Вы находитесь на странице: 1из 9

ARTICLE IN PRESS

Computers & Geosciences 32 (2006) 1411–1419


www.elsevier.com/locate/cageo

CR1Dmod: A Matlab program to model 1D complex resistivity


effects in electrical and electromagnetic surveys$
Thomas Ingeman-Nielsena,, Franc- ois Baumgartnera,b
a
The Arctic Technology Center, Technical University of Denmark, Building 204, DK-2800 Kgs. Lyngby, Denmark
b
Institute of Geophysics, Faculty of Earth Sciences and Environment, University of Lausanne, CH-1015 Lausanne, Switzerland
Received 28 October 2004; received in revised form 19 September 2005; accepted 9 January 2006

Abstract

We have constructed a forward modelling code in Matlab, capable of handling several commonly used electrical and
electromagnetic methods in a 1D environment. We review the implemented electromagnetic field equations for grounded
wires, frequency and transient soundings and present new solutions in the case of a non-magnetic first layer. The
CR1Dmod code evaluates the Hankel transforms occurring in the field equations using either the Fast Hankel Transform
based on digital filter theory, or a numerical integration scheme applied between the zeros of the Bessel function. A
graphical user interface allows easy construction of 1D models and control of the parameters. Modelling results are in
agreement with other authors, but the time of computation is less efficient than other available codes. Nevertheless, the
CR1Dmod routine handles complex resistivities and offers solutions based on the full EM-equations as well as the quasi-
static approximation. Thus, modelling of effects based on changes in the magnetic permeability and the permittivity is also
possible.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Complex resistivity; Forward modelling; Terrestrial resistivity; EM coupling; Geophysical prospecting

1. Introduction software with the ability to take into account CR


effects is very scarce.
Lately, the Arctic Technology Center has had an The U.S. Geological Survey has published reports
interest in evaluating the application of the complex of modelling codes written in FORTRAN for
resistivity (CR) method to groundwater prospecting several electrical and electromagnetic methods,1 of
and permafrost mapping in Greenland. During the which the EMCUPL dipole–dipole modelling rou-
course of our research it became apparent that the tine (Kauahikaua and Anderson, 1979) is one of the
availability of free, easy to use forward modelling few available codes to incorporate CR effects.

$ 1
Code available from server at http://www.iamg.org/CGEditor/ Bankey, V., Anderson, W.L., 1995. A bibliography of some
index.htm. geophysical computer programs, databases, and maps from the
Corresponding author. Tel.: +45 45252251; U.S. Geological Survey, 1971–1994. Open-File report 95-77, U.S.
fax: +45 45885935. Geological Survey, available at: http://pubs.usgs.gov/of/1995/
E-mail address: tin@byg.dtu.dk (T. Ingeman-Nielsen). ofr-95-0077/of-95-77.html.

0098-3004/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cageo.2006.01.001
ARTICLE IN PRESS
1412 T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419

Modelling and field data from a loop–loop system dependence (no unit). In practice, electromagnetic
over a polarizable, layered half-space in the frequency theory calls for the use of conductivities rather than
domain were reported by Hohmann et al. (1970), and resistivities. In our definition of complex earth
Descloitres et al. (2000) report modification of a properties, the complex conductivity is simply the
TEM forward modelling routine to accommodate inverse of the CR.
CR effects and its application to the interpretation of
negative transients in a volcanic environment. None 3. Theoretical basis
of these codes has been made available to the
scientific community in an organized manner. The traditional way of measuring the CR of the
We therefore decided to implement the ground is by measuring the complex impedance in
CR1Dmod code, which would handle some of the the frequency domain between two grounded wires
most commonly encountered 1D forward modelling in the co-linear dipole configuration. However, as
problems with CR. The Matlab environment was the complex properties of the ground may poten-
chosen for the development of the code. Although, tially manifest themselves in any electromagnetic
computationally slower than similar implementa- measurement, a range of different methods and
tions in FORTRAN or C, the Matlab environment configurations have been implemented in
offers structure and easily programmable GUIs. CR1Dmod. The following sections summarize the
Furthermore, with Matlab widely used in educa- theoretical background and some practical consid-
tional activities at many universities, it will be fairly erations of the implemented algorithms.
easy for researchers, teachers and students to
modify the configurations or algorithms of the 3.1. EM coupling between grounded wires
CR1Dmod to accommodate specific needs or to
implement new ones altogether. According to Sunde (1968), the electromagnetic
coupling between two grounded wires in an arbitrary
2. The complex resistivity model configuration on the surface of the earth can be
calculated as an integration of mutual impedances of
That resistivity (or electrical impedance) measure- virtual dipoles along the paths of the wires:
ments of geological materials are complex and vary Z B Z b 
with the frequency of the applied current is a well q2 QðrÞ
Z¼ PðrÞ cos x þ ds dS
established fact. Over the years, several models have A a qs qS
been proposed for the description of the dispersive Z BZ b
behavior of the CR (Marshall and Madden, 1959; ¼ PðrÞ cos x ds dS
A a
Zonge, 1972; Wong, 1979), but the most widely used
þ QðjAajÞ  QðjBajÞ  QðjAbjÞ þ QðjBbjÞ, ð2Þ
seem to be the Cole–Cole model, which was
originally developed by Cole and Cole (1941) to where A; B and a,b are the end points of the trans-
describe dielectric dispersion. Pelton et al. (1978) mitter and receiver wires (the grounding points), dS
showed that this model would accurately describe and ds are infinitesimal elements of the two wires (the
the resistivity dispersion observed in field data virtual dipoles), x is the angle between the wire elements
obtained at low frequencies from areas with metallic and r is the distance between them. The P and Q
mineral content. It has later been shown to also fit functions are given by (Ward and Hohman, 1988)
the spectra obtained from laboratory measurements Z
z^0 1 l
on sedimentary samples polluted with organic PðrÞ ¼ ½1 þ RTE 0  J 0 ðlrÞ dl, (3)
4p 0 u0
solutions (Vanhala et al., 1992).
The Cole–Cole model is the resistivity model Z  
1 1  
TM u0
 
TE z^0
chosen for implementation in the CR1Dmod code QðrÞ ¼ 1  R0  1 þ R0
4p 0 y^ 0 u0
and is given by
   1
1  J 0 ðlrÞ dl. ð4Þ

r ¼ r0 1  m 1  , (1) l
1 þ ðio tÞc Here, J 0 is the Bessel function of order 0 and l is the
where r is the CR, r0 is the resistivity in the direct variable of integration. RTE n and Rn
TM
are the general-
current (DC) limit ðO mÞ, m is the chargeability (no ized reflection coefficients for the transverse electric and
unit), t is the time constant ðsÞ and c is the frequency transverse magnetic modes looking into layer n þ 1
ARTICLE IN PRESS
T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419 1413

Here, mn is the magnetic permeability of layer


µ0, ε0
Air
n ðH m1 Þ and n is the permittivity ðF m1 Þ. The
Ground
0 recursion is started with RTM N ¼ 0. The relations for
TE
RTE
n and c n are obtained when z^n is exchanged for y^ n
σ1, µ1, ε1 h1 in cTM
n .
Z1
A few special cases exist for which simplifications
of the general equations can be obtained. In the case
σ2, µ2, ε2 h2 of a non-magnetic half-space (m1 ¼ m0 ), the P
function can be evaluated analytically. If a layered
Z2 model has a non-magnetic top layer, the homo-
geneous half-space term, based on the properties of
the first layer, can be subtracted from the kernel
function and the corresponding analytical expres-
sion is added outside the transform:
Zn-1 "Z
z^0 1
4RTE1 u1
σn, µn, εn hn
PðrÞ ¼ 2 2 2
lJ 0 ðlrÞ dl
4p 0 RTE 1 ðk 1  k 0 Þ þ ðu0 þ u1 Þ
Zn 2
þ ððik1 r þ 1Þeik1 r
ðk21  k20 Þr3
σn+1, µn+1, εn+1 hn+1 #

Zn+1 ðik0 r þ 1Þeik0 r Þ . ð6Þ

In this formulation, the Hankel transform is a


correction term to the homogeneous half-space
solution, corresponding to the summed effect of
ZN-1 the additional layers.
If, in addition to the above assumption, frequen-
σN, µN, εN cies are sufficiently low that distances involved in
Z the problem are much shorter than the free-space
wavelength, so that the dielectric properties in the
Fig. 1. Basis of the model is a horizontally stratified medium of
N layers. air can be neglected (y^ 0  0, k0  0), both the
P- and Q-functions can be further simplified as
"Z
(Fig. 1). These depend on the layering of the subsurface z^0 1
4RTE
1 u1
and the electromagnetic properties of each layer and PðrÞ ¼ 2 2
lJ 0 ðlrÞ dl
4p 0 RTE 1 k1 þ ðl þ u1 Þ
are given by the recursive relations #
RTM TM 2
nþ1 þ cnþ1
ik1 r
RTM ¼ e2un hn , þ 2 ððik1 r þ 1Þe  1Þ , ð7Þ
n
RTM TM k 1 r3
nþ1 cnþ1 þ 1
un unþ1 "Z "
 1 1
RTE
1 z^0
y^ n y^ nþ1 QðrÞ ¼ 
cTM ¼ un unþ1 , ð5Þ 4p 0 RTE 2 2
nþ1 1 k1 þ ðl þ u1 Þ
þ # #
y^ n y^ nþ1 RTM 4u1 2^z0
1
þ TM J 0 ðlrÞ dl þ 2 . ð8Þ
where hn is the thickness of layer n (h0 ¼ 0), and: ðR1 þ 1Þy^ 1 l k1 r
y^ n ¼ sn þ ion ; y^ 0 ¼ io0 ,

z^n ¼ iomn ,
3.2. TEM calculations
k2n 2
¼ ^zn y^ n ¼ o n mn  iomn sn ,
CR1Dmod includes a functionality to model
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
transient soundings (TEM-soundings) in the central
un ¼ l2  k2n . loop configuration. The implemented algorithm is
ARTICLE IN PRESS
1414 T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419

based on that of Ryu et al. (1970) and Anderson loop as observed by a dipole receiver. This functio-
(1981), in which the vertical field at the center of a nality uses integration of the H z -field from virtual
circular transmitting loop is calculated in the dipoles along the transmitter loop, in a way similar
frequency domain and transformed to time domain to the one described for the electric field of the
by means of an inverse cosine or sine transform. For grounded dipoles (however without grounding
consistency, we have rewritten the equations using points). Access to this type of calculation has not
the same notation as for the case of the grounded yet been included in the graphical user interface, but
dipoles. The frequency domain field calculation is is planned for a future version.
given by
Z 1  3.2.1. Time-domain calculations
l 1 I
H z ðoÞ ¼ Ia ð1 þ RTE
0 Þ  lJ 1 ðlaÞ dl þ Since the transient response is a causal function
0 2u0 2 2a
Z 1 ! ðhz ðtÞ ¼ 0 for tp0Þ, the transformation of the
TE 1
ðR1 þ 1Þðu0 þ ð^z0 =^z1 Þu1 Þ 1 frequency response to the time domain can be
¼ Ia TE TE

0 R1 c1 þ 1 2 obtained by means of a sine or cosine transform,
and following Anderson (1974, 1981):
I Z
lJ 1 ðlaÞ dl þ . ð9Þ 2 11
2a hz ðtÞ ¼  Im½H z ðoÞ cosðotÞ do. (12)
Here, a is the radius of the transmitter loop, and the p 0 o
term l=2 which is subtracted in the Hankel trans- In Eq. (12) step-current excitation has been
form kernel in Eq. (9) corresponds to the static field assumed. Assuming that the field does not vary
in the DC limit. The term can be integrated over the area of the receiver loop, the relation
analytically, and the resulting expression is added between the vertical field and the induced voltage is
outside the Hankel transform. This operation given by
ensures definite convergence of the integrand (Ryu
dhz ðtÞ
et al., 1970). vðtÞ ¼ m0 npb2 , (13)
Analogous to the grounded wire case, the kernel dt
function of the H z ðoÞ function can be simplified in where b is the radius of the receiver loop and n is the
the case of a non-magnetic first layer (m1 ¼ m0 ), by number of turns. After making suitable substitu-
subtracting a homogeneous half-space term from tions, the resulting mutual impedance can be
the kernel function and adding the analytic expres- expressed as
sion outside the transform, giving Z    
2nb2 1 2g 2a
Z 1 ! ZðtÞ ¼  Im H z sinðgtÞ dg.
3
s1 a 0 s1 m0 a 2 I
2RTE
1 u1 l 1
H z ðoÞ ¼ Ia 
RTE 2 2 2 2 (14)
0 1 ðk 1  k 0 Þ þ ðu0 þ u1 Þ
I Here, t ¼ 2tðs1 m0 a2 Þ1 and g ¼ 12s1 m0 oa2 , where s1
lJ 1 ðlaÞ dl þ is the DC conductivity of the first layer.
2a
I
þ 2 ððk20 a2  3ik0 a  3Þeik0 a 3.3. FDEM calculations
ðk1  k20 Þa3
 ðk21 a2  3ik1 a  3Þeik1 a Þ. ð10Þ The horizontal co-planar (HCP) loop system, also
At sufficiently low frequencies (k0  0), the expres- called a HLEM system, is popular in near-surface
sion further simplifies to electromagnetic prospecting. Theoretically, the coils
Z 1 ! of the system are usually treated as vertical magnetic
2RTE
1 u 1 l 1 dipoles. The general expression for the z-component
H z ðoÞ ¼ Ia 2 2

0 RTE
1 k1 þ ðl þ u1 Þ
2 of the magnetic field, generated by such a system
I placed on the surface of a layered half-space, is
lJ 1 ðlaÞ dl þ given by Ward and Hohman (1988):
2a
Z
I
ð3 þ ðk21 a2  3ik1 a  3Þeik1 a Þ. ð11Þ m 1 l3
 Hz ¼ ½1 þ RTE  J 0 ðlrÞ dl, (15)
k21 a3 4p 0 0
u0
The implemented calculation routine also has the where m is the magnetic moment of the transmitter,
capability of modelling the response from a square and all other parameters are as previously defined.
ARTICLE IN PRESS
T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419 1415

Wait (1952) pointed out that for the case of a non- with
magnetic half-space, Eq. (15) can be solved analy- Z 1
r1
tically. Thus, similar to the case of the P-term of the QðrÞ ¼ K 1 ðlÞJ 0 ðlrÞ dl, (21)
2p 0
grounded dipoles and the TEM central loop
expression, the homogeneous half-space term can where
be subtracted from the kernel function and the snþ1
analytical expression added outside the transform: K nþ1 þ tanhðlhn Þ
si
"Z K n ðlÞ ¼ snþ1 ; K N ¼ 1.
m 1
4RTE
1 u1
þ K nþ1 tanhðlhn Þ
Hz ¼ 2 2 2
l3 J 0 ðlrÞ dl sn
4p 0 RTE 1 ðk1  k 0 Þ þ ðu0 þ u1 Þ
Computational efficiency of Eq. (21) is usually
2 increased by subtracting the homogeneous half-
þ ðð9 þ 9ik0 r  4k20 r2  ik30 r3 Þeik0 r
ðk21  k20 Þr5 space response from the kernel and adding it outside
#
the Hankel transform:
ð9 þ 9ik1 r  4k21 r2  ik31 r3 Þeik1 r Þ . ð16Þ Z 1 
r1 1
QðrÞ ¼ ðK 1 ðlÞ  1ÞJ 0 ðlrÞ dl þ . (22)
2p 0 r
The expression further simplifies at sufficiently low
frequencies (k0  0):
"Z
m 1
4RTE
1 u1
Hz ¼ 2 2
l3 J 0 ðlrÞ dl 4. Calculation of Hankel and cosine transforms
4p 0 RTE 1 k1 þ ðl þ u1 Þ
#
2 The general form of the Hankel transform, which
2 2 3 3 ik1 r
þ 2 ð9  ð9 þ 9ik1 r  4k1 r  ik1 r Þe Þ . occurs in all the presented EM-field equations, can
k 1 r5 be written as
ð17Þ Z 1
f ðr; FÞ ¼ Kðl; FÞJ n ðlrÞ dl, (23)
Typically, data from magnetic dipole–dipole mea- 0
surements are given as mutual coupling ratios,
where K is the kernel function, J n is the Bessel func-
defined as the measured impedance normalized by
tion of order n, F represents the model parameters
the theoretical impedance of the system in free-
(here the layer properties) and r is the calculation
space. The mutual coupling ratio can be calculated
distance. The inverse Fourier transform, in the form
by dividing the expression for H z at the position of
of cosine or sine transform, which is used in the
the receiver by the corresponding primary field, the
time-domain calculations, is obtained by exchan-
field that would be measured in free-space. For the
ging J n with the cos or sin functions.
HCP configuration the primary field is given by
Two different approaches exist for the calculation
(Ward and Hohman, 1988)
of Hankel and harmonic transforms: numerical
m ik0 r 2 2 integration and digital filtering. CR1Dmod includes
H Pz ¼ e ðk0 r  ik0 r  1Þ. (18)
4pr3 implementations of both approaches.
In case of quasi-static conditions, this expression
simplifies to 4.1. Numerical integration
m
H Pz ¼  . (19)
4pr3 Numerical integration of Hankel transforms have
been reported by a number of authors (Frischknecht,
1967; Dey and Ward, 1970; Hohmann et al., 1970).
3.4. DC geoelectrical calculations The general approach is to evaluate the integral in
segments between the zeros of the Bessel function.
Forward modelling capability of DC geoelectrical Usually the first few segments, where the kernel
soundings has been included for completeness. The function shows the largest variations, are computed
algorithm used to calculate the DC resistance is using an adaptive quadrature rule, and the remainder
given by Koefoed (1979): of the integral, usually limited to a fixed number of
segments, is calculated using a non-adaptive Gaus-
R ¼ QðjAajÞ  QðjBajÞ  QðjAbjÞ þ QðjBbjÞ (20) sian quadrature rule. If the kernel function is well
ARTICLE IN PRESS
1416 T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419

behaved, an alternating convergent series is even- and sine transforms). They have intentionally been
tually obtained, to which some sort of convergence kept separate from the remaining code, so that they
improvement scheme (Itoh, 1989; Chave, 1983) can can be used in other applications. A separate driver
be applied to speed up the rate of convergence and program for a vertical magnetic dipole configura-
minimize computation time. Several methods have tion over a homogeneous half-space is included with
been suggested but the most widely used in CR1Dmod to show the efficiency and precision of
geophysical literature is the Euler method. the FHT versus the NHT routine. This program
The NHT routine of CR1Dmod uses Matlabs also serves to demonstrate the use of FHT and
built-in adaptive Simpsons quadrature rule NHT routines for other applications.
(QUAD) to evaluate the first few segments. The
remaining segments are evaluated using the built-in 5. Spline interpolation applied to dipole integrations
adaptive Lobatto quadrature rule (QUADL). The
Euler method of convergence improvement, in the In order to increase the speed of computations in
form presented by Press et al. (1986), is applied to CR1Dmod, spline interpolation schemes have been
the series. The integration is terminated when the implemented for two situations.
last term added to the series is less than a tolerance EM calculations of certain wire configurations
specified by the user, or when a predefined number comprise integration over virtual electrical dipoles
of segments have been calculated. along the wires. Such integrations often result in a
The routine uses precalculated locations of the very large number of evaluations of the field
first 300 zeros of J n (n ¼ 0; 1). If more than 300 equations. In CR1Dmod, the user can choose to
segments are required for the calculations, further use cubic spline interpolation to significantly reduce
zeros are approximated by adding multiples of p. the number of calculations.
However, usually 300 segments are more than The FHT method offers an easy and quick way of
adequate to obtain the desired accuracy. calculating transforms of multiple r distances.
However, the calculated distances are logarithmi-
4.2. Fast Hankel transform filters cally spaced at the sampling interval of the filter
coefficients used. For maximum flexibility, but at
The main disadvantage of the numerical integration the cost of computation speed, we have chosen to let
scheme is the time of computation. Many evaluations the user set the number of evaluations per decade of
of the kernel function are needed in order to obtain a distance, and calculate these individually.
reasonable precision. The development of the fast The interval of r distances ½R1 ; R2  on which the
Hankel transform technique (Gosh, 1971; Johansen spline is based, is determined automatically based
and Sørensen, 1979; Anderson, 1979; Christensen, on the configuration, such that R1 is the second
1990), based on digital filtering theory, has resulted in largest regular distance smaller than the smallest
a much faster way to calculate Hankel transforms. By distance needed for the calculation. Similarly, R2 is
choosing appropriate substitutions, the Hankel trans- the second smallest regular distance larger than the
form can be converted to a convolution integral, greatest distance needed.
which is evaluated in the Fourier domain as a simple For the inverse Fourier transform (implemented
multiplication of a set of precalculated filter coeffi- as a sine transform) used in the time-domain
cients and calculated values of the kernel function. calculations, CR1Dmod allows for cubic spline
This approach result in a great increase in speed, interpolation of the frequency domain response in
but also puts a restraint on the maximum accuracy a similar way.
obtainable with a given set of filter coefficients. As the frequency interval needed for the trans-
The function implemented in CR1Dmod is based form is not known a priori, CR1Dmod lets the user
on a Fortran routine originally written by N.B. specify the limits in terms of an induction number
Christensen (pers. comm., 2004) and the filter values range. The induction number is taken to be
of Christensen (1990). qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B ¼ 12om0 s1  a, where s1 is the DC resistivity of
4.3. Comments on the transform routines the first layer and a the radius of the transmitter
loop. After the calculations have completed, the
CR1Dmod contains independent routines to evaluations of the frequency domain expressions
calculate NHT and FHT (also capable of cosine can be plotted together with the cubic spline
ARTICLE IN PRESS
T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419 1417

evaluations, to check that the specified range and Table 1


sampling interval were appropriate. Configurations available in CR1Dmod

Configuration Calc. type


6. Program structure
DC FD TD
CR1Dmod consists of two main windows, one Wenner x
that controls the model parameters and configura- Schlumberger x
tion of the measurement, and the other that controls Dipole–dipole x x (o)
the calculation specific parameters. GSA x x (o)
The left-hand side of the model window features TEM central loop x x
Sq. Tx loop + m. dip. (x) (x)
an interactive plot of the half-space model (Fig. 2). HCP FDEM (HLEM) x
By clicking the mouse in the plot, the user can insert
layer boundaries and drag these to the desired (x): method is implemented in modelling routine but not in GUI.
position. In the right-hand side of the window, the (o): method is planned for a future version.
lower section gives input fields for the layer
parameters, including Cole–Cole model parameters,
relative permittivity (r ¼ =0 ) and magnetic sus- input fields. The topview window also lets the user
ceptibility (w ¼ m=m0  1), and allows for adding add and move segments of the receiver and
and deleting layers in the model. The upper right- transmitter wires. Thus, the influence caused by
hand section of the window gives control of the the placement of the wires can be taken into account
measurement configurations. Table 1 lists the in the forward modelling of frequency domain
available configurations and the type of calculation responses. The user should take care when using
available for each configuration. this option that Tx and Rx wires do not cross
The general surface array (GSA) allows for themselves or each other.
arbitrary location of receiver and transmitter All grounded wire configurations, except the
electrodes on the surface of the layered half-space. GSA, allows for multiple input of the configuration
A special window is called up by pressing the parameters such as A-spacing and N-spacing.
‘‘topview’’ button. This window allows the user to Combined with the option to spline interpolate the
place the electrodes, either by dragging them with wire integrations, this allows for fast calculation of
the mouse, or by entering new coordinates in the the response of many configurations in one batch.
In the calculation window (Fig. 3), parameters
specific to the type of calculation can be adjusted
prior to calling the computational routines. For
configurations allowing a choice (Table 1), the
domain of calculation is selected (DC, FD, TD) as
well as the ‘‘full’’ or ‘‘quasi-static’’ modes. In full
mode, CR1Dmod selects either the full solution or
the non-magnetic first layer solution depending on
the magnetic susceptibility specified for the first
layer. In quasi-static mode, the program assumes
both non-magnetic and quasi-static approximations
regardless of the specified susceptibility.
Most input fields in this window accept any
regular Matlab expression. For instance, the ex-
pression logspace(-6,-3,30) in the Times input
field will result in 30 logarithmically spaced times
from 1 ms to 1 ms.
The forward modelling routines are coded in
separate functions:
Fig. 2. Main model window of CR1Dmod features a graphical
representation of the model, in which the user can add or delete
layers and move boundaries. Right-hand side of the window gives  emgsafwd.m calculates frequency domain re-
control of the configuration and layer parameters of the model. sponse of the grounded wire configurations.
ARTICLE IN PRESS
1418 T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419

perhaps to interpolate between data points of


laboratory CR-measurements.

7. Conclusion

We have developed a Matlab application,


CR1Dmod, capable of forward modelling of the
response of several commonly used electrical and
electromagnetic methods in a 1D environment. We
have obtained results in agreement with other
authors (Kauahikaua and Anderson, 1979; Ander-
son, 1981; Christensen and Auken, 1992; Auken
et al., 2001), but the time of computation is not as
efficient as other routines. Nevertheless, our code
handles CRs defined by a Cole–Cole resistivity
dispersion model, and offers solutions based on the
full EM-equations as well as the quasi-static
approximation. As a result, it is also possible to
Fig. 3. Calculation window of CR1Dmod gives user control over
calculation domain, transform types, tolerances and spline model effects due to changes in the magnetic
interpolations. permeability and permittivity of subsurface layers,
and a better understanding of the physical processes
can be obtained.
 temmfwd.m calculates frequency and time-do-
The graphical user-interface of CR1Dmod gives
main responses of the central loop configuration.
an easy way to build and modify layered models,
It also has the capability of calculating the
and offers intimate control of calculation para-
response of a square transmitter loop and a
meters. The Hankel transforms can be calculated
magnetic dipole receiver located at the center, but
either as numerical integrations between the zeros of
this option has not been implemented in the GUI
the Bessel function or by means of the fast Hankel
yet.
transform technique. The forward modelling rou-
 fdemfwd.m calculates frequency domain response
tines and the Hankel transform functions are
of the HCP loop–loop configuration.
separate units that can be incorporated into other
 dcgsafwd.m calculates DC response of grounded
applications.
wire configurations.
With Matlab being the preferred language of
numerical computing for geophysical applications
These functions setup the needed parameters,
at many universities, we hope that this freely
including pointers to the kernel functions, and pass
distributed code will be found useful for both
this information to either the NHT or FHT
research and educational purposes.
routines, which perform the actual transforms.
After the calculations have ended, the results are
saved to a binary Matlab file along with the model, Acknowledgement
configuration and calculation parameters. Finally,
the calculated response is plotted to the screen. In We wish to thank Mr. Niels Bøie Christensen for
the plots, the impedances of grounded wires are making his Hankel transform codes available to us.
multiplied by a geometric factor to yield the
apparent resistivities.
The Cole–Cole resistivity dispersion model used References
by CR1Dmod may not be applicable to all
problems. Therefore, the computation of the com- Anderson, W.L., 1974. Electromagnetic fields about a finite
electric wire source. USGS-GD 74-041, U.S. Geological
plex conductivities used by the forward modelling
Survey, 205pp.
routines has been implemented in a separate Anderson, W.L., 1979. Computer program: numerical integra-
function (Z_cr.m). The user is free to change this tion of related Hankel transforms of orders 0 and 1 by
function to accommodate different models or adaptive digital filtering. Geophysics 44 (7), 1287–1305.
ARTICLE IN PRESS
T. Ingeman-Nielsen, F. Baumgartner / Computers & Geosciences 32 (2006) 1411–1419 1419

Anderson, W.L., 1981. Calculation of transient soundings for a Itoh, T. (Ed.), 1989. Numerical Techniques for Microwave and
central induction loop system (program TCILOOP). Open- Millimeter-Wave Passive Structures. Wiley, New York,
File Report 81-1309, U.S. Geological Survey, 80pp. 720pp.
Auken, E., Breiner, M., Nebel, L., Pellerin, L., Thomsen, P., Johansen, H.K., Sørensen, K., 1979. Fast Hankel transforms.
Sørensen, K.I., 2001. EMMA—electromagnetic modelling Geophysical Prospecting 27, 876–901.
and analysis. In: EEGS Birmingham Proceedings. EEGS, Kauahikaua, J., Anderson, W.L., 1979. Programs EMCUPL and
Birmingham, UK, pp. 114–115. SCHCOPL: computation of electromagnetic coupling on a
Chave, A.D., 1983. Numerical integration of related Hankel layered halfspace with complex conductivities. Open-File
transforms by quadrature and continued fractions expan- Report 79-1430, U.S. Geological Survey, 91pp.
sions. Geophysics 48 (12), 1671–1686. Koefoed, O., 1979. Geosounding Principles, 1. Methods in
Christensen, N.B., 1990. Optimized fast Hankel transform filters. Geochemistry and Geophysics. vol. 14A, Elsevier, Amster-
Geophysical Prospecting 38 (5), 545–568. dam, 276pp.
Christensen, N.B., Auken, E., 1992. SELMA - Simultaneous Marshall, D.J., Madden, T.R., 1959. Induced polarization, a
Electromagnetic Layered Modelling and Analysis. In: Jack- study of its causes. Geophysics 24 (4), 790–816.
obsen, B.H. (Ed.), Proceedings of Interdisplinary Inversion Pelton, S.H., Ward, S.H., Hallof, P.G., Sill, W.R., Nelson, P.H.,
Workshop 1 Aarhus 1992, Methodology and Applications 1978. Mineral discrimination and removal of inductive
within Geophysics, Astronomy, and Geodesy, Geoskrifter 41, coupling with multifrequency IP. Geophysics 43, 588–609.
pp. 49–56. Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T.,
Cole, K.S., Cole, R.H., 1941. Dispersion and absorption in 1986. Numerical Recipes, The Art of Numerical Computing.
dielectrics, I. Alternating current fields. Journal of Chemical Cambridge University Press, Cambridge 818pp.
Physics 9, 341–353. Ryu, J., Morrison, F., Ward, S.H., 1970. Electromagnetic fields
Descloitres, M., Guérin, R., Albouy, Y., Tabbagh, A., Ritz, M., about a loop source of current. Geophysics 35 (5), 862–896.
2000. Improvement in TDEM sounding interpretation in Sunde, E.D., 1968. Earth Conduction Effects in Transmission
presence of induced polarization. A case study in resistive Systems. Dover, New York 370pp.
rocks of the Fogo volcano, Cape Verde Islands. Journal of Vanhala, H., Soininen, H., Kukkonen, I., 1992. Detecting
Applied Geophysics 45, 1–18. organic chemical contaminants by spectral-induced polariza-
Dey, A., Ward, S.H., 1970. Inductive sounding of a layered earth tion method in glacial till environment. Geophysics 57 (8),
with a horizontal magnetic dipole. Geophysics 35 (4), 1014–1017.
660–703. Wait, J.R., 1952. Current-carrying wire loops in a simple
Frischknecht, F.C., 1967. Fields about an oscillating magnetic inhomogeneous region. Journal of Applied Physics 23 (4),
dipole over a two-layer earth, and application to ground and 497–498.
airborne electromagnetic surveys. Quarterly of the Colorado Ward, S.H., Hohman, G.W., 1988. Electromagnetic Theory for
School of Mines 62 (1), 326. Geophysical Applications, vol. 1. Society of Exploration
Gosh, D.P., 1971. The application of linear filter theory to the Geophysicists, pp. 131–311 (Chapter 4).
direct interpretation of geoelectrical resistivity sounding Wong, J., 1979. An electrochemical model of the induced-
measurements. Geophysical Prospecting 19 (2), 192–217. polarization phenomenon in disseminated sulfide ores.
Hohmann, G.W., Kintzinger, P.R., Van Voorhis, G.D., Ward, Geophysics 44 (7), 1245–1265.
S.H., 1970. Evaluation of the measurement of induced Zonge, K., 1972. Electrical properties of rocks as applied to
electrical polarization with an inductive system. Geophysics geophysical prospecting. Ph.D. Thesis, The University of
35 (5), 901–915. Arizona, 156pp.