Plane Wave Propagation Problems in Electrically ... · Plane Wave Propagation Problems in...
Transcript of Plane Wave Propagation Problems in Electrically ... · Plane Wave Propagation Problems in...
Plane Wave Propagation Problems in Electrically Anisotropic
and Inhomogeneous Media with Geophysical Applications
Glenn Andrew Wilson
BAppSc (Physics) CQU
School of Microelectronic Engineering
Faculty of Engineering and Information Technology
Submitted as fulfilment of the requirements for the degree of Doctor of Philosophy
December 2002
ii
Abstract
Boundary value problems required for modelling plane wave propagation in
electrically anisotropic and inhomogeneous media relevant to the surface impedance
methods in electromagnetic geophysics are formally posed and treated. For a
homogeneous TM-type wave propagating in a half space with both vertical and
horizontal inhomogeneities where the TM-type wave is aligned with one of the
elements of the conductivity tensor, it is shown using exact solutions that the shearing
term in the homogeneous Helmholtz equation for inclined anisotropic media:
.0cossin 11 2
cossinsincos
,,
2
,
2,
2
,
2
,
2
2,
2
,
2
,
2
=−∂∂
∂
−+
∂∂
++
∂∂
+
mxmx
mmmtn,m
mx
mt
m
mn
mmx
mt
m
mn
m
Hjzy
Hσ
zH
yH
ωµαασ
σα
σα
σα
σα
unequivocally vanishes and solutions need only be sought to the homogeneous
Helmholtz equation for biaxial media:
.0cossinsincos
,2,
2
,
2
,
2
2,
2
,
2
,
2
=−∂
∂
++
∂∂
+ mx
mx
mt
m
mn
mmx
mt
m
mn
m HjzH
yH
ωµσ
ασ
ασ
ασ
α
This implies that those problems posed with an inclined uniaxial conductivity
tensor can be identically stated with a fundamental biaxial conductivity tensor,
provided that the conductivity values are the reciprocal of the diagonal terms from the
Euler rotated resistivity tensor:
mxxmxxmx ,1
,, σρσ == − ,
mzz
mzymyzmyymyy
mn
m
mt
mmy
,
,,,
1,
1
,
2
,
2
,sincos
σσσ
σρσ
ασ
ασ −==
+= −
−
,
myy
myzmzymzzmzz
mn
m
mt
mmz
,
,,,
1,
1
,
2
,
2
,cossin
σσσ
σρσ
ασ
ασ −==
+= −
−
.
iii
The applications of this consequence for numerical methods of solving arbitrary
two-dimensional problems for a homogeneous TM-type wave is that they need only to
approximate the homogeneous Helmholtz equation and neglect the corresponding
shearing term. The self-consistent impedance method, a two-dimensional finite-
difference approximation based on a network analogy, is demonstrated to accurately
solve for problems with inclined uniaxial anisotropy using the fundamental biaxial
anisotropy equivalence.
The problem of a homogeneous plane wave at skew incidence upon an inclined
anisotropic half space is then formally treated. In the half space, both TM- and TE-type
waves are coupled and the linearly polarised incident TM- and TE-type waves reflect
TE- and TM-type components. Equations for all elements of the impedance tensor are
derived for both TM- and TE-type incidence. This offers potential as a method of
predicting the direction of anisotropic strike from tensor impedance measurements in
sedimentary environments.
iv
Statement of Originality
This work has not previously been submitted for any other degree or diploma at
this or any other University, College or Institution. To the best of my knowledge and
belief, this thesis contains no material previously published or written by any other
person, except where due reference and acknowledgment is made in the thesis itself.
___________________________________
Glenn Andrew Wilson
v
Run rabbit run!
Dig that hole forget the sun,
And when at last the work is done,
Don�t sit down. It�s time to start another one.
Breathe, by Pink Floyd, 1973.
vi
List of Contents
Abstract�����������������������..��.. ii Statement of Originality�����������������.�.� iv
List of Contents���������������������..�.. vi
List of Figures���������������������..�� ix
Acknowledgements�������������������..�� xiii
Introduction�������������...��������...�� 1
0.1. Overview����.����������������� 1
0.2. Outline of thesis�������.�����������. 3
0.3. Detailed outline of thesis���.������������3
0.4. Publications��������������������. 7
0.5. References�������������������...� 7
Chapter 1. Introductory theory����������������...9
1.1. The Maxwell equations����������������9
1.1.1. The Maxwell equations������������.9
1.1.2. Auxiliary potential functions����������14
1.1.3. Boundary conditions������������� 15
1.1.4. Frames of reference�������������.. 16
1.1.5. Further notation and definitions��������... 19
1.2. Surface impedance�����������������.. 21
1.3. Electrical anisotropy�����������������23
1.4. References�������������������...� 27
Chapter 2. Exact solutions for one-dimensional electrical anisotropy
problems�������������..�������... 30
2.1. Introduction��������������������. 30
2.2. The problem of a homogeneous half space��������. 33
2.2.1. The Chetaev method of functional rotation�..���34
2.2.2. The method of tensor rotation���������.. 38
2.2.3. Discussion����������������� 43
2.3. The problem of a horizontally layered half space���.�.�. 43
2.4. References������������������..��. 46
vii
Chapter 3. Exact solutions for two-dimensional electrical anisotropy
problems�������������.�������... 47
3.1. Introduction��������������������. 47
3.2. Exact formulation: preliminary considerations��..����. 48
3.3. Exact formulation: perfectly insulating basement solution�.�52
3.3.1. Inclined anisotropic conductivity������...�.. 52
3.3.2. Fundamental anisotropic conductivity���.��� 62
3.4. Exact formulation: perfectly conducting basement solution...�65
3.4.1. Inclined anisotropic conductivity�����...��.. 65
3.4.2. Fundamental anisotropic conductivity���.�...�. 75
3.5. Discussion������������������..��. 78
3.6. References������������������.��.. 79
Chapter 4. Approximate solutions for two-dimensional electrical
anisotropy problems����.������..�����.. 81
4.1. Introduction��������������������. 81
4.2. Formulation: self-consistent impedance method�����.�82
4.3. Results: inclined anisotropy in inhomogeneous media��..�. 88
4.3.1. Homogeneous layer above a perfect electric
conductor������������.�����.88
4.3.2. Horizontal homogeneous layers above a perfect
electric conductor���������.����� 92
4.3.3. Vertical dyke embedded in a homogeneous layer
above a perfect electric conductor�������... 92
4.4. Discussion���������������.�����.. 93
4.5. References���������������..�����. 94
Chapter 5. Exact solutions for three-dimensional electrical anisotropy
problems������������������...��...95
5.1. Introduction��������������������. 95
5.2. General solutions for the electromagnetic fields in an
anisotropic medium�����������������. 96
5.3. General solutions for the electromagnetic fields in air.���.. 104
5.4. The problem of a TE-type incident field��������.� 105
5.5. The problem of a TM-type incident field�����.��..�.108
5.6. Discussion�����������������...��� 110
5.7. Conclusions�������������..����...�� 121
viii
5.8. References��....���������������...�� 121
Chapter 6. Conclusions��������������..�����.. 123
6.1. Contributions��������������.���.��.123
6.2. Future research�������..�����������.. 125
6.2.1. Geophysical applications�����������..125
6.2.2. Complex media.���������..�����. 127
6.3. References�����������.���������.. 127
Appendix 1. Co-ordinate rotations��������������..�129
Appendix 2. Tensor rotation������������������132
Appendix 3. Solutions to one-dimensional magnetic field coefficients� 133
Appendix 4. Solutions to the Helmholtz equation�������..�� 135
Appendix 5. Solution to Fourier series coefficients������..��. 138
Appendix 6. Finite-difference operators�����������.�� 141
Appendix 7. Inhomogeneous plane waves�����.�������.143
ix
List of Figures
Figure 1.1. Fundamental co-ordinate system for the solution to
problems�����������������..���17
Figure 1.2. Inclined co-ordinate system relative to the fundamental
co-ordinate system with Euler angles about the x-axis
indicated���������������...����.. 18
Figure 1.3. Inclined co-ordinate system relative to the fundamental
co-ordinate system with Euler angles about the x-axis
indicated���������������...����.. 18
Figure 1.4. Inclined co-ordinate system relative to the fundamental
co-ordinate system with Euler angles about the x-axis
indicated���������������...����.. 19
Figure 2.1. Effective conductivity for 0 calculated using o90≤≤ mα
equations (2.1) and (2.2) for a homogeneous half space
where 001.0=tσ S/m and 4=λ ����������..31
Figure 2.2. Surface impedance and magnitude of a homogeneous
TM-type wave incident upon a homogeneous half space
with the anisotropic parameters 001.0=tσ S/m, 4=λ
and ������������..������. 32 o30=α
Figure 2.3. Surface impedance and magnitude of a homogeneous
TM-type wave incident upon a two-layered half space
where the upper layer is isotropic with conductivity of 0.1
S/m and 10 km thick. The bottom layer, of infinite depth
extent, has the anisotropic parameters 001.0=tσ S/m,
4=λ and ����������������.33 o30=α
Figure 2.4. Geometry for the inclined anisotropic half space. The
source fields are incident parallel to the x-axis. The z-axis
is normal to the surface of the earth�����.����.. 34
Figure 2.5. Geometry for a two-layer horizontally stratified half space
where each layer exhibits inclined uniaxial anisotropic
conductivity��������������.��.��..45
Figure 3.1. Geometry and parameters of the dyke embedded in a
homogeneous layer����������.������48
x
Figure 3.2. Fourier series representations, satisfying the )(zgm
boundary conditions that for and
�����������.�.�.��.�����.. 53
0)(, =zH amx 0=z
hz =
Figure 3.3. Fourier series representations, satisfying the )(zgm
boundary conditions that for and 0)(, =zH amx 0=z
0)(, =
∂∂
zhH a
mx at z = h��������.�..�����..67
Figure 4.1. Schematic diagram in the y-z plane showing the numbering
convention for the impedance mesh. The dotted line shows
the integration path in the x-z plane for the application of
Ampere�s Law�������������..����.. 84
Figure 4.2. Surface impedance magnitude and phase response for an
anisotropic layer of 1 000 m thickness with 001.01, =tσ
S/m and 01.01, =nσ S/m above a perfectly conducting
basement. The anisotropic conductivity tensor is rotated
through the range . The impedance method o900 ≤≤ α
solutions for 10 Hz (+), 100 Hz (x), 1 kHz (o) and 10 kHz
(*) are presented. The exact solutions for 10 Hz (dotted),
100 Hz (solid), 1 kHz (dash-dotted) and 10 kHz (double-
dashed) are also shown. Note that at 1 kHz and 10 kHz, the
surface impedance is equal to the intrinsic impedance of the
layer and the phase is equal to 45o���...������.. 89
Figure 4.3. Surface impedance magnitude and phase response for an
anisotropic layer of 20 m thickness with 01, 15εε =n ,
01.01, =nσ S/m, 01, 5εε =t and 001.01, =nσ S/m above a
perfect electrically. The anisotropic complex conductivity
tensor is rotated through the range . The o900 ≤≤ α
impedance method solutions for 104 Hz (o), 105 Hz (*) and
106 Hz (∆) are presented. The exact solutions for 104 Hz
(solid), 105 Hz (dashed) and 106 Hz (double dashed) are
also shown. Note that at due to the effects of displacement
currents at 106 Hz, the surface impedance is equal to the
xi
intrinsic impedance of the layer and the phase is not equal
to 45o����������������..����� 90
Figure 4.4. Surface impedance magnitude and phase response at 100
Hz for two anisotropic layers overlying a perfectly
conducting basement using the self-consistent impedance
method (o) and corresponding exact solutions (solid line).
The upper layer thickness was varied and the total depth
was kept constant at 1 000 m. The upper layer has
σt,1 = 0.01 S/m, σn,1 = 0.001 S/m and α1 = 30o. The lower
layer has σt,2 = 0.05 S/m, σn,2 = 0.5 S/m and α2 = 60o�..... 91
Figure 4.5. Surface impedance magnitude and phase response at a
single boundary at 10 kHz for a 500 m wide vertical dyke
with 01.01, =tσ S/m and 001.01, =nσ S/m embedded in an
otherwise homogeneous layer with 001.01, =tσ S/m and
01.01, =nσ S/m, terminated with a perfectly conducting
basement. The depth of the dyke and layer is 200 m. The
impedance method response (dots) is shown above the
exact response (solid line)������.�������. 93
Figure 5.1. Geometry for the inhomogeneous plane wave incident at a
skew angle θ to the anisotropic half space inclined at angle
α about the x-axis����������������.. 97
Figure 5.2. xyZ of a 10 kHz homogeneous TM-type plane wave as a
function of θ for a half space with 001.0=tσ S/m,
01.0=nσ S/m, and =α 45o���..��������� 112
Figure 5.3. of a 10 kHz homogeneous TM-type plane wave as a xyZ∠
function of θ for a half space with 001.0=tσ S/m,
01.0=nσ S/m, and =α 45o�����..������� 113
Figure 5.4. yyZ of a 10 kHz homogeneous TM-type plane wave as a
function of θ for a half space with 001.0=tσ S/m,
01.0=nσ S/m, and =α 45o�����..������� 114
xii
Figure 5.5. of a 10 kHz homogeneous TM-type plane wave as a yyZ∠
function of θ for a half space with 001.0=tσ S/m,
01.0=nσ S/m, and =α 45o�������..����� 115
Figure 5.6. Angle of orientation of the surface impedance meter ψ
with respect to the skew angle of incidence θ and strike
of the anisotropic half space in the xy plane������.. 116
Figure 5.7. Normalized observed (measured) magnetic field of a 10 kHz
homogeneous TM-type plane wave as a function of ψ for
a half space with 001.0=tσ S/m, 01.0=nσ S/m, =α 45o
and θ =60o�������������������. 117
Figure 5.8. Normalized observed (measured) electric field of a 10 kHz
homogeneous TM-type plane wave as a function of ψ for
a half space with 001.0=tσ S/m, 01.0=nσ S/m, =α 45o
and θ =60o�������������������.. 118
Figure 5.9. Normalized observed (measured) magnetic field of a 19.8
kHz VLF wave as a function of ψ measured at different
times of day at the Callide Mine, Trap Gully B Area in July
1997 from [5.13]����������������� 119
Figure 5.10. Normalized observed (measured) electric field of a 19.8
kHz VLF wave as a function of ψ measured at different
times of day at the Callide Mine, Trap Gully B Area in July
1997 from [5.13]����������������� 120
xiii
Acknowledgments
During this thesis, I�ve been fortunate to have the opportunity to meet and work
with many great and inspiring people.
Foremost, my grateful acknowledgments to Professor D. V. Thiel for his
ongoing professional and personal encouragement, guidance and commitment to
supervise this thesis as Principal Supervisor, and to Dr S. G. O�Keefe for his role as
Associate Supervisor. I am most appreciative to Dr P. G. Keleher for taking on the role
of External Associate Supervisor whilst I was employed in Central Queensland during
2000. During 2002, I visited the Department of Geology and Geophysics, University of
Utah, Salt Lake City, as an exchange graduate student. I am indebted to Professor M. S.
Zhdanov for his hospitality and guidance, as well as for accepting the role of External
Associate Supervisor. Special thanks are also extended to Mr W. J. F. Nichols of
Callide Coalfields Pty Ltd, Biloela, for his unwavering support during the course of this
candidature.
As a self-funded candidate, there is absolutely no way this thesis could have
been completed without the financial support of several organisations, to which I am all
very grateful and would like to acknowledge here. Callide Coalfields Pty Ltd have
financially supported this work and their funding was primarily responsible for
enabling the exchange to the University of Utah. The International Centre of Griffith
University provided a long stay scholarship as part of the International Experience
Incentive Scheme. In 2001, I was awarded an Australian Institute of Geoscientists
Postgraduate Bursary and I acknowledge that support here.
I am also indebted to a great number of teachers, colleagues, associates and
fellow students, from whom along the way, I have learnt from. I wish to firstly
acknowledge my gratitude to those organisations which provided me with employment
and study opportunities during this thesis; Central Queensland University, Callide
Coalfields Pty Ltd, Griffith University and the University of Utah; but importantly,
those people to whom I have had the pleasure to meet, know and work with: Drs N. G.
Golubev, D. A. James, and Messrs A. Chernyavskiy, G. J. Durnan, H. Ebersbach, B. R.
Hanna, Y. Hill, A. V. Gribenko, S. A. Mehanee, S. Mukherjee, E. Peksen, S. A. Saario,
R. W. Schlub and E. Tolstaya.
Finally, but certainly not least, to my family for their on-going support and
encouragement, which is always appreciated.
1
Introduction
0.1. Overview
The importance of studying electrically anisotropic and inhomogeneous media
is of direct relevance to a wide range of electrodynamic disciplines, such as photonic
crystals [0.9] and inductive prospecting methods [0.10]. A variety of methods for
solving the boundary value problems associated with the electrodynamics of
anisotropic and inhomogeneous media have been developed over the last half-century.
These methods can be broadly described as being one of either:
• exact (or analytical) solutions to partial differential equations with boundary
value conditions;
• approximate (or numerical) solutions to partial differential equations with
boundary value conditions;
• simulations using RC network analogies; or
• high frequency scale modelling using electrolytic troughs.
In this thesis, only exact and approximate methods for the solutions to boundary
value problems related to plane wave propagation in electrically anisotropic and
inhomogeneous media with specific applications to low frequency electromagnetic
geophysical methods will be considered.
The first comprehensive catalogue of exact solutions for plane wave
propagation problems in electrically anisotropic and inhomogeneous media was
detailed by Porstendorfer [0.11, 0.12]. These same catalogues of models also detail
some of the earlier works on approximate and scale modelling solutions. The most
comprehensive catalogue and study of approximate solutions for plane wave
propagation problems in electrically inhomogeneous media, including finite-difference,
finite-element and integral equation methods, was conducted as part of the Comparison
of Modelling Methods for Electromagnetic Induction Problems (COMMEMI) project
[0.23]. However, no studies of approximate solutions for problems including
electrically anisotropic media were studied as part of the COMMEMI project.
2
In the geophysical community, both academic and industrial, there is now
significant interest in the development of three-dimensional approximate solutions to
problems including both electrically anisotropic and inhomogeneous media. It may be
argued that studying exact solutions to the electrodynamics of plane wave propagation
in anisotropic and inhomogeneous media for a few ideal models will not yield
significant information about the more general induction problem of arbitrarily
anisotropic and inhomogeneous media. However, the viewpoint is taken that the
principles one needs to understand the more complicated induction problems can be
developed one by one by first handling the simpler class of induction problems.
Methods of exactly solving the boundary value problems for the
electrodynamics of isotropic and inhomogeneous media for geophysical applications
have existed circa 1950. Following these early works, the mathematical foundations for
solving the exact boundary value problems for the electrodynamics of anisotropic and
inhomogeneous media for geophysical applications were developed circa 1960.
Literally, hundreds of references may be cited and in view of the great diversity of
references available in the literature, it would be difficult to cite every author who has
contributed to the discipline over the last half century. However, it is important to
mention here some of the outstanding developments in the literature; namely:
• A. N. Tikhonov�s [0.17] exact solution of a horizontal electric dipole on the
surface of an anisotropic half space;
• the extensive works of D. N. Chetaev [0.2-0.8] devoted to the electrodynamics
of anisotropic media; and
• the tensor surface impedance concept for magnetotelluric prospecting by T.
Cantwell [0.1] and I. I. Rokityanski [0.14].
These particular works have had an especially strong influence on the studies of
wave propagation in anisotropic media for geophysical applications, and specifically,
this thesis relies heavily on the works of D. N. Chetaev for developing exact solutions
to plane wave propagation problems in electrically anisotropic and inhomogeneous
media. However, it is recognised that exact solutions do have limitations, particularly
when one is trying to simulate the propagation of fields in complicated models. It is for
this reason that interest in approximate solutions has increased significantly in recent
years. However, this thesis only contains part of the material that constitutes
approximate solutions to anisotropic electrodynamic problems. For investigations of
3
approximate solutions, this thesis will only consider a two-dimensional finite-difference
method. Discussions and comparisons to corresponding finite-element methods are
presented. The theories of integral equations are not considered in this thesis.
0.2. Outline of thesis
This thesis structurally consists of two parts:
(1) an exact and approximate study of inclined and coplanar electrical anisotropy in
one- and two-dimensional wave propagation problems via vector function
solutions to boundary value problems, with particular reference to homogeneous
plane wave excitation (Chapters 2, 3 and 4). Consideration is only given to
those problems where the source field is incident parallel to one of the axes of
the conductivity tensor, allowing the TE- and TM-type waves to propagate
independently; and
(2) an exact study of the problem of inclined and coplanar electrical anisotropy at a
skew incidence to an incident homogeneous plane wave, derived using potential
function solutions to boundary value problems (Chapter 5). Consideration is
given to both TE- and TM-type source fields and equations for the coupled
reflected and transmitted waves are derived. Results from very low frequency
surface impedance surveys are also presented to verify the formulations
obtained.
0.3. Detailed outline
This thesis consists of six chapters. In some detail, we now outline the relevant
material in each chapter.
Chapter 1.
Electromagnetic fields are defined by the Maxwell equations. There exist a wide
variety of definitions and nomenclature used in the different scientific literatures, which
may lead to some confusion for the reader of this thesis. Due to constant referral to the
Maxwell equations throughout this thesis, and those formulae and conditions derived
4
from them, this thesis commences with a formal treatment of the equations of
electrodynamics, followed by brief reviews of the surface impedance and electrical
anisotropy concepts. Thus, Chapter 1 is auxiliary to the later chapters of this thesis,
where the different classes of plane wave propagation problems are treated.
Chapter 2.
In this chapter, we consider the surface impedance problem of a homogeneous
plane wave incidence upon a homogeneous half space with inclined electrical
anisotropy. It is assumed that the source field is polarised such that it is aligned parallel
to one of the elements of the inclined conductivity tensor, meaning that the TE- and
TM-type waves propagate independently. First, Chetaev�s method of functional
rotation [0.2] is considered, where the Maxwell equations are considered in the inclined
co-ordinate system and rotated into the fundamental co-ordinate system. The method of
tensor rotation is then considered where the inclined conductivity tensor is rotated into
the fundamental co-ordinate system with an Euler rotation.
It is shown that the class of problem with an inclined conductivity tensor is
equivalent to a class of problem with a biaxial conductivity tensor in the fundamental
co-ordinates, where:
zz
zyyzyyyy
nty σ
σσσρ
σα
σασ −==
+= −
−1
122 sincos , (0.1)
and
yy
yzzyzzzz
ntz σ
σσσρ
σα
σασ −==
+= −
−1
122 cossin . (0.2)
It is then shown that for homogeneous TM-type incidence, the surface
impedance depends only upon the conductivity given by equation (0.1). For incidence
of the TE-type, the surface impedance is independent of the anisotropy parameters.
These formulations are extended to the problem of a horizontally stratified half space
using Wait�s [0.18] transmission line analogy. This chapter has been published as
[0.19].
5
Chapter 3.
In the previous chapter, it was shown that for homogeneous TM-type waves, the
problem of an inclined conductivity tensor is equivalent to a class of problem with a
biaxial conductivity tensor in the fundamental co-ordinates, provided that the medium
is horizontally homogeneous. In this chapter, Chetaev�s method of functional rotation is
extended to consider the fields of a homogeneous plane wave in a laterally
inhomogeneous medium. This exact solution is based on Rankin�s [0.13] Fourier series
solution for TM-type incidence. TE-type incidence is not considered in this chapter
since, as was shown in Chapter 2, TE-type waves propagate independently of the
inclined anisotropic conductivity. Of principle interest in this chapter are the properties
of the anisotropic Helmholtz equation:
.0cossin 11 2
cossinsincos
2
2
222
2
222
=−∂∂
∂
−+
∂∂
++
∂∂
+
xx
tn
x
tn
x
tn
Hjzy
Hσ
zH
yH
ωµαασ
σα
σα
σα
σα
(0.3)
Using separation of variables and by imposing known boundary conditions, it is
shown that in laterally inhomogeneous media, the shearing term containing the mixed
partial derivative in equation (0.3) vanishes and only solutions to the bianisotropic
Helmholtz equation need be found:
0cossinsincos2
222
2
222
=−∂
∂
++
∂∂
+ x
x
tn
x
tn
HjzH
yH ωµ
σα
σα
σα
σα . (0.4)
From equations (0.1) and (0.2), one observes that equation (0.4) can be cast in a
form where the inclined conductivity tensor is equivalent to a biaxial conductivity
tensor in fundamental co-ordinates. The exact solution for a horizontally
inhomogeneous layer is then derived. The result is similar to the formulation by Rankin
[0.13]. The work presented in this chapter has been published as [0.21].
6
Chapter 4.
Chapter 4 commences by formally introducing the self-consistent impedance
method; a two-dimensional finite-difference method for TM-type propagation problems
developed by Thiel & Mittra [0.15]. This approximate method is presented in such
form that it can be applied to a wide range of electrodynamic problems beyond the
scope of the specific geophysical ones presented in this thesis. The formulation
presented can include displacement currents through a complex conductivity.
When other authors developed approximate methods for solving two-
dimensional TM-type propagation problems in inclined anisotropic media, they
considered approximate solutions to equation (0.3). In the formulation of the self-
consistent impedance method presented in this thesis, one is able to observe that only
approximate solutions to equation (0.4) need be sought. In previous chapters, it has
been shown that for homogeneous TM-type waves, the class of problem with an
inclined conductivity tensor is equivalent to a class of problem with a biaxial
conductivity tensor in the fundamental co-ordinates.
It is therefore concluded that by approximating equation (0.4), which is
identified as the Helmholtz equation for bianisotropic media in the fundamental co-
ordinate system, one is equivalently approximating equation (0.3), which has been
identified as the Helmholtz equation for inclined anisotropic media. Results are
compared with the corresponding exact solutions for inclined anisotropic media as
presented in Chapters 2 and 3. Whilst in this thesis, we have only chosen the self-
consistent impedance method to demonstrate this formulation, it should be noted that
the principle of the inclined anistropic/bianisotropic equivalence would hold for any
approximate method. The work presented in this chapter has been presented in papers
[0.16, 0.20, 0.22].
Chapter 5.
In this chapter, Chetaev�s method of auxiliary potentials is implemented to
solve for the fields of homogeneous plane waves at skew incidence to an inclined
anisotropic half space. This class of problem is a logical extension of the class of
problems considered in Chapter 2, which can be considered as a special class of the
solutions presented in this chapter. When a TM/TE-type field is incident at a skew
angle upon an inclined anisotropic half space, both TM- and TE-type field components
7
exist in the lower half space. As a result, both TM/TE- and TE/TM-type fields are
reflected from the surface into the upper half space. In effect, this means that a linearly
polarised field is reflected as an elliptically polarised field from an anisotropic half
space. First, the general expressions for all electromagnetic field components that exist
in the anisotropic half space are considered. Then, the separate problems of
inhomogeneous TE- and TM-type incident fields are treated, and equations for the
surface impedance tensor elements at the surface of the anisotropic half space are
derived. The response of one-dimensional surface impedance measurements are then
analysed and discussions related to the practical applications of this technique for
identifying the direction of anisotropic strike are presented. This chapter has not yet
been prepared for publication.
Chapter 6.
In this chapter, a summary of the thesis is presented. Achievements and specific
contributions to the solution of plane wave propagation problems in electrically
anisotropic and inhomogeneous media are presented. The thesis is concluded with some
brief suggestions for future research in the electrodynamics of anisotropic media.
0.4. Publications
Papers [0.19-0.21] have been prepared by the author in collaboration with the
thesis supervisors, and submitted to international journals. Papers [0.16, 0.22] have
been prepared by the author in collaboration with the thesis supervisors, and submitted
to international conferences.
0.5 References
[0.1] T. Cantwell, PhD thesis, Massachusetts Institute of Technology, 1960.
[0.2] D. N. Chetaev, Bull. Acad. Sci. USSR Geophys. Ser. 4, 407 (1960).
[0.3] D. N. Chetaev, Radioteck. Elektron. 8, 64 (1963).
[0.4] D. N. Chetaev, Phys. Solid Earth 2, 233 (1966); 2, 651 (1966).
[0.5] D. N. Chetaev, Sov. Phys. Dokl. 12, 555 (1967).
[0.6] D. N. Chetaev, in Estestvennoye Electromagnitoe Pole i Issledovaniya
Vnutrennego Stroeniya (Nauka, Moscow, 1971), pp. 15-39.
8
[0.7] D. N. Chetaev & M. G. Savin, Sov. Phy. Dokl. 12, 564 (1967).
[0.8] D. N. Chetaev & B. N. Belen�kaya, Phys. Solid Earth 7, 212 (1971); 8, 535-538
(1972).
[0.9] J. D. Joannopoulos, R. D. Meade & J. N. Winn, Photonic Crystals: Modeling
the Flow of Light (Princeton Uni. Press, Princeton, 1995).
[0.10] J. G. Negi & P. D. Saraf, Anisotropy in Geoelectromagnetism (Elsevier,
Amsterdam, 1989).
[0.11] G. Porstendorfer, Principles of Magneto-telluric Prospecting (Geopublication
Associates, Berlin, 1975).
[0.12] G. Porstendorfer, in Geoelectric and Geothermal Studies (East-Central Europe,
Soviet Asia, edited by A. Adam (Akad. Kiado, Budapest, 1976), pp. 152-164.
[0.13] D. Rankin, Geophysics 27, 666, 1962.
[0.14] I. I. Rokityanski, Bull. Acad. Sci. USSR Geophys. Ser. 10, 1050 (1961).
[0.15] D. V. Thiel & R. Mittra, Radio Sci. 36, 31 (2001).
[0.16] D. V. Thiel & G. A. Wilson, presented at URSI Nat. Sci. Meet., Boulder, CO,
Jan. 2002.
[0.17] A. N. Tikhonov, Sov. Phys. Dokl. 4, 566 (1959).
[0.18] J. R. Wait, Electromagnetic Waves in Stratified Media, 2nd ed. (Permagon Press,
Oxford, 1970).
[0.19] G. A. Wilson & D. V. Thiel, Radio Sci. 37, 1029, 2001RS002535 (2002).
[0.20] G. A. Wilson & D. V. Thiel, IEEE Trans. Geosci. Remote Sensing, submitted.
[0.21] G. A. Wilson & D. V. Thiel, J. Electromagn. Waves Applic., submitted.
[0.22] G. A. Wilson & D. V. Thiel, presented at IEEE Int. Antennas Propagat. Symp.,
San Antonio, TX, Jun. 2002.
[0.23] M. S. Zhdanov, I. M. Varenstov, J. T. Weaver, N. G. Golubev & V. A. Krylov,
J. App. Geophys. 37, 133 (1997).
9
Chapter 1
Introductory Theory
The material discussed in this thesis requires the knowledge of fundamental
concepts from electromagnetic theory. As we will continually be referring to the
Maxwell equations throughout this thesis, this chapter starts by defining the Maxwell
equations in the most general cases. We then obtain forms of these equations that are
used regularly throughout this thesis. After this, the frames of reference, further
notations and other definitions are presented. An introductory discussion on surface
impedance and electrical anisotropy in electromagnetic geophysics is then presented.
1.1. The Maxwell equations
1.1.1. The Maxwell equations
The theory for electrical geophysics has been developed from the laws
governing the behaviour of electromagnetic fields in an inhomogeneous conducting
earth. The Maxwell equations, which are the fundamental, mathematical formulations
describing the behaviour of electromagnetic fields, are the basis of this theory, e.g., see
[1.33]. For the sake of a thorough exposition, this section follows the formal
mathematical treatment of Zhdanov [1.32]. In differential form, the Maxwell equations,
excluding extraneous currents and sources in a homogeneous medium are:
t∂∂+=×∇ DJH , (1.1)
t∂∂−=×∇ BE , (1.2)
0=⋅∇ B , (1.3)
qρ=⋅∇ D , (1.4)
10
where qρ is the density of all charges in the medium. Equation (1.1) is an expression of
Ampere�s Law which relates the magnetic field intensity H with the current density J
and time-dependent displacement field D; equation (1.2) is an expression of Faraday�s
Law which relates the electric field intensity E with the time-dependent magnetic flux
density B; equation (1.3) states the non-existence of magnetic monopoles; and equation
(1.4) is Gauss� or Coulomb�s Law stating the existence of electric monopoles.
Equations (1.3) and (1.4) represent the continuous properties of the electromagnetic
fields. Equations (1.1) to (1.4) can also be expressed in integral form as:
sDJlH dt
dSC
⋅
∂∂+=⋅ ∫∫
, (1.5)
td
C ∂∂−=⋅∫φ
lE , (1.6)
0
=⋅∫SdsB , (1.7)
qdS
=⋅∫ sD , (1.8)
where q is the total charge and φ is the total flux through the surface S. As a result of
equations (1.1) and (1.4), the Maxwell equations are conservative such that:
tq
∂∂−
=⋅∇ρ
J . (1.9)
In addition to the four fundamental Maxwell equations, two constitutive
equations exist to relate the magnetic flux density B and magnetic field intensity H, and
the displacement field D and electric field intensity E vectors:
HB µ= , (1.10)
ED ε= , (1.11)
11
where µ and ε represent the permeability and permittivity of the medium. The current
density J in an extended medium is linearly proportional to the electric field vector E.
This is Ohm�s Law, and is stated as:
EJ σ= , (1.12)
where σ is the conductivity of the medium, and is the reciprocal of the resistivity ρ of
the medium. In equations (1.10) to (1.12), the terms µ, ε and σ respectively can have
either scalar or tensor properties. It is observed that equations (1.3) and (1.4) follow
from equations (1.1) and (1.2) with continuity equation (1.9).
To seek both exact and approximate solutions, it is useful to separate the
Maxwell equations (1.1) to (1.4) into equations of the electric and magnetic fields. To
do this, one can first re-write equations (1.1) and (1.2) exclusively in terms of E and H
by substituting constitutive relations (1.10) and (1.11), as well as equation (1.12). By
applying the curl operator to both sides of equation (1.2) and substituting the result into
equation (1.1), one obtains:
02
2
=∂∂+
∂∂+×∇×∇
ttEEE µσµε , (1.13)
for the electric field. Similarly, by applying the curl operator to both sides of equation
(1.1) and substituting into equation (1.2), one obtains:
02
2
=∂∂+
∂∂+×∇×∇
ttHHH µσµε , (1.14)
for the magnetic field. Using the vector identity:
EEE ∆−⋅∇∇=×∇×∇ )( ,
where ∆ is the Laplacian operator (n.b., ∇ is also commonly used in scientific
literature), and with Maxwell equation (1.4), equation (1.13) can be re-written as:
2
12
qttρ
εµσµε ∇=
∂∂−
∂∂−∆ 1
2
2 EEE . (1.15)
Free charges cannot exist in a homogeneous medium without extraneous
electric currents and sources, so equation (1.15) reduces to:
02
2
=∂∂−
∂∂−∆
ttEEE µσµε . (1.16)
One obtains a similar relationship for the magnetic field in equation (1.14),
using Maxwell equation (1.3) to obtain:
02
2
=∂∂−
∂∂−∆
ttHHH µσµε . (1.17)
Equations (1.16) and (1.17) are called the homogeneous telegraph equations for
a homogeneous medium in the absence of extraneous currents and charge. In many
problems in electromagnetic geophysics, the fields vary with time sufficiently slowly
that the ∂ term can be ignored. Such quasi-static electromagnetic fields satisfy
the Maxwell equations:
22 / t∂
JH =×∇ , (1.18)
t∂∂−=×∇ BE , (1.19)
0=⋅∇ B , (1.20)
and the continuity condition ∇ for an inhomogeneous medium with conductivity
gradients:
0=⋅ J
σσ
∇⋅−=⋅∇ EE 1 , (1.21)
13
in the absence of extraneous currents and sources. Equations (1.18) to (1.21) lead to the
homogeneous diffusion equations:
0=∂∂−∆
tEE µσ , (1.22)
0=∂∂−∆
tHH µσ . (1.23)
If one considers the Maxwell equations for a perfect electric insulator (i.e.,
0=σ ), then equation (1.1) takes the form of:
t∂∂=×∇ DH . (1.24)
In a homogeneous medium, equations (1.16) and (1.17) then take the form of
the homogeneous wave equations:
02
2
=∂∂−∆
tEE µε , (1.25)
02
2
=∂∂−∆
tHH µε . (1.26)
It is now assumed that all fields are monochromatic and vary with a time
dependence of )exp( tjω type where 1−=j . With few exceptions, any field that is
an arbitrary function of time can be synthesised using a sum of cisoidal functions since
any reasonably well-behaved function of time can be replaced with either a Fourier
integral or Fourier series of its spectral components [1.32].
In a homogeneous medium without extraneous currents and charges, the
telegraph, diffusion and wave equations all satisfy the homogeneous Helmholtz
equations:
02 =+∆ EE k , (1.27)
14
02 =+∆ HH k , (1.28)
where k is the wave number, and has the most general expression:
ωµσεµω jk −= 22 , (1.29)
when magnetic conductivity is neglected. To prevent exponentially divergent solutions
in both E and H with increasing values of either time or distance, the values of k are
chosen such that . In a homogeneous medium, the wave number can also be
written as:
0Re >k
*2 ωµσjk −= , (1.30)
where ωεσσ j+=* is called the complex conductivity. Hence, it is possible to obtain
frequency-dependent telegraph equation solutions using the diffusion equation provided
the conductivity is complex.
1.1.2. Auxiliary potential functions
Since B is rotational in the full space, as expressed in equation (1.3), it may
therefore be defined unequivocally by specifying a vector potential function A [1.32]:
AB ×∇= . (1.31)
Since , then equation (1.3) is satisfied identically, and it is noted
that A may have both rotational and lamellar parts. The use of A requires the
specification of the rotational and lamellar parts for unique representation of the field
quantities. The rotational part is normally specified based on the properties of the field.
The lamellar part must then be specified to be consistent with the field equations.
Substituting equation (1.31) into equation (1.2), then:
0( =×∇⋅∇ A)
0)( =+×∇ AE ωj . (1.32)
Since for any scalar Φ, equation (1.32) can be integrated to obtain: 0=Φ∇×∇
15
Φ∇−−= AE ωj , (1.33)
where Φ is a scalar function of position and is called the scalar potential function.
Equation (1.15) satisfies equation (1.3) identically. Since Φ and ∇ are as yet
arbitrary, we can choose a relationship between them with a view to simplifying the
equations satisfied by A. One condition is:
A⋅
0=⋅∇ A , (1.34)
and this is known as the Coulomb gauge condition. The second condition is:
Φ−=⋅∇ ωµεjA , (1.35)
and this relationship is known as the Lorentz gauge condition. It should be noted
however that equation (1.35) is by no means the only unique solution for the Lorentz
gauge condition. Any function which satisfies the Lorentz gauge condition then
satisfies equations (1.2) and (1.4) identically as they lead to the continuity equation, and
hence conservative properties of the Maxwell equations [1.6].
1.1.3. Boundary conditions
As written in equations (1.1) to (1.8), the Maxwell equations are only valid in a
homogeneous medium where the properties µ, ε and σ vary slowly in space such that
E, D, B and H can be differentiated with respect to spatial and time co-ordinates only.
If the properties do vary discontinuously, the Maxwell equations must be augmented
with appropriate boundary conditions. At a boundary between different regions in the
earth, µ, ε and σ can change abruptly. Referring to equations (1.10) to (1.12), this
results in a discontinuous change in several of the fields. This is considered as a
physical discontinuity in material properties between two media, referred to as either
medium 1 or medium 2. It is therefore essential to have boundary conditions to define
the fields at points immediately adjacent to each other on either side of a boundary
across which the properties of the medium changes. E is governed by equation (1.4)
and (1.11). By re-writing this in terms of the electric field intensity alone, the interface
16
conditions for the tangential t and normal n components of the field can be written
respectively as:
tt EE 21 = , (1.36)
snn DD ρ=− 21 , (1.37)
where sρ is the surface charge density at the surface. Thus the tangential component of
E is continuous regardless of the material properties. The normal component of D is
discontinuous and the discontinuity is equal to the surface charge density at the
boundary. In the particular case when 0=sρ , the normal component of E is
continuous across the boundary. The boundary conditions for B are similarly obtained.
However, in this case, the surfaces between the two media may contain a surface
current density instead of a surface charge density such that:
tt BB 21 = , (1.38)
snn JBB =− 21 , (1.39)
where is the surface current density at the surface. Thus the tangential component of
B is continuous regardless of the material properties. The normal component of B is
discontinuous and the discontinuity is equal to the surface current density at the
boundary. In the particular case when , the normal component of B is continuous
across the boundary. Equations (1.36) to (1.39) are a complete representation of the
boundary conditions for the tangential and normal components of E, D, B and H.
sJ
0=sJ
1.1.4. Frames of reference
In this thesis, all problems are considered within the Cartesian co-ordinate
system. The fundamental co-ordinate system is defined as the right-hand positive {x, y,
z} co-ordinate system, as shown in Figure 1.1. The xy-plane is parallel to the surface of
the earth and the z-axis is perpendicular to the surface of, and is directed into the earth.
Another right-handed positive Cartesian co-ordinate system exists and can be
17
represented by {x�, y�, z�} where the y�z�-plane is inclined at an angle α about the x-
axis, the x�z�-plane is inclined at an angle β about the y-axis and the x�y�-plane is
inclined at an angle θ about the z-axis. This is defined as the inclined co-ordinate
system (Figures 1.2, 1.3 and 1.4). α, β and θ represent the three elementary Euler angles
about the x, y and z axes respectively, which can be used to rotate one frame of
reference to another (see Appendix 1).
x y z
Figure 1.1. Fundamental co-ordinate system for the solution to problems.
18
x= x� y α α z� y� z
Figure 1.2. Inclined co-ordinate system relative to the fundamental co-ordinate
system with the Euler angles about the x-axis indicated.
x x� β y = y� β z� z
Figure 1.3. Inclined co-ordinate system relative to the fundamental co-ordinate
system with the Euler angles about the y-axis indicated.
19
x x� θ y θ y� z = z�
Figure 1.4. Inclined co-ordinate system relative to the fundamental co-ordinate
system with the Euler angles about the z-axis indicated.
1.1.5. Further notation and definitions
All dimensional quantities are expressed explicitly in the SI units of
measurement. All definitions used in this thesis conform to the IEEE Standard
Definitions of Terms for Radio Wave Propagation, IEEE Std 211-1997TM. However,
in the interest of clarifying some of those definitions, we now define some of the
important concepts, and their respective mathematical formulations.
Homogeneous plane waves are those waves where the planes of constant
magnitude and constant phase are parallel. Time harmonic homogeneous plane waves
propagating in the z-direction are characterised by the notation:
)exp()exp( zjktj z−ω , (1.40)
where is the wave number in the z-direction. Inhomogeneous plane waves are those
waves where the planes of constant magnitude and constant phase are not parallel.
Time harmonic inhomogeneous plane waves propagating in the x- and z-directions are
characterised by the notation:
zk
20
)exp()exp()exp( zjkxktj zx −−ω , (1.41)
where is the wave number in the x-direction. In the scientific literature, there is
considerable ambiguity in the definitions of anisotropy and inhomogeneity as a
descriptor for media, e.g. [1.7], especially when discussing macro-anisotropy. Here, the
definitions for anisotropy and inhomogeneity are clarified as they relate to the problems
discussed in this thesis.
xk
The full space is considered as two horizontal half spaces. The upper half space
will be air and have properties such that 0=σ , 0εε = and 0µµ = . The lower half
space will be earth and have properties such that 0≠σ , 0εεε r= and 0µµµ r= .
The lower half space is defined as homogeneous when the half space consists
only of one medium. The lower half space is defined as inhomogeneous when the half
space can be considered a juxtaposition of two or more media in any spatial
dimension(s) and the boundaries between the different media may be unconformal.
A medium is defined as isotropic when all values of the conductivity,
permittivity and permeability are represented by the scalar constants σ, ε and µ
respectively, and are position independent everywhere inside that medium.
A medium is defined as coplanar anisotropic when any, or all of the values of
the conductivity, permittivity and permeability are represented by the second rank
diagonal tensors σ� , ε� and µ� respectively, where each element of the tensor is
coplanar with the fundamental co-ordinate system, and are position independent
everywhere inside that medium. When two elements of the diagonal tensors are equal,
and different to the third element, the medium is defined as uniaxial anisotropic. When
all three elements of the diagonal tensors are different, the medium is defined as biaxial
anisotropic, or bianisotropic.
A medium is defined as inclined anisotropic when any, or all of the values of
the conductivity, permittivity and permeability are represented by the second rank
diagonal tensors σ� , ε� and µ� respectively, inclined at an arbitrary angle to the
fundamental co-ordinate system, and are position independent everywhere inside that
medium. In fundamental co-ordinates, the tensor will contain non-diagonal elements.
Most earth media are non-magnetic so unless explicitly stated otherwise for the
particular problems considered in this thesis, it is assumed that I0� µµ = in the full
space, where I is the identity matrix. In addition to the definitions for inhomogeneous
21
that have been presented above, the term also extends to the description for the right-
hand side of a differential equation not equalling zero.
1.2. Surface impedance theory
By examining the relationship between two orthogonal field components of an
electromagnetic wave incident upon the surface of the earth, it is possible to gain
information about the subsurface conductivity profile beneath the point of measurement
[1.4, 1.23]. This method of mapping the subsurface was initially developed from the
telluric method of geophysical mapping where voltage probes were used to measure the
potential difference at points across the earth�s surface [1.22]. It was realised that the
source fields for these earth currents were geomagnetic pulsations and associated
ionospheric currents, where the fields can be considered to propagate as homogeneous
plane waves [see 1.1]. It should be noted that there has been considerable debate on the
appropriateness of the homogeneous plane wave model [1.9, 1.25, 1.27, 1.28]. As a
result, a method of normalising the measured electric field data was achieved by also
measuring the associated magnetic field components. The surface impedance Zmn is
defined as the ratio between the horizontal electric Em and magnetic Hn field
components:
n
mmn H
EZ = , (1.42)
where yxnm , , = . Originally, all interpretation was conducted for one-dimensional
models of horizontally stratified, isotropic media [1.30]. Such models are referred to as
Tikhonov-Cagniard models. Since the inception of this method, it has been applied
across the frequency range from 10-4 Hz to 100 kHz using both natural radiation
sources such as geomagnetic pulsations, which includes the magnetotelluric (MT)
[1.24, 1.26] and audiomagnetotelluric (AMT) methods, and atmospheric based
techniques [1.11]. Artificial sources have also been introduced, and include the
controlled-source audiomagnetotelluric (CSAMT) [1.13], very low frequency (VLF)
[1.15], and radio magnetotelluric (RMT) [1.10] methods. In geophysics, it is common
to express the surface impedance as an apparent resistivity ρa defined by:
22
0
2
ωµρ mn
a
Z= , (1.43)
It should be noted that this definition of apparent resistivity has been questioned
e.g., [1.21] but in this thesis, where used, the definition of equation (1.43) is implied.
The concept of the apparent resistivity is that it is equal to the resistivity of the medium
as if it is a homogeneous medium.
If the radiation source is not linearly polarised or the earth�s subsurface is
laterally anisotropic, hence not corresponding to the Tikhonov-Cagniard model, the
perpendicularity of E and H does not hold [1.5, 1.19]. In such a case, each electric field
component is the linear sum of both components of the horizontal magnetic field, and
can be expressed as the linear relationship:
yxyxxxx HZHZE += , (1.44)
yyyxyxy HZHZE += , (1.45)
where the second rank impedance tensor Z� has been introduced:
=
yyyx
xyxx
ZZZZ
Z� . (1.46)
The tensor elements are defined by equation (1.43). It is specifically noted that
equation (1.46) is defined in the xy-plane. The components and are called the
principal impedances, and the components Z and are called the additional
impedances [1.20]. However, the tensor components change as the co-ordinate system
is rotated. If a right-handed rotation about the positive z-axis is applied to equation
(1.46) (see Appendix 1), then:
xyZ yxZ
xx yyZ
θθθθ 2sin)(21sincos)( 22
'' yxxyyyxxxx ZZZZZ +−+= , (1.47)
θθθθ 2sin)(21sincos)( 22
'' yyxxyxxyyx ZZZZZ −+−= , (1.48)
23
θθθθ 2sin)(21sincos)( 22
'' yyxxxyyxxy ZZZZZ −+−= , (1.49)
θθθθ 2sin)(21sincos)( 22
'' yxxyxxyyyy ZZZZZ +−+= . (1.50)
There are three expressions obtained from the impedance tensor that are
independent of the rotation of the measurement axes. These are called invariants [1.20]
and are denoted by I:
2
1 effyxxyyyxx ZZZZZI =−= , (1.51)
yyxx ZZI +=2 , (1.52)
ZZZI xyxy 23 =−= , (1.53)
where and 2effZ Z are called the effective and mean impedance values, respectively. If
the radiation source is linearly polarised and/or the earth�s subsurface is laterally
isotropic, then and . 0== yyZxxZ yxxy ZZ −=
1.3. Electrical anisotropy
The concepts of homogeneity and isotropy play an important role in current
electromagnetic plane wave modelling and inversion. Usually, multi-dimensional
models are considered to be a juxtaposition of elements that are homogeneous and
isotropic, e.g., see [1.34]. It has been well documented that in layered or laminated
sedimentary rocks, electrical current circulates with less resistance parallel to the
bedding planes than perpendicularly to them [e.g., 1.14, 1.18]. Several mechanisms
account for this conductivity difference including sorting (i.e., graded deposition), rock
matrix grain geometry, pore geometry and the preferential orientation and grain
geometry of conductive clay minerals. For convenience, early studies concerning
electrical anisotropy assumed the anisotropic model based upon the uniaxial crystal
structure in optics, e.g. [1.3, 1.16], where, in inclined co-ordinates, the uniaxial
conductivity tensor can be represented by:
24
=
n
t
t
σσ
σσ
000000
� , (1.54)
where tσ is the conductivity tangential to the bedding plane, and nσ is the
conductivity normal to the bedding plane. The coefficient of anisotropy [1.18] is
defined as:
n
t
σσλ = . (1.55)
This assumption is based upon the view that depositional conditions in different
lateral directions, representing different points in space at the same instant in time, have
less variation than depositional conditions in the vertical direction, representing the
same points in space at different instants in time. Whilst a uniaxial anisotropic
conductivity is clearly a reasonable approximation based on a simple stratigraphic
model, the assumption is also mathematically convenient. The Maxwell equations can
be decoupled in certain frames of reference, leading to relatively simple exact solutions.
If a more general model were to be considered, it would be easy to assume depositional
environments where the conditions vary laterally [1.12] and it would be necessary to
consider the biaxial conductivity tensor:
=
n
yt
xt
σσ
σσ
000000
� ,
,
. (1.56)
In this thesis, only uniaxial anisotropic media is considered. However, the
following discussion about the properties of the conductivity tensor is equally
applicable to both uniaxial and biaxial anisotropic conductivity tensors. Consider a
conductivity tensor '�σ analogous to either equations (1.54) or (1.56) in an inclined co-
ordinate system {x�, y� z�}. The conductivity tensor can be rotated into the fundamental
co-ordinate system {x, y, z} using the three Euler rotations [1.2]:
),,( '� ),,(� T αβθσαβθσ RR= , (1.57)
25
where ),,( αβθR is the Euler co-ordinate rotation matrix (see Appendix 1):
)()()(),,( αβθαβθ RRRR = . (1.58)
The resultant conductivity tensor σ� takes the most general form of:
=
zzzyzx
yzyyyx
xzxyxx
σσσσσσσσσ
σ� , (1.59)
where ) , , , ,( θβασσσ ntijij f= . The corresponding resistivity tensor ρ� is defined as
the corresponding inverse of the conductivity tensor:
I='� '� ρσ , (1.60)
and it follows that:
IRIRRR == ),,( ),,(),,('� '� ),,( TT αβθαβθαβθρσαβθ , (1.61)
IRRRR =),,('� ),,( ),,( '� ),,( TT αβθραβθαβθσαβθ , (1.62)
I=∴ ρσ � � . (1.63)
Equation (1.63) holds for any co-ordinate rotation. Furthermore, to define the
resistivity tensor, it is demonstrated that:
11 � � � � −− = ρρρσ I , (1.64)
1� � −= ρσ II , (1.65)
1�� −=∴ ρσ . (1.66)
26
However, it should be noted that and except when 1,,
−≠ jiji ρσ jiji ,1
, ρσ ≠−
IR = ),,( αβθ . σ� has the specific properties of non-negativity and symmetry [1.31].
σ� is symmetric whenever the magnetic field does not influence the conduction process
[1.17]. Symmetry is thus granted for purely ohmic conduction. However, in the
presence of Hall currents, such as in plasma, σ� will be asymmetric [1.29]. To
demonstrate this symmetry, an example from Dekker & Hastie [1.8] is used. Let us
consider a laterally anisotropic surface where the ohmic functions:
θθθ sincos)( yx EEE += , (1.67)
θθσθθσθ sin)(cos)()( yx EEJ += , (1.68)
exist, where )(θσ is the effective conductivity in the direction of θ , and is defined
simply as:
)()()(
θθθσ
EJ= . (1.69)
Now, if , then 0=xE θθσ sin)()0( yEJ = and:
∫∫−−
===2
2
2
2
)( cossin)0( cos1)0(π
π
π
π
θθσθθθθσ ddJEE
J
yyxy . (1.70)
If , then 0=yE θθσπ cos)()2
( xEJ = and:
∫∫ ===ππ
θθσθθθπθ
π
σ00
)( cossin)2
( sin1)2
(ddJ
EE
J
xxyx . (1.71)
We can then observe that:
27
∫∫ −=−−
ππ
π
θθσθθθθσθθσσ0
2
2
)( cossin)( cossin ddyxxy . (1.72)
If )()( θσθσ −= , then 0=− yxxy σσ . If not, then there will be rectification of
the electrical current in the direction of θ . Now, consideration is given to the second
condition of non-negativity in σ� . Assuming no magnetic conductivity exists, any
losses in fields are exclusively ohmic (i.e., from electrical conductivity). From
Poynting�s theorem, the expression for the specific energy dissipation u due to ohmic
losses, known as Joule�s Law, can be written as [1.16]:
2 � EEJ σ=⋅=u , (1.73)
Since from thermodynamic considerations, we observe that the time averaged
specific energy dissipation
0>u
u can be written as:
0�21
21 >⋅⋅=⋅= E*EJ*E σu , (1.74)
which implies that σ� must be positive semi-definite. This means that only positive
conductivity values can exist in the diagonal elements of σ� .
1.4. References
[1.1] M. N. Berdichevskiy & M. S. Zhdanov, Advanced Theory of Deep
Geomagnetic Sounding (Elsevier, Amsterdam, 1984).
[1.2] F. W. Byron & R. W. Fuller, Mathematics of Classical and Quantum Physics
(Dover Publications, New York, 1992).
[1.3] M. Born & E. Wolf, Principles of Optics (Permagon Press, Oxford, 1965).
[1.4] L. Cagniard, Geophysics 18, 605 (1953).
[1.5] T. Cantwell, PhD thesis, Massachusetts Institute of Technology, 1960.
[1.6] D. N. Chetaev, Phys. Solid Earth 2, 555 (1966).
[1.7] D. N. Chetaev & B. N. Belen�kaya, Phys. Solid Earth 7, 212 (1971); 8, 535
(1972).
28
[1.8] D. L. Dekker & L. M. Hastie, Geophys. J. R. astron. Soc. 61, 11 (1980).
[1.9] V. I. Dmitriev & M. N. Berdichevsky, Proc. IEEE 67, 1034-1044 (1979).
[1.10] A. Dupis, A. Choquier & G. Borruet, Bull. Soc. Geol. France 166, 231 (1995).
[1.11] S. J. Garner & D. V. Thiel, Explor. Geophys. 31, 173 (2000).
[1.12] S. Gianzero, D. Kennedy, L. Gao & L. SanMartin, Petrophys. 43, 172 (2002).
[1.13] M. A. Goldstein & D. W. Strangway, Geophysics 40, 669 (1975).
[1.14] G. V. Keller & F. C. Frischknecht, Electrical Methods in Geophysical
Prospecting (Permagon Press, Oxford, 1966).
[1.15] R. J. King, IEEE Trans. Antennas Propagat. 24, 115 (1976).
[1.16] L. D. Landau, E. M. Lifshitz & L. P. Pitaevskii, Electrodynamics of Continuous
Media, 2nd ed. (Butterworth-Heinemann, Oxford, 1982).
[1.17] L. Onsager, Phys. Rev. 37, 405 (1931).
[1.18] E. I. Parkhomenko, Electrical Properties of Rocks (Plenum Press, New York,
1967).
[1.19] I. I. Rokityanski, Bull. Acad. Sci. USSR Geophys. Ser. 10, 1050 (1961).
[1.20] I. I. Rokityanski, Geoelectromagnetic Investigation of Earth's Crust and Mantle
(Springer-Verlag, Berlin, 1982).
[1.21] B. R. Spies & D. E. Eggers, Geophysics 51, 1462 (1986).
[1.22] D. V. Thiel, IEEE Trans. Antennas Propagat. 48, 1517 (2000).
[1.23] A. N. Tikhonov, Dokl. Akad. Nauk SSSR 73, 295 (1950).
[1.24] A. N. Tikhonov & M. N. Berdichevsky, Phys. Solid Earth 2, 93 (1966).
[1.25] A. N. Tikhonov, D. N. Chetaev, V. A. Morgunov, I. K. Chantladze, S. V.
Shamanin & Ye. A. Gerasimovich, Dokl. Akad. Nauk SSSR Zemli Nauk 217,
28 (1974).
[1.26] K. Vozoff, Geophysics 37, 98 (1972).
[1.27] J. R. Wait, Geophysics 19, 281 (1954).
[1.28] J. R. Wait, J. Res. Nat. Bur. Stand. D. Radio Propagat. 66D, 509 (1962).
[1.29] J. R. Wait, Electromagnetics and Plasmas (Holt, Rinehart and Winston, New
York, 1968).
[1.30] J. R. Wait, Electromagnetic Waves in Stratified Media, 2nd ed. (Permagon Press,
Oxford, 1970).
[1.31] P. Weidelt, in Three-Dimensional Electromagnetics, edited by M. Oristaglio &
B. Spies (Soc. Explor. Geophys., Tulsa, 1999), pp. 119-137.
[1.32] M. S. Zhdanov, Integral Transforms in Geophysics (Springer-Verlag, Berlin,
1988).
29
[1.33] M. S. Zhdanov and G. V. Keller, The Geoelectrical Methods in Geophysical
Exploration (Elsevier, Amsterdam, 1994).
[1.34] M. S. Zhdanov, I. M. Varenstov, J. T. Weaver, N. G. Golubev & V. A. Krylov,
J. App. Geophys. 37, 133 (1997).
30
Chapter 2
Exact solutions for one-dimensional anisotropy problems
2.1. Introduction
The effect of fundamental uniaxial anisotropic conductivity in a horizontally
layered half space excited by an inhomogeneous plane wave has been previously
studied [2.4, 2.5, 2.14-2.16]. Chetaev [2.2, 2.3] was the first to consider the problem of
a homogeneous half space with inclined uniaxial conductivity anisotropy excited by a
homogeneous plane wave. Chetaev�s method of solution (referred to as the Chetaev
method of functional rotation) was later extended to a two-layer model by Sinha [2.20],
where only the basement was assumed to have inclined anisotropic conductivity, and
later by Negi & Saraf [2.9] where the upper layer was assumed to have inclined
anisotropic conductivity. Reddy & Rankin [2.17] extended the formulation to a multiple
layered half space where each layer was assumed to have inclined anisotropic
conductivity. Further discussion on horizontally layered problems with inclined
anisotropy subsequently followed [2.1, 2.6-2.8, 2.10-2.13, 2.18, 2.23-2.24]. Singh
[2.19] presented an album of sounding curves investigating the effect of conductive and
resistive basements. Thiel [2.21] observed that from Singh�s [2.19] results using the
formulation derived by Sinha [2.20], the effective horizontal conductivity of an inclined
anisotropic layer m was given by the equation:
1
,
2
,
2 sincos−
+=
mn
m
mt
mm σ
ασ
ασ , (2.1)
where mα is the angle of inclination of the anisotropic conductivity, measured with
respect to the horizontal plane. mt ,σ and mn,σ are the tangential and normal
conductivities of the medium and were discussed in Chapter 1. Thiel [2.21] observed
that equation (2.1) is the yyρ element of a uniaxial resistivity tensor rotated into the
horizontal plane. Thiel [2.21] suggested that perhaps the effective horizontal
conductivity should be given by the yyσ element of the uniaxial conductivity tensor
rotated into the horizontal plane, i.e.:
31
mmnmmtm ασασσ 2,
2, sincos += , (2.2)
which would produce a different set of results for the surface impedance than those
presented by Singh [2.19] when 0 . Figure 2.1 presents the effective
horizontal conductivity for calculated using equations (2.1) and (2.2) for
a homogeneous half space where
o90<< mαo90
001.0=t
0 ≤≤ mα
σ S/m and 4=λ . Thiel [2.21] based his
argument on the belief that the conductivity tensor is a more fundamental quantity than
the resistivity tensor, since the conductivity describes the relationship between H and E
in the Maxwell equation (1.1). The discussion was dismissed (without scientific
argument) by Singh [2.19], and no further discussion on this problem has arisen in the
literature.
Figure 2.1. Effective conductivity for 0 calculated using equations
(2.1) and (2.2) for a homogeneous half space where
o90≤≤ mα
001.0=tσ S/m
and 4=λ .
32
The difference in the surface impedance magnitude of a homogeneous TM-type
wave across the magnetotelluric frequency band (10-4 Hz to 10 Hz) for a half space
with inclined conductivity anisotropy is shown in Figure 2.2 with effective
conductivities given by both equations (2.1) and (2.2). The homogeneous half space has
the anisotropic parameters 001.0=tσ S/m, 4=λ and . For a homogeneous
half-space, the phase is equal to and is constant for all frequencies. Figure 3
presents the response for the same frequency band for a two-layered half-space, where
the upper layer has isotropic conductivity of 0.1 S/m and is 10 km thick, overlaying the
homogeneous half-space with inclined anisotropic conductivity described above.
o30=αo45
Figure 2.2. Surface impedance and magnitude of a homogeneous TM-type wave
incident upon a homogeneous half space with the anisotropic
parameters 001.0=tσ S/m, 4=λ and . o30=α
Now, given that some formulations rely on the rotation of the resistivity tensor,
based on Chetaev [2.2, 2.3], and other formulations rely on the rotation of the
conductivity tensor, e.g., [2.25], it is important to clarify under what conditions the two
methods are equal. The observation that the two rotations yield different results for the
33
same conductivity and angle of inclination is one that appears to have been overlooked.
In this chapter, the Chetaev method of function rotation [2.2, 2.3] and the method of
tensor rotation [2.13] will be reviewed to derive expressions for the surface impedance
of homogeneous TE- and TM-type waves incident upon horizontally homogeneous
inclined anisotropic media.
Figure 2.3. Surface impedance and magnitude of a homogeneous TM-type wave
incident upon a two-layered half space where the upper layer is
isotropic with conductivity of 0.1 S/m and 10 km thick. The bottom
layer, of infinite depth extent, has the anisotropic parameters
001.0=tσ S/m, 4=λ and . o30=α
2.2. The problem of a homogeneous half space
We will consider the surface impedance of a homogeneous, monochromatic
TM-type wave with time variance of )exp( tjω incident upon a homogeneous half
space with inclined uniaxial anisotropic conductivity. It is assumed that the
component of the TM-type wave is co-incident with the
xH
xxσ element of the inclined
34
conductivity tensor, and so propagates independently of any TE-type waves (see Figure
2.4). It is assumed that the component of the TE-type wave is co-incident with the xE
xxσ element of the inclined conductivity tensor, and so propagates independently of
any TM-type waves.
'E y
'Ez
σt x
y
α
σt y�
z� σn
z
Figure 2.4. Geometry for the inclined anisotropic half space. The source fields
are incident parallel to the x-axis. The z-axis is normal to the surface
of the earth.
2.2.1. The Chetaev method of functional rotation
In this section, the formulation of Chetaev [2.2] is followed. First, from
equations (1.18) and (1.19), we write the forms of Maxwell�s equations for a
homogeneous TM-type wave in the inclined co-ordinate system:
'1 '
zH x
t ∂∂
=σ
, (2.3)
'1 '
yH x
n ∂∂−=
σ, (2.4)
35
0'' ''' =+
∂∂
−∂
∂x
yz Hjz
Ey
E ωµ . (2.5)
Differentiating equations (2.3) and (2.4) with respect to z� and y� respectively,
and substituting the results into equation (2.5) obtains:
0'
1'
1'2
'2
2'
2
=−∂
∂+
∂∂
xx
t
x
n
HjzH
yH ωµ
σσ, (2.6)
which is the homogeneous Helmholtz equation for anisotropic media. Now, since the
half space is assumed to be infinitely deep, a solution to equation (2.6) in the form of a
down-going homogeneous plane wave is considered:
)cos'sin'(exp)exp( 1111' αα zykCzkCHH xx +−=−== , (2.7)
which satisfies equation (2.6) provided the wave number is given by: 1k
122
1sincos
−
+=
nt
jkσ
ασ
αωµ , (2.8)
and provided that Re to prevent an exponentially divergent solution in ; and
where is an arbitrary constant, which can be complex. Following equations (2.3)
and (2.4), the electric fields in inclined co-ordinates can be expressed as:
01 >k xH
1C
'1
' cos xt
y HkE ασ−
= , (2.9)
'1
' sin xn
z HkE ασ
= . (2.10)
Using a co-ordinate rotation (see Appendix 1), the horizontal electric field in
fundamental co-ordinates can be written as:
36
αα sincos '' zyy EEE −= . (2.11)
After substituting equations (2.9) and (2.10) into equation (2.11), then:
xnt
y HkE sincos 22
1
+−=
σα
σα , (2.12)
is obtained, and it follows that the surface impedance of the TM-type wave can be
expressed as:
+−=
ntyx jZ
σα
σαωµ
22 sincos . (2.13)
We present here another method of solving this problem using rotation of the
Maxwell equations. Consider the x-axis co-ordinate rotation (see Appendix 1):
−=
−=
zyx
zyx
zyx
mm
mm )(cossin0sincos0
001
'''
ααααα R , (2.14)
such that all the partial derivatives of any function f can be re-written as:
mm zf
yf
yf αα sincos' ∂
∂+∂∂=
∂∂ , (2.15)
mmmm zf
zyf
yf
yf αααα 2
2
222
2
2
2
2
sincossin2cos' ∂
∂+∂∂
∂+∂∂=
∂∂ , (2.16)
mm zf
yf
zf αα cossin' ∂
∂+∂∂−=
∂∂ , (2.17)
mmmm zf
zyf
yf
zf αααα 2
2
222
2
2
2
2
coscossin2sin' ∂
∂+∂∂
∂−∂∂=
∂∂ . (2.18)
37
Now, equation (2.6) can be re-written in y, z co-ordinates as:
.0cossin 11 2
cossinsincos
2
2
222
2
222
=−∂∂
∂
−+
∂∂
++
∂∂
+
xx
tn
x
tn
x
tn
Hjzy
Hσ
zH
yH
ωµαασ
σα
σα
σα
σα
(2.19)
From equation (2.19), the term containing the mixed partial derivative:
zyH
σx
tn ∂∂∂
−
2
cossin 11 2 αασ
,
is commonly called the shearing term, from the mechanics analogy of forces acting on
a beam carrying a load, where the diagonal terms of the force tensor are pressures, or
tensions, and the non-diagonal terms are shearing forces.
A general solution for a homogeneous plane wave, like equation (2.7), is
satisfied provided that the wave number is given by equation (2.8) and for
the same reasons as stated before. For any medium, m, the intrinsic impedance is
defined as:
0Re 1 >k
mm k
jZ ωµ−= . (2.20)
By substituting equation (2.8) into equation (2.20), the surface impedance is
identically obtained to equation (2.13). Now, the same formulation as shown above will
be considered but with considerations to a down-going homogeneous plane wave of the
of TE-type. From equations (1.18) and (1.19):
zE
jH x
y ∂∂−=
ωµ1 , (2.21)
yE
jH x
z ∂∂
=ωµ1 , (2.22)
and the Helmholtz equation for Ex is simply:
38
xxx Ek
yE
zE 2
22
2
2
2
=∂
∂+
∂∂
, (2.23)
which has a solution in the form of a down-going homogeneous plane wave:
)exp( 22 zkCEx −= , (2.24)
since the half space is assumed to be infinitely deep. Following equation (1.18), the
wave number k is given by: 2
tjk ωµσ=2 , (2.25)
provided to prevent an exponentially divergent solution in ; and where
is an arbitrary constant which can be complex. Since E
0Re 2 >k xE
2C x is laterally homogeneous,
Hz vanishes and the solution for Hx is simply:
xy Ejk
Hωµ
2= , (2.26)
and the surface impedance can be written as:
txy
jZσωµ= , (2.27)
which is independent of the anisotropic parameters α and σn.
2.2.2 Method of tensor rotation
Let us now consider the quasi-static Maxwell curl equations (1.18) to (1.19) and
expand them respectively to obtain:
xyz J
zH
yH
=∂
∂−
∂∂ , (2.28)
39
yzx J
xH
zH
=∂
∂−
∂∂
, (2.29)
zxy J
yH
xH
=∂
∂−
∂∂
, (2.30)
xyz Hj
zE
yE ωµ−=
∂∂
−∂
∂ , (2.31)
yzx Hj
xE
zE ωµ−=
∂∂
−∂
∂, (2.32)
zxy Hj
yE
xE
ωµ−=∂
∂−
∂∂
. (2.33)
From Ohm�s Law, equation (1.12), and for a full conductivity tensor, equation
(1.59), one can write the current density components as:
zxzyxyxxxx EEEJ σσσ ++= , (2.34)
zyzyyyxyxy EEEJ σσσ ++= , (2.35)
zzzyzyxzxz EEEJ σσσ ++= . (2.36)
Since we are considering a homogeneous half space excited by a homogeneous
plane wave, all partial derivatives with respect to x and y are equal to zero. From
equations (2.28) to (2.33), then it follows that:
zxzyxyxxxy EEE
zH
σσσ ++=∂
∂− , (2.37)
zyzyyyxyxx EEE
zH σσσ ++=∂
∂, (2.38)
40
xy Hj
zE
ωµ=∂
∂, (2.39)
yx Hj
zE ωµ−=∂
∂, (2.40)
and by implication it is observed that:
0=zH , (2.41)
0=++ zzzyzyxzx EEE σσσ . (2.42)
Equations (2.41) and (2.42) simply state that the vertical magnetic field and
vertical current density in the half space are equal to zero, when the source field is a
homogeneous plane wave. By differentiating equations (2.39) and (2.40) with respect to
z and substituting the results into equations (2.37) and (2.38), one obtains:
( zxzyxyxxxx EEEj
zE σσσωµ ++−
∂∂
= 2
2
0 ), (2.43)
( zyzyyyxyxy EEEj
zE
σσσωµ ++−∂
∂= 2
2
0 ). (2.44)
From equation (2.36), it is possible to write:
yzz
zyx
zz
zxz EEE
σσ
σσ
−−= . (2.45)
After substituting equation (2.45) into equations (2.43) and (2.44), two coupled
second order differential equations for the horizontal electric fields are obtained [2.13]:
yzz
zyxzxyx
zz
zxxzxx
x EjEjzE
−−
−−
∂∂
=σ
σσσωµ
σσσσωµ2
2
0 , (2.46)
41
yzz
zyyzyyx
zz
zxyzyx
y EjEjzE
−−
−−
∂∂
=σ
σσσωµ
σσσ
σωµ2
2
0 . (2.47)
Let us now assume that problem is posed with the inclined conductivity tensor
only being rotated about the x-axis by α such that the conductivity tensor can be written
as:
=
zzyz
zyyy
xx
σσσσ
σσ
00
00� , (2.48)
where the elements of the conductivity tensor then have the form:
txx σσ = , (2.49)
ασασσ 22 sincos ntyy += , (2.50)
ασασσ 22 cossin ntzz += , (2.51)
αασσσσ cossin)( ntzyyz −== , (2.52)
where 0==== zxxzyxxy σσσσ . With such conditions on the conductivity tensor,
equations (2.46) and (2.47) take the forms of the uncoupled second order ordinary
differential equations:
xxxx Ej
zE ωµσ−
∂∂
= 2
2
0 , (2.53)
yzz
zyyzyy
y EjzE
−−
∂∂
=σ
σσσωµ2
2
0 . (2.54)
Now, if one considers a solution for equation (2.53) of the form of a down-
going homogeneous plane wave of type:
42
)exp( 33 zkCEx −= , (2.55)
where is an arbitrary constant which may be complex, then it is easily observed
that:
3C
tjk ωµσ=3 , (2.56)
provided that to prevent an exponentially divergent solution in . The
surface impedance can then be written as:
0Re 3 >k xE
txy
jZσωµ= , (2.57)
which corresponds identically to the surface impedance of the TE-type wave, derived
earlier as equation (2.27). If one now considers a solution for equation (2.54) of the
form of the down-going homogeneous plane wave:
)exp( 44 zkCEy −= , (2.58)
where C is an arbitrary constant which may be complex, then from equations (2.49) to
(2.52), it can be determined that:
4
122 sincos
−
+=−
ntzz
zyyzyy σ
ασ
ασ
σσσ , (2.59)
implying that the yyρ term of the rotated resistivity tensor can be written in terms of
yyσ , yzσ , zyσ and zzσ , and vice versa. It then follows simply that:
122
4sincos
−
+=
nt
jkσ
ασ
αωµ , (2.60)
43
provided to prevent an exponentially divergent solution in . The surface
impedance can then be written as:
0Re 4 >k yE
+−=
ntyx jZ
σα
σαωµ
22 sincos , (2.61)
which corresponds identically to equation (2.13).
2.2.3 Discussion
For TM-type incidence in sub-Sections 2.21 and 2.22, one can observe that the
inclined uniaxial anisotropic conductivity tensor can be written as a corresponding
fundamental bianisotropic conductivity tensor where:
zz
zyyzyyyy
nty σ
σσσρ
σα
σασ −==
+= −
−1
122 sincos ,
yy
yzzyzzzz
ntz σ
σσσρ
σα
σασ −==
+= −
−1
122 cossin .
This is a very important concept. We can therefore state that for a homogeneous
plane wave propagating in a homogeneous half space, the inclined conductivity tensor
is equivalent to a bianisotropic conductivity tensor in the fundamental co-ordinates.
2.3. The problem of a horizontally layered half space
Thiel [2.21] suggested that the Wait�s [2.22] surface impedance solution for a
horizontally stratified half space consisting of isotropic layers could be used to
calculate the surface impedance of a horizontally stratified half space consisting of
layers with inclined anisotropy, provided the conductivity used in Wait�s [2.22]
expressions were that of the effective horizontal conductivity. This argument is a valid
one, given that for a homogeneous TM-type plane wave, it has been shown in Section
2.2 that the surface impedance is a function of the yyρ term only, automatically
implying that non-diagonal conductivity tensor terms are considered. One can then
44
write the surface impedance Z1,s expression for a horizontally stratified half space (e.g.,
see Figure 2.5) consisting of M layers with inclined anisotropy using Wait�s [2.22]
transmission line analogy:
)tanh()tanh(
11,21
111,21,1 hkZZ
hkZZZZ
s
ss +
+= , (2.63)
)tanh()tanh(
22,32
222,32,2 hkZZ
hkZZZZ
s
ss +
+= , (2.64)
������������
)tanh()tanh(
,1
,1
mmsmm
mmmsmmms hkZZ
hkZZZZ
+
+
++
= , (2.65)
������������
)tanh()tanh(
111
11111
−−−
−−−−− +
+=
MMMM
MMMMMM hkZZ
hkZZZZ , (2.66)
where m = 1, 2, �.., M, indicating the layer number from the surface and σn,m, σt,m and
αm are the conductivity and anisotropy inclination values for the mth layer.
45
x y z σt,1 Zs,1 z = 0 α1 σt,1 Z1 σn,1 α1 σt,2 Zs,2 z = h1 α2 Z2 σn,2 α2 σt,2
Figure 2.5. Geometry for a two-layer horizontally stratified half space where
each layer exhibits inclined uniaxial anisotropic conductivity.
Equations (2.63) to (2.66) can be used to exactly reproduce the results of Singh
[2.19], Negi & Saraf [2.9] and Sinha [2.20]. Equations (2.63) to (2.66) are derived on
the basis of the continuity of the tangential field components at the boundaries, with the
conditions that:
mxmx HH ,1, =− , (2.67)
zH
zH mx
mn
m
mt
mmx
mn
m
mt
m
∂∂
+=
∂∂
+ −
−
−
−
− ,
,
2
,
21,
1,
12
1,
12 sincossincos
σα
σα
σα
σα
. (2.68)
Boundary conditions (2.67) and (2.68) then lead to 2(M-1) equations which are
linear in their plane wave amplitude coefficients, to solve for 2(M-1) unknowns in
terms of the known incident plane wave amplitude. For waves of the TE-type, the wave
number will simply contain the σt,m terms, identically satisfying all other conditions
required for equations (2.63) to (2.66).
46
2.4. References
[2.1] N. M. Al�tgauzen (Althausen), Phys. Solid Earth 5, 510 (1969).
[2.2] D. N. Chetaev, Bull. Acad. Sci. USSR Geophys. Ser. 4, 107 (1960).
[2.3] D. N. Chetaev, Phys. Solid Earth 2, 233 (1966).
[2.4] D. L. Dekker & L. M. Hastie, Geophys. J. R. astron. Soc. 61, 11 (1980).
[2.5] R. J. King, App. Phys. 5, 187 (1974).
[2.6] R. J. King & J. R. Wait, Symp. Math. 17, 107 (1976).
[2.7] V. T. Levadnyy & I. V. Pavlova, Geomag. Aeron. 21, 657 (1981); 22, 689
(1982).
[2.8] D. Loewenthal & M. Landisman, Geophys. J. R. astron. Soc. 35, 195 (1973).
[2.9] J. G. Negi & P. D. Saraf, Geophys. Prosp. 20, 785 (1972).
[2.10] J. G. Negi & P. D. Saraf, Radio Sci. 11, 787 (1976).
[2.11] J. G. Negi & P. D. Saraf, Phys. Earth Planet. Int. 44, 324 (1986).
[2.12] D. P. O�Brien & H. F. Morrison, Geophysics 32, 668 (1967).
[2.13] J. Pek & F. A. M. Santos, presented at 19th DGG Coll. Electromagn. Depth Inv.,
Burg Ludwigstein, Germany, Oct. 2001.
[2.14] V. Petr, Stud. Geophys. Geod. 11, 291 (1967).
[2.15] O. Praus & V. Petr, Can. J. Earth Sci. 6, 759 (1969).
[2.16] G. G. Pukhov, Geol. Geofiz. 7, 68 (1965).
[2.17] I. K. Reddy & D. Rankin, Geophys. Prosp. 19, 84 (1971).
[2.18] P. D. Saraf & J. G. Negi, Geophys. J. R. astron. Soc. 74, 809 (1983).
[2.19] R. P. Singh, Geophys. Prosp. 33, 369 (1985); 34, 925 (1986).
[2.20] A. K. Sinha, Geoexpl. 7, 9 (1969).
[2.21] D. V. Thiel, Geophys. Prosp. 34, 923 (1986).
[2.22] J. R. Wait, Electromagnetic Waves in Stratified Media, 2nd ed. (Permagon Press,
Oxford, 1970).
[2.23] J. R. Wait, IEEE Trans. Antennas Propagat. 39, 268 (1991).
[2.24] J. R. Wait, J. Electromagn. Waves Applic. 10, 871 (1996).
[2.25] P. Weidelt, in Three-Dimensional Electromagnetics, edited by M. Oristaglio &
B. Spies (Soc. Explor. Geophys., Tulsa, 1999), pp. 119-137.
[2.26] G. A. Wilson & D. V. Thiel, Radio Sci. 37, 1029, 2001RS002535 (2002).
47
Chapter 3
Exact solutions for two-dimensional inclined anisotropy problems
3.1. Introduction
As discussed in Chapter 2, the exact one-dimensional surface impedance
expression for homogeneous plane wave incidence above a horizontally stratified
(vertically inhomogeneous) earth has been well documented. Exact and approximate
solutions for the surface impedance anomaly above a horizontally inhomogeneous half
space with isotropic conductivity have been previously investigated for both TE- and
TM-type waves [3.2, 3.3, 3.7-3.9, 3.13-3.16, 3.18-3.23]. The issue of a sloping contact
was considered theoretically by Dmitrieyev & Zakharov [3.4] and Geyer [3.5], and
numerically by Reddy & Rankin [3.12]. However, with the exception of the
fundamental uniaxial anisotropy solution of Obukhov [3.10], and the review of
d�Erceville & Kunetz�s [3.3] solution for inclined uniaxial anisotropic conductivity by
Grubert [3.6], no significant attention has been given to the problem of exactly solving
for the surface impedance anomaly above a vertical contact between two conductive
media that have inclined anisotropic conductivity.
The exact inclined anisotropic problem has been investigated for vertically
inhomogeneous problems (see Chapter 2). It is the purpose of this chapter to present the
exact quasi-static solution for the surface impedance response of a conducting layer
with inclined anisotropic conductivity, with lateral inhomogeneities. In a similar way
that the solutions of Weaver et al. [3.22, 3.23] were developed as control models for the
COMMEMI project, it is suggested that the development of an exact solution for a two-
dimensional control model with inclined uniaxial anisotropy will serve as a benchmark
for other approximate methods of solution. We will do this using an extension of the
Fourier series method presented by previous authors for isotropic and coplanar
anisotropic media. The formulation by Rankin [3.11] is used as the basis for the
formulation presented here.
It is assumed that the xxσ element of the conductivity tensor is parallel to the
strike of the dyke and the linearly polarised field, as this then allows one to solve
for the linearly polarised TM-type homogeneous plane wave, as the TE-type
homogeneous plane wave is uncoupled and will be propagated independently. By
xH
48
assuming that the lateral inhomogeneities have an infinite strike length, this reduces the
problem to a two-dimensional one and makes it only necessary to solve the Maxwell
equations in the region . 0≥z
bσ
-
σt,2
σt,2
3.2. Exact formulation: preliminary considerations
Consider a homogeneous dyke with inclined conductivity anisotropy extending
infinitely into the x-direction, embedded in an otherwise homogeneous layer, which
also exhibits inclined anisotropic conductivity. The common depth of the dyke and the
layer is h and they are both underlain by a homogeneous half space, or basement, with
isotropic conductivity . Figure 3.1 shows the solution space and respective variables.
By considering a homogeneous plane wave as the source field, then all partial
derivatives with respect to x are equal to zero.
x l/2 +l/2 y z σt,1 σt,2 z = 0 α2 α1 α2 σt,1 σt,2
α2 α1 α2 σn,2 σn,1 σn,2 z = h
σb
Figure 3.1. Geometry and parameters of the dyke embedded in a homogeneous
layer.
In the local inclined co-ordinate system characterised by angle of inclination 1α
about the x-axis, the conductivity of the dyke is represented with the uniaxial
conductivity tensor:
49
=
1,
1,
1,
1
000000
�
n
t
t
σσ
σσ . (3.1)
Similarly, in the local inclined co-ordinate system characterised by angle of
inclination 2α about the x-axis, the conductivity of the host layer is represented with
the uniaxial conductivity tensor:
=
2,
2,
2,
2
000000
�
n
t
t
σσ
σσ . (3.2)
t subscripts denote the conductivity tangential to the inclined horizontal axis of
the medium, and n subscripts denote the conductivity normal to the inclined horizontal
axis of the medium. When general solutions to the Maxwell equations are considered in
this section, subscript m is introduced to designate the medium, where m = 1 for the
dyke and m = 2 for the host layer. The equations presented here can be considered to
satisfy an arbitrary inclined co-ordinate system {x�, y�, z�} rotated through an angle mα
about the x-axis. Firstly, the forms of the Maxwell equations in the inclined co-ordinate
system are considered. From the frequency domain statement of Maxwell equation
(1.2):
HE ωµj−=×∇ , (3.3)
it follows that for a linearly polarised TM-type homogeneous plane
wave where is parallel to the
0,, == mzmy HH
mx,H xxσ component of mσ� . From equations (3.3) and
(1.18), we then obtain the family of equations:
'',','
,' zE
yE
Hj mymzmx ∂
∂−
∂∂
=− ωµ , (3.4)
'1 ,'
,,' y
HE mx
mtmz ∂
∂−=σ
, (3.5)
50
'1 ,'
,,' z
HE mx
mnmy ∂
∂=
σ. (3.6)
By differentiating equations (3.5) and (3.6) with respect to y' and z' respectively,
and by substituting these results into equation (3.4), one obtains the homogeneous
Helmholtz equation for anisotropic media:
0'
1'
1,'2
,'2
,2
,'2
,
=−∂
∂+
∂∂
mxmx
mt
mx
mn
HjzH
yH
ωµσσ
. (3.7)
From the co-ordinate rotations for transforming {x', y', z'} co-ordinates to {x, y,
z} co-ordinates (see Appendix 1), the co-ordinate rotations for transforming {x, y, z}
co-ordinates to {x', y', z'} co-ordinates can be written as:
−=
−=
zyx
zyx
zyx
mm
mm )(cossin0sincos0
001
'''
ααααα R , (3.8)
such that the partial derivatives of any function f can be re-written as:
mm zf
yf
yf αα sincos' ∂
∂+∂∂=
∂∂ , (3.9)
mmmm zf
zyf
yf
yf αααα 2
2
222
2
2
2
2
sincossin2cos' ∂
∂+∂∂
∂+∂∂=
∂∂ , (3.10)
mm zf
yf
zf αα cossin' ∂
∂+∂∂−=
∂∂ , (3.11)
mmmm zf
zyf
yf
zf αααα 2
2
222
2
2
2
2
coscossin2sin' ∂
∂+∂∂
∂−∂∂=
∂∂ . (3.12)
Equation (3.7) can be written in {x, y, z} co-ordinates as:
51
.0cossin 11 2
cossinsincos
,,
2
,
2,
2
,
2
,
2
2,
2
,
2
,
2
=−∂∂
∂
−+
∂∂
++
∂∂
+
mxmx
mmmtn,m
mx
mt
m
mn
mmx
mt
m
mn
m
Hjzy
Hσ
zH
yH
ωµαασ
σα
σα
σα
σα
(3.13)
The presence of the lateral inhomogeneities will generate anomalous magnetic
fields across the strike of the dyke. At an infinite distance from the dyke, the anomalous
fields must reduce to zero and the solution for the magnetic field will be identical to
that of a horizontally homogeneous medium. Near the dyke, the general solution for the
total magnetic field can be written as the sum of the background (b) and
anomalous (a) fields:
),(, zyH mx
),()(),( ,,, zyHzHzyH amx
bmxmx += , (3.14)
where is the anomalous field that exists due to the lateral inhomogeneities.
Since satisfies equation (3.13), then and must also
satisfy a form of equation (3.13) as linear sums of solution. In this particular problem,
the anisotropic medium is bound by a perfectly insulating layer (i.e., air) at the upper
boundary ( ), and a basement with isotropic conductivity
),(, zyH amx
),(, zyH mx
0=z
)(, zH bmx ),(, zyH a
mx
bσ at the lower
boundary ( ). A general solution for the background magnetic field , is a
homogeneous plane wave with both up-going and down-going components of the form:
hz = )(, zbmH x
)exp()exp()(, zkBzkAzH mmmmb
mx +−= , (3.15)
which satisfies equation (3.13) provided that the wave number is given by:
1
,
2
,
2 cossin−
+=
mt
m
mn
mm jk
σα
σαωµ , (3.16)
and to prevent an exponentially divergent solution in . It should be
noted that and in equation (3.15) are independent of {x, y, z, t} and are not
0Re >mk
mA
)(, zH bmx
mB
52
related to components of the vector potential A or magnetic flux density B. The
solution for is easily identified as the solution for the horizontally
homogeneous problem, as presented in Chapter 2.
)(, zH bmx
))
hhk
m
m
))
hkhk
m
m−
sinh((sinhhkhk
m
m
0)(, =zm
0)0 H=
()( gy mm
3.3. Exact formulation: perfectly insulating basement solution
3.3.1. Inclined anisotropic conductivity
If 0=bσ , then at and this boundary condition is
equivalent to the top of the basement being a perfect magnetic conductor [3.19]. It
follows (see Appendix 3) that the coefficients for equation (3.15) are given by:
0),(, =zyH mx hz =
sinh(2exp(0
kH
Am = , (3.17)
sinh(2exp(0H
Bm−
= , (3.18)
where is the magnetic field magnitude at , and is a constant which may be
complex. Equation (3.15) can then be written as:
0H 0=z
))
)( 0,
zHzH b
mx−
= . (3.19)
It is easily observed from equation (3.19) that when .
Similarly, when . This also implies that when
since . By separation of variables, the anomalous magnetic field will be
written as the product of two independent functions, and :
0)(, =zH bmx
)(, =zH amx
)(zgm
hz =
0=zH ax
,(y
hz = 0
,H mx
)(yfm
)),(, zfzyH amx = , (3.20)
53
where can be expressed as a Fourier series of sine terms with an argument of )(zgm
hznπ :
∑∞
=
=
1, sin)(
nnmm h
znAzg π ,
where n is the mode number (1 ) and where are the Fourier series
coefficients, and should not be confused with in equation (3.17), a coefficient of the
magnetic field wave equations. For convenience, we will include in the values of
at each n. As shown in Figure 3.2, this series satisfies the boundary conditions
that the anomalous magnetic field vanishes at both and .
∞ ...., 3, 2, , nmA ,
z =
mA
nmA ,
h
)(yfm
0=z
z = 0 z = h n = 1 n = 2 n = 3 n = 4
Figure 3.2. Fourier series representations, satisfying the boundary
conditions that for and .
)(, zg nm
0)(, =zH amx 0=z hz =
Equation (3.20) can then be written as:
=∑
∞
= hznyfzyH
nnm
amx
π1
,, sin)(),( . (3.21)
54
For , each term of equation (3.21) must satisfy a form of
equation (3.13) as a linear sum of solutions. For each term from equation (3.21), we
have the partial derivatives:
∞= ...., 3, 2, ,1n
∂∂
=∂
∂h
zny
yfy
zyH nma
nmx πsin)(),(
2,
2
2,,
2
, (3.22)
−=
∂∂
hznyf
hn
zzyH
nm
anmx ππ sin)(
),(,2
22
2,,
2
, (3.23)
∂∂
=∂∂
∂h
zny
yfh
nzy
zyH nma
nmx ππ cos)(),( ,,,
2
. (3.24)
The form of equation (3.13) that the anomalous fields must satisfy is then
written as:
.0sin)(
cos)(
cossin 11 2
sin)(cossin
sin)(sincos
,
,
,
,2
22
,
2
,
2
2,
2
,
2
,
2
=
−
∂∂
−+
+−
∂∂
+
hznyfj
hzn
yyf
hn
σ
hznyf
hn
hzn
yyf
nm
nmmm
mtn,m
nmmt
m
mn
m
nm
mt
m
mn
m
πωµ
ππαασ
ππσ
ασ
α
πσ
ασ
α
(3.25)
At the and boundaries, 0=z hz = 0sin =
hznπ and the shearing term:
0cos)(
cossin 11 2 ,
,
=
∂∂
−
hzn
yyf
hn
σnm
mmmtn,m
ππαασ
. (3.26)
At , 0=z 1cos =
hznπ , so equation (3.26) can only vanish for three possible
cases:
55
(a) if:
011
,
=−mtn,mσ σ
, (3.27)
which is the special case for an isotropic solution; or
(b) if:
0cossin =mm αα , (3.28)
which is only the special case of either 0=mα or , corresponding to
fundamental anisotropic solutions; or else,
o90=mα
(c) if:
ny
yf nm 0)(, ∀=
∂∂
. (3.29)
As a general solution for the inclined anisotropic problem is sought, equations
(3.27) and (3.28) are invalid (as they are special conditions) implying that equation
(3.29) must hold true in all cases. This means that equation (3.25) can be reduced to:
.0sin)(
sin)(cossin
sin)(sincos
,
,2
22
,
2
,
2
2,
2
,
2
,
2
=
−
+−
∂∂
+
hznyfj
hznyf
hn
hzn
yyf
nm
nmmt
m
mn
m
nm
mt
m
mn
m
πωµ
ππσ
ασ
α
πσ
ασ
α
(3.30)
A solution for must satisfy both equations (3.29) and (3.30). To satisfy
equation (3.30), a solution to is of form:
)(, yf nm
f )(, ynm
56
+
−=
hyq
bh
yqayf nm
nmnm
nmnm,
,,
,, expexp)( , (3.31)
provided that:
++
+=mmnmmt
mmnmmtmznm nhkq
ασασασασ
π 2,
2,
2,
2,2222
,, sincoscossin
, (3.32)
where is the vertical wave number: mzk ,
1
,
2
,
2
,sincos
−
+=
mt
m
mn
mmz jk
σα
σαωµ .
This solution satisfies the condition that equation (3.21) vanish for ∞=y , if we only
take the a-type terms for positive y and b-type terms for negative y both in m = 2, whilst
both positive and negative exponentials can exist in the finite region of m = 1.
Symmetry conditions at the 2ly ±= boundaries then permit us to write:
nmnm ba ,, = , (3.33)
which will ensure that is an even function about y = 0; i.e.,
. Employing this boundary condition is equivalent to using
one of the boundaries for solving the continuity of the magnetic field components, with
the remaining boundary condition to be available for solving the remainder of the
coefficients [3.11]. If the
),(, zyH mx
),(),( ,, zyHzyH mxmx −=
2l=y boundary is considered, then following from equation
(3.33), the use of symmetry implies that:
∑∞
=
−=
−−
112
,2,2
,1,1 sin
2exp
2cosh2
n
nn
nn HH
hzn
hlq
ahlq
a π , (3.34)
57
where:
hkzhkH
HHm
mbmxm sinh
)(sinh0,
−=≡ ,
for m = 1, 2. The expansion of into a sine series of argument 12 HH −h
znπ is written
as:
=− ∑
∞
= hznCHH
nn
π1
12 sin , (3.35)
where is a complex constant yet to be determined. At the boundary, term-by-term
must be equated, so both equations:
nC
nn
nn
n Ch
lqa
hlq
a =
−−
2
exp2
cosh2 ,2,2
,1,1 , (3.36)
02
exp2
sinh2 ,22,,2,2
,11,,1,1 =
−+
h
lqqa
hlq
qa nzznn
nzznn ρρ , (3.37)
must be satisfied, where equation (3.37) is obtained from the Maxwell equation:
yH
E mxmzzmz ∂
∂−= ,
,, ρ , (3.38)
stating the continuity of the tangential electric field across the boundary 2ly ±= .
Solutions for the a-type coefficients are then:
+
=
hlq
hlq
Ca
n
n
nzz
nzz
zznn
2sinh2
2cosh2 ,1
,2
,11,
,12,
2,,1
ρρ
ρ, (3.39)
58
+
−
=
hlq
hlq
hlq
hlq
Ca
nzz
nzz
nnzzn
n
2sinh
2cosh
2sinh
2exp
,11,
,12,
,1,21,
,2
ρρ
ρ. (3.40)
Following the expansion of into an odd Fourier series with argument 12 HH −
hznπ , one obtains:
∫
−=
h
n dzh
znHHh
C0
12 sin)(2 π , (3.41)
where:
)sinh()(sinh
)sinh()(sinh
1
10
2
2012 hk
zhkHhk
zhkHHH
−−
−=− . (3.42)
Following through with equation (3.41) using integration by parts (see
Appendix 5), the solution for is: nC
+
+
−−=
2
222
12
222
22
21
220 )(2
hnk
hnkh
kknHCn ππ
π. (3.43)
Substituting equation (3.43) into equations (3.39) and (3.40), solutions for the a-
type coefficients now take the form:
+
+
+
−−=
hlq
hlq
hnk
hnkh
kknHa
n
n
nzz
nzz
zzn
2sinh
2cosh
)(
,1
,2
,11,
,12,2
222
12
222
22
21
222,0
,1
ρρππ
πρ,
(3.44)
59
+
+
+
−−
=
hlq
hlq
hnk
hnkh
hlq
hlq
kknHa
n
n
nzz
nzz
nnzz
n
2cosh
2sinh
2sinh
2exp)(2
,1
,1
,22,
,11,2
222
12
222
22
,1,221
221,0
,2
ρρππ
πρ.
(3.45)
For 22lyl ≤≤− , the anomalous magnetic field is written as:
=∑
∞
= hzn
hyq
azyHn
nn
ax
π1
,1,11, sincosh2),( . (3.46)
The total magnetic field can then be written as:
+
−= ∑
∞
= hzn
hyq
ahk
zhkHzyH
n
nnx
π1
,1,1
1
101, sincosh2
)sinh()(sinh
),( . (3.47)
From Maxwell equations (3.5) and (3.6) rotated about the x-axis, we have:
zH
E xyyy ∂
∂= 1,
1,1, ρ , (3.48)
so the horizontal electric field can be written as:
.coscosh2
)sinh()(cosh
),(
1
,1,11,
1
101,11,
+
−−=
∑∞
= hzn
hyq
ah
n
hkzhkHk
zyE
n
nnyy
yyy
ππρ
ρ
(3.49)
At the air-half space interface ( ), equation (3.47) reduces to: 0=z
01, )0,( HyH x = , (3.50)
and is constant ∀ . For , equation (3.49) reduces to: y 0=z
60
∑∞
=
+
−=
1
,1,11,
1
101,11, .cosh2
)sinh()cosh(
),(n
nnyy
yyy h
yqa
hn
hkhkHk
zyE πρρ
(3.51)
The surface impedance is defined as:
−= ∑
∞
= hyq
Ha
hnhkkyZ nn
nyyyyyx
,1
0
,1
11,11,1 cosh2)coth()0,( πρρ . (3.52)
In equation (3.52), the term:
)coth(sincos)coth( 11,
12
1,
12
11,1 hkjhkknt
yy
+=
σα
σαωµρ , (3.53)
is identified as the surface impedance of a laterally homogeneous layer above a perfect
magnetic conducting basement (see Chapter 2). Once equation (3.44) is substituted into
(3.52), one obtains:
.
2sinh
2cosh
cosh
)(2
)coth(sincos)0,(
1 ,1
,2
,11,
,12,2
222
12
222
2
,12
1,2,2
12
23
2
11,
12
1,
12
∑∞
=
+
+
+
×
−+
+=
n n
n
nzz
nzz
n
yyzz
ntyx
hlq
hlq
hnk
hnk
hyq
n
kkh
hkjyZ
ρρππ
ρρπ
σα
σαωµ
(3.54)
For 2ly ≥ and
2ly −≤ , the constant term is the same as for medium 1, however
the subscripts are interchanged. One writes the anomalous magnetic field as:
−=∑
∞
= hzn
hyq
azyHn
nn
ax
π1
,2,22, sinexp),( , (3.55)
61
where . The total magnetic field can then be written as: 0Re ,2 >nq
−+
−= ∑
∞
= hzn
hyq
ahk
zhkHzyH
n
nnx
π1
,2,2
2
202, sinexp
)sinh()(sinh
),( . (3.56)
From Maxwell equations (3.5) and (3.6) rotated about the x-axis, we have:
zH
E xyyy ∂
∂= 2,
2,2, ρ , (3.57)
so the horizontal electric field is written as:
.cosexp
)sinh()(cosh
),(
1
,2,22,
2
202,22,
−+
−−=
∑∞
= hzn
hyq
ah
n
hkzhkHk
zyE
n
nnyy
yyy
ππρ
ρ
(3.58)
At the air-half space interface ( ), equation (3.56) reduces to: 0=z
02, )0,( HyH x = , (3.59)
and is constant ∀ . For , equation (3.58) reduces to: y 0=z
.exp
)sinh()cosh(
),(
1
,2,22,
2
202,22,
∑∞
=
−+
−=
n
nnyy
yyy
hyq
ah
n
hkhkHk
zyE
πρ
ρ
(3.60)
The surface impedance is defined as:
−−= ∑
∞
= hyq
Ha
hnhkkyZ nn
nyyyyyx
,2
0
,2
11,22,2 exp)coth()0,( πρρ . (3.61)
In equation (3.52), the term:
62
)coth(sincos)coth( 22,
22
2,
22
22,2 hkjhkknt
yy
+=
σα
σαωµρ , (3.62)
is identified as the surface impedance of a laterally homogeneous layer above a perfect
magnetic conducting basement (see Chapter 2). Once equation (3.45) is substituted into
(3.61), one obtains:
.
2cosh
2sinh
2sinh
2expexp
)(2
)coth(sincos)0,(
1 ,1
,1
,22,
,11,2
222
12
222
2
,1,2,22
1,2,2
12
23
2
22,
22
2,
22
∑∞
=
+
+
+
×
−+
+=
n n
n
nzz
nzz
nnn
yyzz
ntyx
hlq
hlq
hnk
hnk
hlq
hlq
hyq
n
kkh
hkjyZ
ρρππ
ρρπ
σα
σαωµ
(3.63)
Equation (3.63) is the complete exact solution for the surface impedance of a
vertical dyke with inclined anisotropic conductivity embedded in an otherwise
homogeneous layer above a perfect magnetic conductor.
3.3.2. Fundamental anisotropic conductivity
The formulations derived in Section 3.3.1 will now be reduced to the problem
of a vertical dyke with fundamental uniaxial anisotropy. In this case, it can be assumed
that 021 ==αα . Firstly, it is noted that the wave number for medium m is now given
by:
mtm jk ,ωµσ= , (3.64)
where to prevent exponentially divergent solutions. Now equation (3.28) is
satisfied and a solution to equation (3.30) is sought. Equation (3.31) is still a valid
solution, and equation (3.32) reduces to:
0Re >mk
63
mt
mnmznm nhkq
,
,2222,, σ
σπ+= , (3.64)
where:
mnmz jk ,, ωµσ= .
We can re-write equation (3.64) as:
2
2222
,m
mnm
nhkqλ
π+= , (3.65)
where mλ is the coefficient of anisotropy of medium m. We note that equation (3.65) is
identical to the solution by Obukhov [3.10]. It follows that:
+
+
+
−−=
hlq
hlq
hnk
hnkh
kknHan
n
n
n
nnn
2sinh
2cosh
)(
,1
,2
,1
1,
2,,12
222
12
222
22
21
220
,1
σσππ
π,
(3.66)
+
+
+
−−
=
hlq
hlq
hnk
hnkh
hlq
hlq
kknHa
n
n
n
n
nn
nn
n
2cosh
2sinh
2sinh
2exp)(2
,1
,1
,2
2,
1,,12
222
12
222
22
,1,221
220
,2
σσππ
π,
(3.67)
and that for 22lyl ≤≤− , the surface impedance can be written as:
64
.
2sinh
2cosh
cosh
)(2
)coth()0,(
1 ,1
,2
,1
1,
2,,12
222
12
222
2
,12
1,2
12
23
2
11,
∑∞
=
+
+
+
×
−+
=
n n
n
n
n
nn
n
n
tyx
hlq
hlq
hnk
hnk
hyq
n
kkh
hkjyZ
σσππ
ρπ
σωµ
(3.68)
For 2ly ≥ and
2ly −≤ , the surface impedance can be written as:
.
2cosh
2sinh
2sinh
2expexp
)(2
)coth()0,(
1 ,1
,1
,2
2,
1,,12
222
12
222
2
,1,2,22
2,2
12
23
2
22,
∑∞
=
+
+
+
×
−+
=
n n
n
n
n
nn
nnn
n
tyx
hlq
hlq
hnk
hnk
hlq
hlq
hyq
n
kkh
hkjyZ
σσππ
ρπ
σωµ
(3.69)
Equations (3.68) and (3.69) were identically obtained by Obukhov [3.10].
Further, it can be observed that when mmnmt σσσ == ,, , then for 22lyl ≤≤− , the
surface impedance is given by:
65
.
2sinh
2cosh
cosh
)(2
)coth()0,(
1 ,1
,2
,1,12
222
12
222
2
,12
12
12
23
2
11
∑∞
=
+
+
+
×
−+
=
n n
n
nn
n
yx
hlq
hlq
hnk
hnk
hyq
n
kkh
hkjyZ
ππ
ρπ
σωµ
(3.70)
For 2ly ≥ and
2ly −≤ , the surface impedance can be written as:
,
2cosh
2sinh
2sinh
2expexp
)(2
)coth()0,(
1 ,1
,1
,2,12
222
12
222
2
,1,2,22
22
12
23
2
22,
∑∞
=
+
+
+
×
−+
=
n n
n
nn
nnn
tyx
hlq
hlq
hnk
hnk
hlq
hlq
hyq
n
kkh
hkjyZ
ππ
ρπ
σωµ
(3.71)
which are identical equations to those derived by Rankin [3.11]. It is noted that
Rankin�s [3.11] solution was derived using the cgs electromagnetic units (emu) in
which µ is dimensionless and equal to unity in free space.
3.4. Exact formulation: perfectly conducting basement solution
3.4.1. Inclined anisotropic conductivity
If ∞=bσ , then at , and this boundary condition is
equivalent to the top of the basement being a perfect electric conductor. It follows (see
Appendix 3) that the coefficients for equation (3.15) are given by:
0),(, =zyE my hz =
66
)cosh(2)exp(0
hkhkH
Am
mm = , (3.72)
)cosh(2)exp(0
hkhkH
Bm
mm
−= , (3.73)
where is the magnetic field magnitude at , and is constant which may be
complex. Equation (3.15) can then be written as:
0H 0=z
)cosh()(cosh
)( 0, hk
zhkHzH
m
mbmx
−= . (3.74)
It is easily observed that 0)(, =
∂∂
zhH b
mx . Similarly, 0)(, =
∂∂
zhH a
mx . Also,
since . By separation of variables, the anomalous
magnetic field can be written as:
0)0,(, =yH amx 0, )0,( HyH mx =
)()(),(, zgyfzyH mma
mx = , (3.75)
where can be expressed as a Fourier series of sine terms with an argument of )(zgm
hn
22( z)1 π+ :
∑∞
=
+=
0, 2
)12(sin)(n
nmm hznAzg π ,
where n is the mode number (0,1 ) and where are the Fourier
series coefficients, and should not be confused with in equation (3.72), a coefficient
of the magnetic field wave equations. As shown in Figure 3.3, this series satisfies the
boundary conditions that the anomalous magnetic field vanishes at and that
∞ ...., 3, 2, ,
A
nmA ,
m
0=z
0)(, =
∂∂
=hz
amx
zzH
.
67
Equation (3.75) can then be written as:
+=∑
∞
= hznyfzyH
nnm
amx 2
)12(sin)(),(0
,,π . (3.76)
For , each term of equation (3.76) must satisfy a form of
equation (3.13). From equation (3.76), we have the derivatives:
∞= ...., 3, 2, ,1 ,0n
+
∂∂
=∂
∂h
zny
yfy
zyH nma
nmx
2)12(sin
)(),(2
,2
2,,
2 π , (3.77)
++−=
∂∂
hznyf
hn
zzyH
nm
anmx
2)12(sin)(
4)12(),(
,2
22
2,,
2 ππ , (3.78)
+
∂∂+=
∂∂∂
hzn
yyf
hn
zyzyH nm
anmx
2)12(cos
)(2
)12(),( ,,,2 ππ . (3.79)
z = 0 z = h n = 0 n = 1 n = 2 n = 3
Figure 3.3. Fourier series representations, satisfying the boundary
conditions that for and
)(, zg nm
0)(, =zH amx 0=z 0
)(, =∂
∂z
hH amx at . hz =
68
The form of equation (3.13) that the anomalous fields must satisfy is then
written as:
.02
)12(sin)(
2)12(cos
)()12(cossin 11
2)12(sin)(
4)12(cossin
2)12(sin
)(sincos
,
,
,
,2
22
,
2
,
2
2,
2
,
2
,
2
=
+−
+
∂∂+
−+
++
+−
+
∂∂
+
hznyfj
hzn
yyf
hn
σ
hznyf
hn
hzn
yyf
nm
nmmm
mtn,m
nmmt
m
mn
m
nm
mt
m
mn
m
πωµ
ππαασ
ππσ
ασ
α
πσ
ασ
α
(3.80)
At the boundary, 0=z 02
)12( =
+
hzn πsin and the shearing term:
02
)12(cos)()12(cossin 11 ,
,
=
+
∂∂+
−
hzn
yyf
hn
σnm
mmmtn,m
ππαασ
. (3.81)
At , 0=z 12
)12(cos =
+
hzn π , so equation (3.81) can only vanish if either
equations (3.27) or (3.28) are valid, or else if:
ny
yf nm 0)(, ∀=
∂∂
, (3.29)
is valid. The same arguments as used in Section 3.3 to justify the property of equation
(3.29) are valid. This means that equation (3.80) can be reduced to the equation:
.02
)12(sin)(
2)12(sin)(
4)12(cossin
2)12(sin
)(sincos
,
,2
22
,
2
,
2
2,
2
,
2
,
2
=
+−
++
+−
+
∂∂
+
hznyfj
hznyf
hn
hzn
yyf
nm
nmmt
m
mn
m
nm
mt
m
mn
m
πωµ
ππσ
ασ
α
πσ
ασ
α
(3.82)
69
A solution for must satisfy both equations (3.29) and (3.82). To satisfy
equation (3.82), a solution to is of form:
)(, yf nm
f )(, ynm
+
−=
hyq
bh
yqayf nm
nmnm
nmnm 2exp
2exp)( ,
,,
,, , (3.83)
provided that:
++
++=mmnmmt
mmnmmtmznm nhkq
ασασασασ
π 2,
2,
2,
2,2222
,, cossincossin
)12(4 , (3.84)
where is the vertical wave number: mzk ,
1
,
2
,
2
,sincos
−
+=
mt
m
mn
mmz jk
σα
σαωµ .
This solution satisfies the condition that equation (3.76) vanish for ∞=y , if we only
take the a-type terms for positive y and b-type terms for negative y both in m = 2, whilst
both positive and negative exponentials can exist in the finite region of m = 1.
Symmetry conditions at the 2ly ±= boundaries then permit us to write:
nmnm ba ,, = , (3.85)
which will ensure that is an even function about y = 0; i.e.,
. Employing this boundary condition is equivalent to using
one of the boundaries for solving the continuity of the magnetic field components, with
the remaining boundary condition to be available for solving the remainder of the
coefficients. If the
),(, zyH mx
),(),( ,, zyHzyH mxmx −=
2ly = boundary is considered, then following from equation (3.85),
the use of symmetry implies that:
70
∑∞
=
−=
+
−−
012
,2,2
,1,1 2
)12(sin4
exp4
cosh2n
nn
nn HH
hzn
hlq
ahlq
a π , (3.86)
where:
hkzhkH
HHm
bmxm sinh
)(sinh0,
−=≡ ,
for m = 1, 2. The expansion of into a sine series of argument 12 HH −h
zn2
)12( π+ is
written as:
+=− ∑
∞
= hznCHH
nn 2
)12(sin0
12π , (3.87)
where is a complex constant yet to be determined. At the boundary, term-by-term
must be equated, so both equations:
nC
nn
nn
n Ch
lqa
hlq
a =
−−
4
exp4
cosh2 ,2,2
,1,1 , (3.88)
04
exp4
sinh2 ,22,,2,2
,11,,1,1 =
−+
h
lqqa
hlq
qa nzznn
nzznn ρρ , (3.89)
must be satisfied, where equation (3.89) is obtained from the Maxwell equation:
yH
E mxmzzmz ∂
∂−= ,
,, ρ , (3.38)
stating the continuity of the tangential electric field across the boundary 2ly ±= .
Solutions for the a-type coefficients are then:
71
+
=
hlq
hlq
Ca
n
n
nzz
nzz
zznn
4sinh2
4cosh2 ,1
,2
,11,
,12,
2,,1
ρρ
ρ, (3.90)
+
−
=
hlq
hlq
hlq
hlq
Ca
nzz
nzz
nnzzn
n
4sinh
4cosh
4sinh
4exp
,11,
,12,
,1,21,
,2
ρρ
ρ. (3.91)
Following the expansion of into an odd Fourier series with argument 12 HH −
hzn
2)12( π+ , one obtains:
∫
+−=
h
n dzh
znHHh
C0
12 2
)12(sin)(1 π , (3.92)
where:
)cosh()(cosh
)cosh()(cosh
1
10
2
2012 hk
zhkHhk
zhkHHH
−−
−=− . (3.93)
Following through with equation (3.92) using integration by parts (see
Appendix 5), the solution for is: nC
++
++
−+−=
2
222
12
222
22
21
220
4)12(
4)12(
)()12(
hnk
hnkh
kknHCn ππ
π. (3.94)
Substituting equation (3.94) into equations (3.90) and (3.91), solutions for the a-
type coefficients now take the form:
,
4sinh
4cosh
4)12(
4)12(2
)()12(
,1
,2
,1
2,
1,,12
222
12
222
22
21
220
,1
+
++
++
−+−=
hlq
hlq
hnk
hnkh
kknHan
n
n
zz
zznn
ρρππ
π
(3.95)
72
.
4cosh
4sinh
4)12(
4)12(
2sinh
2exp)()12(
,1
,1
,2
1,
2,,12
222
12
222
22
,1,221
220
,2
+
++
++
−+−
=
hlq
hlq
hnk
hnkh
hlq
hlq
kknHa
n
n
n
zz
zzn
nn
n
ρρππ
π
(3.96)
For 22lyl ≤≤− , the anomalous magnetic field is written as:
+
=∑
∞
= hzn
hyq
azyHn
nn
ax 2
)12(sin2
cosh2),(0
,1,11,
π . (3.97)
The total magnetic field can then be written as:
+
+
−= ∑
∞
= hzn
hyq
ahk
zhkHzyH
n
nnx
π)12(sin2
cosh2)cosh(
)(cosh),(
0
,1,1
1
101, . (3.98)
From Maxwell equations (3.5) and (3.6) rotated about the x-axis, we have:
zH
E xyyy ∂
∂= 1,
1,1, ρ , (3.48)
so the horizontal electric field can be written as:
.2
)12(cos2
cosh2
)12(
)cosh()(sinh
),(
0
,1,11,
1
101,11,
+
++
−−=
∑∞
= hzn
hyq
ah
n
hkzhkHk
zyE
n
nnyy
yyy
ππρ
ρ
(3.99)
At the air-half space interface ( ), equation (3.98) reduces to: 0=z
01, )0,( HyH x = , (3.100)
and is constant ∀ . For , equation (3.99) reduces to: y 0=z
73
∑∞
=
++−
=0
,1,11,
1
101,11, .
2cosh)12(
)cosh()sinh(
),(n
nnyy
yyy h
yqa
hn
hkhkHk
zyE πρρ
(3.101)
The surface impedance is defined as:
+−= ∑∞
= hyq
Ha
hnhkkyZ nn
nyyyyyx 2
cosh)12()tanh()0,( ,1
0
,1
01,11,1
πρρ . (3.102)
In equation (3.102), the term:
)tanh(sincos)tanh( 11,
12
1,
12
11,1 hkjhkknt
yy
+=
σα
σαωµρ , (3.103)
is identified as the surface impedance of a laterally homogeneous layer above a perfect
electric conducting basement (see Chapter 2). Once equation (3.95) is substituted into
(3.103), one obtains:
.
2sinh
2cosh
4)12(
4)12(
cosh)12(
)(2
)tanh(sincos)0,(
0 ,1
,2
,1
2,
1,,12
222
12
222
2
,12
1,2
12
23
2
11,
12
1,
12
∑∞
=
+
++
++
+
×
−+
+=
n n
n
n
zz
zzn
n
yy
ntyx
hlq
hlq
hnk
hnk
hyq
n
kkh
hkjyZ
ρρππ
ρπ
σα
σαωµ
(3.104)
For 2ly ≥ and
2ly −≤ , the constant term is the same as for medium 1, however
the subscripts are interchanged. One writes the anomalous magnetic field as:
+
−=∑
∞
= hzn
hyq
azyHn
nn
ax 2
)12(sin2
exp),(0
,2,22,
π , (3.105)
74
where . The total magnetic field can then be written as: 0Re ,2 >nq
+
−+
−= ∑
∞
= hzn
hyq
ahk
zhkHzyH
n
nnx 2
)12(sin2
exp)cosh(
)(cosh),(
0
,2,2
2
202,
π . (3.106)
From Maxwell equations (3.5) and (3.6) rotated about the x-axis, we have:
zH
E xyyy ∂
∂= 2,
2,2, ρ , (3.57)
so the horizontal electric field is written as:
.2
)12(cos2
exp2
)12(
)cosh()(sinh
),(
0
,2,22,
2
202,22,
+
−++
−−=
∑∞
= hzn
hyq
ah
n
hkzhkHk
zyE
n
nnyy
yyy
ππρ
ρ
(3.107)
At the air-half space interface ( ), equation (3.105) reduces to: 0=z
02, )0,( HyH x = , (3.108)
and is constant ∀ . For , equation (3.107) reduces to: y 0=z
.2
exp2
)12(
)cosh()sinh(
),(
0
,2,22,
2
202,22,
∑∞
=
−++
−=
n
nnyy
yyy
hyq
ah
n
hkhkHk
zyE
πρ
ρ
(3.109)
The surface impedance is defined as:
−+−= ∑
∞
= hyq
Ha
hnhkkyZ nn
nyyyyyx 2
exp2
)12()tanh()0,( ,2
0
,2
01,22,2
πρρ . (3.110)
In equation (3.110), the term:
75
)tanh(sincos)tanh( 22,
22
2,
22
22,2 hkjhkknt
yy
+=
σα
σαωµρ , (3.111)
is identified as the surface impedance of a laterally homogeneous layer above a perfect
electric conducting basement (see Chapter 2). Once equation (3.96) is substituted into
(3.110), one obtains:
.
4cosh
4sinh
4)12(
4)12(
4sinh
4exp
2exp)12(
)(
)tanh(sincos)0,(
0 ,1
,1
,2
1,
2,,12
222
12
222
2
,1,2,22
2,2
12
23
2
22,
22
2,
22
∑∞
=
+
++
++
+
×
−+
+=
n n
n
n
zz
zzn
nnn
zz
ntyx
hlq
hlq
hnk
hnk
hlq
hlq
hyq
n
kkh
hkjyZ
ρρππ
ρπ
σα
σαωµ
(3.112)
Equation (3.112) is the complete exact solution for the surface impedance of a
vertical dyke with inclined anisotropic conductivity embedded in an otherwise
homogeneous layer above a perfect electric conductor.
3.4.2. Fundamental anisotropic conductivity
The formulations derived in Section 3.4.1 will now be reduced to the problem
of a vertical dyke with fundamental uniaxial anisotropy. In this case, it can be assumed
that 021 ==αα . Similar to Section 3.3.2, it is noted that the wave number for medium
m is now given by:
mtm jk ,ωµσ= , (3.64)
where to prevent exponentially divergent solutions. Now equation (3.28) is
satisfied and a solution to equation (3.80) is sought. Equation (3.83) is still a valid
solution, and it is noticed that equation (3.84) reduces to:
0Re >mk
76
mt
mnmznm nhkq
,
,2222,, )12(4
σσ
π++= , (3.113)
where:
mnmz jk ,, ωµσ= .
We can also write equation (3.113) as:
2
2222
,)12(4
m
mnm
nhkqλ
π++= , (3.114)
where mλ is the coefficient of anisotropy of medium m. We note that equation (3.114)
is identical to the solution of Obukhov [3.10]. It follows that:
,
4sinh
4cosh
4)12(
4)12(2
)()12(
,1
,2
,1
1,
2,,12
222
12
222
22
21
220
,1
+
++
++
−+−=
hlq
hlq
hnk
hnkh
kknHan
n
n
n
nnn
σσππ
π
(3.115)
+
++
++
−+−
=
hlq
hlq
hnk
hnkh
hlq
hlq
kknHa
n
n
n
n
nn
nn
n
4cosh
4sinh
4)12(
4)12(
4sinh
4exp)()12(
,1
,1
,2
2,
1,,12
222
12
222
22
,1,221
220
,2
σσππ
π,
(3.116)
and that for 22lyl ≤≤− , the surface impedance can be written as:
77
.
4sinh
4cosh
4)12(
4)12(
2cosh)12(
)(2
)tanh()0,(
0 ,1
,2
,1
1,
2,,12
222
12
222
2
,12
1,2
12
23
2
11,
∑∞
=
+
++
++
+
×
−+
=
n n
n
n
n
nn
n
n
tyx
hlq
hlq
hnk
hnk
hyq
n
kkh
hkjyZ
σσππ
ρπ
σωµ
.
(3.117)
For 2ly ≥ and
2ly −≤ , the surface impedance can be written as:
.
4cosh
4sinh
4)12(
4)12(
4sinh
4exp
2exp)12(
)(2
)tanh()0,(
0 ,1
,1
,2
2,
1,,12
222
12
222
2
,1,2,22
2,2
12
23
2
22,
∑∞
=
+
++
++
+
×
−+
=
n n
n
n
n
nn
nnn
n
tyx
hlq
hlq
hnk
hnk
hlq
hlq
hyq
n
kkh
hkjyZ
σσππ
ρπ
σωµ
(3.118)
Further, it can be observed that when mmnmt σσσ == ,, , then for 22lyl ≤≤− ,
the surface impedance is given by:
.
4sinh
4cosh
4)12(
4)12(
2cosh)12(
)(2
)tanh()0,(
0 ,1
,2
,1,12
222
12
222
2
,12
12
12
23
2
11
∑∞
=
+
++
++
+
×
−+
=
n n
n
nn
n
yx
hlq
hlq
hnk
hnk
hyq
n
kkh
hkjyZ
ππ
ρπ
σωµ
. (3.119)
78
For 2ly ≥ and
2ly −≤ , the surface impedance can be written as:
,
4cosh
4sinh
4)12(
4)12(
4sinh
4exp
2exp)12(
)(2
)coth()0,(
0 ,1
,1
,2,12
222
12
222
2
,1,2,22
22
12
23
2
22,
∑∞
=
+
++
++
+
×
−+
=
n n
n
nn
nnn
tyx
hlq
hlq
hnk
hnk
hlq
hlq
hyq
n
kkh
hkjyZ
ππ
ρπ
σωµ
(3.120)
which are identical equations to those derived by Rankin [3.11]. Again, it is noted that
Rankin�s [3.11] solution was derived using the cgs electromagnetic units (emu) in
which µ is dimensionless and equal to unity in free space.
3.5. Discussion
In all discussions here, we have considered only the propagation of
homogeneous TM-type waves. In Chapter 2, it was demonstrated that for a half space,
and for a horizontally stratified half space, the inclined uniaxial anisotropic
conductivity tensor can be written as a corresponding fundamental bianisotropic
conductivity tensor where:
mxxmxxmx ,1
,, σρσ == − , (3.121)
mzz
mzymyzmyymyy
mn
m
mt
mmy
,
,,,
1,
1
,
2
,
2
,sincos
σσσ
σρσ
ασ
ασ −==
+= −
−
, (3.122)
myy
myzmzymzzmzz
mn
m
mt
mmz
,
,,,
1,
1
,
2
,
2
,cossin
σσσ
σρσ
ασ
ασ −==
+= −
−
. (3.123)
79
It has been proven in this chapter that this same tensor equivalence exists for
horizontally inhomogeneous media. It has been demonstrated that the shearing term in
the rotated Helmholtz equation vanishes for horizontally inhomogeneous media, as in
the case for horizontally homogeneous media. Further, if one compares equations
(3.54) and (3.63) with equations (3.68) and (3.69) respectively, it should be noticed that
the equations are identical with the exception that in equations (3.54) and (3.63), the
conductivity terms are given by equations (3.122) and (3.123). Hence, it is concluded
that two-dimensional problems with inclined uniaxial anisotropic conductivity can be
equivalently described as a two-dimensional problem with fundamental biaxial
anisotropic conductivity. This has important applications to approximate methods of
solution. These applications will be discussed in Chapter 4.
3.6. References
[3.1] D. N. Chetaev, Bull. Acad. Sci. USSR Geophys. Ser. 4, 107 (1960).
[3.2] P. C. Clemmow, Phil. Trans. R. Soc. Lond. A 246, 1 (1953).
[3.3] I. d�Erceville & G. Kunetz, Geophysics 27, 651 (1962); 28, 490 (1963).
[3.4] V. I. Dmitriev & E. V. Zakharov, Phys. Solid Earth 6, 719 (1970).
[3.5] R. G. Geyer, Geophysics 37, 337 (1972).
[3.6] D. Grubert, presented at 12th Workshop Electromagn. Induction Earth, Brest,
France, Aug. 1994.
[3.7] F. W. Jones & A. T. Price, Geophys. J. R. astron. Soc. 20, 317 (1970).
[3.8] G. V. Koschlakov, Geol. Geofiz. 11, 119 (1970).
[3.9] G. G. Obukhov, Izv. Akad. Nauk SSSR Fiz. Zemli 4, 89 (1969).
[3.10] G. G. Obukhov, Izv. Akad. Nauk SSSR Fiz. Zemli 4, 106 (1969).
[3.11] D. Rankin, Geophysics 27, 666 (1962); 28, 490 (1963).
[3.12] I. K. Reddy & D. Rankin, Pure App. Geophys. 105, 847 (1973).
[3.13] R. C. Robertson, IEEE Trans. Geosci. Remote Sensing 27, 369 (1989).
[3.14] E. E. S. Sampaio & D. Dias, Geophys. Prosp. 49, 107 (2001).
[3.15] E. E. S. Sampaio & J. T. Fokkema, J. Geophys. Res. 97, 1953 (1992).
[3.16] U. Schmucker, in Protokoll Kolloqium Elektromagnetische Tiefenforschung
Hochst im Odenwald, edited by K. Bahr & A. Junge (Deutsche Geofys.
Gesellschaft, Germany, 1994), pp. 3-26.
[3.17] R. Truemann, Acta Geod. Mont. Acad. Sci. Hung. 5, 61 (1970).
[3.18] J. R. Wait & K. P. Spies, J. Geomagn. Geoelec. 26, 449 (1974).
80
[3.19] J. T. Weaver, Geophysics 28, 1386 (1963).
[3.20] J. T. Weaver, Can. J. Phys. 41, 484 (1963).
[3.21] J. T. Weaver, B. V. Le Quang & G. Fischer, Geophys. J. R. astron. Soc. 87, 263
(1985).
[3.22] J. T. Weaver, B. V. Le Quang & G. Fischer, Geophys. J. R. astron. Soc. 87, 917
(1986).
81
Chapter 4
Approximate solutions for two-dimensional electrical anisotropy problems
4.1. Introduction
The methods of approximate solutions for modelling a system use a
mathematical model that can be solved numerically to represent the system developed.
The results obtained from the approximate solution are then interpreted in terms of the
original system and serve to develop an understanding of the physical processes
involved [4.20]. As discussed in Chapter 2, one-dimensional exact solutions for
inclined anisotropy, where the directions of anisotropy are at arbitrary angles to the
fundamental co-ordinate system employed, have also been presented previously and
used to explain experimental observations. Further, in Chapter 3, exact solutions were
derived for a two-dimensional dyke embedded in a host layer. In all of these solutions,
the conductivity boundaries correspond to continuous analytical functions in the
electromagnetic fields.
In practical problems however, the inclined anisotropy occurs in structures with
arbitrary shape for which an exact solution may not be obtained easily, if at all. For
such problems, it is essential to employ approximate (or numerical) methods of
solution. Finite-element, finite-difference and integral equation solutions have all been
employed in modeling arbitrary two- and three-dimensional geological structures
[4.19]. However in these instances, the inhomogeneous models were considered as a
spatial juxtaposition of different media with isotropic conductivity. The modelling of
inhomogeneous media with isotropic conductivity is physically intuitive as far as
understanding multi-dimensional electromagnetic induction in complex objects is
concerned. Of specific interest to the geophysical community now is the ability to
model inhomogeneous media with anisotropic conductivity. Frequency-domain
differential equation methods of approximate solutions for such media have been
previously presented using finite-element [4.4, 4.9, 4.11, 4.18] and finite-difference
methods [4.1, 4.2, 4.6-4.10, 4.14, 4.15, 4.19].
It has been demonstrated in Chapters 2 and 3 that two-dimensional problems
involving TM-type incidence on media with inclined uniaxial anisotropic conductivity
were equivalent to problems involving TM-type incidence on media with fundamental
biaxial anisotropic conductivity. The effective horizontal and vertical conductivity
82
values were obtained from the diagonal components of the Euler rotation of the
resistivity tensor into the horizontal and vertical planes respectively, and were shown to
be functions of the diagonal and non-diagonal terms from the Euler rotation of the
conductivity tensor into the horizontal and vertical planes respectively. In this chapter,
we will introduce the self-consistent impedance method of Thiel & Mittra [4.13], based
on a finite-difference approximation of an RC circuit analogy, to model inhomogeneous
media with inclined uniaxial anisotropic conductivity in two-dimensions. RC network
analogies are quite lucid tools for simulating induction problems, particularly when one
is faced with some of the more abstract mathematical concepts behind finite-element,
integral equation and hybrid methods of approximate solution. The work presented in
this chapter has been published as [4.17] and submitted for publication as [4.16].
4.2. Formulation: self-consistent impedance method
A two-dimensional self-consistent form of the impedance method was recently
published and its applications to surface impedance modeling of isotropic media were
discussed [4.13]. The definition of self-consistent is that the magnetic field is assumed
to be unknown everywhere in the solution space (with the exception of the source
terms) and is independent of the model. Previous formulations of the impedance
method [4.3] assumed that the magnetic field everywhere was known and was
dependent of the model.
The self-consistent method requires the solution space to be divided into
rectangular cells bound by four impedance elements as
shown in Figure 4.1. The applied field is a magnetic field impressed on one or more
cells with an implied time variance of exp(
kikikiki ZZZZ ,,4,,3,,2,,1 and , , ,
)tjω . In the formulation presented in this
chapter, it is assumed that 0µµ = everywhere in the solution space. For a quasi-static
solution, displacement currents are ignored everywhere in the solution space. However,
displacement currents can be introduced simply by considering a complex conductivity.
Adopting the finite-difference approach (see Appendix 6), we introduce the notation
(i,k) for the y-z directions, to avoid using ( j,k) since j is the 1− in this thesis. Each
cell has sides and lies in the yz plane. By applying Faraday�s Law to the
(i,k)
k,iki zy , and ∆∆th cell, one can write:
83
, )( )()()(
,,,0,,4,1,
,,31,,,,2,1,,,11,,
kikikikikiki
kikikikikikikikiki
zyHjZIIZIIZIIZII∆∆−=−+
−+−+−
−
++−
ωµ (4.1)
where is the current in the (i,k)kiI ,th element, is the magnetic field component H
normal to the (i,k)
kiH , x
th cell, and are the cell dimensions and , for m = 1, 2,
3 and 4, are the element impedance values given by:
kiy ,∆ kiz ,∆ kimZ ,,
kikikim
kikim zx
yZ
,,,,
,,, ∆∆
∆=
σ, (4.2)
where kim ,,σ is the conductivity of the material in the (i,k)th cell and is the
direction normal to the cell. By applying a finite-difference formulation of Ampere�s
Law in integral form along a rectangular path in the xz plane around (see Figure
4.1), the current in impedance element is given by:
kix ,∆
kiZ ,,1
kiI ,,1
kikikikikiki xHHIII ,1,,1,,,,1 )( ∆−=−= −− , (4.3)
which is recognized as a finite-difference form of boundary condition (1.39). Using
equation (4.3), equation (4.1) can be re-written as:
( ) ( ) ( ) ( )
( ) ( ) ,0
122
2,,,
,12
,,,
1,
2,,,
,12
,,,
1,,2
,,,2
,,,
=∆
−∆
−∆
−∆
−
+
∆+
∆
−+
+−
kikiy
ki
kikiz
ki
kikiy
ki
kikiz
kiki
kikiykikiz
ykH
zkH
ykH
zkH
Hykzk
(4.4)
which can be identified as the finite-difference approximation to the Helmholtz
equation, identical in form to equation (3.30), where and k are the propagation
coefficients in the y and z directions respectively, written as:
kiyk ,, kiz ,,
kiykiy jk ,,0,, σωµ= , (4.5)
kizkiz jk ,,0,, σωµ= . (4.6)
84
∆zi,k
∆zi,k
Z2,i,k Z4,i,k
i-1,k
i,k-1 Z1,i,k
∆x
i+1,k
Figure 4.1. Schematic
convention
integration
Law (after [
The wave numbers a
divergent solutions. For a s
can write the matrix equatio
0JH S = ,
i,k
Z3,i,k
i,k+1
∆yi,k
diagram in the y-z plane showing the numbering
for the impedance mesh. The dotted line shows the
path in the x-z plane for the application of Ampere�s
4.13]).
re chosen such that to prevent exponentially
olution space consisting of N cells, from equation (4.4), one
n:
0Re ,),,( >kizyk
(4.7)
85
where H is a 1×N matrix of unknown magnetic field elements in the solution space and
is the 1×N matrix of applied current densities (i.e., source terms) which must have
non-zero terms to prevent non-trivial solutions in H. For example, in plane wave
incidence, the impressed magnetic field is generated from a series of applied current
densities in the x-direction. S is a sparse square matrix N
0J
2 in size, and is called the
propagation matrix, since it is expressed in terms of the propagation coefficients and
cell dimensions. The unknown magnetic field values are solved with the matrix
equation:
01JSH −= ,
where can be solved by any number of matrix inversion algorithms. The
formulation presented in this thesis was written in Matlab, and solves for the matrix
inversion using a standard LU decomposition method. Dirichlet boundary conditions
are used to define the magnetic field as the incident homogeneous plane wave.
Neumann boundary conditions are employed to terminate the boundaries in the other
directions. From equation (4.4), it is possible to observe that the diagonal elements of S
have the form:
1−S
( ) ( ) 1222
,,,2
,,,
+∆
+∆ kikixkikiz ykzk
, (4.8)
and the non-zero off-diagonal elements have the forms of:
( )2,,
1yk kiy ∆
− , (4.9)
( )2,,
1zk kiz ∆
− . (4.10)
It is clear then, from equations (4.8) to (4.10), that the self-consistent impedance
method can accommodate fundamental uniaxial and biaxial anisotropic media, where
. The current through is calculated from equation (4.2) so the
horizontal electric field can be calculated using the field form of Ohm�s Law:
kizkiykix kkk ,,,,,, ≠≠ kiZ ,,1
kiE ,,1
86
kikiy
kikiki z
HHE
,,,
1,,,,1 ∆
−= −
σ. (4.11)
The surface impedance Z at the earth-air interface, defined at the top of the
(i,k)
yx
th cell, can now be written as:
kikikiy
kiki
x
yyx zH
HHHE
Z,1,,,
1,,
∆−
==−
−
σ. (4.12)
It should be noted that the term in equation (4.12) is measured above the
surface of the conducting half space and the term is measured at the surface of the
conducting half space. It is observed that having at least two rows of air cells above the
surface of the conducting half space provides accurate surface impedance results. Since
the medium above the half space is perfectly insulating, is approximately uniform
over distances less than one-tenth of the wavelength in free-space above the surface. If
one seeks to introduce a semi-implicit approximation for the magnetic field, similar to
the finite-difference time-domain method [4.12]:
1, −kiH
kiE ,,1
xH
21,,
2/1,−
−
+= kiki
ki
HHH , (4.13)
then it is easily observed that equation (4.12) reduces to:
)()(2
1,,,,,
1,,
+
−
+∆−
==kikikikiy
kiki
x
yyx HHz
HHHE
Zσ
.
In the above form, the self-consistent impedance method can solve for
fundamental biaxial anisotropy problems, and those inclined uniaxial anisotropy
problems that reduce to fundamental biaxial anisotropy problems. For the inclined
anisotropy problem, the substitution of equations (2.59) and an analogous vertical
conductivity value obtained from the substitution of instead of from equation
(2.45), into equations (4.5) and (4.6) reduces the problem to a fundamental one.
zE yE
87
Other authors who have considered two-dimensional numerical modelling of
inclined anisotropic media, such as Shize & Shengkai [4.11], have considered
approximate solutions to equation (2.19), the rotated anisotropic Helmholtz equation.
If we were to consider the simulation of horizontally inhomogeneous media with
inclined anisotropy, then according to other authors, equation (4.4) must be replaced by
the finite-difference approximation to equation (2.19):
( ) ( )
( ) [ ]
( ) [ ]
[ ] .0112
cossin
cossin1
sincos1
1cossin2sincos2
1,11,11,11,12,,
2,,,,
,,
1,1,2,,
,2
2,,
,2
2,
,1,12,,
,2
2,,
,2
2,
,2,,
,2
2,,
,2
22,,
,2
2,,
,2
2,
=−−+
−
∆∆−
−
+
∆−
−
+
∆−
+
+
∆+
+
∆
−−+++−−+
−+
−+
kikikikikitkinkiki
kiki
kikikit
ki
kin
ki
ki
kikikit
ki
kin
ki
ki
kikit
ki
kin
ki
kit
ki
kin
ki
ki
HHHHxz
HHz
HHx
Hzx
γγαα
γα
γα
γα
γα
γα
γα
γα
γα
(4.14)
Specifically, it should be noted that in equation (4.14), the so-called shearing
term:
[ ]1,11,11,11,12,,
2,,,,
,, 112
cossin−−+++−−+ −−+
−
∆∆ kikikikikitkinkiki
kiki HHHHyz γγαα
, (4.15)
vanishes unequivocally (see Chapter 3) and equation (4.14) reduces to equation (4.4)
identically. Hence, there is no need to consider the partial term of (4.15) in any two-
dimensional finite-difference formulation for TM-type waves. It should be noted that in
this case, equation (4.14) defines the magnetic fields in both fundamental biaxial
anisotropic media, as well as inclined uniaxial anisotropic media.
The basic field approximation, as given by equation (4.4), is formally identical
to that resulting from a finite-difference approximation with a cell-centered TM-type
magnetic field. This approach is not common in two-dimensional finite-difference
modelling practice, but standard three-dimensional staggered-grid modelling
approaches [4.8] reduce to this case when homogeneity is assumed in one horizontal
direction. Thus, based upon the approximation of equation (4.4), two-dimensional
accuracy tests for staggered grid solutions could be made.
88
4.3. Results: inclined anisotropy in inhomogeneous media
To demonstrate that an inclined uniaxial anisotropic conductor can be simulated
as a fundamental biaxial anisotropic conductor, a number of cases involving inclined
anisotropic conductivity are presented in which the surface impedance of the
conducting half space can be expressed as an analytical function to assess the accuracy
of this new formulation.
4.3.1. Homogeneous layer above a perfect electric conductor
The exact solution for the surface impedance of a TM-type wave normally
incident upon a horizontally layered half space was presented in Chapter 2. For a
homogeneous layer with intrinsic impedance:
122
1sincos
−
+=
nt
jZσ
ασ
αωµ , (4.16)
terminated at z = h by a perfect electric conductor, the surface impedance at z = 0 is
given by the transmission line analogy:
)tanh(sincos1
122
,1 hkjZnt
s
−
+=
σα
σαωµ , (4.17)
where:
+=
nt
jkσ
ασ
αωµ22
1sincos , (4.18)
and to prevent an exponentially divergent solution in . A comparison
between the surface impedance responses obtained from equation (4.17) and with the
self-consistent impedance method is shown in Figure 4.2 for frequencies in the range
, and is presented as a function of the angle of inclination of the
0Re 1 >k
10Hz ≤≤ f
sZ ,1
kHz 10
89
anisotropic conductivity, . The anisotropic layer was assumed to have o900 ≤≤ α
001.01, =tσ S/m and 01.0=1,nσ S/m. The approximate model used 3 air cells, and
earth cells, and the depth between the air-half space interface and the perfect
electric conducting basement was constant at 1 000 m. The cell sizes selected were
m and m.
2×
1003×
20=∆y 10=∆z
.0=
01.01, =nσ
o90≤0 ≤ α
Figure 4.2. Surface impedance magnitude and phase response for an
anisotropic layer of 1 000 m thickness with 0011,tσ S/m and
S/m above a perfectly conducting basement. The
anisotropic conductivity tensor is rotated through the range
. The impedance method solutions for 10 Hz (+), 100 Hz
(x), 1 kHz (o) and 10 kHz (*) are presented. The exact solutions for
10 Hz (dotted), 100 Hz (solid), 1 kHz (dash-dotted) and 10 kHz
(double-dashed) are also shown. Note that at 1 kHz and 10 kHz, the
surface impedance is equal to the intrinsic impedance of the layer
and the phase is equal to 45o.
90
The surface impedance magnitude agrees very well with the corresponding
exact solutions for all frequencies considered. The surface impedance phase however,
is more accurate at lower frequencies than at high frequencies. This difference is
suggested to be a resultant from the cell size chosen. For the lower frequencies, the cell
size was chosen to be several orders of magnitude less than the wavelength in the half
space. At the higher frequencies presented, the cell size was approximately in the same
order of magnitude as the wavelength considered. Such discretisation errors are
introduced in all approximate techniques and are not therefore unique to the self-
consistent impedance method.
Figure 4.3. Surface impedance magnitude and phase response for an
anisotropic layer of 20 m thickness with 01, 15εε =n , 01.01, =nσ S/m,
01, 5εε =t and 001.01, =nσ S/m above a perfect electrically. The
anisotropic complex conductivity tensor is rotated through the
range . The impedance method solutions for 10o900 ≤ ≤α 4 Hz (o),
105 Hz (*) and 106 Hz (∆) are presented. The exact solutions for 104
Hz (solid), 105 Hz (dashed) and 106 Hz (double dashed) are also
shown. Note that at due to the effects of displacement currents at 106
Hz, the surface impedance is equal to the intrinsic impedance of the
layer and the phase is not equal to 45o.
91
To demonstrate the ability of the self-consistent impedance method to include
complex conductivity, the surface impedance of a homogeneous layer with 01, 15εε =n ,
01.01, =nσ S/m, 01, 5εε =t and 001.01, =nσ S/m above a perfect electrically
conducting basement, for f = 104, 105 and 106 Hz are presented in Figure 4.3, as a
function of the angle of inclination, 0 . The thickness of the anisotropic layer
was assumed to be 20 m and a uniform cell size of ∆x = ∆z = 0.5 m was used. The
solution space was 3 cells wide and 42 cells deep, allowing for 2 air cells above the
anisotropic interface for surface reflections. The corresponding exact solutions are also
presented [4.17].
o90≤≤ α
Figure 4.4. Surface impedance magnitude and phase response at 100 Hz for two
anisotropic layers overlying a perfectly conducting basement using
the self-consistent impedance method (o) and corresponding exact
solutions (solid line). The upper layer thickness was varied and the
total depth was kept constant at 1 000 m. The upper layer has σt,1 =
0.01 S/m, σn,1 = 0.001 S/m and α1 = 30o. The lower layer has σt,2 =
0.05 S/m, σn,2 = 0.5 S/m and α2 = 60o.
92
4.3.2. Horizontal homogeneous layers above on a perfect electric conductor
One may extend the transmission line analogy as discussed in the previous
section to any number of horizontal, homogeneous layers. Figure 4.4 presents a
comparison of the exact and approximate results for a two-layered, horizontally
homogeneous, inclined anisotropic earth above a perfectly conducting basement for a
frequency of 10 kHz for a variable upper layer thickness . The second layer thickness
is varied in each case such that h m. The upper layer has
1h
2h 000 121 =+ h 01.01, =tσ
S/m and 001.01, =nσ S/m. The second layer has 05.02, =tσ S/m and 5.0=2,nσ S/m.
The angle of inclination in the upper layer is 30o and in the lower layer is 60o, and is
kept constant in all models. The cell sizes selected were m and ∆ m,
and the total solution space was 3 × 102 cells where 2 air cells were assigned in the
model. There is good agreement between the impedance method and the corresponding
exact solutions, though at one particular point ( = 100 m), the phase does differ by as
much as 16% from the corresponding exactly derived value and the magnitude varies
by 19%. All other points have variations less than these just described, and reduce to nil
variance for the single layer case when = 1 000 m.
10=∆y 20=z
1h
1h
4.3.3. Vertical dyke embedded in a homogeneous layer above a perfect electric
conductor
The exact solution for the surface impedance of a homogeneous plane wave
incident upon a vertical dyke with inclined electrical anisotropy embedded in an
otherwise homogeneous layer, above a perfect electric conducting basement was
presented in Chapter 3. It was demonstrated in that chapter that the inclined electrical
anisotropy problem is equally posed as a fundamental electrical anisotropy problem. In
Figure 4.5, the surface impedance at 10 kHz is presented for a dyke that is 500 m wide
with 01.0=tσ S/m and 001.0=nσ S/m embedded in an otherwise homogeneous layer
with 00.0=t 1σ S/m and 01.0=nσ S/m. The common depth of the dyke and
homogeneous layer to the perfect electrically conducting basement is 200 m. The cell
sizes selected were m and ∆ m, and the total solution space was 100 × 42
cells where 2 air cells were assigned in the model. There is good agreement in the
surface impedance magnitude between the impedance method and the corresponding
exact solution, but the surface impedance phase has some inaccuracies.
5 z=∆y 5=
93
Figure 4.5. Surface impedance magnitude and phase response at a single
boundary at 10 kHz for a 500 m wide vertical dyke with 01.01, =tσ
S/m and 001.01, =nσ S/m embedded in an otherwise homogeneous
layer with 001.01, =tσ S/m and 01.01, =nσ S/m, terminated with a
perfectly conducting basement. The depth of the dyke and layer is
200 m. The impedance method response (dots) is shown above the
exact response (solid line).
4.5. Discussion
In this chapter, the principle of modelling the surface impedance of an
inhomogeneous half space with inclined uniaxial electrical anisotropy as an equivalent
half space with fundamental electrical biaxial anisotropy has been demonstrated
successfully. The self-consistent impedance method has been introduced, and shown to
accurately model the surface impedance response of these two-dimensional induction
problems over a broad range of frequencies. Whilst the impedance method has been
introduced for this modelling concept in this chapter, it is strongly emphasised that the
modelling principles introduced in this chapter can be applied to any two-dimensional
94
approximate method. The surface impedance magnitude can be accurately modelled
using the impedance method, but the surface impedance phase is shown to be
inaccurate for laterally inhomogeneous and anisotropic models. It is importantly noted
here that this chapter presents the first comparison of any approximate technique with
the corresponding exact solution for a laterally inhomogeneous and anisotropic half
space.
4.6. References
[4.1] M. Eisel, V. Haak, J. Pek & V. Cerv, J. Geophys. Res. 106B, 16061 (2001).
[4.2] W. Heise & J. Prous, Geophys. J. Int. 147, 610 (2001).
[4.3] D. A. James, PhD thesis, Griffith University, 1998.
[4.4] Y. Li, Geophys. J. Int. 148, 389 (2002).
[4.5] R. L. Mackie, J. T. Smith & T. R. Madden, Radio Sci. 29, 923 (1994).
[4.6] G. A. Newman & D. L. Alumbaugh, Geophysics 67, 484 (2002).
[4.7] A. M. Osella & P. Martinelli, Geophys. J. Int. 115, 819 (1993).
[4.8] J. Pek & T. Verner, Geophys. J. Int. 128, 505 (1997).
[4.9] I. K. Reddy & D. Rankin, Geophysics 40, 1035 (1975).
[4.10] P. D. Saraf, J. G. Negi & V. Cerv, Phys. Earth Planet. Int. 43, 196 (1986).
[4.11] X. Shizhe & Z. Shengkai, Acta Seismolog. Sinica 7, 80 (1985).
[4.12] A. Taflove, Computational Electrodynamics: The Finite Difference Time
Domain Method (Artech House, Norwood, 1995).
[4.13] D. V. Thiel & R. Mittra, Radio Sci. 36, 31 (2001).
[4.14] T. Wang & S. Fang, Geophysics 66, 1386 (2001).
[4.15] P. Weidelt, in Three-Dimensional Electromagnetics, edited by M. Oristaglio &
B. Spies (Soc. Explor. Geophys., Tulsa, 1999), pp. 119-137.
[4.16] G. A. Wilson & D. V. Thiel, IEEE Trans. Geosci. Remote Sensing, submitted.
[4.17] G. A. Wilson & D. V. Thiel, presented at IEEE Int. Antennas Propagat. Symp.,
San Antonio, TX, Jun. 2002.
[4.18] C. Yang, Xibei Dizhen Xuebao 19, 27 (1997).
[4.19] C. Yin & H. M. Maurer, Geophysics 66, 1405 (2001).
[4.20] M. S. Zhdanov, I. M. Varenstov, J. T. Weaver, N. G. Golubev & V. A. Krylov,
J. App. Geophys. 37, 133 (1997).
95
Chapter 5
Exact solutions for three-dimensional anisotropy problems
5.1. Introduction
As presented in the previous chapters of this thesis, the physical basis for the
surface impedance of a homogeneous plane wave polarised parallel or perpendicular to
the strike of a two-dimensional half space is that the electric and magnetic fields
generated are orthogonal. In any other situation, the electric and magnetic field vectors
are not orthogonal, indicating that either the source field or half space is three-
dimensional and inhomogeneous, or that the source field is not polarised parallel or
perpendicular to the strike of the half space [5.4, 5.5, 5.17]. Below 1 kHz, the source
fields are elliptically polarised [5.16] and this can lead to some difficulties in surface
impedance data reduction and interpretation. Above 1 kHz, the principle source field
for surface impedance measurements is the radiation from lightning discharges or
artificial signals from navigation beacons and radio transmitters [5.22].
These very low frequency (VLF) waves propagate with very little loss in the
earth-ionosphere waveguide as a series of waveguide modes. Cloud-to-ground lightning
discharges and VLF antennae are effectively vertical electric dipoles, and as such only
launch linearly polarised TM-type waves. At imperfect ionospheric and terrestrial
boundaries, the TM-type waves can be reflected as TE-type waves, which propagate
independently and with higher attenuation than the lower order TM-type waves that
dominate VLF propagation over very large distances in the earth-ionosphere waveguide
[5.24]. However, the extent of this mechanism for elliptical polarization has been
calculated to be less than 1% [5.12], and the minor/major axis ratio of the magnetic
field polarization ellipse has been measured to less than 1% [5.14].
The one-dimensional surface impedance of VLF radio waves is commonly
measured using portable surface impedance meters [5.1, 5.18] that measure the
horizontal electric and magnetic field magnitudes, and the phase difference between
them. In the interpretation of the VLF surface impedance, the earth is usually assumed
to be isotropic. However, some authors have presented methods of interpreting one-
dimensional VLF surface impedance data above a laterally anisotropic half space [5.11,
5.18, 5.19]. However, in these formulations, it has been assumed that the horizontal
magnetic field component maintains its linear polarization. This is not exactly true in
96
the presence of an arbitrary anisotropic half space. Further, no analyses have ever been
presented for the surface impedance of an inclined anisotropic half space at arbitrary
skew angles.
In this chapter, the method of auxiliary potentials, developed by Chetaev &
Belen�kaya [5.2, 5.9, 5.10], will be applied to solve for the fields of a linearly polarised,
homogeneous plane wave incident at a skew angle of incidence upon an inclined
anisotropic half space. In the anisotropic half space, all field components will be
coupled and hence a TM-type wave incident upon the half space will generate a
reflected TE-type wave in addition to the reflected TM-type wave. This will influence
the measured electric and magnetic field components in one-dimensional surface
impedance measurements. This work is a logical extension of Chapter 2, where the
source field was assumed to be parallel and perpendicular to the strike of the
anisotropic half space. Expressions for all elements of the impedance are derived, and
their applications to VLF surface impedance measurements are further discussed.
5.2. General solutions for the electromagnetic fields in an anisotropic medium
By developing Chetaev�s method of auxiliary potentials [5.7], we will consider
the surface impedance of homogeneous, monochromatic plane waves characterised by
, incident at a skew angle upon a homogeneous half space with inclined
uniaxial anisotropic conductivity (Figure 5.1). Solutions will be presented for both TE-
and TM-type incident plane waves. This class of problem is a natural extension to the
class of models presented in Chapter 2 of this thesis.
)exp( kz−
Given the Maxwell equations (1.1) to (1.4) for monochromatic fields in the
inclined co-ordinate system of a uniaxial anisotropic medium, characterised by a
uniaxial tensor of complex conductivity given by equation (1.54), where no free
charges or extraneous currents exist, we introduce the potentials:
'' AB ×∇= , (1.31)
Φ−∇−= '' AE ωj . (1.33)
Substituting equations (1.31) and (1.33) into the Maxwell equations (1.2) and
(1.3) results in:
97
x x� θ z = 0 y α α z� y� z
Figure 5.1. Geometry for the homogeneous plane wave incident at a skew angle
θ to the anisotropic half space inclined at angle α about the x-axis.
0�'�)'(' =Φ∇−−⋅∇−∇∆ σµσωµ AAA j , (5.1)
and substituting equations (1.31) and (1.33) into the Maxwell equation (1.1) and (1.4)
results in:
0'�� =⋅∇+Φ∇⋅∇ Aσωσ j , (5.2)
where σ�⋅∇ can not be abbreviated when σ� is anisotropic [5.23]. We introduce the
optimal Lorentz gauge condition [5.6]:
'1 A⋅∇−=Φtµσ
, (5.3)
and substitute equation (5.3) into equation (5.1) to obtain:
0)'( 1�
'�' =⋅∇∇
−+−∆ AAA
t
jσσσωµ . (5.4)
98
for the vector potential A in {x� ,y�, z�} co-ordinates. For and , which are
associated with
'xA 'yA
tσ , equation (5.4) reduces to the homogeneous Helmholtz equations:
0'' =−∆ xtx AjA ωµσ , (5.5)
0'' =−∆ yty AjA ωµσ . (5.6)
From the optimal Lorentz gauge condition, the full condition on the vector
potential is stated as:
Φ=∂∂
−'
1 '
zAz
tµσ, (5.7a)
0'''' =
∂∂
+∂∂
yA
xA yx . (5.7b)
For , which is associated with 'zA nσ , equation (5.4) reduces to:
0'
)1( 2'
22
'' =∂
∂−Λ+−∆
zAAjA z
znz ωµσ , (5.8)
where λσ
σ 1==Λt
n , the reciprocal of the coefficient of anisotropy. When we rotate
the co-ordinates from the fundamental co-ordinate system to the inclined co-ordinate
system (see Appendix 1), equations (5.5) and (5.6) are invariant, whilst we introduce
the functional rotations:
θα sinsin' z
fxf
∂∂−=
∂∂ ,
θα 222
2
2
2
sinsin' z
fxf
∂∂−=
∂∂ ,
99
θα cossin' z
fyf
∂∂−=
∂∂ ,
θα 222
2
2
2
cossin' z
fyf
∂∂−=
∂∂ ,
αcos' z
fzf
∂∂−=
∂∂ ,
α22
2
2
2
cos' z
fzf
∂∂−=
∂∂ .
Equation (5.7) will take the form of:
0sincossinsin '' =∂
∂+
∂∂
− αθαθz
Az
A yx , (5.9)
and equation (5.8) takes the form of:
0cos)1( 22
'2
2'' =
∂∂
−Λ+−∆ αωµσzAAjA z
znz , (5.10)
since all partial derivatives of the homogeneous plane wave field with respect to x and y
are equal to zero. When one considers equations (5.5) and (5.6) with the spatial
variance described by , then and are observed to satisfy the wave
number:
)exp( 1zk− 'xA 'yA
tjk ωµσ=21 , (5.11)
which is the wave number for an ordinary wave, provided Re to prevent
exponentially divergent solutions in and . As a result of the spatial variances,
from equation (5.9), the relation between the vector potential components of the
ordinary wave:
01 >k
'xA 'yA
100
θtan'' xy AA = , (5.12)
is obtained. From co-ordinate rotations (see Appendix 1), one can write for the ordinary
wave:
αθθ cossincos '' yxx AAA += , (5.13)
αθθ coscossin '' yxy AAA +−= , (5.14)
αsin'yz AA = . (5.15)
By substituting equation (5.12) into equations (5.13) to (5.15), the vector
potential components in fundamental co-ordinates can be written as:
+=θ
αθθcos
cossincos 22
'xx AA , (5.16)
)1(cossin' −= αθxy AA , (5.17)
αθ sintan'xz AA = . (5.18)
As a consequence of equation (5.7), it is observed for the ordinary wave that
, which implies: 0=Φ
AE ωj−= . (5.19)
As a result, the components of the electric fields of the ordinary wave in
fundamental co-ordinates can be written as:
+−=θ
αθθωcos
cossincos 22
'xx AjE , (5.20)
101
)1(cossin' −−= αθω xy AjE , (5.21)
αθω sintan'xz AjE −= . (5.22)
From equation (1.31), the components of the magnetic flux density of the
ordinary wave in fundamental co-ordinates can be written as:
)1(cossin'1 −= αθxx AjkB , (5.23)
+−=θ
αθθcos
cossincos 22
'1 xy AjkB , (5.24)
0=zB . (5.25)
After considering equation (5.10) with the spatial variance described by
, then it is observed that satisfies the wave number: )exp( 2 zk− 'zA
αλωµσ
222
2 sin)1(1 −+= tj
k , (5.26)
which is the wave number for an extraordinary wave, provided to prevent
exponentially divergent solutions in . For the extraordinary wave, it is observed that
the scalar potential is not zero but is given by:
0Re 2 >k
'zA
αµσ
cos'2
zt
Ajk=Φ . (5.27)
From equation (5.27), it follows that:
k� cos'
22
=Φ∇ α
µσ zt
Ak
. (5.28)
102
From co-ordinate rotations (see Appendix 1), one can write for the
extraordinary wave:
αθ sinsin'zx AA −= , (5.29)
αθ sincos'zy AA −= , (5.30)
αcos'zz AA = . (5.31)
The components of the electric field for the extraordinary wave are obtained by
substituting equations (5.29) to (5.31) into equation (1.33):
αθω sinsin'zx AjE = , (5.32)
αθω sincos'zy AjE = , (5.33)
αµσ
cos'
22
21
zt
z Akk
E
−= . (5.34)
From equation (1.31), the components of the magnetic flux density of the
ordinary wave in fundamental co-ordinates can be written as:
αθ sincos'2
zx AjkB −= , (5.35)
αθ sinsin'2
zy AjkB = , (5.36)
0=zB . (5.37)
The observed fields will be the sum of the ordinary and extraordinary fields:
aryextraordinordinary EEE += , (5.38)
103
aryextraordinordinary HHH += . (5.39)
From equations (5.20) to (5.25) and (5.32) to (5.37), we can write the
components of the total electric and magnetic fields in fundamental co-ordinates in the
inclined anisotropic half space as:
αθωθ
αθθω sinsincos
cossincos'
22
' zxx AjAjE +
+−= , (5.40)
αθωαθω sincos)1(cossin '' zxy AjAjE +−−= , (5.41)
αµσ
αθω cossintan '
22
21
' zt
xz Akk
AjE
−+−= , (5.42)
αθαθ sincos)1(cossin '2'1 zxx AjkAjkB −−= , (5.43)
αθθ
αθθ sinsincos
cossincos'2
22
'1 zxy AjkAjkB +
+−= , (5.44)
0=zB . (5.45)
Equation (5.45) is expected, following from equation (2.41). Also, equations
(5.40) to (5.32) are observed to satisfy equation (2.42):
0=++ zzzyzyxzx EEE σσσ , (2.46)
provided the elements of the conductivity tensor are given by:
ααθσσσ sincossin)( ntzx −= , (5.47)
ααθσσσ sincoscos)( ntzy −= , (5.48)
104
ασασσ 22 cossin ntzz −= , (5.49)
which states that the vertical current density of a homogeneous plane wave is equal to
zero.
5.3. General solutions for the electromagnetic fields in air
In the upper half space, the general solution for waves characterised by equation
(5.4) are the superposition of the TE- and TM-type waves, which propagate
independently but satisfy the same wave number:
εµω220 =k . (5.50)
where . From the Maxwell equations (1.1) and (1.2), the components of the
homogeneous TM-type waves will be:
ℜ∈ 0k
zE
tB xy
∂∂
−=∂
∂, (5.51)
zB
tE yx
∂∂
−=∂
∂µ
ε 10 . (5.52)
We will choose the general solutions for the magnetic field component of the
homogeneous TM-type waves to take the form:
)exp()exp( 00 zjkNzjkMBy +−= , (5.53)
so the corresponding electric field component can be written as:
)}exp()exp({ 000 zjkNzjkM
kEx −−=
ωµε, (5.54)
105
where M and N are (complex) constants independent of {x, y, z, t}, representing the
coefficients for the down- and up-going homogeneous plane waves. From the Maxwell
equations (1.1) and (1.2), the components of the homogeneous TE-type wave will be:
zB
tE xy
∂∂−=
∂∂
µε 1
0 , (5.55)
zE
tB yx
∂∂
−=∂
∂. (5.56)
We will choose the general solutions for the electric field component of the
homogeneous TE-type waves to take the form:
)exp()exp( 00 zjkQzjkPEy +−= , (5.57)
so the corresponding magnetic field component can be written as:
)}exp()exp({ 000 zjkQzjkP
kBx −−=
ω, (5.58)
where P and Q are (complex) constants independent of {x, y, z, t}, representing the
coefficients for the down- and up-going homogeneous plane waves.
5.4. The problem of a TE-type incident field
Let us consider a TE-type wave incident upon the anisotropic half space at a
skew angle of incidence. Since the ordinary and extraordinary waves in the anisotropic
half space couple all field components, then a TM-type wave must also be reflected, in
addition to the reflected TE-type wave. The problem is to determine the constants N, Q,
and in terms of the amplitude of the incident TE-type wave, P. The continuity
of is ensured by the continuity of across the interface. We now have four
continuity conditions for the horizontal field components. First, we must express in
terms of from continuity of the horizontal components of the TM-type wave that is
generated in the lower half space and reflected into the upper half space. So, we
'xA 'zA
'xA
zB yE
'zA
106
consider the boundary and set in equation (5.53), so only an up-going
TM-type wave in the upper half space is considered. The fields of the reflected TM-
type wave are then equated with the corresponding field components in the anisotropic
half space. From equations (5.53) and (5.54), one can write:
0=z 0=M
2 sinsinθ
αθ
−
++
111
kcosαθ
−
1
21 k
kk
NBzy ==0
, (5.59)
Nk
Ezx µωε0
00
−==
. (5.60)
We now equate equations (5.59) and (5.60) with equations (5.44) and (5.40)
respectively to obtain the relation:
'21
2' sincos
cossin11
zx AkkA
αθθ
+++
= . (5.61)
From equations (5.43) and (5.61), one can write:
'xTMx AFB = , (5.62)
where
+=2
12
22
cossincos
kk
kjFTM θθ . (5.63)
Similarly, one can write:
'xTMx AGB = , (5.64)
where
++
= θθα sin11
cossin 2kjGTM . (5.65)
107
Now, all field components for the electromagnetic field at the surface of the half
space can be expressed in terms of : xB
xTMTM
x BG
jF
jE
+
+−= αθωθ
αθθω sinsin1cos
cossincos1 22
, (5.66)
xTMTM
y BG
jF
jE
+−−= αθωαθω sincos1)1(cossin1 , (5.67)
xTMtTM
z BG
kkF
jE
−+−= α
µσαθω cos1sintan1 2
22
1' , (5.68)
xTMTM
y BG
jkF
jkB
+
+−= αθθ
αθθ sinsin1cos
cossincos12
22
1 , (5.69)
0=zB . (5.70)
It is noticed that when =θ 0o or 180o, then:
txy
jZσωµ−= , (5.71)
}sin)1(1{ 22 αλσωµ −+−=
tyx
jZ . (5.72)
When =θ 90o or 270o, then:
}sin)1(1{ 22 αλσωµ −+=
txy
jZ . (5.73)
0=yxZ . (5.74)
108
Equations (5.71) to (5.74) agree with the expected values of the surface
impedance when the source fields are parallel and perpendicular to the strike of the
anisotropic half space.
5.4. The problem of a TM-type incident field
Let us now consider a TM-type wave incident upon the anisotropic half space at
a skew angle of incidence. Since the ordinary and extra-ordinary waves in the
anisotropic half space couple all field components, then a TE-type wave must also be
reflected, in addition to the reflected TM-type wave. The problem is to determine the
constants P, N, and in terms of the amplitude of the incident TM-type wave, M.
The continuity of is ensured by the continuity of across the interface. We now
have four continuity conditions for the horizontal field components. First, we must
express in terms of from continuity of the horizontal components of the TE-
type wave that is generated and reflected. So, we consider the boundary and set
in equation (5.57) so only an up-going TE-type wave in air is considered. We
equate the fields of the reflected TE-type wave with the corresponding field
components in the anisotropic half space. From equations (5.57) and (5.58), we write:
'xA
B
'zA
'xA
z yE
'zA
0=z
0=P
QEzy ==0
, (5.75)
Qk
Bzx ω
00
−==
. (5.76)
We now equate equations (5.75) and (5.76) with equations (5.41) and (5.43)
respectively to obtain the relation:
'10
20' )1(cossin)(
sincos)(zx A
kkkk
A−−
−=
αθαθ
. (5.77)
From equations (5.44) and (5.77), one can write:
'xTEy AFB = , (5.78)
109
where
ααθ
µσαθ
tan)()1(costan)(
)1(cossin20
102
22
11 kk
kkkkjjkF
tTE −
−−
−+−= . (5.79)
Similarly, one can write:
'zTEy AGB = , (5.80)
where
αµσ
αθ cossincos)(2
22
1201
−+−=
tTE
kkjkkjkG . (5.81)
Now, all field components for the electromagnetic field at the surface of the half
space can be expressed in terms of : xB
xTETE
x BG
jF
jE
+
+−= αθωθ
αθθω sinsin1cos
cossincos1 22
, (5.82)
xTETE
y BG
jF
jE
+−−= αθωαθω sincos1)1(cossin1 , (5.83)
xTEtTE
z BG
kkF
jE
−+−= α
µσαθω cos1sintan1 2
22
1' , (5.84)
xTETE
y BG
jkF
jkB
+
+−= αθθ
αθθ sinsin1cos
cossincos12
22
1 , (5.85)
0=zB . (5.86)
It is noticed that when =θ 0o or 180o, then:
110
}sin)1(1{ 22 αλσωµ −+=
txy
jZ , (5.87)
tyx
jZσωµ= . (5.88)
When =θ 90o or 270o, then:
0=xyZ , (5.89)
}sin)1(1{ 22 αλσωµ −+=
tyx
jZ . (5.90)
Equations (5.87) to (5.90) agree with the expected values of the surface
impedance when the source fields are parallel and perpendicular to the strike of the
anisotropic half space.
5.6. Discussion
From equations (5.66) to (5.70), and (5.82) to (5.86), the elements of the
impedance tensor can be calculated:
x
xxx B
EZ
µ= , (5.91)
y
xxy B
EZ
µ= , (5.92)
x
yyx B
EZ
µ= , (5.93)
y
yyy B
EZ
µ= . (5.94)
111
As we are principally concerned with TM-type propagation problems related to
VLF propagation and surface impedance measurements, polar diagrams of one of the
principal components and additional components of the surface impedance tensor for a
10 kHz homogeneous plane wave at all angles of skew incidence upon a homogeneous
half space with 001.0=tσ S/m, 01.0=nσ S/m, and =α 60o are presented in Figures
5.2 to 5.5. It is observed that the values converge to the expected impedances for
=θ 0o, 90o, 180o and 270o. It should be observed that Figures 5.2 to 5.5 correlate to the
additional impedance polar diagrams expected from a two-dimensional structure [5.3].
The principal impedance polar diagram exhibits the phase of a one-dimensional half
space, but have the magnitude characteristics of a two-dimensional half space [5.3].
This reinforces the well-known principle of the existence of ambiguity in surface
impedance measurements, where it is difficult to differentiate between the surface
impedance response of a one-dimensional homogeneous and anisotropic half space, and
a two-dimensional homogeneous (and anisotropic) half space.
We will now examine the effect of an anisotropic half space on the measured
electric and magnetic fields in an ideal one-dimensional surface impedance
measurement. We introduce ψ as the angle between the orientation of the surface
impedance meter and the direction of propagation, i.e., the angle of the instrument with
respect to θ (see Figure 5.6).
112
Figure 5.2. xyZ of a 10 kHz homogeneous TM-type plane wave as a function of
θ for a half space with 001.0=tσ S/m, 01.0=nσ S/m, and =α 45o.
113
Figure 5.3. of a 10 kHz homogeneous TM-type plane wave as a function
of
xyZ∠
θ for a half space with 001.0=tσ S/m, 01.0=nσ S/m, and
=α 45o.
114
Figure 5.4. yyZ of a 10 kHz homogeneous TM-type plane wave as a function of
θ for a half space with 001.0=tσ S/m, 01.0=nσ S/m, and =α 45o.
115
Figure 5.5. of a 10 kHz homogeneous TM-type plane wave as a function
of
yyZ∠
θ for a half space with 001.0=tσ S/m, 01.0=nσ S/m, and
=α 45o.
The fields induced in the horizontal electric and magnetic dipoles, short
insulated wires and small loops respectively [5.18], will be proportional to the electric
and magnetic field components of the total fields in the direction of the dipole axis.
Ignoring the influences of the radiation patterns of the electric and magnetic dipoles, for
any ψ , components of the observed electric and magnetic fields will be:
ψψ sincos yxobs EEE += , (5.95)
ψψ cossin yxobs HHH +−= . (5.96)
If the homogeneous half space was isotropic, then equations (5.95) and (5.96)
would simply reduce to:
116
x x� ψ y θ y�
Figure 5.6. Angle of orientation of the surface impedance meter ψ with respect
to the skew angle of incidence θ and strike of the anisotropic half
space in the xy plane.
ψsinxobs EE = , (5.97)
ψcosyobs HH = , (5.98)
and the observed surface impedance at any sZ ψ would be given by:
ψtany
xs H
EZ = , (5.99)
which exhibits asymptotes when 90=θ or 180o, as expected for linearly polarised
fields. We will now consider a 10 kHz homogeneous TM-type plane wave incident
upon a homogeneous half space with 001.0=tσ S/m, 01.0=nσ S/m, and =α 60o,
where θ = 45o, 90o and 135o. The fields can be calculated using the expressions derived
in Section 5.4. From equations (5.97) and (5.98), the one-dimensional surface
impedance can be calculated, and the corresponding polar diagrams for the normalized
117
magnitude of the magnetic and electric fields are shown in Figures 5.7 and 5.8
respectively.
Figure 5.7. Normalized observed (measured) magnetic field of a 10 kHz
homogeneous TM-type plane wave as a function of ψ for a half
space with 001.0=tσ S/m, 01.0=nσ S/m, =α 45o and θ =60o.
It is observed in Figure 5.7 that at VLF frequencies, the polarization of the
horizontal magnetic field of a homogeneous TM-type plane wave is independent of the
anisotropy of the half space. This verifies the assumptions of previous authors [5.11,
5.12, 5.15, 5.19] who assumed that the magnetic field maintained linear polarization
above an anisotropic half space.
118
Figure 5.8. Normalized observed (measured) electric field of a 10 kHz
homogeneous TM-type plane wave as a function of ψ for a half
space with 001.0=tσ S/m, 01.0=nσ S/m, =α 45o and θ =60o.
In Figure 5.7, it is observed that at VLF frequencies, the polarization of the
horizontal electric fields of a homogeneous TM-type plane wave are elliptically
polarized and are dependent upon the anisotropy of the half space. For the model
presented here, the major axis of the electric field ellipse is observed to rotated
approximately 12o about the horizontal plane with respect to the magnetic field,
maximizing for values of θ = 45o and 135o. It is suggested that by measuring the polar
radiation fields of the electric field and magnetic fields in one-dimensional surface
impedance measurements, information about the presence of anisotropy in the earth
may be ascertained.
To practically demonstrate this effect of anisotropy in a typical VLF survey, the
surface impedance of a VLF radio wave was measured at 10o instrument orientations
with respect to the direction of propagation of the VLF fields [5.13]. Measuring the
19.8 kHz fields of the North West Cape (Western Australia) VLF transmitter, the
surveys were conducted at Callide Mine, Biloela, Queensland, Australia in July 1997 at
different times of day (5:15 am, 6:15 am, 7:15 am, 8:15 am, 9:15 am, 4:25 pm). The
119
magnitudes of the electric and magnetic fields, and the phase difference between them
were measured using a meter similar to [5.18]. The magnetic field was measured using
a ferrite cored multi-turn loop antenna. The electric field was measured using an
electrically short, insulated dipole antenna, which has been demonstrated to be an
effective antenna for measuring the horizontal electric field [5.20]. Figures 5.9 and 5.10
respectively present the normalized magnetic and electric fields from these surveys.
Figure 5.9. Normalized observed (measured) magnetic field of a 19.8 kHz VLF
wave as a function of ψ measured at different times of day at the
Callide Mine, Trap Gully B Area in July 1997 from [5.13].
120
Figure 5.10. Normalized observed (measured) electric field of a 19.8 kHz VLF
wave as a function of ψ measured at different times of day at the
Callide Mine, Trap Gully B Area in July 1997 from [5.13].
It is noted that during the measurements between 6:15 am and 8:15 am relate to
the sunrise period along the 3.5 Mm propagation path between the transmitter and
receiver sites, explaining degradation in the magnetic and electric field polarizations at
those times. This is a result of the destructive interference that occurs between TM01
and TM02 modes during periods of sunrise and sunset [5.21]. However is observed in
Figure 5.9 that the magnetic fields demonstrate a linear polarization for stable
waveguide propagation conditions (5:15 am, 9:15 am, 4:25 pm). Comparing Figure
5.10 to Figure 5.9, it is observed that the horizontal electric field is elliptically
polarised, with the major axis of the electric field polar pattern rotated approximately
10o from the magnetic field polar pattern. As discussed earlier in the model study, this
suggests the presence of anisotropy in the local earth. However, from the surface
impedance data available, it is not possible to determine whether the anisotropy is due
to inclined or lateral anisotropy in the local earth.
121
5.7. Conclusions
In this chapter, the general expressions for the fields of TE- and TM-type
homogeneous plane waves at a skew angle of incidence upon an inclined anisotropic
half space have been derived. Previous analyses have only considered fields of
homogeneous plane waves in the problems of a laterally anisotropic half space (i.e.,
=α 90o), and have not considered the problem of an inclined anisotropic half space.
Further, previous analyses have not considered the effect of mode conversions at the
air-half space boundary, and in the case of TM-type wave, have assumed linear
polarization of the magnetic field is maintained. The results presented in this chapter
have shown that the assumption that the linear polarization of the magnetic field is
maintained, and further the solutions obtained have been shown to converge on the
expected values for the special cases presented in Chapter 2 of this thesis and can be
considered as general solutions for the homogeneous plane wave incident upon an
inclined anisotropic half space.
5.8. References
[5.1] S. A. Arcone & A. J. Delaney, Radio Sci. 15, 1 (1980).
[5.2] B. N. Belen�kaya, Phys. Solid Earth 8, 252 (1972).
[5.3] M. N. Berdichevsky & V. I. Dmitriev, Magnetotellurics in the Context of the
Theory of Ill-Posed Problems (Soc. Explor. Geophys., Tulsa, 2002).
[5.4] D. E. Boerner, R. D. Kurtz & A. G. Jones, Geophysics 58, 924 (1993).
[5.5] T. Cantwell, PhD thesis, Massachusetts Institute of Technology, 1960.
[5.6] D. N. Chetaev, Phys. Solid Earth 2, 233 (1966).
[5.7] D. N. Chetaev, Phys. Solid Earth 2, 651 (1966).
[5.8] D. N. Chetaev, Sov. Phys. Dokl. 12, 555 (1967).
[5.9] D. N. Chetaev & B. N. Belen�kaya, Phys. Solid Earth 7, 212 (1971).
[5.10] D. N. Chetaev & B. N. Belen�kaya, Phys. Solid Earth 8, 535 (1972).
[5.11] G. Fischer, B. V. Le Quang & I. Muller, Geophys. Prosp. 31, 977 (1983).
[5.12] J. Galejs, Radio Sci. 4, 1047 (1967).
[5.13] W. J. F. Nichols, MAppSc thesis, Central Queensland University, 2001.
[5.14] G. S. Parks, G. H. Price, A. L. Whiston & H. W. Parker, Measurement of VLF
Wavefront Components over Long Paths (Stanford Res. Inst., Rep. No. 3590,
1964).
122
[5.15] I. B. Ramaprasada Rao, R. R. Mathur & N. S. Patangay, in Deep
Electromagnetic Exploration, edited by K. K. Roy, S. K. Verman & K. Mallick
(Narosoa Publ. House, New Delhi, 1998), pp. 607-614.
[5.16] D. Rankin & I. K. Reddy, Pure App. Geophys. 78, 58 (1970).
[5.17] I. I. Rokityanski, Bull. Acad. Sci. USSR Geophys. Ser. 10, 1050 (1961).
[5.18] D. V. Thiel, Geoexpl. 17, 285 (1979).
[5.19] D. V. Thiel, Explor. Geophys. 15, 43 (1984).
[5.20] D. V. Thiel, IEEE Trans. Antennas Propagat. 48, 1517 (2000).
[5.21] D. V. Thiel & I. J. Chant, Geophysics 47, 60 (1982); 48, 1697 (1983); 49, 1389
(1984).
[5.22] D. V. Thiel, M. J. Wilson & C. J. Webb, Geoexpl. 25, 163 (1988).
[5.23] A. N. Tikhonov, Sov. Phys. Dokl. 4, 566-570 (1959).
[5.24] J. R. Wait, Electromagnetic Waves in Stratified Media, 2nd ed. (Permagon Press,
Oxford, 1970).
123
Chapter 6
Conclusions
6.1. Contributions
This thesis has investigated the electrodynamics of plane wave propagation in
electrically anisotropic and inhomogeneous media for applications in the surface
impedance methods of electromagnetic geophysics. The thesis has been written in a
theoretical context, emphasising physical principles over technical details or model
studies. No geological aspects of interpreting electrical anisotropy from experimental
data have been presented.
For a homogeneous TM-type wave propagating in a half space (where no
extraneous currents or charges exist) with both vertical and horizontal inhomogeneities,
where the TM-type wave is aligned with one of the elements of the conductivity tensor,
it has been shown using exact solutions (Chapters 2 and 3) that the shearing term in the
homogeneous anisotropic Helmholtz equation:
.0cossin 11 2
cossinsincos
2
2
222
2
222
=−∂∂
∂
−+
∂∂
++
∂∂
+
xx
tn
x
tn
x
tn
Hjzy
Hσ
zH
yH
ωµαασ
σα
σα
σα
σα
(6.1)
unequivocally vanishes and one need only seek solutions to the homogeneous
Helmholtz equation:
.0 cossinsincos2
222
2
222
=−∂
∂
++
∂∂
+ x
x
tn
x
tn
HjzH
yH ωµ
σα
σα
σα
σα (6.2)
This implies that those problems posed with an inclined uniaxial conductivity
tensor can be identically stated with a fundamental biaxial conductivity tensor [6.13,
6.15], provided that the conductivity values are the reciprocal of the diagonal terms
from the Euler rotated resistivity tensor:
124
xxxxx σρσ == −1 , (6.3)
zz
zyyzyyyy
nty σ
σσσρ
σα
σασ −==
+= −
−1
122 sincos , (6.4)
yy
yzzyzzzz
ntz σ
σσσρ
σα
σασ −==
+= −
−1
122 cossin . (6.5)
Exact solutions have been derived for a family of one- and two-dimensional
problems involving a homogeneous, linearly polarised TM-type field [6.11, 6.13, 6.15].
It has been demonstrated that problems with vertical inhomogeneity (i.e., horizontally
stratified), where the each layer exhibits inclined uniaxial anisotropy, can be solved
with a transmission line analogy provided the conductivity in each expression is given
by equation (6.4). The surface impedance response of a vertical dyke in an otherwise
homogeneous layer has also been derived using Fourier series expansion, where both
the dyke and the layer exhibit inclined uniaxial anisotropy.
These exact solutions have provided the physical intuition to identify the
inclined uniaxial anisotropy equivalence to fundamental biaxial anisotropy present in
two-dimensional TM-type propagation problems. This implies those approximate
methods of solving arbitrary two-dimensional problems for a homogeneous TM-type
wave need only to solve the homogeneous Helmholtz equation and can neglect the
corresponding shearing term. It is identified that this inherent non-uniqueness will pose
a significant challenge for any inversion routine, and the requirement for significantly
more detailed geological a priori information to be included in the different
regularisation methods will be essential.
The self-consistent impedance method, a two-dimensional finite-difference
approximation based on a network analogy, has been demonstrated to accurately solve
for such TM-type propagation problems [6.9, 6.12, 6.14]. Whilst other approximate
techniques more accurate than the self-consistent impedance method exist, the context
of its introduction has been to demonstrate the principle of modelling two-dimensional
inhomogeneous media with inclined uniaxial anisotropy.
The problem of a homogeneous plane wave at skew incidence upon an inclined
anisotropic half space has also been treated in this thesis. When a wave is incident upon
a half space with inclined uniaxial anisotropy, both TM- and TE-type waves are
125
coupled and the linearly polarised incident TM- and TE-type waves reflect both TE-
and TM-type (i.e., elliptically polarised) components. Expressions for all elements of
the impedance tensor have been derived for both TM- and TE-type incidence as
homogeneous waves. This simple model offers potential as a method of predicting the
direction of anisotropic strike and dip from a circular survey of one-dimensional
surface impedance measurements, making it particularly applicable to VLF surveys in
sedimentary environments.
6.2. Future research
This thesis has provided a contribution to the fundamental knowledge of the
electrodynamics of plane wave propagation in electrically anisotropic and
inhomogeneous media. However, there is significant scope for continued research in
this interesting discipline. We conclude this thesis by reviewing some of the possible
topics for future research.
6.2.1. Geophysical applications
In the realistic application of geophysical methods to mineral exploration, the
effects of electrical anisotropy and inhomogeneity must be considered. Studies into
these phenomena will be essential for the future interpretation of resistivity, induced
polarization, surface impedance, borehole induction, and transient and airborne
electromagnetic surveys. To this end, there are three main directions of possible future
research.
To able to fully understand the extent and range of electrical anisotropy in
different geological environments, particularly Australian, the first direction must be
targeted towards acquiring high quality petrophysical data from different rocks and
environments than is currently available. It is well known that all sedimentary rocks
exhibit a form of uniaxial anisotropy (though some biaxial anisotropy has been
observed) [6.5], and most minerals, at the crystalline scale, exhibit uniaxial, if not
biaxial anisotropy. Without this basic appreciation of the existence of electrical
anisotropy, future research and applications of more advanced interpretation methods
may end up to be futile.
The second direction of future research is the continued development of
efficient three-dimensional approximate forward modelling techniques that can
126
simulate the electromagnetic response of complex objects with arbitrary shape and
arbitrary anisotropy for any electromagnetic survey method. This is already an active
research area for magnetotelluric [6.10] and borehole induction logging problems
[6.16], but very little attention has been given to the transient electromagnetic
techniques, particularly airborne methods. To be truly useful to the geophysicist, it is
critical that the approximate forward modelling methods be able to capture the full
geological complexity of the earth, such as topography, induced polarisation effects as
captured in frequency-dependent complex conductivities, high conductivity contrasts
and non-ideal transmitter/receiver characteristics. Integral equations offer some
advantages over differential equation methods, but difficulty exists in being able to
construct (and solve) truly arbitrary and complex models with high conductivity
contrasts. Whilst finite-difference methods are conceptually simple to construct, they
cannot necessarily capture the entire geological complexity of the earth, such as
irregular boundaries. Modern finite-element methods [6.7] are able to capture the entire
geological complexity through curvi-linear cell structures, and probably offer the best
opportunity for realising the goals of developing a general and computationally
efficient approximate forward modelling method.
The third, and most industrially relevant direction of future research is the
development of three-dimensional inversion methods that can solve for inhomogeneous
anisotropic conductivity. Whilst this is currently an intense research area for borehole
induction logging problems, relatively little focus has been given to the inversion of
anisotropic parameters in other electromagnetic techniques. Currently, the majority of
inversion methods developed for surface and airborne techniques seek only those
inverse solutions with isotropic conductivity distributions. Ultimately, as we seek to
solve for realistic inverse models from survey data, it will be essential to have inverse
methods that can capture the necessary geological complexity that exists in reality. The
most challenging problem for inverting data with anisotropic and inhomogeneous
conductivity distributions will be the inherent non-uniqueness that exists, in addition to
the problems of instability and existence. To be able to extract such useful information
though, the ability to input sufficiently more geological a priori data into the inverse
models through different regularisation methods [6.15] will be critical.
127
6.2.2. Complex media
The study of microwave and optical communications has recently been subject
to rapid growth, in terms of both fundamental theory and industrial applications. New
techniques for guiding waves for communication purposes have been of significant
interest. This particular field of study has renewed studies in the broad discipline of
high frequency wave propagation in complex media, inclusive of anisotropic,
bianisotropic and chiral media, photonic crystals, photonic band-gap structures and so-
called metamaterials.
At microwave frequencies, there has been significant interest in the studies of
printed circuits (antennas and transmission lines) on anisotropic substrates [6.1, 6.3,
6.6]. At optical frequencies, there has been significant interest in the studies of glass
[6.2] and silicon-on-insulator [6.8] waveguides. Underlying this whole discipline are
the theoretical investigations into the electrodynamics of complex media [6.4].
Whilst this thesis has not concerned itself with studying these specific problems,
it is observed that there are similarities between these problems, and those geophysical
ones considered in this thesis. Possible future research efforts could include
investigations into integrated optical propagation problems, such as turning mirrors and
waveguides, in anisotropic silicon-on-insulator. Also, metamaterial-filled waveguides
for optical and microwave propagation will provide an interesting research topic.
Electromagnetic wave propagation in periodic and infinite structures (e.g., photonic
crystals) for optical applications has been studied for many years now. However, there
is difficulty in being able to manufacture such devices and correlating the experimental
results to theory. This will continue to be an area of active research. Other possible
research efforts could include further investigations of planar antennas and transmission
lines on anisotropic and bianisotropic substrates.
6.3. References
[6.1] V. Losada, R. R. Boix & F. Medina, IEEE Trans. Antennas Propagat. 49, 1603
(2001).
[6.2] C. J. Markman, Y. Ren, G. Genty & M. Kristensen, IEEE Photonics Tech. Lett.
14, 1294 (2002).
[6.3] F. Mesa, D. R. Jackson & M. J. Friere, IEEE Trans. Microwave Theory Tech.
50, 94 (2002).
128
[6.4] F. Olyslager & I. V. Lindell, IEEE Antennas Propagat. Mag. 44, 48 (2002).
[6.5] E. I. Parkhomenko, Electrical Properties of Rocks (Plenum Press, New York,
1967).
[6.6] R. Pregla, IEEE Trans. Microwave Theory Tech. 50, 1469 (2002).
[6.7] F. Sugeng, Explor. Geophys. 29, 615 (1998).
[6.8] Y. Z. Tang, W. H. Wang, T. Li & Y. L. Wang, IEEE Photonics Tech. Lett. 14,
68 (2002).
[6.9] D. V. Thiel & G. A. Wilson, presented at URSI Nat. Sci. Meet., Boulder, CO,
Jan. 2002.
[6.10] P. Weidelt, in Three-Dimensional Electromagnetics, edited by M. Oristaglio &
B. Spies (Soc. Explor. Geophys., Tulsa, 1999), pp. 119-137.
[6.11] G. A. Wilson & D. V. Thiel, Radio Sci. 37, 1029, 2001RS002535 (2002).
[6.12] G. A. Wilson & D. V. Thiel, IEEE Trans. Geosci. Remote Sensing, submitted.
[6.13] G. A. Wilson & D. V. Thiel, J. Electromagn. Waves Applic., submitted.
[6.14] G. A. Wilson & D. V. Thiel, presented at IEEE Int. Antennas Propagat. Symp.,
San Antonio, TX, Jun. 2002.
[6.15] M. S. Zhdanov, Geophysical Inversion Theory and Regularization Problems
(Elsevier, Amsterdam, 2002).
[6.16] M. S. Zhdanov, D. Kennedy & E. Peksen, Petrophys. 42, 118 (2001).
- The End -
129
Appendix 1
Co-ordinate rotations
�The angles ( ),, ψθϕ are called the Euler angles. Their definition varies widely
� the probability is small that two distinct authors� general rotation matrix will be the
same.� [A1.1]
Let us consider another Cartesian co-ordinate system that is represented by {x�,
y�, z�} where the y�z�-plane is inclined at an angle α about the x-axis, the x�z�-plane is
inclined at an angle β about the y-axis and the x�y�-plane is inclined at an angle θ about
the z-axis. We call this the inclined co-ordinate system (Figures 1.2-1.4). α, β and θ
represent the three elementary Euler angles about the x, y and z axes respectively,
which can be used to rotate one frame of reference to another. When 0==θβ , we
can write:
'xx = , (A1.1)
αα sin'cos' zyy −= , (A1.2)
αα cos'sin' zyz += , (A1.3)
which can be written in matrix notation as:
=
'''
)(zyx
zyx
αR , (A1.4)
where:
−=
ααααα
cossin0sincos0001
)(R . (A1.5)
130
Similarly, when 0== βα , we can write:
θθ sin'cos' yxx += , (A.16)
θθ cos'sin' yxy +−= , (A1.7)
'zz = , (A1.8)
which can be written in matrix notation as:
=
'''
)(zyx
zyx
θR , (A1.9)
where:
−=
1000cossin0sincos
)( θθθθ
θR . (A1.10)
Similarly, when 0==θα , we can write:
ββ sin'cos' zxx −= , (A1.11)
'yy = , (A1.12)
ββ cos'sin' zxz += , (A1.13)
which can be written in matrix notation as:
=
'''
)(zyx
zyx
βR , (A1.14)
131
where:
−=
ββ
βββ
cos0sin010
sin0cos)(R . (A1.15)
When 0≠α , 0≠β and 0≠θ , one can write:
.coscoscossinsin
sincossincossinsinsinsincoscoscossinsincoscossinsinsinsincoscossincoscos
)()()(),,(
−+−
−−−=
=
βαβαβαθβαθβαθαθβθβαθαθβαθαθβθ
αβθαβθ RRRR
(A1.16)
Note that we can also rotate from the fundamental co-ordinates into the inclined
co-ordinates analogous to equation (A1.16), but would have to rotate the fundamental
co-ordinates by the negative of the Euler angles.
Reference
[A1.1] F. W. Byron & R. W. Fuller, Mathematics of Classical and Quantum Physics
(Dover Publications, New York, 1992).
132
Appendix 2
Tensor rotation
We can write the field form of Ohm's Law in {x�, y�, z�} co-ordinates as:
' '�' EJ σ= , (A2.1)
where '�σ is the second rank conductivity tensor, and the primes denote quantities in the
inclined co-ordinate system. To write equation (A2.1) in the fundamental co-ordinate
system, we can write equation (A2.1) as [A2.1]:
' '� ),,(' ),,( ERJR σαβθαβθ = . (A2.2)
Since , then we can write equation (A2.2) as: IRR =),,( ),,(T αβθαβθ
ERRJ ),,('� ),,( T αβθσαβθ= , (A2.3)
or as:
EJ � σ= , (A2.4)
where , and where J and E are the current density and
electric field intensity, respectively, in the fundamental co-ordinate system. We note
here that
),,('� ),,(� T αβθσαβθσ RR=
),,( αβθR is an orthogonal matrix, i.e., R . ),,( ),,( -1T αβθαβθ R=
Reference
[A2.1] M. S. Zhdanov, D. Kennedy & E. Peksen, Petrophys. 42, 588 (2001).
133
Appendix 3
Solution to one-dimensional magnetic field coefficients
Consider the expression for the background magnetic field:
)exp()exp()(, zkBzkAzH mmmmb
mx +−= . (3.15)
At the air-half space interface for z = 0, we have:
mm BAH +=0 , (A3.1)
where is the (complex) amplitude of the magnetic field in free space. If 0H 0=bσ
then the vanishing magnetic field at z = h is equivalent to the statement that the top of
the basement is a perfect magnetic conductor. For z = h:
( ) ( hkBhkA mmmm expexp0 +−= ) . (A3.2)
From equation (A3.1), we can obtain the relations and
. Respectively substituting these into equation (A3.2), we then obtain the
solutions to the field equation coefficients above a perfectly insulating basement as:
mm AHB −= 0
mm BHA −= 0
( )( )hk
hkHA
m
mm sinh2
exp0= , (A3.3)
( )( )hk
hkHB
m
mm sinh2
exp0 −−= . (A3.4)
If ∞=bσ then at z = h which implies that 0, =myE 0, =∂
∂
=hz
mx
zH
which we can
write as:
( ) ( hkBhkA mmmm expexp0 −−= ) , (A3.5)
134
and equation (A3.1) still applies. Using the same relations as used before we then
obtain solutions to the field equation coefficients above a perfectly conducting
basement as:
( )( )hk
hkHA
m
mm cosh2
exp0= , (A3.6)
( )( )hk
hkHB
m
mm cosh2
exp0 −−= . (A3.7)
135
Appendix 4
Solutions to the Helmholtz equation
To solve Helmholtz equation (3.25) of the anomalous magnetic field in an
inclined uniaxial anisotropic medium rotated into the fundamental co-ordinates, we
must satisfy both equations:
,0sin)(
sin)(cossin
sin)(sincos
,
,2
22
,
2
,
2
2,
2
,
2
,
2
=
−
+−
∂∂
+
hznyfj
hznyf
hn
hzn
yyf
nm
nmmt
m
mn
m
nm
mt
m
mn
m
πωµ
ππσ
ασ
α
πσ
ασ
α
(A4.1)
and
. 0)(, =
∂∂
yyf nm (A4.2)
We will consider a solution of the form:
+
−=
±±
hyq
bh
yqayf nm
nmnm
nmnm,
,,
,, expexp)( , (A4.3)
where the ± superscript above correspond to the positive and negative roots of
to be determined. From equation (A4.3), we can observe that:
nmq ,
2,nmq
)()(
,2
2,
2,
2
yfh
qy
yfnm
nmnm =∂
∂. (A4.4)
Substituting equation (A4.4) into equation (A4.1) results in the equation:
136
,0sin)(
sin)(cossin
sin)(sincos
,
,2
22
,
2
,
2
,2
2,
,
2
,
2
=
−
+−
+
hznyfj
hznyf
hn
hznyf
hq
nm
nmmt
m
mn
m
nmnm
mt
m
mn
m
πωµ
ππσ
ασ
α
πσ
ασ
α
, (A4.5)
where the and )(, yf nm
hznπsin terms cancel, leaving us with the equation:
.0 cossin
sincos
2
22
,
2
,
2
2
2,
,
2
,
2
=−
+−
+ ωµπ
σα
σα
σα
σα
jh
nh
q
mt
m
mn
mnm
mt
m
mn
m (A4.6)
The coefficient q can be isolated from equation (A4.6) by first writing: nm,
,0sincos
cossinsincos
1
,
2
,
2
2
22
,
2
,
21
,
2
,
2
2
2,
=
+−
+
+−
−
−
mt
m
mn
m
mt
m
mn
m
mt
m
mn
mnm
j
hn
hq
σα
σαωµ
πσ
ασ
ασ
ασ
α
(A4.7)
where we identify the wave number of the medium:
1
,
2
,
22 sincos
−
+=
mt
m
mn
mm jk
σα
σαωµ , (A4.8)
and then re-arrange equation (A4.7) to the form of:
0cossinsincos 22
,
2
,
21
,
2
,
2222
, =−
+
+−
−
hknq mmt
m
mn
m
mt
m
mn
mnm σ
ασ
ασ
ασ
απ . (A4.9)
We can identify that:
137
mmnmmt
mmnmmt
mt
m
mn
m
mt
m
mn
m
ασασασασ
σα
σα
σα
σα
2,
2,
2,
2,
,
2
,
21
,
2
,
2
sincoscossincossinsincos
++
=
+
+
−
,
which simply states that:
yy
zzzzyy σ
σρρ =−1 ,
so we can then write equation (A4.9) as:
++
+±=±
mmnmmt
mmnmmtmnm nhkq
ασασασασ
π 2,
2,
2,
2,2222
, cossincossin
. (A4.10)
Now, we seek to demonstrate when equation (A4.3) satisfies equation (A4.2).
We have the derivative:
+
−−=
∂∂ ±±±±
hyq
bh
qh
yqa
hq
yyf nm
nmnmnm
nmnmnm ,
,,,
,,, expexp
)(, (A4.11)
and from equation (A4.2), we can then write:
−=
±±
hyq
ah
yqb nm
nmnm
nm,
,,
, expexp , (A4.12)
which is satisfied when we write:
−=
+−
hyq
ah
yqb nm
nmnm
nm,
,,
, expexp ,
where and are the negative and positive roots of respectively to
prevent exponentially divergent solutions, since it is later observed due to symmetry
conditions that a .
−nmq ,
+nmq ,
nm, =
2,nmq
nmb ,
138
Appendix 5
Solution to Fourier series coefficients
Let us consider the integral:
( )∫
−=
h
n dzh
znHHh
C0
12 sin2 π , (A5.1)
where ( ) ( )
−−
−=−
hkzhk
hkzhk
HHH1
1
2
2012 sinh
sinhsinh
sinh . We can write equation (A5.1)
as:
( ) ( )
−
−
−
= ∫∫ dzh
znhk
zhkdz
hzn
hkzhk
hH
Chh
nππ
0 1
1
0 2
20 sinsinh
sinhsin
sinhsinh2
. (A5.2)
Let us now consider the solution to the general integral:
( ) dzh
znkh
zhk
−
∫πsin
sinhsinh ,
which we will write as ( ) dzh
znzhkkh
−∫
πsinsinhsinh
1 . Let
and
( )zhku −= sinh
dzh
zndv
= πsin so that we can use the integration by parts method,
∫ ∫−= vduuvudv , to then obtain:
( ) ( )
( ) .coscoshsinh
cossinhsinh
sinsinhsinh
1
dzh
znzhkkhn
khh
znzhkkhn
hdzh
znzhkkh
−−
−−=
−
∫
∫π
π
ππ
π
(A5.3)
139
Now let u and ( zhk −= cosh ) dzh
zndv
= πcos so that we can use the
integration by parts method, ∫ ∫−uv=udv vdu , to then obtain:
( ) ( )
( )
( ) ,sinsinhsinh
sincoshsinh
cossinhsinh
sinsinhsinh
1
22
22
22
2
dzh
znzhkhn
hkh
znzhkkhn
khh
znzhkkhn
hdzh
znzhkkh
−−
−−
−−=
−
∫
∫
πγπ
ππ
ππ
π
(A5.4)
which we can then write as:
( )
( ) ( ) ,sincoshsinh
cossinhsinh
sinsinh1sinh
1
22
2
22
22
Ch
znzhkkhn
khh
znzhkkhn
h
dzh
znzhkn
hkkh
+
−−
−−
=
−
+ ∫
ππ
ππ
ππ
(A5.5)
where we introduce C as a constant of integration. Equation (A5.5) can then expressed
as:
( )
( ) ( ) ,sincoshsinh
cossinhsinh
sinsinhsinh
12
222
Ch
znzhkkh
kh
znzhkkhh
n
dzh
znzhkh
nkkh
+
−−
−−
=
−
+ ∫
πππ
ππ
(A5.6)
so then we have solved for the general integral as:
140
( ) ( )
( ) .sincoshsinh
cossinhsinh
sinsinhsinh
1
2
222
2
222
Ch
znzhk
hnkkh
k
hznzhk
hnkkhh
ndzh
znzhkkh
+
−
+
−
−
+
−=
−∫
ππ
ππ
ππ
(A5.7)
Now we apply the bounds between 0 and h to obtain the general definite
integral:
( )
+
=
−∫
2
2220
sinsinhsinh
1
hnkh
ndzh
znzhkkh
h
πππ . (A5.8)
We can now solve equation (A5.2) as:
( )
+
+
−−=
2
222
22
222
12
21
2202
hnk
hnkh
kknHCn ππ
π. (A5.9)
An identical solution is obtained for the integral:
( )∫
+−=
h
n dzh
znHHh
C0
12 2
)12(sin1 π , (A5.10)
which has the solution:
( )
++
++
−+−=
2
222
22
222
12
21
220
4)12(
4)12(4
)12(
hnk
hnkh
kknHCn ππ
π. (A5.11)
141
Appendix 6
Finite-difference operators
Consider an arbitrary function f (x, y) in a curvilinear co-ordinate system.
Consider a Taylor series expansion about the point (x, y) such that:
...!3)(
!2)(
!1)(
3
3
32
2
2
+∆+∆+∆+=∆+ xdx
fdxdx
fdxdxdffxxf , (A6.1)
...!3)(
!2)(
!1)(
3
3
32
2
2
+∆−∆+∆−=∆− xdx
fdxdx
fdxdxdffxxf , (A6.2)
...!3)(
!2)(
!1)(
3
3
32
2
2
+∆+∆+∆+=∆+ ydy
fdydy
fdydydffyyf , (A6.3)
...!3)(
!2)(
!1)(
3
3
32
2
2
+∆−∆+∆−=∆− ydy
fdydy
fdydydffyyf . (A6.4)
Subtracting equation (A6.2) from equation (A6.1), then we obtain the first order
derivative with respect to x:
3)(2
)()( xOx
xxfxxfdxdf ∆+
∆∆−−∆+= . (A6.5)
Subtracting equation (A6.4) from equation (A6.3), then we obtain the first order
derivative with respect to y:
3)(2
)()( yOy
yyfyyfdydf ∆+
∆∆−−∆+= . (A6.6)
Adding equations (A6.2) and (A6.1), then we obtain the second order derivative
with respect to x:
142
422
2
)()(
)()(2)( xOx
xxfxfxxfdx
fd ∆+∆
∆−+−∆+= . (A6.7)
Adding equations (A6.4) and (A6.3), then we obtain the second order derivative
with respect to y:
422
2
)()(
)()(2)( yOy
yyfyfyyfdy
fd ∆+∆
∆−+−∆+= . (A6.8)
If we neglect and higher order terms in equation (A6.5) and re-write it
as:
3)( xO ∆
xxxf
xxxf
dxdf
∆∆−−
∆∆+=
2)(
2)( , (A6.9)
then apply equation (A6.6), in which we have also neglected O , then we obtain
the mixed derivatives with respect to x and y:
3)( x∆
zyyyxxfyyxxfyyxxfyyxxf
zyf
∆∆∆−∆−+∆+∆−−∆−∆+−∆+∆+=
∂∂∂
4),(),(),(),(2
.
(A6.10)
143
Appendix 7
Inhomogeneous plane waves
Definition 1: A homogeneous plane wave is: a wave for which the planes of constant
amplitude and constant phase are parallel. Homogeneous plane waves are sometimes
called uniform plane waves [A7.1].
Definition 2: An inhomogeneous plane wave is: a wave for which the planes of
constant amplitude and planes of constant phase are not parallel. Sometimes called a
heterogeneous plane wave, but this use is depreciated [A7.1].
Consider a plane wave A incident upon a half space in orthogonal {x, y, z} co-ordinates,
at an arbitrary (complex) angle θ with respect to (from) the normal of the x-y plane.
Definition 3: The wave vector is: the complex vector kr
in expressions for wave
propagation using the exponential factor )](exp[ rkrr
⋅− j [A7.1].
For the plane wave, one can write:
}�'exp{ 'zkzjA −= , (A7.1)
where:
xx =' , (A7.2)
θθ sincos' zyy −= , (A7.3)
θθ cossin' zyz += . (A7.4)
Equation (A7.1) is a homogeneous plane wave in the ' -plane. We can re-write
equation (A7.1) in {x, y, z} co-ordinates as:
' yx
}�)cossin(exp{ 'zk θθ zyjA +−= ,
144
which we will re-write as:
}exp{}exp{ })cos(exp{})sin(exp{
zjkyjkzkjykjA
zy −−=−−= θθ
(A7.5)
where θsink=yk and θcosk=zk are the vector components of k. Equation (7.5) is
an inhomogeneous plane wave in the xy-plane. By virtue of equation (7.5), in the xy-
plane, the plane of constant amplitude is not parallel to the plane of constant phase. It is
specifically noted that equation (A7.5) is of the form of the Zenneck surface wave
[A7.2].
Remark: At the surface of a half space, an inhomogeneous plane wave can sometimes
be defined by a homogeneous plane wave at oblique incidence upon the surface of the
half space.
References
[A7.1] IEEE Std 211-1997TM, IEEE Standard Definitions of Terms for Radio Wave
Propagation.
[A7.2] J. R. Wait, IEEE Antennas Propagat. Mag. 40, 7 (1998).