The investigation identified a total of 237 cases of acute diarrheal illness during the outbreak period. The age distribution of cases was striking: more than half of those affected, 55.2 percent, were individuals under 20 years of age. This pattern is consistent with the epidemiology of many enteric infections, where younger people, with less developed hygiene habits, closer contact in household and school settings, and greater susceptibility, often bear a disproportionate share of the disease burden. The concentration of cases among children and adolescents also raises concerns about the potential for onward transmission within households and community institutions, a dynamic that the investigators would go on to examine through the lens of spatial statistics.

Environmental investigations formed a central pillar of the outbreak response. The team sampled water sources across the affected area, including the municipal supply maintained by the Public Health Engineering Department and household-level storage containers. The results were revealing: microbiological contamination was detected in household-stored drinking water and in a household well, whereas no contamination was found in the sampled PHED water supply source. This dissociation between a clean central supply and contaminated point-of-use water is a well-recognized hazard in rural water systems, where water that leaves a treatment facility in good condition can be recontaminated during storage and handling in the home. Taken together with the observed clustering of cases, the findings suggest that localized environmental exposure, potentially associated with point-of-use water contamination, played a role in the outbreak.

The microbiological picture that emerged from patient samples was more complex than a single-organism explanation would allow. Stool samples from 22 patients were tested, and Norovirus GII was identified in 22.7 percent of them. Blood samples from 18 patients were screened for viral hepatitis, and Hepatitis A virus was detected in 5.6 percent of those samples. Escherichia coli was detected in all 16 stool samples that underwent microbiological analysis, although the authors caution that these organisms were probably normal gut flora and that the samples were not further screened for specific diarrheagenic strains; at the same time, they note that the possibility of pathogenic E. coli strains contributing to illness cannot be completely ruled out. This multi-pathogen etiology complicates the interpretation of the outbreak but is not unusual in settings where water, sanitation, and hygiene infrastructure is strained and multiple enteric pathogens can circulate simultaneously.

The detection of Norovirus GII is particularly noteworthy. Noroviruses are a leading cause of acute gastroenteritis worldwide, responsible for explosive outbreaks in settings ranging from cruise ships to schools, and the GII genogroup contains the strains most commonly associated with human disease. Yet in rural Bihar, diarrheal outbreaks have traditionally been attributed to bacterial pathogens, particularly Vibrio cholerae and various enteric bacteria, and viral etiologies have often gone undetected simply because diagnostic capacity for them has been limited. The identification of Norovirus GII in more than a fifth of tested patients suggests that the etiological profile of diarrheal disease in the region may be evolving, or at least that improved molecular diagnostics are revealing pathogens that earlier investigations would have missed. Either way, the finding underscores the importance of maintaining and expanding viral diagnostic capacity, such as that provided by the Viral Research and Diagnostic Laboratory network, in state surveillance systems.

What sets this investigation apart methodologically is its use of geostatistical analysis to characterize the spatial structure of transmission. The researchers began by applying Global Moran’s I statistics to the geographic coordinates of cases, a test that asks whether the spatial distribution of disease intensity deviates from complete spatial randomness. The answer was a clear yes: the analysis confirmed significant global spatial clustering, with a Moran’s I value of 0.1161, a z-score of 3.64, and a p-value below 0.001. In practical terms, this means that cases were not scattered randomly across the landscape but tended to occur near other cases, a pattern consistent with a localized environmental exposure or with person-to-person transmission operating at the neighborhood scale.

Global statistics, however, can only say that clustering exists somewhere in the study area; they cannot say where. To pinpoint the locations of intense transmission, the team turned to the Getis-Ord Gi* statistic, a local indicator of spatial association that evaluates each location against its neighbors and assigns a measure of statistical confidence to the degree of clustering. Combined with Inverse Distance Weighting, a technique that interpolates disease intensity across the map based on nearby observations, the Getis-Ord Gi* analysis identified high-intensity clusters at 99 percent confidence, corresponding to the highest bin in the classification scheme, specifically within the Paswan and Manjhi Tolas. These are settlements of marginalized Mahadalit communities, and their identification as hot spots directs attention to the intersections of social vulnerability, environmental exposure, and infectious disease risk that shape outbreak patterns in the region.

The spatial findings, interpreted alongside the microbiological and environmental evidence, led the investigators to a nuanced conclusion about the outbreak’s dynamics. Rather than a single confirmed point source, such as one contaminated well or one contaminated batch of water, the pattern of statistically significant clustering within specific settlements, combined with contamination found at the household level rather than in the central supply, supports a hypothesis of environmental exposure followed by subsequent person-to-person transmission. This kind of mixed dynamic is common in norovirus outbreaks, where an initial environmental introduction can seed chains of transmission that spread efficiently through close contact, contaminated surfaces, and the characteristic aerosolization that accompanies vomiting. Recognizing such dynamics matters operationally, because interrupting person-to-person transmission requires interventions, including hygiene promotion, isolation of cases, and disinfection, that go beyond fixing a single water source.

Beyond its findings about this particular outbreak, the study offers a replicable template for outbreak investigation in settings where surveillance data are often aggregated at coarse administrative levels, obscuring the fine-grained spatial structure of transmission. By applying localized spatial autocorrelation to case-level coordinates, the team transformed routine surveillance data into precise, statistically robust maps of disease clustering. The authors argue that integrating such advanced spatial statistics into routine outbreak investigations enhances the precision of public health interventions, allowing resources to be targeted to the specific neighborhoods and communities where transmission is most intense. In a state like Bihar, where diarrheal disease outbreaks remain a persistent public health challenge and where marginalized communities such as the Mahadalit tolas face compounding environmental and social risks, this approach promises a more objective and equitable basis for response.

The investigation was conducted as a government-mandated public health response, exempted from individual ethics review under the ICMR National Ethical Guidelines as fundamental public health practice, and authorized by the Office of the District Magistrate of Jehanabad. It drew on the coordinated efforts of district surveillance units, rapid response teams, block health officials, and community health workers, supported technically by the Asian Development Research Institute and health system strengthening funding from the Gates Foundation. As molecular diagnostics become more widely available and spatial analytical tools become more accessible to district-level teams, the Bihar experience suggests that the future of outbreak response in resource-constrained settings may lie in combining these approaches: identifying not only what pathogen is making people sick, but precisely where transmission is concentrated, and tailoring interventions to the social and environmental realities of the communities most affected.

Subject of Research: Geostatistical analysis of a norovirus GII-associated acute diarrheal outbreak in Jehanabad, Bihar, India

Article Title: Spatiotemporal dynamics and multi-pathogen etiology of a norovirus GII-associated diarrheal outbreak in Bihar: a geostatistical analysis of localized spatial autocorrelation

Article References: Kumar, A., Gupta, N., Mishra, R., Sharma, A., Razdan, D., Kumar, A., Rai, A., Pandey, A., Prasad, D., Mishra, R., & Mahapatra, S. (2026). Spatiotemporal dynamics and multi-pathogen etiology of a norovirus GII-associated diarrheal outbreak in Bihar: a geostatistical analysis of localized spatial autocorrelation. BMC Infectious Diseases. https://doi.org/10.1186/s12879-026-14497-8

Image Credits: AI Generated

DOI: 10.1186/s12879-026-14497-8

Keywords: Norovirus GII, spatial autocorrelation, Getis-Ord Gi*, Moran's I, outbreak investigation, Bihar, diarrheal disease, Hepatitis A virus, water contamination, epidemiology, geostatistics, public health

The research, published in Discover Geoscience by Dionisio Uendro Carlos, César Augusto Moreira, and Guilherme Libardi Maestrelo, introduces a moving-window ordinary kriging (MW-OK) algorithm designed specifically for airborne and ground radiometric data. Conventional interpolators such as inverse distance weighting and minimum curvature are fast and widely used, but they share a fundamental weakness: they do not explicitly account for the local spatial structure of the data. In practice, that means they can oversmooth short-wavelength variability, blurring exactly the sharp radiometric gradients that mark boundaries between rock units or zones of hydrothermal alteration.

Kriging, by contrast, is a geostatistical method built on a statistical model of how a property varies in space. Because radioactive decay is inherently a stochastic process, the authors argue, an interpolator that respects this randomness is desirable. Yet standard kriging typically relies on a single global variogram, the mathematical function describing spatial correlation, assumed to hold across the entire survey area. Radiometric variables rarely obey that assumption at regional scale. Lithological boundaries, weathering domains, and survey-related effects all conspire to make the spatial structure change from place to place, so a global model can over-smooth local features or produce biased estimates where the structure shifts.

MW-OK sidesteps this problem by abandoning the global model altogether. For every prediction node on the output grid, the algorithm defines a circular window of investigation, in this case with a radius of 750 meters, roughly 1.5 times the 500-meter flight-line spacing, so that each window intersects two adjacent flight lines. Within that window, the method assumes second-order stationarity holds only approximately and locally. An empirical semivariogram is computed from all unique pairs of samples in the neighborhood, binned into eight lag classes, with any bin containing fewer than three valid pairs rejected. An exponential model is then fitted to each local semivariogram by bounded non-linear least squares, yielding a nugget effect, partial sill, and range parameter tailored to that specific patch of ground.

The computational scale is striking. The study area, a 40 by 40 kilometer subset of the Sul-Riograndense Shield Aerogeophysical Project flown by the Geological Survey of Brazil in 2010, contains 50,374 measurement points. Because the algorithm estimates a local variogram for each measurement point, it effectively generates 50,374 variograms. Neighbor retrieval is accelerated with a k-dimensional tree using ball queries, and safeguards keep the local kriging systems well conditioned: a minimum of eight neighbors, a maximum of thirty, removal of duplicate coordinates to prevent singular matrices, and a tiny ridge term added to the kriging matrix diagonal for numerical robustness. If a non-linear variogram fit fails, the workflow falls back on physically sensible defaults derived from the local data variance and median nearest-neighbor spacing.

The fitted local variograms themselves carry geological information. For potassium, the nugget effect was essentially zero at the scale of the data support, meaning microscale variability and measurement noise are minor compared with the spatially structured component, and correlation decayed over a practical range of about 1.3 kilometers. Equivalent uranium showed an even longer continuity, around 1.7 kilometers, consistent with smoother spatial behavior, while equivalent thorium displayed a larger partial sill, indicating higher structured variance, with a moderate range near 1.5 kilometers. These parameters confirm that, within each window, the radiometric variability is predominantly spatially structured rather than random noise.

When the researchers applied MW-OK to the survey data, the resulting potassium, uranium, and thorium maps reproduced the broad patterns that geologists expect. Potassium enrichment above 4 percent traced the São Sepé Granite in the north, elevated values of roughly 4 percent marked the Ramada Granite in the southwest, and low concentrations coincided with the metatuff-dominated Bossoroca Complex and the sandstones of the Maricá Group. Thorium highs of about 21 parts per million again outlined the São Sepé Granite, with additional anomalies over the Cerro da Cria Granite and nearby rhyolites. But the kriged surfaces went further than the conventional minimum-curvature grids: they better represented local gradient variations near lithological contacts and revealed internal heterogeneity within the felsic bodies that the smoother deterministic algorithm muted. Notably, three east-west linear artifacts aligned with the survey tie lines, visible in the minimum-curvature uranium map, were suppressed in the kriged surface.

A key advantage of kriging is that it also delivers an estimate of its own uncertainty. The kriging variance maps for all three radioelements bore the clear imprint of the acquisition geometry: uncertainty was minimized along and near the north-south flight lines and increased in the gaps between them, telling interpreters exactly where the estimates are most reliable. To quantify how closely MW-OK matched the industry-standard approach, the team performed pixel-wise comparisons with the minimum-curvature grids. The agreement was remarkably strong, with coefficients of determination of 0.9938 for potassium, 0.9727 for equivalent uranium, and 0.9946 for equivalent thorium, and residual histograms tightly centered near zero. Ordinary least-squares fits revealed a subtle but systematic amplitude scaling, with slopes slightly below unity, suggesting MW-OK modestly attenuates the highest values while remaining close to the identity overall. The authors are careful to frame this as evidence of methodological consistency rather than broad superiority: the two methods agree at the domain scale, and their differences emerge primarily at local scales.

Beyond the three radioelement grids, the workflow generates the interpretive products that exploration geologists actually use. It computes standard radiometric ratios (K/eTh, K/eU, eU/eTh) and the F-parameter, a widely used proxy for hydrothermal alteration that relates potassium abundance to the uranium-to-thorium ratio. It also produces ternary radioelement maps in three complementary forms: the classical K-eU-eTh display, normalized ternaries, and compositional-data transformations using additive, centered, and isometric log-ratios that respect the closure constraint and reduce spurious correlation when treating the three elements as parts of a composition. In the study area, a historically significant gold district in southern Brazil where mineralization is hydrothermal and structurally controlled in quartz veins hosted by schists and granites, these products delineated the main lithological contacts and showed spatial correspondence with known gold-mineralized zones and areas of gold enrichment identified by stream-sediment prospecting.

Perhaps most importantly for the wider community, the entire workflow is open source and exports standard deliverables, including GXF grids, GeoTIFF rasters, KMZ overlays, and PNG figures, with proper spatial metadata, making it directly interoperable with both open-source tools such as Python, QGIS, and Google Earth and commercial platforms like Oasis Montaj, Intrepid Geophysics, and ArcGIS. The authors acknowledge the method’s limitations: effective spatial resolution remains constrained by flight-line spacing and sampling interval, computational cost grows with finer grids and larger neighborhoods, and sparsely sampled sectors can yield under-populated windows with unstable variograms, which the workflow mitigates through minimum-neighbor thresholds and k-nearest-neighbor fallbacks. Looking ahead, they plan to incorporate co-regionalized cokriging constrained by lithology and structure, and deep-kriging approaches, within the same moving circular-window framework. For now, MW-OK stands as a robust, locally adaptive alternative that offers incremental rather than transformative gains over deterministic gridding, but gains that could make the difference between a blurred contact and a clearly resolved exploration target.

Subject of Research: A locally adaptive geostatistical interpolation method for processing airborne and ground gamma-ray spectrometric data

Article Title: A moving-window ordinary kriging approach for airborne and ground gamma-ray spectrometric data processing

Article References: Carlos, D. U., Moreira, C. A., & Maestrelo, G. L. (2026). A moving-window ordinary kriging approach for airborne and ground gamma-ray spectrometric data processing. Discover Geoscience, 4(1), Article 289. https://doi.org/10.1007/s44288-026-00674-4

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00674-4

Keywords: gamma-ray spectrometry, kriging, geostatistics, radiometric mapping, mineral exploration, semivariogram, interpolation, gold prospecting, geological mapping, Brazil, open-source software, compositional data

The problem that motivated the study is a familiar one in polar science: radar, the workhorse tool for measuring ice thickness, struggles in exactly the places where the answers matter most. Radio-echo sounding works beautifully over gentle, well-behaved beds, but the Denman Glacier Trough is deep, narrow, and steep-walled, roughly 20 kilometers wide and 110 kilometers long, and its flanks generate off-nadir echoes and layover that scramble the return signal. Crevasse fields near the grounding line scatter the pulse, warm or water-saturated basal sediments attenuate it, and beneath the floating ice shelf the ocean simply swallows the radar beam entirely. As a result, bed observations in the region are sparse and unevenly distributed, and the continent-scale compilations that most ice sheet models rely upon, BedMachine and the newer Bedmap3, must interpolate or extrapolate across large gaps. Interpolation smooths sharp relief and can bridge across steep troughs, underestimating both depth and the steepness of the sidewalls, while mass-conservation methods that fuse radar thickness with satellite velocities can recover deep troughs but remain highly non-unique where surface velocity gradients are small.

To break this impasse, the team turned to a completely independent physical signal: gravity. During the Australian Denman Terrestrial Campaign in the 2023/24 austral summer, the researchers hauled a Scintrex CG-5 Autograv gravimeter across the ice surface, collecting nine high-precision gravity measurements along a roughly 14-kilometer transect that crosses the deepest part of the trough, with stations spaced about 1.7 kilometers apart. Ground-based gravity has a crucial advantage over airborne surveys: because the instrument sits on the ice rather than flying kilometers above it, it retains the short-wavelength gravity signals that airborne data lose to mandatory filtering and upward continuation. The team tied their readings to the absolute gravity reference at Casey Station, corrected for instrument drift using repeated base measurements at Bunger Hills, and computed free-air anomalies on the WGS84 ellipsoid with careful solid-Earth tide and atmospheric-pressure corrections. These ground data were then merged with the extensive ICECAP airborne gravity, magnetic, and radar archive collected between 2008 and 2018.

The heart of the method is a two-scale, ensemble-based gravity inversion built on a Markov Chain Monte Carlo framework. Gravity inversion is notoriously non-unique, meaning an infinite family of density and depth combinations can reproduce the same measured anomaly, so the authors attacked the ambiguity head-on. First, they used sequential Gaussian simulation, a geostatistical technique, to generate an ensemble of plausible non-terrain gravity disturbances, the long-wavelength background signal produced by regional geology rather than by the bed itself. A smooth regional trend was fitted to locations with reliable radar-based topographic control, the residuals were characterized with a directional variogram, and hundreds of statistically realistic residual fields were simulated. Subtracting each of these from the isostatically corrected gravity disturbance yielded an ensemble of target terrain effects, one hundred for the regional inversion at 2-kilometer resolution and fifty for the high-resolution local inversion at 1-kilometer resolution focused on the trough.

Each terrain effect was then fed into a random-walk Metropolis-Hastings MCMC inversion. Starting from a BedMachine-based bed plus a smooth Gaussian perturbation, the algorithm repeatedly proposed small, spatially correlated Gaussian patches of bed change, recomputed the forward gravity response only where it mattered for efficiency, and accepted or rejected each proposal using a joint likelihood that weighs both the gravity misfit and the radar bed picks, with adopted uncertainties of 1.5 mGal for gravity and 30 meters for radar. By progressing from large update stencils to small ones, the multiscale scheme captured the long-wavelength geometry first and then layered on finer detail. The result is not a single map but an ensemble of bed realizations, each internally consistent with the data, together with a standard-deviation field that honestly quantifies where the bed is well constrained and where it remains contested. Ensemble spread is low, below about 200 meters, where radar and gravity jointly constrain the bed, and climbs above 900 meters along the steep, poorly imaged trough interior.

What the ensemble reveals is a fundamentally different landscape from the one in the standard products. Where BedMachine and Bedmap3 depict the Denman trunk as a single, continuous, elongated depression, the gravity-derived reconstruction shows a compartmentalized, rugged terrain: multiple localized depressions and pockets of varying depth, subdued ridges, and steeper trough walls with greater lateral relief than either compilation. Along cross-trough profiles near the grounding line, the best-fit gravity bed generally falls between the shallower Bedmap3 and the deeper BedMachine interpretations, yet consistently exhibits sharper, more sharply defined geometry than either. Perhaps most strikingly, the ensemble hints at a subdued bedrock high within the central trough that may partition it into two connected basins, and it traces a deep, trough-like channel continuing beneath the ice shelf, a feature resolved by neither BedMachine nor Bedmap3 because no radar data exist there. The best-fit gravity model reproduces the observed gravity field with a mean absolute error of 1.6 mGal and honors the radar picks to within about 100 meters across most of the domain.

The geology beneath the ice adds a second layer of insight. Using Euler deconvolution of reprocessed ICECAP magnetic data, the team estimated the depth to magnetic sources across the region. On the western flank of the trough, magnetic sources lie several kilometers deep, consistent with crystalline basement, while the eastern side shows subdued magnetic responses characteristic of sedimentary or metasedimentary units. A cluster of subvertically aligned magnetic-source solutions in the central profile suggests a steeply dipping fault zone, likely a lithological boundary separating the two terranes, lying within 1 to 2 kilometers of the previously proposed Scott Fault. The Denman Glacier Trough, the authors argue, appears to exploit this ancient structural weakness, echoing earlier work suggesting that a mechanically weak, potentially water-saturated sedimentary bed along the Knox Rift, a failed rift from the separation of India from East Gondwana with up to 6 to 7 kilometers of infill, helped focus ice flow into this corridor.

The dynamical implications are sobering. Because much of the Denman system is grounded below sea level, its stability hinges on bed geometry through the mechanism known as Marine Ice Sheet Instability. When a grounding line retreats onto a bed that slopes downward inland, the ice at the grounding line becomes thicker, the outflow flux increases nonlinearly with thickness, and retreat feeds on itself in a positive feedback loop. The gravity-derived geometry, with its steep flanks, retrograde inland-sloping basin, and strong asymmetry, reinforces the glacier’s susceptibility to this instability, although the authors are careful to note that geometry alone cannot prove instability is underway; pinning points such as along- and across-trough ridges could temporarily stabilize the grounding line by adding basal and lateral drag. The observed context is already unsettling: the glacier’s grounded portion has accelerated by 174 percent over the past five decades, and its grounding line has retreated more than 5 kilometers since 1996.

Beyond Denman itself, the study carries a broader message for Antarctic science. Ice sheet models are only as good as the bed maps beneath them, and the discrepancies between the gravity-constrained geometry and widely used compilations suggest that many current models may rely on overly smoothed bed representations, potentially underestimating grounding line sensitivity in fast-flowing outlet systems where modest bed changes translate into major dynamical consequences. The authors argue that geophysical inversion methods deserve a place alongside radar interpolation and mass conservation in future continental bed-mapping efforts, and their ensemble of bed realizations is immediately usable as boundary conditions for ice-flow modeling and grounding-line stability assessments. As the team points out, well-placed new constraints, whether longer ground gravity profiles, magnetotelluric soundings, or seismic data, would do more to sharpen the picture than modest noise reductions. For now, the hidden landscape beneath Denman Glacier stands revealed in sharper relief than ever before, and its steep, segmented, seaward-plunging geometry is a reminder that some of the biggest uncertainties in sea-level projections lie buried under kilometers of ice.

Subject of Research: Geostatistical gravity inversion of subglacial topography beneath the Denman Glacier, East Antarctica

Article Title: Gravity topography modeling of the Denman Glacier region using a geostatistical approach

Article References: Lösing, M., Aitken, A., Field, M., MacKie, E., & Li, L. (2026). Gravity topography modeling of the Denman Glacier region using a geostatistical approach. The Cryosphere, 20(10), 5629-5652. https://doi.org/10.5194/tc-20-5629-2026

Image Credits: AI Generated

DOI: 10.5194/tc-20-5629-2026

Keywords: Denman Glacier, East Antarctica, subglacial topography, gravity inversion, Markov Chain Monte Carlo, radio-echo sounding, BedMachine, Bedmap3, Marine Ice Sheet Instability, sea-level rise, grounding line, geostatistics

The problem the researchers identified is called smoothing-induced weak annotation, or SiWA. In data-driven mineral prospectivity mapping, positive training samples are labeled locations where mineralization is known to occur. Ideally, the area or volume of each labeled sample would exactly match the footprint of the deposit it contains. In practice, at national and continental scales, this is nearly impossible. A single grid cell in a pan-Canadian study might cover more than five square kilometers, while the deposit inside it could occupy a tiny fraction of that space. The result is a positive sample that is overwhelmingly composed of barren ground, diluting the true geological signature of mineralization with noise from its surroundings.

Zhang and Parsa describe this with a simple analogy: a shoebox labeled as containing a tennis ball, when it actually holds a tennis ball buried among many other objects. The label is not wrong, but it is imprecise, and any learning system trained on it must guess which part of the box matters. In geostatistical terms, the annotated sample has a much larger support than the target it is meant to represent. When covariates are averaged over that larger area, the distinctive nonlinear fingerprints of mineralization are attenuated into the background, a process mathematically analogous to low-pass filtering a signal.

The theoretical consequences are striking. As the ratio of annotated area to deposit footprint grows, the averaging of many mostly negative subsamples invokes the central limit theorem, pushing the joint distribution of predictors and labels toward a multivariate normal distribution. The conditional expectation of the predictand then becomes a linear function of the predictors. In other words, smoothing artificially linearizes the relationship the machine learning model is trying to learn. The hypothesis the algorithm fits is no longer the true relationship between mineral-system evidence and mineralization, but a smoothed, linearized approximation of it. This mirrors a well-known effect in time-series forecasting, where averaging over windows longer than the autocorrelation timescale erases high-frequency, nonlinear patterns and erodes the advantage of nonlinear models.

To test these ideas empirically, the authors turned to two pan-Canadian datacubes, one targeting flake graphite and the other magmatic nickel, with or without copper, cobalt, and platinum-group elements. Both mineral systems host deposits believed to be small relative to the resolution of the data, making them naturally vulnerable to SiWA. The datasets contained 1,239 and 1,240 positive samples respectively, drawn from federal, provincial, and territorial geological surveys, ranging from minor occurrences to operating mines. The covariates included geological, geophysical, and geochronological layers such as crustal thickness, gravity and magnetic data, density models, and depth to the Mohorovicic discontinuity, all integrated on the H3 discrete global grid system at resolution 7, where each hexagonal cell averages 5.16 square kilometers.

The experimental design was deliberately rigorous. Rather than hand-picking a single algorithm, the researchers employed a deep ensemble workflow comprising 2,000 unique model configurations per target, spanning eight machine learning algorithms from logistic regression and naive Bayes to random forests, support vector machines, Gaussian processes, and neural networks, combined with 25 feature-space variants, five negative-labeling schemes, and two hyperparameter-tuning metrics. Smoothing was then applied exclusively to positive samples by expanding their labels into progressively larger neighborhoods of hexagonal cells, from the origin cell out to k-rings spanning an average centroid distance of 21.4 kilometers, roughly two orders of magnitude larger than the likely footprint of the deposits. Negative samples and covariates were left untouched, isolating annotation weakness as the sole experimental variable.

The results were counterintuitive and, in some ways, alarming. As annotation resolution decreased, apparent model performance actually improved, with both the area under the receiver operating characteristic curve and the weighted F1-score rising in an asymptotic fashion toward an elbow near k equals 3. At the same time, the prospective area shrank dramatically, with high-probability zones vanishing first, particularly in regions far from known positive samples. The sensitivity of the mapping outcome to annotation resolution proved to be on the same order as the sensitivity to the choice of machine learning algorithm itself, and an order of magnitude greater than the sensitivity to negative samples, tuning metrics, or feature dimensionality. In short, an aspect of study design often treated as innocuous turned out to rival the most heavily engineered component of the entire workflow.

The mechanism behind these trends became clear when the authors stratified their ensembles by performance. The lowest-performing quartile of workflows was dominated by simple, almost exclusively linear models such as logistic regression and naive Bayes configured with few features, and these models were the least affected by smoothing, retaining target areas most similar to the unsmoothed baseline. The highest-performing, most expressive nonlinear models, by contrast, lost nearly all of their advantage once smoothing was applied, contributing essentially nothing to the merged ensembles beyond the baseline. This is the signature of linearization: when the learnable relationship has been flattened by averaging, a linear model can capture it as well as a deep network, and the extra capacity of nonlinear methods is wasted. Meanwhile, workflow-induced uncertainty, measured as the dissent among the 2,000 equiprobable models, spiked precisely where positive samples were located, indicating that models shifted from broad agreement near known mineralization to fundamental disagreement as annotation weakened.

The implications ripple outward in several directions. First, the probabilistic meaning of a prospectivity map depends directly on annotation strength. If cells are coarser than deposit footprints, the chance of actually finding a deposit within any flagged cell is the model’s posterior probability multiplied by an annotation probability that can be vanishingly small; for the Canadian datasets, the authors estimate odds as poor as one in nearly 38,000 for the smallest targets. Field validation campaigns must therefore acquire far more samples to reach statistical significance when annotation is weak. Second, performance comparisons between studies conducted at different resolutions, on different targets, or in different regions are fundamentally incomparable unless annotation strength is explicitly controlled, a caution that strikes at the heart of the benchmarking and review culture in machine learning for geoscience. Third, and perhaps most practically, model complexity should be matched to annotation strength rather than maximized, because expressive models trained on heavily smoothed data learn relationships that are artifacts of the labeling strategy rather than properties of the Earth.

Solutions are not straightforward. Approaches developed in medical image segmentation and other weak-supervision domains, such as blending strong and weak labels, refining annotations, or rejecting noisy samples, assume that strongly annotated data exist somewhere in the pipeline, an assumption that fails at continental scales where deposit footprints are unknown and covariate resolution is costly to improve. The authors argue that the most reliable remedy, at least for now, is the one already familiar from time-series forecasting: choose simpler, more linear models when annotation strength is low, and never let model expressivity exceed what the data can support. Over the longer term, improving the resolution of primary and secondary datasets and profiling the geostatistical footprints of positive samples could reduce the severity of SiWA at its source. What is certain is that annotation quality, long a quiet afterthought in mineral prospectivity mapping, must now be recognized as a first-order control on what these powerful AI maps actually mean, and on whether the critical minerals they promise can truly be found.

Subject of Research: The effect of smoothing-induced weak annotation on data-driven mineral prospectivity mapping models

Article Title: The Effects of Smoothing-Induced Weak Annotation in Mineral Prospectivity Mapping

Article References: Zhang, S. E., & Parsa, M. (2026). The Effects of Smoothing-Induced Weak Annotation in Mineral Prospectivity Mapping. Natural Resources Research. https://doi.org/10.1007/s11053-026-10760-6

Image Credits: AI Generated

DOI: 10.1007/s11053-026-10760-6

Keywords: mineral prospectivity mapping, machine learning, weak annotation, geodata science, data quality, smoothing, critical minerals, graphite, magmatic nickel, uncertainty, geostatistics, mineral exploration

The stakes of this comparison are enormous. Accurate maps of soil pH, organic carbon, texture, and nutrients underpin sustainable agriculture, environmental management, and land-use planning. Conventional soil mapping techniques often fail to capture the fine spatial variability of soils and the complex interactions between soil properties and landscape features. The three methods reviewed represent fundamentally different philosophies: HASM is a deterministic approach grounded in differential geometry, BME is a probabilistic geostatistical framework, and machine learning is a data-driven pattern finder. Until now, the literature had never systematically compared all three, leaving researchers without an evidence base for choosing among them.

The mathematics behind each method reveals why they behave so differently. HASM rests on the fundamental theorem of surface theory, which states that a surface is uniquely determined by its first and second fundamental forms. The first form, expressed through coefficients E, F, and G, describes how rapidly a soil property such as nutrient concentration changes in the east-west and north-south directions, enabling calculations of curve lengths, angles, and surface areas. The second form, with coefficients L, M, and N, captures the local curvature of the soil surface, which governs how anomalies such as nutrient hotspots or erosion-prone zones are rendered on the predictive surface. This geometric rigor allows HASM to honor observed data points exactly and produce smooth, physically realistic surfaces.

BME takes an entirely different route. It treats soil properties as a random field and maximizes Shannon entropy, the mathematical measure of uncertainty, subject to the constraints imposed by available data. The method begins with a maximally non-committal prior probability distribution reflecting general knowledge about spatial variability, then updates it with hard, measured soil samples to produce a posterior distribution. The result is a prediction that is probabilistically coherent, explicitly quantifying uncertainty while integrating both exact measurements and softer, less certain information. This makes BME particularly powerful for combining heterogeneous data sources, a persistent headache in soil science where field surveys, legacy maps, and remote sensing products rarely agree.

Machine learning, by contrast, learns patterns directly from data. Random Forest, an ensemble of unpruned decision trees whose predictions are averaged, has become the workhorse of digital soil mapping because of its robustness with high-dimensional inputs. The Euclidean Distance Field enhancement addresses a critical weakness: raw X-Y coordinates tell a model little about spatial structure. EDF transforms coordinates into distance vectors, computing the Euclidean distance between each query point and sample locations, or distances to the corners and center of the sampling rectangle. Studies reviewed show that this transformation substantially improves prediction accuracy by letting algorithms perceive spatial autocorrelation they would otherwise miss.

The review, conducted under PRISMA 2020 guidelines, screened 701 records down to 190 included studies from Scopus, Web of Science, and PubMed. Machine learning appeared in 28 percent of studies, making it by far the most popular choice, while HASM and BME were used in only 3 percent and 2 percent respectively. Bibliometric analysis revealed a strongly upward publication trend and dense collaborative networks, with Chinese institutions such as Wuhan University, Beijing Normal University, and the Chinese Academy of Sciences forming the largest cluster. Notably, Global South institutions in Ethiopia, Benin, Nigeria, and India are increasingly visible, reflecting how soil degradation and food security concerns drive interest well beyond wealthy nations.

Performance analysis showed that HASM, machine learning, and BME consistently outperformed other approaches, but with striking context dependence. BME excels at handling spatial dependency and outliers, making it well suited to regional mapping, yet it is sensitive to skewness and requires ample sample sizes for stable variogram estimation. HASM delivers exceptional fine-scale accuracy and handles skewness well, but it depends on dense sampling and falters in the presence of outliers. Machine learning is robust to skewness and outliers when datasets are large, but its performance becomes irregular under strong spatial dependency unless spatial information is explicitly engineered into the model through EDF features. Reported R² values ranged widely, from 0.45 to 0.95 for machine learning, 0.78 to 0.80 for HASM, and 0.35 to 0.50 for BME, suggesting that study-specific factors often matter more than the algorithm itself.

The review also uncovered troubling reporting gaps that undermine the field’s credibility. Thirty-four percent of studies did not specify their data sources, and roughly 20 percent relied on Google databases of questionable provenance. Thirty-six percent omitted soil nutrients entirely, and a remarkable 72 percent failed to report soil sampling depth, hindering reproducibility. Publication bias analysis using funnel plots, Egger’s regression test, and the trim-and-fill method suggested that machine learning accuracy may be overestimated by around 5 to 7 percent, with an adjusted median R² of 0.73 after accounting for potentially missing weaker studies. Sensitivity analyses, including leave-one-out testing, confirmed that no single study drove the central findings.

The authors argue that the future lies in hybrid modeling that fuses the pattern-finding strength of machine learning, the uncertainty quantification of BME, and the geometric precision of HASM. Such combinations, they contend, can deliver more accurate and robust predictions across diverse data quality conditions than any single paradigm. Emerging deep learning architectures, including convolutional neural networks for spatial imagery and recurrent networks for temporal soil measurements, remain largely untested against these conventional benchmarks and deserve systematic comparison. Proximal soil sensing and remote sensing could supply the high-resolution data needed to feed all three approaches.

For practitioners, the practical message is clear: model choice should follow data characteristics, not fashion. Regional mapping with spatially correlated data favors BME, provided distributions are transformed when skewed. Fine-scale studies with dense sampling suit HASM. Highly variable datasets with large samples favor machine learning, ideally with Euclidean Distance Fields to encode spatial structure. Above all, the review calls for transparent data reporting, standardized validation protocols, and interdisciplinary collaboration, warning that without these foundations even the most sophisticated algorithms will produce maps that look convincing but cannot be trusted.

Subject of Research: Comparative performance of HASM, Euclidean-enhanced machine learning, and Bayesian Maximum Entropy methods for spatial prediction of soil properties

Article Title: A systematic review of high accuracy surface modeling Euclidean enhanced machine learning and Bayesian maximum entropy for soil property prediction

Article References: Kuse, K. A., Agbangba, C. E., & Kakaï, R. G. (2026). A systematic review of high accuracy surface modeling Euclidean enhanced machine learning and Bayesian maximum entropy for soil property prediction. Discover Soil, 3(1), Article 104. https://doi.org/10.1007/s44378-026-00256-3

Image Credits: AI Generated

DOI: 10.1007/s44378-026-00256-3

Keywords: soil prediction, digital soil mapping, machine learning, HASM, Bayesian Maximum Entropy, Euclidean distance fields, geostatistics, random forest, soil nutrients, systematic review, spatial modeling, remote sensing

The research team, led by Arshdeep Singh Atwal and Aarush Lal of CSK Himachal Pradesh Krishi Vishvavidyalaya, Palampur, focused on cauliflower, a nutrient-hungry cool-season crop that dominates the vegetable economy of the Shivalik foothill zone. Indora block, spanning roughly 298 square kilometres between elevations of 261 and 765 metres, receives about 1000 millimetres of rain annually and enjoys a semi-humid subtropical climate that has made it an important vegetable belt. Yet the region’s agriculture is constrained by tiny landholdings, erratic water availability, and fertiliser use that is often both low and imbalanced. Continuous cropping, excessive tillage, residue removal and skewed fertiliser regimes have accelerated nutrient mining and organic matter depletion, making systematic soil characterisation an urgent priority rather than an academic luxury.

Between August 2022 and 2023, the researchers sampled the top 20 centimetres of soil from ninety-nine farming households growing the Megha cauliflower variety on plots of at least 500 square metres, covering twenty-two village councils and thirty-seven villages. The samples, air-dried and sieved, were subjected to a full battery of standard analyses: bulk density by core sampler, texture by hydrometer, water holding capacity by the Keen and Raczkowski box method, pH by glass electrode, organic carbon by dichromate wet oxidation, available nitrogen by alkaline potassium permanganate, phosphorus by Olsen’s method, potassium by flame photometry, and sulphur by turbidimetry, alongside exchangeable calcium and magnesium. The result was one of the most comprehensive physico-chemical datasets ever assembled for this corner of the lower Himalayas.

The descriptive statistics painted a picture of generally healthy but unevenly distributed fertility. Soil pH ranged from 6.34 to 7.56 with a mean of 7.16, placing the soils in the neutral to slightly alkaline range that suits vegetable production well. Electrical conductivity, averaging 0.26 decisiemens per metre, stayed safely below the 0.8 threshold considered problematic for crops. Organic carbon ranged from 7.00 to 13.60 grams per kilogram, a medium-to-high level the authors attribute to the regular addition of farmyard manure and plant residues by local growers. Available nitrogen averaged 284 kilograms per hectare, phosphorus a modest 12.15 kilograms per hectare, and potassium 280 kilograms per hectare, while sulphur, calcium and magnesium varied considerably, with phosphorus showing the highest relative variability of all measured properties at a coefficient of variation of 38.28 percent.

Correlation analysis revealed the web of interdependencies that governs how these soils behave. Bulk density, a measure of compaction, correlated positively with particle density and available potassium, the latter likely reflecting potassium-bearing minerals such as feldspars and micas that simultaneously increase particle packing and nutrient supply. It correlated strongly and negatively with porosity, as expected, since denser soils have less pore space. Most strikingly, organic carbon correlated strongly and positively with available nitrogen at 0.65, confirming that organic matter is the engine of nitrogen supply in these systems, releasing mineralisable nitrogen in proportion to its abundance. Sulphur tracked positively with pH, consistent with enhanced mineralisation of organic sulphur under near-neutral conditions, while exchangeable calcium and magnesium rose together with indicators of improved fertility and microbial activity.

To compress this multivariate complexity into interpretable structure, the team applied principal component analysis, which extracted six components with eigenvalues above one that together explained 69.43 percent of the total variance. The first component, dominated by bulk density alone, signalled that soil structural condition, shaped by management decisions about tillage and compaction, is the single strongest axis of differentiation across the landscape. The second was driven by organic carbon, reflecting the heavy reliance of smallholders on organic manure as their primary nutrient pathway. Later components captured particle density with phosphorus and potassium, the pH-sulphur-calcium-magnesium cluster tied to base saturation and leaching dynamics, electrical conductivity, and finally nitrogen with porosity. A complementary hierarchical cluster analysis grouped the ninety-nine sampling sites into seven major clusters, whose long branch linkages pointed to substantial multivariate dissimilarity driven by topography, parent material and divergent management environments.

The study’s most innovative contribution came from spatial statistics. Using Global Moran’s I, a measure of spatial autocorrelation computed in the GeoDa software with a six-nearest-neighbour weight matrix and 999 permutations, the researchers tested whether each soil property was randomly distributed or spatially structured. Six properties passed the significance test: pH, electrical conductivity, organic carbon, porosity, nitrogen and sulphur. Positive Moran’s I values for these attributes mean that similar values cluster together on the map, so high-pH fields neighbour high-pH fields and nitrogen-poor patches form coherent zones rather than isolated points. The remaining properties, including bulk density, water holding capacity, phosphorus, potassium, calcium and magnesium, appeared spatially random, a finding the authors link to the region’s complex topography, which creates microclimatic and edaphic variation over short distances, and to individual farm management decisions that override natural gradients.

Local Indicator of Spatial Association analysis then pinpointed exactly where the clusters lie. Sulphur expressed the strongest clustering of any property, with 47 significant sampling sites and 23.23 percent of the study area falling into the high-high category, concentrated toward the north, a pattern the authors attribute to gypsiferous parent material, higher organic matter and fertilisers such as single superphosphate. Electrical conductivity formed high-value clusters in the north, where erratic and low rainfall allows salts to accumulate, while nitrogen showed high clusters in the central and northern regions and low clusters toward the south-west, a patchiness that implicates nutrient management practices as the dominant driver rather than any natural gradient. The team even notes that naturally occurring Acacia catechu trees, which fix atmospheric nitrogen, may act as local nitrogen hotspots. Porosity was the only variable with more low-low than high-high area, with compacted zones in the northern half pointing to the cumulative effect of seasonal tillage and field traffic.

Translated into practice, the cluster typology becomes a management blueprint. Low-low zones, where a property is significantly depressed and surrounded by similarly depleted neighbours, are the highest-priority targets: nitrogen-poor clusters call for increased fertilisation or green manures such as Sesbania or Trifolium, organic-carbon-poor zones for farmyard manure, compost and legume cover crops between cauliflower seasons, and sulphur-deficient patches for sulphur-coated urea and oilseed cakes. Porosity low-low zones indicate compaction that could be remedied with subsoiling or deep tillage combined with organic amendments. High-high clusters, by contrast, may need optimisation to avoid toxicity-like conditions, while high-low outliers, anomalously enriched sites within deficient neighbourhoods, deserve investigation to understand how they arose. To visualise all of this, the team generated continuous fertility surfaces using inverse distance weighting in QGIS with a weighting power of two and a fine spatial resolution of 0.0001 degrees, producing thematic maps of every measured property clipped to the block boundary.

The authors are candid about the limitations: the survey covered a single block, a single season, and only the surface layer, so the findings cannot be generalised to other agro-climatic zones without further work. Even so, the framework is deliberately replicable, and the maps now serve as a baseline against which future soil health monitoring, climate-driven fertility shifts and the effects of changing nutrient regimes can be measured. For a region where farmers manage fragments of land under unpredictable monsoon rainfall, the message is quietly transformative: soil fertility in the lower Himalayas is not a uniform backdrop but a structured, mappable landscape of hotspots and deficits, and knowing where those zones lie is the first step toward feeding crops precisely what they need, where they need it, while protecting fragile mountain soils for the long term.

Subject of Research: Spatial variability and geostatistical mapping of soil physico-chemical properties in vegetable-cultivated Himalayan soils

Article Title: Spatial characterization and geostatistical analysis of soil properties in vegetable cultivated soils of Indora block (Kangra district), Himachal Pradesh, India

Article References: Atwal, A. S., Lal, A., Kapoor, R., Sandal, S. K., Sharma, R., Sepehya, S., & Kumari, P. (2026). Spatial characterization and geostatistical analysis of soil properties in vegetable cultivated soils of Indora block (Kangra district), Himachal Pradesh, India. Discover Soil, 3(1), Article 116. https://doi.org/10.1007/s44378-026-00271-4

Image Credits: AI Generated

DOI: 10.1007/s44378-026-00271-4

Keywords: soil science, spatial variability, geostatistics, Moran's I, LISA, precision agriculture, Himalayan farming, cauliflower, soil fertility, nutrient management, Himachal Pradesh, organic carbon

Ribeiro’s scientific foundation was laid in geostatistics, the branch of applied mathematics developed originally for mining estimation and later adopted across the environmental sciences. His 1992 doctoral thesis at Instituto Superior Técnico, written within the framework of mining engineering, was a substantial work of more than four hundred pages devoted to the geostatistical characterisation of hydrogeological systems, including contributions to structural model inference under nonstationary geostatistics. This was a technically demanding area: nonstationary models allow the statistical structure of a field, such as a groundwater level surface, to vary across space rather than assuming uniform behaviour, which is far closer to how real aquifers behave. As early as 1989, Ribeiro had co-authored work with F. Muge on a geostatistical approach to modelling a piezometric field, presented in the influential Geostatistics book series, demonstrating how kriging-based methods could be brought to bear on the spatial structure of groundwater heads.

That quantitative grounding soon produced practical tools that spread well beyond Portugal. Perhaps the best known is the Susceptibility Index, a method Ribeiro developed to assess the vulnerability of aquifers to diffuse agricultural pollution, first applied in the Alentejo region of southern Portugal and published in the proceedings of the 2001 Future Groundwater Resources at Risk conference. Vulnerability mapping itself has a distinguished lineage, traceable to the French work of Margat in 1968 and Albinet and Margat in 1970, and later systematised in the American DRASTIC scheme of 1987. Ribeiro’s contribution was to adapt the concept to the specific dynamics of agricultural contamination, where nitrate leaching from fertilisers and irrigation return flows threatens shallow aquifers. The method was subsequently applied in strikingly different settings, including the Daule aquifer in Ecuador, where Ribeiro and colleagues published a vulnerability assessment in the Science of the Total Environment in 2017, showing how a Portuguese-designed index could be transferred to a tropical coastal aquifer facing salinisation and agrochemical pressure.

The vulnerability work was inseparable from a broader research programme on groundwater quality and monitoring, much of it conducted with Stigter and Manuel João Carvalho Dill. In a pair of papers published in Hydrogeology Journal and the Journal of Hydrology in 2006, the team evaluated intrinsic and specific vulnerability methods against observed salinisation and nitrate contamination in two agricultural regions of southern Portugal, and then developed a groundwater quality index as a communication tool for agro-environmental policy. The point was not merely academic: indices translate complex multivariate chemistry into a form that regulators, farmers and the public can act upon. Later work pushed the statistical sophistication further, with factorial regression models built to explain and predict nitrate concentrations under agricultural land, published in the Journal of Hydrology in 2008, and a 2011 study in Environmental Science & Technology that examined the efficiency of monitoring programmes for nitrate-contaminated groundwater, asking hard questions about whether existing networks actually detect what they are supposed to detect.

Monitoring network design became a speciality in its own right. Together with Luís Nunes and Maria da Conceição Cunha, Ribeiro published a series of papers in the mid-2000s on optimising groundwater monitoring networks through redundancy reduction and on optimal space-time coverage under exploration cost constraints, appearing in the Journal of Water Resources Planning and Management and Environmental Monitoring and Assessment. The underlying problem is one that every environmental agency faces: sensors and sampling campaigns are expensive, yet removing the wrong wells destroys the information content of the entire network. By combining variance-reduction criteria from geostatistics with optimisation algorithms, the team provided a quantitative basis for deciding where each measurement earns its keep. A 2007 paper in Computer-Aided Civil and Infrastructure Engineering compared variance-reduction and space-filling design approaches, clarifying the trade-offs between statistical efficiency and spatial coverage.

Ribeiro’s geostatistical toolkit also reached back toward its origins in reservoir characterisation. A 1990 paper in Mathematical Geology, co-authored with colleagues including Amílcar Soares, applied geostatistical and multivariate data analysis techniques to improving reservoir description, connecting hydrogeology with petroleum geoscience. Multivariate statistics became a recurring thread throughout his career. In Colombia, working with Teresita Betancur-Vargas and her students at the Universidad de Antioquia, principal component analysis was used to untangle the hydrochemistry of the Urabá aquifer, published in the Journal of Geochemical Exploration in 2013, and to identify regional flow systems in the Gulf of Urabá. Similar approaches informed studies of pesticide exposure in the shallow groundwater of the Tagus vulnerable zone in Portugal, published in Environmental Science and Pollution Research in 2012, and of natural background levels assessment combined with indicator kriging for groundwater quality interpretation, published with Italian colleagues in the Science of the Total Environment in 2016.

Climate change entered his research agenda early and stayed there. With Luís Veiga da Cunha, Ramiro Oliveira and João Nascimento, Ribeiro produced some of the first assessments of climate change impacts on groundwater resources in mainland Portugal, published in Portuguese water-sector literature in the mid-2000s and in IAHS proceedings in 2007. The work extended to the Madeira archipelago through the CLIMAAT II project and to the Mediterranean, where Stigter led a comparative assessment of climate change impacts on three coastal aquifers published in Regional Environmental Change in 2014. Further afield, Ribeiro contributed to evaluating piezometric trends using the seasonal Kendall test, a nonparametric method robust to seasonality, in the alluvial aquifers of the Elqui river basin in north-central Chile, published in Hydrological Sciences Journal in 2015, and to analysing how transmissivity uncertainty affects the reliability of models of the Northwestern Sahara Aquifer System, published in the Journal of African Earth Sciences in 2017 with Mounira Zammouri.

One of the most forward-looking strands of his later career concerned groundwater-dependent ecosystems and the living world hidden inside aquifers. Ribeiro co-edited the volume Groundwater and Ecosystems in the International Association of Hydrogeologists’ Selected Papers in Hydrogeology series, published by CRC Press in 2013, and worked with colleagues on groundwater ecosystems and bio-indicators. With Mafalda Shapouri, Luís Cancela da Fonseca and others, he studied the variation of stygofauna, the specialised animals inhabiting groundwater, along a gradient of salinisation risk in a coastal Mediterranean aquifer, published in Hydrology Research in 2016. Estuarine biodiversity was examined as an indicator of groundwater discharge in Estuarine, Coastal and Shelf Science in 2012. These interests anticipated by years the growing international recognition of groundwater as an ecosystem in its own right, a view crystallised in a 2023 Global Change Biology review describing groundwater as a hidden global keystone ecosystem. Ribeiro also contributed methodological work on identifying groundwater-dependent ecosystems in Portugal for the implementation of the European Water Framework Directive, presented at the 44th IAH Congress in Dubrovnik in 2017.

What truly set Ribeiro apart, and what gives the tribute its unusual warmth, is the cultural dimension of his legacy. He was a lifelong student of the relationship between water and music, tracing how composers from Handel to Toru Takemitsu drew inspiration from water in its many forms. He presented a radio programme on Antena 2 in Portugal devoted to Shostakovich, created the Musicágua project exploring water as a source of inspiration in musical creation, and in 2021 published a chapter on music inspired by groundwater and other components of the hydrological cycle in a Springer volume on geoethics and groundwater management. In the same volume he revisited ancestral groundwater techniques, arguing that traditional collection and distribution systems deserve renewed attention as nature-based solutions for modern water management, an argument he illustrated with a study of a public groundwater supply in the São Pedro do Sul municipality based on an ancestral capture system.

The tribute’s authors frame this breadth not as dilettantism but as a coherent humanistic vision: the conviction that groundwater science serves society best when it is quantitative yet humble, technically rigorous yet culturally literate. His career, spanning Portugal, Spain, Morocco, Chile, Colombia, Ecuador and beyond, and his mentorship of students and collaborators across three continents, built a community as much as a publication record. In an era when hydrogeology faces intensifying pressures from climate change, agricultural intensification and ecosystem degradation, the example of a scientist who could design a monitoring network, map an aquifer’s vulnerability, champion the unseen life within groundwater and then explain why a Shostakovich symphony and a spring share the same spirit, stands as a reminder of what the discipline, at its best, can be.

Subject of Research: The interdisciplinary career and legacy of hydrogeologist Luís Ribeiro in groundwater geostatistics, vulnerability assessment and water culture

Article Title: Luís Ribeiro: A visionary mind bridging groundwater science, culture and music through an interdisciplinary and humanistic legacy

Article References: Stigter, T. Y., Nascimento, J., de Melo, M. T. C., & Betancur-Vargas, T. (2026). Luís Ribeiro: A visionary mind bridging groundwater science, culture and music through an interdisciplinary and humanistic legacy. Hydrogeology Journal. https://doi.org/10.1007/s10040-026-03137-9

Image Credits: AI Generated

DOI: 10.1007/s10040-026-03137-9

Keywords: Luís Ribeiro, hydrogeology, geostatistics, groundwater vulnerability, Susceptibility Index, groundwater-dependent ecosystems, stygofauna, climate change, monitoring network design, nitrate contamination, water and music, Portugal

The logic behind a baseline study is deceptively simple but scientifically crucial. Every patch of ground on Earth emits ionizing radiation, driven by naturally occurring radionuclides such as uranium-238, thorium-232, and potassium-40 in soils and building materials, along with a contribution from cosmic rays. When a facility that handles radioactive material begins operating, regulators and the public will inevitably ask whether the environment has changed. Without a rigorous, statistically defensible picture of conditions before operations start, it becomes nearly impossible to distinguish pre-existing geological variability from any future operational influence. The new study addresses that problem head-on by characterizing the ambient gamma-radiation field at the South Cikarang site with unusual methodological care.

The researchers designated nineteen monitoring locations across terrain that is anything but uniform, mixing engineered fill, natural clay, and vegetated ground. Because one designated point, labeled Point B, was physically inaccessible, direct measurements were ultimately obtained at eighteen locations. At each accessible point, the team measured the ambient dose equivalent rate, written H*(10), which is the standard operational quantity for estimating external exposure to penetrating radiation. Crucially, measurements were taken at two heights: at the ground surface and at one meter above ground, the height at which a standing human body would receive most of its external dose. Each detector-and-height configuration was repeated ten times, allowing the team to quantify measurement uncertainty rather than relying on single spot readings.

What makes the study technically interesting is its dual-detector strategy. The team used two fundamentally different radiation detection technologies in parallel: a Geiger-Müller tube, a gas-filled detector prized for ruggedness and simplicity, and a sodium iodide scintillation detector, NaI(Tl), which converts gamma-ray energy into flashes of light and offers higher sensitivity. The two detector types respond differently to the energy spectrum of environmental gamma rays, so agreement between them is a powerful internal check. Across all detector systems and both measurement heights, the ambient dose equivalent rates ranged from 0.030 to 0.120 microsieverts per hour, values comfortably within the range typical of natural background worldwide. The median reduction in dose rate from the surface to one meter height was 11.8 percent for the GM detector and 14.2 percent for the NaI(Tl) system, a pattern consistent with gamma rays originating primarily from radionuclides in the soil itself.

The statistical agreement between the two instruments was striking. Pearson correlation coefficients reached 0.958 at ground level and 0.919 at one meter, indicating that despite their different physical principles, the two detector systems told essentially the same story across the site. In radiation metrology, this kind of cross-validation matters enormously. A single detector can drift, suffer from energy-response artifacts, or be fooled by site-specific conditions. When two independent technologies converge on the same spatial pattern with nearly identical rankings of hot and cold spots, confidence in the resulting baseline rises dramatically. The authors also subjected their data to formal uncertainty analysis, following international guidance from the International Atomic Energy Agency on quantifying uncertainty in nuclear analytical measurements, which strengthens the study’s defensibility in a regulatory context.

Perhaps the most visually compelling part of the work is its use of geostatistics. Rather than treating the eighteen measurement points as isolated numbers, the team applied Ordinary Kriging, a spatial interpolation technique borrowed from mining geology and soil science. Kriging does not simply draw smooth contours between points; it models the underlying spatial structure of the data, using a variogram to describe how quickly radiation values become uncorrelated with distance. The analysis identified meaningful spatial dependence over a range of approximately 100 meters, meaning that measurements taken closer together than that distance carry information about one another, while points farther apart are effectively independent. That length scale is itself a finding: it tells future monitoring programs how densely they need to sample to capture the site’s true variability.

The geostatistical mapping also revealed a clear physical explanation for the radiation pattern. Higher dose rates clustered over areas of engineered sand and crushed-stone fill, materials that typically contain mineral grains enriched in naturally occurring radionuclides, while lower values were generally associated with natural clay-dominated and vegetated ground. This association between construction fill and elevated gamma readings echoes findings from other parts of Indonesia, including previous work by overlapping research teams in the Mamuju region of West Sulawesi, which is known for unusually high natural radiation. In a rapidly industrializing landscape like Cikarang, where ground is routinely reshaped and imported fill is common, understanding that human-modified ground can shift the local radiation field is essential context for any future facility monitoring.

To translate the measurements into human-health terms, the team performed screening-level dose and risk assessments. Annual external dose estimates for a hypothetical person at the site ranged from 0.052 to 0.172 millisieverts per year, and excess lifetime cancer risk estimates ranged from 1.76 times ten to the minus four to 5.86 times ten to the minus four. These figures should be read as conservative screening values rather than predictions of harm; they sit well within the range of doses people receive from natural background radiation everywhere on the planet, which averages a few millisieverts per year globally according to United Nations Scientific Committee assessments. The point of the exercise is not alarm but calibration: by quantifying the risk landscape before the cyclotron is built, the researchers have created a reference line against which any future change can be judged.

Medical cyclotrons occupy a distinctive niche in the radiological world. Unlike nuclear power plants, they accelerate charged particles to bombard targets and produce short-lived isotopes such as fluorine-18 and technetium precursors used in diagnostic imaging. Their routine emissions are modest, but neutron activation, activated components, and gaseous releases during target irradiation mean that regulators, following standards set by Indonesia’s nuclear regulator BAPETEN and international guidance from the IAEA and the International Commission on Radiological Protection, still require careful environmental surveillance. A credible pre-operational baseline is the foundation of that surveillance, and the South Cikarang study demonstrates how to build one with limited resources: replicate measurements, dual detectors, height-resolved sampling, and spatial statistics.

The broader significance of the study lies in its transferability. The authors describe their approach as a framework integrating measurement validation, uncertainty analysis, and spatial modeling for environmental surveillance around planned radiological facilities, and the recipe is deliberately generic. Any country planning a cyclotron, a radioisotope production plant, or even a research reactor could adapt the same workflow, adjusting the sampling density to the spatial correlation scale revealed by its own variogram. As nuclear medicine expands across Southeast Asia and the global south, demand for radiopharmaceuticals is growing faster than the infrastructure to produce them, and with that growth comes a parallel need for environmental science that earns public trust. This study, supported by BRIN and an ITB research grant, with site access provided by the state pharmaceutical company PT Bio Farma, shows that meticulous baseline work can be done before the first beam is ever switched on, ensuring that when questions arise about a facility’s environmental footprint, the answer will rest on data rather than doubt.

Subject of Research: Pre-operational environmental gamma-radiation baseline characterization at a proposed medical cyclotron facility in Indonesia

Article Title: Environmental radiological baseline characterization of a proposed medical cyclotron facility in Indonesia using dual-detector measurements and geostatistical analysis

Article References: Purnama, D. S., Muliawan, D., Santoso, M., Fitriana, R., Nugraha, E. D., Abdullah, A. R. A., & Permana, S. (2026). Environmental radiological baseline characterization of a proposed medical cyclotron facility in Indonesia using dual-detector measurements and geostatistical analysis. Environmental Geochemistry and Health, 48(15), Article 617. https://doi.org/10.1007/s10653-026-03505-0

Image Credits: AI Generated

DOI: 10.1007/s10653-026-03505-0

Keywords: medical cyclotron, radiological baseline, ambient dose equivalent, Geiger-Müller detector, NaI(Tl) scintillator, Ordinary Kriging, geostatistics, environmental monitoring, radiation protection, nuclear medicine, Indonesia, West Java

The stakes are high. Arsenic is a naturally occurring metalloid that becomes toxic to humans at relatively low exposures, and the World Health Organization recognizes it as a major public health concern. In Guizhou, the problem has a notorious history: chronic arsenic poisoning has previously been documented in villages where residents burned coal with exceptionally high arsenic content indoors. When arsenic sits in the topsoil of farmland, it can enter the food chain through crops, be inhaled as dust, or leach into the groundwater that threads through karst aquifers. Mapping its spatial distribution is therefore not an academic exercise but a prerequisite for protecting food safety in one of China’s most geologically distinctive agricultural regions.

The research team, led by Zhizhuo Liu and Lang Zhang with colleagues from Beijing Normal University, the Institute of Geophysical and Geochemical Exploration, and Tianjin Chengjian University, collected agricultural topsoil samples across Zhijin County and measured arsenic concentrations in the laboratory. The results revealed a striking range: values spanned from as low as 1.79 milligrams per kilogram of soil to more than 40 milligrams per kilogram, with a mean of 17.7 milligrams per kilogram. That spread matters, because it means some fields are relatively clean while others approach or exceed thresholds of concern, and the difference between them is not random. Understanding what drives that heterogeneity was the central question of the study.

Before any modeling could begin, the researchers had to confront a statistical challenge common in soil geochemistry: arsenic concentration data are typically skewed, with a long tail of high values that can distort conventional analyses. The team applied a Yeo-Johnson transformation, a flexible mathematical procedure that reshapes the distribution to reduce skewness. The transformation helped, but formal tests still rejected normality of the transformed data, with a p-value of 0.003. This detail is more than statistical housekeeping. It signals that arsenic in these soils is genuinely patchy and structured by underlying processes, rather than varying smoothly and randomly, which shaped the choice of methods that followed.

To characterize the spatial structure, the researchers turned to semivariograms, the workhorse tool of geostatistics. A semivariogram describes how similar soil values are as a function of the distance between sampling points, and fitting a mathematical model to it reveals the scale over which the variable behaves predictably. Among the candidate models, an exponential semivariogram fit best, achieving a coefficient of determination of 0.820. Two parameters stood out. The nugget-to-sill ratio of 0.627 indicated that a substantial fraction of the variation occurs at very short distances or within measurement error, a signature of strong local heterogeneity. The range of 37.02 kilometers showed that arsenic values remain spatially correlated over tens of kilometers, implying that broad regional forces, not just field-scale quirks, shape the pattern.

Using that fitted model, the team produced maps of arsenic across the county with ordinary kriging, a geostatistical interpolation technique that weights nearby observations according to the modeled spatial structure. After back-transforming the predictions to the original concentration scale, the maps revealed relatively high arsenic values concentrated mainly in the northern, northeastern, and central-eastern parts of Zhijin County. These hotspots provide exactly the kind of actionable intelligence that environmental agencies need: instead of monitoring uniformly, regulators can focus verification sampling and agricultural inspections on the zones where the geostatistical model suggests arsenic is most likely to be elevated.

Mapping where arsenic is high is only half the story; the other half is explaining why. For this, the researchers employed Geodetector, a statistical framework designed specifically to quantify how much of the spatial variation of a variable can be explained by a categorical environmental factor. The method computes a q statistic that measures explanatory power, and it can also test whether pairs of factors interact to explain more variation together than either does alone. In the Zhijin analysis, several factors showed statistically significant associations with arsenic patterns, with nominal p-values at or below 0.003: mercury concentration, soil organic carbon, distance to mining sites, elevation, distance to rivers, and lithology, the underlying rock type from which the soils developed.

The single most powerful factor was mercury, with a q statistic of 0.4241, the largest of any individual variable tested. The pairing of mercury and arsenic is geologically meaningful, because both elements are often enriched together by the same mineralization and coal-related geological processes that characterize parts of Guizhou. Even more intriguing was the interaction analysis: the combination of mercury and lithology produced the largest joint q statistic of the study, 0.5030, meaning that rock type and mercury together explained half of the spatial variation in soil arsenic. This suggests that arsenic accumulation in Zhijin is not driven by a single cause but by the interplay of geological substrate and geochemical processes that concentrate multiple potentially toxic elements simultaneously.

To push the explanatory analysis further, the team compared five candidate machine learning regressors for predicting arsenic concentrations from environmental covariates, including satellite-derived and terrain-based variables such as Sentinel-2 imagery, elevation from the Shuttle Radar Topography Mission, and river networks from OpenStreetMap. The winner was XGBRegressor, an implementation of extreme gradient boosting, a tree-based ensemble method that builds many sequential decision trees to capture nonlinear relationships. It achieved a pooled out-of-fold coefficient of determination of 0.4197, a root mean square error of 6.7793 milligrams per kilogram, and a mean absolute error of 4.5831 milligrams per kilogram, outperforming the other four models on every metric.

Crucially, the researchers did not treat the model as a black box. They applied SHAP, or SHapley Additive exPlanations, a technique borrowed from game theory that assigns each input variable a signed contribution to every individual prediction. At the global level, SHAP ranked mercury, elevation, and lithology as the three most influential predictors, and mercury and elevation remained among the top three across all five cross-validation folds, indicating that the ranking was stable rather than an artifact of a particular data split. The team then went a step further and produced sample-level SHAP maps, which visualize how each factor pushes arsenic predictions up or down at specific locations. These maps revealed heterogeneous signed contributions across the county, showing that the same factor can raise predicted arsenic in one area and lower it in another, a nuance that global averages completely obscure.

The authors are careful, and appropriately so, about what these findings do and do not prove. Predictive performance was moderate, with the best model explaining roughly 42 percent of the variance, and spatial transferability to other counties remains unverified. More fundamentally, the study distinguishes between spatial explanatory power, predictive contribution, and causality: the fact that mercury and lithology statistically explain arsenic patterns does not by itself establish a causal mechanism. What the results do support is targeted verification and zoned monitoring. For a region where karst hydrology can rapidly transport contaminants and where millions of people depend on local agriculture, that is a pragmatic and valuable outcome. The study also demonstrates a methodological template, pairing geostatistics, Geodetector, gradient boosting, and SHAP-based interpretability, that other regions with complex geology and legacy mining can adapt to trace the hidden geography of soil contamination.

Subject of Research: Spatial distribution and environmental drivers of arsenic in karst agricultural soils

Article Title: Spatial structure and multiscale environmental associations of arsenic in karst agricultural soils of Zhijin County, southwestern China

Article References: Spatial structure and multiscale environmental associations of arsenic in karst agricultural soils of Zhijin County, southwestern China. (n.d.). https://doi.org/10.1007/s10661-026-15960-4

Image Credits: AI Generated

DOI: 10.1007/s10661-026-15960-4

Keywords: arsenic, karst soils, soil contamination, geostatistics, ordinary kriging, Geodetector, XGBoost, SHAP, heavy metals, Guizhou, agricultural soil, environmental monitoring

The research, led by Aaron Jerome Rampersad of the University of Canterbury with Christianne Marie-Claire Faith Zakour of the Loss and Damage Youth Coalition, addresses a problem that has haunted Caribbean hydrology for decades. Rainfall networks across the region are sparse, records are riddled with gaps, and conventional methods for building design rainfall curves quietly assume data that simply do not exist. The stakes are not abstract. Hurricane Ivan in 2004 damaged or destroyed approximately 89 percent of Grenada’s housing stock, inflicting losses near 900 million US dollars, roughly twice the national GDP. Hurricane Beryl in 2024 caused an estimated 218 million dollars in damage and triggered parametric insurance payouts of 55.6 million dollars. Designing infrastructure against the wrong rainfall statistics has direct, measurable consequences.

The team assembled an archive of 28 rain gauges drawing on records from Grenada’s National Water and Sewerage Authority and the Grenada Airports Authority. Before any analysis, the raw material was daunting: 7,671 observed station-days from the water authority network alongside 14,885 daily values from the airport, with most non-airport stations suffering gaps ranging from isolated days to entire missing years. Only Point Salines, with an approximately continuous 40-year record, approached the completeness that standard frequency analysis assumes. For many stations, the eventual curated dataset guaranteed a minimum of 12 years of usable records, a thin foundation for estimating rainfall quantities associated with 50- or 100-year return periods.

The workflow begins with a diagnostic innovation. Rather than relying on the classical semivariogram, the geostatistical workhorse that bins station pairs by separation distance, the authors computed site-specific Pearson correlations directly between every pair of stations. This approach, adapted from techniques used to study non-stationary spatial correlation in earthquake ground motions, sidesteps a known weakness: with so few stations, lag-bin averaging obscures the very structure the analysis is meant to reveal. The verdict was clear. Daily rainfall dependence in Grenada is governed primarily by how far apart two stations are, with elevation acting as a secondary influence whose effect shifts with the seasons. March, one of the driest months, showed strikingly coherent spatial rainfall, while June, as the Intertropical Convergence Zone migrates northward, produced far more scattered behavior.

Those diagnostics fed directly into the gap-filling stage. The team fitted spatial correlation models to the daily rainfall field using Bayesian inference, implemented in Python with the NumPyro library and the No-U-Turn Sampler, treating model parameters as probability distributions rather than fixed numbers. Ordinary kriging driven by these Bayesian-inferred models then reconstructed missing daily values, and it outperformed a full bench of competitors including inverse distance weighting, radial basis functions, and Gaussian process regression. Adding elevation dissimilarity as a covariate delivered marginal but consistent gains. Strict quality control followed: imputations were only retained when at least six donor stations contributed and the kriging prediction variance stayed below 25 percent of the marginal daily variance, lifting every station’s completeness above 70 percent.

With the reconstructed dataset in hand, the researchers turned to extreme value theory. Instead of the Gumbel distribution used in earlier Grenadian studies, which effectively fixes the shape parameter at zero and can underestimate rare rainfall quantiles, they fitted the full generalized extreme value distribution to annual maxima and the generalized Pareto distribution to peaks over threshold, the latter using declustering to ensure that exceedances separated by less than 24 hours counted only once. Because most stations contributed just 12 years of annual maxima, weakly informative priors were specified through an empirical Bayes-style strategy informed by preliminary analytical fits, stabilizing inference on the tail-shape parameter that controls how heavy the rainfall distribution’s upper end truly is. Markov chain Monte Carlo sampling, with four chains and extended warm-up, produced full posterior distributions for every parameter.

The physical signals that emerged are plausible for a mountainous Caribbean island. The shape parameter correlated moderately with elevation, at 0.25 for the generalized extreme value model and 0.34 for the generalized Pareto model, consistent with orographic enhancement steepening the upper tail of the rainfall distribution where moisture-laden trade winds are forced over the central highlands. Interpolating the posterior return levels across the island required choosing among kriging variants, and intrinsic collocated cokriging with a residual correlogram, which exploits elevation as a secondary variable, won on leave-one-out cross-validation. Five poorly constrained stations with only about four years of reliable data were excluded after sensitivity testing showed they destabilized the fitted spatial structures. The final maps show the highest predicted extremes in the island’s northeast, broadly matching Grenada’s known climatic zoning, though accompanied by appropriately large uncertainty estimates.

The study is unusually candid about its own limitations. A sensitivity analysis traced every annual maximum and threshold exceedance back to its source, classifying each as observed or imputed, and then refitted the models using observed extremes only. Where imputed values dominated a station’s extreme sample, return levels shifted dramatically, with differences approaching 80 percent for some generalized Pareto estimates and exceeding 50 percent for some extreme value estimates; at the 25-year return period, most stations stayed within roughly 25 percent. The authors attribute this partly to the smoothing inherent in kriging, which produces conditional-mean predictions rather than stochastic realizations and can dampen localized extremes. They suggest that future work propagate imputation uncertainty directly, through empirical Bayesian kriging, multiple conditional realizations, or hierarchical models treating missing rainfall as latent quantities.

What elevates the work beyond a single-country case study is its transferability. The complete workflow, from site-specific correlation diagnostics through Bayesian model fitting to island-wide interpolation, is publicly available through GitHub and Zenodo repositories, and the underlying Grenada Daily Rainfall Database has been released on Zenodo. The authors also outline an engineering validation path, proposing two-dimensional flood simulations in flood-prone catchments such as St. John’s and Great River to test whether the estimated rainfall fields produce physically reasonable inundation. For small island developing states facing intensifying hurricanes and rising adaptation costs, the message is straightforward: with Bayesian methods, even a fragmented, decades-old network of rain gauges can yield defensible, spatially explicit design rainfall, provided the uncertainties are confronted rather than hidden.

Subject of Research: Bayesian estimation of multi-station rainfall intensity-duration-frequency curves and extreme daily rainfall mapping in data-limited island settings, applied to Grenada

Article Title: A Bayesian workflow for multi-station IDF curve development in data-limited island settings: application to Grenada

Article References: Rampersad, A. J., & Zakour, C. M.-C. F. (2026). A Bayesian workflow for multi-station IDF curve development in data-limited island settings: application to Grenada. Theoretical and Applied Climatology, 157(10), Article 633. https://doi.org/10.1007/s00704-026-06558-4

Image Credits: AI Generated

DOI: 10.1007/s00704-026-06558-4

Keywords: IDF curves, extreme rainfall, Bayesian inference, Grenada, kriging imputation, generalized extreme value distribution, generalized Pareto distribution, spatial correlation, small island states, geostatistics, return levels, climate risk