Professional Documents
Culture Documents
Abstract
The objective of this paper was to compare Wenner and dipole–dipole configurations in
delineating an underground cavity at a site near the University of Malaya, Malaysia. A
three-dimensional electrical resistivity imaging survey was carried out along seven parallel
lines using Wenner and dipole–dipole arrays. A three-dimensional least-squares algorithm,
based on the robust inversion method, was used in the inversion of the apparent resistivity data.
In the inverted model, both the horizontal and vertical extents of the anomalous zones were
displayed. Results indicate the superiority of the Wenner array over the dipole–dipole array for
determining the vertical distribution of the subsurface resistivity, although the dipole–dipole
array produced a better lateral extent of the subsurface features. The results show that the
three-dimensional electrical resistivity imaging survey using both the Wenner and
dipole–dipole arrays, in combination with an appropriate three-dimensional inversion method
and synthetic model analysis, can be highly useful for engineering and environmental
applications, especially for underground three-dimensional cavity detection.
2. Site description
Figure 2. Pseudo-section of the measured and calculated apparent resistivity data using a Wenner array along seven parallel lines.
31
A Neyamadpour et al
(a) (b)
(c) (d)
(e) (f)
(g) (h)
Figure 4. The 3D resistivity model obtained by a Wenner array as horizontal depth slices. The depth (m) for layers a through h are as
follows: (a) 0–0.5, (b) 0.5–1.1, (c) 1.1–1.7, (d) 1.7–2.5, (e) 2.5–3.4, (f ) 3.4–4.4, (g) 4.4–5.5, (h) 5.5–6.9.
32
Comparison of Wenner and dipole–dipole arrays in the study of an underground three-dimensional cavity
(a)
Figure 5. Iso-resistivity surface of the highest resistivity zones obtained by (a) Wenner array and (b) dipole–dipole array.
(a) (b)
(c) (d)
(e) (f)
(g) (h)
Figure 6. The 3D resistivity model obtained by a dipole–dipole array as horizontal depth slices. Two vertical columns are denoted by A and
B. The depth (m) for layers a through h are as follows: (a) 0–0.5, (b) 0.5–1.0, (c) 1.0–1.6, (d) 1.6–2.3, (e) 2.3–3.0, (f ) 3.0–3.9, (g) 3.9–5.0,
(h) 5.0–6.9.
3. Data acquisition to acquisition noise and higher signal strength, and it requires
fewer measurement points for the same line coverage than the
Pole–dipole and pole–pole arrays were not suitable because dipole–dipole array.
the field conditions did not allow for the placement of remote The dipole–dipole array has better horizontal resolution
electrodes at a sufficient distance from the survey lines. The and better depth coverage at the ends of the lines. On the
Wenner array has better depth resolution, less susceptibility other hand, at the investigation site, both the horizontal and
33
A Neyamadpour et al
Figure 7. The sensitivity values of the cells used in the inversion model for the Wenner array. The depth (m) for layers a through h are as
follows: (a) 0–0.5, (b) 0.5–1.1, (c) 1.1–1.7, (d) 1.7–2.5, (e) 2.5–3.4, (f ) 3.4–4.4, (g) 4.4–5.5, (h) 5.5–6.9.
(d)
(e) (f)
(g) (h)
Figure 8. The sensitivity values of the cells used in the inversion model for the dipole–dipole array. The depth (m) for layers a through h are
as follows: (a) 0–0.5, (b) 0.5–1.0, (c) 1.0–1.6, (d) 1.6–2.3, (e) 2.3–3.0, (f ) 3.0–3.9, (g) 3.9–5.0, (h) 5.0–6.9.
the vertical subsurface structures are expected. Therefore, the electrode spacings were measured (at 1, 2 and 3 times the
Wenner and dipole–dipole arrays were both used for this case minimum spacing), which gave a data set of 134 data points for
study. each line (n = 1 to n = 6). In addition, in the Wenner array, five
Measurements were collected along seven parallel lines, different electrode spacings were measured in the x-direction
and the electrode cables were oriented in the x-direction, which (at 1, 2, 3, 4 and 6 times the minimum spacing), which gave
is approximately north to south (figure 1). Each cable covers a data set of 52 data points for each line. The background
38 m between the first and the last take-out, with a spacing of noise levels were checked before measurements began. The
2 m in the x-direction. Roll-along measurements using a y- vertical stack was set to four. The relative standard deviation
spacing of 2 m were carried out until a grid of 20 × 7 electrodes for each stack was checked during the measurements because
was covered. Thus, the total investigation area was 38 × 12 m, it is a valid indicator of the quality of the measurements (Tohon
and 140 electrode positions were used. Measurements were et al 2004). The standard deviation of the measurements was
taken only in the x-direction in order to limit the time needed generally less than 2% and 4% for the Wenner and dipole–
for data acquisition. In the dipole–dipole array, three different dipole arrays, respectively. All measured data sets were
34
Comparison of Wenner and dipole–dipole arrays in the study of an underground three-dimensional cavity
(A)
Figure 9. Two synthetic models used in this study. The vertical columns shown in figure 1 are denoted by A and B in these models.
merged in order to perform the 3D inversion using RES3Dinv subsurface structure (Dahlin et al 2002). Thus, we performed
software (Loke 2007). Two separate final sets of 938 and a 3D inversion of the apparent resistivity data to obtain a more
364 apparent resistivity measurements were collected using a reliable image of the cavity structure using the robust inversion
Terrameter SAS 4000 instrument (Dahlin 1996). technique.
In order to determine the subsurface resistivity from the
4. Inversion of the data set measured apparent resistivity, the widths of the interior blocks
in the top two layers of the inversion model were set at half the
The measured apparent resistivity data obtained by the Wenner unit electrode spacing, while the widths of the blocks in the
and dipole–dipole arrays are displayed as a pseudo-section in deeper layers were kept equal to the electrode spacing (Dahlin
figures 2 and 3, respectively. Although the pseudo-section et al 2002). The thickness of the first layer and the factor
can give some information about the locations of the cavity used to increase the layer thickness were set at 1.5 m and
and the horizontal pipe, their size, depth and extent cannot be 1.15 m, respectively. We used 0.15 and 0.01 for the initial and
correctly estimated. Furthermore, the pseudo-section can be minimum damping factors, respectively, in the data inversion.
misleading if used as the basis for an interpretation of the true These factors determine the relative importance given to
35
A Neyamadpour et al
(a)
(b)
(d)
(e)
Figure 10. Inversion results of the Wenner data for model (1). The depth (m) for layers a through h are as follows: (a) 0–1.0, (b) 1.0–2.2, (c)
2.2–3.5, (d) 3.5–5.0, (e) 5.0–6.8.
reducing the data misfitting and the smoothness of the model 5.5 m. The resistive zone, on the south side of the slices
(Ellis and Oldenburg 1994). The finite difference method (figure 4), is also about 4.4 m in depth. High resistivities are
(Dey and Morrison 1979) was used to calculate the apparent not observed near the middle of the site.
resistivity values as well as the elements of the Jacobian In order to display the 3D extent of the cavity and
matrix (McGillivray and Oldenburg 1990) for the inversion horizontal pipe, an iso-resistivity surface was also produced
method. The number of iterations to recalculate the Jacobian corresponding to resistivities higher than 1550 m. The
matrix was also set to three. The investigation site consists isosurface connects data points of equal resistivity values
of discrete subsurface structures with sharp boundaries (figure 5(a)). As illustrated in the figure, the horizontal pipe
between different bodies. For such cases, a robust model and cavity have been clearly detected.
inversion is more suitable than the smoothness-constrained
method. 5.2. The dipole–dipole array
As a comparative study, the 3D resistivity imaging data
5. Results obtained by the dipole–dipole array were also inverted using
the robust inversion method. After nine iterations, the
5.1. Wenner array
inversion process converged with a RMS misfit of 14.09%,
The inversion process using the robust technique converged which is greater than the RMS misfit for the inversion obtained
after six iterations with a RMS misfit of 4.52% for the apparent by the Wenner array (4.52%). This difference can be explained
resistivity values. The 3D resistivity model is shown as by the greater signal strength of the Wenner array than the
horizontal depth slices in figure 4. The presence of the vertical dipole–dipole array, especially when a high resistivity contrast
column A, shown earlier in figure 1, is not clear in these slices, exists.
although the cavity with a width of 11 m in its upper part From the inverted data, we extracted horizontal depth
(layer a) and 5 m in its lower part (layer g) is observed. slices (figure 6) in order to display the lateral extent of the high
According to this figure, the depth of the cavity is almost resistivity zones. However, while all features are almost well
36
Comparison of Wenner and dipole–dipole arrays in the study of an underground three-dimensional cavity
(a)
(b)
(c)
(e)
(f)
Figure 11. Inversion results of the dipole–dipole data for model (1). Corresponding depth (m) for layers a to h are as follows: (a) 0–0.7, (b)
0.7–1.5, (c) 1.5–2.5, (d) 2.5–3.7, (e) 3.7–5.2, (f ) 5.2–7.0.
resolved in these depth slices, the inversion of the resistivity to resistivities higher than 1550 m (figure 5(b)). According to
data obtained by the dipole–dipole array produces a greater the figure, the cavity has been clearly detected, but the vertical
depth for the resistive material surrounding the horizontal pipe dimension of the pipe was not reliable. Although the dipole–
(>6.9 m) and a smaller depth for the cavity (3.9 m) compared dipole array produced a better lateral extent for the subsurface
to the inversion result obtained by the Wenner array. It is features, the results illustrate the superiority of the Wenner
clear that the Wenner array has better depth resolution than the array over the dipole–dipole configuration for determination
dipole–dipole array. On the other hand, the vertical columns A of the vertical distribution of subsurface resistivity. The
and B are observed in the first slice (layer a). These two inversion of the data set for each configuration was carried
columns are not observed in the inversion obtained from the out on a 1.73 GHz Dual Core PC, and the system required
Wenner array. This can be explained by the better horizontal approximately 6 min for the program to converge to a
resolution of the dipole–dipole array compared to the Wenner satisfactory model.
configuration. In these slices, the cavity with a width of 14 m
in its upper part (layer a) and about 5 m in its lower part 6. Discussion
(layer f) is observed. Although the horizontal changes in
resistivity are in accordance with the findings obtained from To study the suitability of different arrays in resistivity surveys,
known information about the investigation area, insufficient several parameters must be evaluated including the vertical
agreement in the vertical resistivity distribution was found, and horizontal resolutions. A common way to study such
especially on the south side of the site. parameters is to calculate a sensitivity function. The sensitivity
An iso-resistivity surface was also produced to display the function shows the degree to which a change in the resistivity
3D extent of the cavity and horizontal pipe, which corresponds of the subsurface will influence the potential measured by
37
A Neyamadpour et al
the array (Edwards 1977, Loke 2009). The difference in the models (1) and (2) is the presence of high resistivity zones
contour pattern of the sensitivity function plot is studied in with a resistivity of 1500 m in synthetic model (2). This
order to explain the response of the different arrays to different difference was incorporated in order to study the effect of
types of structures and compare the suitability of various arrays construction materials left after the construction of the pipe.
for a particular survey. The sensitivity values of the cells used Figures 10 and 11 show the inversion results of the Wenner
in the inversion model for the Wenner and the dipole–dipole data and the dipole–dipole data for model (1), respectively.
arrays are shown in figures 7 and 8, respectively. In these The results indicate that both arrays could not resolve the
figures, the model cells in the areas of the high resistivity horizontal resistivity on the right side of the grid (the pipe).
zones located on both sides of the grid have sensitivity This can be explained by relatively poor data coverage of
values ranging from 0.4 to 1.3 for the dipole–dipole array the arrays on both sides of the grid, especially for deeper
(figures 8(e)–(h)). These values are higher than the slices. In contrast, the cavity is almost well observed by both
sensitivity values, ranging from 0.04 to 0.5, for the Wenner arrays along with the vertical columns A and B in two upper
array (figures 7(e)–(h)). Thus, the results of the dipole– slices.
dipole array for both sides of the grid might be more reliable The 3D resistivity models obtained by the inversion of the
than the Wenner array. Since the sensitivity values of these Wenner and dipole–dipole data are shown as horizontal depth
regions for both arrays are relatively small, we considered two slices in figures 12 and 13, respectively. Comparing these
synthetic test models in order to attain a reliable interpreted pictures with figures 4 and 5, it is clear that the cavity is well
model. resolved as a high resistive zone on the left side of slices in
The background resistivity of the synthetic models shown the model. The high resistivity zone on the right side of slices
in figure 9 is 400 m, and the horizontal resistivity zone (figures 12 and 13) is also obvious in these figures. Therefore,
with width 6 m is located at a depth ranging from 5 m to we conclude that the high resistivity zones observed on the
6 m. The high resistivity zone on the left side of both models right side of grid in inversion results of field data is probably
was considered in order to observe the effect of the cavity on due to the construction material left after construction of the
the inversion results. The main difference between synthetic pipe.
38
Comparison of Wenner and dipole–dipole arrays in the study of an underground three-dimensional cavity
39
A Neyamadpour et al
Claerbout J F and Muir F 1973 Robust modeling with erratic data non-linear inverse problem: a comparative study Geophys.
Geophysics 38 826–44 Prospect. 38 499–524
Cook J C 1974 Yes, we can locate solution cavity boundaries Rock Mochales T, Casas A M, Pueyo E L, Pueyo O, Roman M T, Pocovi
Mechanics and Geophysics: 4th Symposium on Salt vol 2 A, Soriano M A and Anson D 2008 Detection of underground
ed A H Coogan (Cleveland, USA: Northern Ohio Geological cavities by combining gravity, magnetic and ground
Society) pp 33–40 penetrating radar surveys: a case study from the Zaragoza area,
Dahlin T 1996 2D resistivity surveying for environmental and NE Spain Environ. Geol. 53 1067–77
engineering applications First Break 14 275–84 Monteiro Santos F A, Andrade Afonso A R and Dupis A 2007 2D
Dahlin T and Bernstone C 1997 A roll-along technique for 3D joint inversion of dc and scalar audio-magnetotelluric data in
resistivity data acquisition with multi-electrode arrays Proc. the evaluation of low enthalpy geothermal fields J. Geophys.
SAGEEP’97 (Symp. on the Application of Geophysics to Eng. 4 53–62
Engineering and Environmental Problems) (Reno, NV) Monteiro Santos F A and Sultan A S 2008 On the 3D inversion of
pp 927–35 vertical electrical soundings: application to the south Ismailia
Dahlin T, Bernstone C and Loke M H 2002 A 3-D resistivity area—Cairo desert road, Cairo, Egypt J. Appl.
investigation of a contaminated site at Lernacken, Sweden Geophys. 65 97–110
Geophysics 67 1692–700 Nyári Z and Kanli A I 2007 Imaging of buried 3D objects by using
40