GEOPHYSICS, VOL. 88, NO. 3 (MAY-JUNE 2023); P. F39–F49, 20 FIGS., 1 TABLE.

10.1190/GEO2022-0058.1

A computational package to compute the electrical resistivity tomography

response for regular bodies immersed in a homogeneous half-space

Hector Octavio Augusto Hernández Contreras1 and E. Leticia Flores Márquez2

near-surface geophysical method in fields such as geotechnics, geo-

ABSTRACT logic structural purposes, bedrock and sand channel localization to
monitor pollution plumes (Benson, 1995; Benson et al., 1997), and
A 3D forward algorithm is presented to model electrical in countless hydrogeologic studies (Clement and Moreau, 2016).
resistivity tomography (ERT) measurements with the capabil- The usual way to apply the ERT is to use conventional field survey
ity to use different electrode array configurations. We have methods, and its interpretation is achieved with commercial soft-
implemented the algorithm as an open-source package with ware, which uses numerical inversion processes to convert field data
the assumption that the medium can be modeled as a homo- to a model. However, despite their technical progress and the multi-
geneous half-space with immersed bodies of different resis- ple advantages of this technique, the interpretation of the images of
tivities with regular geometry. This algorithm incorporates a the apparent electrical resistivity of the subsurface is not simple,
novel numerical approximation by using the volume integral even when inversion processes are used. The experience that an op-
method, with the hypothesis of weak scattering, where the erator needs to interpret these methods is very important. In addition
secondary currents generated in the bodies due to the main to proficiency in software utilities, the expert technician must use
field are not considered. The algorithm also includes a his or her good judgment in assigning the proper resistivity to geo-
new method called cubature to calculate volume integrals logic marker formations, which agree with the geologic information
and the ERT response. In addition, the code has a module of the study area. In this paper, we introduce the mathematical back-
to display the results as color resistivity profiles. This method ground for direct 3D modeling of regular bodies buried in a homo-
avoids the limitation of the complex geometry of the medium geneous half-space and the volume integral model of the electric
or even of the immersed bodies. The method is useful for response (VIMER) package to compute their ERT response, with
planning a field survey, exploring the use of a specific elec- the intention that even undergraduate earth science students can per-
trode array, or implementing new arrays. The package, called form their exercises in this technique, without the need for expen-
the volume integral model of the electric response, is compu- sive specialized software. This package also can contribute to the
tationally efficient and can be easily ported to any operating synthetic testing of electrode arrays with unconventional configu-
system that supports MATLAB. rations, given the significant advancements that these techniques
have had in the past 40 years, foremost in automatically generating
optimized electrode array configurations.
The equations that allow modeling the ERT techniques belong to
INTRODUCTION the partial differential equations (PDEs), and there exist a large num-
ber of numerical techniques to solve them, in particular, the finite-
Electrical resistivity tomography (ERT) has become very popular difference methods. This method has been helpful, mainly in 2D
in innumerable applications of earth science studies (Barker, 1991; problems, as in the works of Mufti (1976), Dey and Morrison
Tsourlos and Ogilvy, 1999; Loke, 2004; Binley et al., 2005, pp. (1979), Spitzer (1995), and Loke and Barker (1996). Another way
129–156; Chambers et al., 2013; Loke et al., 2013; Hörning to obtain a solution is with the finite-element method; good examples
et al., 2020; Troiano and Di Giuseppe, 2020; Li and Yang, are given in Coggon (1971), Fox et al. (1980), Pridmore et al. (1980),
2021). For instance, ERT has been elevated to an obligatory, Sasaki (1994), Tsourlos and Ogilvy (1999), Li and Spitzer (2002,

Peer-reviewed code related to this article can be found at this paper’s Supplemental Materials link.
Manuscript received by the Editor 13 April 2022; revised manuscript received 6 November 2022; published ahead of production 7 February 2023; published
online 13 April 2023.
1
Universidad Nacional Autónoma de México, Posgrado en Ciencias de la Tierra, Coyoacán, México. E-mail: [email protected].
2
Universidad Nacional Autónoma de México, Instituto de Geofísica, Coyoacán, México. E-mail: [email protected] (corresponding author).
© 2023 Society of Exploration Geophysicists. All rights reserved.

F39

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 F40 Hernández Contreras and Flores Márquez

2005), Marescot et al. (2008), Pidlisecky and Knight (2008), Ren and where ∇ ¼ ðð∂=∂xÞ; ð∂=∂yÞ; ð∂=∂zÞÞ; δðr̄Þ is the impulse function;
Tang (2010), Karaoulis et al. (2013), Blanchy et al. (2020), and Loke I is the electrical current; and r̄ denotes the distance between the
et al. (2022). The PDE also can be solved by using the surface/volume source and the observed point. The boundary conditions are ex-
integral method for the direct current (DC) problem, which was widely pressed as follows:
used in the 1970s, with the pioneering works of Hohmann (1971,
1975), Dey and Morrison (1979), Zhdanov and Keller (1994), and ∂Uðr̄Þ
Li and Oldenburg (1996) and integral equations such as those in ¼ 0; z ¼ 0; Uðr̄Þ ¼ 0 when r̄ → ∞ and z ≥ 0: (2)
∂z
the studies of Snyder (1976), Eskola (1992), Pérez-Flores (1995),
Pérez-Flores et al. (2001), León-Sánchez (2004), Pidlisecky et al.
There is a solution for the potential UðrÞ, using Green’s functions
(2007), Pidlisecky and Knight (2008), and Huo et al. (2014).
Gðr̄; r̄ 0 Þ, for a region defined by a semisphere in terms of a volu-
This paper presents the computational analytical procedure and
metric integral:
an open MATLAB code to compute the ERT response using the
integro-differenciate algorithm proposed by Flores-Márquez et al. Z
(2015), which is based on previous works (Hohmann, 1971, 1975; Uðr̄Þ ¼ − ρðr̄ 0 ÞJ̄ðr̄ 0 Þ · ∇ 0 Gðr̄; r̄ 0 ÞdV; (3)
Gómez-Treviño, 1987; Kauffman, 1992; Pérez-Flores et al., 2001;
V
Tejero-Andrade, 2002; León-Sánchez, 2004; Hernández-Contreras,
2018) addressing the same topic. The method is based on the in-
where ρðr̄ 0 Þ is the resistivity and J̄ðr̄ 0 Þ is the electric current density.
tegral solution of the forward DC geoelectric response for 3D target
Equation 3 is implemented by Flores-Márquez et al. (2015) as the
bodies in a half-space based on Green’s functions. The hetero-
geneity of the half-space and the complex geometry of the bodies VIM method to obtain the electric resistivity response of the half-
inside it have been important limitations to the technique of the in- space. To use this equation, it must be assumed that the electric
tegral equation and in the use of the finite-difference or finite- current density vector J̄ðr̄Þ is known, and the secondary currents
element methods. The disadvantage of those methods is the size that compose it are negligible, denoting J̄ s as the secondary current
of the linear equation systems that consume large computational and J̄p as the primary current, which means J̄s ðr̄Þ ≪ J̄p ðr̄Þ. This
resources or even the nonlinear equation systems. assumption is the weak scattering (Eskola, 1992, pp. 23–26) formu-
Due to the weak scattering assumption proposed by Eskola (1992, lation.
p. 23), the integral equation becomes a viable option for direct mod- Here, J̄ p is expressed as follows:
eling, and in addition, the use of new techniques to solve volume " #
integrals, even complex geometries of bodies, is no longer a limita- I ðx − xs Þ^i þ ðy − ys Þ^j þ ðzÞk^
tion. Moreover, Flores-Márquez et al. (2015) propose a numerical J̄ p ðr̄Þ ¼ : (4)
2π ½ðx − xs Þ2 þ ðy − ys Þ2 þ ðzÞ2 3=2
analytical implementation of this method. The problem is solved with
two kinds of integral equations by introducing the resistivity contrast
between the bodies and the homogeneous half-space and considering By substituting the gradient of Green’s function (Kauffman, 1992;
the concepts of the additive potential sources for immersed bodies Tejero-Andrade, 2002) in equation 3 (refer to equations 8 and 10 in
and surface charge density, which yield two types of solutions, Flores-Márquez et al., 2015), it becomes
the surface integral method and volume integral method (VIM).
There remains the problem of the computational implementation
Z " #
of both; however, the proposal has the advantage that it is not neces- I ðx 0 −xs Þðx−x 0 Þþðy 0 −ys Þðy−y 0 Þ−z 02
Uðr̄Þ¼− ρðx 0 ;y 0 ;z 0 Þ dv;
4π 2 ½ðx 0 −xs Þ2 þðy 0 −ys Þ2 þz 02  ½ðx−x 0 Þ2 þðy−y 0 Þ2 þz 02 
3 3
sary to discretize the medium because the model is based on the elec- 2 2
v
trical response of the immersed bodies.
(5)
In this work, we combine the VIM implementation of Flores-
Márquez et al. (2015) with a relatively new method to compute vol-
ume integrals called cubature by Jayan and Nagaraja (2015). This where I is the electrical current; ðxs ; ys Þ are the coordinates of the
method has the capacity to eliminate the requisite use of the com- source within a fixed reference system ðx; y; zÞ, which is called the
plex geometries of bodies and offers better results compared with global coordinate system (GCS); and ðx 0 ; y 0 ; z 0 Þ are within a local
traditional routines to calculate multidimensional integrals and al- coordinate system (LCS) with an origin at the middle of the
lows for easier application in any programming language. To make prismatic body and are the variables in the integral of equation 4.
it easier and available, we have implemented the algorithm as an The transformation (Figure 1) between both coordinate systems
open-source MATLAB package. (refer to Figure 2 in Flores-Márquez et al., 2015) is performed
by using the auxiliary vectors ra ¼ ðxa ; ya ; za Þ; r 0 ¼ ðx 0 ; y 0 ; z 0 Þ,
and r ¼ ðx; y; zÞ and by assuming an arbitrary point
THEORETICAL RESPONSE OF ELECTRICAL Pðxp ; yp ; zp Þ in the space, whose vector position is ra . Equation 5
RESISTIVITY OF A HETEROGENEITY can be used to obtain the potential that is produced by a point source
anywhere on the z = 0 plane.
For a nonhomogeneous, 3D half-space with resistivity ρðr̄Þ, the Resistivity contrasts ðΔρÞ between the isolated bodies and
electric potential (U) produced by a source point on the surface for the half-space are considered under the superposition principle
z = 0 is expressed as (Flores-Márquez et al., 2015) for an array with two current electrodes (A, B) and two potential

 electrodes (M, N) (refer to the “Different types of electrode arrays”
∇Uðr̄Þ
∇· ¼ −Iδðr̄Þ; z ≥ 0; (1) section). Then, the following expression is obtained (Flores-
ρðr̄Þ Márquez et al., 2015) for the four electrode distances involved:

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 VIMER program to compute ERT response F41

Z " " # NUMERICAL INTEGRATION (CUBATURE)

jIj ðx 0 − xcs
0
Þðxp0 − x 0 Þ þ ðy 0 − ycs
0
Þðyp0 − y 0 Þ − ðz 0 þ za Þ2
ΔUM ¼ 2 Δρ −
N
4π 0 0 0 0 0 0 0 0 0
½ðx − xcs Þ þ ðy − ycs Þ þ ðz þ za Þ2 ½ðxp − x Þ þ ðyp − y Þ þ ðz þ za Þ 2
v
2 2 3
2 2 0 2 3

AM For this algorithm, we used the “cubature” proposed by Jayan and

" #
ðx 0 − xcs
0
Þðxp0 − x 0 Þ þ ðy 0 − ycs
0
Þðyp0 − y 0 Þ − ðz 0 þ za Þ2 Nagaraja (2015), who present a general formula that allows the evalu-
þ
½ðx 0 − xcs
0 2
Þ þ ðy 0 − ycs
0 2
Þ þ ðz 0 þ za Þ2 ½ðxp0 − x 0 Þ2 þ ðyp0 − y 0 Þ2 þ ðz 0 þ za Þ2 2
3 3

BM
ation of triple integrals over a variety of regions, such as cubes, tetra-
" #
ðx − xcs00
Þðxp0 − x 0 Þ þ ðy 0 − ycs
0
Þðyp0 − y 0 Þ − ðz 0 þ za Þ2 hedrons, prisms, pyramids, or any other region bounded by straight or
þ
½ðx 0 − xcs
0 2
Þ þ ðy 0 − ycs
0 2
Þ þ ðz 0 þ za Þ2 ½ðxp0 − x 0 Þ2 þ ðyp0 − y 0 Þ2 þ ðz 0
3
þ za Þ2 2
3

AN
curved boundaries. In addition, the 1D quadrature points published
" # #
ðx 0 − xcs
0
Þðx0p − x 0 Þ þ ðy 0 − ycs
0
Þðyp0 − y 0 Þ − ðz 0 þ za Þ2 by Ma et al. (1993) are used; they assign a group of nodes and
− 0 0 2 0 0 2 0
dV; (6)
½ðx − xcs Þ þ ðy − ycs Þ þ ðz þ za Þ ½ðxp0 0 2
−x Þ þ ðyp0 0 2 0
− y Þ þ ðz þ za Þ  weights as the best approximation to a specific set of functions.
3 2 3
2 2
BN

The generic integral of a function fðx; y; zÞ to be evaluated is

where AM = MN = NB is the distance between the electrodes and expressed as
ΔU is the potential difference (refer to Figure 2). Z Z Z
b g2 ðxÞ h2 ðx;yÞ
Note that the integral never cancels its denominator, and it is not I¼ fðx; y; zÞdzdydx: (8)
singular; the range of integration is defined, and the kernel never a g1 ðxÞ h1 ðx;yÞ
reaches a singularity within this range.
When the source coordinates are ðxcs 0 ; y 0 Þ on the surface (z = 0) The integration domain
cs
for the current electrode and the coordinates on the surface for the
potential electrode are ðxp0 ; yp0 Þ, then the apparent resistivity ρa is Ω¼fðx;y;zÞja≤x≤b;g1 ðxÞ≤y≤g2 ðxÞ;h1 ðx;yÞ≤z≤h2 ðx;yÞg
expressed as (9)
 
2π ΔUM is transformed into a cube from zero to one in a space defined by
ρa ¼ ρ m þ N
; (7) ξ; η, and γ as follows:
1
AM − 1
BM − 1
AN þ 1
BN
jIj
C ¼ fðξ; η; γÞj0 ≤ ξ ≤ 1; 0 ≤ η ≤ 1; 0 ≤ γ ≤ 1g: (10)
where ρm is the resistivity of the half-space.
This VIM implementation combined with a relatively new The coordinate transformation ðx; y; zÞ → ðξ; η; γÞ is clearly de-
method to evaluate the volume integral of equation 6, called cuba- scribed in Jayan and Nagaraja (2015). Then, the resulting Jacobian
ture, is implemented in the algorithm proposed here. With the use of is a function of both coordinate systems and is positive.
these techniques, the two main problems of direct modeling can be Jayan and Nagaraja (2015) apply a Lagrange polynomial base for
solved. First, the use of an integral equation avoids the problem of simple cases, such as a pyramid geometry, and quoting from them
having a massive system of equations, and second, the complex “Any of the Gaussian Quadrature nodes and weights can be substi-
geometry of the bodies in the half-space is no longer necessary. tuted in the formula.” In our case for equation 6, we use 10 nodes
Of course, the evaluation of a volume integral for a body with a and 10 weights of the polynomial products of fractional exponent
complex geometry is still quite difficult because, as will be ex- functions published by Ma et al. (1993). We have conducted some
plained in the next section, the boundaries of the body must be de- tests by using 5, 10, and 15 nodes and their respective weights, con-
fined by a function, but this is possible. cluding that better results are obtained for 10 nodes.
The limits of integration are functions of the transformation. It is
possible to evaluate that integral over a wide range of regions, even
with curved surfaces. It is only necessary to analytically define the
function that describes the shape of the region projected onto the
corresponding axis. In our case, we use the tetrahedron case, which
is equivalent to transforming the region into a cube, and after ap-
plying the formula for the Gaussian quadrature, the integral in equa-
tion 8 is rewritten with the next set of equations (Jayan and
Nagaraja, 2015):

Figure 2. The quadripole Wenner array, distance AM = MN = NB,

Figure 1. The GCS and LCS. and investigation depth (n = data level).

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 F42 Hernández Contreras and Flores Márquez

a ¼ 0; b ¼ a; g1 ¼ 0; g2 ¼ a − x; DIFFERENT TYPES OF ELECTRODE ARRAYS

h1 ¼ 0; h2 ¼ a − x − y: (11) To proceed with the numerical implementation, it is necessary to
decide what type of electrode array should be simulated. This array
is formed mainly by electrodes from which the electric current flows
Then, the region is described as into the half-space and another electrode over which the potential is
measured.
T ¼ fðx; y; zÞj0 ≤ x ≤ a;0 ≤ y ≤ a − x; 0 ≤ z ≤ a − x − yg: In the literature, there are numerous possible configurations
(12) of electric and potential electrodes; in most cases, four electrodes
(Figure 2) are used. Usually, A and B denote the current electrodes
and M and N denote those of the potential (Orellana Silva, 1982;
The integral becomes Telford et al., 1990; Zhdanov and Keller, 1994). Therefore, the
quadripole array is the most prevalent (Loke et al., 2010, 2013),
Z aZ a−x
Z a−x−y X
N3 and the assignment of the depth of investigation is an important item
I¼ fðx;y;zÞdzdydx≈ cm fðxm ;ym ;zm Þ; in all cases that is linked to the specific electrode array and is re-
0 0 0 m¼1 ferred to as the data level, considering n = 1 as the first data level. In
turn, each level is interpreted as a depth in terms of the distance
(13)
between the two electrodes.
Some authors, such as Szalai and Szarka (2008), have performed
where
the tasks of documenting and classifying different types of arrays,
considering parameters such as superposition, focusing, and colli-
cm ¼ wi1 wj2 wk3 ðb − aÞðd − cÞðf − eÞ; (14) nearity, and have documented approximately 100 configurations
that have been deployed in a variety of exploration studies. Com-
monly, the type of array is classified as conventional, nonconven-
tional, and optimal sets of array configurations (Loke et al., 2013;
xm ¼ aξi; (15) Abdullah et al., 2019); however, this discussion is beyond the scope
of this paper.
The VIMER package simulates four of the most common arrays
in geophysical exploration: Wenner-Schlumberger, dipole-dipole,
ym ¼ að1 − ξi Þηj ; (16) pole-dipole, and pole-pole (Orellana Silva, 1982; Telford et al.,
1990). However, given that the coded algorithm is divided into
modules, any other conceivable array can be added.

zm ¼ að1 − ξi − ηj þ ξi ηj Þγ k : (17)
NUMERICAL IMPLEMENTATION
The nodes are ξi ; ηj , and γ k , and the respective weights are wi1 ; wj2 ,
Our algorithm is called the VIMER (Hernández-Contreras,
and wk3 .
2023); it models the response of the underground in an ERT
scheme. The program was developed in the
MATLAB® environment and has a visualization
module that presents the results in a color pseu-
dosection format.
The program has four main parts: (1) initializa-
tion (body characteristics, type of array, number,
and electrode orientation), a box window that
displays the data input; (2) coordinate assign-
ment (determined by the selected array);
(3) (equation 6) equation solution; and (4) graphi-
cal display of the results. Figure 3 shows a flow-
chart indicating the logical sequence of the
program. Each module is subsequently de-
scribed.

Initialization
A window appears to input several parameters,
such as the type of array and orientation, as
shown in Figure 4, number of electrodes and
the separation between them, and the resistivity
Figure 3. Flow diagram showing the logical sequence of the VIMER package. Modified of the half-space and the number of bodies in it.
from Hernández-Contreras (2018). Depending on this number of bodies, a panel

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 VIMER program to compute ERT response F43

with the same number of tabs is displayed next; in this tab, the po- Visualization
sition and characteristics of each body, including their resistivity,
are entered. A couple of functions are executed to prepare the matrix for
the MATLAB’s contourf command. To make this possible, the
“image dilation” method is used (Gonzalez and Woods, 1992,
Coordinates assignment
pp. 519–523), where every pixel of the image is reproduced or
This method establishes two coordinate systems, GCS and LCS, dilated toward the right, down, and right-down direction, which
which are defined during the early stages of the program; but to is necessary because the obtained matrix of data is of the sparse
present the images in a more comprehensible way and to easily de- type, and each data entry is displayed as a pixel (Figure 5).
termine the position of the bodies by the user, the first electrode of The dilation does not replace a computed element in the model,
the chosen array on the surface must be considered as the origin of and the image is swapped from top to bottom and from left to right.
the coordinate system, and the algorithm will make the proper trans- Thus, if a point obtained from the dilation has the same position as
formations. the point calculated by the model, this last point overwrites the pre-
The coordinate system is shown in Figure 4. The x-axis increases vious point.
to the east, the y-axis increases to the north, and the z-axis increases
upward; the length is measured in m, the angle is measured in de-
SYNTHETIC EXAMPLES
grees with decimals (°), and resistivities are measured in ohm.m.
With the aim of illustrating the validity of the forward 3D VIMER
Solution of the equation modeling developed for this work, this section is devoted to showing
the response of this code to a series of synthetic data. To demonstrate
According to the type of array chosen, one of the four routes the capabilities of the VIMER package, concerning the types of ar-
shown in the flowchart is executed, and as needed, the potential rays and contrast resistivities, different conditions of immersed
difference among 2, 3, or 4 electrodes is calculated using the cu- bodies are presented. One hundred tests have been performed, but
bature on each attribution point. To determine the depth of the in- only five of them are presented. The electric responses obtained
vestigation (data level) on the z-axis, the depth of investigation Ze by the VIMER program for four synthetic tests, as well as plots
reported by Edwards (1977) is used. at the first depth level for quantitative assessment, are shown in color
The cubature function uses the integration limits determined by graphic displays. In addition, an analytic test is presented, where the
the body dimensions as parameters for integration. The integral is numerical response was compared with the analytical response of a
evaluated with 10 nodal points and their weight coefficients wi that two-layer model (Telford et al., 1990, pp. 540–542).
were obtained from the family of functions, “polynomial products
and logarithmic functions,” published by Ma et al. (1993). Example of a two-layer model
The cubature is calculated for each prism at each data point, and
their contribution is then added to the electric response. As the first test to verify the capabilities of the code, as has been
conducted by some authors (Gernez et al., 2020), we compared the
VIMER numerical response with an analytical solution from
Telford et al. (1990, pp. 540–541) for a two-layer model. The model
consists of a 100 m thick layer with a homogeneous resistivity of
100 ohm.m over a deeper infinitely thick second layer with a resis-
tivity of 300 ohm.m and uses a Wenner array configuration. The
analytic solution is obtained for the segment between one current
electrode (A) and one potential electrode (M), the distance AM
being denoted by the parameter L, as shown in Figure 6, where
a scheme of this model and the comparative plot between the ana-
lytical solution from Telford et al. (1990, p. 540, equation 8.38) and
the VIMER results are displayed. The numerical comparison be-
tween both solutions establishes a global difference of approxi-
mately 7%, which we consider a very good approximation of
the analytical solution by the VIMER algorithm. Because the
VIM method proposed by Flores-Márquez et al. (2015) uses

Figure 4. Scheme of coordinate systems and angle of the profile of Figure 5. Example of the dilation method growing a pixel toward
the electrodes in the VIMER code. the right, down, and diagonally.

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 F44 Hernández Contreras and Flores Márquez

Green’s functions to solve equation 1, the solution to the integral is of differences between the analytical response and the VIMER
dominated by the behavior of Green’s functions, which intrinsically model does not change depending on the resistivity values.
have a Dirac delta function. Then, the electrical response suddenly This comparison shows coincidences in the computed resistivity
increases due to the deepest layer as one approaches this level. It is values at all investigation depths or data levels and central assigna-
expected that the electrical response increases near the layer, which tion points. It is important to clarify that the VIM method was de-
is not the case for the analytical solution. The same behavior for the veloped to model the effects of 3D immersed bodies in a
analytical solution and VIMER solution is expected when the upper homogeneous half-space and that the analytical solution also is a
layer is more resistive than the deeper layer because the percentage direct method but assumes a 1D problem. Then, the second layer
is considered as a body with dimensions of 500 m × 500 m and a
thickness of 1000 m in the VIMER code; this assumption could
explain the differences between the models. We conclude that, in
general, the approximation with VIMER follows the same behavior
as the analytical solution and is sufficiently accurate.

Example of a resistive block (Wenner and pole-dipole

arrays)
The model used in the following two examples is shown in Fig-
ure 7; it consists of a 100 ohm.m cube buried in a half-space with a
resistivity value of 10 ohm.m. In both experiments, the array con-
sists of 21 electrodes with a spacing of 1 m. The first example has a
Figure 6. Two-layer model validation. (a) Scheme of two layer and Wenner alpha configuration, and the second example is a pole-di-
the array used; the electrode locations are represented by arrows at pole array.
the surface. (b) Comparison between the analytical solution (Telford
et al., 1990, p. 540) and numerical values of the VIMER code of the In this example, we observe an anomaly at a depth of 1 m in the
apparent resistivity ρa obtained with a Wenner array versus L (elec- model; this anomaly has a resistivity value of 16 ohm.m in the
trode spacing). VIMER model (see Figure 8). The shape of the anomaly shows
higher values between 5 and 15 m in the horizontal direction
and between 0.5 and 1.5 m in the depth.
Figure 9 shows a plot of the values at the second level; the largest
resistivity value is approximately 17 ohm.m at a horizontal distance
of approximately 10 m.
Figure 10 shows the same case for a pole-dipole array, where we
observe the largest resistivity values of approximately 20 ohm.m
and the peak value of this anomaly is at a depth of approximately
2 m. To show the second level of the ERT from the VIMER, we plot
the resistivities along this level, as shown in Figure 11, where the
maximum resistivity is between a horizontal distance of 10 m and a
horizontal distance of 11 m. The VIMER code is sensitive to the
asymmetry of the pole-dipole array.

Figure 7. Schematic representation of a cube buried in a homo- Example of a conductive block (dipole-dipole array)
geneous half-space (10 ohm.m) with a resistivity value of 100
ohm.m. The third synthetic experiment is a conductive block with the
same shape as that in Figure 6. However, in this case, the resistivity
values are reversed, and the half-space is given a value of 100
ohm.m and has a 10 ohm.m block inside (Figure 12). The electrode

Figure 8. The two-dimensional output of resistivity values obtained Figure 9. Graph of resistivity values against the distance for the
for the Wenner alpha configuration by using the VIMER code. second data level of the model (n = 2, investigation depth ∼ 1.04 m).

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 VIMER program to compute ERT response F45

array in this case is the same as the previous examples but in a in the middle of the bodies. Several tests were performed to show
dipole-dipole configuration. the capabilities of the VIMER code for the superposition of the
In this case, the VIMER output (Figure 13) is a resistivity bodies with different resistivities and locations, obtaining appropri-
anomaly of 86 ohm.m at a depth of approximately 1 m, with a lat- ate results in all cases.
eral extension from 8 to 13 m in the horizontal direction.
The graph of Figure 14 shows the minimum of the model at the CASE STUDY: SITE OF XOCHIMILCO, MEXICO
second level, which is approximately 85.5 ohm.m situated at the
center of the array. This quantitative assessment allows us to con- To demonstrate the capabilities of the VIMER to reproduce the
clude that, in all cases, the maximum (or minimum) resistivity anomalies that could be obtained in a real scenario, it was tested in
anomaly is properly situated in the middle of the body location. the Xochimilco study. This example was selected from published
studies because it satisfied the criteria that it has survey data and
Example of two blocks (Wenner array)
The algorithm also is tested for two blocks of different shapes;
these blocks have the same resistivity value of 50 ohm.m. These
bodies are immersed at different depths in a half-space with a con-
stant resistivity of 100 ohm.m. The Wenner array contains 48 elec-
trodes with a 2 m spacing between each electrode (Figure 15).
This example has the purpose of showing the principle of super-
position; its results are shown in Figure 16. We observe two minima
coincident to the location of the middle of each block.
Figure 17 shows that at the first level of the model, it indicates
two minima coincident with the location of the middle of each block
that are located at 32 and 64 m horizontal distance marks. Here,
again, it can be observed that, even in the case of the superposition Figure 12. Schematic representation for a cube with a 10 ohm.m
of bodies, the location of the minimum of the resistivity anomaly is resistivity buried in a homogeneous 100 ohm.m half-space.
correct at the middle of each body location. In the opposite case of
having two resistive bodies, the code also shows the minima located

Figure 13. Resistivity values calculated with the VIMER program

for a dipole-dipole configuration traversing the conductive block
model.
Figure 10. Resistivity values obtained for the pole-dipole configu-
ration by modeling with the VIMER program.

Figure 11. Graph of resistivity values of the second level of the Figure 14. Graph of resistivity values at the second level of the
model in the case of the pole-dipole configuration (n = 2, investi- model for the dipole-dipole configuration (n = 2, investigation depth
gation depth ∼ 1.04 m). ∼ 1.04 m).

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 F46 Hernández Contreras and Flores Márquez

a previous model that resulted from an independent numerical in- borehole: basalt at depth, basalt with sand and gravel, and a mixture
version process. We also considered that the results of the inversion of gravel and silty sands at the top.
were verified against a typical geologic depth section of the area. Considering Figure 18, we model three basalt blocks at the bottom
The Xochimilco study (Barrera, 2015) consists of an ERT section of the section, with resistivities of 2000 ohm.m as reported by Barrera
in the town of San Luis Tlaxialtemaco, Xochimilco, Mexico, and (2015), and three shallow blocks with resistivities of 50 and 88 ohm.m,
had the purpose of identifying a good location for an artificial re- all immersed in a half-space of 300 ohm.m (Figure 19). In this way, the
charge well. The EarthImager® commercial code was selected as the three main units are represented. The position and characteristics of
inversion software. these bodies (see Figure 19b) are provided in Table 1.
The results of the study are shown in Figure 18, which were veri- Figure 20 shows the results and a comparison of the model with
fied with mechanical borehole logging reported in the publication. the data acquired in the field using the Wenner-Schlumberger con-
Three main geologic units were identified because of this proof figuration; we observe the correspondence between the apparent re-
sistivities in the 2D section. It is important to emphasize that the
VIMER computes a direct model in 3D form and that the computed
apparent resistivities are displayed just as a section over the center-
line of the model.
Notably, the theoretical development used here had previously
been implemented in Fortran code (Flores-Márquez et al., 2015)

Figure 15. Schematic representation of the homogeneous half-

space (100 ohm.m) with two cubes buried in it with a resistivity
value of 50 ohm.m. Figure 18. Inverted resistivity section of a Wenner-Schumberger
traverse accomplished in the Town of San Luis Tlaxialtemaco,
Xochimilco, Mexico (Barrera, 2015). The lithology is shown on
the left.

Figure 16. Resistivity values calculated for a Wenner configuration

for two conductive blocks with the VIMER code.

Figure 19. Resistivity values profiles: (a) apparent measured resis-

tivity, (b) calculated resistivity values obtained from the inverted
Figure 17. Graph of resistivity values at the second level for the resistivity section, (c) resistivity model obtained by EarthImager®
Wenner configuration with two conductive blocks in a more resis- and published by Barrera (2015), and (d) the proposed bodies sat-
tive half-space (100 ohm.m, n = 2, investigation depth ∼ 2.08 m). isfying the previously inverted resistivity section.

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 VIMER program to compute ERT response F47

Table 1. Geometric characteristics of the modeled bodies in The VIM is able to reproduce the half-space response with one or
the Xochimilco case study. more heterogeneities represented by well delimited prisms.
Position (m) Dimensions (m) The synthetic test, against a two-layer (ρ1 < ρ2 ) model, shows that
the VIMER reproduces a good approximation of one analytical
Body x y Depth x y Height σ (Ωm) solution. Furthermore, in a real scenario, the VIMER is able to
1 32 0 21 32 32 16 2000 represent a shallow conductive layer using several bodies. However,
a weakness of the VIMER is that the resistivities reproduced
2 97 0 24 36 36 13 2000
by the VIMER code differ from those expected in the proposed
3 157 0 17 8 8 5 2000 bodies.
4 30 0 4.6 8 8 7 50 The forward models performed by the VIMER package can be
5 92 0 4.6 48 48 7 88 helpful when planning a field survey or justifying the use of a spe-
6 166 0 4 28 28 7 50 cific electrode array with a set number of electrodes or determining
optimum electrode distances when considering the expected struc-
tures. The VIMER also is very helpful for those starting their re-
search on the DC geoelectric methods.
The applicability of the VIMER package in the case of Xochi-
milco’s study shows that a scenario mimicking the recorded geo-
logic borehole logs can be reproduced even with the complexity of
several layers. More complex scenarios, such as displacements be-
tween prisms (faults) and dipping structures, can be added if
needed. At the beginning of this paper, we mentioned that the cu-
bature could be implemented not only for parallelograms but also
for bodies with curved faces, and this capability could be useful
when representing a very complex half-space. Nevertheless, we
demonstrate that this capability was not necessary for Xochimil-
co’s study because good results could be achieved by modeling
Figure 20. Resistivity values along a profile obtained with the Wen- several regular bodies to represent the underground structures.
ner array configuration: (a) the model obtained by Barrera (2015) The more complex the half-space is, the larger the number of
modeling with the EarthImager and (b) modeling with the VIMER blocks needed.
code.
The VIMER code has a module to display the results as resistivity
profiles in color, which use the dilation technique to fill in the re-
with other types of quadrature methods for integration, and their sulting sparse matrix needed to display the bidimensional sections.
results also were compared with electrical responses obtained by Although it is very useful, anyone implementing this algorithm can
other algorithms based on different theoretical approaches (e.g., replace this technique module with another method of interpolation.
Pridmore et al., 1980; Tsourlos and Ogilvy, 1999; Pérez-Flores The use of windows to enter the parameters needed by the algo-
et al., 2001), reproducing the results within a good approximation rithm is very convenient for the user; in this way, the script is not
for all cases. This finding gives confidence about the theoretical altered every time a new model is calculated. The display of the
development. The actual numerical approach of the VIMER pack- resulting profiles avoids the requirement of additional software
age not only reproduces the old examples but also replicates new to display the data, allowing the running of several models in an
cases (Karaoulis et al., 2013; Blanchy et al., 2020; Gernez et al., easy way.
2020) and has the capability to display the profiles in color. Sim- The conducted tests with the synthetic data indicate that the
ilarly, the VIMER is able to reproduce the analytical response for VIMER package produces good results compared with the pub-
stratified layers, particularly two and three layers (Pridmore et al., lished results for similar problems obtained by other algorithms,
1980; Telford et al., 1990; Gernez et al., 2020), even though the even those produced by finite-element methods.
theoretical development is performed for the closed finite bodies. The runtime performance modeling of the synthetic example of
In addition, the VIMER package ensures an easy implementation; Figure 7 is a total time of 1.451 s. After the graphical user interface
it has the possibility of being portable on different operating sys- (GUI), the most expensive function is the cubatura (cubature), with
tems where the MATLAB environment runs. 252 calls and a self-time of 0.220 s. For the synthetic example of
Figure 15, the runtime is 3.890 s and the cubatura has 2880 calls and
a self-time of 2.360 s. Recall that the function cubatura is invoked
CONCLUSION
every time a quadripole must be estimated; the performance of the
As an objective of this study, we present an open-source code in algorithm is directly dependent on the number of quadripoles used
the MATLAB environment, which is called the VIMER package, to in the array. The number of bodies also has an impact because more
compute the electric DC response of a half-space through direct bodies imply more quadripoles to compute. All of the models were
modeling using Green’s functions. The implementation of this al- simulated using an Intel i5-6200U chip at 2.3 GHz and 12 GB
gorithm is easy and does not require additional software to show the DDR4 memory on a 64-bit Windows 10 OS.
results in a graphical format. The model half-space may contain The VIMER program fulfills the requirements to compute the
heterogeneities, each one represented by regular prisms of constant resistivity response in 3D spaces without the high cost or heavy
resistivity. computational engagement.

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 F48 Hernández Contreras and Flores Márquez

ACKNOWLEDGMENTS entered in the main window to visualize different models or

different array configurations, and the program can be run
We thank the Instituto de Geofísica of the National Autonomous again to compare the results.
University of Mexico (Universidad Nacional Autónoma de México, • The exit button is hit to close the VIMER GUI.
UNAM) for their infrastructure facilities. HOAHC thanks
CONACYT México for granting his scholarship during his Master
in Sciences studies; this work is based on his M.S. thesis. The au-
thors are grateful to the editorial team of GEOPHYSICS and the anony- REFERENCES
mous reviewers for their help in the improvement of this paper.
Abdullah, F. M., M. H. Loke, M. Nawawi, and K. Abdullah, 2019, Improv-
ing the resolution of 3-D resistivity surveys along the perimeter of a con-
DATA AND MATERIALS AVAILABILITY fined area using optimized arrays: Pure and Applied Geophysics, 176,
1701–1715, doi: 10.1007/s00024-018-2061-0.
Data associated with this research are available and can be ac- Barker, R. D., 1991, Depth of investigation of collinear symmetrical
four-electrode arrays: Geophysics, 54, 1031–1037, doi: 10.1190/1
cessed via the following URL: https://doi.org/10.5281/zenodo .1442728.
.7567075. Barrera, J. A., 2015, Empleo del método geofísico tomografía de resistividad
eléctrica para la ubicación de un sitio de recarga de acuífero en Xochi-
milco, México: Bachelor’s degree thesis, Facultad de Ingeniería, UNAM,
APPENDIX A https://tesiunam.dgb.unam.mx/F/?func=find-b&find_code=WTT&
request=Empleo+del+m%C3%A9todo+geof%C3%ADsico+Tomograf%
C3%ADa+de+Resistividad+El%C3%A9ctrica+para+la+ubicaci%C3%
QUICK GUIDE B3n+de+un+sitio+de+recarga+de+acu%C3%ADfero+en
+Xochimilco&local_base=TES01, accessed 6 March 2023.
The modeling program needs the dimensions and characteristics Benson, A. K., 1995, An integration of geophysical methods and geochemi-
of a body with flat parallel walls, and a coordinate system x, y, and z cal analysis to map acid mine drainage — A case study: Exploration and
Mining Geology, 4, 411–419.
is established; z increases upward, x increases eastward, and y in- Benson, A. K., K. L. Payne, and M. A. Stubben, 1997, Mapping ground-
creases toward the north. Distance dimensions are expressed in m, water contamination using dc resistivity and VLF geophysical methods —
angles are expressed in decimal degrees (°), and resistivities are ex- A case study: Geophysics, 62, 80–86, doi: 10.1190/1.1444148.
Binley, A., A. Kenma, Y. Rubin, and S. Hubbard, 2005, Chapter 5, DC re-
pressed in ohm.m. The first electrode laid on the surface is defined sistivity and induced polarization methods, in Y. Rubin and S. S. Hubbard,
as the origin of the coordinate system, which must be considered eds., Hydrogeophysics: Springer, 129–156.
when choosing the position and kind of array. Blanchy, G., S. Saneiyan, J. Boyd, P. McLachlan, and A. Binley, 2020,
ResIPy: An intuitive open-source software for complex geoelectrical
inversion/modeling: Computers and Geosciences, 137, 104423, doi: 10
• When starting the program, a window will appear. At the left .1016/j.cageo.2020.104423.
of the window, the user will define the layout of the electrode Chambers, J. E., P. B. Wilkinson, S. Penn, P. I. Meldrum, O. Kuras, M. H.
on the surface and the kind of array to be simulated. Loke, and D. A. Gunn, 2013, River terrace sand and gravel deposit reserve
estimation using three-dimensional electrical resistivity tomography for
• The first electrode is the coordinate (0, 0, 0), after which fol- bedrock surface detection: Journal of Applied Geophysics, 93, 25–32,
lows the electrode line. doi: 10.1016/j.jappgeo.2013.03.002.
•
Clement, R., and S. Moreau, 2016, How should an electrical resistivity
In the first two fields, the number of electrodes and the dis- tomography laboratory test cell be designed? Numerical investigation
tance between them are entered. of error on electrical resistivity measurement: Journal of Applied
• In the third field, the angle from the y-axis (north) to the elec- Geophysics, 127, 45–55, doi: 10.1016/j.jappgeo.2016.02.008.
Coggon, J. H., 1971, Electromagnetic and electrical modelling by the finite
trode line is entered. element method: Geophysics, 36, 132–155, doi: 10.1190/1.1440151.
• The type of array is selected from the pop-up list. Dey, A., and H. F. Morrison, 1979, Resistivity modeling for arbitrarily
• The number of bodies to be considered in the half-space is shaped three-dimensional structures: Geophysics, 44, 753–780, doi: 10
.1190/1.1440975.
selected. According to the typed number, a group of tabs Edwards, L. S., 1977, A modified pseudosection for resistivity and IP:
should appear below inside a sub panel in the main window. Geophysics, 42, 1020–1036, doi: 10.1190/1.1440762.
•
Eskola, L., 1992, Geophysical interpretation using integral equations:
Only a maximum of six bodies can be entered into the pro- Chapman & Hall.
gram for performance convenience. Flores-Márquez, E. L., A. Tejero-Andrade, A. León-Sánchez, C. Arango-
• For the tabs mentioned previously, the user will establish the Galván, and R. Chávez-Segura, 2015, Two algorithms to compute the
electric resistivity response using Green’s functions for 3D structures:
dimensions and characteristics of each body in their respec- Geofísica Internacional, 54, 7–20, doi: 10.1016/j.gi.2015.04.006.
tive tab. The first two fields are reserved for the positions x Fox, R. C., G. W. Hohmann, T. J. Killpack, and L. Rijo, 1980, Topographic
and y of the geometric center of the body, and the next three effects in resistivity and induced-polarization surveys: Geophysics, 45,
75–93, doi: 10.1190/1.1441041.
fields determine the shape of the body: height (z-axis), width Gernez, S., A. Bouchedda, E. Gloaguen, and D. Paradis, 2020, Aim4res: An
(y-axis), and length (x-axis). open-source 2.5D finite differences MATLAB library for anisotropic
•
electrical resistivity modeling: Computers and Geosciences, 135,
In the depth field, the distance between the surface and the 104401, doi: 10.1016/j.cageo.2019.104401.
top of the body is specified. Gómez-Treviño, E., 1987, Nonlinear integral equations for electromagnetic
• The last field defines the resistivity of the body. inverse problems: Geophysics, 52, 1297–1302, doi: 10.1190/1.1442390.
Gonzalez, R., and R. Woods, 1992, Digital image processing: Addison-Wes-
• After entering all the required data, the run button at the bot- ley Publishing Company.
tom left of the window is hit to initiate the process. Hernández-Contreras, H. O. A., 2018, Desarrollo de un algoritmo para el
• After processing the data, the program will show a window cálculo de resistividades del subsuelo por medio de tomografía eléctrica:
M.S. thesis, Posgrado en Ciencias de la Tierra, UNAM, https://repositorio
with the electric response. At the bottom right of this win- .unam.mx/contenidos/modelado-de-la-respuesta-de-resistividad-
dow, there is a “Save” button. To save the results as an xyz electrica-de-un-caso-de-geometria-regular-de-estructuras-3d-usando-
funciones-101417, accessed 6 March 2023.
Ascii file, this button is hit, and the name of the file is en- Hernández-Contreras, H. O. A., 2023, Volume integral model of the electric
tered. The file will be stored in the same directory in which response (VIMER), Version 1.5.1, [software]: GitHub, https://github.com/
the program was run. New parameters or new bodies can be Octavio-Augusto/VIMER, accessed 16 January 2023.

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user
 VIMER program to compute ERT response F49

Hohmann, G. W., 1971, Electromagnetic scattering by conductors in the Marescot, L., S. P. Lopes, S. Rigobert, and A. G. Green, 2008, Nonlinear
earth near a line source of current: Geophysics, 36, 101–131, doi: 10 inversion of geoelectric data acquired across 3-D objects using a finite-
.1190/1.1440150. element approach: Geophysics, 73, no. 3, F121–F133, doi: 10.1190/1
Hohmann, G. W., 1975, Three-dimensional induced polarization and .2903836.
electromagnetic modeling: Geophysics, 40, 309–324, doi: 10.1190/1 Mufti, I. R., 1976, Finite-difference resistivity modeling for arbitrarily
.1440527. shaped two-dimensional structures: Geophysics, 41, 62–78, doi: 10
Hörning, S., L. Gross, and A. Bárdossy, 2020, Geostatistical electrical re- .1190/1.1440608.
sistivity tomography using random mixing: Journal of Applied Geophys- Orellana Silva, E., 1982, Prospección geoeléctrica en corriente continua,
ics, 176, 104015, doi: 10.1016/j.jappgeo.2020.104015. Parte 1: Paraninfo, 578.
Huo, Z.-H., S. K. Dai, and Q. Y. Jiang, 2014, Integral equation forward mod- Pérez-Flores, M. A., 1995, Inversión rápida en 2-D de datos de resistividad,
eling in geophysical electromagnetic field: Progress in Geophysics, 29, magnetotelúricos y electromagnéticos de fuente controlada a bajo número
742–747. de inducción: Ph.D. thesis, Centro de Investigación Científica y de
Jayan, S., and K. V. Nagaraja, 2015, A general and effective numerical in- Estudios Superiores de Ensenada, Ensenada, México.
tegration method to evaluate triple integrals using generalized Gaussian Pérez-Flores, M. A., S. Méndez-Delgado, and E. Gómez-Treviño, 2001, Im-
quadrature: Procedia Engineering, 127, 2041–1047, doi: 10.1016/j aging low frequency and DC electromagnetic fields using a simple linear
.proeng.2015.11.457. approximation: Geophysics, 66, 1067–1081, doi: 10.1190/1.1487054.
Karaoulis, M., A. Revil, P. Tsourlos, D. D. Werkema, and B. J. Minsley, Pidlisecky, A., E. Haber, and R. Knight, 2007, RESINVM3-D: A MATLAB
2013, IP4DI: A software for time-lapse 2D/3D DC-resistivity and induced 3-D resistivity inversion package: Geophysics, 72, no. 2, H1–H10, doi: 10
polarization tomography: Computers and Geosciences, 54, 164–170, doi: .1190/1.2402499.
10.1016/j.cageo.2013.01.008. Pidlisecky, A., and R. Knight, 2008, FW2_5D: A MATLAB 2.5-D electrical
Kauffman, A. A., 1992, Geophysical field theory and method: Part A — resistivity modeling code: Computers and Geosciences, 34, 1645–1654,
Gravitational, electric and magnetic fields: Academic Press. doi: 10.1016/j.cageo.2008.04.001.
León-Sánchez, A. M., 2004, Modelación de la respuesta eléctrica de estruc- Pridmore, D. F., G. W. Hohmann, S. H. Ward, and W. R. Sill, 1980,
turas 3D en un semiespacio conductor: Bachelor’s degree thesis, Facultad An investigation of finite-element modeling for electrical and electromag-
de Ingeniería, UNAM. netical data in 3D: Geophysics, 46, 1009–1024, doi: 10.1190/1.1441239.
Li, Y., and D. W. Oldenburg, 1996, 3-D inversion of magnetic data: Geo- Ren, Z., and J. Tang, 2010, 3D direct current resistivity modeling with un-
physics, 61, 394–408, doi: 10.1190/1.1443968. structured mesh by adaptive finite-element method: Geophysics, 75, no. 1,
Li, Y., and K. Spitzer, 2002, 3D direct current resistivity forward modeling H7–H17, doi: 10.1190/1.3298690.
using finite element in comparison with finite-difference solutions: Geo- Sasaki, Y., 1994, 3D resistivity inversion using the finite-element method:
physical Journal International, 151, 924–934, doi: 10.1046/j.1365-246X Geophysics, 59, 1839–1848, doi: 10.1190/1.1443571.
.2002.01819.x. Snyder, D. D., 1976, A method for modeling the resistivity and IP response
Li, Y., and K. Spitzer, 2005, Finite-element resistivity modeling for 3D struc- of two-dimensional bodies: Geophysics, 41, 997–1015, doi: 10.1190/1
tures with arbitrary anisotropy: Physics of the Earth and Planetary Inte- .1440677.
riors, 150, 15–27, doi: 10.1016/j.pepi.2004.08.014. Spitzer, K., 1995, A 3D finite-difference algorithm for DC resistivity mod-
Li, Y., and D. Yang, 2021, Electrical imaging of hydraulic fracturing fluid eling using conjugate gradient methods: Geophysical Journal
using steel-based wells and a deep-learning method: Geophysics, 86, International, 123, 903–914, doi: 10.1111/j.1365-246X.1995.tb06897.x.
no. 4, E315–E332, doi: 10.1190/geo2020-0178.1. Szalai, S., and L. Szarka, 2008, On the classification of surface geoelectric
Loke, M. H., 2004, Tutorial: 2-D and 3-D electrical imaging surveys, https:// arrays: Geophysical Prospecting, 56, 159–175, doi: 10.1111/j.1365-2478
www.researchgate.net/publication/264739285_Tutorial_2-D_and_3- .2007.00673.x.
D_Electrical_Imaging_Surveys, accessed 25 October 2022. Tejero-Andrade, A., 2002, Modelación de la respuesta magnetométrica em-
Loke, M. H., and R. D. Barker, 1996, Practical techniques for 3D resistivity pleando aproximación de Born con conductividad de referencia variable:
surveys and data inversion: Geophysical Prospecting, 44, 499–523, doi: Ph.D. thesis, Instituto de Geofísica, U.N.A.M.
10.1111/j.1365-2478.1996.tb00162.x. Telford, W. M., L. P. Geldart, and R. E. Sheriff, 1990, Applied geophysics,
Loke, M. H., J. E. Chambers, D. F. Rucker, O. Kuras, and P. B. Wilkinson, vol. 1, 2nd ed.: Cambridge University Press.
2013, Recent developments in the direct-current geoelectrical imaging Troiano, A., and M. G. Di Giuseppe, 2020, Application of principal com-
method: Journal of Applied Geophysics, 95, 135–156, doi: 10.1016/j ponent analysis to geo-electrical recordings: Journal of Applied Geophys-
.jappgeo.2013.02.017. ics, 178, 104038, doi: 10.1016/j.jappgeo.2020.104038.
Loke, M. H., P. B. Wilkinson, and J. E. Chambers, 2010, Fast computation Tsourlos, P. I., and R. D. Ogilvy, 1999, An algorithm for the 3-D inversion of
of optimized electrode arrays for 2D resistivity surveys: Computers and tomographic resistivity and induced polarization data: Preliminary results:
Geosciences, 36, 1414–1426, doi: 10.1016/j.cageo.2010.03.016. Journal of the Balkan Geophysical Society, 2, 30–45.
Loke, M. H., P. B. Wilkinson, J. E. Chambers, S. Uhlemann, T. Dijkstra, and Zhdanov, M. S., and G. V. Keller 1994, The geoelectrical methods in
T. Dahlin, 2022, The use of asymmetric time constraints in 4-D ERT in- geophysical exploration: Elsevier.
version: Journal of Applied Geophysics, 197, 104536, doi: 10.1016/j
.jappgeo.2022.104536.
Ma, J., V. Rokhlin, and S. Wandzura, 1993, Generalized Gaussian quadra-
ture rules for systems of arbitrary functions: Yale University. Biographies and photographs of the authors are not available.

Downloaded from http://pubs.geoscienceworld.org/seg/geophysics/article-pdf/88/3/F39/5872987/geo-2022-0058.1.pdf

by Texas A&M University user