Equations
59/296
1 x2 ddx
P00n 22x d
dxP0n n n 1 12
d
dxPn0 (1:222)
Next, we differentiate this expression again, observing the same rules andgathering terms,
1x2 d2dx2
P00n 2x d
dxP00n
4 d
dxP0n 4x
d2
dx2P0n
n n 1 2 d
2
dx2Pn 0
(1:223)
1 x2
d2
dx2P00n 2x
d2
dx2P0n 4x
d2
dx2P0n n n 1 2 4
d2
dx2Pn0
(1:224)
1 x2 d2dx2
P00n 6xd2
dx2P0n n n 1 6
d2
dx2Pn0 (1:225)
which, as we did with (1.222), we can write in the form
1 x2 d2dx2
P00n 23xd2
dx2P0n n n 1 23
d2
dx2Pn0 (1:226)
On following step-by-step in the same manner, we get after the third
differentiation
1 x2 d3dx3
P00n 2 4 xd3
dx3P0n n n 1 3 4
d3
dx3Pn0 (1:227)
Equations (1.222), (1.226), and (1.227) all have the same form. The higher-
order differentiation is accompanied by systematically different constants. By
extension, differentiating (1.219)m times (wherem n) yields the differential
equation
1 x2 dmdxm
P00n 2 m 1 x dm
dxmP0n n n 1 m m 1
dm
dxmPn0(1:228)
Now let themth-order differentiation ofPnbe written as
dm
dxmPnx Qx
1
x2
m=2
(1:229)
Substitution of this expression into (1.228) gives a new differential equation
involving Q(x). We need to determine bothdm=dxmP0n anddm=dxmP00n , sorst we differentiate (1.229) with respect tox:
44 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
60/296
-
7/26/2019 Guide to Geophysical Equations
61/296
The functionsQ(x) involve two parameters, the degreenand orderm, and are
writtenPn,m(x). Thus
1 x2 d2dx2
Pn;m x 2x ddx
Pn;m x n n 1 m2
1 x2
Pn;m x 0
(1:237)
This is the associated Legendre equation. The solutions Pn,m(x) orPn,m(cos ),
wherex = cos , are calledassociated Legendre polynomials, and are obtained
from the ordinary Legendre polynomials using the denition ofQ in (1.229):
Pn;mx 1 x2 m=2 dm
dxmPnx (1:238)
Substituting Rodriguesformula (1.218) forPn(x) into this equation gives
Pn;mx 1 x2 m=2
2nn!
dnm
dxnm x2 1 n (1:239)
The highest power ofx in the function (x2 1)n isx2n. After 2ndifferentiations
the result will be a constant, and a further differentiation will give zero.Therefore n + m 2n, and possible values ofm are limited to the range 0 m n.
1.15.1 Orthogonality of associated Legendre polynomials
For succinctness we again writePn,m(x) as simplyPn,m. The dening equations
for the associated Legendre polynomialsPn,mand Pl,mare
1 x2 Pn;m 00 2x Pn;m 0 n n 1 m21 x2 Pn;m0 (1:240)1 x2 Pl;m 00 2x Pl;m 0 l l 1 m2
1 x2
Pl;m0 (1:241)
As for the ordinary Legendre polynomials, we multiply (1.240) by Pl,m and
(1.241) byPn,m:
1 x2
Pn;m 00Pl;m 2x Pn;m 0Pl;m n n 1 m
2
1 x2 Pn;mPl;m0(1:242)
46 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
62/296
-
7/26/2019 Guide to Geophysical Equations
63/296
-
7/26/2019 Guide to Geophysical Equations
64/296
1.16 Spherical harmonic functions
Several geophysical potential
elds
for example, gravitation and geomagnet-ism satisfy the Laplace equation. Spherical polar coordinates are best suited
for describing a global geophysical potential. The potential can vary with
distance r from the Earths center and with polar angular distance and
azimuth (equivalent to co-latitude and longitude in geographic terms)
on any concentric spherical surface. The solution of Laplaces equation in
spherical polar coordinates for a potential U may be written (see Section
2.4.5,(2.104))
UX1n0
Xnm0
Anrn Bn
rn1
amn cosm bmn sinm
Pmn cos (1:252)
HereAn,Bn,amn, and b
mn are constants that apply to a particular situation. On the
surface of the Earth, or an arbitrary sphere, the radial part of the potential of a
point source at the center of the sphere has a constant value and the variation
over the surface of the sphere is described by the functions in and . We are
primarily interested in solutions outside the Earth, for which An is zero. Also we
can set the constantBnequal toRn+1
, whereRis the Earths mean radius. Thepotential is then given by
UX1n0
Xnm0
R
r
n1amn cosm bmn sinm
Pmn cos (1:253)
Let thespherical harmonic functionsCmn ; and Smn ; be dened as
C
m
n ; cosm Pm
n cos Smn ; sinm Pmn cos
(1:254)
The variation of the potential over the surface of a sphere may be described by
these functions, or a more general spherical harmonic function Ymn ; thatcombines the sine and cosine variations:
Ymn ; Pmn cos cosm
sin
m
(1:255)
Like their constituent parts the sine, cosine, and associated Legendre
functions spherical harmonic functions are orthogonal and can be
normalized.
1.16 Spherical harmonic functions 49
-
7/26/2019 Guide to Geophysical Equations
65/296
1.16.1 Normalization of spherical harmonic functions
Normalization of the functions Cmn ; and Smn ; requires integrating thesquared value of each function over the surface of a unit sphere. The element ofsurface area on a unit sphere is d = sin dd (Box 1.3) and the limits of
integration are 0 and 0 2. The integral is
ZZS
Cmn ; 2
d Z
0
Z20
Cmn ; 2
sin dd
Z
0
Z20
cosmPmn cos 2 sin dd (1:256)
Letx = cos in the associated Legendre polynomial, so thatdx = sin dand
the limits of integration are 1 x 1. The integration becomes
Z1
x1 Z2
0
cos2md
8>:
9>=>; P
mn x
2dx Z
1
x1
Pmn x 2
dx (1:257)
Normalization of the associated Legendre polynomials gives the result in
(1.248), thus
ZZS
Cmn ; 2
d 22n 1
n m!n m ! (1:258)
The normalization of the function Smn ; by this method delivers the sameresult.
Spherical harmonic functions make it possible to express the variation of a
physical property (e.g., gravity anomalies,g(, )) on the surface of the Earth as
an innite series, such as
g ; X1n0
Xnm0
amnCmn ; bmnSmn ;
(1:259)
The coefcients amn and bmn may be obtained by multiplying the functiong ;
by Cmn ; or Smn ; , respectively, and integrating the product over thesurface of the unit sphere. The normalization properties give
50 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
66/296
-
7/26/2019 Guide to Geophysical Equations
67/296
nodal lines of latitude and the longitudinal lines separate sectors in which the
potential is greater or less than the uniform value. An example of a sectorial
spherical harmonic is Y5
5
;
, shown inFig. 1.12(b).
In the general case (m 0, n m) the potential varies with both latitude and
longitude. There aren m nodal lines of latitude andm nodal great circles (2m
meridians) of longitude. The appearance of the spherical harmonic resembles a
patchwork of alternating regions in which the potential is greater or less than the
uniform value. An example of atesseralspherical harmonic isY45 ; , whichis shown inFig. 1.12(c).
1.17 Fourier series, Fourier integrals, andFourier transforms
1.17.1 Fourier series
Analogously to the representation of a continuous function by a power series
(Section 1.10), it is possible to represent a periodic function by an innite sum
of terms consisting of the sines and cosines of harmonics of a fundamental
frequency. Consider a periodic function(t) with periodthat is dened in the
interval 0 t , so that (a)(t) is
nite within the interval; (b) (t) is periodicoutside the interval, i.e., (t + ) = (t); and (c) (t) is single-valued in the
interval except at a nite number of points, and is continuous between these
points. Conditions (a)(c) are known as the Dirichlet conditions. If they are
satised,(t) can be represented as
ft a02
X1n1
ancos nt bnsin nt (1:262)
where = 2/ and the factor 1
2 in the rst term is included for reasons ofsymmetry. This representation of (t) is known as a Fourier series. The
orthogonal properties of sine and cosine functions allow us to nd the coef-
cientsan andbn of thenth term in the series by multiplying (1.262) by sin(nt)
or cos(nt) and integrating over a full period:
an2
Z=2t
=2
ftcos nt dt
bn2
Z=2t=2
ftsin nt dt(1:263)
52 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
68/296
Instead of using trigonometric functions, we can replace the sine and
cosine terms with complex exponentials using the denitions in (1.7), i.e., we
write
cos nt exp int expint 2
sin nt exp int expint 2i
(1:264)
Using these relationships in (1.262) yields
ft X1n0
an2
exp int expint bn2i
exp int expint
X1n0
an ibn2
exp int
X1n0
an ibn2
expint
(1:265)
The summation indices are dummy variables, so in the second sum we can
replacen by n, and extend the limits of the sum to n = ; thus
ft X1n0
an ibn2
exp int
X0
n1
an ibn2
exp int
X1
n1
an an i bn bn 2
exp int (1:266)
If we denecnas the complex number
cn an
a
n
i bn
b
n
2 (1:267)the Fourier series (1.262) can be written in complex exponential form as
ft X1
n1cnexp int (1:268)
In this case the harmonic coefcientscnare given by
cn1
Z=2t=2
ftexpint dt (1:269)
1.17 Fourier series, integrals, and transforms 53
-
7/26/2019 Guide to Geophysical Equations
69/296
-
7/26/2019 Guide to Geophysical Equations
70/296
Box 1.5. Transition from Fourier series to Fourier integral
The complex exponential Fourier series for a function(t) is
ft X1
n1cnexp int (1)
where the complex coefcientscnare given by
cn1
Z=2
t=2
ftexpint dt (2)
In these expressions = 2/is the fundamental frequency and is the
fundamental period. From one value ofn to the next, the harmonic frequency
changes by = 2/, so the factor preceding the second equation can be
replaced by 1/= /(2). To avoid confusion when we insert (2) into (1), we
change the dummy variable of the integration tou, giving
cn
2Z=2
u=2fuexpinu du (3)
After insertion, (1) becomes
ft X1
n1
2
Z=2u=2
fuexpinu du
0B@
1CA
exp int
X1
n1
2
Z=2u=2
fuexp in t u du0B@
1CA (4)
We now dene the function within the integral as
F Z=2
u=2
fuexp in t u du (5)
The initial Fourier series becomes
1.17 Fourier series, integrals, and transforms 55
-
7/26/2019 Guide to Geophysical Equations
71/296
f t 12 X
1
1
F (6)
We now let the incremental frequency become very small, tending in the
limit to zero; this is equivalent to letting the periodbecome innite. The
indexnis dropped becauseis now a continuous variable; the discrete sum
becomes an integral and the functionf(t) is
f t 12
Z1
1
F d (7)
while the functionF() from (5) becomes
F Z1
u1fuexp i t u du (8)
On insertingF() into (7) we get
f t 12
Z11
Z1u1
fuexp i t u du24 35d
12
Z11
Z1u1
fuexpiu du24
35 exp it d (9)
The quantity in square brackets, on changing the variable from uback tot, is
g 12
Z1t1
ftexpit dt (10)
and the original expression can now be written
f t Z1
1g exp it d (11)
The equivalence of these two equations is known as the Fourier integral
theorem.
56 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
72/296
Fourier series that represent odd or even functions consist of sums of sines or
cosines, respectively. In the same way, there are sine and cosine Fourier integrals
that represent odd and even functions, respectively. Suppose that the function (t)
iseven, and let us replace the complex exponential in (1.272) using(1.5):
g 12
Z1t1
ftcos t isin t dt (1:273)
The sine function is odd, so, if(t) is even, the product(t)sin(t) is odd, and the
integral of the second term is zero. The product(t)cos(t) is even, and we can
convert the limits of integration to the positive interval:
g 12
Z1t1
ftcos t dt
1
Z1t0
ftcos t dt (1:274)
Thus, if(t) is even, then g() is also even. Similarly, one nds that, if(t) is
odd,g() is also odd.
Now we expand the exponential in (1.271) and apply the same conditions of
evenness and oddness to the products:
ft Z1
1g cos t isin t d
2Z1
0
g cos t d (1:275)
If we were to substitute (1.275) back into (1.274), the integration would be
preceded by a constant 2/, the product of the two constants in these equations.
Equations (1.275) and (1.274) form a Fourier-transform pair, and it does not
matter how the factor 2/ is divided between them. We will associate it here
entirely with the second equation, so that we have the pair of equations
ft Z1
0
g cos t d
g 2
Z1t0
ftcos t dt(1:276)
1.17 Fourier series, integrals, and transforms 57
-
7/26/2019 Guide to Geophysical Equations
73/296
The even functions(t) andg() areFourier cosine transformsof each other.
A similar treatment for a function (t) that is oddleads to a similar pair of
equations in which the Fourier transform g() is alsooddand
ft Z1
0g sin t d
g 2
Z1t0
ftsin t dt(1:277)
The odd functions(t) andg() areFourier sine transformsof each other.
f u r t h e r r e a d i n g
Boas, M. L. (2006).Mathematical Methods in the Physical Sciences, 3rd edn. Hoboken,
NJ: Wiley, 839 pp.
James, J. F. (2004). A Students Guide to Fourier transforms, 2nd edn. Cambridge:
Cambridge University Press, 135 pp.
58 Mathematical background
-
7/26/2019 Guide to Geophysical Equations
74/296
2
Gravitation
2.1 Gravitational acceleration and potential
The Universal Law of Gravitation deduced by Isaac Newton in 1687 describes
the force of gravitational attraction between two point masses m and Msepa-
rated by a distancer. Let a spherical coordinate system (r,,) be centered on
the point mass M. The force of attraction F exerted on the point mass m acts
radially inwards towards M, and can be written
F G mMr2
er (2:1)
In this expression, Gis the gravitational constant (6.674 21 1011 m3 kg1 s2),
eris the unit radial vector in the direction of increasing r, and the negative sign
indicates that the force acts inwardly, towards the attracting mass. The gravita-
tional accelerationaGat distanceris the force on a unit mass at that point:
aG GM
r2 er (2:2)
The acceleration aGmay also be written as the negative gradient of a gravita-
tional potentialUG
aG rUG (2:3)
The gravitational acceleration for a point mass is radial, thus the potential
gradient is given by
UGr
G Mr2
(2:4)
UG GM
r (2:5)
59
-
7/26/2019 Guide to Geophysical Equations
75/296
In Newtons time the gravitational constant could not be veried in a laboratory
experiment. The attraction between heavy masses of suitable dimensions is weak
and the effects of friction and air resistance relatively large, so the rst successful
measurement of the gravitational constant by Lord Cavendish was not made until
more than a century later, in 1798. However, Newton was able to conrm the
validity of the inverse-square law of gravitation in 1687 by using existing
astronomic observations of the motions of the planets in the solar system.
These had been summarized in three important laws by Johannes Kepler in
1609 and 1619. The small sizes of the planets and the Sun, compared with the
immense distances between them, enabled Newton to consider these as point
masses and this allowed him to verify the inverse-square law of gravitation.
2.2 Keplers laws of planetary motion
Johannes Kepler (15711630), a German mathematician and scientist, formu-
lated his laws on the basis of detailed observations of planetary positions by
Tycho Brahe (15461601), a Danish astronomer. The observations were made
in the late sixteenth century, without the aid of a telescope. Kepler found that the
observations were consistent with the following three laws (Fig. 2.1).
1. The orbit of each planet is an ellipse with the Sun at one focus.
2. The radius from the Sun to a planet sweeps over equal areas in equal intervals
of time.
S
Q
Q*
p
a
b
r P(r,)
P*
aphelion perihelion
Fig. 2.1. Illustration of Keplers laws of planetary motion. The orbit of each planetis an ellipse with the Sun at its focus (S); a,b, andpare the semi-major axis, semi-
minor axis, and semi-latus rectum, respectively. The area swept by the radius to a
planet in a given time is constant (i.e., area SPQ equals area SP*Q*); the square of
the period is proportional to the cube of the semi-major axis. After Lowrie (2007).
60 Gravitation
-
7/26/2019 Guide to Geophysical Equations
76/296
3. The square of the period is proportional to the cube of the semi-major axis of
the orbit.
The fundamental assumption is that the planets move under the inuence of acentral, i.e., radially directed force. For a planet of mass mat distancerfrom the
Sun the forceF can be written
F md2r
dt2 frer (2:6)
The angular momentumh of the planet about the Sun is
h
r
m
dr
dt (2:7)
Differentiating with respect to time, the rate of change of angular momentum is
dh
dt m
d
dt r
dr
dt
m
dr
dt
dr
dt
m r
d2r
dt2
(2:8)
The rst term on the right-hand side is zero, because the vector product of a
vector with itself (or with a vector parallel to itself) is zero. Thus
dh
dt r m
d2r
dt2 (2:9)
On substituting from (2.6) and applying the same condition, we have
dh
dt r frer fr r er 0 (2:10)
This equation means thath is a constant vector; the angular momentum of the
system is conserved. On taking the scalar product ofh and r, we obtain
r h r r mdr
dt
(2:11)
Rotating the sequence of the vectors in the triple product gives
r h mdr
dt r r 0 (2:12)
This result establishes that the vectorr describing the position of a planet isalways perpendicular to its constant angular momentum vectorh and therefore
denes a plane. Every planetary orbit is therefore a plane that passes through the
Sun. The orbit of the Earth denes theeclipticplane.
2.2 Keplers laws of planetary motion 61
-
7/26/2019 Guide to Geophysical Equations
77/296
2.2.1 Keplers Second Law
Let the position of a planet in its orbit be described by polar coordinates ( r,)
with respect to the Sun. The coordinates are dened so that the angleis zero atthe closest approach of the planet to the Sun (perihelion). The angular momen-
tum at an arbitrary point of the orbit has magnitude
h mr2d
dt (2:13)
In a short interval of time t the radius vector from the Sun to the planet
moves through a small angle and denes a small triangle. The area A of
the triangle is
DA 1
2r2 D (2:14)
The rate of change of the area swept over by the radius vector is
dA
dt lim
Dt!0
DA
Dt
limDt!0
1
2r2D
Dt
(2:15)
dA
dt
1
2 r
2d
dt (2:16)
On inserting from (2.13) we get
dA
dt
h
2m (2:17)
Thus the area swept over by the radius vector in a given time is constant. This is
Keplers Second Law of planetary motion.
2.2.2 Keplers First Law
If just the gravitational attraction of the Sun acts on the planet (i.e., we ignore the
interactions between the planets), the total energy Eof the planet is constant.
The total energy E is composed of the planets orbital kinetic energy and its
potential energy in the Suns gravitational eld:
1
2 m
dr
dt 2
1
2 mr2 d
dt 2
Gm
S
r E (2:18)
The rst term here is the planets linear (radial) kinetic energy, the second term
is its rotational kinetic energy (with mr2 being the planets moment of inertia
62 Gravitation
-
7/26/2019 Guide to Geophysical Equations
78/296
about the Sun), and the third term is the gravitational potential energy. On
writing
drdt
drd
ddt
(2:19)
and rearranging terms we get
dr
d
2d
dt
2 r2
d
dt
2 2G
S
r 2
E
m (2:20)
Now, to simplify later steps, we make a change of variables, writing
u 1r
(2:21)
Then
dr
d
d
d
1
u
1
u2du
d
r2
du
d
(2:22)
Substituting from (2.22) into (2.20) gives
r2ddt
2dud
2 r2 ddt
2 2G Sr 2Em (2
:23)
With the result of (2.13) we have
r2d
dt
h
m
r d
dt
1
r
h
m
u
h
m
(2:24)
On replacing these expressions, ( 2.23) becomes
h
m
2du
d
2 u2
h
m
2 2uGS 2
E
m (2:25)
du
d
2 u2 2uGS
m2
h2 2
Em
h2 (2:26)
The rest of the evaluation is straightforward, if painstaking. First we add a
constant to each side,
du
d
2 u2 2uGS
m2
h2 GS
m2
h2
2 2
Em
h2 GS
m2
h2
2(2:27)
2.2 Keplers laws of planetary motion 63
-
7/26/2019 Guide to Geophysical Equations
79/296
du
d
2 u GS
m2
h2
2 2
Em
h2 GS
m2
h2
2(2:28)
Next, we move the second term to the right-hand side of the equation, giving
du
d
2 2
Em
h2 GS
m2
h2
2 u GS
m2
h2
2(2:29)
du
d
2 GS
m2
h2
21
2Eh2
G2S2m3
u GS
m2
h2
2(2:30)
Now, we dene some combinations of these terms, as follows:
u0 GSm2
h2 (2:31)
e2 1 2Eh2
G2S2m3 (2:32)
Using these dened terms, (2.30) simplies to a more manageable form:
du
d 2
u2
0e
2
u u0
2
(2:33)
du
d
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiu20e
2 u u0 2
q (2:34)
The solution of this equation, which can be tested by substitution, is
u u0 1 e cos (2:35)
The angleis dened to be zero at perihelion. The negative square root in (2.34)
is chosen because, asincreases,rincreases andu must decrease. Let
p 1
u0
h2
GSm2 (2:36)
r p
1 e cos (2:37)
This is the polar equation of an ellipse referred to its focus, and is the proof of
Keplers First Law of planetary motion. The quantityeis the eccentricity of the
ellipse, whilep is thesemi-latus rectumof the ellipse, which is half the length of
a chord passing through the focus and parallel to the minor axis (Fig. 2.1).
These equations show that three types of trajectory around the Sun are
possible, depending on the value of the total energyEin (2.18). If the kinetic
64 Gravitation
-
7/26/2019 Guide to Geophysical Equations
80/296
energy is greater than the potential energy, the value ofEin (2.32) is positive,
andeis greater than 1; the path of the object is a hyperbola. If the kinetic energy
and potential energy are equal, the total energy is zero and e is exactly 1; the path
is a parabola. In each of these two cases the object can escape to innity, and the
paths are called escape trajectories. If the kinetic energy is less than the potential
energy, the total energy Eis negative and the eccentricity is less than 1. In this
case (corresponding to a planet or asteroid) the object follows an elliptical orbit
around the Sun.
2.2.3 Keplers Third Law
It is convenient to describe the elliptical orbit in Cartesian coordinates (x, y),
centered on the mid-point of the ellipse, instead of on the Sun. Dene thex-axis
parallel to the semi-major axis a of the ellipse and they-axis parallel to the semi-
minor axisb. The equation of the ellipse in Fig. 2.1is
x2
a2
y2
b2 1 (2:38)
The semi-minor axis is related to the semi-major axis by the eccentricity e, so
that
b2 a2 1 e2
(2:39)
The distance of the focus of the ellipse from its center is by denition ae. The
lengthpof the semi-latus rectum is the value ofyfor a chord through the focus.
On settingy = p and x = ae in (2.38), we obtain
p2
b2 1
ae 2
a2 1 e2
(2:40)
p2 a2 1 e2 2
(2:41)
Now consider the application of Keplers Second Law to an entire circuit of the
elliptical orbit. The area of the ellipse isab, and the period of the orbit is T, so
dA
dt
ab
T (2:42)
Using (2.17),
h
m
2ab
T (2:43)
2.2 Keplers laws of planetary motion 65
-
7/26/2019 Guide to Geophysical Equations
81/296
From (2.36) and (2.43) we get the value of the semi-latus rectum,
p
1
GS
h
m 2
1
GS
2ab
T 2
(2:44)
Substituting from (2.41) gives
a 1 e2
42a2b2
GST2
42a4
GST2 1 e2
(2:45)
After simplifying, we nally get
T2
a3
42
GS (2:46)
The quantities on the right-hand side are constant, so the square of the period is
proportional to the cube of the semi-major axis, which is Keplers Third Law.
2.3 Gravitational acceleration and the potential
of a solid sphere
The gravitational potential and acceleration outside and inside a solid spheremay be calculated from the Poisson and Laplace equations, respectively.
2.3.1 Outside a solid sphere, using Laplaces equation
Outside a solid sphere the gravitational potentialUG satises Laplaces equation
(Section 1.9). If the density is uniform, the potential does not vary with the
polar angle or azimuth . Under these conditions, Laplaces equation in
spherical polar coordinates ( 2.67) reduces to
r r2
UG
r
0 (2:47)
This implies that the bracketed quantity that we are differentiating must be a
constant,C,
r2UG
r C (2:48)
UG
r
C
r2 (2:49)
The gravitational acceleration outside the sphere is therefore
66 Gravitation
-
7/26/2019 Guide to Geophysical Equations
82/296
-
7/26/2019 Guide to Geophysical Equations
83/296
UG
r
4
3Gr (2:58)
aG r5R UG
r 4
3Gr
er (2:59)
This shows that the gravitational acceleration inside a homogeneous solid
sphere is proportional to the distance from its center. At the surface of the
sphere,r= R, and the gravitational acceleration is
aG R 4
3GR er
GM
R2 er (2:60)
where the massMof the sphere is
M4
3R3 (2:61)
To obtain the potentialinside the solid sphere, we must integrate ( 2.58). This
gives
UG 23Gr2 C2 (2:62)
The constant of integrationC2is obtained by noting that the potential must be
continuous at the surface of the sphere. Otherwise a discontinuity would exist
and the potential gradient (and force) would be innite. Equating (2.54) and (
2.62) atr= R gives
2
3GR2
C
2
GM
R
4
3GR2 (2:63)
C2 2GR2 (2:64)
The gravitational potential inside the uniform solid sphere is therefore given by
UG 2
3Gr2 2GR2 (2:65)
UG
2
3G r2 3R2 (2:66)
A schematic graph of the variation of the gravitational potential inside and
outside a solid sphere is shown in Fig. 2.2.
68 Gravitation
-
7/26/2019 Guide to Geophysical Equations
84/296
2.4 Laplaces equation in spherical polar coordinates
In the above examples the sphere was assumed to have uniform density so that
only the radial term in Laplaces equation had to be solved. This is also the case
when density varies only with radius. In the Earth, however, lateral variations of
the density distribution occur, and the gravitational potential UG is then a
solution of the full Laplace equation
1
r2
rr2
UG
r
1
r2 sin
sin
UG
1
r2 sin2
2UG
2 0 (2:67)
This equation is solved using the method ofseparation of variables. This is a
valuable mathematical technique, which allows the variables in a partial differ-
ential equation to be separated so that only terms in one variable are on one side
of the equation and terms in other variables are on the opposite side. A trial
solution forUGis
UG r; ; < r (2:68)
Here , , and are all functions of a single variable only, namelyr,, and,
respectively. Multiplying (2.67) byr2 and inserting (2.68) forUGgives
rr2