Geology, Geophysics & Environment 2013 Vol. 39 No.

4 331–339

http://dx.doi.org/10.7494/geol.2013.39.4.331

2D Electrical Resistivity Tomography

interpretation ambiguity –
example of field studies supported
with analogue and numerical modelling

Grzegorz Bania & Michał Ćwiklik

AGH University of Science and Technology, Faculty of Geology,

Geophysics and Environmental Protection, Department of Geophysics;
al. Mickiewicza 30, 30-059 Krakow, Poland;
e-mail: [email protected], [email protected]

Abstract: Single Electrical Resistivity Tomography (ERT) survey was carried out in  the  Manor
and Park Complex in Nowa Huta (Krakow Branice, Poland). It was applied at a small distance and pa-
rallel to the longer wall of a monumental building containing an empty 3 m deep basement. Analogue
modelling was performed in order to recreate the field study at the proper scale. The laboratory set-up
consisted of a water tank where electrodes were mounted to the particular plate, which rested on wa-
ter surface. The basement model was made out of a non-conducting material. The default and robust
inversions were tested and these variants were also considered with the use of numerical modelling.
Laboratory experiments have confirmed that zones visible in the interpreted field section are caused
by the influence of the building cellar located next to the survey line. Zones of this kind are additio-
nally disturbed by the local geological structure. The experiment has pointed out, among others, that
as the distance between the survey line and the underground body increases, the inversion results are
still burdened by an object influence. Thus, similar situations can be verified with the use of analogue
modelling presented in this paper or 3D numerical one.
Key words: Electrical Resistivity Tomography (ERT), ambiguity, 2D inversion, analogue and nume-
rical modelling

331
332 G. Bania & M. Ćwiklik

INTRODUCTION
The Electrical Resistivity Tomography (ERT) is currently the most frequently used geo-
electrical method (Dahlin 1996, Loke 2010a). It is broadly used in geomorphology (Schrott
& Sass 2008), engineering (Daily & Ramirez 2000), archaeology (Cardarelli 2002, Mościcki
2008) and  environmental studies (Dahlin et al. 1996, Amidu & Dunbar 2008, Georgaki et
al. 2008, Sechman et al. 2013). Generally, a 2D variant is performed. As a result, we obtain
a so-called apparent resistivity pseudosection. The section presents the distribution of regis-
tered apparent resistivity as a function of distance on a survey line and a so-called pseudodepth
or the size of spacing AB/2 (Loke 2010b). Apparent resistivity pseudosection is the qualitative
reflection of apparent resistivity distribution in the centre. By carrying out inversion, we obtain
the quantitative distribution of electrical resistivity as a function of distance and depth (Loke
& Barker 1996). The ERT inversion is carried out under an assumption that the structure of
the centre is approximated with a 2D model (x, z). For y direction, we assume the continuity
of structures presented in a 2D section (Loke 2010b). It is a simplification, as the real centre
is 3D. The inversion is an iteration process where the initial, theoretical, model of the centre is
being successively modified. The process ends when the compatibility of the theoretical data,
calculated from a model, and field data is satisfactory. It is then estimated by the Root Mean
Square (RMS) error, which determines the discrepancy between the theoretical apparent resis-
tivity pseudosection (calculated for the given model) and the field one. The characteristic fea-
ture of the inversion results is the smooth distribution of electrical resistivity in the centre. It is
related to the inversion technique – the so-called smoothness constrain least squares method
(Loke 2010b). It can be noticed that for problems connected with e.g. underground water pol-
lution, the assumption may be right but in case where sharp geological boundaries are present,
it is rather not. We should also remember that in the ERT method we deal with the ambiguity
problem, similar to Vertical Electrical Sounding method (Koefoed 1979), but it has a more
complicated 2D character. The ambiguity depends on the fact that the centre can be attributed
by many correct geoelectrical models which are situated in the accepted RMS error limits.
In practice, an interpreted resistivity section is related to a centre present directly under
the survey line. This assumption is true only for situations where the geological structure is
uncomplicated enough and can be approximated with 2D model.
In this paper, we present an example of field data where the ERT results are disturbed
by an object not present directly under the survey line. In the described case, the survey line
is situated next to the building with a cellar. Numerical modelling and analogue modelling at
a laboratory scale were performed in the attempt to explain and discuss the obtained results.

METHOD
The field survey was carried out in the Manor and Park Complex in Nowa Huta (Kra-
kow Branice, Poland). The field measure was applied along a single survey line located par-
allel to the longer wall of a building. The building dimensions, containing an empty 3 m deep
2D Electrical Resistivity Tomography interpretation... 333

basement, are 10.0 m by 11.5 m. The survey line was located 0.5 m from the wall and was
95 m long (Fig. 1). The ERT measurement was performed with the GDRM-VAR resistivity
meter and 96 electrodes. The basic electrode spacing Δx was 1 m and the Wenner alpha spac-
ings were a = 1 m, 2 m, 3 m, 4 m, 5 m, 6 m, 8 m, 11 m, 15 m and 20 m.

Fig. 1. ERT field data inversion results: LR – low resistivity zone, HR – high resistivity zone

Analogue modelling was performed in order to recreate the field study (presence of

the basement) at the (laboratory) scale of 1:67. The laboratory set-up consisted of a water
tank, with the dimensions of 2.0 m by 1.5 m by 1.4 m (Fig. 2A). On the water surface, which
practically is a  homogenous conductivity medium, there was a  plate made from a  non-
conductor. 98 electrodes were linked to the plate creating a single ERT profile. The ERT
profile was 145.5 cm long and the basic electrode spacing Δx was 1.5 cm. The survey was
carried out with the  GEOMES-RR5 resistivity meter. The  Wenner array spacings were
a = 1.5 cm, 3.0 cm, 4.5 cm, 6.0 cm, 9.0 cm, 12.0 cm and 16.5 cm. The basement model
was made out of a non-conducting material. The model dimensions are 17.3 cm by 15.0 cm
by 4.5 cm (Fig. 2B).
334 G. Bania & M. Ćwiklik

A B

Fig. 2. Laboratory settings (A) and analogue modelling inversion results (B): LR – low resistivity
zone, HR – high resistivity zone

The laboratory measurements were conducted on four profiles. In the first case, the mod-
el was moved 7.5 mm from the survey line. In the next cases, the distance was 15.0 mm,
22.5 mm and 30.0 mm. For the field surveys it is equivalent to: 0.5 m, 1.0 m, 1.5 m and 2.0 m.
In the laboratory experiments some scaling problems related to limited dimensions of the wa-
ter tank, electrodes size, etc. occurred. In order to limit the measurement errors, each ERT
survey was conducted twice. At stage one, a measurement was carried out without the mod-
el (N – uniform half space + noise). Afterwards, the survey was repeated with the model
(A – anomaly + noise). All survey results were normalized. Apparent resistivity values for
the anomaly were divided by corresponding apparent resistivity values for the noise (A/N).
The normalization allowed distinguishing mainly the anomaly coming from the model.
The field and  laboratory ERT data were inverted with RES2DINV software (Loke
2010a). Default settings, like e.g. mesh grid size, were used. The default and robust inver-
sion (Loke et al. 2003) options were tested. These variants were also considered with the use
of numerical modelling which was performed with RES2DMOD software (Loke 2002).
For this, four geometrical models were made (Fig. 3A). They were characterised by sharp
2D Electrical Resistivity Tomography interpretation... 335

boundaries between the modelled body and the surroundings. Models were created for two
versions of the resistivity contrast – 1000:1 and 1:1000. The modelling procedure was car-
ried out for the Wenner array spacing a = 1d, 2d, 3d, …, 20d. The calculated distribution of
apparent resistivity was then inverted.

Fig. 3. Numerical models (A), default (B) and robust inversion results (C) for two different resistivity
contrasts between the  modelled body and  the  background: H –  high resistivity body (ρ2/ρ1 = 103),
L – low resistivity body (ρ2/ρ1 = 10−3)

RESULTS
Figures 3B and 3C show inversion results of numerical models (Fig. 3A). In Figure 3B,
for the default inversion, the interpreted resistivity distribution is blurred and does not reflect
the real shape of the anomalous body. For the robust inversion (Fig. 3C), the distribution of
336 G. Bania & M. Ćwiklik

the isolines is closer to the real object shape than in the previous case. Inversion results show,
apart from the real objects (models) mapping, false anomalies as well. These anomalies are
often beyond the model boundaries. Anomalies for the default variant are clearly larger than
for the robust variant.
Object mapping differs for different interpretation variants. It depends on the geometri-
cal situation. The object mapping is the most precise for the M2 model (Fig. 3B, C). Inversion
results for the M1 model show that the anomaly is slightly deeper than the modelled body. In
the M3 case, the anomaly concentrates in the upper half of the assumed object. In inversion
results for the M4 model, we can observe a clear difference for opposite resistivity contrasts.
For good conductive objects (L), there are considerably lower interpreted resistivity zones
than the resistivity of the assumed object. This can be explained by the complicated character
of the current flow for such geometrical configuration.
The robust inversion proves to be useful in those cases where the model has clearly sharp
boundaries. The inversion variant is dedicated to situations where some elements, e.g. geo-
logical structure, is known and could be treated in the inversion process as a benchmark (Loke
2010b). In this research, field and laboratory body dimensions (the model and the basement)
are known. The resistivity contrast between the modelled body and the background is very
high. As a consequence, the robust inversion was applied as advisable for presented cases.
The results of the  field data inversion are shown in  Figure 1. A  few characteristic
zones revolved on the resistivity section. The HR1 and HR2 zones are situated between 51
and 64 meter of the profile and their resistivity value is about 400 Ω·m. The HR2 zone is lim-
ited in the bottom, so its depth range could not be estimated. Between these zones, the LR1
zone is distinguished. It ranges between 1.5–3.0 m below the surface and the resistivity value
is 10–50 Ω·m. Two relatively low resistivity zones, LR2 (10–25 Ω·m), are clearly visible on
both sides of the HR2 zone. The Z1 zone is another typical zone with quite regular shapes.
The inversion results of the laboratory ERT data are shown in Figure 2B. In the shallow
part of P1 section, we can observe a relatively high resistivity zone HR1. Its horizontal extent
is almost equal to the modelled body width. The HR1 zone is also separated from the surface
by a thin low resistivity layer. The green colour HR2 zone is connected with the HR1 zone
at two spots and it spans the section boundaries. The low resistivity oval shaped zone LR1
can be observed between the HR1 and HR2 zones. Other typical anomalies, which have rela-
tively low resistivity values in comparison to the background, are two LR2 zones, which are
clearly visible in all four sections. The situation is almost the same for P2 section, but HR1
zone is characterized by smaller amplitude. Additionally, the zone is asymmetrical. This in-
dicates that the modelled body was not placed ideally parallel to the survey line. The array is
very sensitive to geometrical deficiency in those cases where the scale is very small. In the P3
section the HR1 zone is almost faded. The HR2 and LR1 zones are amplified. Moreover,
the HR2 zone shape has changed and, as a consequence, its lower boundary can be observed.
In the P4 profile located farthest from the survey line, the  HR1 anomaly is not observed.
The HR2 zone amplitude increases considerably and its outlines are sharpened with regards
to the previous sections.
2D Electrical Resistivity Tomography interpretation... 337

DISCUSSION
Analogue modelling inversion results (Fig. 2B) show a  considerable similar-
ity to the field data inversion results (Fig. 1). The field survey line should be compared
to the laboratory P1 profile as it corresponds to the field data situation in the appropriate
scale. The interpreted HR1, HR2 and LR1 zones in P1 section (Fig. 2B) have similar distri-
bution and shapes like HR1, HR2 and LR1 zones in the field section (Fig. 1). We can also
observe that in both sections LR2 zones are visible. In laboratory sections they are placed
a little deeper. The zones of this kind are often observed in the inversion results in a situ-
ation where the  centre has a  regularly-shaped body with a  much higher resistivity than
resistivity of the background. Similar effects can be observed in ERT inversion sections
from the Krakow Main Square Market, where the water tank from World War II was found
(Mościcki 2008). It should be noted that HR2 zone in the field section (Fig. 1) has practi-
cally the same shape and it is not limited at the base as the object interpreted from numeri-
cal modelling (Fig. 3C – M4, H). However, we do not observe analogical LR2 zones next
to the object. The appearance of such zones would require additional explanation.
The analysis of the laboratory results (Fig. 2B) shows that as the distance between sur-
vey line and model increases, all distinguished zones (expect of LR2) change their amplitudes
and shapes or even vanish (HR1). This results from the underground object being present
in some distance from the survey line that has difficult to evaluate influence on the survey
results. Therefore, the 2D section, obtained as a result of ERT surveys, cannot be unambigu-
ously treated as information that comes directly from below the survey line. It is because we
are also dealing with the influence of the survey line surroundings. Cardarelli & Di Filippo
(2009) reported similar observations. As a result, there are some interpreted zones, which are
not present in reality under the survey lines in every analysed section.
The Z1 zone in  field section (Fig. 1), with the  resistivity value of ca. 70 Ω·m, has
a regular shape. It should be noted that this zone should not be treated as an interpreted effect
from a regular rectangular object. Its appearance can be a side effect of applying the robust
inversion (Loke et al. 2003). This option resulted in  sharp edges; also, this fragment was
attributed the shape of a regular geometric block. We observe similar effects in interpreted
sections showing the presence of cavities in the eastern part of Saudi Arabia as discussed by
Metwaly & AlFouzan (2013).

CONCLUSIONS
Laboratory experiments (Fig. 2) confirmed that interpreted zones visible in field section
(Fig. 1) are caused by the influence of the building cellar being placed next to the survey line.
It should be noted that zones of this kind are additionally disturbed by the local geological
structure. The experiment pointed out that as the distance between the survey line and the un-
derground body increases, the inversion results are still suffering from object influence. In all
338 G. Bania & M. Ćwiklik

laboratory cases, there were no physical objects under ERT survey lines shapes and distribu-
tion of which could correspond to those visible in the interpreted sections.
Authors also turned their attention to the problem connected with zones such as LR2
which are visible in inversion results (Figs 1, 2B). They occur in field data and analogue
modelling results when some contrasting (with background) underground body is present
under or next to the survey line. We do not observe these zones in numerical modelling re-
sults. The 2D ERT surveys results are carried out by 2D inversion, which simplifies the 3D
structure of the centre with the use of 2D section. In practice, the interpreted image is treated
as everything that is located directly under the survey line. This assumption is true only for
the situation where the geological structure is not complicated enough and can be estimated
as 2D model. In reality, such situation can be found very rarely.
While conducting ERT surveys, the data is influenced by many objects located at some
distance from the survey line. Therefore, e.g. when surveys are being made on limited area
and next to buildings or walls, we should take care about influence of such objects on the ob-
tained results. Situations of this kind can be verified with the  use of analogue modelling
(presented in the article) or 3D numerical one.

Authors are grateful to dr. Włodzimierz Jerzy Mościcki for help and advice.

The work was financially supported by Dean’s Grants no. 15.11.140.334 and
no. 15.11.140.335.

REFERENCES
Amidu S.A. & Dunbar J.A., 2008. An evaluation of the electrical-resistivity method for wa-
ter-reservoir salinity studies. Geophysics, 73, 4, G39–G49.
Cardarelli E., 2002. Geoelectric survey by ERT to  investigate marble ruins at the  Roman
forum, Rome, Italy. European Journal of Environmental and Engineering Geophysics,
7, 219–228.
Cardarelli E. & Di Filippo G., 2009. Integrated geophysical methods for the characterisation
of an archaeological site (Massenzio Basilica – Roman forum, Rome, Italy). Journal
of Applied Geophysics, 68, 508–521.
Dahlin L.T., 1996. 2D resistivity surveying for environmental and engineering applications.
First Break, 14, 7, 275–283.
Dahlin L.T., Forsberg K., Nilsson A. & Flyhammer P., 2006. Resistivity Imaging for Map-
ping of Groundwater Contamination at the Municipal Landfill La Chureca, Managua,
Nicaragua. Near Surface 2006, 12th European Meeting of Environmental and Engineer-
ing Geophysics, 4–6 September, Helsinki, Finland.
Daily W. & Ramirez A.L., 2000. Electrical imaging of engineered hydraulic barriers. Geo-
physics, 65, 1, 83–94.
2D Electrical Resistivity Tomography interpretation... 339

Georgaki I., Soupios P., Sakkas N., Ververidisa F., Trantasa E., Vallianatos F. & Manios T.,
2008. Evaluating the use of electrical resistivity imaging technique for improving CH4
and CO2 emission rate estimations in landfills. Science of the Total Environment, 389,
522–531.
Koefoed O., 1979. Geosounding Principles. Elsevier, Amsterdam.
Loke M.H., 2002. RES2DMOD ver. 3.01. Rapid 2D resistivity forward modelling using
the finite-difference and finite-elements method. Geotomo Software. Manual.
Loke M.H., 2010a. RES2DINV ver. 3.59 for Windows XP/Vista/7, Rapid 2-D Resistivity & IP
inversion using the least-squares method. Geotomo Software. Manual.
Loke M.H., 2010b. Tutorial: 2-D and 3-D electrical imaging surveys. Geotomo Software.
Loke M.H. & Barker R.D., 1996. Least squares deconvolution of apparent resistivity pseudo-
sections. Geophysics, 60, 6, 1682–1690.
Loke M.H., Ackworth I. & Dahlin T., 2003. A comparison of smooth and blocky inversion
methods in 2D electrical imaging surveys. Exploration Geophysics, 34, 182–187.
Metwaly M. & AlFouzan F., 2013. Application of 2-D geoelectrical resistivity tomography
for subsurface cavity detection in the eastern part of Saudi Arabia. Geoscience Fron-
tiers, 4, 4, 469–476.
Mościcki W.J., 2008. The application of resistivity investigations in archaeology – two case
studies from Krakow, Poland. Near Surface 2008, 14th European Meeting of Environ-
mental and Engineering Geophysics, 15–17 September 2008 Krakow, Poland. Extended
Abstracts & Exhibitors’ Catalogue, P58.
Schrott L. & Sass O., 2008. Application of field geophysics in geomorphology: Advances
and limitations exemplified by case studies. Geomorphology, 93, 55–73.
Sechman H., Mościcki W.J. & Dzieniewicz M., 2013. Pollution of near-surface zone in the vi-
cinity of gas wells. Geoderma, 197–198, 193–204.