A Practical approach for self-potential data acquisition,

processing, and visualization

Stéphanie Barde-Cabusson, Anthony Finizola, Niels Grobbe

To cite this version:

Stéphanie Barde-Cabusson, Anthony Finizola, Niels Grobbe. A Practical approach for self-potential
data acquisition, processing, and visualization. Interpretation, 2021, 9, pp.T123-T143. �10.1190/int-
2020-0012.1�. �hal-03003692�

HAL Id: hal-03003692

https://hal.univ-reunion.fr/hal-03003692v1
Submitted on 17 Jun 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
A practical approach for self-potential data acquisition, processing, and
visualization
Stéphanie Barde-Cabusson1, Anthony Finizola2, and Niels Grobbe3

Abstract
We have developed a comprehensive methodology for the acquisition and processing of self-potential (SP)
data, as well as some keys for the interpretation of the results. The wide applicability of the SP method and its
low cost make it a popular method to use in a variety of natural environments. Despite its versatility and the fact
that various published journal papers describe the method and its applications, we believe that there is an im-
portant need for a dedicated, peer-reviewed SP acquisition, processing, and visualization/interpretation paper in
the scientific literature. We have identified great interest from the scientific community for such a journal paper
as a guide for existing and new practitioners with their SP survey design, data acquisition, robust processing,
and initial interpretation steps. We have developed a step-by-step methodology for SP data acquisition and
processing, combined with practical guidance for the interpretation of collected and processed SP data, includ-
ing an evaluation of common errors and typical sources of uncertainty. Our examples are based on studies in
volcanic environments (e.g., hydrothermal systems); however, the processing steps and methodology are fully
applicable and transferable across disciplines to SP data acquired in any environment, and for a wide variety of
applications. We evaluated the field acquisition method and the low-cost equipment, the reference and closure
corrections, their meaning for the SP signal, and their effect on the data set. The benefits of interpolating SP data
in two steps are examined. Combining map visualization, SP versus distance, and SP versus elevation graphs
appears to be a highly effective strategy to interpret the signal in terms of hydrogeologic and hydrothermal
domains and to highlight structural limits in volcanic contexts as well as in other environments.

Introduction words, there is no off-the-shelf SP acquisition system

The self-potential (SP) method is a passive geoelec- available, to our knowledge. Training for the SP method
trical method that is sensitive to underground fluid is often taught by word of mouth and field experience,
flows such as hydrogeologic and hydrothermal circula- and equipment has to be fabricated in house or gath-
tions. It is convenient to apply in the field, and the data ered from various suppliers. However, this lack of
are relatively easy to process. Contrary to most other standardization also offers the benefit of flexibility:
geophysical methods, for the SP method, no official There is total freedom to adapt, for example, the choice
manual is typically delivered with the equipment. Some of devices, data acquisition strategy, and data process-
geophysical equipment companies have started to list ing to the needs of each user, study setting, and target.
SP among the methods for which they offer devices. Historically, the SP method has been used in economic
However, this is relatively recent (less than 5 years) mineral exploration. Now, it also expands to environ-
and the products are mainly electrodes and wires origi- mental geophysics and other applications, in which
nally sold for electric resistivity or magnetotelluric sur- for the above described reasons it remains “the ugly
veys, which need to be combined in a customized duckling” that can transform into a swan (Nyquist
fashion for SP acquisition purposes by the user. In other and Corry, 2002) if proper attention is dedicated to it.

1
University of Hawai‘i at Mānoa, Hawai‘i Institute of Geophysics and Planetology, School of Ocean and Earth Science and Technology,
Honolulu, Hawaii 96822 2217, USA. E mail: [email protected] (corresponding author).
2
Université de La Réunion, Laboratoire GéoSciences Réunion, F 97744 Saint Denis, France and Université de Paris, Institut de Physique du Globe
de Paris, CNRS, F 75005 Paris, France. E mail: [email protected].
3
University of Hawai‘i at Mānoa, Hawai‘i Institute of Geophysics and Planetology, School of Ocean and Earth Science and Technology,
Honolulu, Hawaii 96822 2217, USA and University of Hawai‘i at Mānoa, Water Resources Research Center, Honolulu, Hawaii 96822 2217, USA. E
mail: [email protected].
Historical evolution of SP applications context (Rao, 1943; Dessau, 1950), but this did not lead
As Sheriff (1978) writes, “Electrical exploration is to consideration of the SP drift to be corrected before
not used a great deal in the United States, although the generating SP maps. The first SP maps have been per-
French and the Soviets use it extensively.” This was at formed related to mining and ore body context (Kruger
the beginning of the expansion of the method, when and Lacy, 1949; Dessau, 1950; Gay, 1967; Henriet, 1971),
some fundamentals of the SP theory and application but without applying the loop correction. In the early
were starting to be published in more detail (e.g., Seme- 1970s, SP prospection started in geothermal explora-
nov, 1980). Finizola (2017) describes the history of the tion (Banwell, 1970; Anderson and Johnson, 1973;
SP method, showing that the SP method was initially Zohdy et al., 1973; Corwin, 1975), and important book
used for prospecting in mining, using rudimentary chapters and an entire Ph.D. thesis were dedicated to
equipment (Fox, 1830). It consisted of metal copper the SP method (Corwin, 1973; Parasnis, 1973; Gex,
plates used as electrodes and a galvanometer measur- 1977). However, it is in 1976 that the concept of closure
ing the electrical current between them. Nowadays, offset and loop correction appeared for the first time.
with more modern devices, SP is still a common, useful This new data processing, preliminary to elaborating
tool for ore deposit detection (see Sato and Mooney an SP map was specified in a geothermal exploration
[1960] for the theory of the electrochemical mechanism in Long Valley, California (Anderson and Johnson,
and, e.g., Eppelbaum, 2019). As more experience was 1976), and also in the first SP publication applied in vol-
gained with the method, SP became a useful tool for canology, on Kilauea in Hawaii (Zablocki, 1976). Since
geothermal studies, primarily in the United States (e.g., this period, the concept of SP loop correction has been
Zohdy et al., 1973; Corwin, 1976; Anderson and John- adopted in various fields (volcanology, hydrogeology,
son, 1979; Corwin and Hoover, 1979) but in the same etc.), but it was only mentioned by one or a few senten-
period, this cheap and user-friendly method raised in- ces in the data processing sections of the publications
terest for applications on active volcanoes in the United (Ballestracci, 1982; Schiavone and Quarto, 1984; Four-
States (Zablocki, 1976), and shortly thereafter in nier, 1989). The most extended explanation about the
Europe (e.g., Van Ngoc et al., 1980; Ballestracci, loop correction is concentrated in one short paragraph
1982; Lénat et al., 1982; Soudoplatoff et al., 1982; Halb- (see Zablocki, 1976, p. 1302).
wachs, 1983; Lénat, 1987). As for most of the existing
geophysical methods, the SP results are sometimes dif- Instrumentation developments
ficult to interpret as a stand-alone data set. Nowadays, Because the initial copper plate electrodes were
despite these challenges, the SP method is successfully used in early mining studies, the devices used for SP
used in many studies covering a wide variety of do- measurements evolved significantly. The first use of
mains such as mineral exploration (e.g., Biswas, 2017), nonpolarizable electrodes was proposed by Barus
geothermal activity and resources (e.g., Zohdy et al., (1882). With the developments in the field of geother-
1973; Corwin, 1976; Anderson and Johnson, 1979; Cor- mal studies, nonpolarizable electrodes were improved
win and Hoover, 1979; Revil and Pezard, 1998; Ishido and were systematically used in SP acquisition, replac-
et al., 2010; Bolós et al., 2019), hydrology, hydrogeology, ing the metal electrodes (Gex, 1977). Indeed, the metal
and hydrogeophysics (e.g., Aubert and Atangana, 1996; electrodes were shown to generate parasitic currents
Doussan et al., 2002; Revil et al., 2005; Suski et al., 2006; through redox reactions occurring between the metal
Jardani et al., 2009; Jouniaux et al., 2009), studies of and the ground, which induces significant errors on the
sinkholes (e.g., Jardani et al., 2006), caves, conduits, measurements. “Porous pot”-type copper-copper-
and faults (Saribudak and Hauwert, 2017; Saribudak, sulphate nonpolarizable electrodes were the first type
2019), dam leakage detection (Gex, 1980; Bolève et al., used and were subsequently improved in Europe (Gex,
2011), subglacial flows (Kulessa et al., 2003), dormant 1977). Other types of electrodes were tested and used
monogenetic volcanic fields (e.g., Bolós et al., 2012; for prospecting and/or monitoring (Perrier et al., 1997),
Barde-Cabusson et al., 2014), and active volcanoes in such as the Petiau lead-lead chloride electrodes (Petiau,
general (e.g., Zablocki, 1976; Van Ngoc et al., 1980; Lé- 2000), which are considered to be more stable in time,
nat, 1987; Finizola et al., 2002, 2003, 2004, 2009, 2017; and they are usually recommended for monitoring pur-
Ishido, 2004; Hase et al., 2005; Aizawa, 2008; Revil et al., poses. The high-impedance multimeters typically used
2008, 2011; Barde-Cabusson et al., 2009, 2012; Bennati for SP measurements need to have a 10–100 MΩ imped-
et al., 2011; Brothelande et al., 2014; Gonzales et al., ance, a basic accuracy of ±0.09%, and a DC voltage in
2014; Villasante-Marcos et al., 2014; Chaput et al., the range of 0.1 mV to 1000 voltage direct current. They
2019; Grobbe and Barde-Cabusson, 2019; just to men- can be further augmented with a filter for unwanted 50–
tion a few). 60 Hz signals usually present in inhabited areas. These
More specifically speaking about the processing of filters are often required to avoid the effect of electric
SP data, reference and loop/closure corrections have lines disrupting the SP signal. Indeed, the SP method
not always been applied to the raw data. In the 1940s, suffers from a series of limitations associated with envi-
few authors mentioned the SP signal variability, in the ronmental factors; for example, poor coupling of the
electrical logging of oil wells (Dickey, 1944; Mounce electrodes with the ground, the presence of tree roots,
and Rust, 1945) and associated with the mineralization or the biological activity of ants around the electrodes all
can disrupt the signal. Furthermore, small-scale hetero- can be generated by redox potentials associated with
geneities of the ground can induce significant differences ore bodies or the metallic casing of boreholes or con-
in the measured SP values when moving the electrode taminant plumes that are rich in organic matter (Sato
position around by a few centimeters (Boudoire et al., and Mooney, 1960; Linde and Revil, 2007; Minsley,
2018). It is important to be aware of the instrumentation 1997; Castermant et al., 2008). Another contribution
limitations and potential sources of uncertainty in the SP is the rapid fluid disruption effect described by Johns-
data, such that one can do their very best during data ton et al. (2001) based on the previous works by Blan-
acquisition to minimize these influences (e.g., ensure chard (1964) in nonporous media. They propose that
good coupling with the ground). Furthermore, with each charge generation by fluid vaporization/disruption con-
evolution of the devices used for SP measurements, and tribute to the SP signal. Because liquids are vaporized
because the method has gained attention from or removed as droplets by gas transport away from hot
professionals in a variety of fields, these limitations have sources, charge generation and local increase in electri-
become increasingly better addressable. cal resistivity by removal of fluids should occur and
Despite the abundant literature on the SP method in contribute to electric fields. However, this new SP
applied geophysics and a growing knowledge of the source has been criticized and might be explained by
theory behind SP signal generation (e.g., Revil and the electrokinetic effect (Revil, 2002).
Jardani, 2013), unfortunately, some significant mistakes
have occurred in the past while acquiring and/or Field acquisition
processing SP data, such as not using closed and inter- All of the measurements performed along a continu-
connected measurements or not correcting the data for ous line are hereafter called “profile.” To perform SP
the drift before generating SP maps. measurements, we use a pair of nonpolarizable electro-
We, and other geophysicists, have been refining the des — usually Cu/CuSO4 electrodes. The Petiau electro-
whole process until we reached an fast and efficient des (Petiau, 2000) are considered more stable for long-
way to acquire and process SP data. The number of term monitoring, but Cu/CuSO4 electrodes offer many
inquiries that we received concerning SP data acquisi- benefits for field prospecting as well as with respect
tion and processing was large, such that we realized the to their design, such as a low price and easy homemade
lack of or inaccessibility of comprehensive literature re- construction. They provide an enhanced electrical con-
garding SP data acquisition and processing and, most tact with the ground due to a greater porous surface, and
importantly, the demand from the scientific community they allow us to improve percolation by manually in-
for such literature. This paper aims to reply to this creasing the pressure of the electrolyte inside the elec-
demand by describing the steps from data acquisition, trode as the operator pushes the rubber plug of the
via processing and data correction steps, to the final electrode. The porous pot electrodes such as those used
visualization of SP results, as well as some important for magnetotelluric measurements have a bigger diam-
keys for their interpretation. Our experience with the eter than “slim” pencil-shape SP electrodes, and they
method was mostly obtained in volcanic environments; theoretically offer a slightly greater or similar porous
consequently, most of the examples presented here surface. However, due to their flat base, any imperfec-
deal with volcanoes. However, given the general nature tion of the ground will lift and tilt the electrode, resulting
of this paper and its focus on data processing and visu- in a significant worsening of the contact (thereby in-
alization, its contents are transferable and applicable to creasing the contact resistance) if not noticed or if not
any study using SP data, expanding its usefulness to any corrected. The pencil-shape electrodes have sharp or
user of the SP method. slightly rounded preeminent porous tips that allow for
optimization of the surface of contact. In volcanic fields,
Overview of SP principles in which the ground is often rocky and irregular, the
The SP method is based on the measurement of nat- problem with “porous pots” is common, and we have
urally occurring electrical potential differences in the found that pencil-shape electrodes allow for getting a
earth. Some SP signals result from the presence of an good electrical contact, quickly and easily. With SP pro-
electrical double layer at the microscale of a rock for- spection, in which many measurements are done each
mation that is saturated to some degree with a pore day, this is all the more important. Cu/CuSO4 electrodes
fluid. In this case, we typically assume that the electro- have been used for the examples presented in this paper.
kinetic effect that occurs in such a system is the dom- The microporous end of these electrodes is made of per-
inant signal-generating mechanism; an assumption that meable wood (or ceramic), which is put into contact
is often valid in most geoscientific settings (e.g., Leroy with the ground (Figure 1). Wood with high permeability
and Revil, 2004; Revil and Jardani, 2013). However, it is is preferred over ceramics due to its durability and be-
important to realize that other contributions to the gen- cause it ensures better percolation and thus better elec-
eration of the SP signal exist, including the thermoelec- trical contact.
tric effect (Nourbehecht, 1963; Sill, 1983), which can be The difference in the electrical potential between the
significant in environments with strong thermal gra- reference electrode (arbitrarily placed at the beginning
dients, such as geothermal fields or active volcanoes. of the profile) and the moving electrode is measured
In addition, SP signals with an electrochemical origin with a high-impedance multimeter, and it is typically
in the order of a few millivolts up to a few tens of milli- any significant benefit or drawback in using a thinner or
volts. Before each SP survey, it is important to check thicker wire on the SP measurement itself. However,
that the multimeter is operating properly. Malfunction with these characteristics, a 300 m wire appears to have
arises from aging of the electronic components or from a good weight/efficiency ratio avoiding heavy coils of
infiltration of water inside the multimeter during field- wire in the field, especially in uneven or otherwise chal-
work. Two types of tests can be done to ensure that the lenging terrains, while still enabling the coverage of
multimeters are properly working: (1) a test using a cur- long distances in the field without having to change
rent calibrator in the laboratory or (2) the comparison the reference too often. Using a long wire (several hun-
of the measured values for at least three different multi- dreds of meters) allows us to reduce possible errors
meters, for the same measurement in the field: The that may occur due to a too-frequent change of the
three multimeters must show the same SP value. Note reference (fixed) electrode. Reducing the number of
that it is very important to always connect the moving reference changes during a survey is recommended to
electrode to the positive port (VDC) of the multimeter reduce possible differences between sections (e.g.,
and the reference electrode (through the connection on noise), associated with different environmental condi-
the wire reel) to the negative port (COM). Keeping this tions at each reference. This is usually insignificant,
convention in mind is of utmost importance to interpret but we can reduce the risk by using a long wire, thereby
the relative positive and negative peaks correctly. minimizing the amount of required reference changes.
A long wire (usually approximately 320 m, approxi- The length of the wire obviously has a trade-off as well
mately 420 m, or approximately 520 m) is used to con- because, the longer the wire, the harder it becomes in
nect the two electrodes to the multimeter. These are rough terrain to carry and to roll-collect the wire for the
deliberately not chosen as exact multiples of 100 m to subsequent section of the profile; for long wires, the
allow for tying the cable at the beginning of the profile chance of it getting stuck behind vegetation, rough rock
to something heavy or immovable, as well as to use the surfaces, etc., becomes higher. In inhabited areas, it is
extra wire to potentially move the new reference loca- also important to take into account that the longer the
tion, in the case of a bad electrode-soil contact at the wire, the higher the electrical noise. The spacing be-
end of the profile. For example, 320 m of cable corre- tween two successive measurement points depends
sponds to a usual 300 m measurement profile in addi- on the size of the anomalies that we expect. Commonly,
tion to having 10 m of extra cable on both ends. The spacing of 20 m is optimal to get good resolution, that is,
core of the wire that we usually use is made of copper, adapted to the detection of the main structural limits,
and its section is usually of 1.5 mm2. We did not notice hydrothermal anomalies, and hydrogeologic flows (Fin-
izola et al., 2002, 2003; Bennati et al., 2011), while ensur-
ing reasonable data acquisition time on areas of several
square kilometers. To save time and maintain the pre-
cision of the spacing between measurements, the cable
can be marked with pieces of colored tape, with the
number of marks corresponding to the distances from
the last reference, for example, the first five marks = 0 m,
one mark = 20 m from the reference, two marks = 40 m,
three marks = 60 m, four marks = 80 m, five marks
again = 100 m, one mark again = 120 m, and so on. At
each measurement location, a small hole (approximately
10 cm deep) is dug to reach the natural moisture of the
soil and thus to improve the electrical contact between
the electrode and the ground. As an alternative to the
fixed reference acquisition method, the gradient method
consists of moving both electrodes forward along the pro-
file at each measurement, with a constant measurement
spacing (e.g., Thanassoulas, 1989; Minsley, 1997; Lowrie,
2007). However, this method increases the chances of
one of those electrodes having a bad contact resistance
for the next measurement, increasing the risk to impact
all the consecutive measurements it is connected to. To
reduce this risk, the leading electrode can also become
the trailing electrode at each new measurement. This
Figure 1. (a) Schematic representation of SP data acquisition. technique can be used to minimize errors but the gradient
COM and DCV indicate the correct connection to the multi
meter. (b) Example of field data acquisition (photo courtesy
method anyway involves significant cumulative errors on
A. Finizola). (c) Schematic representation of a Cu CuSO4 non long distances, compared to the fixed reference electrode
polarizable electrode (modified from Barde Cabusson, 2007). acquisition technique. Repeating each measurement at
least three times at each station helps to reduce the error
but does not cancel it and adds significantly to the time trical resistance prior to each SP measurement to check
necessary at each measurement point. This is critical for the electrical contact between the electrodes. This sec-
survey areas of several square kilometers, where thou- ond option allows for getting a direct assessment of the
sands of measurements are necessary. The gradient quality of the electrical contact, and it significantly re-
method should thus be used with extreme caution in duces the duration of each measurement. This is appli-
small-dimension surveys with no or very limited environ- cable only to nonpolarized electrodes because other
mental noise. With a long wire, switching the electrodes electrode types would need a sometimes long relaxa-
at each new reference is performed to keep the electrode tion time and, in the long term, become permanently
located at the last measurement untouched in the ground polarized. Usually, the resistance should be on the or-
while starting with the next section of measurements. der of a few kilo ohms or tens of kilo ohms. If the re-
This minimizes the risk to worsen the contact without sistance is much higher, the first verification consists of
noticing it. It also compensates for the small difference checking the wires connected to the multimeter by cre-
of potential possibly existing between the two electrodes ating a closed circuit with the wire and checking for a
themselves. The significantly lower noise affecting the maximum resistance of only a few ohms. Usually, the
data with the fixed-base method (with respect to the gra- contact of the moving electrode with the ground should
dient method) is crucial in studies involving the repetition be enhanced when reaching a few hundred kilo ohms
of measurements in time, for example, to monitor the or, alternatively, the location of the measurement
evolution of the SP signal along a profile. Chaput et al. should be modified by a few tens of centimeters if the
(2019) show the reproducibility of their SP measure- resistance value exceeds approximately 1 mega ohm.
ments in areas not affected by variations of the hydrother- These threshold values are indicative of and closely
mal or hydrogeologic activity, whereas the method depend on the geologic nature of the surface or subsur-
allows for detecting some variations from a few millivolts face and the moisture of the ground. When the resis-
to several tens of millivolts, mapping a caldera collapse tance measurement is satisfying, the SP measurement
before the collapse with 1 m of precision. With the noise can be made by switching the multimeter to the VDC
affecting the signal with the gradient method, such pre- position. The operator should read the value when it
cision would be lost. is stable (usually after a few seconds for nonpolarizable
Water or a bentonite/water/salt water mixture can be electrodes). In the case of doubtful SP values, problems
used to improve the contact resistance when the regarding the quality of the SP data can be easily de-
ground is particularly dry, but this is usually unneces- tected when, while making several measurements, the
sary. The natural moisture of the ground is generally SP variation appears to be directly linked to the resis-
sufficient to provide a good coupling, and the use of tance variation. In that case, the electrode needs to be
a high-impedance multimeter allows for getting good- moved to another location. However, risks of malfunc-
quality measurements even with varying contact resis- tioning of the instrumentation are usually easy to detect
tances from point to point. For SP, one can even argue and mitigate because of the several tests and checks
that the artificial wetting of the ground is not recom- done during a survey and during the preparation of
mended: It can change the local geochemistry and elec- the survey: (1) Before the survey, each wire and each
trical double-layer effects that are at the foundation of pair of electrodes are checked carefully for a good elec-
the SP coupling, and these changes may therefore erro- trical connection, (2) each multimeter that is used is
neously impact the SP measurement. Therefore, we calibrated with a current calibrator or, if one is not
would only recommend using wetting of some kind in available, some measurements are performed in the
case everything else fails. From a practical point of view, field at the same location using different multimeters
measurements should be planned beginning at the higher to check their accuracy, (3) at the beginning of every
elevation and ending downslope. This makes the winding day in the field, the electrodes are placed tip to tip to
of the wire significantly easier and faster because the check for an electrical potential difference of less than
weight of the wire itself would otherwise increase the 1 mV, and (4) if any sudden variation of the SP value
friction with the ground, whereas, going downhill, grav- occurs in the field, we perform repeated measurements
ity will actually help in pulling this weight. in a space of a few centimeters and some instrument
With Cu/CuSO4 (and other nonpolarizable) electro- checks are done again. Each measurement location is
des, two main techniques exist to ensure the quality registered with a handheld GPS and recorded along
of the SP data. The first option is to perform several SP with the measurement numbering, the distance from
measurements in a small area for each data point; this the initial reference, the measured resistance, and the
method can be used for any electrode type, but it is time measured SP value. A good practice is to keep the elec-
consuming and does not ensure a real control on the trodes, and in particular the reference electrode, pro-
quality of the electrical contact. In addition, there is the tected from direct sunlight to avoid possible
ambiguity of which SP value to choose. The repeated disruptions due to temperature variations and rapid
measurement technique is a good practice for the refer- evaporation of the electrolyte from the electrodes. SP
ence electrode change at the end of a section of wire, in data have to be collected when the moisture of the
complement of the second option presented thereafter. ground is good enough in the first few centimeters of
The second option is to measure the value of the elec- the ground. We usually dig a maximum of 10–15 cm,
and at that scale, the vertical electrical field gradient is The whole measurement network of an SP survey has
lower than SP environmental noise (Boudoire et al., to be interconnected. To do so, SP measurements are
2018; Chaput et al., 2019). If the electrode contact is sat- performed forming loops, that is, closed profiles or pro-
isfying, there is no reason to dig more than 15 cm in files that are connected at both ends to other closed pro-
depth for SP mapping. If the contact is not satisfying files (Figure 2b). During data processing, this will allow
in the shallow part of the ground, it is better to postpone for assessment and correction of the drift undergone dur-
a survey to a rainier season. ing the acquisition of the data. The only exception to
A new reference station is established every time the closed-loop acquisition is for the “floating profiles” that
end of the wire is reached. At the last measurement are connected to the main network by only one end, or
along the wire, the moving electrode is left in place for single isolated profiles that are measured in a short
in the ground to become the new reference electrode time period (of a few hours maximum) for which we typ-
and the former reference electrode that stayed at the ically can assume that no significant modification of the
beginning of the profile is picked up and is then used environmental conditions takes place. However, this
as the moving electrode for the next section, until should be done only exceptionally and only when closing
the end of the wire is reached again. Using this tech- the profile is absolutely not possible. A profile connected
nique helps to avoid error accumulations related to at both ends to a single water body is also considered a
the potential difference existing between the two elec- closed profile, considering that this water body is an
trodes put tip to tip, that is, the error induced by the equipotential surface (Figure 2b). This is true if we con-
electrodes themselves. This error is very small but sider that the pH and mineral composition are reason-
can accumulate if the reference and measuring electro- ably similar in the whole water body. In general, it is
des are not switched at each section. Switching ensures thus preferable to close the profiles at the same geo-
that the electrode error cancels out over two sections graphic location. By convention, the 0 mV SP value is
(one section corresponds to the length of the wire). Al- generally attributed to the sea, a lake, or surface of an
ternatively, the potential difference between the two aquifer, and a positive or negative SP anomaly will refer
electrodes can be measured regularly and the data to an anomaly with respect to this body. In the case of
can be corrected from the measured value. Note that the absence of any water body in the investigated area,
after switching, the previous reference electrode be- the 0 mV potential is taken arbitrarily. This can be an
area located outside or at the margin of the sector where
comes a moving electrode and must be connected to
the main study is focused (where SP variations are ex-
the positive port (VDC) of the multimeter, whereas
pected to be minor) or where the values are approxi-
the moving electrode becomes the new reference and
mately the average of the data set.
must therefore be connected to the negative port
After the survey, all of the SP values of the intercon-
(COM) through the coil of wire.
nected profiles are corrected through
several steps that will connect the whole
data set to a unique reference station
(e.g., with respect to the sea or any other
main water body of the investigated
area, when possible), which is set to
0 mV. This process requires two con-
secutive corrections of the raw SP data:
(1) the reference correction and (2) the
closure (or loop, or drift) correction. We
use a spreadsheet application, such as
Excel or Grapher (Golden Software),
to process the data, and we use a con-
touring and 3D surface mapping pro-
gram to interpolate the data and
create maps and graphics (e.g., overlay-
ing the 2D SP map on a 3D surface) for
the visual presentation and evaluation of
the results. Besides maps, the data can
be displayed using plots (graphs) as
well. In a plot, the SP data are presented
as a function of the acquisition distance
Figure 2. (a) Schematic representation of the reference correction of SP data. to, for example, precisely locate struc-
The orange arrows represent the shift of the data, section by section. This must
be applied to any SP profile composed of several sections, that is, with several
tural features, or as a function of the al-
references, for example, those presented in (b) (SP profiles in an aerial view). titude to, for example, highlight areas
(b) Schematic representation of a profile closed on itself, on the sea at both ends, with different geology or hydrogeologic
or on any other SP profile. or hydrothermal characteristics.
Corrections of an SP profile of potential between the two electrodes themselves,
Reference correction etc.); consequently, a drift in the measurements will
During acquisition of the data, each time that we ar- be observed.
rive at the end of the wire with the moving electrode Without any other prior information, we assume that
(which would define one section of a profile), a new the drift increases linearly from the time when the first
reference electrode is established, that is, the last posi- measurement was taken to the time of the last one; that
tion of the moving electrode will become the new refer- is, along the period of time of the acquisition, the drift
ence station for the next section, until the end of the increases continuously. It remains difficult to provide a
wire is reached again. The new reference is considered very accurate correction due to the variability of events
to be the new 0 mV reference for the whole new section provoking the drift. Among those, localized rain events
(Figure 2), relative to which all other measurements of that may affect only part of a profile (in particular in the
that section are made. Therefore, we obtain data with case of kilometer-long profiles), variations of the hydro-
several arbitrary 0 mV values, at the beginning of each thermal activity (in geothermal fields and volcanoes),
new section of a profile, which must be properly com- and spatiotemporal variations of the sun radiation,
bined and corrected to obtain continuity in the data set which are all nonlinear variations. Thus, we consider
along each profile. In other words, the reference correc- that the most efficient approximation, in terms of accu-
tion is applied to connect all of the sections of a single racy and acquisition and processing times, is to use the
SP profile together. It must be applied for each SP pro-
file, section by section, to connect each section to the
end of the previous one. Only the first section will re-
main unchanged, whereas the others will be corrected
section by section along the direction in which the mea-
surements were acquired. This correction allows us to
link all of the sections of one profile to the same initial
reference potential (0 mV at the beginning of the
profile).
Let us now take a closer look at how we can bring
these theoretical concepts to practice. In a spreadsheet,
we isolate each profile in individual worksheets (Fig-
ure 3). We want to highlight that the formulas used
are transferable to any (open-source) spreadsheet soft-
ware. The reference correction is made on each profile
separately, section by section. In the column containing
the raw SP data, the first section is not modified and the
last value of this section is added to all of the measure-
ments of the second section. Thereafter, the third sec-
tion is shifted by the last corrected value of the second
section and so on. In the example of Figure 3, the loca-
tions of the references are highlighted by the orange
rows and correspond to a data acquisition length of
300 m (the maximum length of the wire) per section.
As pointed out previously, the initial section (rows
3–18 in this example) depends directly on the first refer-
ence so no correction is applied to these data points.
The second section (from rows 19 to 33) is shifted to
the last value of the initial section; section 3 (rows
34–48) is shifted to the last value of section 2.
Figure 3. Reference correction of the SP data along a single
profile, using Microsoft Excel. The references locations and
Closure (or loop, or drift) correction for one
values are highlighted by the orange rows that mark the limits
profile of each section of 300 m (the maximum length of the wire in
In the case of a closed profile, the first point is geo- this example). The initial section (rows 3 18) does not need a
graphically the same as the last one: The measured SP reference correction (note that columns F and G are identical
value should theoretically be the same as the well. This for that section). The second section (rows 19 33) is added to
would be true if no environmental perturbations oc- the last value of the initial section. Likewise, section 3 (rows
curred within the time span when the first and last mea- 34 48) is added to the last value of section 2. The $ sign is the
formula that is used in Excel to keep the cell coordinates (the
surements were made. However, during a survey, the column letter and the row number) constant when the for
measurement conditions can change (e.g., rain events mula is copied through the rest of the section. However, any
inducing variations in the soil moisture, the soil temper- of those formulas are transferable to any other open source
ature, instrument error such as the progressive increase spreadsheet software.
distance along the profile to distribute the drift. Usually, very last measurement point of the profile, after appli-
we expect a drift of a few millivolts in an SP loop ac- cation of the reference correction (Figure 4a). The cor-
quired over the duration of a few hours without drastic rection of this drift is applied to all of the data points of
variations of the environmental conditions (in particu- the profile by linearly increasing the correction factor
lar, heavy rain), although loops performed over several from the first to the last point of the profile. The initial
days and with changing conditions may show drifts of point is the reference of the profile so no correction is
several tens of millivolts. It remains difficult to identify applied to it (considered as point 0). In our example of
the various phenomena producing the drift and to Figure 4, the first measurement point is corrected for
evaluate their individual contributions; regardless, it one-tenth of the total drift, in which 10 is the total num-
is considered parasitic and must be corrected through ber of data points without counting the reference. The
the closure correction. second point is corrected for two-tenths of the total
In the example presented in Figure 4, we imagine a drift, and so on until the last point. The last point is cor-
profile consisting of 10 measurement points, in which rected for the total drift, resulting in it obtaining the
the last measurement point (10) is geographically the same value as the initial point, as it is supposed to be.
same as the reference (0), forming a closed profile (for Looking at a practical example of the aforemen-
examples of closed profiles, see Figure 2b), that is, the tioned theoretical correction, the Excel spreadsheet
first and last measurement points that are either at the presented in Figure 5 shows that the reference point
same location or are both linked by an isopotential line (row 3) and the last point (row 48) have the same
(e.g., sea, lake, and aquifer). The drift is defined by the GPS coordinates but a different SP value after reference
difference between the SP values at the reference and correction (see cells G3 and G48). As said, this differ-
ence is the drift that we must correct the data for. The
correction is presented in column H, and the corre-
sponding equation is shown on the right side of the
figure.
First, we must create a column with continuous nu-
meration of the data points, beginning at zero for the
reference (column A). The numbers in this column will
be used for determining the correction factor applied to
each point. The reference itself does not need a closure
(or drift) correction, which is why the correction factor
for this SP value is zero (cell A3). The closure correc-
tion formula is first defined in cell H48, and this formula
is applied across the column, up to the first reference
point. We can immediately validate whether or not the
formula has been applied correctly because the result-
ing value of the last data point (cell H48) should be iden-
tical to the value of the reference point (here 0 mV).
Some elements of the equation are constant, but others
vary as the correction is applied to each point of the
profile.
The closure correction equation is

SPc ¼ SPr − ðD∕NÞ  n; (1)

where SPc is the SP value after the closure correc-

tion, SPr is the SP value after the reference correction,
D is the observed drift, N is the total number of data
points in the profile excluding the reference, and n is
the place of the data point in the profile, determining,
together with D and N, the correction factor.
Figure 4. Schematic representation of (a) an SP profile com
posed of 10 measurements in which points 0 and 10 are either
In the example shown in Figure 5, the correction for-
geographically the same or are linked by an isopotential line mula in Excel is composed as follows:
(e.g., sea, lake, or aquifer); the difference observed in the SP
value is the drift. (b) The linear distribution of the drift from
1) Drift
the reference (0 fractions of the drift) to point 10 (10 fractions
of the drift, i.e., the total drift), in which one green arrow is D ¼ G48 − G3: (2)
one fraction of the total drift. (c) The application of the clo
sure correction for the 10 data points, with points 0 and 10 This is a constant in the equation for the whole pro-
now displaying the same SP value. file, which is why we use the $ sign that keeps the
 row or column following this sign unmodified when Closure correction for multiprofile surveys
copying the formula to the top of the column. Overview
2) Divide the drift by the number of data points in the As discussed before, in a complete SP survey, mea-
profile (N; a constant) surements are performed forming loops (closed profiles
or interconnected profiles) to monitor the drift and cor-
¼ ðG48 − G3Þ∕A48: (3) rect for it. When a data set is composed of several pro-
files in an area to build an SP map, each profile must
first be corrected independently for the reference cor-
3) Multiply by the place of the data point in the profile rection, as described in the “Reference correction” sec-
(n; variable) tion. In the following description, we assume that this
reference correction has been applied to all of the
¼ ððG48 − G3Þ∕A48Þ  A48; (4) profiles.
After the reference correction, each profile has its
where n is a variable and A48 in our example cor- initial point as a reference with a value of 0 mV. To con-
responds to the closure correction applied to the nect the data from all of the profiles together, we
last point of the profile. When copying the formula choose an initial closed profile/loop that will form the
to the top of the column in the Excel sheet, the cor- basis for the subsequent closure corrections of other
rection factor will change as a function of the row. closed profiles (see step 1 for Figure 7a–7d). Under op-
4) Subtract this value from the SP value for which the timal conditions, the initial loop should be completed in
reference correction has already been applied (SPr) a short amount of time. However, it can be composed of
measurements acquired over several days. This loop is
¼ G48 − ððG48 − G3Þ∕A48Þ  A48: (5) the first to which the closure correction is applied,

This is also a variable that will

change per data point.
Note that equation 3 gives the abso-
lute value corresponding to the length
of a single green arrow in Figure 4.
In a real case study (Figure 6), we ob-
serve that the closure correction affects
the whole profile but does not affect
the shape of the SP curve; that is, the
maxima and minima remain geographi-
cally unchanged. The drift is exagger-
ated here, compared to the length of
the profile, to get a perception of the ef-
fect of the closure correction. In a real
case, the closure correction would usu-
ally be lower. However, performing the
closure correction is important because
even an uncorrected small drift can in-
duce big artifacts on an SP map. The
importance of the location of the SP
maxima and minima will be discussed
in the following sections. It is also im-
portant to keep in mind that the SP mea-
surements are not absolute; after the
application of the reference correction,
Figure 5. Closure correction of the SP data along one closed profile (or loop),
the SP values are relative to the first using Excel. References are highlighted by the orange rows that mark the limits
reference of the profile that they belong of each section of 300 m (the maximum length of the wire in this example). In
to and, after the application of the clo- column G, the first (cell G3) and the last (cell G48) SP measurements are at the
sure correction, the values of the whole same geographic location but they show different SP values (values after the
data set are relative values with respect reference correction). This difference (G48 G3) corresponds to the drift of this
profile. Column H shows the closure correction applied to correct for this drift.
to one unique reference that we choose
When applied to the whole loop, the first (cell H3) and the last (cell H48) display
out of the area of interest of the study or the same SP value (here, 0 mV). The $ sign is the formula that is used (in Excel) to
that is placed at the sea, a lake, or an- keep the cell coordinates (the column letter and the row number) constant when
other water body. the formula is copied through the rest of the column.
which also gives a sense of the quality of the measure- c5, and d5 correspond to the complete data sets ac-
ments; the other profiles have to be connected at both quired in the field. The green sections are the loops con-
ends to this initial loop, or to another profile that is al- nected step by step to the data that have already been
ready corrected by the closure correction. Thus, the corrected (the black sections) in a previous step. The
profiles are forming new loops and are
connected step by step to parts of the
survey for which the data have already
undergone the closure correction. To op-
timize this procedure and to make sure
proper data corrections can be made, a
precise planning of SP data acquisition
in the field always involves foreseeing
the correction phase. This involves plan-
ning interconnected profiles and profiles
that preferably do not cross each other.
Profiles crossing each other create addi-
tional sections, each of which have to
be processed as individual loops. The
processing can therefore become time-
consuming. Moreover, each intersection
of the profiles and, in general, each con-
nection between different profiles of
the SP survey should be performed at
a common measuring point in the field.
To do so, it is recommended to flag those
connections points in the field (e.g.,
the references of each section can be
marked by placing a flag or a discreet
ring of stones) to be able to connect a
new profile at the exact same measure-
ment point of an older profile (i.e., using Figure 7. Two examples of simple data networks corrected by the closure
correction. (a and b) Two alternatives of the closure correction for a simple grid.
the same hole). (c and d) Two alternatives of the closure correction for a “volcano type” net
In the successive steps proposed in work, that is, circular profiles around the crater and the base of the cone + radial
Figure 7, the black dotted lines are the profiles on the flanks. In the successive steps, the black dotted lines are the data
data to which the closure correction that have already been corrected. The green dotted lines are the data currently
has already been applied; steps a4, b4, being corrected.

Figure 6. Example of the closure correction for an actual closed SP profile. The black line is the data set that has been previously
corrected for the reference correction, and the orange line is the same data set after the application of the closure correction. The
drift is exaggerated here, compared to the length of the profile, to get a perception of the effect of the closure correction. Note that
the last point (at a 4380 m distance) is corrected to reach the same value as the reference point (at a 0 m distance; that is geo
graphically the same point because it is a closed profile). Also note how this correction changes the SP values gradually without
affecting the location of the SP maxima and minima.
rule to keep in mind is that after being corrected for the Step 1) Correct an initial loop 1, which is a closed pro-
closure correction, a data point is fixed and its value file or a profile connected at both extremities to the
with respect to the unique reference of the data set can- same equipotential surface (e.g., the sea).
not be modified during the following steps of data Step 2) Connect a second loop (loop 2) to loop 1 by
processing. choosing one of the two extremities of loop 2
coinciding geographically with a point of loop 1.
Connect a new loop to already-corrected data The SP value will have to match the SP value of
As discussed before, after the reference correction, loop 1 at the connection location: This is done by
all the individual profiles have their reference (the first adding or subtracting the adequate SP value to all of
point of the profile) at 0 mV. After the first step of the the data point of loop 2.
closure correction (see the “Closure [or loop, or drift] Step 3) Distribute the drift along this new loop.
correction for one profile” section and the “Overview” Steps 4 and 5) Connect the next loop to the already-
section), we choose a second loop to be corrected. Be- corrected data (e.g., at step 4, loop 1 and loop 2);
cause the reference was set arbitrarily to 0 mV, the SP that is, choose the next loop to be corrected from
values at the connections between the two loops are not the remaining, uncorrected data, and apply steps 2
equal (Figure 8a). We first have to connect one extrem- and 3 to this new selected loop. The data that have
ity of the new loop (the green line in Figure 8) to the already been corrected for the closure correction
data that have previously been corrected (the black line are fixed and do not change. This step is repeated
in Figure 8) following a similar procedure as the one until the whole data set is properly combined and
described in the “Closure (or loop, or drift) correction corrected for the closure correction.
for one profile” section for the reference correction. For
that purpose, the SP value difference at the intersection
For practical reasons, a survey does not necessarily
is added to all data points of the new profile such that
start at the point that ultimately will be considered as
this extremity of the new profile/loop has an SP value
equal to one at the same geographic point in the al-
ready-corrected data. In our example (Figure 8), 4 mV
is added to each measurement point of the green loop to
make the value of the first point coincide with the black,
already-corrected loop (Figure 8b). After this operation,
the electrical potential difference remains the same be-
tween each measurement point and the reference of the
profile; that is, the same value has been added to each
data point of the profile.
In Figure 8b, we observe that, after connecting the
data of the new loop (i.e., adding the difference in
SP values of the connection point to the whole profile),
the SP values of the other extremity are not equal to the
same geographic point in the initial black loop. This ob-
served SP difference is the drift that must be distributed
along the new profile (see the “Closure [or loop, or drift]
correction for one profile” section). Figure 8c shows the
final result of the closure correction with the two ends
of the new loop displaying the same SP value as those
initially encountered at the same locations on the black
loop (Figure 8a). In the spreadsheet (Figure 9), we first
have to add the SP value from the already-corrected
loop at the connection, to all points of the new loop
(loop 2). Successively, the drift is distributed such that
the two extremities of the newly corrected loop (K2 and
Figure 8. Schematic decomposition of the closure correction
K6) display the same SP values as the reference loop for in a data set of two profiles. The black circles are data points
these identical geographic locations (D3 and D7; see already corrected for the closure correction. The green dots
also Figure 8c). From this step onward, the data belong- are the data to be corrected for the closure correction. The
ing to this loop are fixed and can be used as data points numbers correspond to the SP values for each data point
to which other loops can be connected in subsequent (in mV). (a) The two tone arrowheads represent the progres
steps of the closure correction procedure, as schemati- sion of the measurements in the field, and REF refers to the
reference (or initial point) of the new profile. The first step (a
cally displayed in Figure 7. and b) consists of connecting a new profile to a corrected one,
The closure correction of a multiprofile survey can and the second step (b and c) consists of the repartition of the
be summarized as follows: drift along the new profile.
the unique reference of the SP data set. When the refer- give a good idea of the meaning of some SP anomalies
ence and closure corrections have been applied to all in different contexts (e.g., Zlotnicki and Nishida, 2003;
collected data, the last step thus consists of assigning Jouniaux et al., 2009; Grobbe and Barde-Cabusson,
a 0 mV value to the point that we choose as the final, 2019). This section aims to give an overview of the vari-
unique reference of the data set. Preferably, this point ous possibilities to visualize SP data and extract useful
should be located at a water body existing in the study information, including a discussion on some common
area (sea, lake, or aquifer) or at any particular measure- elements of the interpretation of SP anomalies in envi-
ment point where no major SP variation is observed. ronmental science studies.
During this last step, we add or subtract the SP value
of the chosen point to the whole SP data set, so that SP maps
this point will be 0 mV and the relative SP values from SP data files contain data acquired in the field, often
one point to another do not change. with an irregular geographic distribution. These data
Finally, the SP values can be copied to a new Excel must be converted into an evenly spaced grid before
sheet along with a numbering column corresponding to being able to create a map representation. A grid file
the data points, the GPS coordinates, the names of the contains a list of Z values (here, our SP data), and
profiles, and any other information that the user may the spatial position assigned to each value into the grid
want to use when presenting the data. These SP values corresponds to the x and y geographic coordinates of
are the final product of the data processing and can be the data points in the field. To create the grid and the
used to visualize the data. corresponding SP map, we use dedicated data mapping,
modeling, and analysis software. The input data file is a
Data visualization and interpretation worksheet or text file in which all of the corrected SP
Now that we have successfully processed the raw SP data are presented in one column. The corresponding x
data, we can visualize the SP anomalies. and y coordinates fill two other columns. The work-
With relatively small measurement spacing, data sheet can contain title rows and other columns that may
plots/graphs can provide detailed information. How- be used to label the data points. For example, an addi-
ever, with a data set well distributed in the field, we tional column created with the numeration of individual
can also present SP data in maps, using a gridding/in- profiles is useful. Making mistakes transferring the data
terpolation, contouring, and surface mapping program. from a field notebook to a computer is very common,
The small measurement spacing and the great spatial and the GPS can also occasionally have low precision in
coverage give valuable information on different levels. the field. This produces artifacts in the resulting SP map
This section does not intend to detail all possible inter- if the coordinates are not corrected. A good idea is to
pretations of SP anomalies. Other publications already create a graph of the geographic coordinates of the data
points, labeled with the numeration of
individual profiles. A missing or mis-
placed point along a profile usually cor-
responds to a mistake in its coordinates
that can be corrected directly in the data
set. This check can sound trivial, but it
avoids important mistakes when the
data are interpolate, and it especially
helps to avoid the interpretation of SP
anomalies that are actually artifacts.
When the data set has been quality-con-
trolled accordingly, it is ready for the
gridding procedure.
As mentioned before, SP mapping in
uneven terrains (i.e., having slopes, de-
pressions, rivers, and inhabited areas),
as is the case most of the time, results
in data sets with heterogeneous spatial
coverage. As a result, on the one hand,
Figure 9. Closure correction of the SP data for the example schematically dis we obtain small measurement spacings
played in Figure 8. Loop 1 is the reference initial loop (the black color in Fig (usually 20 m or less) along each profile,
ure 8), and loop 2 is the new loop (to be corrected; the green color in Figure 8). whereas, on the other hand, profiles are
The upper part of the figure refers to the connection of the SP data performed by sometimes separated several hundreds
adding the intersection SP value from loop 1, to all data points of loop 2 (i.e.,
connecting the data). The lower part corresponds to the drift distribution. The
of meters from each other. The gaps be-
$ sign is the formula that is used (in Excel) to keep the cell coordinates (the tween profiles can be significant, and a
column letter and the row number) constant when the formula is copied through single interpolation of the data can lead
the rest of the column. to the appearance of artifacts that can
be visually misleading for the interpretation of SP maps. be used to perform a second interpolation/gridding, but
Although many gridding and filtering methods exist, a this time with smaller grid spacing (usually two times
solution to minimize the effect of heterogeneous data the spacing used along the SP profiles = 40 m). The re-
coverage is to perform a double interpolation of the sulting grid can be used to create the final SP map. The
data set (Figure 10). Kriging usually creates some arti- aim when creating a mesh of approximately 10 times
facts on the resulting map. These artifacts are often due the measurement step is to create a smoothed grid be-
to the limited search neighborhood typically used, as tween the SP profiles before the final interpolation (Fin-
described by Neufeld and Wilde (2005). The authors izola et al., 2002). Obviously, in interpolation, one can
proposed the use of all of the samples for kriging to gen- always make different choices for algorithms. However,
erate artifact-free maps. For data interpolations to gen- our key message here is to take a widely adopted inter-
erate SP maps, we thus suggest using a very large radius polation algorithm (kriging) implemented in many
of interpolation encompassing approximately half of all commercial software packages and demonstrate the
of the data set (e.g., a 1 km radius or more for the SP benefits of using our double-gridding method in addi-
data of our Stromboli volcano example),
with four sectors of search (northeast,
northwest, southwest, and southeast
quarters), a maximum of 64 data points
to use from all the sectors, and a maxi-
mum of 16 data points to use from each
sector. With a reduced radius of search,
artifacts appear between the measure-
ment profiles independently of any krig-
ing/variogram options used. Moreover,
kriging can be significantly affected
by the choice of variogram model (Gor-
sich and Genton, 2000; McClean and
Evans, 2000; Pyrcz and Deutsch, 2001;
Leuangthong and Deutsch, 2004). To
avoid any subjective selection of a vario-
gram model, we suggest using double
linear interpolation as a simple and ef-
fective method. It consists of two steps
of (1) creating a regional mesh that will
smooth high-frequency artifacts and
(2) a second interpolation that preserves
the detailed information measured
along the profiles. The challenge, espe-
cially for SP interpolation maps, is to
preserve the high-amplitude and short-
wavelength SP anomalies. Indeed, these
can be interpreted in terms of structural
limits (e.g., fault systems), which is one
of the key applications of the SP method
(Finizola et al., 2002, 2003, 2009; Bennati
et al., 2011). Basically, field data are
used to produce a grid with large spac-
ing (e.g., 10 times the minimum spacing
of the field data, e.g., 200 m for a field
data acquisition spacing of 20 m). From
this grid, a new series of data is ex-
tracted by converting the grid to an AS-
CII XYZ file, which can be edited in a
spreadsheet program. The data must Figure 10. Double interpolation/gridding of the SP data for an irregular distri
be combined with the original field data bution of the measurement points in the field. Step 1 creates a grid with a large
in a unique spreadsheet. This new data spacing (200 m). The resulting data are added to the original data in step 2 to
finally create a grid with spacing corresponding to two times the measurement
set created has a more even distribution spacing along individual profiles (40 m). The white dots represent the SP meas
of the data points in the whole study urement points. SP maps are extracted from the Stromboli SP map (modified
area and preserves the high resolution from Finizola et al. [2002]; see the whole map in Figure 11 with details of the
of the data along the SP profiles. It can differences and benefits of the double gridding at a greater scale).
tion to kriging. Here, the two gridding steps are per- box 4 in Figure 11) and the first-order information is
formed with the kriging method, which is also a good revealed. However, even with an enhanced representa-
option to interpolate unevenly sampled data. Combined tion, it is important to remember that the interpretation
with the double-gridding approach, the results show a of such maps should always be done keeping in mind
faithful representation of the data close to the profiles the acquisition geometry. There are many different
and smoothed artifacts far from them. High and low choices to make with respect to the interpolation. Our
filters, to filter out anomalous outliers, can also be used example demonstrates that different choices can im-
on the corrected SP data, to smooth the final result prove the final results and interpretability of the data.
(Thanassoulas, 1989). A downside of this approach One direction to explore, for example, is the addition
would be that the user has to define the upper and of topography, or other prior information, as a cokrig-
lower bounds of these filters, and it will be a bit arbi- ing property.
trary unless the user has clear physical reasons to select Once a basic map is created, several steps can be car-
these specific values. Filtering also means that anoma- ried out to enhance it and highlight as much information
lously high or low SP values, even though they can be as possible, such as blanking the map away from the
realistic, may be removed from the data. data points (i.e., removing interpolated areas outside
Figure 11 is an example case study using SP data of the area of interest, where no data are available), or
from Stromboli volcano. The general aspect of the creating a custom color scale. This will not modify the
map is relatively similar, but a rapid comparison of the information obtained from and represented by the data,
two maps exemplifies the usefulness of a double-gridd- but it will ease the perception of the first-order results in
ing process. Calculating a grid directly from the cor- the map, and it will aid in the interpretation of SP data.
rected SP data generates a high-frequency signal that
can mislead scientific interpretation. Indeed, the ap- SP versus distance graphs
pearance of short-wavelength SP anomalies in areas not SP data can be presented as graphs/plots, showing
well covered by field measurements can confuse the SP values as a function of the distance along a profile,
reader (e.g., the yellow anomaly in box 1 in Figure 11). or as a function of the altitude. As a function of the dis-
Artifacts can sometimes be very close to actual data and tance, graphs can be used to precisely localize, for ex-
look like large-scale anomalies (box 2 in Figure 11). ample, anomalies of the SP signal related to structural
This two-step procedure preserves the density of the discontinuities of the ground, such as faults and fissures
data along the profiles and the position and extension (Finizola et al., 2002; Bennati et al., 2011), to delineate
of the real SP anomalies (e.g., box 3 in Figure 11). The hydrothermally active zones (e.g., the typical W-shaped
general aspect of the map is greatly improved because SP signal on volcanoes) (Ishido, 2004; Gonzales et al.,
most of the high-frequency artifacts are removed (e.g., 2014), or to define the spatial extent of preferential

Figure 11. Study case example of a single gridding, directly to a 40 m resolution (the left side) compared with a double gridding
with interpolation at 200 and 40 m (the right side) of SP data from Stromboli (modified from Finizola et al., 2002). The white dots
represent the SP measurement points. The dashed boxes show artifacts that are being removed (boxes 1, 2, and 4) without altering
the resolution and accuracy of the map in areas well covered by the measurements in the field (box 3). The maps in Figure 10 show
the central part of the present figure.
groundwater flow pathways (Revil et al., 2005). This is interpreting an SP profile. We also want to emphasize
where the effort of acquiring data with short measure- that for SP versus distance graphs, the shorter the
ment spacing really pays off. Studies with 1 m measure- measurement spacing, the better the SP data can detect
ment spacing (Finizola et al., 2009), although time small-scale structures.
consuming, enable us to locate structural discontinu-
ities in great detail and allow for assessing potential lim- SP versus elevation graphs
its of collapses associated with preexisting or incipient The effect of the elevation variations or the so-called
fault planes (Barde-Cabusson et al., 2012; Chaput topographic effect (e.g., Zlotnicki and Nishida, 2003) in-
et al., 2019). volves a constant linear SP/elevation relationship (e.g.,
In the Garrotxa, a dormant monogenetic volcanic Lénat, 2007) explained by a water table whose depth
field located in northern Spain, the Rocanegra (RN) and increases regularly with elevation in a homogeneous
Puig Subià (PS) volcanic cones are thought to be built medium. In the W-shapes signal observed on active vol-
on the same north-northeast–south-southwest eruptive canoes, the external branches of the “W” are then ex-
fissure (Barde-Cabusson et al., 2014). They are two plained as the decrease of the signal while climbing
well-preserved volcanic edifices aligned along a the slopes of the volcano. SP versus elevation graphs
north-northeast–south-southwest direction that is also can be built using the whole SP data set of the study
reflected by gravimetry data acquired at the same study area and the elevation information at the same meas-
site. In the alignment of those two volcanic edifices, the urement points. It is useful in a hydrogeologic context
authors identified a clear SP minimum on one SP profile as well as in areas showing hydrothermal activity, to
intersecting the fissure (Figure 12b, profile a-b; the fis- delineate different hydrogeologic and hydrothermal do-
sure is represented as the dashed orange line). Profile mains. A simple approach consists of representing the
a-b crosses the fissure in a rough southwest–northeast SP versus elevation data in a graph. The strategy here is
direction, and, even if the signal is noisy, it clearly high- to identify and extract SP values and the corresponding
lights the area affected by the fissure. Usually, an SP elevation of the measurements from different key areas
anomaly associated with a fault would rather show of the study site. Those key areas are delimited by a gen-
up as a narrow peak in the SP data, but in this case, the eral knowledge of the geologic and structural features
ground may be fractured over a broader area due to of the study site or, more interestingly, the SP versus
past volcanic or tectonic activity. The northern half elevation plot itself will help in identifying different
of profile c-d follows a path nearly parallel to the align- groups of data corresponding to, for example, areas
ment of RN and PS volcanic cones, and then it shows with different hydrogeologic characteristics. Figure 13
decreasing SP values when approaching the eruptive shows two examples on the active Piton de la Fournaise
fissure. This SP profile shows a 150 mV minimum, volcanic cone (Réunion Island, France; Barde-Cabus-
well-defined by at least 35 measurement points, and in- son et al., 2012) and dormant Garrotxa monogenetic
terpreted as being associated to the eruptive fissure field (Catalonia, Spain; Barde-Cabusson et al., 2014).
(Barde-Cabusson et al., 2014). However, in this latter On the active Piton de la Fournaise area, which is a hy-
case, nothing confirms that profile c-d actually crosses drothermal context, different groups of the SP versus
the eruptive fissure and the SP minimum observed can elevation data are clearly identified because they show
be called a “relative minimum,” that is,
relative to the rest of the data in this pro-
file. This minimum indicates that the
profile passes close to the eruptive fis-
sure and, in a case in which no other
data would support the location of the
fissure, the information could be used
to plan follow-up data acquisition. This
is an important discussion for the inter-
pretation of SP profiles because (1) we
define “real” and “relative” minima that,
respectively, correspond to the actual
intersection of an SP profile with a fault
or fissure, or to an anomaly in the vicin-
ity of this structural limit and (2) we
highlight the fact that getting results of Figure 12. (a) Section of the SP map acquired in Garrotxa (Catalonia, Spain). (b
other geophysical or geochemical meth- and c) SP versus distance graphs along selected SP profiles (a b and c d). StaM,
ods available at the same location can RN, and PS stand for the Santa Margarida, Rocanegra, and Puig Subià volcanoes,
respectively, and ref is the reference for the SP data set. The dashed orange line
substantially help the interpretation of is the eruptive fissure between RN and PS, and the small white dots are the SP
such SP plots, and in any case, knowl- measurements. Profile a b defines a real SP minimum, whereas c d shows a rel
edge of the general geology of the study ative SP minimum (see the text for an explanation; [a and b] has been modified
area is a prerequisite for any attempt at from Barde Cabusson et al., 2014).
Figure 13. (a and b) The SP versus elevation graphs of the Piton de la Fournaise cone (Réunion Island, France) and Garrotxa
(Catalonia, Spain), respectively. (c and d) The data points displayed on the corresponding SP map and digital elevation map. The
colors of the data points in the corresponding graphs and maps (a and c, and b and d) are the same ([a and c] are modified from
Barde Cabusson et al., 2012; [b and d] are modified from Barde Cabusson et al., 2014).

different trends. Those groups correspond to different data, Lénat (2007) shows that computing the SP/eleva-
intensities of the hydrothermal activity likely due to tion gradient is useful for qualitative interpretation of
more (nearly vertical trend) or less (flatter trend) in- SP data; the author shows that the gradient is more sen-
tense hydrothermal flow. Low hydrothermal activity on sitive to variations of the SP signal associated to the
the flanks of the cone is characterized by a +1.7 mV/m presence of perched aquifers, to lateral variations of
gradient (corresponding to the slope of the linear re- underground physical properties, or to upward hydro-
gression of this group; Figure 13a). The summit area thermal circulations of fluids.
shows a different trend with a greater general slope,
and the two summit craters show a nearly vertical trend
Conclusion
associated to an intense hydrothermal activity. In the
The SP method is a simple method whether regard-
Garrotxa volcanic zone (Barde-Cabusson et al., 2014),
ing the acquisition of the measurements in the field, or
two major groups of data were identified with an
the processing of the data. However, the reference and
eastern zone (the black data points in Figure 13b) show- the closure corrections of SP data are vital to SP
ing a −0.19 mV/m trend characteristic of hydrogeologic processing and the significance of a good planning of
zones (Lénat, 2007). A western group (the gray data SP acquisition in view of applying those corrections,
points in Figure 13b) shows more chaotic behavior. that is, in particular that the prepared plan must include
Some local anomalies, such as the high infiltration closed profiles (loops) interconnected to common
along the Puig Subià eruptive fissure, can contrast with measurement points in the field for the purpose of ap-
the main trend (the blue data points in Figure 13b). The plying the closure correction. The reference correction
two zones are interpreted to be areas with different hy- allows us to connect different sections of a profile,
drogeologic contexts (Barde-Cabusson et al., 2014). whereas the closure correction allows us to correct
Going further in the processing of SP versus elevation the drift of SP measurements and sets a unique refer-
ence for the whole data set. This is an essential condi- the NSF EPSCoR ‘Ike Wai project and 50% by the Uni-
tion to visualize the result on a map. We show that data versity of Hawai‘i at the Mānoa Water Resources Re-
acquisition spacing is a critical parameter that will de- search Center. Support for the Hawai‘i EPSCoR
termine the size of the geologic and structural features Program is provided by the National Science Founda-
that we will be able to identify. Acquiring data with tion’s Research Infrastructure Improvement (RII)
small spacing along the survey lines (or profiles), even Track-1: ‘Ike Wai: Securing Hawaii’s Water Future
if the profiles are distant from each other, will give an Award # OIA-1557349.
opportunity to interpret the data with a great amount of This study contributes to the IdEx Université de
detail. Although SP versus distance graphs can locate Paris ANR-18-IDEX-0001. The work of A. Finizola is
structural limits precisely, maps and SP versus eleva- IPGP contribution #4141.
tion graphs identify hydrogeologic or hydrothermal
domains of different intensities at a greater scale. A pre-
Data and materials availability
liminary step consisting of double gridding the cor-
Confirm that the manuscript has been submitted
rected data set is not a compulsory step, but it has
solely to this journal and is not published, in press, or
shown to be useful to counteract the effects of an un-
submitted elsewhere.
even distribution of the measurements in the field in ad-
dition to the use of a method such as kriging. Using both
References
allows us to keep detailed information close to the data
points and to smooth artifacts far from it. SP maps will Aizawa, K., 2008, Classification of self-potential anomalies
highlight first-order information and can help to identify on volcanoes and possible interpretations for their sub-
key areas for continued studies (using SP and/or other surface structure: Journal of Volcanology and Geother-
geophysical methods and data types). mal Research, 175, 253 268, doi: 10.1016/j.jvolgeores
The robust practical data acquisition and processing .2008.03.011.
method that is proposed here will help several research- Anderson, L. A., and G. R. Johnson, 1973, Application of
ers across many disciplines to plan their SP surveys, the self-potential method in the exploration for geother-
acquire good-quality data, get reliable results with the mal energy in Long Valley, California: Eos, Transactions
method through an efficient processing, visualize the American Geophysical Union, 54, 1212.
data, and, finally, interpret the results for their own Anderson, L. A., and G. R. Johnson, 1976, Application of
fields of research. the self-potential method to geothermal exploration
in Long Valley, California: Journal of Geophysical Re-
Acknowledgments search, 81, 1527 1532, doi: 10.1029/JB081i008p01527.
We want to thank the members of the team led by A. Anderson, L. A., and G. R. Johnson, 1979, Some observa-
Finizola in the Aeolian Islands, who participated in data tions of the self-potential effect in geothermal areas in
acquisition, some of which is unpublished, and particu- Hawaii and Nevada: Geothermal Resources Council,
larly for the data acquisition resulting in the SP map of Transactions, 2, 9 12.
Stromboli that is used in this paper to demonstrate the Aubert, M., and Y. Q. Atangana, 1996, Self potential-method
two-step interpolation of SP maps. In the same way, we in hydrogeological exploration of volcanic areas:
are grateful to the numerous individuals who contrib- Ground Water, 34, 1010 1016, doi: 10.1111/j.1745-6584
uted to data acquisition on Piton de la Fournaise and .1996.tb02166.x.
at the Garrotxa volcanic field for the great teamwork Ballestracci, R., 1982, Self-potential survey near the craters
in acquiring the data used to illustrate some of the ex- of Stromboli volcano (Italy). Inference for internal
amples in this paper. The authors would like to thank M. structure and eruption mechanism: Bulletin of Volcan-
Aubert, J.-F. Lénat, A. Revil, E. Delcher, and the works ology, 45, 349 365, doi: 10.1007/BF02597257.
performed in collaboration with STRATAGEM974, for Banwell, C. J., 1970, Geophysical techniques in geothermal
very interesting discussions over several years. We
exploration: UN Symposium on the Development and
thank the associate editor of Interpretation V. Egorov
Utilization of Geothermal Resources 1, 32.
and our reviewers P. Wolfgram, M. Pezzoli, M. Saribu-
Barde-Cabusson, S., 2007, Formation de caldera par fluage
dak, K. Titov, and an anonymous reviewer for reviews
d’un système hydrothermal volcanique (in French):
that truly improved the final version of this manuscript.
This is contributed paper #2434 of the Hawai‘i Insti- Ph.D. thesis, Université Blaise Pascal, Clermont-Fer-
tute of Geophysics and Planetology, #11171 of the rand (France), 203.
School of Ocean and Earth Science and Technology, Barde-Cabusson, S., A. Finizola, A. Peltier, M. Chaput, N.
and WRRC-CP-2021-06 of the Water Resources Re- Taquet, S. Dumont, Z. Duputel, A. Guy, L. Mathieu, S.
search Center, University of Hawai‘i at Mānoa, Hono- Saumet, F. Sorbadère, and M. Vieille, 2012, Structural
lulu, Hawai‘i, USA. control of collapse events inferred by self-potential
S. Barde-Cabusson is currently 100% funded by the mapping on the Piton de la Fournaise volcano (La Ré-
National Science Foundation’s Established Program union Island): Journal of Volcanology and Geothermal
to Stimulate Competitive Research (NSF EPSCoR) Research, 209–210, 9 18, doi: 10.1016/j.jvolgeores.2011
‘Ike Wai project, and N. Grobbe is funded for 50% by .09.014.
Barde-Cabusson, S., A. Finizola, A. Revil, T. Ricci, S. Pis- variability of soil CO2 flux: Implication for monitoring
citelli, E. Rizzo, B. Angeletti, M. Balasco, L. Bennati, S. strategy: Journal of Volcanology and Geothermal Re-
Byrdina, N. Carzaniga, A. Crespy, F. Di Gangi, J. Morin, search, 366, 13 26, doi: 10.1016/j.jvolgeores.2018.10
A. Perrone, M. Rossi, E. Roulleau, B. Suski, and N. Vil- .001.
leneuve, 2009, New geological insights and structural Brothelande, E., A. Finizola, A. Peltier, E. Delcher, J.-C. Ko-
control on fluid circulation in La Fossa cone (Vulcano, morowski, F. Di Gangi, G. Borgogno, M. Passarella, C.
Aeolian Islands, Italy): Journal of Volcanology and Trovato, and Y. Legendre, 2014, Fluid circulation pat-
Geothermal Research, 185, 231 245, doi: 10.1016/j tern inside La Soufrière volcano (Guadeloupe) inferred
.jvolgeores.2009.06.002. from combined electrical resistivity tomography, self-
Barde-Cabusson, S., J. Gottsman, J. Martí, X. Bolós, A. G. potential, soil temperature and diffuse degassing
Camacho, A. Geyer, L. L. Planagumà, E. Ronchin, and A. measurements: Journal of Volcanology and Geothermal
Sanchez, 2014, Structural control of monogenetic vol- Research, 288, 105 122, doi: 10.1016/j.jvolgeores.2014
canism in the Garrotxa volcanic field (Northeastern .10.007.
Spain) from gravity and self-potential measurements: Castermant, J., C. A. Mendonça, A. Revil, F. Trolard, G.
Bulletin of Volcanology, 76, 788, doi: 10.1007/s00445- Bourrié, and N. Linde, 2008, Redox potential distribu-
013-0788-0. tion inferred from self potential measurements during
Barus, C., 1882, On the electrical activity of ore bodies. the corrosion of a burden metallic body: Geophysical
Geology of the Comstock lode and the Washoe District: Prospecting, 56, 269 282, doi: 10.1111/j.1365-2478
U.S. Geological Survey, Monographs of the United .2007.00675.x.
States Geological Survey, 309 367. Chaput, M., A. Finizola, A. Peltier, N. Villeneuve, M. Cro-
Bennati, L., A. Finizola, J. Walker, D. Lopez, C. Higuiera- visier, and S. Barde-Cabusson, 2019, Where does a
Diaz, C. Schütze, F. Barahona, R. Cartagena, V. Conde, volcano break? Using self-potential reiteration in fore-
R. Funes, and C. Rios, 2011, Fluid circulation in a com- casting the precise location of major destructive events
plex volcano-tectonic setting, inferred from self-poten- on active volcanoes: The case study of the Piton de la
tial and soil CO2 flux surveys: The Santa María Cerro Fournaise 2007 caldera collapse: Volcanica, 2, 151 159,
Quemado Zunil volcanoes and Xela caldera doi: 10.30909/vol.02.02.151159.
(northwestern Guatemala): Journal of Volcanology Corwin, R. F., 1973, Offshore application of self-potential
and Geothermal Research, 199, 216 229, doi: 10.1016/ prospecting: Ph.D. thesis, University of California, 303.
j.jvolgeores.2010.11.008. Corwin, R. F., 1975, Use of the self-potential method for geo-
Biswas, A., 2017, A review on modeling, inversion and in- thermal exploration: U.S. Energy Research and Develop-
terpretation of self-potential in mineral exploration and ment Administration Report, Technical information
tracing paleo-shear zones: Ore Geology Reviews, 91, division, Lawrence Berkeley Laboratory, University of
21 56, doi: 10.1016/j.oregeorev.2017.10.024. California, 53.
Blanchard, D. C., 1964, Charge separation from saline Corwin, R. F., 1976, Self-potential exploration for geother-
drops on hot surfaces: Nature, 201, 1164 1166. mal reservoirs: Proceedings of the 43rd Workshop on
Bolève, A., F. Janod, A. Revil, A. Lafon, and J.-J. Fry, 2011, Geothermal Reservoir Engineering, U.S. Government
Localization and quantification of leakages in dams us- Printing Office, Washington, D.C. 2, 937 945.
ing time-lapse self-potential measurements associated Corwin, R. F., and D. B. Hoover, 1979, The self-potential
with salt tracer injection: Journal of Hydrology, 403, method in geothermal exploration: Geophysics, 44,
242 252, doi: 10.1016/j.jhydrol.2011.04.008. 226 245, doi: 10.1190/1.1440964.
Bolós, X., S. Barde-Cabusson, D. Pedrazzi, J. Martí, A. Ca- Dessau, G., 1950, Some results of geophysical prospecting
sas, M. Himi, and R. Lovera, 2012, Investigation of the conducted for the Geological Survey of India from 1945-
inner structure of La Crosa de Sant Dalmaimaar (Cata- 1948: Geophysics, 15, 704 731, doi: 10.1190/1.1437635.
lan Volcanic Zone, Spain): Journal of Volcanology and Dickey, P. A., 1944, Natural potentials in sedimentary rocks:
Geothermal Research, 247–248, 37 48, doi: 10.1016/j Transactions of the AIME, 155, 39 48, doi: 10.2118/
.jvolgeores.2012.08.003. 944039-G.
Bolós, X., G. Cifuentes, J. L. Macías, G. Sosa-Ceballos, F. Doussan, C., L. Jouniaux, and J.-L. Thony, 2002, Variations
Garcia-Tenorio, and M. Albor, 2019, Geophysical imag- of self-potential and unsaturated water flow with time in
ing of fluid circulation and its relation with the struc- sandy loam and clay loam soils: Journal of Hydrology,
tural system of Cerritos Colorados geothermal field, 267, 173 185, doi: 10.1016/S0022-1694(02)00148-8.
La Primavera caldera (Mexico): Journal of Volcanology Eppelbaum, L. V., 2019, Advanced analysis of self-potential
and Geothermal Research, 369, 238 249, doi: 10.1016/j data in ore deposits of the South Caucasus: ANAS
.jvolgeores.2018.11.015. Transactions, Earth Sciences, 2, 21 35, doi: 10.33677/
Boudoire, G., A. Finizola, A. Di Muro, A. Peltier, M. Liuzzo, ggianas20190200029.
F. Grassa, E. Delcher, C. Brunet, P. Boissier, M. Chaput, Finizola, A., 2017, Approche pluridisciplinaire des circula-
V. Ferrazzini, and S. Gurrieri, 2018, Small-scale spatial tions de fluides au sein de volcans actifs: Habilitation à
 diriger des recherché (in French): Ph.D. thesis, Univer- Grobbe, N., and S. Barde-Cabusson, 2019, Self-potential
sité de La Réunion (France), 378. studies in volcanic environments: A cheap and efficient
Finizola, A., M. Aubert, A. Revil, C. Schütze, and F. Sortino, method for multiscale fluid-flow investigations: Interna-
2009, Importance of structural history in the summit tional Journal of Geophysics, 2019, 1 19, doi: 10.1155/
area of Stromboli during the 2002-2003 eruptive crisis 2019/2985824.
inferred from temperature, soil CO2, self-potential, and Halbwachs, M., 1983, Electrical and electromagnetic meth-
electrical resistivity tomography: Journal of Volcanol- ods, Chapître 34 in forecasting volcanic events, in H.
ogy and Geothermal Research, 183, 213 227, doi: 10 Tazieff and J.-C. Sabroux, eds., Developments in volcan-
.1016/j.jvolgeores.2009.04.002. ology 1: Elsevier, 507 528.
Finizola, A., J.-F. Lénat, O. Macedo, D. Ramos, J.-C. Hase, H., T. Hashimoto, S. Sakanaka, W. Kanda, and Y. Ta-
Thouret, and F. Sortino, 2004, Fluid circulation and naka, 2005, Hydrothermal system beneath Aso volcano
structural discontinuities inside Misti volcano (Peru) in- as inferred from self-potential mapping and resistivity
ferred from self-potential measurements: Journal of structure: Journal of Volcanology and Geothermal Re-
Volcanology and Geothermal Research, 135, 343 360, search, 143, 259 277, doi: 10.1016/j.jvolgeores.2004.12
doi: 10.1016/j.jvolgeores.2004.03.009. .005.
Finizola, A., F. Sortino, J.-F. Lénat, M. Aubert, M. Ripepe, Henriet, J. P., 1971, Large self-potential anomalies in the
and M. Valenza, 2003, The summit hydrothermal system Noir Ri basin (Ardennes, Belgium): Bulletin de la
of Stromboli. New insights from self-potential, temper- Société Belge de Géologie, de Paléontologie et d’hydro-
ature, CO2 and fumarolic fluid measurements, with logie, 80, 85 96.
structural and monitoring implications: Bulletin of Ishido, T., 2004, Electrokinetic mechanism for the
Volcanology, 65, 486 504, doi: 10.1007/s00445-003- “W”-shaped self-potential profile on volcanoes: Geo-
0276-z. physical Research Letters, 31, L15616, doi: 10.1029/
Finizola, A., F. Sortino, J.-F. Lénat, and M. Valenza, 2002, 2004GL020409.
Fluid circulation at Stromboli volcano (Aeolian Islands, Ishido, T., Y. Nishi, and J. W. Pritchett, 2010, Application of
Italy) from self-potential and soil gas surveys: Journal of self-potential measurements to geothermal reservoir
Volcanology and Geothermal Research, 116, 1 18, doi: engineering: Characterization of fractured reservoirs:
10.1016/S0377-0273(01)00327-4. 35th Workshop on Geothermal Reservoir Engineering,
Fournier, C., 1989, Spontaneous potentials and resistivity Stanford University, SGP-TR-188.
surveys applied to hydrogeology in a volcanic area: Jardani, A., A. Revil, F. Akoa, M. Schmutz, N. Florsch,
Case history of the chaîne des Puys (Puy de Dôme, and J. P. Dupont, 2006, Least squares inversion
France): Geophysical Prospecting, 37, 647 668, doi: of self-potential (SP) data and application to the
10.1111/j.1365-2478.1989.tb02228.x. shallow flow of ground water in sinkholes: Geophysi-
Fox, R. W., 1830, XXV. On the electro-magnetic properties cal Research Letters, 33, L19306, doi: 10.1029/
of metalliferous veins in the mines of Cornwall: Philo- 2006GL027458.
sophical Transactions of the Royal Society of London, Jardani, A., A. Revil, W. Barrash, A. Crespy, E. Rizzo, S.
120, 399 414, doi: 10.1098/rstl.1830.0027. Straface, M. Cardiff, B. Malama, C. Miller, and T. John-
Gay, S. P., Jr., 1967, A 1,800 millivolt self-potential anomaly son, 2009, Reconstruction of the water table from self-
near Hualgayoc, Peru: Geophysical Prospecting, 15, potential data: A Bayesian approach: Groundwater, 47,
236 245, doi: 10.1111/j.1365-2478.1967.tb01786.x. 213 227, doi: 10.1111/j.1745-6584.2008.00513.x.
Gex, P., 1977, La Méthode de polarisation spontanée, essai Johnston, M. J. S., J. D. Byerlee, and D. Lockner, 2001,
de réhabilitation: Ph.D. Thèse, Université de Lausanne Rapid fluid disruption: a source of selfpotential anoma-
(Switzerland), 97. lies on volcanoes: Journal of Geophysical Research,
Gex, P., 1980, Phénomènes d’électrofiltration liés à quel- 106, 4327 4335.
ques sites de barrages: Bulletin de la Société Vaudoise Jouniaux, L., A. Maineult, V. Naudet, M. Pessel, and P. Sail-
des Sciences Naturelles, 75, 39 50. hac, 2009, Review of self-potential methods in hydro-
Gonzales, K., A. Finizola, J.-F. Lénat, O. Macedo, D. Ramos, geophysics: Comptes Rendus Géoscience, 341, 928
J.-C. Thouret, N. Fournier, V. Cruz, and K. Pistre, 2014, 936, doi: 10.1016/j.crte.2009.08.008.
Asymmetrical structure, hydrothermal system and Kruger, F. C., and W. C. Lacy, 1949, Geological explanation
edifice stability: The case of Ubinas volcano, Peru, re- of geophysical anomalies near Cerro de Pasco, Peru: Eco-
vealed by geophysical surveys: Journal of Volcanology nomic Geology, 44, 485 491, doi: 10.2113/gsecongeo.
and Geothermal Research, 276, 132 144, doi: 10.1016/j 44.6.485.
.jvolgeores.2014.02.020. Kulessa, B., B. Hubbard, and G. H. Brown, 2003, Cross-
Gorsich, D. J., and M. G. Genton, 2000, Variogram model coupled flow modeling of coincident streaming and
selection via nonparametric derivative estimation: electrochemical potentials and application to subglacial
Mathematical Geology, 32, 249 270, doi: 10.1023/A: self-potential data: Journal of Geophysical Research,
1007563809463. 108, 2381, doi: 10.1029/2001JB001167.
Lénat, J.-F., 1987, Structure et dynamique interne d’un vol- Petiau, G., 2000, Second generation of lead-lead chloride
can basaltique intraplaque océanique: Le Piton de la electrodes for geophysical applications: Pure and
Fournaise (île de la Réunion): Thèse d’Etat, Univ. Blaise Applied Geophysics, 157, 357 382, doi: 10.1007/
Pascal, Clermont-Ferrand (France), 337. s000240050004.
Lénat, J.-F., 2007, Retrieving self-potential anomalies in a Pyrcz, M. J., and C. V. Deutsch, 2001, Two artifacts of prob-
complex volcanic environment: An SP/elevation gra- ability field simulation: Mathematical Geology, 33, 775
dient approach: Near Surface Geophysics, 5, 161 170, 799, doi: 10.1023/A:1010993113807.
doi: 10.3997/1873-0604.2006028. Rao, M. R., 1943, Spontaneous polarization surveys near
Lénat, J.-F., M. Aubert, and C. Archambault, 1982, Rela- Guddadarangavvanahalli, Chitaldrug, Mysore State, In-
tions spatiales et temporelles de la polarisation sponta- dia: Transactions of the AIME, 164, 107 113.
née (P.S.) et des températures à l’Etna. Perspectives de Revil, A., 2002, Comment on “Rapid fluid disruption: A
la méthode P.S. en tant que méthode de surveillance: source for self-potential anomalies on volcanoes” by
9ème Réunion Annuelle des Sciences de la Terre, Paris M. J. S. Johnston, J. D. Byerlee, and D. Lockner: Journal
(Soc. géol. Fr. ed.), 376. of Geophysical Research, 107, ETG 2-1 ETG 2-5, doi: 10
Leroy, P., and A. Revil, 2004, A triple-layer model of the .1029/2001JB000788.
surface electrochemical properties of clay minerals: Revil, A., L. Cary, Q. Fan, A. Finizola, and F. Trolard, 2005,
Journal of Colloid and Interface Science, 270, 371 Self-potential signals associated with preferential
380, doi: 10.1016/j.jcis.2003.08.007. ground water flow pathways in a buried paleo-channel:
Leuangthong, O., and C. V. Deutsch, 2004, Transformation Geophysical Research Letters, 32, L07401, doi: 10.1029/
of residuals to avoid artifacts in geostatistical modelling 2004GL022124.
with a trend: Mathematical Geology, 36, 287 305, doi: Revil, A., A. Finizola, S. Piscitelli, E. Rizzo, T. Ricci, A.
10.1023/B:MATG.0000028438.48852.b0. Crespy, B. Angeletti, M. Balasco, S. Barde-Cabusson,
Linde, N., and A. Revil, 2007, Inverting self potential L. Bennati, A. Bolève, S. Byrdina, N. Carzaniga, F. Di
data for redox potentials of contaminant plumes: Gangi, J. Morin, A. Perrone, M. Rossi, E. Roulleau,
Geophysical Research Letters, 34, doi: 10.1029/ and B. Suski, 2008, Inner structure of La Fossa di
2007GL030084. Vulcano (Vulcano Island, southern Tyrrhenian Sea,
Lowrie, W., 2007, Fundamentals of geophysics, 2nd ed.: Italy) revealed by high-resolution electric resistivity
Cambridge University Press. tomography coupled with self-potential, temperature,
McClean, C. J., and I. S. Evans, 2000, Apparent fractal di- and CO2 diffuse degassing measurements: Journal of
mensions from continental scale digital elevation mod- Geophysical Research, 113, B07207, doi: 10.1029/
els using variogram methods: Transactions in GIS, 4, 2007JB005394.
361 378, doi: 10.1111/1467-9671.00061. Revil, A., A. Finizola, T. Ricci, E. Delcher, A. Peltier, S.
Minsley, B. J., 1997, Modeling and inversion of self-poten- Barde-Cabusson, G. Avard, T. Bailly, L. Bennati, S. By-
tial data: B.S. Applied Physics, Purdue University. rdina, J. Colonge, F. Di Gangi, G. Douillet, M. Lupi, J.
Mounce, W. D., and W. M. Rust, 1945, Natural potentials in Letort, and E. Tsang Hin Sun, 2011, Hydrogeology of
well logging: Transactions of the AIME, 164, 288 294. Stromboli volcano, Aeolian Islands (Italy) from the in-
Neufeld, C. T., and B. J. Wilde, 2005, A global Kriging pro- terpretation of resistivity tomograms, self-potential, soil
gram for artifact-free maps: Centre for Computational temperature, and soil CO2 concentration measure-
Geostatistics, Report 7, University of Alberta, 403-1 ments: Geophysical Journal International, 186, 1078
401-8. 1094, doi: 10.1111/j.1365-246X.2011.05112.x.
Nourbehecht, B., 1963, Irreversible thermodynamic effects Revil, A., and A. Jardani, 2013, The self-potential method
in inhomogeneous media and their applications in cer- theory and applications in environmental geosciences,
tain geoelectric problems: Ph.D. thesis, Massachusetts 1st ed.: Cambridge University Press, 383.
Institute of Technology. Revil, A., and P. A. Pezard, 1998, Streaming potential
Nyquist, J. E., and C. E. Corry, 2002, Self-potential: The anomaly along faults in geothermal areas: Geophysical
ugly duckling of environmental geophysics: The Lead- Research Letters, 25, 3197 3200, doi: 10.1029/
ing Edge, 21, 446 451, doi: 10.1190/1.1481251. 98GL02384.
Parasnis, D. S., 1973, Mining geophysics: Elsevier, 394. Saribudak, M., 2019, Surface geophysics, Edwards and
Perrier, F. E., G. Petiau, G. Clerc, V. Bogorodsky, E. Erkul, Trinity aquifers, central Texas, in J. M. Sharp Jr., R.
L. Jouniaux, D. Lesmes, J. Macnae, J. M. Meunier, D. T. Green, and G. M. Schindel, eds., The Edwards aquifer,
Morgan, D. Nascimento, G. Oettinger, G. Schwarz, H. the past, present and future of a vital resource: Geologi-
Toh, M. J. Valiant, K. Vozoff, and O. Yazici-Cakin, cal Society of America Memoir, 215.
1997, A one-year systematic study of electrodes for long Saribudak, M., and N. W. Hauwert, 2017, Integrated geo-
period measurements of the electric field in geophysical physical investigations of Main Barton Springs, Austin,
environments: Journal of Geomagnetism and Geoelec- Texas, USA: Journal of Applied Geophysics, 138, 114
tricity, 49, 1677 1696, doi: 10.5636/jgg.49.1677. 126, doi: 10.1016/j.jappgeo.2017.01.004.
Sato, M., and M. H. Mooney, 1960, The electrochemical (Sicile): Comptes rendus de l’Académie des Sciences,
mechanism of sulfide self-potentials: Geophysics, 25, 290D, 443 446.
226 249, doi: 10.1190/1.1438689. Villasante-Marcos, V., A. Finizola, R. Abella, S. Barde-
Schiavone, D., and R. Quarto, 1984, Self-potential prospec- Cabusson, M. J. Blanco, B. Brenes, V. Cabrera, B. Casas,
ting in the study of water movements: Geoexploration, P. De Agustin, F. Di Gangi, I. Dominguez, O. Garcia, A.
22, 47 58, doi: 10.1016/0016-7142(84)90005-X. Gomis, J. Guzman, I. Iribarren, G. Levieux, C. Lopez, N.
Semenov, A. S., 1980, Electrical exploration: Natural Luengo-Oroz, I. Martin, M. Moreno, S. Meletlidis, J.
polarization method (inRussian), 446. Morin, D. Moure, J. Pereda, T. Ricci, E. Romero, C. Schü-
Sheriff, R. E., 1978, A first course in geophysical explora- tze, B. Suski-Ricci, P. Torres, and P. Trigo, 2014, Hydro-
tion and interpretation (in German): International Hu- thermal system of Central Tenerife volcanic complex,
man Resources Development Corporation, 313. Canary Islands (Spain), inferred from self-potential mea-
Sill, W. R., 1983, Self potential modeling from primary surements : Journal of Volcanology and Geothermal Re-
flows: Geophysics, 48, 76 86, doi: 10.1190/1.1441409. search, 272, 59 77, doi: 10.1016/j.jvolgeores.2013.12.007.
Soudoplatoff, S., J.-F. Lénat, and J. P. Vilbonnet, 1982, Car- Zablocki, C. J., 1976, Mapping thermal anomalies on an
tographie des anomalies thermiques du Piton de la active volcano by the self-potential method, Kilauea,
Fournaise. Résultats obtenus par thermographie in- Hawaii: Proceedings of the 2nd U.N. Symposium of
frarouge et par cartographie de la polarisation sponta- the development and use of geothermal resources 2,
née (P.S.): 9ème Réunion Annuelle des Sciences de la 1299 1309.
Terre, Paris (Soc. géol. Fr. ed.), 583. Zlotnicki, J., and Y. Nishida, 2003, Review on morphologi-
Suski, B., A. Revil, K. Titov, P. Konosavsky, M. Voltz, C. Dagès, cal insights of self-potential anomalies on volcanoes:
and O. Huttel, 2006, Monitoring of an infiltration experi- Surveys in Geophysics, 24, 291 338, doi: 10.1023/B:
ment using the self-potential method: Water Resources GEOP.0000004188.67923.ac.
Research, 42, W08418, doi: 10.1029/2005WR004840. Zohdy, A. A. R., L. A. Anderson, and L. J. P. Muffler, 1973,
Thanassoulas, C. P., 1989, Application of the self poten- Resistivity, self-potential, and induced-polarization
tial technique over the Milos geothermal test site: Geo- surveys of a vapor-dominated geothermal system: Geo-
thermics, 18, 497 505, doi: 10.1016/0375-6505(89) physics, 38, 1130 1144, doi: 10.1190/1.1440400.
90052-7.
Van Ngoc, P., D. Boyer, and G. Kieffer, 1980, Sur l’existence
d’une zone de forte conductivité électrique, en relation Biographies and photographs of the authors are not
avec les éruptions sur le versant nord-est de l’Etna. available.