The PLOS One study examined whether one pseudo-dipole survey could reduce the need to collect several standard electrode-array datasets separately. The analysis combined analytical solutions, finite-element forward modeling, inversion and controlled noise experiments. It was not a field trial, so the results describe performance under modeled homogeneous, vertical-contact, multi-anomaly and four-zone resistivity conditions rather than performance at a hydrogeological site.

Forward and inverse calculations were performed with GeoElectroLab, a finite-element program developed by the research team. The inversion module used a smoothness-constrained Gauss–Newton algorithm, an unstructured mesh and the PARDISO parallel solver. Numerical tests varied electrode numbers and spacings, background and target resistivities, random noise, anomaly depth and inversion settings. Monte Carlo simulations used 100 realizations to examine filtering behavior across homogeneous half-space, vertical-contact and multi-anomaly models under noise levels from 1% to 30%.

The pseudo-dipole configuration fixes a reference current electrode and a reference potential electrode at opposite ends of the survey line. A mobile current-potential pair is then moved along the line in an inverted-trapezoidal measurement pattern, with one additional series obtained by interchanging the mobile electrodes. The study used this arrangement to generate apparent-resistivity pseudosections and inverted models, and compared them with a conventional dipole array and with Wenner α, β and γ arrays.

Under a homogeneous half-space model with 128 electrodes at 2.5 m spacing, a background resistivity of 100 Ω·m and 10% random noise, the average relative error of the apparent-resistivity analytical solutions was 2.5% for both the dipole and pseudo-dipole arrays. The dipole array produced a cumulative error approximately half that of the pseudo-dipole array, but the pseudo-dipole dataset was about twice as large. In a separate vertical-contact model, both arrays closely followed the theoretical response, with slightly lower average relative errors for the pseudo-dipole configuration, although the numerical and analytical discrepancy was larger for that configuration.

Resolution was evaluated with a model containing 11 low-resistivity anomalies, each 10 m by 10 m in a 1,000 Ω·m background. In the deeper layer with anomaly tops at 25 m, the pseudo-dipole inversion had an average full width at half maximum of 14.3 m, compared with 22.7 m for the dipole array; the true anomaly width was 10 m. This corresponded to a reported 37% improvement in lateral resolution. Semi-quantitative anomaly-recognition scores declined with depth for both configurations, from 0.85 to 0.41 for the pseudo-dipole array and from 0.83 to 0.23 for the dipole array as targets were moved from 5 m to 35 m depth. At 25 m depth, the dipole inversion clearly resolved four of seven anomalies and also produced false anomalies, whereas the pseudo-dipole inversion retained central anomalies with greater shape and value accuracy.

The transformation method reconstructs a target array by selecting the relevant dipole measurements, applying the required linear-combination coefficients and calculating apparent resistivity with the target array’s geometric factor. Reciprocal pseudo-dipole measurements were used for quality control because their values are not expected to be equal. A relative-difference threshold of 0.2 retained more than 90% of reliable data and excluded more than 95% of severely contaminated measurements across the tested model and noise conditions. In a multi-anomaly model with 10% noise, tightening the threshold from 0.2 to 0.15 changed normalized root-mean-square error by less than 5%, whereas relaxing it to 0.25 increased the error by more than 15%. The authors cautioned that site-specific noise could still affect the appropriate threshold.

Transformation accuracy was tested with a four-zone model containing resistivities of 10, 100, 200 and 1,000 Ω·m. With 5% random noise, relative differences between standard and transformed Wenner datasets were 2.5% for Wenner α, 3.1% for β and 1.8% for γ. Direct and transformed inversions produced generally consistent spatial patterns, although the pseudo-dipole inversion was closest to the true model. Its final root-mean-square error was 1.04, compared with 1.95 for standard Wenner α, 2.32 for standard β and 5.44 for standard γ. The corresponding transformed-array errors were 2.29, 2.02 and 6.10, respectively.

The study also estimated acquisition requirements rather than recording field time. For a 64-electrode survey with 61 pseudosection layers, the estimated number of current injections was 1,953 for the Wenner dataset and 496 for a four-channel pseudo-dipole configuration. Transformation amplified errors by approximately 2.0 to 2.2 for noise levels up to 20%, and the linear relationship was not expected to hold above 30% noise. The analysis therefore supports the numerical feasibility and modeled resolution of the approach, but the authors emphasized that field validation across diverse geological settings is still required. The current implementation also assumes two-dimensional subsurface structure, and secure placement of two reference electrodes at the line ends may be impractical in some environments.

Subject of Research: Numerical evaluation of a pseudo-dipole high-density DC resistivity array and its transformation into standard Wenner array data

Article Title: A comparative study of pseudo-dipole arrays and standard arrays based on the high-density DC resistivity method

Article References: Jiang, X., Jiang, S., Chen, H., Gong, J., & Liu, S. (2026). A comparative study of pseudo-dipole arrays and standard arrays based on the high-density DC resistivity method. PLOS One, 21(10), e0360161. https://doi.org/10.1371/journal.pone.0360161

Image Credits: AI Generated

DOI: 10.1371/journal.pone.0360161

Keywords: high-density DC resistivity, pseudo-dipole array, electrical resistivity tomography, Wenner array, array transformation, numerical modeling, aquifer mapping, inversion, quality control, hydrogeophysics, comparative, pseudo-dipole

The study, published in the journal Plant and Soil, monitored three wheat genotypes with contrasting root architectures through a full season of terminal drought, the period during grain filling when water supply becomes progressively restricted and yield potential is sealed. Rather than relying on qualitative snapshots of soil wetness, the team converted time-lapse resistivity images into depth-resolved estimates of root water uptake, achieving a cross-validated accuracy of roughly 0.019 cubic metres of water per cubic metre of soil against independent sensor measurements. The result is a non-invasive phenotyping framework that can distinguish not just how much water a genotype extracts, but when, and from where in the profile.

The technical achievement rests on solving a long-standing calibration problem. ERT works by injecting electrical current into the soil through arrays of electrodes and measuring the resulting voltage differences; because soil electrical resistivity is strongly sensitive to water content, drying zones light up as resistivity hotspots. But translating resistivity into actual water content has traditionally required fixed petrophysical relationships, such as Archie-type models, which assume a single universal equation links the two quantities. In heterogeneous field soils, where texture and pore structure change with depth, that assumption breaks down. The Israeli team instead trained a Random Forest regression model that used both resistivity and measurement depth as predictors, allowing the data itself to define how the resistivity-moisture relationship shifts between soil layers.

The calibration dataset came from 278 paired observations between ERT-derived resistivity and Time-Domain Reflectometry sensors installed at depths from 0.2 to 1.5 metres. When the researchers compared five candidate models using cross-validation grouped by survey date, a strategy that prevents temporal data leakage, the Random Forest dramatically outperformed the analytical alternatives, which produced errors roughly three times larger. The machine learning model achieved a correlation of 0.974 with directly measured soil water content, with negligible bias, demonstrating that the geophysical signal could be trusted as a quantitative, not merely qualitative, record of soil drying.

The field experiment itself was a feat of logistical intensity. Under a mobile rain-out shelter at the university’s Rehovot experimental farm, 64 stainless steel electrodes lined a 32-metre transect divided into three 6-metre genotype zones. The team conducted 44 ERT surveys, at least three per week, from January through crop maturity, each requiring about 52 minutes. Terminal drought was imposed in mid-February by excluding all rainfall, leaving the plots with just 142 millimetres of total available water. Because rainfall was excluded and the coarse, sandy soil limited capillary rise and drainage, the researchers could attribute changes in soil water content almost entirely to root uptake, converting the calibrated moisture time series into cumulative extraction estimates for each 10-centimetre soil layer.

What the images revealed were three fundamentally different water-use strategies. The genotype WM-203, which combined a wide root angle with high root biomass, displayed what the authors describe as an aggressive extraction strategy: it engaged multiple soil layers early, simultaneously, and in a coordinated fashion, with correlation coefficients between daily uptake at different depths reaching 0.80 to 0.98. Its cumulative uptake climbed steeply from 11 millimetres in early February to roughly 85 millimetres by early April, the highest seasonal total, representing about 60 percent of the available water. This spatial redundancy, drawing from many depths at once, meant the plant never depended on a single moisture reservoir as the profile dried.

WM-140, a narrow-angled, high-biomass genotype, played a waiting game. Its uptake was dominated early by a single deep layer at 0.8 metres, which contributed a remarkable 47 percent of total uptake in February, the highest single-depth proportion observed anywhere in the experiment. As the season progressed, extraction broadened across mid- and deep layers, and cumulative uptake surged in mid-March, precisely when atmospheric vapour pressure deficit climbed toward its seasonal peak of 2.8 kilopascals. The third genotype, IPLR-760, with wide roots but low biomass, showed the most fragmented behaviour: erratic switching between shallow and deep extraction zones, with peak depths shifting by more than 0.6 metres between consecutive sampling dates, and weak hydraulic coupling in the mid-profile, where depth-pair correlations fell as low as 0.52.

The physiological consequences tracked the underground patterns with striking fidelity. WM-203 sustained photosynthetic capacity through grain filling and produced the highest grain yield at 7.0 tonnes per hectare, with remarkably low variability, translating to a water productivity of about 8.2 kilograms of grain per cubic metre of extracted water. WM-140 reached 6.3 tonnes per hectare, though with high replicate-to-replicate variability that hints at the risk inherent in its dependence on late-season deep moisture. IPLR-760, despite extracting a comparable total volume of water to WM-140, roughly 75 millimetres, collapsed physiologically: its chlorophyll fluorescence plunged to the lowest value recorded in the entire dataset during the critical grain-filling window, and its yield crashed to just 1.5 tonnes per hectare, with water productivity of a mere 1.9 kilograms per cubic metre.

That last comparison may be the study’s most consequential finding. Two genotypes extracted nearly identical volumes of water over the season, yet one produced four times more grain than the other. Total water uptake, in other words, is not the story; the timing, depth distribution, and coordination of uptake are. The authors argue that drought adaptation hinges on synchronising root hydraulic activity with canopy demand during grain filling, and that structural traits such as rooting depth and root biomass, while useful, are insufficient predictors of field performance on their own. A sparse root network, they suggest, may simply lack the density of active uptake sites needed to sustain water supply under the steep hydraulic gradients of a drying profile, a hypothesis consistent with broader soil-plant-atmosphere theory in which root conductance becomes the bottleneck under deficit.

The implications for breeding are considerable. Wheat production currently stands near 799 million tonnes annually, and meeting demand projected for 2050 will require at least a 60 percent increase, much of it from water-limited environments where every millimetre of extracted soil water counts. If ERT-based phenotyping can be scaled to larger breeding populations, breeders could select directly for functional water-use coordination rather than proxy traits measured in pots. The framework is not without limitations: the genotype zones along the transect were not replicated across independent plots, inversion smoothing and petrophysical model choice influence the depth resolution of the estimates, and the absence of direct root measurements leaves some structural interpretations open. The authors themselves call for integration with root imaging, isotopic water-source partitioning, and replicated transects in future work. But the core demonstration stands: a technique borrowed from hydrogeology, married to machine learning, has made the hidden half of the wheat plant visible, quantitative, and breedable, at exactly the moment a warming climate is making every root’s strategy a matter of global food security.

Subject of Research: Non-invasive quantification of root water uptake in field-grown wheat under terminal drought using electrical resistivity tomography and machine learning calibration

Article Title: ERT-based root water uptake quantification in field-grown wheat under terminal drought

Article References: Akaerue, E., Kashi, O., Sadeh, R., Peleg, Z., & Schwartz, N. (2026). ERT-based root water uptake quantification in field-grown wheat under terminal drought. Plant and Soil. https://doi.org/10.1007/s11104-026-09078-2

Image Credits: AI Generated

DOI: 10.1007/s11104-026-09078-2

Keywords: Electrical Resistivity Tomography, root water uptake, wheat, terminal drought, Random Forest, soil water content, root system architecture, drought phenotyping, water productivity, grain yield, hydrogeophysics, plant and soil

The stakes could hardly be higher. Groundwater quantity within Ibeator is already insufficient for its growing population, and inadequate investigation has produced a string of failed boreholes. Without a clear picture of how thick the protective layers above an aquifer are, how deep the water sits, and how readily contaminants could migrate downward, planners are essentially drilling blind. The study set out to close that knowledge gap by pairing a classic geophysical field technique with modern data-driven modeling, offering a template that could be transferred to regions with similar geology.

The fieldwork rested on twelve vertical electrical soundings conducted with an ABEM Tetrameter SAS 1000 resistivity meter using the Schlumberger configuration, with current electrode spacing reaching 900 meters. This method injects current into the ground and measures the resulting voltage differences, allowing researchers to reconstruct how electrical resistivity changes with depth. Because the resistivity of a layer depends on its lithology, porosity, water content, and salinity, these soundings act as a non-invasive X-ray of the subsurface. Several soundings were deliberately placed near existing boreholes so the interpreted geoelectric sections could be checked against observed lithology, a validation step that boosted confidence when the predicted depth to the aquifer of 90 and 81 meters compared favorably with the 110 meters recorded in the geological log at Okorobi.

Interpretation of the soundings revealed a seven-layer subsurface model across the study area, with resistivity values spanning from about 72 ohm-meters in conductive clays to more than 8,000 ohm-meters in resistive sands. At one representative station, the sequence progressed from lateritic topsoil through clay, silty sand, dry sand, wet sand, and finally a thick saturated sandstone aquifer extending to depths of around 170 meters before a basal clay was encountered. From these geoelectric parameters, the team derived the Dar-Zarrouk quantities, longitudinal conductance and transverse resistance, which link the electrical behavior of the subsurface to its hydraulic properties through Darcy-type relationships. Those derived parameters, including hydraulic conductivity, transmissivity, and storativity, became the raw material for the machine learning comparison.

The spatial patterns that emerged are striking. Aquifer thickness is greatest in the eastern part of the study area around Umudime, reaching 90 to 210 meters, while the western communities of Okorobi, Okahia, and Uhuala sit above much shallower aquifers. Overall aquifer depths ranged from about 45.6 meters to 408.85 meters, averaging 227.21 meters. Resistivity of the aquifer material was highest, between 10,000 and 28,000 ohm-meters, around Okorobi and Uhuala, and lowest, between 1 and 2,000 ohm-meters, toward the east and northeast. Crucially, hydraulic conductivity, with values between roughly 0.000174 and 0.000215 meters per second, showed an inverse relationship with resistivity: where the subsurface resists electrical current, it also tends to resist the flow of water.

Transmissivity and storativity maps echoed the same geography, with transmissivity values from about 0.0035 to 0.0455 square meters per second and storativity from roughly 0.0077 to 0.324 square meters per second, both declining from north to south and lowest at the western and eastern edges. These variations matter because they translate directly into vulnerability. A shallow aquifer with thin protective cover and high transmissivity offers contaminants a fast track to the water table, whereas deeper, better-shielded units are naturally more resilient. The lithologic shifts observed from sand and sandstone in the northeast toward shale-rich sequences in the northwest, near the contact between the Imo Shale and the Benin Formation, further modulate how water moves and how well each zone is protected.

With the dataset assembled, the researchers built two families of predictive models. The artificial neural network used a feedforward, supervised architecture with backpropagation learning, taking three inputs, aquifer depth, aquifer resistivity, and aquifer thickness, through a hidden layer of three nodes to a single output. Seventy percent of the samples trained the network and thirty percent were reserved for testing, with the architecture tuned experimentally by adjusting neuron counts and training proportions. The multivariate linear regression approach, by contrast, fit explicit equations relating the same three predictors to each hydraulic property, with predictor selection guided by Pearson correlation coefficients and standard assumption checks for linearity, multicollinearity, homoscedasticity, normality of residuals, and independence of errors.

The head-to-head results were unambiguous. For hydraulic conductivity, the ANN managed an R-squared of only 0.0869, with training errors of 4.037 times ten to the minus five in root mean squared error and 0.343 in mean absolute percentage error, while the MLR achieved an R-squared of 0.9775 with far lower training errors. For transmissivity, the ANN reached an R-squared of 0.9157 against 0.9958 for the regression model, and for storativity the MLR again led with 0.9868 versus 0.9157. The authors attribute the neural network’s underperformance to the small dataset: ANNs typically require large volumes of data to learn complex nonlinear patterns, and with only twelve soundings the network likely underfit the underlying relationships, whereas the regression equations captured the dominant linear structure with remarkable precision.

Sensitivity analysis of the trained models also revealed which inputs mattered most, with aquifer depth and resistivity ranking highest in different orders depending on the target property. The practical payoff is a set of explicit equations that water managers in Ibeator and geologically similar terrain can apply to new geoelectric data to estimate hydraulic conductivity, transmissivity, and storativity without expensive pumping tests. That, in turn, enables targeted protection: communities over shallow, highly transmissive aquifers can be prioritized for stricter sanitation controls and careful borehole siting, while deeper, better-protected zones can absorb more intensive development.

The broader lesson resonates well beyond southeastern Nigeria. In an era when deep learning dominates headlines, this study is a reminder that model sophistication must match data availability, and that transparent statistical models can deliver both accuracy and interpretability where data are scarce. The authors suggest that the hybrid strategy, using neural networks to probe nonlinear structure and regression to quantify the influence of each factor, offers a holistic picture of aquifer vulnerability and a practical path toward sustainable groundwater management, improved water security, and environmental conservation for the region’s growing population.

Subject of Research: Prediction of aquifer protective capacity and hydraulic parameters in southeastern Nigeria using artificial neural networks and multivariate linear regression applied to vertical electrical sounding data

Article Title: Enhancing aquifer protective capacity prediction over Ibeator and environ, Southeastern Nigeria using artificial neural networks and multivariate linear regression analysis

Article References: Usman, A. O., Akakuru, O. C., Azuoko, G.-B., Abraham, E. M., Chinwuko, A. I., & Chizoba, C. J. (2024). Enhancing aquifer protective capacity prediction over Ibeator and environ, Southeastern Nigeria using artificial neural networks and multivariate linear regression analysis. BMC Environmental Science, 1(1), Article 13. https://doi.org/10.1186/s44329-024-00013-3

Image Credits: AI Generated

DOI: 10.1186/s44329-024-00013-3

Keywords: groundwater, aquifer vulnerability, artificial neural network, multivariate linear regression, vertical electrical sounding, hydrogeophysics, hydraulic conductivity, transmissivity, storativity, Nigeria, geoelectric survey, water resource management