geosciences

Review
Deep Electrical Resistivity Tomography for Geophysical
Investigations: The State of the Art and Future Directions
Marianna Balasco 1 , Vincenzo Lapenna 1, *, Enzo Rizzo 2 and Luciano Telesca 1

1 Institute of Methodologies for Environmental Analysis of CNR, I-85050 Tito, Italy

2 Department of Earth Sciences, University of Ferrara, I-40100 Ferrara, Italy
* Correspondence: [email protected]

Abstract: Electrical Resistivity Tomography (ERT) is a robust and well-consolidated method largely
applied in near-surface geophysics. Nevertheless, the mapping of the spatial resistivity patterns of
the subsurface at a depth greater than 1 km was performed in just a few cases by the ERT method,
called deep ERT (DERT). Since, in many cases, the term DERT was adopted with ambiguity for
geoelectrical explorations varying in depth within a range of 0–500 m, the main goal of this review
is to clearly define the DERT method, identifying a threshold value in the investigation depth. The
study focuses both on the purely methodological aspects (e.g., geoelectrical data processing in low
noise-to-signal ratio conditions; tomographic algorithms for data inversion) and on the technological
features (e.g., sensor layouts, multi-array systems), envisaging the future directions of the research
activity, especially that based on machine learning, for improving the geoelectrical data processing
and interpretation. The results of the more significant papers published on this topic in the last
20 years are analyzed and discussed.

Keywords: geoelectrical methods; electrical resistivity tomography; deep geophysical explorations

Citation: Balasco, M.; Lapenna, V.;

Rizzo, E.; Telesca, L. Deep Electrical
Resistivity Tomography for
Geophysical Investigations: The State
1. Introduction
of the Art and Future Directions. To date, there has been growing attention paid to novel applications of geophysical
Geosciences 2022, 12, 438. https:// tomography for investigating complex geological environments and poorly understood
doi.org/10.3390/geosciences12120438 geophysical processes in the shallow part of the Earth’s crust (0–10 km). The exploration of
the “Earth Thin Skin” is vital for human life and has an extraordinary social and economic
Academic Editors: Thomas Hermans
and Jesus Martinez-Frias
impact (e.g., natural hazards, sustainable geo-energy and geo-resources, climate change and
environmental protection, etc.), coherent with the UN Sustainable Development Goals [1].
Received: 28 September 2022 Electrical and electromagnetic methodologies are currently applied to capture high-
Accepted: 23 November 2022 resolution 2D, 3D and 4D images of the subsurface resistivity patterns. One of them,
Published: 28 November 2022 Electrical Resistivity Tomography (ERT), is a robust and well-consolidated method for near-
Publisher’s Note: MDPI stays neutral surface geophysics, with a wide spectrum of applications in the geological, engineering
with regard to jurisdictional claims in and environmental sciences. Technological advances (e.g., multi-channel arrays, innovative
published maps and institutional affil- sensors) and novel tomographic algorithms for data inversion have rapidly transformed
iations. ERT into one of the most employed geophysical methods [2].
The first scientific works on the ERT method date back to the period 1990–2000 [3–5].
After these pioneer activities, the yearly number of articles focused on this exploration
technique published in top-leading journals increased exponentially, testifying the great
Copyright: © 2022 by the authors. interest of the scientific community. The ERT method has found an impressive number
Licensee MDPI, Basel, Switzerland.
of applications in near-surface geophysics, from hydrogeology to precision farming, from
This article is an open access article
engineering geology to geohazards, from CO2 storage to the study of the effects of climate
distributed under the terms and
change on soil and the subsurface [6–10].
conditions of the Creative Commons
However, most of these applications used the ERT method for investigation depths in
Attribution (CC BY) license (https://
the range of 0–200 m (Figure 1), while a very limited number of experimental works have
creativecommons.org/licenses/by/
been carried out for exploring deep geological environments (investigation depth greater
4.0/).

Geosciences 2022, 12, 438. https://doi.org/10.3390/geosciences12120438 https://www.mdpi.com/journal/geosciences

Geosciences 2022, 12, x FOR PEER REVIEW 2 of 19

Geosciences 2022, 12, 438 2 of 19

in the range of 0–200 m (Figure 1), while a very limited number of experimental works
have been carried out for exploring deep geological environments (investigation depth
greater
than than
500 m). 500 Information
m). Information on theon electrical
the electrical properties
properties of rocks
of the the rocks beyond
beyond ~1 km~1 km
depth
depth has been obtained only from direct soundings or passive
has been obtained only from direct soundings or passive electromagnetic measurementselectromagnetic meas-
urements (e.g., Magnetotelluric—MT)
(e.g., Magnetotelluric—MT) so far. Nevertheless,
so far. Nevertheless, the MT are
the MT surveys surveys are strongly
strongly affected by
affected by anthropic
anthropic noise in urbanized
noise in urbanized areas; furthermore,
areas; furthermore, the qualitytheofquality
the MTof the MT esti-
estimation of the
mation of thedistribution
resistivity resistivity distribution
in the depthin the depth
range of 0–1 range of 0–1limited
km is quite km is quite limited [11],
[11], especially if not
especially
integrated if not
withintegrated
other activewith other active
geophysical geophysical
techniques. techniques.
Typically, the MTTypically,
method isthe usedMT first
method is used first as a “pathfinder” role in a large-scale exploration
as a “pathfinder” role in a large-scale exploration and then, in the shared depth range of and then, in the
shared depth range
investigation, theofERT
investigation,
method can thegreatly
ERT method can greatly
contribute contribute
to defining to defining
in detail in
the electrical
detail the electrical
properties properties
of subsoil. of subsoil.
Therefore, Therefore,
it is necessary it is necessary
to improve to improve
the efficiency and thethe effi-
capability
ciency and
of the ERT themethod
capability of thegeological
in deep ERT method in deep geological investigations.
investigations.

Figure 1. (Left)
Figure Geological
1. (Left) map
Geological of the
map Montaguto
of the earthflow
Montaguto (Campania
earthflow (Campaniaregion, southern
region, Italy)
southern with
Italy) with
thethe
location of direct and indirect investigations. Red lines: ERT profiles. Green line: water drainage
location of direct and indirect investigations. Red lines: ERT profiles. Green line: water drainage
structures. Green dot: borehole. Blue triangle: piezometers. (Right) Fence diagram showing the
structures. Green dot: borehole. Blue triangle: piezometers. (Right) Fence diagram showing the
locations of the 2D ERT (ERT 3–ERT 13) within a 3D space. (Reprinted with permission from [7],
locations of the 2D ERT (ERT 3–ERT 13) within a 3D space. (Reprinted with permission from [7],
Copyright (2018) Elsevier).
Copyright (2018) Elsevier).
TheThepresent study
present reviews
study reviewsthethe
most important
most important scientific results
scientific obtained
results obtainedwith thethe
with
ERT method in exploring geological environments with an investigation depth
ERT method in exploring geological environments with an investigation depth greater greater
than 500500
than m. Furthermore,
m. Furthermore,a critical analysis
a critical of theof
analysis methodological and technological
the methodological lim-
and technological
its limits
underlying the application of ERT for deep geological investigations is presented
underlying the application of ERT for deep geological investigations is presented and
discussed. Finally,Finally,
and discussed. the possible and more
the possible andpromising future future
more promising directions of the research
directions ac-
of the research
tivities about this topic are identified and discussed.
activities about this topic are identified and discussed.

2. Deep ERT
2. Deep Method:
ERT Data
Method: Processing
Data Processing
WhenWhenthethe
deep ERT
deep ERTmethod
method (DERT)
(DERT)is applied, it isitnecessary
is applied, to increase
is necessary thethe
to increase spac-
spac-
inging
between
between thethe
electrodes
electrodesused forfor
used injecting
injectingcurrents
currentsintointo
thethe
ground
ground (A (A
andand
B) and forfor
B) and
receiving
receivingthethe
voltage
voltagesignals
signals(M(Mandand
N).N).
However,
However, thisthis
procedure is affected
procedure by by
is affected technical
technical
problems
problems due
dueto to
the
theuse
useofoflong
longelectrical
electrical cables
cables and the presence
and the presence of
ofinduced
inducedcurrents.
currents.For
Forthis
thisreason,
reason,inin
thethe DERT
DERT method,
method, thethe emitting
emitting andand receiving
receiving systems
systems are generally
are generally decou-
pled and the dipole–dipole array configuration is generally adopted. The injecting dipole
is physically separated from the receiving one and their mutual distance r is gradually
Geosciences 2022, 12, 438 3 of 19

increased along a selected profile on the surface to obtain the 2D resistivity patterns of deep
geological environments. Other array configurations can be adopted for obtaining the 3D
resistivity models and/or for minimizing the effects of topographic obstacles.
Considering that the amplitude of the voltage signals produced by the emitting sources
strongly decreases when the distance r increases (~1/r3 for dipole–dipole array), it is easy
to understand that the low signal-to-noise ratio in voltage recordings represents the main
problem for the DERT method [12,13]. Furthermore, the presence of conductive zones in
the subsurface and of anthropic sources of electromagnetic noise (e.g., pipelines, railway,
etc.) on the surface could make this condition worse, reducing the possibility to carry out
DERT surveys [13].
For these reasons, the processing and filtering of the noisy voltage recording is the first
key action to be approached in any application of the DERT method. The voltage signals
recorded during the DERT survey can be modeled as follows:

v(t) = s(t) + n(t) (1)

where s(t) is the useful signal produced by the square wave of a DC current injected
into the ground by the emitting dipole and n(t) is the electrical noise. The period of the
square wave is generally selected in a range varying from 5 to 60 s. Generally, the noise
is due to: (a) a natural component related to the time-dependent changes of the telluric
currents induced by the variation of the Earth’s magnetic field; (b) an anthropic component
produced by spurious electrical phenomena and man-made activities. To avoid the effects
of the electrode polarization, s(t) is a square wave of the DC current with a cyclic inversion
of polarity. Previous studies have clearly demonstrated that the geoelectrical noise is not
stationary nor Gaussian [14,15].
The amplitude ∆s of the useful signal s(t) depends on the intensity of current (I), the appar-
ent resistivity ($a) of the investigated subsurface, and the geometry of the electrode array (K):

∆s = ($a × I)/K (2)

For the dipole–dipole array configuration:

K = π (n)(n + 1)(n + 2) a (3)

where a is the spacing of each dipole and n a number for indicating the distance between the
electrodes as a multiple of the spacing a. Briefly, the presence of the conductive geological
environment, the technological limits for the power intensity of the energizing system and
the large distances between the dipoles are the factors that can strongly limit the application
of the DERT method.
The main tasks of DERT data acquisition and processing are the following: (i) record-
ings of the voltage signals with the receiving dipole (MN) in a synchronous mode with the
injection of the square wave current by means of the dipole (AB); (ii) removal of the outlier
spikes due to the man-made electrical noise and/or technical problems in the electrode
contacts; (iii) removal of the long-period drift; and (iv) estimation of the amplitude of the
useful signal s(t) [13–15].
The first task requires attention to the positioning of the electrodes into the ground
to avoid errors in the estimates of the geometrical factor K and perfect synchronization
between the emitting and receiving systems with GNSS systems. A DC power generator
used to inject currents into the ground with an intensity varying in the range of 5–20A and a
set of data loggers able to measure very low-voltage signals (10−3 –10−6 Volt) represent the
basic hardware for the DERT method. The second task is aimed at controlling the quality of
the voltage recordings with a real-time evaluation of the noise level and the use of automatic
algorithms for identifying and removing artificial peaks in the voltage signals. The length of
the voltage records is modulated on the basis of the noise intensity level [15,16], and this
aspect is fundamental to reduce the error in the estimate of the useful signals s(t).
 nals. The length of the voltage records is modulated on the basis of the noise intensit
level [15,16], and this aspect is fundamental to reduce the error in the estimate of th
Geosciences 2022, 12, 438
useful signals s(t). 4 of 19

The third task is quite simple and generally approached using the classical poly
nomial fitting to remove the drift in the voltage recordings [17]. Finally, the last and mor
The third task is quite simple and generally approached using the classical polynomial
relevant task is the estimation of the amplitude of the useful signal s(t) that is generall
fitting to remove the drift in the voltage recordings [17]. Finally, the last and more relevant
performed by the stacking, the periodogram, the Fast Fourier Transform (FFT), th
task is the estimation of the amplitude of the useful signal s(t) that is generally performed
Maximum
by Entropy
the stacking, and the autoregressive
the periodogram, the Fast Fourier model [18,19].
Transform (FFT), A
thesimplified
Maximum sequence
Entropy of th
results
and obtained during
the autoregressive model the different
[18,19]. tasks ofsequence
A simplified the voltage
of thedata processing
results is reported i
obtained during
Figure
the 2. tasks of the voltage data processing is reported in Figure 2.
different

Figure2.2.(a)(a)
Figure Example
Example of aofvoltage
a voltage recording
recording affected
affected by trend.by(b)trend. (b) The
The trend trend estimation
estimation (red line) by(red
a line) b
polynomial fitting. (c) An example of a signal recording with a good signal-to-noise ratio. (d) Spectral ratio. (d
a polynomial fitting. (c) An example of a signal recording with a good signal-to-noise
Spectral amplitude
amplitude of the FourierofTransform
the Fourier Transform
of the of theinsignal
signal reported (c), thereported
peak at theinenergization
(c), the peak at the energiza
frequency
tion frequency
(0.025 (0.025
Hz) representing theHz) representing
amplitude the amplitude
of the useful signal s(t). (e)ofAnthe usefulofsignal
example a signals(t). (e) Anwith
recording example of
signal recording with a very low signal-to noise ratio. (f) Fourier analysis of the
a very low signal-to noise ratio. (f) Fourier analysis of the signal reported in (e); the main peak is not at signal reported i
the energization frequency. (Reprinted with permission from [16], Copyright (2019), Taylor and Francis).from [16
(e); the main peak is not at the energization frequency. (Reprinted with permission
Copyright (2019), Taylor and Francis).
Lapenna et al. [15] demonstrated that the most suitable and robust method for evalu-
Lapenna
ating the et al.
amplitude [15]useful
of the signal ∆s is the
demonstrated thatperiodogram
the most suitable
techniqueand
and robust method fo
the relative
error. Since the root-mean-square error associated with the estimate of ∆s depends on the
evaluating the amplitude of the useful signal Δs is the 2periodogram technique and th
variance of the noise and the total number of sampled data (~σ /N), it strongly decreases
relative error. Since the root-mean-square error associated with the estimate of Δs de
with the increase in the time duration of the voltage recordings and, consequently, with the
pends of
number onsamples
the variance
available offor
the
thenoise and the
estimation total number
algorithms of sampled data
[14,15]. Unfortunately, during(~σ2/N), i
strongly
the decreasesa with
field acquisition, strongthe increase
increase intime
in the the duration
time duration of the voltage
of the energization and recordings
voltage and
Geosciences 2022, 12, 438 5 of 19

data acquisition leads to technical problems, as well as to a relevant increase in the field trip
costs. Thus, finding a good compromise between the need to reduce the errors in estimating
the amplitude of the useful signals and the duration of the measurement campaigns and
cost of field trips, represents a crucial task.
As it concerns the tomographic inversion of the 2D and 3D apparent resistivity val-
ues obtained during the field surveys, the published works applied robust and well-
consolidated algorithms adopted for the near-surface applications of the ERT method.
However, DERT is generally characterized by a low density of data, so that automatic
parameterization by near-surface-dedicated software might not be optimal, and adverse
effects of regularization might appear in the inversion.
Several inversion methods are well defined in this context, such as the RES2DINV
software by Geotomo [3] that is based on the smoothness-constrained least-squares inver-
sion implemented by a quasi-Newton optimization technique, the DCI2D by UBC-GIF [20],
which consists of iterative procedures of systems of equations based on objective functions,
the BERT software [21], in which the cost function is minimized using a Gauss–Newton
scheme, and the ERT-Lab by Multi-Phase Technologies and Geostudi Astier [16], which uses
a finite elements (FEM) approach to model the subsoil by adopting a mesh of hexahedrons to
correctly incorporate complex terrain topography and is based on a smoothness-constrained
least-squares algorithm with Tikhonov model regularization. The only additional critical
element is given by the removal of the apparent resistivity values derived from the esti-
mates of the useful signals affected by large errors. Generally, a threshold value of 10% is
adopted as the relative error in the estimate of the ∆s.

3. Deep ERT Method: Applications

Most of the ERT applications concern near-surface investigations (0–200 m) and are
based on the use of the multi-electrode systems and multi-channel cables for data acquisi-
tion. The same approach is adopted for investigating electrical resistivity variations with
an investigation depth in the range of 200–500 m [22–25]. The increase in the length of
cables and the difficulties of the field acquisition activities that are also time-consuming
strongly limit the use of ERT for deeper aims. In fact, to investigate deeper than a 500 m
depth threshold, it is necessary to avoid the use of multi-channel cables, to adopt decoupled
energizing and measuring systems and to apply advanced approaches for data acquisition
and analysis, representing surely a challenging task. Thus, in this section, we review the
most relevant works focused on the application of the 2D and 3D DERT method based on
the use of decoupled transmitting and receiving systems, for investigation depths greater
than 500. It is remarkable to highlight that few papers have been published on this topic
in the last 20 years, and only in recent years have novel applications of this method been
made available.
One of the first applications of deep electrical imaging was carried out for studying
the Ahuachapan–Chipilapa geothermal field (El Salvador) [26]. A total of four resistivity
profiles with a maximum length of 10 km were obtained with a dipole–dipole array config-
uration. In total, two different dipole spacings of 500 m and 1000 m were used, and the
measurements were performed using a 35 KW power generator for injecting a square-wave
current into the ground. The results of the 2D resistivity images helped to reconstruct
the geometry of conductive zones associated with the presence of geothermal fluids and
supported the interpretation of the magnetotelluric soundings carried out in the same area.
Another pioneering work was planned for investigating the volcanic structure of Mt.
Vesuvius (Italy) [27], which is considered the most dangerous volcano in the world. An
apparent resistivity pseudosection was carried out across the N–S profile of Vesuvius; the
length of the profile was 13 km, and the basic electrode spacing about 500 m.
The 2D pseudosection resistivity profile is reported in Figure 3, where a pseudodepth
of about 3 km was depicted. The limit of this application is the absence of a tomographic
inversion procedure. However, the general apparent resistivity pattern defined a roughly
horizontal alternation of conductive and resistive bodies within the investigated depth
 Geosciences 2022, 12, x FOR PEER REVIEW 6 of 19

The 2D pseudosection resistivity profile is reported in Figure 3, where a pseudo-

depthThe 2D pseudosection
of about resistivity
3 km was depicted. Theprofile
limit ofisthis
reported in Figure
application is the3,absence
where of a pseudo-
a tomo-
Geosciences 2022, 12, 438 depth of about 3 km was depicted. The limit of this application is the
graphic inversion procedure. However, the general apparent resistivity pattern definedabsence of a 6tomo-
of 19a
graphic
roughly inversion
horizontalprocedure.
alternationHowever, the general
of conductive apparent
and resistive resistivity
bodies withinpattern defined a
the investigated
roughly horizontal alternation of conductive and resistive bodies within
depth range. The main geoelectrical result highlighted the largely extended, relatively the investigated
depth
range. range.
low resistivity The
The mainzonemain geoelectrical
in the central
geoelectrical result
part
result of thehighlighted
highlighted profile, the largely
closely
the largely extended,
in correspondence
extended, relatively
relatively tolowthe
low resistivity
Somma caldera,
resistivity zone in the
zone inincluding, central
the centralinpart part
the middle, of the profile,
the topclosely
of the profile, terminal closely in correspondence
part of Vesuvius.toFinally,
in correspondence to
the Somma the
these
Somma
caldera, caldera,
results were
including,including,
compared
in theand in the the
middle,
integrated
middle, with
top theSelf
topPotential
terminal terminal partand
(SP)
part of Vesuvius. of Vesuvius.
MT Finally,
ones to
Finally, these
investigate
these results
results
were wereand
the shallow
compared compared and integrated
deepintegrated
and regions ofwith with
the volcanic
Self Self Potential
area.
Potential (SP) and (SP)MTandones
MT ones to investigate
to investigate the
the shallow
shallow and deep
and deep regions
regions of theofvolcanic
the volcanic
area.area.

Figure 3. 2D Apparent resistivity pseudosection carried out on Vesuvius (Reprinted with permis-
Figure 3.3.2D
sion from
Figure 2D Apparent
[27], resistivity
Copyright
Apparent pseudosection
(1998),pseudosection
resistivity Elsevier). carried
carried outout
on on Vesuvius
Vesuvius (Reprinted
(Reprinted withwith permis-
permission
sion from [27], Copyright (1998), Elsevier).
from [27], Copyright (1998), Elsevier).
Another one of the pioneering activities that used the DERT method concerned a
study Another
performed
Another oneofclose
one of
thethe
to pioneering
the site of
pioneering activities
the thatthat
German
activities used the Deep
Continental
used the DERT DERT
method method
Drilling concerned
Project
concerned studya
a (KTB).
study
Storz et
performedperformed
al., 2000toclose
close [28] to the
carried
the site site ofa the
2DGerman
outGerman
of the ERT Continental
profile
Continental using
Deep 44 Deep Drilling
dipoles;
Drilling theyProject
Project (KTB).
adapted
(KTB). the
Storz
etStorz et al., 2000 [28] carried out a 2D ERT profile using 44 dipoles;
classical ERT array configuration using separate injecting and receiving dipoles (Figure
al., 2000 [28] carried out a 2D ERT profile using 44 dipoles; they adapted they
the adapted
classical the
ERT
classical
4). The
array ERT array
length
configuration configuration
of theusing
profile was 22using
separate separate
km, the
injecting injecting
electrode
and andwas
distance
receiving dipolesreceiving
500 m,dipoles
(Figure theThe
4). (Figure
maximum
length
4). The length of the profile was 22 km, the electrode distance was
amplitude of the injected current was 15 A and the square wave period of the current
of the profile was 22 km, the electrode distance was 500 m, the 500
maximum m, the maximum
amplitude of
amplitude
injected
the injected of
into the injected
the
current ground
was 15 wascurrent
A and was
20the 15 temporal
s. square
The A wave
and the square
duration
period wave theperiod
ofcurrent
of the of theinto
registrations
injected current
wasthea
injected
maximum
ground wasinto
of the
2025 ground
s. min.
The temporalwas duration
20 s. The of temporal duration
the registrations wasofa the registrations
maximum was a
of 25 min.
maximum of 25 min.

Figure 4. Schematic description of the field procedure adopted for the deep ERT carried out close to
Figure 4. Schematic description of the field procedure adopted for the deep ERT carried out close to
Figure
the KTB4. Schematic description
drill project of the
facility. field procedure
(Reprinted with adopted for the
permission deep[28],
from ERT carried out close
Copyright to
(2000),
the
theKTB drilldrill
KTB project facility.facility.
project (Reprinted with permission
(Reprinted with from [28], Copyright
permission from (2000),
[28], Wiley-Blackwell
Copyright (2000),
Wiley-Blackwell Publishing Ltd.).
Wiley-Blackwell
Publishing Ltd.). Publishing Ltd.).
The data
The dataprocessing
processing of of968
968voltage
voltagerecordings
recordingswaswascarried
carriedout
outusing
usingthe
theclassical
classical
The data
flowchart processing
based on the of potential
drift 968 voltage recordings
corrections, was carried
recursive notchout using and
filtering the outliers’
classical
flowchart based on the drift potential corrections, recursive notch filtering and outliers’
flowchart based on the
removal, band-pass
band-pass drift potential
filtering corrections,
andstacking.
stacking. torecursive
Due to the low notch filtering and
low signal-to-noise
signal-to-noise outliers’
ratio of the
the
removal, filtering and Due the ratio of
removal, band-pass filtering and stacking. Due to the low signal-to-noise ratio of the
voltage recordings and the choice to accept only useful signals affected by a limited error
(smaller than 10%), more than 50% of the voltage recordings was eliminated. Finally, the
apparent resistivity values were calculated and inverted for obtaining a 2D resistivity image
of the subsurface. A simultaneous iterative reconstruction technique (SIRT) was adopted; it
is based on the discretization of the subsurface into N grid elements, where the cells are
Geosciences 2022, 12, x FOR PEER REVIEW 7 of 19

voltage recordings and the choice to accept only useful signals affected by a limited error
(smaller than 10%), more than 50% of the voltage recordings was eliminated. Finally, the
Geosciences 2022, 12, 438 7 of 19
apparent resistivity values were calculated and inverted for obtaining a 2D resistivity
image of the subsurface. A simultaneous iterative reconstruction technique (SIRT) was
adopted; it is based on the discretization of the subsurface into N grid elements, where
the cells
small are small
compared compared
to the to the dipole
dipole spacing spacing
[29]. The [29]. The
maximum maximumdepth
investigation investigation
reached
depth reached during this first test of deep ERT was about 4 km (Figure 5).
during this first test of deep ERT was about 4 km (Figure 5).

Figure 5.
Figure 5. Deep
Deep ERT
ERT image
image obtained
obtained close
close to
to KTB
KTB facility
facility (Germany);
(Germany); the
the conductive anomaly in
conductive anomaly in the
the
central part of the profile is associated to the presence of a fault zone with fluids. (Reprinted with
central part of the profile is associated to the presence of a fault zone with fluids. (Reprinted with
permission from [28], Copyright (2000), Wiley-Blackwell Publishing Ltd.).
permission from [28], Copyright (2000), Wiley-Blackwell Publishing Ltd.).

The interpretation
The interpretation of of the
theresistivity
resistivityimage
imagewas waswell
wellcorrelated
correlatedwith withotherother geophysi-
geophysical
cal models, with the results of downhole logging in the KTB drillholes
models, with the results of downhole logging in the KTB drillholes and with resistivity and with resistiv-
ity measurements
measurements on core
on core geological
geological sounding
sounding [30,31].
[30,31]. The The spatial
spatial resistivity
resistivity distribution
distribution was
was able
able to reconstruct
to reconstruct the extension
the extension of theofconductive
the conductive structures
structures (ρ < 10 (ρ Ωm)
< 10 with
Ωm) awithsteep a
steep inclination related to the presence of faults. In these highly
inclination related to the presence of faults. In these highly fractured zones, there is the fractured zones, there is
the presence
presence of permeable
of permeable material
material and andfluidfluid circulation.
circulation. The The
rapid rapid increase
increase in theinresistivity
the resis-
tivity values (ρ > 3000 Ωm) at depths between 2 and 3 km
values (ρ > 3000 Ωm) at depths between 2 and 3 km was associated with the presence was associated with the pres-of
enceFalkenberg
the of the Falkenberg
Granite Granite
Massif. Massif.
This work can
This work can be
be considered
considered the the first experimental test
first experimental test of
of the
the DERT method with
DERT method with anan
investigation depth
investigation depth ofof 44 km;
km; however
however the the spatial
spatial resolution
resolution of of the
the resistivity
resistivity pattern
pattern was
was
strongly limited
strongly limited by few apparent
by few apparent resistivity
resistivity values
values obtained
obtained forfor the
the large spacings of
large spacings the
of the
dipole–dipole array
dipole–dipole array system. system.
Colella
Colella etetal.,al.,
20042004 [32] applied
[32] applied the DERT themethod
DERTformethod
obtainingfor theobtaining the first
first high-resolution
high-resolution
resistivity imagingresistivity
of the HighimagingAgriof the High
Valley basinAgri ValleyItaly),
(southern basin representing
(southern Italy), one repre-
of the
senting
more one ofcomponents
complex the more complex components
of the Quaternary faultofnetwork
the Quaternary fault network
of the Apennine of the
chain. During
Apennine
the chain. During
field campaigns, the DERT
six 2D field campaigns, six 2D DERT profiles
profiles perpendicularly orientedperpendicularly
to the longitudinal ori-
ented direction
basin to the longitudinal
were carried basin
out direction
using the were carried out
dipole–dipole using
array the dipole–dipole
configuration. array
The spacing
configuration.
of dipoles was The 200 m,spacing of dipoles
the maximum was 200
distance m, the emitting
between maximum and distance
receiving between
dipolesemit-
was
ting
n andthe
times receiving
basic electrode was n times
dipoles spacing, withthe basic investigation
a mean electrode spacing,
depthwith a mean
of about 500 investi-
m. The
gation depth
lengths of aboutwere
of the profiles 500 in
m.the The lengths
range of 4–7 of km.
the Algorithms
profiles were forin the rangeand
detrending of 4–7 km.
filtering
Algorithms
voltage for detrending
recordings and the FFT andmethod
filtering werevoltage
used forrecordings
extracting and thethe FFTvoltage
useful method were
signals.
usedhigh
The for extracting
resolutionthe of theuseful voltage
electrical signals.
images The highthe
delineated resolution
complexofgeometry
the electrical
of theimages
High
delineated the complex geometry of the High Agri Valley basin (Figure 6).
Agri Valley basin (Figure 6).
Geosciences 2022, 12,
Geosciences2022, x FOR PEER REVIEW
12, 438 8 of 19 8 of 1

Figure
Figure 6. 6. ERTs
ERTs carried
carried outout in the
in the northernmost
northernmost (ERT(ERT 1, 23)and
1, 2 and and3)southernmost
and southernmost
(ERT 4, (ERT
5 and4, 5 and
part of the High Agri Valley basin. The relatively high resistivity contrast between
6) part of the High Agri Valley basin. The relatively high resistivity contrast between the shallow the shallow a
luvial deposits
alluvial depositsandandthethe pre-Quaternary
pre-Quaternary bedrock
bedrock highlights
highlights the geometry
the geometry of the fault-bonded
of the fault-bonded basin. basi
The
The resistivity
resistivity zones
zones on shallow
on the the shallow
layerslayers are associated
are associated with thewith the presence
presence of fans onofthe
fans on the
flanks of flanks
thebasin.
the basin. The
The shape
shape of basin
of the the basin
floorfloor is asymmetric
is asymmetric and theand the maximum
maximum sedimentsediment
thickness thickness
occurs occu
along the northeastern margin (Reprinted with permission from [32], Copyright
along the northeastern margin (Reprinted with permission from [32], Copyright (2004), Elsevier). (2004), Elsevier).

AA comparison
comparison between
between the results
the results of the
of the 2D DERT 2Dprofiles
DERTand profiles
some and some
seismic seismic refle
reflection
tion profiles carried out in the same area confirmed the high spatial resolution
profiles carried out in the same area confirmed the high spatial resolution of the DERT. The of th
electrical imaging highlights the irregular shape of the basin, which is bordered by shallow
DERT. The electrical imaging highlights the irregular shape of the basin, which is bo
faults and filled with Pleistocene alluvial deposits.
dered by shallow faults and filled with Pleistocene alluvial deposits.
In the longitudinal cross-section, the basin appears as a mosaic of fault-bounded blocks
forming In three
the longitudinal cross-section,
different depocenters separatedthebybasin appearshighs.
intrabasinal as a mosaic of fault-bounde
In the transverse
blocks forming three different depocenters separated by intrabasinal
cross-section, the basin is an irregular graben, locally asymmetric to the northeast, and highs. In the tran
verse
with cross-section,
secondary grabensthe basin
due is an irregular
to antithetic graben,
faults. The locallystudy
geophysical asymmetric to the
of the High Agrinortheas
and with
Valley basinsecondary grabens with
has been improved due new
to antithetic
DERT field faults. The geophysical
measurements [33]. One study
of themof the Hig
isAgri
27 km Valley basinanhas
long with been improved
electrode distance ofwith
aboutnew400 DERT
m. Thefield
papersmeasurements
describe the new [33]. One o
them is 27 km long with an electrode distance of about 400 m. The papers describe th
acquisition system composed of a multi-electrode and a multichannel automatic system
with
new12acquisition
channels. Tosystem
date, a 3D resistivity of
composed model of the basin is available
a multi-electrode [16].
and a multichannel automat
The same methodological approach was adopted in other interesting studies concern-
system with 12 channels. To date, a 3D resistivity model of the basin is available [16].
ing the investigation of geothermal areas and groundwater aquifer in a karst environment
locatedThe same
in the methodological
southern Apennine chain approach was adopted in other interesting studies con
(Italy) [34,35].
cerning the investigation of geothermal areas and groundwater aquifer in a karst env
ronment located in the southern Apennine chain (Italy) [34,35].
Balasco et al. [36] applied the DERT and the MT methods to investigate the Atern
Valley (central Italy) that was struck by the 2009 L’Aquila earthquake. The integration o
these two different methods allowed the authors to increase the spatial resolution of th
Geosciences 2022, 12, 438 9 of 19

Geosciences 2022, 12, x FOR PEER REVIEW

Balasco et al. [36] applied the DERT and the MT methods to investigate the Aterno 9 of 19
Valley (central Italy) that was struck by the 2009 L’Aquila earthquake. The integration of
these two different methods allowed the authors to increase the spatial resolution of the
subsurface resistivity images. The MT method is suitable for investigating deep geological
ical structures
structures in a in a depth
depth range range of 0–10
of 0–10 km,km,but but
it is itcharacterized
is characterized by low
by low spatial
spatial resolu-
resolution
tion
in in shallow
shallow subsurface
subsurface investigations.
investigations. Therefore,
Therefore, using the using
DERT themethod
DERT in method in the
the common
common depth range of investigation (01.2 km), a resistivity model
depth range of investigation (01.2 km), a resistivity model with a high resolution could with a high resolu-
tionobtained,
be could be suggesting
obtained, suggesting that integrating
that integrating DERT with DERT MTwith
seems MTtoseems
optimizeto optimize
the spatialthe
spatial resolution of the resistivity subsurface pattern at different
resolution of the resistivity subsurface pattern at different investigation depths. investigation depths.
The length
The lengthof ofthe
theDERT
DERTprofile
profilewas was 8 km,
8 km, thethe electrode
electrode spacing
spacing was was
400 m 400 m the
and andmax-
the
maximum
imum distance
distance betweenbetween the injecting
the injecting and receiving
and receiving systemssystems
was eight wastimeseight
the times the
electrode
electrodeThe
spacing. spacing. The measurements
measurements were performed werewith
performed with a dipole–dipole
a dipole–dipole array con-
array configuration. The
figuration.
system The system
consisted consisted ofstation,
of a transmitting a transmitting station, awhich
which injected injected current,
square-wave a square-wave
with a
current, with a maximum energizing current of 5A, into the
maximum energizing current of 5A, into the ground and a multichannel receiver systemground and a multichannel
receiver system
composed of fourcomposed
remotely of four remotely
multichannel multichannel
dataloggers dataloggers
controlled controlled
by a laptop. by a
For each
laptop. For
current each current
injection, injection,
eight voltage eight voltage
recordings lastingrecordings
from 5 tolasting
20 minfromwere5simultaneously
to 20 min were
simultaneously
acquired. A totalacquired. A total
of 112 voltage of 112 voltage
recordings, relatedrecordings, related
to the different to theofdifferent
position posi-
the electrodes
tion ofthe
along theprofile,
electrodes
werealong the profile,
collected. The data were collected.was
processing Thecarried
data processing
out using the wasclassical
carried
out usingand
stacking theFFT
classical stacking and FFT methods.
methods.
The apparent resistivity values
The apparent resistivity values werewere inverted
inverted by by means
means of of the
the inversion
inversion algorithm
algorithm
[20,37]. Figure 7 shows the resistivity tomographic image that
DCIP2D [20,37]. Figure 7 shows the resistivity tomographic image that is characterized
is characterized by
by resistivity
resistivity values
values varying
varying fromfrom10 Ωm
10 Ωm to over
to over 2500 ΩmΩm
2500 andand
manymany lateral
lateral discontinui-
discontinuities
ties related
related to manyto many regional
regional faults,
faults, bothboth thrusts
thrusts and and successive
successive extensional
extensional features.
features. To
To the
the end
end of theofDERT
the DERT profile,
profile, several several
highlyhighly resistive
resistive (ρ > 1000 (ρ >Ωm)
1000 Ωm) separated
zones, zones, separated
by relativeby
conductive (200–500 Ωm)
relative conductive (200–500 Ωm) areas, are observed, testifying a complex tecton-
areas, are observed, testifying a complex tectonic-stratigraphical
ic-stratigraphical
structure. At about structure. At about
7000 m along 7000 mprofile,
the DERT along the the sharp
DERTvertical
profile,resistivity
the sharpcontrast
vertical
resistivity contrast highlights the Paganica Fault.
highlights the Paganica Fault.

Figure 7. DERT image elaborated by the DCIP2D inversion code. Major features are labeled with
Figure 7. DERT image elaborated by the DCIP2D inversion code. Major features are labeled with
letters. (A) represents alluvial deposits and the Messinian turbidite complex; (B) Unfractured cal-
letters. (A) represents alluvial deposits and the Messinian turbidite complex; (B) Unfractured
careous block. The more resistive blocks are probably associated to Miocene units. Black dashed
calcareous block.
lines indicate The more
presumed resistive
faults [36]. blocks are probably associated to Miocene units. Black dashed
lines indicate presumed faults [36].
In this area of Central Italy, Pucci et al. [38] studied and improved the knowledge of
In this area of Central Italy, Pucci et al. [38] studied and improved the knowledge
the complex geometry of the Paganica–San Demetrio basin using the DERT method.
of the complex geometry of the Paganica–San Demetrio basin using the DERT method.
They carried out the 2D DERT measurements along three different profiles perpendicu-
They carried out the 2D DERT measurements along three different profiles perpendicularly
larly oriented to the eastern edge of the Quaternary basin and the Paganica fault (Figure
oriented to the eastern edge of the Quaternary basin and the Paganica fault (Figure 8).
8).
Geosciences 2022,
Geosciences 12, 438
2022, 12, x FOR PEER REVIEW 10
10 of 19
of 19

Figure 8.
Figure 8. (a)
(a)Model
Modelofofthe
thefirst
firstERT
ERTprofile obtained
profile using
obtained a Wenner-alpha
using a Wenner-alpha array configuration;
array (b)
configuration;
ERTERT
(b) interpretation of of
interpretation thethe
tomographic
tomographicimage
imagewith
withthe
thebedrock/basin
bedrock/basininfill
infillbodies
bodiesand
and main
main faults
faults
highlighted (Reprinted
highlighted (Reprinted with
with permission
permission from
from [38],
[38], Copyright
Copyright (2016)
(2016) Oxford
Oxford University
UniversityPress).
Press).

A multi-electrode
A multi-electrode2D 2Ddevice
devicebybymeansmeans of of
thethe
ABEMABEM Terrameter
Terrameter SAS-4000
SAS-4000 instru-
instrument
ment connected
connected to ABEM to ABEM
ES1064ES1064 C multiplexer
C multiplexer was adopted
was adopted for the
for the field datafield data acquisi-
acquisition. They
tion. aThey
used used
2.52 km a 2.52
long cable,km long
with cable,
a set of 64with a setsteel
stainless of 64electrodes
stainlessandsteel electrodes
offset of 40 m,and
andoffset
a roll-
of 40 m,
along and afor
scheme roll-along scheme for
data acquisition withdata acquisition and
Wenner-alpha withpole-dipole
Wenner-alpha arrayand pole-dipole
configurations.
array configurations. The length of the three profiles was,
The length of the three profiles was, respectively, 2.52 km, 6.38 km and 3.8 km. respectively, 2.52 km, 6.38 km
and 3.8 km.
They processed apparent resistivity data using the X2IPI software for data filtering,
whileThey
dataprocessed
inversion apparent
was carried resistivity
out withdata using thesoftware
RES2DINV X2IPI software
[2]. Thefor data filtering,
interpretation of
while
the 2Ddata inversion
resistivity was with
models carried out withconstraints
geological RES2DINVcontributed
software [2]. to The
better interpretation
describing the of
the 2Dofresistivity
shape the basin.models with suggested
The results geologicalaconstraints
southeastward contributed
deepening to better
of thedescribing
Paganica–San the
shape of the
Demetrio basin.
Basin from The results suggested
∼200–300 m to the maximuma southeastward
depocenter deepening
of ∼600 m, of the Paganica–San
largely exceeding
Demetrio
the knownBasin fromof∼200–300
thickness m to the
the continental maximum
sequence. depocenter
Complex lateralofand ∼600 m, largely
vertical heteroge-ex-
ceeding
neous the known
resistivity thickness
regions of the
suggested thecontinental
existence ofsequence. Complex
several tectonic lateralthat
features and vertical
displaced
heterogeneous
the resistivity regions
original Meso-Cenozoic suggested
multilayer the existence
producing of several tectonic
buried cumulative landformsfeatures that
and con-
displaced
trolling thethe original Meso-Cenozoic
development of depressions,multilayer
later filledproducing
with sequencesburiedofcumulative
Plio-Quaternaryland-
forms and controlling
continental deposits. The thegeological
development of depressions,
interpretation later filled with
of the tomographic imagessequences
showedof a
Plio-Quaternary
structural continental
style consisting deposits.
of SW dipping, Theleading
geological
normal interpretation
faults fromof the tomographic
which NE dipping,
images showed
antithetic a structural
faults splay style consisting
out, defining grabens actingof SWasdipping,
sedimentary leadingtraps.normal faults from
which NE dipping, antithetic faults splay out, defining grabens acting
Carrier et al. [39] adopted, for the first time, the new Fullwaver as sedimentary
instrument of the Iris
traps.
company for exploring the geothermal field in the Greater Geneva Basin (Switzerland). A
Carrier
2D DERT et al. with
profile [39] adopted,
a length for thekm
of 4.8 firstand
time,anthe new Fullwaver
investigation depth instrument
of about 1ofkm thewasIris
companyinfor
planned anexploring
urbanizedthe areageothermal
affected by field in the Greater
electrical noise. Geneva Basin (Switzerland). A
2D DERT
The new profile
systemwithconsists
a length of of
25 4.8 km and an
V Fullwaver investigation
nodes depth
for receiving of about
voltage 1 km
signals andwas1I
Fullwaver nodes for the injection of the
planned in an urbanized area affected by electrical noise.current with reverse polarity into the subsurface
with The
a maximum
new system intensity of 10ofA.25
consists ForV each receiving
Fullwaver node,
nodes forareceiving
set of three alignedsignals
voltage electrodes
and
with spacing of 50 m is installed. This new system is extremely
1I Fullwaver nodes for the injection of the current with reverse polarity into the subsur- user-friendly and can be
easily adopted
face with for DERTintensity
a maximum surveys of in 10
urban areas.
A. For Furthermore,
each receiving node, the system
a set of does not aligned
three require
long cableswith
electrodes and spacing
any fixed ofarray
50 mconfigurations,
is installed. This 2Dnewandsystem
3D DERT surveys can
is extremely be carried
user-friendly
out
and can be easily adopted for DERT surveys in urban areas. Furthermore, the voltage
without any geometrical constraints. During the field surveys, a set of 2318 system
recordings was obtained; a post-processing and a quality control of the data were carried
does not require long cables and any fixed array configurations, 2D and 3D DERT sur-
out using the PROSYS software and ERT-Lab [40]. After this step, a total of 1746 useful
veys can be carried out without any geometrical constraints. During the field surveys, a
voltage signals were selected and used for estimating the apparent resistivity values. Finally,
set of 2318 voltage recordings was obtained; a post-processing and a quality control of
the BERT algorithm [37] allowed the authors to obtain the 2D resistivity model (Figure 9).
the data were carried out using the PROSYS software and ERT-Lab [40]. After this step, a
Geosciences 2022, 12, x FOR PEER REVIEW 11 of 19

total of 1746 useful voltage signals were selected and used for estimating the apparent
Geosciences 2022, 12, 438
resistivity values. Finally, the BERT algorithm [37] allowed the authors to obtain the 2D
11 of 19

resistivity model (Figure 9).

Figure 9. 9.
Figure (a):(a):Results
Resultsof
of the deepERT
the deep ERTtomographic
tomographic inversion.
inversion. (b) Coverage
(b) Coverage (log10)(log10) of the penetra-
of the penetration
tiondepth
depth (Reprinted with permission from [39], Copyright (2019) Oxford University
(Reprinted with permission from [39], Copyright (2019) Oxford University Press). Press).

TheTheresult
resultofofthe
the2D2D DERT highlightsa ahigh-resistivity
DERT highlights high-resistivity zonezone in the
in the SE part
SE part of theof the
profile and at a depth between 250 m and 650 m. In the central part of the DERTthe
profile and at a depth between 250 m and 650 m. In the central part of the DERT profile, profile,
low resistivity values are related to the presence of groundwater, while
the low resistivity values are related to the presence of groundwater, while the low re- the low resistivity
values at greater depths crosscut the Upper Cretaceous–Lower Cenozoic limits. For this
sistivity values at greater depths crosscut the Upper Cretaceous–Lower Cenozoic limits.
study, the good agreement between the results obtained with the DERT method with other
Formore
this robust
study,andthe consolidated
good agreement between
geophysical the results
methods, obtained
the gravity and with
seismictheexploration
DERT method
with other more robust
methods, is remarkable. and consolidated geophysical methods, the gravity and seismic
exploration methods,
Lajaunie is remarkable.
et al. [41] applied the deep ERT method for investigating the Sèchilienne
Lajaunie
slope, which et al. [41]
is one of theapplied the most
largest and deepactive
ERT landslides
method for investigating
in the European Alps the(France).
Sèchilienne
slope, which is one of the largest and most active landslides in the European Alps
The geological and hydrogeological settings of this complex landslide were largely studied
(France). The geological
with different geophysicaland hydrogeological
methods, but the surveyssettings
producedofonly
this1Dcomplex landslide
and 2D images of the were
subsurface. This work represents one of the first applications of the
largely studied with different geophysical methods, but the surveys produced only 1D deep 3D ERT method
to investigate a landslide area. They obtained a 3D resistivity model of the Sèchilienne
and 2D images of the subsurface. This work represents one of the first applications of the
slope up to a depth of 500 m.
deep 3D TheERT methodfield
geophysical to investigate
campaign was a landslide
carried outarea.
using They obtained
a FullWaver a 3D
system resistivity
made of
model of the Sèchilienne slope up to a depth of 500 m.
23 V receiving nodes with two orthogonal MN dipoles with spacing of 50 m and a 1 I
The geophysical
energizing node withfield campaign
a fixed electrodewasand carried
a movingout using aB FullWaver
electrode along 30 selectedsystem made of
points
23 within
V receiving nodes area.
the landslide withThetwoPROSYS
orthogonal
and theMN BERTdipoles with
software spacing
were of the
used for 50 voltage
m and a 1 I
energizing node with
data processing a fixed
and the electrode
resistivity and a moving
data inversion, electrode
respectively B along
(Figure 10). 30 selected points
within the landslide area. The PROSYS and the BERT software were used for the voltage
data processing and the resistivity data inversion, respectively (Figure 10).
Geosciences 2022,
Geosciences 12, 438
2022, 12, x FOR PEER REVIEW 1212of
of 19
19

Figure 10. 3D
Figure 10. 3D representation
representation of the interpreted
of the interpreted model
model superimposed
superimposed on
on the
the geological
geological map
map of
of the
the
slope. The squares represent the resistivity (log10) of the surface waters sampled in spring or at
slope. The squares represent the resistivity (log10) of the surface waters sampled in spring or at
surface runoff locations. The black lines represent the major faults. (Reprinted with permission
surface runoff locations. The black lines represent the major faults. (Reprinted with permission
from [41], Copyright (2019) Oxford University Press).
from [41], Copyright (2019) Oxford University Press).
The 3D model is characterized by two different resistivity anomalies. An anomaly
The 3D model is characterized by two different resistivity anomalies. An anomaly with
with high resistivity values was identified in the western sector of the model and associ-
high resistivity values was identified in the western sector of the model and associated with
ated
very with veryand
fractured fractured and dryThe
dry material. material.
absenceThe absence
of the waterof the water
sources sources
in this in this area
area confirms the
confirms the high of
high permeability permeability of the slope
the slope material. material.
At the surface,At thethe surface,
high the high
resistivity valuesresistivity
seem to
values seem
be aligned to be
with thealigned with the
major faults. Themajor faults. The conductive-zone
conductive-zone anomaly at the east anomaly
of theatlandslide
the east
of the landslide (Sabot fault) was detected and interpreted as a perched
(Sabot fault) was detected and interpreted as a perched aquifer. On the basis of geological aquifer. On the
basis of geological and hydrogeological studies, the unstable zones are
and hydrogeological studies, the unstable zones are correspond with the resistive anomaly. correspond with
the resistive anomaly. Notwithstanding the poor spatial resolution
Notwithstanding the poor spatial resolution of the resistivity measurements, the results of the resistivity
measurements, the results
contributed to better contributed
describing to better describing
the hydrogeological settingthe hydrogeological
of the investigated area. setting of
the investigated area.
Troiano et al. [42] studied the complex geometry of the central sector of the Campi
Troiano
Flegrei calderaet al. [42] studied
(Italy), applyingthe complex
a deep geometry
3D ERT of the central
tomographic sectorThey
approach. of theadopted
Campi
Flegrei caldera (Italy), applying a deep 3D ERT tomographic approach.
the new Iris FullWaver instrument system consisting of 12 dual-channel receivers with They adopted the
new Iris FullWaver instrument system consisting of 12 dual-channel
a sampling rate of 10 ms and an Iris VIP 10,000 electrical transmitter for injecting DC receivers with a
sampling
current into rate
theofground,
10 ms andwithan Iris VIPof10,000
a change electrical
polarity every 2transmitter for injecting
s. In a selected area of theDC cur-
Campi
rent intocaldera,
Flegrei the ground, with
the field a change of polarity
measurements everyout
were carried 2 s.with
In a aselected area ofdistribution
dense spatial the Campi
Flegrei caldera,
of receiving the field
stations; for measurements
each point, twowere carried
dipoles without with a dense
a spacing of 100spatial distribution
m perpendicularly
of receiving stations; for each point, two dipoles with a spacing of
oriented were installed. The scholars adopted the principal component decomposition 100 m perpendicularly
oriented
method towere installed.
remove The scholars
the electrical noise andadopted the principal
extracting the usefulcomponent decomposition
voltage signals.
methodThetoinversion
remove the electrical
of the apparent noise and extracting
resistivity the carried
values was useful voltage
out usingsignals.
the commercial
The inversion
software ERTlab3D,ofand the aapparent resistivity
3D resistivity values
imaging of was carried out using
the investigated area wasthe commer-
obtained
cial software
(Figure 11). TheERTlab3D,
maximum andinvestigation
a 3D resistivitydepthimaging
was aboutof the
400investigated
m, which was areaevaluated
was ob-
tained
using a(Figure
robust and11).quantitative
The maximum methodinvestigation depth
for evaluating thewas
depth about 400 m, which was
of investigation.
Geosciences 2022, 12, x FOR PEER REVIEW 13 of 19

Geosciences 2022, 12, 438 13 of 19

evaluated using a robust and quantitative method for evaluating the depth of investiga-
tion.

Figure 11.11.
Figure (a)(a)
Electrical
Electricalresistivity 3D model
resistivity 3D modelobtained
obtained through
through thethe
deepdeep
ERTERT imaging.
imaging. (b) Electrical
(b) Electrical
resistivity
resistivity cross-sections along selected traces. (c) Map of the area resolved by the ER survey [42]. [42].
cross-sections along selected traces. (c) Map of the area resolved by the ER survey

TheThe3D 3Dresistivity
resistivity model
model ofofthe theinvestigated
investigated areaarea clearly
clearly identifies
identifies two verytwo very conduc-
conductive
tive zones: the Solfatara maar and the Agnano Plain. The high resolution of the tomo-
zones: the Solfatara maar and the Agnano Plain. The high resolution of the tomographic
image allowed the authors to reconstruct the geometry of these structures. Furthermore, a
graphic image allowed the authors to reconstruct the geometry of these structures. Fur-
resistive zone is clearly visible with a conical shape that is associated with the Pisciarelli
thermore, a resistive zone is clearly visible with a conical shape that is associated with the
fumarole. The quality of the results confirms the capacity of deep ERT to contribute to
Pisciarelli
describing fumarole.
the geometry The quality of the
of geological results confirms
structures the capacity
in active volcanic areas. of deep ERT to con-
tribute Mazzini
to describing the geometry of geological structures in
et al. [43] applied the deep ERT method for studying an interesting active volcanic areas.
volcanic area
Mazzini
located etEast
in the al. [43]
Javaapplied the deep ERT
Basin representing method
the largest for studying
eruptive an interesting
clastic system volcanic
on Earth. The
area located in
geoelectrical the East
survey Java Basin
was planned representing
for investigating the largest
the Lusi eruptive eruptive
site, whereclastic system on
a geothermal
Earth.
systemThewasgeoelectrical
pierced onto the survey
surface was
alongplanned for investigating
the Watukosek fault system in the
May Lusi
2006.eruptive site,
where aDuring
geothermal
the fieldsystem was pierced
campaigns, the Fullwaveronto system
the surface
of thealong the Watukosek
IRIS company was usedfault
for sys-
carrying out 3D deep ERT in an area of 15 km2 . The main characteristics and performances
tem in May 2006.
of the hardware and the software of this system were already described in the previous
During the field campaigns, the Fullwaver system of the IRIS company was used for
paragraphs. A set of 25 V Fullwaver receiving nodes and 1 I Fullwaver energizing nodes
carrying
was used outfor3D deep ERT
obtaining 3596 in an area
values of theof 15 km
useful
2. The main characteristics and perfor-
voltage signals. After the pre-processing
mances of the procedures
and filtering hardware to and the software
remove the voltage ofrecordings
this system withwere
a veryalready described in the
low signal-to-noise
previous paragraphs.
ratio, a set of 2331 values A setwasofselected
25 V Fullwaver
to perform receiving
the inversion nodes
with and 1 I Fullwaver
the BERT software. ener-
gizing nodes
The main was usedoffor
results theobtaining
3D ERT opened 3596 thevalues
way of forthe useful voltage
investigating signals.
the possible After the
correla-
pre-processing
tion between the and filtering
presence of procedures
the Watukosek to fault
remove the the
system, voltage recordings
movements of the with
deep a very low
fluids
signal-to-noise ratio, a set of 2331 values was selected to perform the inversion with the
and the formation of the Lusi geothermal system (Figure 12). Furthermore, the results have
great
BERT implications for supporting the local authorities in planning and managing hazard
software.
mitigation activity.
The main results of the 3D ERT opened the way for investigating the possible cor-
relation between the presence of the Watukosek fault system, the movements of the deep
fluids and the formation of the Lusi geothermal system (Figure 12). Furthermore, the
Geosciences 2022, 12, x FOR PEER REVIEW 14 of 19

Geosciences 2022, 12, 438 14 of 19

results have great implications for supporting the local authorities in planning and
managing hazard mitigation activity.

Figure 12.12.
Figure 3D3Dprojection of the
projection acquired
of the deep
acquired electrical
deep resistivity
electrical tomography
resistivity tomography data. Extracted
data. Extractedcells
cells
with
withresistivities
resistivitiesbelow
below2.5
2.5 Ohm.m underneath
underneaththe theactive
active vents
vents highlight
highlight the area
the area of collapse
of collapse (dashed
(dashed black arrows).
black arrows). The red-shaded
The red-shaded area indicates
area indicates the Watukosek
the Watukosek faultthat
fault system system that intersects
intersects Lusi [43].
Lusi [43].
Rizzo et al. [44] carried out an innovative experiment with the 3D deep ERT method
forRizzo et al. [44]
exploring carried out geothermal
the Larderello an innovative experiment
area with
(Italy), the the 3D
oldest deep ERTfield
geothermal method
in the
forworld
exploring
underthe Larderelloforgeothermal
exploitation area (Italy),
power production. the first
For the oldest
time,geothermal field in the
a 3D Surface-Hole Deep
world underResistivity
Electrical exploitation for power(SH-DERT)
Tomography production. wasFor the first
designed andtime, a 3D
carried outSurface-Hole
for studying a
Deep Electrical
complex Resistivity
geothermal fieldTomography
and operating(SH-DERT) was designed and
in extreme environmental carried out for
conditions.
studying a complex geothermal field and operating in extreme environmentalanconditions.
The injecting and receiving systems were distributed on a surface with extension of
2 and in the Venelle-2 well with a depth of 1.6 km. A log cable with the capacity to
The injecting and receiving systems were distributed on a surface with an extension
6 km
of operate
6 km2 and in the
at high Venelle-2 well
temperatures waswith
useda to
depth
manageof 1.6 km.
and A log12
control cable with thesteel
cylindrical capacity to
electrodes
operate
with aatdiameter
high temperatures
of 20 mm and was used to
a length of manage
1.5 m. Allandthecontrol 12 cylindrical
measurements steel elec-
were carried out by
trodes
meanswith
of aamulti-channel
diameter of 20datalogger
mm and asynchronized
length of 1.5withm. All
thethe measurements
injecting were car-
systems (square-wave
of DC current with a maximum intensity of 10 A and polarity
ried out by means of a multi-channel datalogger synchronized with the injecting inversion of 32 s).systems
A dipole–
dipole array configuration was adopted with a maximum distance
(square-wave of DC current with a maximum intensity of 10 A and polarity inversion of between injecting and
receiving dipoles of 1.600 m.
32 s). A dipole–dipole array configuration was adopted with a maximum distance be-
The dataand
tween injecting processing
receivingofdipoles
the voltage signals
of 1.600 m. recorded during the field surveys was
performed
The data processing of the voltage signals filtering
with classical methods (detrending, recordedand FFT technique).
during A total of
the field surveys was10%
of the data
performed collected
with was
classical removed
methods due to the very
(detrending, lowand
filtering signal-to-noise
FFT technique).ratio.ATototal
perform
of
the resistivity data inversion, the authors applied the ERT-Lab software. In a first step,
10% of the data collected was removed due to the very low signal-to-noise ratio. To per-
they inverted only the apparent resistivity data estimated with data coming from dipoles
form the resistivity data inversion, the authors applied the ERT-Lab software. In a first
installed on the surface; in a second step, they used all data available, including the apparent
step, they inverted only the apparent resistivity data estimated with data coming from
resistivities calculated using the sensors in the borehole too (Figure 13). The investigation
dipoles installed on the surface; in a second step, they used all data available, including
depth of the surface 3D DERT was about 800 m, while the 3D SH-DERT was 1600 m. A
the apparent resistivities calculated using the sensors in the borehole too (Figure 13). The
shallow conductive nucleus (ρ < 10 Ωm) bound by zones with relatively high resistivity
investigation depth of the surface 3D DERT was about 800 m, while the 3D SH-DERT was
zones is clearly identified. The DERT measurements carried out with electrodes in the
1600 m. A shallow conductive nucleus (ϱ < 10 Ωm) bound by zones with relatively high
borehole and on the surface strongly increase the spatial resolution of the 3D resistivity
resistivity zones is clearly identified. The DERT measurements carried out with elec-
pattern and are of great interest for the scientific community involved in the study of
trodes in the borehole and on the surface strongly increase the spatial resolution of the 3D
geothermal fields.
resistivity pattern and are of great interest for the scientific community involved in the
study of geothermal fields.
Geosciences 2022, 12,
Geosciences 2022, 12, x438
FOR PEER REVIEW 1515ofof 19
19

Figure
Figure 13.
13. (a) 3D SH-DERT;
(a) 3D SH-DERT;(b) (b)Resistivity
Resistivityisosurfaces
isosurfacesobtained
obtained using
using only
only surface
surface electrodes;
electrodes; (c) (c)
3D
3D full-data
full-data DERT;
DERT; (d)(d) Resistivity
Resistivity isosurfacesobtained
isosurfaces obtainedusing
usingboth
both surface
surface and
and borehole
borehole electrodes.
electrodes.
(Reprinted
(Reprinted with
with permission
permission from
from [44],
[44], Copyright
Copyright (2022,
(2022, Elsevier).
Elsevier).

4.
4. Discussion
Discussion
To date, the
To date, theapplication
applicationofof
thethe DERT
DERT method
method still still remains
remains limited
limited to case
to a few a few case
studies
studies
with anwith an investigation
investigation depth greater
depth greater than
than 500 m. 500
Them. Thecharacteristics
main main characteristics
of the of the
DERT
DERT applications
applications discussed
discussed in the previous
in the previous paragraph
paragraph are summarized
are summarized in Table in1.Table 1.

Table
Table 1.1. Simplified
Simplifiedclassification
classificationofofthe
themain
main characteristics
characteristics (geological
(geological context,
context, array
array configura-
configurations,
tions, algorithms
algorithms forprocessing
for data data processing and inversion,
and inversion, investigation
investigation depth) ofdepth) of the
the DERT DERT applications
applications presented
presented and discussed
and discussed in Section in
3. Section 3.
Geological Context, Array and Electrode Investigation
Authors Geological Context, Array and Electrode Processing; Inversion Investigation Depth
Authors Country
Country Spacing
Spacing
Processing; Inversion Depth
Geothermal, El Sal- dipole–dipole, 500 m up 2 km (pseudo-
Floresetetal.,
al.,1997
1997 Geothermal, El dipole–dipole, 500 m pseudosection
Flores vador
Salvador uptoto 1000 m
1000 m
pseudosection depth)
2 km (pseudodepth)
Di Maio et al., 1998 Volcanic, Italy 500 m stacking; pseudsection 3 km
3 km (pseudo-
(pseudodepth)
Di Maio et al., 1998 Volcanic, Italy 500 m500 m stacking;
stacking;pseudsection
Storz et al., 2000 Faults, Germany dipole–dipole, SIRT depth)
4 km
detrending, filtering and
Storz et
Colella al.,2004
et al., 2000 Faults,
Faults,Germany
Italy dipole–dipole,
dipole–dipole, 200500
mm stacking; SIRT 0.54 km
km
FFT; RES2DINV
detrending, filtering and
Colellaetetal.,
Balasco al.,2011
2004 Faults, Italy
Geothermal, Italy dipole–dipole,
dipole–dipole, 400200
mm
stacking and FFT 0.5km
1.2 km
FFT; RES2DINV
method; DCIP2D
Wenner-alpha and stacking and FFT
X2IPI software method;
for data
Balasco
Pucci et et
al.,al.,
20162009 Geothermal,
Faults, Italy Italy dipole–dipole,
pole–dipole, 40 400
m m filtering;DCIP2D
RES2DINV 1.2 km
~0.75 km
Geothermal, Full-wave viewer
Carrier et al., 2019 Wenner-alpha
dipole–dipole,and
50 mpole– X2IPI software for data ~1 km
Pucci et al., 2016 Faults, Italy
Switzerland software; BERT ~0.75 km
dipole, 40 m filtering; RES2DINV
Full-wave viewer
Lajaunie et al., 2019 Landslide, France dipole–dipole, 50 m 0.5 km
Geothermal, Swit- Full-wave viewer soft-
software; BERT
Carrier et al., 2019 dipole–dipole, 50 m ~1 km
Troiano et al., 2019 zerland
Volcanic, Italy 100 m
Principalware; BERT
component
0.4 km
decomposition, ERTlab3D
Full-wave viewer soft-
Launie et
etal.,
al.,2021
2019 Landslide, France dipole–dipole, 50 m Full-wave viewer 0.5kmkm
Mazzini Volcanic, Indonesia dipole–dipole, 50 m ware;BERT
software; BERT 0.5
Principal component
detrending, filtering and de-
Troiano
Rizzo et et
al.,al.,
20222019 Volcanic, Italy
Geothermal, Italy 100 m 50 m
dipole–dipole, 0.4km
1.6 km
composition, ERTlab3D
FFT; ERTlab
Full-wave viewer soft-
Mazzini et al., 2021 Volcanic, Indonesia dipole–dipole, 50 m 0.5 km
However, in the last five years, this topic raised ware; BERT attention in the scientific
important
community, with an increase in interesting and detrending,
challengingfiltering
works.and
The main applications
Rizzo et al., 2022 Geothermal, Italy dipole–dipole, 50 m 1.6 km
concern the investigation of the complex geometryFFT; of theERTlab
seismically active structures, the
mapping of fluids and gases in volcanic zones and the exploration of the geothermal fields
However,
for energy in the last
production. five years,
Moreover, this topic
in these recentraised important
papers, attention
technological in the permitted
advances scientific
community, with an increase in interesting and challenging works. The main applica-
tions concern the investigation of the complex geometry of the seismically active struc-
Geosciences 2022, 12, 438 16 of 19

an increase in the maximum investigation depth. For this reason, we can define a survey
carried out with an exploration depth of at least 500 m as deep ERT (DERT) investigation.
As it concerns the field survey, distributed and high-spatial-density sensor networks of
measuring and energizing dipoles are generally adopted. The measuring dipoles are equipped
with instruments to record low-voltage signals (10−3 –10−6 V), while a mobile system equipped
with a high-electrical-power generator is used for the injection of the current square wave
(max 20 A) into subsoil with the polarity regularly inverted. The frequency of these inversions
is extremely low, and the approximation of the direct electrical current (f = 0) is acceptable.
Furthermore, there is great attention paid to the sensors realized with materials to ensure
an optimal contact between electrodes and hosting to reduce the anthropic noise, especially
when they are installed in boreholes. All the sensors for measuring electrical potential are
remotely controlled by means of a multi-channel datalogger.
As it concerns methodological advances, tomographic inversion is carried out with
robust and well-assessed algorithms used for the application of the ERT method in near-
surface investigations. Recently, there has been a growing interest in open-source soft-
ware based on Python tools and extraordinary opportunities related to the use of high-
performance computing facilities. Then, there are no limits or critical problems in applying
the large spectrum of commercial and open-sources software already available for DERT
data inversion.
In this scenario, a critical aspect is the absence of innovative algorithms for the data
processing and filtering of the voltage recordings with a low signal-to-noise ratio. They are
generally based on standard statistical and spectral decomposition techniques, as well as
detrending, outlier removal and Fourier analysis for evaluating the amplitude of the useful
voltage signals.
One of the most promising future directions of research on the DERT method could
be represented by the application of Machine Learning algorithms for noise removal and
signal identification in the voltage recordings [45–48]. The novel distributed sensors that
are remotely controlled can acquire a long time series of voltage measurements, and this
large amount of data could be easily processed with Machine Learning algorithms.
Another promising research activity could be based on the application of advanced
statistical approaches for better characterizing the inner dynamical structure of the electrical
noise that is generally assumed to be Gaussian, although it is widely recognized that this
assumption is too simple for many geophysical signals [49–51].

5. Conclusions
In this work, a critical analysis of the main advantages and limitations concerning the
applications of the deep ERT method is presented and discussed.
The main technological advances are related to the use of a dense network of receiving
and injecting dipoles that can easily installed on the surface and in boreholes without
any geometrical limitations. The dipoles are decoupled and separately distributed, so
that the connection with long cables is not necessary. Furthermore, a new class of power
generators for injecting DC currents into the ground with a maximum intensity of 20A is
available. These technological advances make it possible to increase the distance between
emitting and receiving systems and, consequently, to investigate the resistivity pattern of
the subsurface at greater depths.
The main limitation of the DERT method is related to the degradation of the quality of
voltage recordings when the distance between the dipoles increases. The amplitude of the
useful voltage signals generated by the current strongly decrease (~1/r3 ) and it becomes
gradually contaminated by the electrical noise; the extreme low signal-to-noise ratio makes
an accurate estimate of the useful signal impossible.
To reduce the errors associated with the extraction of the useful signals from the
voltage time series and, as a consequence, to obtain apparent resistivity values for deep
layers, it is mandatory to adopt advanced statistical and mathematical tools. One of the
most promising research directions is the introduction of algorithms based on Artificial
Geosciences 2022, 12, 438 17 of 19

Intelligence and Machine Learning methods that could strongly improve the ability to
process large numbers of voltage recordings.
Finally, an improvement of the DERT investigation depth (>1 km) could open the
way for a more relevant contribution for studying a wide class of challenging geological
problems (e.g., monitoring CO2 storage, migration of deep gases and fluids in volcanic
areas, evaluation of the ice thickness in polar regions).

Author Contributions: Conceptualization, M.B.; V.L.; E.R.; L.T.; methodology, M.B.; V.L.; E.R.; L.T.;
writing—original draft preparation, M.B.; V.L.; E.R.; L.T.; supervision and project administration, V.L.
All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Capello, M.A.; Shaughnessy, A.; Caslin, E. The Geophysical Sustainability Atlas: Mapping geophysics to the UN Sustainable
Development Goals. Lead. Edge 2021, 40, 1. [CrossRef]
2. Loke, M.H.; Chambers, J.E.; Rucker, D.F.; Kuras, O.; Wilkinson, P.B. Recent developments in the direct-current geoelectrical
imaging method. J. Appl. Geophys. 2013, 95, 135–156. [CrossRef]
3. Loke, M.H.; Barker, R.D. Rapid least-squares inversion of apparent resistivity pseudosections using a quasi-Newton method.
Geophys. Prospect. 1996, 44, 131–152. [CrossRef]
4. Loke, M.H.; Barker, R.D. Practical techniques for 3D resistivity surveys and data inversion. Geophys. Prospect. 1996, 44, 499–523.
[CrossRef]
5. Binley, A.; Henry-Poulter, S.; Shaw, B. Examination of solute transport in an undisturbed soil column using electrical resistance
tomography. Water Resour. Res. 1996, 32, 763–769. [CrossRef]
6. Perrone, A.; Lapenna, V.; Piscitelli, S. Electrical resistivity tomography technique for landslide investigation: A review. Earth Sci.
Rev. 2014, 135, 65–82. [CrossRef]
7. Bellanova, J.; Calamita, G.; Giocoli, A.; Luongo, G.; Macchiato, M.; Perrone, A.; Uhlemann, S.; Piscitelli, S. Electrical resistivity
imaging for the characterization of the Montaguto landslide (southern Italy). Eng. Geol. 2018, 243, 272–281. [CrossRef]
8. Finizola, A.; Revil, A.; Rizzo, E.; Piscitelli, S.; Ricci, J.; Morin, B.; Angeletti, L.; Mocochain, L.; Sortino, F. Hydrgeological insights
at Stromboli volcano (Italy) from geoelectrical, temperature, and CO2 soil degassing investigations. Geophys. Res. Lett. 2006,
33, L17304. [CrossRef]
9. Bergmann, P.; Schmidt-Hattenberger, C.; Labitzke, T.; Wagner, F.M.; Just, A.; Flechsig, C.; Rippe, D. Fluid injection monitoring
using electrical resistivity tomography five years of CO2 injection at Ketzin, Germany. Geophys. Prospect. 2017, 65, 859–875.
[CrossRef]
10. Oldenborger, G.A.; LeBlanc, A.M. Monitoring changes in unfrozen water content with electrical resistivity surveys in cold
continuous permafrost. Geophys. J. Int. 2018, 215, 2. [CrossRef]
11. Zhdanov, M. Geophysical Electromagnetic Theory and Methods, 1st ed.; Elsevier: Amsterdam, The Netherlands, 2009.
ISBN 9780444529633.
12. Alfano, L. Dipole-dipole deep geoelectrical soundings over geological structures. Geophys. Prospect. 1980, 28, 283–296. [CrossRef]
13. Lapenna, V.; Satriano, C.; Patella, D. On the methods of evaluation of apparent resistivity under conditions of low message-to-
noise. Geothermics 1987, 16, 487–504. [CrossRef]
14. Cuomo, V.; Lapenna, V.; Macchiato, M.; Patella, D.; Satriano, C.; Serio, C. Statistical analysis of stationary noisy voltage recordings
in geoelectrics. Boll. Geofis. Teor. Appl. 1990, 24, 205–210.
15. Lapenna, V.; Macchiato, M.; Patella, D.; Satriano, C.; Serio, C.; Tramutoli, V. Statistical analysis of non-stationary voltage recordings
in geoelectrical prospecting. Geophys. Prospect. 1994, 42, 917–952. [CrossRef]
16. Rizzo, E.; Giampaolo, V. New Deep Electrical Resistivity Tomography in the High Agri Valley Basin (Basilicata, Southern Italy).
Geomat. Nat. Hazards Risk 2019, 10, 197–218. [CrossRef]
17. Bendat, J.S.; Piersol, A.G. Random Data: Analysis and Measurements Procedures; John Wiley & Sons: New York, NY, USA, 1986.
18. Box, G.E.P.; Jenkins, G.M. Time Series Analysis Forecasting and Control; Holden Day: San Francisco, CA, USA, 1976.
19. Jenkins, G.M.; Watts, D.G. Spectral Analysis and Its Applications; Holden Day: San Francisco, CA, USA, 1968.
20. Oldenburg, D.W.; Mcgillivray, P.R.; Ellis, R.G. Generalized subspace method for large scale inverse problems. Geophys. J. Int. 1993,
114, 12–20. [CrossRef]
21. Rucker, C. Advanced Electrical Resistivity Modelling and Inversion Using Unstructured Discretization. Ph.D. Thesis, University
of Leipzig, Leipzig, Germany, 2010.
22. Goebel, M.; Pidlisecky, A.; Knight, R. Resistivity imaging reveals complex pattern of saltwater intrusion along Monterey coast. J.
Hydrol. 2017, 551, 746–755. [CrossRef]
Geosciences 2022, 12, 438 18 of 19

23. Legaz, A.; Vandemeulebrouck, J.; Revil, A.; Kemna, A.; Hurst, A.W.; Reeves, R.; Papasin, R. A case study of re-sistivity and
self-potential signatures of hydrothermal instabilities, Inferno Crater Lake, Waimangu, New Zealand. Geophys. Res. Lett. 2009,
36, L12306. [CrossRef]
24. Lévy, L.; Maurya, P.K.; Byrdina, S.; Vandemeulebrouck, J.; Sigmundsson, F.; Árnason, K.; Ricci, T.; Deldicque, D.; Roger, M.;
Gilbert, B.; et al. Electrical resistivity tomography and time-domain induced polarization field investigations of geothermal areas
at Krafla, Iceland: Comparison to borehole and laboratory frequency-domain electrical observations. Geophys. J. Int. 2019, 218,
1469–1489. [CrossRef]
25. Revil, A.; Finizola, A.; Piscitelli, S.; Rizzo, E.; Ricci, T.; Crespy, A.; Angeletti, B.; Balasco, M.; Cabusson, S.B.; Bennati, L.; et al. Inner
structure of La Fossa (Volcano Island, southern Tyrrhenian Sea, Italy) revealed by high-resolution electric resistivity tomography
coupled with self-potential, temperature, and CO2 diffuse degassing measurements. J. Geophys. Res. Solid Earth 2007, 113, B07207.
[CrossRef]
26. Flores, M.A.; Gomez Trevino, E. Dipole-dipole resistivity imaging of the Ahuachapan-Chipilapa geothermal field, El Salvador.
Geothermics 1997, 26, 657–680. [CrossRef]
27. Di Maio, R.; Mauriello, P.; Patella, D.; Petrillo, Z.; Piscitelli, S.; Siniscalchi, A. Electric and electromagnetic outline of the Mount
Somma–Vesuvius structural setting. J. Volcanol. Geotherm. Res. 1998, 82, 219–238. [CrossRef]
28. Storz, H.; Storz, W.; Jacobs, F. Electrical resistivity tomography to investigate geological structures of the earth’s upper crust.
Geophys. Prospect. 2000, 48, 455–471. [CrossRef]
29. Weller, A.; Gruhne, M.; Seichter, M.; Boerner, F.D. Monitoring hydraulic experiments by complex conductivity tomography. Eur. J.
Environ. Eng. Geophys. 1997, 1, 209–228. [CrossRef]
30. Bosum, W.; Casten, U.; Fiedberg, F.C.; Heyde, I.; Soffel, H.C. Three-dimensional interpretation of the KTB gravity and magnetic
anomalies. J. Geophys. Res. 1997, 102, 18307–18321. [CrossRef]
31. ELEKTB Group. KTB and the electrical conductivity of the crust. J. Geophys. Res. 1997, 102, 18289–18305. [CrossRef]
32. Colella, A.; Lapenna, V.; Rizzo, E. High-resolution imaging of the High Agri Valley Basin (Southern Italy) with electrical resistivity
tomography. Tectonophysics 2004, 386, 29–40. [CrossRef]
33. Rizzo, E.; Colella, A.; Lapenna, V.; Piscitelli, S. High-resolution images of the fault-controlled High Agri Valley basin (Southern
Italy) with deep and shallow electrical resistivity tomographies. Phys. Chem. Earth 2004, 29, 321–327. [CrossRef]
34. Tamburriello, G.; Balasco, M.; Rizzo, E.; Harabaglia, P.; Lapenna, V.; Siniscalchi, A. Deep electrical resistivity tomography and
geothermal analysis of Bradano foredeep deposits in Venosa area (southern Italy): First results. Ann. Geophys. 2008, 51, 203–212.
[CrossRef]
35. Rizzo, E.; Giampaolo, V.; Capozzoli, L.; Grimaldi, S. Deep electrical resistivity tomography for the hydrogeological setting of
Muro Lucano Mounts Aquifer (Basilicata, Southern Italy). Geofluids 2019, 2019, 6594983. [CrossRef]
36. Balasco, M.; Galli, P.; Giocoli, A.; Gueguen, E.; Lapenna, V.; Perrone, A.; Piscitelli, S.; Rizzo, E.; Romano, G.; Siniscalchi, A.; et al.
Deep geophysical electromagnetic section across the middle Aterno Valley (central Italy): Preliminary results after the 6 April
2009 L’Aquila earthquake. Boll. Geofis. Teor. Appl. 2011, 52, 443–455. [CrossRef]
37. Oldenburg, D.W.; Li, Y. Inversion of induced polarization data. Geophysics 1994, 59, 1327–1341. [CrossRef]
38. Pucci, S.; Civico, R.; Villani, F.; Ricci, T.; Delcher, E.; Finizola, A.; Sapia, V.; De Martini, P.M.; Pantosti, D.; Barde-Cabusson, S.; et al.
Deep electrical resistivity tomography along the tectonically active Middle Aterno Valley (2009 L’Aquila earthquake area, central
Italy). Geophys. J. Int. 2016, 207, 967–982. [CrossRef]
39. Carrier, A.; Fischanger, F.; Gance, J.; Cocchiararo, G.; Morelli, G.; Lupi, M. Deep electrical resistivity tomography for the
prospection of low- to medium-enthalpy geothermal resources. Geophys. J. Int. 2019, 219, 2056–2072. [CrossRef]
40. Fischanger, F.; Morelli, G.; Ranieri, G.; Santarato, G.; Occhi, M. 4d cross-borehole electrical resistivity tomography to control resin
injection for ground stabilization: A case history in Venice (Italy). Near Surf. Geophys. 2013, 11, 41–50. [CrossRef]
41. Lajaunie, J.; Gance, J.; Nevers, P.; Malet, J.P.; Bertrand, C.; Garin, T.; Ferhat, G. Structure of the Sèchilienne unstable slope from
large-scale three dimensional electrical tomography using a Resistivity Distributed Automated System(R-DAS). Geophys. J. Int.
2019, 219, 129–147. [CrossRef]
42. Troiano, A.; Isaia, R.; Di Giuseppe, M.G.; Tramaruolo, F.D.A.; Vitale, S. Deep Electrical Resistivity Tomography for a 3D picture of
the most active sector of Campi Flegrei caldera. Sci. Rep. 2019, 9, 15124. [CrossRef]
43. Mazzini, A.; Carrier, A.; Sciarra, A.; Fischanger, F.; Winarto-Putro, A.; Lupi, M. 3D deep electrical resistivity tomography of the
Lusi eruption site in East Java. Geophys. Res. Lett. 2021, 48, e2021GL092632. [CrossRef]
44. Rizzo, E.; Giampaolo, V.; Capozzoli, L.; De Martino, G.; Romano, G.; Santilano, A.; Manzella, A. 3D deep geoelectrical exploration
in the Larderello geothermal sites (Italy). Phys. Earth Planet. Inter. 2022, 329, 329–330. [CrossRef]
45. Berger, K.J.; Johnson, P.A.; de Hoop, M.V.; Beroza, G.C. Machine Learning for data-driven discovery in solid Earth geosciences.
Science 2019, 363, eaau0323. [CrossRef]
46. Stallone, A.; Cicone, A.; Materassi, M. New insights and best practices for the successful use of Empirical Mode Decomposition,
Iterative Filtering and derived algorithms. Sci. Rep. 2020, 10, 15161. [CrossRef] [PubMed]
47. Ierley, G.; Kostinski, A. Extraction of unknown signals in arbitrary noise. Phys. Rev. 2021, 103, 022130. [CrossRef] [PubMed]
48. Yu, S.; Ma, J. Deep learning for geophysics: Current and future trends. Rev. Geophys. 2021, 59, e2021RG000742. [CrossRef]
49. Guo, X.; Zhao, Q.; Ditmar, P.; Sun, Y.; Liu, J. Improvements in the monthly gravity field solutions through modeling the colored
noise in the GRACE data. J. Geophys. Res. Solid Earth 2018, 123, 7040–7054. [CrossRef]
Geosciences 2022, 12, 438 19 of 19

50. Bogusz, J.; Rosat, S.; Klos, A.; Lenczuk, A. On the noise characteristics of time series recorded with nearby located GPS receivers
and superconducting gravity meters. Acta Geod. Geophys. 2018, 53, 201–220. [CrossRef]
51. Chen, H.-J.; Ye, Z.-K.; Chiu, C.-Y.; Telesca, L.; Chen, C.-C.; Chang, W.L. Self-potential ambient noise and spectral relationship with
urbanization, seismicity, and strain rate revealed via the Taiwan Geoelectric Monitoring Network. J. Geophys. Res. Solid Earth
2020, 125, e2019JB018196. [CrossRef]