Groundwater depletion threatens water security in arid and semi-arid regions, where long monitoring records are often unavailable. Existing forecasting methods either require extensive historical data, ignore spatial correlation, or lack explicit policy linkages. Here, an adaptive ensemble framework is presented that addresses these limitations by combining a hierarchical Bayesian spatio-temporal model (HBARST) with manually tuned Seasonal Autoregressive Integrated Moving Average (SARIMA) models, a combination that, to our knowledge, has not been previously reported to groundwater forecasting in a structurally complex semi-arid aquifer. Ensemble weights are optimized via inverse RMSE. Using 25-year monthly data from 43 observation wells (OWs) in the Kazerun aquifer (Iran), the ensemble achieves a network-wide RMSE of 1.58 m, outperforming automated SARIMA, manual SARIMA alone, and machine learning models (p < 0.01, Diebold-Mariano test). Wavelet analysis reveals dual-porosity behavior with fault-controlled amplification. The framework demonstrates stable performance with reduced training data (10-15 wells, 5-7 years), though these thresholds are aquifer-specific and require local validation for transfer. Scenario analysis shows that a combined abstraction reduction and managed aquifer recharge strategy achieves substantial recovery with a favorable benefit-cost ratio, preserves agricultural jobs, and provides quantitative information that can support SDG 13 implementation. The open-source framework provides a replicable template for operational groundwater management in data-limited regions worldwide. Data are available on Zenodo (10.5281/zenodo.19570759).
Reliable groundwater-level forecasting is critical for water resources management and early warning in regions where monitoring networks are sparse and irregular. However, most machine-learning and time-series approaches assume regularly sampled observations, limiting applicability under realistic monitoring conditions. This study presents a forecasting-oriented hydrogeological digital twin framework designed for irregular multi-well groundwater observations. Measurements were harmonized onto a monthly grid and augmented with missingness-aware features encoding observation availability and staleness. The framework integrates meteorological forcing, SAFER-derived actual evapotranspiration (ETa), GLDAS root-zone soil moisture, and rolling hydroclimatic indicators representing antecedent recharge and wetness conditions. A global long short-term memory (LSTM) neural network with well embeddings was trained using time-ordered validation and leakage-controlled preprocessing. Forecast skill was evaluated using mean absolute error (MAE), root mean squared error (RMSE), and the coefficient of determination (R2). The model achieved stable generalization with training and test RMSE values near 0.42 m, a test MAE of 0.28 m, and a test R2 of 0.88. Per-well evaluation showed robust performance across heterogeneous monitoring conditions, with larger errors concentrated in wells affected by long observation gaps or abrupt regime shifts. Results demonstrate that explicitly accounting for monitoring irregularity and antecedent hydroclimatic conditions improves groundwater forecasting under severe data scarcity. The proposed framework provides a scalable forecasting-oriented digital twin module for operational groundwater monitoring in data-limited settings.
Microplastic (MP) pollution is a global environmental concern. Recent works highlighted the need to consider even natural and regenerated microfibres (MFs). Despite their ecological and hydrological importance, karst systems remain understudied. Being open systems, karst underground environments are particularly vulnerable to contamination through the connections with surface habitats. For the first time, this preliminary research investigated the presence of MPs and other anthropogenic microparticles (AMPs), such as MFs, in cave dripping waters, highlighting the indirect impact of surface human activity. AMPs were counted and characterized via microscopy and spectroscopy. The 80.5% of the analyzed AMPs were cellulosic and 81.6% fibres. MPs ranged from 0 to 23 MP/L, confirming the presence of polyethylene terephthalate (PET) and polyurethane (PU), and were all fragments, except for one PET fibre. Microparticles < 1 mm were the most abundant (89.3%), of which the most had size between 0.49 and 0.10 mm (51.5%). Pollution sources in cave dripping waters could be linked to landfills in sinkholes, garbage, nearby roads, railways, and human activities. These results highlight MPs and other AMP pollution exists in cave dripping waters, impacting water resources, habitats and species. More comprehensive and multidisciplinary monitoring are needed, taking into account surface and subterranean connections in karst areas, aiming at conservation purposes. Promoting collaborations with show caves and cavers can be instrumental in pilot studies, and acquiring new and valuable data.
Extreme hydroclimatic events, such as hurricanes and monsoon-season rainfall, can dramatically alter groundwater recharge patterns and mobilize contaminants, but the mechanisms that distinguish short-term from sustained impacts on aquifer chemistry remain poorly understood. This study investigated how flooding and prolonged wet-season recharge generate distinct contaminant-mobilization regimes in a sandy floodplain aquifer in South Texas. Groundwater and surface-water quality monitoring occurred across three hydroclimatic periods: post-hurricane flooding, dry season, and the summer monsoon. Field measurements were integrated with geochemical and statistical modeling to evaluate hydrologic and geochemical controls on latent controls on contaminant behavior under contrasting recharge conditions. Results show that flooding triggered rapid, transport-dominated spikes in ammonium, Zn2+, and fecal indicators (total coliform, E. coli), whereas the monsoon period was characterized by elevated Pb2+ and phosphate, consistent with redox-driven internal processes and sustained hydrologic connectivity. Modeling results further indicated that contaminant dynamics were governed more strongly by recharge persistence, redox evolution, and nutrient interactions than by total rainfall alone. Notably, surface-water–groundwater exchange likely continued well beyond the flood event, challenging the assumption that flood-driven contamination is a short-lived phenomenon. Finally, the co-occurrence of heavy metals, fecal markers, and phosphate in drinking-water flood plain aquifers indicates that exposure risks persist across seasons in infrastructure-limited communities. These findings highlight the need for long-term groundwater monitoring frameworks that account for both rapid event-scale contamination and prolonged wet-season geochemical evolution.
This study presents integrated isotope-geochemical data on groundwater from the southern part of Western Siberia to identify the factors of uranium mobilization. Groundwater within the multi-aged, complex rock aquifer system is predominantly of HCO3–Mg–Ca type, whereas waters hosted in granitoid units (the Priobsky and Barlak complexes) exhibit SO4–HCO3–Na–Mg–Ca characteristics. Measured 222Rn activity ranges from 1 to 1161 Bq/L. This study systematically evaluated the disequilibrium state of uranium isotopes (234U/238U) across contrasting hydrogeological settings: granitoid massifs, contact zones, and sedimentary complexes of different ages. Maximum 234U excess occurs in granitoid-hosted groundwater, confirming the Priobsky and Barlak complexes as the primary uranium sources in the region. Stable isotope signatures (δ2H, δ18O) confirm the meteoric origin of the groundwater, with recharge of the Paleozoic bedrock aquifer complex being unaffected by short-term seasonal variations. Thermodynamic modeling classifies all studied waters as a single geochemical type: siliceous-Na (Ca–Mg–K). Elevated uranium concentrations are directly linked to granitoid lithologies hosting accessory minerals (e.g., uraninite, monazite, zircon, bastnäsite) that accumulate uranium, thorium, and rare earth elements. Radionuclides are released into groundwater as a result of mineral weathering. Under oxidizing conditions, uranium remains mobile in the form of uranyl ion (UO22+), whereas thorium is immobilized via adsorption onto mineral surfaces or precipitation. Subsequent uranium precipitation at redox boundaries may contribute to the formation of paleovalley-type infiltration uranium deposits. These findings advance the understanding of uranium mobilization pathways and support targeted assessments of groundwater safety and mineral potential in crystalline terrains.
A multi-faceted investigation employing hydrogeochemical, trace-element, and stable isotopic techniques has been conducted in abandoned mines of the Northern Coalfield Limited, Singrauli, India, to identify the contaminant sources and ionic mobilization pathways, to unravel the geochemical processes responsible for Acid Mine Drainage (AMD) formation, and to investigate the hydrogeochemistry of the surrounding groundwater. AMD pits are characterized by highly acidic conditions and high total dissolve solids (TDS), whereas groundwater samples have neutral pH and TDS within the permissible limits. Groundwater was dominated by the Ca2+-HCO3- hydrochemical facies, whereas AMD samples exhibited Ca2+-SO42--Cl- facies, reflecting sulphide oxidation and mineral dissolution from host rock and mine tailings. Stable isotope (δ18O-δ2H) results indicate that most groundwater samples align along the local meteoric water line (LMWL), suggesting precipitation as the primary source of recharge. However, a subset of groundwater samples from the biotite schist terrain deviates slightly below the LMWL, reflecting evaporative enrichment likely driven by slow infiltration and prolonged residence time within the aquifer. In contrast, AMD samples show a more distinct deviation below the LMWL, attributed to enhanced evaporation. Trace metals show elevated concentration, with aluminum exceeding the permissible limit in all groundwater samples and 75% of AMD samples, while nickel exceeds the permissible limits in 25% of AMD samples, indicating dissolution of aluminosilicate and oxidation of sulphide minerals, respectively. Eighty percent of groundwater samples show fluoride concentrations exceeded the permissible drinking limits, whereas, WAWQI revealed that 60% of groundwater samples are unsuitable for drinking purposes (Grade E). Overall, the study demonstrates that geogenic processes together with localized anthropogenic activities influence the sources and ionic mobilization of contaminants in groundwater and AMD through distinct hydrogeochemical pathways. The elevated metal concentrations in the AMD water pose a significant risk to the surrounding environment and associated ecosystems, thereby necessitating appropriate mitigation and management measures.
In semi-arid regions, groundwater quality exhibits complex nonlinear behavior and strong seasonal variability. Accurate prediction is therefore essential for sustainable management. This study proposes an explainable hybrid framework that integrates XGBoost with three advanced metaheuristic optimization algorithms: the Honey Badger Algorithm (HBA), Marine Predators Algorithm (MBA), and Coati Optimization Algorithm (CAO). The framework predicts Drinking Groundwater Quality Index (DGWQI) using 104 samples from 52 monitoring wells in the Melur aquifer, during both the pre-monsoon (PRM) and post-monsoon (POM) seasons in a semi-arid region, Tamil Nadu, India. The optimized models were systematically evaluated for predictive accuracy, convergence behavior, computational efficiency, generalization capability, robustness, explainability, and spatial prediction reliability. Among the evaluated models, XGB-HBA achieved the best predictive performance for the PRM dataset (R2 = 0.990, RMSE = 3.794, and MAE = 3.171). In contrast, XGB-MBA produced the best result for the POM dataset (R2 = 0.9773, RMSE = 5.40, and MAE = 3.828). SHAP analysis revealed distinct seasonal control on groundwater quality. TDS was the dominant predictor during the PRM, whereas groundwater quality in the POM season was influenced by a broader combination of hydrogeochemical variables associated with recharge processes. Groundwater quality maps consistently identified both persistent hotspots and seasonal spatial variations.
Solute breakthrough curves serve as fundamental diagnostic metrics for characterizing mass transport mechanisms in hydrogeological systems. Traditional interpretations relying exclusively on boundary outlet sampling and deterministic inversion inherently suffer from severe mathematical ill-posedness and equifinality. Resolving these structural limitations necessitates a methodological transition toward high-dimensional internal sensing and probabilistic parameterization. This review presents a critical synthesis of how advanced observational constraints and scientific machine learning are fundamentally reshaping solute transport modeling. We evaluate the progression from classical iterative algorithms to ensemble-based data assimilation frameworks capable of reconstructing complex spatial heterogeneity. Furthermore, we critically assess the operational trade-offs of modern artificial intelligence paradigms. This evaluation analyzes the capacity of probabilistic generative surrogates to ensure robust uncertainty quantification and examines the role of physics-informed neural networks in enforcing mass conservation, while explicitly addressing their optimization instabilities when deployed on noisy field datasets. To organize model and algorithm selection, we develop a structured Complexity–Model–Data framework that relates diagnostic evidence for process complexity to candidate transport formulations and the observational support available for inversion. Dimensionless indicators and BTC morphology are used to narrow the set of plausible models, whereas data density, computational demand, and uncertainty requirements inform the selection of deterministic, probabilistic, data-assimilation, or physics-informed methods. A one-dimensional synthetic case illustrates how this reasoning can be implemented under controlled conditions. The framework provides a practical basis for transparent, site-specific model evaluation while preserving the roles of hydrogeological interpretation, measurement quality, management objectives, and expert judgment.
Declining water availability and increasing water demand highlight the need to improve water supply security and adopt sustainable water management approaches involving alternative water resources. Managed aquifer recharge (MAR) uses alternative water resources, enhancing groundwater storage while ensuring its quality. This study presents results from a field-scale soil-aquifer treatment managed aquifer recharge (SAT-MAR) system, representing the first application in Portugal using basins to infiltrate treated wastewater from the Comporta Wastewater Treatment Plant (WWTP). By comparing the performance of different infiltration basins, the study verified infiltration water quality improvement during vadose zone passage and evaluated groundwater deterioration risks. The system performance was examined across three infiltration basins (B). B1 and B2 were ameliorated with reactive organic layers (ROLs), containing a mixture of sand, 30% biochar and 15% granular activated carbon (GAC), respectively, compared to B4 with plain sand. The ROLs, designed as nature-based and energy-free, allowed a sustainable tertiary treatment for alternative water resources. GAC-ROL achieved superior pharmaceutical compounds (PhCs) removal (77%), compared to B1 (15%) and B4 (12%), reducing hydrophobic and polar compounds below quantification limits, including hydrochlorothiazide and sulfamethoxazole. Compound-specific adsorption and biodegradation patterns, characterized by pKa and logKow, confirmed the higher retention capacity of GAC-ROL. The receiving aquifer showed ∼33% reduction of organic matter indicators (TOC, COD, UV254). Median stable pH (∼6.2) and redox (∼87 mV), indicated system buffering capabilities. Nutrients surpassed the groundwater (APA, 2023) regulatory thresholds in both vadose zone and groundwater, emphasizing the need for enhanced optimization at the WWTP. This study provides strong evidence that integrating ROLs into SAT-MAR systems enhances performance and improves TWW quality.
Qarhan Salt Lake (QSL) in the Qaidam Basin is China’s largest source of potassium fertilizer. However, several decades of intensive subsurface brine extraction have triggered severe groundwater depletion, large-scale cones of depression, and disruption of the regional water-salt equilibrium, threatening the sustainability of this resource. This study aimed to resolve the scientifically and practically significant issue of determining a sustainable subsurface brine extraction volume that reconciles potassium production targets with hydrological recovery and environmental constraints. By employing a refined MODFLOW-based numerical model that incorporates actual extraction and recharge dynamics, we simulated five subsurface brine extraction scenarios (4.0 to 6.0×108 m3/a) over a 20-year period. Results demonstrate that an annual brine extraction volume between 4.5 and 5.0×108 m3/a supports stable production of over 5 million tons of potassium fertilizer while providing key ecological-hydrogeological benefits: an average water level rise of 0.5–0.6 m, a reduction of the area of the cone of depression by 35–75 km2, a consistent positive water balance of 0.4–0.5×108 m3/a, and minimized evaporation losses. These findings offer a validated, multi-criteria framework for sustainable brine mining in QSL and present a transferable model for managing hypersaline resources in water-limited regions worldwide.
Groundwater quality and hydrogeochemical controls were assessed using 145 samples from Birimian crystalline basement and Voltaian sedimentary aquifers in central Ghana. An integrated approach combining hydrochemical facies classification, Gibbs diagrams, bivariate ion analysis, PHREEQC saturation modelling, principal component analysis (PCA), and multiple water quality indices was applied. The study indicates that Ca2+-HCO3- facies predominates, implying that groundwater chemistry is primarily governed by water-rock interactions. Saturation index modelling confirms widespread undersaturation of primary silicate and sulphate minerals, while kaolinite and K-mica represent stable secondary weathering products. PCA identifies four principal processes: (i) a weathering-mineralization continuum, (ii) fluoride mobilization via ion exchange, (iii) geogenic Fe-Mn enrichment under reducing conditions, particularly in Voltaian aquifers, and (iv) localized anthropogenic contamination associated with agricultural inputs. Nitrate concentrations exceeding regulatory limits suggest the infiltration of agricultural leachates into the groundwater system. Water quality assessment indicates that most samples are suitable for drinking, with strong agreement across WQI, WPI, and PIG indices. Irrigation assessment shows predominantly low sodium hazard, although localized magnesium enrichment may pose long-term soil risks. These findings provide a robust hydrogeochemical baseline for central Ghana and demonstrate a transferable multi-index assessment framework for groundwater evaluation in similar tropical aquifer systems.
Iron and manganese frequently co-occur in groundwater and are commonly removed by biological filtration. While long-term Mn removal is predominantly driven by abiotic oxidation on birnessite-coated media, the impact of iron on this catalytic system remains poorly understood. This study investigated the structural, mineralogical, and microbiological organization of a full-scale groundwater biofilter treating Fe– and Mn-bearing water seven years after start-up. Scanning electron microscopy revealed a Fe-rich surface overlayer at the top of the filter, overlying a MnOx coating present throughout the bed. Fe K-edge EXAFS analysis demonstrated that this poorly crystalline Fe phase is ferrihydrite-like and exhibits short Fe–Mn distances (∼2.88 Å), consistent with close structural association with underlying birnessite rather than isolated ferrihydrite precipitation. Despite substantial Fe loading, MnOx remained predominantly abiotic turbostratic birnessite (AOS 3.61–3.67). Biofilm analyses showed increased metabolic activity and dominance of filamentous bacteria affiliated with the genus Crenothrix compared to a Mn-only filter media. Biokinetic assays performed on the extracted biofilm revealed similar Mn(II) oxidation rates (k = 1.2 h−1) but a 25-fold reduction in biological Mn oxidation capacity relative to a Mn-only biofilter, indicating a functional shift toward Fe removal rather than Mn accumulation. These findings demonstrate that iron does not merely compete with manganese but restructures the mineral and microbial architecture of the biofilter, altering its long-term functional trajectory. Spatial separation of Fe and Mn removal may help preserve Mn polishing capacity and biofilter stability in groundwater treatment systems.
Understanding the groundwater-ecosystems-surface water relationships at an aquifer scale is crucial for groundwater-supported ecosystems’ incorporation in resource management. This phenomenon is under-researched in data-scarce southern Africa, especially in transboundary aquifer settings. To investigate this interrelationship, this study was initiated in the semi-arid South African portion of the Tuli-Karoo transboundary aquifer, shared with Botswana and Zimbabwe. Water Table Fluctuation (WTF) and Chloride Mass Balance (CMB) methods were employed for recharge; Environmental isotopes (δ 2H, δ18O, and 3H) analysis for recharge mechanism; and baseflow algorithms for groundwater contribution to baseflow. The estimated alluvial aquifer average recharge ranges from 17.3 to 43.9 mm/a (5 – 12.9% of MAP), while for the sandstone aquifer it ranges from 0.4 to 8.2 mm/a (0.12 – 2.5% of MAP). Local precipitation is the source of post-evaporation recharge. Isotope analysis indicated bi-directional groundwater-surface water interactions. Baseflow average is 3.4 m3/s with a low baseflow index of 7.6%. On average, groundwater contributes 0.39 m3/s and 0.22 m3/s during wet and dry periods, respectively. Additionally, groundwater levels were observed to be declining gradually, which could negatively impact groundwater component of baseflow, and riparian and wetland vegetation. It can be concluded that rainfall seasonal fluctuations control groundwater-ecosystems-surface water interactions. These aquifer-specific findings are crucial for comprehensive implementation of integrated water resource management that is inclusive of associated ecosystems and their protection.
Groundwater systems are increasingly threatened by co-existing heavy metals and microplastics, which interact at molecular and geochemical levels, leading to complex environmental risks. This review explores emerging biohybrid remediation strategies that integrate microbial engineering, advanced materials, and catalytic interfaces to enable simultaneous contaminant removal. We discuss molecular interactions between microplastics and metal ions, microbial functionalization for selective sorption and transformation, and the application of materials such as metal-organic frameworks, redox-active polymers, and engineered biochar. Special emphasis is placed on engineered bio-hybrid interfaces, including catalytic membranes and living barriers, for in situ groundwater remediation. However, the transformation pathways, such as polymer degradation, metal precipitation, and geochemical stabilization, are also examined. Eventually, we highlight future directions toward integrated, scalable, and sustainable groundwater remediation strategies based on molecular design and environmental systems engineering.
Water quality (WQ) assessment increasingly relies on statistical analysis to summarize complex hydrochemical datasets, identify dominant processes, and support interpretation of contamination sources. Yet statistical choices are often weakly justified, assumption checks are incompletely reported, and recurring concerns persist about the reproducibility gap in analytical workflows. This critical review reframes statistical analysis in WQ research as a defensible decision pathway rather than a list of techniques. A structured qualitative review of peer-reviewed literature published between 2000 and 2025, identified through Web of Science, Google Scholar, and SciSpace, examined exploratory data analysis, normality assessment, data transformation, and univariate, bivariate, and multivariate methods. These methods were widely used, often in combination, but transparent workflows linking data characteristics, analytical assumptions, and method selection were rarely reported. Building on these findings, this review proposes a conceptual decision-support framework that integrates pre-analysis, method selection, and interpretation. Four framework-guided sensitivity demonstrations illustrate the effects of correlation-method selection, scaling before PCA, transparency in HCA, and censored-data treatment. The framework emphasizes alignment between statistical choices and data structure, sample size, study objectives, and interpretive limits, while embedding checkpoints for diagnostics, preprocessing, uncertainty, and robustness assessment. Because statistical analysis alone cannot fully capture hydrochemical complexity or unequivocally identify contamination sources, the framework should be used alongside hydrogeochemical interpretation, isotopic evidence, spatial analysis, and, where appropriate, machine-learning approaches. Overall, the review provides a critical synthesis and a structured guidance tool with potential to improve defensible WQ assessment, while acknowledging that further empirical testing is required across contrasting monitoring contexts.
Saturated soil hydraulic conductivity (Ksat) is a critical parameter governing water infiltration, groundwater recharge, and land–atmosphere interactions. However, comprehensive continental-scale Ksat data are limited due to marked spatial variability and insufficient field campaigns at continental scale. This study aims to generate the first continent-wide, machine learning-based Ksat dataset for South America, accompanied by independent hydrological validation. Utilizing an ensemble machine learning framework trained on a newly compiled and harmonized dataset of 778 field observations—the most extensive to date for the continent—and integrating environmental covariates selected for their mechanistic relevance to infiltration processes, we systematically assessed three strategies for outlier management: (A) no noise reduction, (B) basic univariate thresholding (>6 m day−1), and (C) a robust multivariate approach using the Minimum Covariance Determinant (MCD). The comparison showed that strategy B adversely affected model performance by excluding hydrologically significant high-conductivity values, whereas strategy C retained the natural range of Ksat while discarding statistically inconsistent data. Among four machine learning algorithms evaluated (Random Forest, SVM, GLMnet, XGBoost), XGBoost applied to strategy C delivered optimal results (RMSE = 0.62; R2 = 0.57), explaining a substantial proportion of the variance in Ksat. The resulting Ksat map demonstrates pronounced spatial gradients, with higher conductivities in humid, deeply weathered soils of the Amazon and Cerrado, and lower values in the Andes and semi-arid northeastern Brazil. Beyond statistical validation, hydrological relevance was established through positive, albeit region-dependent, correspondence with the Baseflow Index (BFI) across multiple watersheds. The produced dataset effectively captures broad regional trends for hydrological modeling purposes across South America, although its application at finer spatial scales requires caution due to smoothing of small-scale variability. Prediction confidence varies spatially, being higher in well-sampled regions such as southeastern Brazil, and lower in data-scarce areas (Amazon, Andes, Patagonia), where extrapolation beyond the training set introduces greater uncertainty.
Karst aquifers represent the principal mountain ranges and source of drinkable water of the Apennines of southern Italy, sustaining the economic and social development. The assessment of current and future availability of groundwater resources of these aquifers is still challenging due to the lack of precipitation measurements at high altitudes which affects reliable estimations. At this scope, a twofold use of data obtained by General Circulation Models and Regional Climate Models combinations is proposed: (i) to estimate empirical annual groundwater recharge coefficient and (ii) to project groundwater recharge trends under future climate scenarios. This study is focused on the 40 principal karst aquifers located in the Apennines mountain range of southern Italy and uses an ensemble of 15 model members from the EURO-CORDEX project. Models time series were analyzed for both the historical period (1950–2005) and future scenarios (2006–2100) under the representative concentration pathways 4.5 and 8.5, across three consecutive decades: short term (2011–2040), medium term (2041–2070) and long term (2071–2100).Results indicate that annual groundwater recharge coefficients refined estimations range between 59% and 93%, with the model outputs validated against observed groundwater outflow measurements. Future projections show a groundwater recharge long-term gradual decline, with the most severe decrease, exceeding approximately 40% under the more severe scenario and 15% under the less severe one. These findings shed light on the vulnerability of karst aquifers of the Apennines of southern Italy to climate change effects, raising awareness about the development of adaptive groundwater management strategies.
This study examines groundwater residence time and seasonal contamination in shallow wells across the floodplain and adjacent uplands of Antananarivo, Madagascar, to clarify how the river–aquifer interactions and rapid recharge affect drinking water quality. The area combines a flood-prone alluvial plain with informal settlements and on-site sanitation, conditions that place pressure on groundwater quality. Groundwater and river water from 35 sites were sampled in three campaigns covering different seasonal conditions, namely the end of the rainy (April 2022) season, the beginning of the rainy season (January 2023), and the dry season (September 2023). Samples were analyzed for various geochemical parameters, hydrocarbons and microbiological indicators, together with residence time tracers (δ2H, δ18O, 3H, 14C, CFCs and SF6). Multi-tracer groundwater residence time interpretation shows that most wells contain young water with mean residence times (MRTs) below 30 years, mixed with an older component in some cases. In contrast, a few wells have MRTs of several decades. Seasonal and spatial differences in contaminants were evaluated with multivariate statistics comparing floodplain wells, upland wells and rivers. Seasonal contrasts indicate that nitrate is persistent and hydrocarbons and fecal bacteria tend to peak at the onset of the rainy season, in a pattern consistent with intense recharge and flooding potentially mobilizing contaminants from the land surface and facilitating intrusion into poorly protected wells, while concentrations decline during the dry season. Floodplain wells do not consistently display more isotopic or chemical similarity to rivers than upland wells, suggesting that large-scale riverbank filtration or floodwater intrusion is unlikely to represent the dominant control of groundwater composition in these riverine wells. Instead, the combined isotopic, chemical and statistical evidence points to rapid recharge and localized wellhead intrusion of stormwater as likely key pathways for contaminant transport to wells.
