Simple Water Tests Could Flag Uranium in Punjab’s Groundwater Before It Reaches the Tap

In the agricultural heartland of Punjab, India, an invisible threat has been seeping through aquifers for years: uranium, dissolved in the very groundwater that millions of people rely on for drinking and irrigation. Testing for uranium has traditionally required specialized laboratory equipment, trained personnel, and significant expense, which means many villages and private wells have simply never been checked. A new study published in Environmental Monitoring and Assessment offers a strikingly practical alternative. Researchers at Central University of Punjab have developed a set of predictive equations that can estimate whether a water sample is likely to exceed uranium safety thresholds using nothing more than routine hydrochemical measurements—parameters such as total dissolved solids, sodium, potassium, magnesium, bicarbonate, nitrate, and pH—that are already collected in most standard water quality surveys.

The research team, led by Anjali Kerketta together with Harmanpreet Singh Kapoor and Prafulla Kumar Sahoo, set out to close a persistent gap in the region’s water safety infrastructure. While previous investigations have thoroughly mapped the extent and geochemical drivers of uranium contamination across Punjab, there has been no simple, cost-effective method for early detection at the local scale. The new approach is designed precisely for that purpose: to allow water managers, health officials, and field technicians to flag potentially dangerous wells without waiting for expensive radiometric or mass-spectrometric analysis. The work also aligns directly with United Nations Sustainable Development Goal 6, which calls for safe and affordable drinking water for all.

To build their predictive tools, the researchers employed two complementary statistical frameworks. The first, a Generalized Additive Model, or GAM, was used to model uranium as a continuous variable, capturing nonlinear relationships between uranium concentrations and the chemical signature of the water. The model performed remarkably well, achieving an adjusted R-squared of 0.73 and explaining 75.7 percent of the deviance in the data. Within this framework, three interacting pairs of parameters emerged as statistically significant: total dissolved solids paired with bicarbonate, total dissolved solids paired with sodium, and bicarbonate paired with sodium. These interactions reveal that uranium enrichment in Punjab’s groundwater is not driven by any single factor but by the combined chemical conditions that prevail in the aquifer.

The underlying geochemistry tells a coherent story. The study found that uranium concentrations in groundwater rose sharply when bicarbonate and sodium levels climbed above roughly 400 to 600 milligrams per liter, and when total dissolved solids reached the range of 1,000 to 1,500 milligrams per liter. In practical terms, this means that the saltier and more carbonate-rich the water, the more uranium it tends to carry. The researchers attribute this pattern to three interlocking mechanisms. Greater ionic strength in solution enhances the desorption of uranium from mineral surfaces, releasing it into the water. Excess bicarbonate promotes the formation of soluble uranyl-carbonate complexes, which keep uranium dissolved and mobile rather than locked into the aquifer matrix. And elevated sodium concentrations further stabilize these complexes in solution, allowing uranium to travel with the groundwater over considerable distances.

For the categorical side of the analysis, the team used stepwise logistic regression to classify water samples into risk categories. They defined two thresholds that matter for public health: 30 micrograms per liter, the limit adopted by India’s Atomic Energy Regulatory Board, and 60 micrograms per liter, the stricter guideline set by the World Health Organization. Samples were labeled Class 1 if they exceeded a threshold and Class 0 if they fell at or below it. The stepwise logistic regression models performed exceptionally well on both tasks, achieving classification accuracies of 94 percent for the 30-microgram threshold and 96 percent for the 60-microgram threshold. These are striking figures for a method that relies entirely on parameters measurable with conventional field and laboratory chemistry.

The end products of the study are two compact predictive equations that any trained technician can apply. For the 30-microgram-per-liter threshold, the probability of exceedance is computed from a logistic function combining the logarithms of total dissolved solids, sodium, potassium, and magnesium, together with pH. For the 60-microgram threshold, the equation instead incorporates total dissolved solids, sodium, nitrate, and pH. The appearance of nitrate in the second equation is chemically meaningful: nitrate acts as an oxidizing agent in aquifers, and previous research has shown that nitrate-dependent mobilization can drive uranium off sediment surfaces and into solution. The prominence of pH in both equations reflects the fact that alkaline conditions favor uranium desorption and carbonate complexation, while acidic waters tend to immobilize it.

Crucially, the researchers did not stop at fitting models to their original dataset. They cross-validated the equations against an independent dataset that the models had never seen, and the equations still achieved accuracies above 80 percent. This external validation is the gold standard for demonstrating that a predictive tool generalizes beyond the specific wells used to train it. It suggests the equations can be deployed with reasonable confidence at unmonitored locations across Punjab, providing a first-line screening capability that can prioritize which sites genuinely need full uranium testing. In a state with thousands of villages and countless private wells, such triage could dramatically reduce the cost and time required to map contamination risk.

The health stakes in Punjab are considerable. Elevated uranium in drinking water has been associated with increasing incidence of kidney ailments and other health problems in the region, and multiple prior studies have documented uranium hotspots across the state’s alluvial aquifers, particularly in the southwest. Uranium is both a chemical toxin, primarily targeting the renal system, and a radioactive element, adding a radiological dose burden for chronic consumers. Because the contamination is geogenic in origin—mobilized naturally from sediments rather than introduced by a single industrial source—there is no simple point of intervention. Early identification of affected supplies is therefore one of the most effective available public health strategies, enabling authorities to redirect communities to safer sources or install treatment before chronic exposure accumulates.

The broader significance of this work extends well beyond Punjab. Across South Asia and in many arid and semi-arid regions worldwide, groundwater uranium contamination is an emerging concern, and the same hydrochemical logic—high ionic strength, carbonate-rich alkaline water, oxidizing conditions—applies in other alluvial basins. The Punjab equations themselves are calibrated to local conditions and should not be transplanted wholesale to other aquifers without recalibration, but the methodological template is portable. Any region with a baseline dataset of paired uranium and routine hydrochemical measurements can, in principle, develop similar screening equations. The approach echoes successful precedents in groundwater science, where logistic regression and related statistical models have been used to predict arsenic, fluoride, and nitrate contamination risk across large territories.

What makes this study especially compelling is its economy. Rather than demanding new instrumentation or exotic data, it extracts maximum predictive value from measurements that water agencies already collect as a matter of routine. A field survey that records pH, total dissolved solids, and major ions can now, with a pocket calculator or a simple spreadsheet, generate an immediate probability estimate for uranium exceedance. That transforms uranium monitoring from a reactive, laboratory-bound exercise into a proactive screening program that can reach remote and underserved communities. For the farmers and families of Punjab who draw their water from the Indus basin aquifers, the study offers something tangible: a faster, cheaper early warning system for one of the region’s most persistent water quality threats, and a model that other contaminated regions can adapt in the fight for safe drinking water.

Subject of Research: Predictive modeling of uranium contamination in Punjab groundwater using routine hydrochemical parameters

Article Title: Development of predictive equations for the early identification of uranium contamination using basic hydrochemical parameters in the groundwater of Punjab, India

Article References: Kerketta, A., Kapoor, H. S., & Sahoo, P. K. (2026). Development of predictive equations for the early identification of uranium contamination using basic hydrochemical parameters in the groundwater of Punjab, India. Environmental Monitoring and Assessment, 198(11), Article 1161. https://doi.org/10.1007/s10661-026-15974-y

Image Credits: AI Generated

DOI: 10.1007/s10661-026-15974-y

Keywords: uranium, groundwater, Punjab, predictive equations, logistic regression, generalized additive model, hydrochemistry, water quality, drinking water safety, uranyl-carbonate complexation, India, SDG 6