See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/354937573

Development and Validation of a Low-cost DC Resistivity Meter for

Humanitarian Geophysics Applications

Article in Geophysics · September 2021

DOI: 10.1190/geo2021-0058.1

CITATIONS READS

3 182

6 authors, including:

Jeffrey Shragge Richard Krahenbuhl

Colorado School of Mines Colorado School of Mines
155 PUBLICATIONS 1,296 CITATIONS 68 PUBLICATIONS 580 CITATIONS

SEE PROFILE SEE PROFILE

Yalo Nicaise
University of Abomey-Calavi
42 PUBLICATIONS 251 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Sedimentary basins of Niger: geodynamic context and geological resources. View project

SubTER View project

All content following this page was uploaded by Jeffrey Shragge on 28 February 2022.

The user has requested enhancement of the downloaded file.

 GEOPHYSICS, VOL. 87, NO. 1 (JANUARY-FEBRUARY 2022); P. WA1–WA14, 12 FIGS., 5 TABLES.
10.1190/GEO2021-0058.1
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Development and validation of a low-cost direct current resistivity meter

for humanitarian geophysics applications

Dana Sirota1, Jeffrey Shragge2, Richard Krahenbuhl2, Andrei Swidinsky2, Nicaise Yalo3, and John
Bradford2

ABSTRACT conductive environments. Our instrument was tested in a Geo-

scientists Without Borders project jointly run between researchers
Insufficient access to safe drinking water is one of the most from Colorado School of Mines (CSM) and Université d’Abo-
DOI:10.1190/geo2021-0058.1

challenging global humanitarian issues. The development of mey-Calavi (UAC), Cotonou, Benin. A key project component
low-cost microcontrollers and the widespread availability of involved CSM and UAC students constructing and validating
cheap electronic components raise the possibility of developing two low-cost DC resistivity meters and then using these instru-
and using low-cost geophysical instrumentation with open-source ments for fieldwork aimed at better characterizing and monitoring
designs and software solutions to circumvent geophysical instru- the health of a local aquifer used as a groundwater source for com-
mentation capital cost issues. To these ends, we alter an existing munities in the Cotonou region. The low-cost instruments were
low-cost direct current (DC) resistivity meter design and develop successfully used alongside a commercial resistivity meter to ac-
an optional modular Raspberry Pi data-logging system to improve quire data for 2D inversion of aquifer hydrostratigraphy, indicat-
the unit’s functionality and usability and to ensure data integrity. ing the presence of a clay-sand contact. The costs of the
Numerical modeling and physical testing demonstrate that the sys- redesigned instrument and data logger, respectively, are $177
tem is more robust than previously published low-cost designs and and $108 (in 2021 USD) with future cost reductions possible,
works in a more diverse range of geologic scenarios — especially which are fractions of the price of commercial resistivity meters.

INTRODUCTION challenges (UNESCO, 2012). Moreover, water-access challenges

are expected to worsen in the coming decades in many parts of
Insufficient access to safe drinking water is one of the most chal- the world (UNEP, 2002), especially in Africa, where resources
lenging and widespread humanitarian issues present throughout the are becoming more scarce due to the rapid population increase,
world today. According to the 2019 United Nations Human Devel- the gradual depletion of existing aquifers, increasing aquifer salin-
opment Report, almost 30% of people do not have reliable access to ity, and the deleterious effects of global and local climate change.
clean and safely managed potable water resources and 55% do not Water scarcity is defined by access to less than 1000 m3 per per-
have access to clean water for sanitation purposes (UNDP, 2019). son per annum (i.e., for all personal, agricultural, and industrial
Like many humanitarian challenges, issues surrounding safe access uses) (UNESCO, 2012), and the challenges of water scarcity are
to groundwater vary geographically, with sub-Saharan African, driven by more than environmental factors alone. In areas where
East Asian, and Andean communities facing the most significant groundwater is available, it is often very expensive to drill wells,

Manuscript received by the Editor 23 January 2021; revised manuscript received 7 April 2021; published ahead of production 29 September 2021; published
online 12 November 2021.
1
Formerly Colorado School of Mines, Department of Geophysics, Golden, Colorado 80401, USA; presently Pacific Northwest National Labs, Richland,
Washington 99354, USA. E-mail: [email protected].
2
Colorado School of Mines, Department of Geophysics, Golden, Colorado 80401, USA. E-mail: [email protected] (corresponding author); rkrahenb@
mines.edu; [email protected]; [email protected].
3
Université d’Abomey-Calavi, Calavi, Bénin. E-mail: [email protected].
© 2022 The Authors. Published by the Society of Exploration Geophysicists. All article content, except where otherwise noted (including republished
material), is licensed under a Creative Commons Attribution 4.0 Unported License (CC BY). See http://creativecommons.org/licenses/by/4.0/. Distribution
or reproduction of this work in whole or in part commercially or noncommercially requires full attribution of the original publication, including its digital
object identifier (DOI).

WA1
 WA2 Sirota et al.

which is compounded by the risk of drilling a dry hole, consuming reducing the barriers to entry for hydrogeophysical investigations
limited resources with no community benefit. In other scenarios in throughout the world.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

which drilling does occur, boreholes often are sited based on com- One of the best-known examples is the low-cost DC resistivity
munity proximity and need rather than, more strategically, on hy- meter developed and published by Clark and Page (2011). Hereafter
drogeologic or hydrogeophysical analyses that could improve referred to as CP2011, this instrument is a simple and elegant DC
overall drilling success. resistivity meter with an open-source design that costs less than
It is now well established that (hydro)geophysics is a valuable $250 to construct. The authors have trained numerous local field
tool for locating new freshwater sources in shallow and deep aqui- crews to build and use the instrument and interpret the results,
fers alike (Martinelli, 1978; Goldman and Neubauer, 1994; Pham and they have successfully found groundwater resources in many
et al., 1994; Vouillamoz et al., 2007; Lindsay et al., 2015; Aré- locations around the world (e.g., Chad, Tanzania, Nigeria, DR
touyap et al., 2019; Cavalcante Messias et al., 2019). Although us- Congo, and Rwanda). Although we are aware that this instrument
ing geophysics can improve drilling success rates, geophysical is in use around the world, it is unclear how well it works for a wide
surveys by themselves can be prohibitively expensive. For example, range of groundwater scenarios because many practitioners do not
the cost of a basic four-electrode direct current (DC) resistivity sys- publish their results.
tem can exceed $500 USD and may not be sufficient for ground- Several other authors have developed and tested low-cost DC re-
water investigations, whereas a high-end commercial DC resistivity sistivity meters with varying degrees of success and applicability
instrument can cost upward of $10,000 USD. A key reason is that (Olowofela et al., 2005; Awotoye and Selemo, 2006; Toll and Has-
commercial-grade geophysical hardware is engineered to work in san, 2012, 2014; Fatahillah et al., 2019). Many of these designs
the toughest physical environments (e.g., temperatures ranging be- were tested and used in laboratory settings (Fatahillah et al.,
tween ±50°C) and for the most challenging subsurface geologic 2019). Some designs include multichannel and automated acquis-
conditions with the associated high-accuracy and -precision require- ition systems (Toll and Hassan, 2012, 2014), use difficult-to-source
ments. However, many options are effectively overengineered for a components that require international shipping, and are challenging
wide variety of hydrogeophysical applications in which more basic to build by users unfamiliar with advanced electronics. Finally,
geophysical instrumentation would suffice. Successful use of hy- while other designs have been used successfully to acquire field
DOI:10.1190/geo2021-0058.1

drogeophysical methods also commonly depends on the availability data appropriate for ultrashallow surveying (Olowofela et al.,
of experienced practitioners to acquire, process, and interpret geo- 2005; Awotoye and Selemo, 2006), none are as widely used as
physical data. Anecdotal evidence from geophysicists working on the CP2011 system.
humanitarian projects suggests that the most significant “barriers to Although low-cost instrumentation offers clear benefits, there are
entry” for undertaking hydrogeophysical work are a lack of access challenges associated with this approach. First, unlike commercial-
to (1) geophysical equipment; (2) geophysical software, computer grade hardware used in a broader range of applications, low-cost
hardware, and the internet; and (3) practical training. instruments are generally engineered for a narrow range of appli-
Of the available geophysical techniques, electrical resistivity cations including groundwater exploration. This can lead to deploy-
(ER) is one of the most widely used and reliable methods for locat- ment scenarios in which one expects a design to work but upon
ing groundwater due to its sensitivity to the presence of subsurface arrival finds the instrument poorly suited to the local environment.
fluids. In particular, DC-resistivity methods have a long history of Second, most field-ready low-cost instruments use handwritten data
use for finding fluids, taking shape as we know them in the early acquisition, due (presumably) to the added complexity of incorpo-
1900s (Schlumberger, 1920). DC-resistivity data generally require rating an automated data logger. This means that geophysical sur-
minimal processing, and inversion can now be performed with a veys completed with these instruments are slow and error-prone.
variety of freely accessible software packages that are available Third, instrumentation designed using components that are low-cost
in multiple (human) languages. The resulting models are often — but only available in a limited geographical area — are sub-
straightforward to interpret in terms of the geology and predicted optimal because the malfunction of even one component can render
water sources. However, like other geophysical instruments, ER the instrument inoperable because of the limited availability of, or
surveying can be expensive and buying or renting commercial knowledge about, possible replacement components. For all of its
ER instruments can cost thousands of dollars (before factoring benefits, the CP2011 instrument suffers from these drawbacks: It is
in labor, expert data interpretation, and commercial software costs). limited in more conductive settings in which one can overload the
Furthermore, commercial-grade equipment is expensive and diffi- system through what is effectively a short circuit through the ground
cult to repair, which can be a significant problem in an international (though circumventing the inverter and only using a 12 V signal
context. Finally, instruments cannot be left in the field for long-term sometimes resolves this problem [J. Clark, personal communica-
monitoring due to the cost and for fear of theft or damage. Overall, tion, 2019]); it requires handwritten data; and it is arguably more
these issues represent barriers to entry for many groups looking to complicated than need be, which makes it more difficult to main-
use DC resistivity methods for groundwater surveying. tain, troubleshoot issues, or replace components.
The recent development of low-cost microcontrollers (e.g., Rasp- The primary contribution of this paper is to improve on the
berry Pi [Rπ], Arduino) and the widespread availability of cheap CP2011 DC resistivity system. Our first goal is to improve system
electronic components raise the possibility to avoid costly geo- reliability and operability so that it can operate in a broader range of
physical hardware: do-it-yourself (DIY) low-cost geophysical in- geologic environments, especially for shallow aquifer systems char-
strumentation using open-source designs and software solutions. acterized by alternating lithologies forming conductive clay/shale
Although this concept is not original and is broadly reflected in to- aquitards and resistive sandy aquifer units from which groundwater
day’s growing “maker culture,” low-cost DIY geophysical instru- extraction is usually cost effective. The second goal is to add a
ments — if done well — offer an alternate path forward for Rπ-based data-logging system to improve system functionality
 Low-cost DC resistivity meter WA3

and usability and data integrity. Our final goal is not only to improve inspired by these designs (Clark and Page, 2011; Clark et al.,
on the initial CP2011 design but also to simplify the system build 2016) and has been modified to meet our project objectives.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

and operation while using near universally available components Discussions with other practitioners who have built and used the
without overly increasing the cost. CP2011 instrument provide anecdotal evidence of other people who
As part of this paper, the redesigned system and add-on automated have obtained accurate, high-quality results when using DC-resistiv-
data-logger module were built, tested, and validated in sandbox ex- ity instruments built off of the original design. However, initial tests of
periments and at the Colorado School of Mines (CSM) Geophysical our equivalent instrument build at multiple field sites in the Golden,
Discovery Lab (GDL). In addition, the instrument was used in a Geo- Colorado, area yielded poor and unusable results. Although the in-
scientists Without Borders (GWB) project jointly run by researchers strument would give accurate and reproducible readings at close elec-
from CSM and the Université d’Abomey-Calavi (UAC) in Cotonou, trode spacings (i.e., <10 m), when deployed in the field using more
Benin, which developed from a decade-long research collaboration widely spaced electrode configurations commonly used in hydrogeo-
between primary investigators at the two institutions stemming from physical investigations, we encountered a debilitating issue of over-
an earlier GWB project. The current GWB project involved instruct- loading the inverter when turning on the system, which rendered the
ing UAC students in the rudiments of hydrogeophysical surveying instrument inoperable. Further experimental analysis suggested that
including the use of DC resistivity, the construction and validation these issues were caused by excessive current flowing into the highly
of low-cost DC resistivity meters, and the subsequent use of the in- conductive Golden-area near-surface clays, which either overloaded
struments in fieldwork aimed at better characterizing a local aquifer the inverter or created growing feedback in the system. Thus, the key
used as a potable groundwater source for communities throughout the contributions of this paper are a redesign of the circuitry behind the
Cotonou region. CP2011 system to forestall these issues and thereby generalize its
The paper begins a review of previous work on developing low- functionality for use in a wider range of geologic environments, es-
cost resistivity instruments and discusses their strengths and limi- pecially more conductive settings.
tations. This is followed by our theoretical modeling, redesign, and
validation of an instrument based on an earlier version (including Instrument redesign
the incorporation and validation of Rπ microcomputers for autolog-
ging). Then, we present results from the CSM-UAC GWB hydro- The redesign required modeling and reconfiguring some compo-
DOI:10.1190/geo2021-0058.1

geophysics project that demonstrate the utility of the instrument for nents in the circuit design, including updates of the current injection
groundwater investigations within an international development box, how the voltmeter and ammeters attach to the individual
context. Finally, we analyze the cost and usefulness of the rede- switches, and the data acquisition procedure. The three guiding de-
signed instrument, review its successes and limitations, and provide sign principles behind the redesign work are to ensure that the final
our recommendations for future design improvements. instrument build (1) remains straightforward to construct by users
with only limited technical experience in electronics, (2) requires a
minimal number of low-cost components, each of which is easily
LOW-COST DC-RESISTIVITY METER DESIGN
accessible around the world, and (3) facilitates accurate data acquis-
One of the most detailed and well-recognized DIY DC-resistivity ition by experts and nonexperts alike. A further design extension
instruments for low-cost geophysical surveys is presented in Clark was developing a modular Rπ-based data-logger system, which al-
and Page (2011) (see therein for a schematic). The C P2011 DC lows the user to choose either to build and use an instrument run-
resistivity instrument can be powered by a 12 V DC car battery ning in the base mode (i.e., handwritten acquisition) or with an
via a standard 120 V (or 220 V) DC-alternating current (AC) automated data-logger system, depending on the user’s skill set,
inverter, or directly with a 120 V (or 220 V) AC power source project needs, and/or financial considerations.
(e.g., a generator). In either case, the AC voltage is then passed The first main requirement of a DC-resistivity system is to inject
across a bridge rectifier, a type of diode bridge that converts the a source current into the ground. The most straightforward way to
alternating current into 120 V (or 220 V) DC. This step is critical inject current is to attach a high-voltage source using high-quality
to the low-cost instrument build because it allows the system to in- electrical wires to two metal electrodes. To obtain a high-voltage
ject significant current into the ground using a high-voltage source, power source, a 12 V DC battery is connected to a power inverter,
thereby producing a stronger signal at the potential electrodes. One which boosts the voltage to an approximately 120 V AC source.
can use off-the-shelf digital multimeters to acquire current and volt- This combination represents the simplest way to obtain a high-volt-
age readings. The system incorporates a two-way switch to flip the age source while satisfying the design criteria described previously;
current direction during the survey so as not to unidirectionally po- however, these are generally the most expensive system compo-
larize the ground or electrodes. By using the switch, current can be nents. In general, any AC power source would suffice and, although
sent in either direction and measurements averaged to mitigate the the system is modeled on 120 V, the instrument will work with other
voltage differences caused by, e.g., self-potential effects. Finally, a high-voltage sources, including when plugged into a 220 V conven-
capacitor is introduced in parallel with the ground to smooth out tional generator or an electrical outlet in countries with a 220 V
sharp on-off transitions and absorb any erratic AC transients. electricity grid.
Clark and Page (2011) test the instrument against the Advanced The power source is then fed through a bridge rectifier (see
Geosciences Inc. (AGI) Sting R1 commercial standard and show Figure 1), a system of four diodes that converts AC power to an
that the data agree within acceptable limits. They use the low-cost equivalent DC current. To flow current through the two electrodes
instrument for several vertical electric soundings (VESs) and make in either direction and to include an “off” setting, a double-pole dou-
interpretations consistent with known geology. Their Wenner sur- ble-throw (DPDT) “on-off-on” switch is used. To protect the inverter
veys can be expanded out to an AB electrode spacing of up to 150 m from being overloaded, we place the electrodes in parallel with a
without a loss of data quality. The instrument discussed herein is resistor and capacitor (RC) circuit system. The large film capacitor
 WA4 Sirota et al.

allows for bidirectional current and smooths out the off-on and on-off approach was to model the RC circuit given the physical and
transients from rapid current injection (Gupta et al., 1972). The re- economic constraints of the updated instrument design.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

sistor slows the capacitor’s charging and reduces the power draw,
thereby allowing the use of a lower wattage (and lower cost) inverter. RC circuit modeling
However, the specific capacitor and resistor requirements need to be
tuned based on the physics of the RC circuit system (see below). To examine the trade-off of different inverter choices, we under-
The second main requirement of the DC-resistivity system is to took a circuit modeling exercise aimed at determining what com-
record the ground response, which can be processed and inverted ponents were necessary to achieve the desired design. This allowed
to estimate the subsurface resistivity distribution. To measure ground us to ask questions such as: How large should the internal resistor
be? And what minimum wattage rating should the inverter have?
response data, a voltage must be acquired between the two electrodes
We emphasize that the basic instrument design remains unchanged
(commonly referred to as MN), which can be accomplished using a
throughout the modeling process, just the size/rating of certain elec-
standard off-the-shelf digital multimeter found in local hardware
tronics components are varied. The numerical models effectively
stores. To use the voltage data described previously, the injected cur-
test whether the instrument can stay below the power limits imposed
rent must also be recorded; thus, an ammeter (i.e., recorded using a
by an inverter for different ground conditions and when using vari-
different digital multimeter setting) is placed in series with the electro-
ous internal resistor ratings. Sirota (2020) presents the theory and
des (see Figure 1). The digital multimeters chosen for use in this
ordinary differential equation (ODE) developed for the RC circuit
project display current and voltage readings with at least 2.5 signifi-
that we solve below.
cant digits. We made this choice because 2.5 and 3.5 digit multimeters
Before proceeding, it is important to provide values and ranges
are relatively inexpensive, standard off-the-shelf products and provide
for key parameters used in this study: (1) based on previous expe-
the precision needed to obtain reliable apparent resistivity values.
rience, we expect resistance values measured between the electrodes
With these two measurements, users can calculate the apparent resis- to fall within the 0.03–15.0 kΩ range; (2) because film capacitors
tivity for the measurement location. are not made with a wide range of capacitance, we assume that this
To be useful for DC resistivity acquisition, the current must be parameter is fixed at C ¼ 1 μF with a fixed source voltage of 120 V;
injected at a steady and high voltage. To enable this in DC, the instru- (3) due to cost considerations, we consider at most a 700 W inverter;
DOI:10.1190/geo2021-0058.1

ment must use an inverter to convert power from a 12 V battery into and (4) the initial condition assumes a charge-free capacitor.
approximately 120 V of AC power. The current is again converted to We solve the associated ODE numerically by starting with our
DC — this time with a 1:1 voltage — using a bridge rectifier. This initial conditions and computing the voltage on the capacitor using
powers an RC circuit consisting of an RC placed in series, which Runga-Kutta and Adams-Bashforth numerical quadrature (Süli and
together are arranged in parallel with two measurement electrodes Mayers, 2003). We evaluated the solution accuracy by examining
connected through the “resistor” that is the earth. An ammeter is at- the difference in the results from the two methods and by seeing
tached in series with the earth, and a switch serves to switch the cur- whether the solutions follow the expected trends. We used the more
rent on, off, and on in the opposite direction (see Figure 1). computationally expensive, but more straightforward to implement,
The power inverter is one of the most important components in Runga-Kutta method because the comparative results showed neg-
the instrument design. From an economic perspective, the cost of an ligible differences. The problem of interest runs over a short “on”
inverter is roughly proportional to its power rating (i.e., an inverter time of the circuit, so computational efficiency was not of primary
capable of outputting 1000 W is about twice as expensive as a concern.
500 W inverter) where the present day (2021) absolute cost is ap- We modeled the RC circuit response from time 0.0 to 0.025 s
proximately $0.10 per watt. Thus, in the interest of lowering costs, a using N ¼ 200 time steps. We then repeated this analysis over a
key design criterion is to use an inverter with sufficient — but not range of R1 and ground resistance Rg values, which permitted in-
excessive — power output. Inverters typically use a rating of the vestigation of the effects of using different resistors in the circuit
maximum sustained power output but can handle momentarily design for different physical earth scenarios. We then determined
higher amounts of power draw. However, instead of conducting the maximum and median power draws of the circuit for each
trial-and-error experiments with different power inverters, our (R1 ,Rg ) configuration to evaluate the circuit requirements based
on the peak and average loads that an inverter
can handle. Color scales were limited to power
draws reasonable for a sub-$100 inverter.
Figure 2a presents the numerical test results in
which the horizontal axis is the on time, the ver-
tical axis is the resistor value R1 (in log10 scale),
and the color shading represents the power (in
watts). For lower R1 values (i.e., 0.1 kΩ), we
note that the power draw has a strong and rapid
transient before asymptoting to a low value. For
Figure 1. Basic circuit diagram of the current injection system. From left to right: the higher R1 values, though, little to no transient
120 V AC power source, assumed to be the output of a power inverter, is directed to a behavior is observed. Figure 2b presents the log-
bridge rectifier, through which the input AC power is output as DC power of the same arithmic growth of the voltage on the capacitor as
voltage. The RC are placed in parallel to the DPDT switch (the dashed rectangle). The
switch sends current between the AB electrode pair in the ground in two different di- a function of time after the power is turned on.
rections, depending on the switch polarity. The ammeter is placed in series and measures The results show that the capacitor charges up to
the current injected into the ground. the value of the 120 V source voltage and that the
 Low-cost DC resistivity meter WA5

power draw of the circuit asymptotically approaches P ¼ V 2s ∕Rg , system cost as of March 2021 is $177, which is broken down by
which is the power draw of the earth once the capacitor is fully component in Table 1. Figure 4 shows a fully constructed system
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

charged. The power draw from the ground is rather low, even in acquiring DC resistivity data. The current injection system is plugged
conductive soils (i.e., as low as 0.03 kΩ). Overall, the power draw into an inverter and attached to a multimeter being used as an am-
is no more than 500 W and is generally under 400 W. This has sig- meter. A second multimeter attached to the MN electrodes measures
nificant cost implications because 400 and 500 W inverters are in- the voltage difference during current injection.
expensive and common, whereas 700 W inverters (and higher
ratings) are higher cost and more difficult to find.
Because the lower bound of the power draw is asymptotically
controlled by the ground resistance (Rg ), we should examine the
effects of R1 on the upper limits of the power draw (Figure 2a).
In this case, we perform tests with a ground resistance of
1.0 kΩ. For lower values of R1 , the maximum power draw is higher
but rapidly decays; higher values of R1 yield gentler curves. The
middle scenario is where the initial power draw starts low and de-
cays fairly quickly around at the middle of Figure 2a in which the
light-blue zone curves inward. This region corresponds to R1 values
of approximately 0.30–0.35 kΩ.
Figure 3a presents the maximum power draw for different values
of circuit R1 and earth Rg resistors. The result shows that the maxi-
mum draw on the power source can be controlled by the choice of
R1 — regardless of the ground conditions. A resistor of >0.1 kΩ
should keep the power draw <1400 W, which is the typical peak
draw of a 700 W inverter. For Rg > 0.32 kΩ, raising the value
of R1 no longer has a significant effect. Figure 3b presents the
DOI:10.1190/geo2021-0058.1

median power draw for different values of R1 and Rg . These results

show that the median draw on the power source is entirely con-
trolled by the ground conditions and is independent of the choice
of R1 . These numerical experiments also show that very conductive
ground will overload a 700 W inverter if the electrodes are too close
together; however, the Rg resistance must be >0.02 kΩ, which is
not often encountered in the field (except in wet saline environ-
ments). The results presented in Figure 3 suggest three key findings:
The resistor R1 should ideally be in the 0.3–0.4 kΩ range; a 700 W
inverter will work in almost any geologic scenario (except for only
the most conductive scenarios such as wet saline environments
where ground resistances are measured of less than 0.02 kΩ); Figure 2. (a) Power draw within the modeled system for various R1
and a 400 W (or even a 300 W) inverter should suffice for inves- resistors. Note that low R1 values lead to strong transient behavior,
whereas high R1 values experience little to no transients. (b) Voltage
tigations in most dry or sandy environments. in the capacitor over time as modeled by Runga-Kutta (RK) and
Numerical modeling of the power required by the circuit proved Adams-Bashford (AB) numerical quadrature. Note that the voltage
to be a useful way to estimate all of the parameters needed for ef- asymptotes to the 120 V source voltage in the modeled scenario.
fective instrumental design. These results align with our field testing
observations in which the low-cost DC resistivity meter uses a 0.33
kΩ resistor due to common availability. In our
field experience, a 700 W inverter always was
sufficient in conductive environments, except
when using a R1 resistor rated for much less than
0.3 kΩ. This demonstrates that the circuit model
developed previously can predict the instrument
behavior to a reasonable and sufficient extent.
Overall, this approach proved useful for design-
ing DIY geophysical instruments and putting
upper (and lower) bounds on the required com-
ponents, which trims down or removes unneces-
sary or overengineered elements and avoids a
trial-and-error approach.
The numerical modeling allowed us to finalize
the various interchangeable components that con- Figure 3. (a) Maximum and (b) median power draws from the inverter for a variety of
tribute to the low-cost instrument design. The total R1 and Rg scenarios. The horizontal and vertical axes are plotted in log10 scale.
 WA6 Sirota et al.

A safety moment should always be a foremost consideration and it is essential to

follow established procedure and not cut corners in the design, test-
Although the DC-resistivity system presented in Figure 4 appears
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

ing, and validation phases (e.g., by using low-quality components

fairly innocuous, it is important to emphasize that handling substan- not designed or rated to handle the throughput wattage).
tial currents at 120 or 240 V is inherently dangerous and incorrect
use can lead to a serious or even lethal injury. We point out that a
major differentiator between commercial and low-cost designs is
INSTRUMENT FIELD TESTS
that the former has additional electrical safety features to reduce The first validation test was conducted at the CSM GDL on the
the risk of operator electrocution. Thus, like with all DIY hardware Kafadar Field (Krahenbuhl et al., 2018). The near-surface geology
projects involving live and powerful electrical sources, safety of the GDL area (Weimer, 1973) comprises thick shale units down
to below the depth of interest (i.e., below 50 m). Overlying these
units is a thin layer of sand and soil in the top 0.5 m with backfill to
Table 1. Cost by item for the components used to construct approximately 2 m depth. During the testing period, water from a
the low-cost DC resistivity system. recent rain was pooling at the surface leading to a saturated near
surface.
Total cost We tested the low-cost instrument to determine whether it would
Item Number ($) record consistent results when connected to different power sources,
given that the CP2011 instrument became overloaded when used at
Film capacitor (1 μF) 1 1.00 this site. We used various wattage inverters as well as generators of
Resistor (0.33 kΩ) 1 1.00 different output voltages while holding all other experimental var-
Bridge rectifier 1 3.00 iables constant, save for using the AB electrode spacings that varied
between 2 and 30 m. Figure 5 presents the test results for the differ-
DPDT switch 1 1.40
ent power scenarios and compares them to the results from a com-
Power cable 1 3.00 mercial-grade AGI Sting R1 system. Overall, the analysis from this
Banana jack binding posts (2 each color; $1.13 4 4.50 experiment leads to a few key observations: (1) The low-cost and
DOI:10.1190/geo2021-0058.1

per unit) commercial DC resistivity instruments measured consistent data;

Power inverter (500 W minimum) 1 85.00
(2) the choice of power source did not really matter, as long as
Digital multimeter (w/RS232 output; $35 per 2 70.00 the inverter had a sufficient wattage rating to push current through
unit)
Spare wire — 5.00 the ground at near and far electrode separations; and (3) lower
(i.e., an inverter) and higher (i.e., a generator) wattage sources per-
Miscellaneous connectors — 3.00
formed similarly. An implication is that a 120 or a 220 V AC gen-
Total — 176.90 erator power source should be able to generate consistent and
accurate DC-resistivity results.
Prices are approximate and are denoted in 2021 USD. Wires, electrodes, and DC
battery power are not included below because these are required components of We conducted a second experiment at the CSM GDL to test how
any DC resistivity system. Note that there is the potential to reduce the total build wide the electrodes can be spaced and still measure consistent data
cost by finding equivalent components priced lower than those presented here. because this metric indirectly corresponds to how deep the resistivity

Figure 5. Power-source tests for the low-cost DC resistivity meter

Figure 4. Field setup of the low-cost DC resistivity system. The with reference results obtained with a Sting R1 system. Note the
current injection box (clear plastic) is plugged into the red power very good consistency between the low-cost and reference measure-
inverter (connected to a standard 12 V car battery) and is attached to ments at the range of AB separations tested in this experiment; how-
the multimeter being used as an ammeter. The other multimeter, ever, the 140 W inverter does not provide enough power at
used as a voltmeter, attaches directly to the MN electrodes to the separations >50 m (note: The 500 W inverter used below success-
left and right out of the field of view. The AB electrodes (not fully resolved these issues). The values begin the diverge before the
shown) are located beyond the MN electrodes. system fails at 55 m.
 Low-cost DC resistivity meter WA7

meter can profile the subsurface geology. Tests were completed using measurements were acquired with each DC resistivity system to
a 700 W inverter power source. Figure 6a presents results for AB evaluate the accuracy and precision of the low-cost instrument.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

separations ranging from 2.5 to 90.0 m acquired at 2.5 m increments. Table 2 presents the sandbox test results. The apparent resistivity
To highlight data similarities and differences, we present the mea- values obtained with the low-cost and Sting R1 instruments were
sured apparent resistivity values on the y-axis in log10 scale. Overall, 1.042 and 1.035 Ωm, respectively. The 0.68% relative difference
the two data sets are quite consistent at the full range of AB electrode likely corresponds to noise in the final significant digit of the multi-
separations. To better highlight the consistency, Figure 6b shows a meter. The low-cost DC resistivity system, though, had higher stan-
crossplot of the same two data sets color coded by AB electrode spac- dard deviation values compared to the Sting R1, which could be
ing. The horizontal and vertical axes show apparent resistivity units in attributed to multimeter significant digit measurement limitations.
log10 scale as measured by the commercial and low-cost instrument, This observation suggests that the low-cost system achieves reason-
respectively. Most of data points fall on the 1:1 line, with only a few able accuracy but at a lower precision than the commercial unit;
outliers visible that correlate with the rapid variations observed in however, in our experience this has not substantially affected the
Figure 5. This experiment illustrates that the low-cost DC-resistivity resulting interpretations. Although this laboratory test confirms that
system performs accurately at this site to 90.0 m AB electrode spac- the low-cost system measures similar apparent resistivity values to a
ing, corresponding to sounding a fairly shallow aquifer (20–30 m) commercial instrument, further testing with different arrays and
and acquires data that are overall comparable with those measured electrode separations is needed to more fully understand its overall
on a commercial-grade system with the degree of interunit variation accuracy and precision.
consistent with a benchmark study of commercial-grade DC-resistiv-
ity systems (Parsekian et al., 2017).

Automated data logging

The increasing availability of CPU microcontrollers and open-
source software suggests that microcontrollers, such as the Rπ,
could be used to create a low-cost robust data logger that simplifies
DOI:10.1190/geo2021-0058.1

overall system use, automates data collection, and could lead to

long-term monitoring stations. Thus, a significant component of
our research was devoted to developing, testing, and validating
an Rπ data logger that can be attached to the low-cost DC-resistivity
instrument in a modular manner. Because the procedure is specific
to Tekpower TP4000ZC digital multimeters with Rs232 outputs, we
refer readers to Appendix B of Sirota (2020) for additional details
and specific build instructions beyond what we sketch out below.
Figure 7a presents a photograph of the low-cost system with the
autologger components attached. The design consists of the existing
low-cost system (the larger plastic box), two connected digital mul-
timeters with RS232 outputs, and the Rπ unit (the smaller plastic
box). Each multimeter outputs a 14 byte data burst every 250 ms in
a series of 112 zeros and ones that are read and decoded by a Python
I/O control script (Sirota, 2020). After decoding the data bursts, the
script averages and stores the four data measurements. Depending
on the multimeter settings, the values may represent either voltage
or current measurements. A basic data acquisition graphical user
interface (GUI) guides users through a series of inputs and prompts
and allows them to designate a survey type and to adjust the elec-
trode spacing. Data measurements are saved to a user-designated
file format suitable for the particular open-source software package
used for data interpretation. Finally, to ensure that the system is
field-operable, the design incorporates a small HDMI-compatible
screen and keypad/mouse for data input.
We initially tested the data-logger system independent of the re-
sistivity system using temperature probes connected to the multi-
meter. The probes were held in the air, on a table, in a hand, and Figure 6. (a) Comparison between the low-cost DC resistivity and
Sting R1 measurement results for AB electrode separations ranging
in hot water to simulate rapid value changes and to differentiate be- from 2.5 to 90.0 m in 2.5 m increments. The y-axis is presented in
tween the two multimeters. For all temperature tests, the data re- log10 scale to better show the similarities and differences in the data.
corded by the autologger matched the values displayed on the Overall, the two data sets are highly similar, with an rms error of
multimeters. The final data-logger tests were performed in a sand- 0.49%. (b) Scatterplot showing the alignment of the low-cost and
Sting R1 instruments measurements. The measurements — color
filled 90 × 46 × 42 cm3 terrarium (Figure 7b). Four electrodes were coded by the AB spacing — closely follow along the blue 1:1 line
placed centered in the tank 20 cm apart. Wenner array measurements except for four anomalous data points. The divergent points are not
were taken using a 10 mA input current for both systems. Three correlated with the spacing.
 WA8 Sirota et al.

As of March 2021, the total cost of the automated data logger This section summarizes the key water resource challenges faced
system is $108, which is broken down by component in Table 3. by residents of the Cotonou, Benin, region and presents the results
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

When added to the cost of the basic low-cost system found in Ta- of the field tests conducted at the UAC campus. We then discuss the
ble 1, the total system cost is $286. Appendix 3 of Sirota (2020) near-surface geology of the Ouédo Field site and present the data
presents a detailed component list and current links to where we and inversion results using a combination of low-cost and commer-
sourced the items. cial DC resistivity instrument sounding data. Because this paper
largely focuses on the development and validation of low-cost in-
strumentation, we refer readers interested in a detailed characteri-
BENIN FIELD PROJECT zation of the regional geology to Blivi et al. (2002), Barthel et al.
(2008, 2009), Boukari et al. (2008), and Adepelumi et al. (2009);
The low-cost instrument design and validation project was for a hydrologic characterization of the Cotonou region to Boukari
funded by a GWB grant, which represented a collaboration between et al. (2008), Silliman et al. (2010, 2011); and for a hydrogeophys-
personnel from CSM and the UAC located in greater Cotonou, Be- ical overview to Lindsay (2015) and Lindsay et al. (2015).
nin (Figure 8). We aimed to develop low-cost instruments suitable
for acquiring geophysical data relevant to hydrologic problems and Water resource challenges in the Cotonou region
to distribute the knowledge of how to build and use the developed
low-cost instrument. As part of the project work plan, UAC students Securing groundwater resources for coastal cities is and will con-
and faculty built, validated, and used the base instruments (i.e., with- tinue to be a challenge because of increasing extraction due to high
out the data logger) to acquire geophysical data at a local field site. population growth, degradation of existing resources through salt-
water intrusion, difficulties in locating new groundwater resources,
and uncertainty associated with aquifer recharge due to climate
Table 2. Field test results of the low-cost and Sting R1 DC change (Silliman et al., 2011). The city of Cotonou, located on
resistivity meters. the Atlantic coastal zone, is one of many regional examples facing
such challenges. The city draws significant groundwater resources
from the Godomey pumping field (Figure 8); however, the past two
Rπ ΔV avg (V) I avg (mA) ΔV/I (Ω) ρa (kΩm) σ (%)
DOI:10.1190/geo2021-0058.1

decades have seen a gradual increase in the salinity of the city’s

1 9.67 11.55 837.6 1.053 0.37 main wells due to saltwater intrusion (Silliman et al., 2010; Houn-
2 8.98 10.81 830.6 1.044 1.11 sinou, 2020). As illustrated in Figure 8, saltwater potentially in-
vades from the Atlantic Ocean, but it is more likely from Lake
3 8.95 10.93 818.6 1.029 3.84
Nokoué to the southeast. This very shallow lake alternates between
Mean 9.20 11.10 828.9 1.042 1.21 freshwater and saltwater depending on the season because it is
Sting R1 — I avg (mA) ΔV/I (Ω) ρa (kΩm) σ (%) hydrologically connected to the Atlantic Ocean (Boukari et al.,
2008). Although the lake is suspected to be one of the main sources
1 — 10 819.3 1.030 0.4 of aquifer recharge (Boukari et al., 2008), the hydrogeology of the
2 — 5 825.4 1.037 0.0 aquifer is not well characterized from a hydrogeophysical perspec-
3 — 5 825.3 1.037 0.0 tive (Lindsay et al., 2015). In addition, surface recharge has been
Mean — 6.67 823.3 1.035 0.4 significantly reduced due to urbanization and the resulting preven-
tion of precipitation from reaching more porous soil (Lind-
say, 2015).
In response to the saltwater intrusion issues in
the Godomey pumping field, the local govern-
ment embarked on a significant capital invest-
ment project to develop the Ouédo pumping
field located approximately 10 km northwest
of the existing well field. The main rationale
for this development was to increase the distance
of the local aquifer from, and remove the hydro-
logic connectivity to, potential sources of salt-
water intrusion. To help the local government
better understand the hydrogeology of the Ouédo
pumping field, members of the UAC Institut Na-
tional de l’EAU have been undertaking hydro-
geophysical investigations. One of the key
longer term goals of the GWB CSM-UAC
project is to use low-cost DC-resistivity instru-
Figure 7. (a) The low-cost DC resistivity system with the basic control box (the larger ments to better characterize the local ground-
clear plastic box), two multimeters, 12 V battery, and inverter. The data-logger compo- water resources and to establish DC-resistivity
nents include the Rπ unit (the smaller clear plastic box), touch screen, and Bluetooth
keyboard with touch pad. (b) Testing of the autologger system in a sand-filled 90 × 46 × stations that could be used for time-lapse inves-
3
42 cm glass aquarium. The four AB and MN electrodes are deployed in a Wenner tigations aimed at monitoring the long-term aqui-
configuration. fer “health.” The efforts made during the first of
 Low-cost DC resistivity meter WA9

two CSM-UAC GWB project seasons in Benin represent the first We used several different constant spread traverses at various
steps in this direction. spacings to acquire additional data on how well our instruments
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

A key component of the CSM-UAC GWB project involved the worked. Figure 9 presents the data comparisons for the low-cost
construction and use of the low-cost DC-resistivity instruments by and Sting R1 DC resistivity instruments. As expected from the
UAC and CSM students. The teams constructed two low-cost DC- CSM field tests, the data from each instrument were consistent, fall-
resistivity instruments, which were tested on the UAC campus ing on average within 1% of each other with a 6% maximum differ-
alongside Sting R1 and IRIS Syscal Kid DC-resistivity meters. ence at one station (Figure 9, datum 19).
To test the meters, we went to a nearby field and acquired readings
for electrodes deployed in a Wenner array at various separations. Ouédo Field site
The Ouédo pumping field site area is located in the southeast
Table 3. Cost by item for the data-logger module that can be portion in the Plateau of Allada in Benin’s coastal sedimentary ba-
added onto the base DC resistivity system. sin. The three main lithologic units that make up the sedimentary
geology of the top 200 m are a clayey sand, a sandy clay, and vari-
Item Number Total cost ($) ous sand units, which sit uncomformably over a hard, highly hetero-
geneous marly basement that represents the lower limit of our
Rπ 4 1 35.00 investigation (Lindsay, 2015). The interbedded units are thought
Micro SD memory card 1 7.50 to form four regional aquifer layers: an unconfined surficial layer
Small HDMI-compatible monitor 1 53.00 overlying three hydraulically connected confined aquifer layers.
Number pad/keyboard 1 13.00 Aquifer separation is inferred from the presence of clay aquitards
encountered in boreholes throughout the region. Clay is typically
Autologger subtotal 1 108.50 regarded as impermeable given its small grain size, but it can form
localized discontinuous lenses that serve to connect the sublayers of
Base low-cost system subtotal 1 176.90
the confined aquifer (Silliman et al., 2010).
Instrument + autologger 1 285.40 Hydrogeophysical investigations were conducted along a road
DOI:10.1190/geo2021-0058.1

with irregularly spaced wells from the Ouédo pumping field (Fig-
Prices are approximate and are denoted in US currency. The autologging system
assumes that the two multimeters with RS232-output capability are used in the ure 10a and 10b). This meant that we worked along a cross section
base low-cost instrument design. The bold values are the subtotal or total costs. of a broad valley as illustrated by the topographic profile presented
in the lower half of Figure 10b. The 30 m deep valley spans a dis-
tance of approximately 4.0 km. We have information about the val-
ley geology because lithologic logs were developed during drilling,
which were interpolated by UAC hydrogeologists into the cross sec-
tion shown in Figure 10a. Because the wells are on average 0.5 km
apart, subsurface structures were interpolated between wells. The
fieldwork surveying goal was to use the low-cost instruments to
tie in the known regional geology and infill knowledge gaps about
the hydrostratigraphy. This is important because knowing the con-
nectivity of aquifers allows development of improved hydrologic
models of saltwater flow beneath the surface and how this poten-
tially affects current and future wells.

Figure 8. Regional map of the greater Cotonou area illustrating the

saltwater intrusion issues potentially originating from the Atlantic
Ocean to the south and Lake Nokoué to the east. The town of God-
omey and the UAC campus are near the center of the figure. The
approximate well sites of the Godomey and Ouédo pumping fields
are illustrated in cyan and blue, respectively. The brown “V” to the
south of the Ouédo pumping field shows the locations of seasonal
rivers that end up feeding into Lake Nokoué. These also represent
topographic features, with the left branch crossing through the
project survey area. Lighter brown areas abutting the Atlantic Ocean
are combined urban areas and coastal mangrove forest. The channel
connecting Lake Nokoué to the Atlantic Ocean is drawn in lighter Figure 9. Results from the UAC campus comparison test between
blue. The red arrows indicate suspected pathways for saltwater the low-cost and Sting R1 DC resistivity meters. Measurements
intrusion into the Godomey pumping field. were taken in a line using a Wenner array with 5 m spacing.
 WA10 Sirota et al.

To constrain the hydrostratigraphy of the Ouédo well field, we resistances) because the different data sets were acquired by
acquired VESs using the low-cost instruments and an IRIS Syscal different crews on different field days.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Kid. Over a four day period, teams of students measured 19 Wenner We expected the data inversions to show distinct zones represent-
arrays with electrode spacings ranging from 15 to 120 m (i.e., a ing the topsoil, an initial clay layer, and underlying sand units as
360 m maximum AB spacing) providing VESs to depths of approx- presented in the cross section in Figure 10a. The sand zone was
imately 80 m. To tie in with the known well-log lithology, the VESs predicted to be interrupted by more conductive and resistive zones
were centered at each well location between F5 and F12 (see Fig- where the clay content and water saturation varied in the rock. In-
ure 10b), with additional VESs acquired between wells to better versions were completed using the RES2DINV (Loke, 2002) com-
constrain the geologic interpretation. Some VESs were measured mercial software package with the topography data and electrode
with only the low cost or the IRIS instrument for survey efficiency. positions calculated from the VES center points. We split the VESs
However, we acquired five collocated VESs using both instruments recorded at and between wells F10 and F12 by the low-cost and
to investigate whether the data were sufficiently close so as to be IRIS instruments to form two different data sets (data acquired
used interchangeably for inversion and interpretation. Figure 11 by a single instrument at a station were included in both data sets).
presents VES data from three collocated well locations. Overall, We then inverted these two data sets independently to test whether
there are larger differences between the measurements than those the results and subsequent interpretations varied significantly. The
observed during the UAC or CSM tests; however, the overall trends two inversion results (Figure 12) are similar with only slight
remain largely the same. Note that there is some error associated differences noted to the northwest near the collocated F10–F12
with electrode station repeatability (including differences in contact wells; the overall interpretation of the inversion results remains un-
changed.
The inversion of the DC-resistivity data (Fig-
ure 12c) shows several deep zones of high con-
ductivity, likely related to areas of higher water
saturation. Resistive zones at the ends of the val-
ley are separated by a conductive zone at its base
that extends to the surface. This is likely due to a
DOI:10.1190/geo2021-0058.1

combination of upwelling from the pumping

wells and surface infiltration from an intermit-
tently flowing seasonal stream. At the boundary
marked in Figure 12c, the resistivity values shift
from more resistive to more conductive, which is
at the inferred depth of the clay-sand interface in
the geologic cross section (Figure 10a). This sug-
gests that subsurface aquifer layers are neither
confined nor isolated; however, further investiga-
tion is needed to validate this interpretation,
e.g., through joint inversion using data from an-
other geophysical method (e.g., seismic).

DISCUSSION
The overall cost of the validated low-cost DC
resistivity instrument was $177, whereas the auto-
mated data-logger system cost a further $108,
bringing the total to $285. To make a complete
system, though, one also needs to purchase two
additional relatively costly components. First,
users would need to buy (or otherwise source)
a 12 V car battery, the price and availability of
which vary regionally. In Benin, the cost of a
new 12 V battery was a relatively expensive
$80. One could also use cheaper motorcycle bat-
teries, which were ubiquitous throughout Benin;
Figure 10. (a) Hydrogeologic cross section (Kotchoni, 2013) through the northwest– however, they likely would not last for a full
southeast 2D transect of hydrogeologic interest constructed using lithographic logs from
10 boreholes. The near-surface geology consists of a humic top soil underlain by a 10– day of surveying. Second, one would need to pur-
25 m thick clay layer. The clay layer is underlain by units ranging from coarse to fine chase four electrical wires/cables, one for each
sands. Several wells intersect supposed lenses of sandy clay and clayey sand. The mea- electrode. Because this is necessary for a commer-
sured static piezometric surface level is shown by the inverted triangles, whereas the cial instrument as well and is not usually included
inferred surface is shown by the dashed line. (b) Well spacing (upper) and the corre-
sponding valley topographic profile (lower). The red, blue, and green dots, respectively, in its purchase, we excluded it from the cost es-
indicate VESs acquired by only the Syscal Kid system, only the low-cost system, and by timations because the associated prices will vary
both systems on different days with different crews. from place to place and the overall required
 Low-cost DC resistivity meter WA11

lengths depend on the survey design. Based on our reconnaissance in current injection problems (i.e., the ammeter reads 0 A) or if there
Benin (and Peru), the price of electronics parts is fairly stable around is no discernible voltage difference reading on the voltmeter.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

the world and may add or subtract only a few dollars to the total cost The Rπ microcomputer is simple to use and does not require
because individual parts are fairly inexpensive. Thus, with the special equipment besides the recommended power cable. The rest
inverter, the multimeters, and all the electronic parts, we find that of the electronic components are easy to obtain and replace, which
$150–$200 is a reasonable cost range for the basic manual box. How- makes the instrument not overly complicated to build (e.g., a half-day
ever, components found online (especially the inverter) may further teaching students in a multilingual environment in Benin). Although
reduce costs. most components are easy to obtain, RS232 multimeters are unfash-
When attempting to lower project costs, one must be careful to ionable and can be difficult to find. Should a different model of multi-
not reduce the functionality of the final product just because it is to meter be used, it is likely that the data stream would have a different
be used by less-experienced operators. During the field work pro- format and require another parser from that presented in Sirota
gram, a participant accidentally crossed wires and blew an inverter
fuse. Although a geophysicist experienced in using the low-cost
DC-resistivity system would recognize what had happened and
know how to fix the problem, an inexperienced operator may
not, which could shut down the surveying until a repair is com-
pleted. Training personnel to build and safely operate DIY instru-
ments will not necessarily equip them with all of the knowledge
needed to fix and improve them. This point illustrates that special-
ized expertise is still needed in most hydrogeophysical surveying
contexts.
Adding in the Rπ-based data logger increases costs by $108.
Most of this overhead comes from the Rπ unit, a display and key-
board, and the markup of the RS232-enabled multimeters. An in-
DOI:10.1190/geo2021-0058.1

triguing idea would be to create an application for mobile phones

that could be installed and used through the Rπ Bluetooth to operate
the system. This would remove $66 from the design and lead to a
sub-$43 data-logger unit. Although these cost reductions would
make the low-cost design relatively more affordable, it is still im-
portant to consider the absolute affordability in terms of, for exam-
ple, the $1217 average nominal per capita earnings ($3446 at
purchasing parity power) in Benin (IMF, 2019). In this context,
building a low-cost system likely remains a significant financial
undertaking and is by no means a panacea for overcoming all
groundwater exploration challenges.
The Ouédo Field tests demonstrated that the data acquired on the
low-cost and commercial-grade instruments are interchangeable at
this site. However, as revealed by the CSM and UAC field tests,
the low-cost instruments are more susceptible to interference from
environmental noise such as power lines and other local electrical
infrastructure. Another example of environmental noise is natural
electric currents in the ground that can cause data drift and odd
values that may affect the low-cost instruments to a greater degree
than their commercial counterparts, which are usually designed
to take these effects into account. However, this problem can
be partially mitigated by modifying the field procedure to
measure the self-potential voltage before and after sampling and
subtracting off the mean value. Overall, that the influence of envi-
ronmental noise cannot be completely removed with low-cost in-
struments remains an advantage of commercial instruments.
Despite this limitation, the suspect data points can be removed
through careful preprocessing without adversely affecting the final
interpretations.
An example of how environmental noise affected our data can be
found in the F12 VES (Figure 11c). In this case, several data points
show obvious errors, whereas the majority align well with the IRIS
data. Because the low-cost instrument is not designed to give clear Figure 11. The VES comparison tests between the IRIS and low-
and coherent error messages, users should be trained to recognize cost DC resistivity instrument data for the (a) F10, (b) F11, and
abnormal readings. For example, users need to learn to diagnose (c) F12 wells (the locations shown in Figure 10b).
 WA12 Sirota et al.

(2020). The increasing availability of Bluetooth-enabled systems, and determine whether the values make sense. In addition, the user
though, could make these a good choice because they can be paired can set parameters such as the electrode spacing and survey type,
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

with the native Rπ capabilities and lead to remote data collection at which makes keeping track of the survey much easier. With the Rπ
distances up to the standard Bluetooth limit of 100 m. data-logger system, the file format is stored in a consistent way and
Component availability may be an issue in various parts of the can easily be modified within a Python script. Using the data logger
world. Ordering online requires internet access and companies that is not necessarily that much faster in the field than surveying with-
facilitate international transactions in often remote locations. At this out it and does make the system more bulky and complex. However,
point, it is unclear how one could ship an Rπ unit to many countries. a data-logging system is a valuable starting point for further auto-
In addition, some of the electronics (e.g., the bridge rectifier) might mation of simple, low-cost ER instruments. As always, adding fur-
be difficult to find, although a capacitor and resistor can be found by ther clarifying warning and error messages into the current software
cannibalizing other electronic devices. It is advisable to carry spares system would be judicious.
for the limited number of system components and to shield the sys- An interesting future extension to the basic DC-resistivity instru-
tem from environmental factors (e.g., moisture, excessive heat) as ment would be to create a manual switch box to allow for quick and
well as possible whenever conducting field surveys. The inverter is easy switching between more than four electrodes, thereby allowing
the most expensive and difficult part to find in the basic low-cost users to conduct 2D and 3D ER tomography surveys using multiple
DC resistivity meter. electrode channels. A switch box would allow a user to select which
The instrument should be tested before use to validate the con- of many input cables should be wired as A, B, M, and N using rotary
struction and ensure data reliability. In cases in which one does not switches. This addition would increase the cost of cables even if not
have a calibrated instrument to use as a control, the resistivity meter overly increasing the cost of the instrument itself (e.g., eight electro-
may be tested across several large resistors. To do this, the instru- des and cables are double the cost of four), but it would represent a
ment should be set up to “inject” current across a resistor of known good intermediate step toward true multichannel capability.
resistance. One can then measure the voltage across the resistor, Another extension would be to incorporate relays (electronic
calculate its resistance, and compare the result to the known rating. switches) into the design. Using relays would mean blending the ba-
The addition of the automated data logger allows users to calcu- sic box with the Rπ system by removing the DPDT switch and hav-
DOI:10.1190/geo2021-0058.1

late and monitor apparent resistivity measurements during a survey ing that function be controlled by the Rπ unit so that a measurement
can be automatically acquired and recorded. Add-
ing this functionality would allow the low-cost
system to replicate the field procedure used with
the most basic commercial instruments. In addi-
tion, electronic switches could replace the rotary
switches in the proposed multichannel switch box
so that the survey could be run automatically with-
out manually turning the current on/off or moving
electrodes. Relays are inexpensive electronic com-
ponents and would not substantially increase the
total cost of the instrument, although its overall
complexity would grow.
The Rπ system could be improved by config-
uring the system to output the data into file for-
mats that can easily be used with various free or
common resistivity inversion programs, includ-
ing RES2DINV (Loke, 2002), SimPEG, and
IPI2WIN (Bobachev, 2002). We aim to improve
the documentation to include detailed build in-
structions and a user guide. This documentation
along with the Python scripts will be publicly
available so that the data logger can be used
along with the DC resistivity system as a low-
cost option for surveying anywhere in the world.
Finally, the first field season of the CSM-UAC
GWB project focused on using the “training the
trainer” model through a two week hydrogeophys-
Figure 12. Inversions of the combined VES data set using RES2DINV software. (a and
b) Low-cost instrument and IRIS data at colocated survey points, respectively, are used. ical workshop focusing on the theory, methods,
The section extends from before well F5 (right/southeast) to well F12 (left/northwest). instrumentation, field practice, and interpretation
Sounding locations are labeled with the black arrows. (c) Interpreted results showing of geophysical data helpful for locating ground-
several deep zones of high conductivity likely related to higher saturation zones. At water resources. Based on our collective experien-
the valley floor and near well F6, there are two large conductive zones, outlined in pur-
ple. There are two resistive zones, circled in orange, at each end of the valley. A black ces and “lessons learned during year 1, we are
line across the entire section highlights an inferred boundary between more resistive and aiming to hold a larger two week hydrogeophysics
more conductive earth, likely due to a clay-sand boundary. workshop at UAC involving faculty, students, and
 Low-cost DC resistivity meter WA13

professionals from universities and agencies throughout francophone Awotoye, K., and A. Selemo, 2006, Design and construction of a resistivity
meter for shallow investigation: Nigerian Journal of Physics, 18, 261–270,
West Africa. The goals of the expanded workshop are to train and doi: 10.4314/njphy.v18i2.38113.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

equip a network of young geoscientists. This will better prepare them Barthel, R., J. Jagelke, J. Gotzinger, T. Gaiser, and A. Printz, 2008, Aspects
to undertake hydrogeophysical surveys and address groundwater of choosing appropriate concepts for modelling groundwater resources in
regional integrated water resources management: Examples from the
challenges in their home countries. Neckar (Germany) and Ouémé catchment (Benin): Physics and Chemistry
of the Earth, Parts A/B/C, 33, 92–114, doi: 10.1016/j.pce.2007.04.013.
Barthel, R., B. Sonneveld, J. Gotzinger, M. Keyzer, S. Pande, A. Printz, and
CONCLUSION T. Gaiser, 2009, Integrated assessment of groundwater resources in the
Ouéme Basin, Benin, West Africa: Physics and Chemistry of the Earth,
Parts A/B/C, 34, 236–250, doi: 10.1016/j.pce.2008.04.001.
In contrast to the high expense of the commercial-grade equip- Blivi, A., E. Anthony, and L. Oyédé, 2002, Sand barrier development in the
ment, a basic DC resistivity system with sufficient precision and Bight of Benin, West Africa: Ocean & Coastal Management, 45, 185–
accuracy for many-to-most hydrogeophysical surveying tasks can 200, doi: 10.1016/S0964-5691(02)00054-6.
Bobachev, C., 2002, IPI2Win: A windows software for an automatic inter-
be constructed for less than $200. In addition, a data autologger pretation of resistivity sounding data: Moscow State University.
device constructed from Rπ microcontroller technology can be gen- Boukari, M., P. Viane, and F. Azonsi, 2008, Three-dimensional modeling of
a coastal sedimentary basin of southern Benin (West Africa), in S. Ade-
erated for less than $110. Numerical modeling and physical tests at lana and A. MacDonald, eds., Applied ground water studies in Africa:
the CSM and at various field sites in Benin demonstrated that the CRC Press, IAH Selected Papers on Hydrogeology 13, 437–456.
system is more robust than previously published low-cost designs Cavalcante Messias, G., J. Agnelo Soares, F. Kipper, I. Fernandes Gomes, V.
Soares Júnior, and J. Alves da Silva, 2019, Hydrogeophysics in fractured
and can work in a more diverse range of geologic environments — crystalline aquifers in English Guyana: Brazilian Journal of Geophysics,
especially the more conductive scenarios commonly encountered in 37, 299–308, doi: 10.22564/rbgf.v37i3.2009.
groundwater exploration. Clark, J., and R. Page, 2011, Inexpensive geophysical instruments support-
ing groundwater exploration in developing nations: Journal of Water Re-
Inexpensive DC-resistivity systems such as the one developed source and Protection, 3, 768–780, doi: 10.4236/jwarp.2011.310087.
during our work can greatly improve the success rates of finding Clark, J., R. Page, R. Franklin, N. Miller, and M. Morken, 2016, Appropriate
geophysics technology: Inexpensive instruments for water exploration at a
safe drinking water as well as provide information about the aquifer local level in developing nations, in G. R. Wessel and J. K. Greenberg,
structure when used by trained and well-equipped technicians. The eds., Geoscience for the public good and global development: Toward a
developed DC resistivity systems may be used to monitor existing sustainable future: Geological Society of America, 171–181.
Fatahillah, D., D. Darsono, and N. Nuryani, 2019, Low-cost multi electrode
groundwater sources and to help characterize zones of saline water
DOI:10.1190/geo2021-0058.1

resistivity meter based on microcontroller for electric resistivity tomog-

encroachment thereby assisting with groundwater management. raphy purpose: Journal of Physics: Conference Series, 1153, 012022,
Finally, low-cost systems require little training to use, can be de- doi: 10.1088/1742-6596/1153/1/012022.
Goldman, M., and F. Neubauer, 1994, Groundwater exploration using inte-
ployed for longer periods, and are easier to repair or replace using grated geophysical techniques: Surveys in Geophysics, 15, 331–361, doi:
parts sourced in developed or developing parts of the world. Ulti- 10.1007/BF00665814.
Gupta, S., J. Bayless, and B. Peikari, 1972, Circuit analysis: Intext Educa-
mately, our assertion (and hope) is that they will contribute to a tional Publishers.
long-term sustainable solution in the search for safe drinking water Hounsinou, S., 2020, Assessment of potential seawater intrusion in a coastal
with a meaningful impact on under-resourced communities in need. aquifer system at Abomey-Calavi, Benin: Heliyon, 6, e03173, doi: 10
.1016/j.heliyon.2020.e03173.
IMF, 2019, World economic outlook database, October 2019, International
Monetary Fund, http://web.archive.org/web/20080207010024/http://www
ACKNOWLEDGMENTS .808multimedia.com/winnt/kernel.htm, accessed 17 November 2019.
Kotchoni, V., 2013, Drilling performance evaluation of the Ouédo catchment
We thank GWB and their generous donors for the grant that made field and the hydrodynamic parameters of the captured aquifer: Interest
for a sustainable development of the local groundwater resource: Master’s
this work possible. We acknowledge CSM for support of the thesis, Abomey-Calavi University.
humanitarian geophysics program, including partial funding for Krahenbuhl, R., B. Passerella, H. Flamme, G. Crookston, and D. Sirota,
the M.S. thesis work of D. Sirota. We thank UAC for institutional 2018, Developing a large underground geophysical education laboratory
at Colorado School of Mines: 88th Annual International Meeting, SEG,
contributions to the field project. Finally, we thank GEOPHYSICS as- Expanded Abstracts, 2687–2691, doi: 10.1190/segam2018-2998344.1.
sistant editor C. Farquharson, associate editor D. Nobes, J. Clark, Lindsay, K., 2015, Seismic imaging to constrain groundwater models for a bet-
and three anonymous reviewers for comments and suggestions that ter understanding and management of water quality in Coastal Benin, West
Africa: A saltwater intrusion problem: M.S. thesis, Boise State University.
improved the quality of the paper. Lindsay, K., J. Bradford, S. Silliman, N. Yalo, and M. Boukari, 2015, Seis-
mic imaging to help understand and manage water quality in Coastal Bé-
nin, West Africa: Geophysics, 80, no. 3, WB35–WB41, doi: 10.1190/
DATA AND MATERIALS AVAILABILITY geo2014-0115.1.
Loke, M., 2002, Res2dinv ver. 3.54: Rapid 2-D resistivity and IP inversion
using the least square method: Geotomo Software.
Data associated with this research are available and can be ob- Martinelli, E., 1978, Groundwater exploration by geoelectrical methods in
tained by contacting the corresponding author. Southern Africa: Bulletin of the Association of Engineering Geologists,
15, 113–124, doi: 10.2113/gseegeosci.xv.1.113.
Olowofela, J., V. Jolaosho, and B. Badmus, 2005, Measuring the electrical
resistivity of the earth using a fabricated resistivity meter: European Journal
REFERENCES of Physics, 26, 501–515, doi: 10.1088/0143-0807/26/3/015.
Parsekian, A., N. Claes, K. Singha, B. Minsley, B. Carr, E. Voytek, R. Har-
Adepelumi, A., B. Ako, T. Ajayi, O. Afolabi, and E. Omotoso, 2009, Delin- mon, A. Kass, A. Carey, D. Thayer, and B. Flinchum, 2017, Comparing
eation of saltwater intrusion into the freshwater aquifer of Lekki Penin- measurement response and inverted results of electrical resistivity tomog-
sula, Lagos, Nigeria: Environmental Geology, 56, 927–933, doi: 10.1007/ raphy instruments: Journal of Environmental and Engineering Geophys-
s00254-008-1194-3. ics, 22, 249–266, doi: 10.2113/JEEG22.3.249.
Arétouyap, Z., D. Bisso, N. Nouck, L. P. Amougou Menkpa, and J. E. As- Pham, N., D. Boyer, T. Nguyen, and G. Nguyen, 1994, Deep ground-water
fahani, 2019, Hydrogeophysical characteristics of Pan-African aquifer investigation by combined VES/MTS methods near Ho Chi Minh City,
specified through an alternative approach based on the interpretation Vietnam: Groundwater, 32, 675–682, doi: 10.1111/j.1745-6584.1994
of vertical electrical sounding data in the Adamawa Region, Central .tb00904.x.
Africa: Natural Resources Research, 28, 63–77, doi: 10.1007/s11053- Schlumberger, C., 1920, Étude sur la prospection électrique du sous-sol:
018-9373-8. Gauthier-Villar et Cie. Ch. VIII, 70–72.
 WA14 Sirota et al.

Silliman, S., B. Borum, M. Boukari, N. Yalo, S. Orou-Pete, D. McInnis, C. UNDP, 2019, Beyond income, beyond averages, beyond today: Inequalities
Fertenbaugh, and A. Mullen, 2010, Issues of sustainability of coastal in human development in the 21st century: United Nations Development
groundwater resources: Benin, West Africa: Sustainability, 2, 2652– Program Human Development Report 2019.
Downloaded 11/17/21 to 138.67.168.140. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

2675, doi: 10.3390/su2082652. UNEP, 2002, Water availability in Africa: UNEP and GRID-Arendal.
Silliman, S., M. Boukari, L. Lougbegnon, and F. Azonsi, 2011, Overview of UNESCO, 2012, World water assessment programme report 4: United Na-
a multifaceted research program in Bénin, West Africa: An International tions.
Year of Planet Earth groundwater project, 175–186. Vouillamoz, J., B. Chatenoux, F. Mathieu, J. Baltassat, and A. Legchenko,
Sirota, D., 2020, Development and validation of a low-cost DC resistivity 2007, Efficiency of joint use of MRS and VES to characterize coastal
meter for humanitarian geophysics applications: Master’s thesis, Colo- aquifer in Myanmar: Journal of Applied Geophysics, 61, 142–154,
rado School of Mines. doi: 10.1016/j.jappgeo.2006.06.003.
Süli, E., and D. Mayers, 2003, An introduction to numerical analysis: Cam- Weimer, R., 1973, A guide to uppermost Cretaceous stratigraphy, central
bridge University Press. Front Range, Colorado: Deltaic sedimentation, growth faulting and early
Toll, D., and A. Hassan, 2014, Data acquisition and control software for Laramide crustal movement: The Mountain Geologist.
automated resistivity measurements: Presented at the Information Tech-
nology in Geo-Engineering.
Toll, D. G., and A. Hassan, 2012, Development of automated multi-electrode
resistivity system for laboratory measurements: Istanbul 2012-
International Geophysical Conference and Oil & Gas Exhibition, 1–4. Biographies and photographs of the authors are not available.
DOI:10.1190/geo2021-0058.1

View publication stats