Aquifer Storage and Recovery (ASR) is a managed aquifer recharge method where water is injected and later extracted using wells. In saline aquifers, ASR performance is often limited by dispersive mixing, which creates a transition zone at the edge of the injected freshwater and buoyancy-driven flow, which causes the freshwater to rise and deform during storage—both reducing recovery efficiency. This study investigates whether horizontal wells can improve ASR performance in saline, low-transmissivity aquifers by achieving acceptable recovery efficiencies and outperforming conventional vertical wells. Three configurations were evaluated numerically with MODFLOW 6: a horizontal well, a fully penetrating vertical well, and a dual well system with a fully penetrating injection well and a partially penetrating extraction well. Models were tested on a large set of parameter combinations from Latin Hypercube Sampling, targeting conditions where vertical wells perform poorly. The horizontal well generally achieved higher recovery efficiencies, with a median of 45% after five ASR cycles, compared to 6% and 16% for the fully and partially penetrating vertical wells. Its advantage was greatest under strong buoyancy conditions, where vertical wells failed to recover any freshwater. While dispersive mixing reduced horizontal well performance by causing earlier saltwater breakthrough, it improved vertical well recovery by stabilizing the injected freshwater. In conclusion, horizontal wells are promising for ASR when hydraulic conditions require multiple vertical wells and when buoyancy-driven flow significantly limits vertical well performance.

Study region: Island of Curaçao, Caribbean Sea Study focus: Groundwater is a critical resource for many Caribbean islands. This study assessed the groundwater quality (main and trace elements, E. coli) of Curaçao during 4 subsequent wet seasons (2020–2024). New hydrological insights for the region: Three main hydrochemical processes were identified with Principal Component Analysis (PCA): 1) seawater influence (median EC = 1900 µS/cm), 2) nutrient contamination, and 3) silicate weathering from altered basalts. Elevated median nitrate concentrations (all 206 samples = 35 mg/L) are highest in urban areas (102 mg/L) suggesting wastewater leakage from cesspits. Lower Cl/B mass ratios (550) in urban areas compared to natural vegetated (1000) areas, indicate a wastewater source derived from reverse osmosis drinking water relatively high in Boron. Enrichment of δ¹⁵N-NO₃ and δ¹⁸O-NO₃ versus wastewater signatures indicates considerable occurrence of denitrification. Phosphate concentrations remain low (∼0.1 mg/L), except for sites irrigated with large volumes of treated wastewater (>1 mg/L). Cl-B endmember mixing model showed 36 % wastewater contribution to groundwater, making it an important additional freshwater source given the semi-arid conditions. The findings underscore the value of comprehensive groundwater monitoring and the need for sanitation improvements. Addressing these challenges will benefit water management practices on Curaçao and similar islands. Additionally, it will improve groundwater quality, and groundwater-affected ecosystems, such as nearshore coral reefs.

Monitored natural attenuation is commonly used to manage petroleum hydrocarbon-contaminated groundwater. However, it requires periodic, costly grab sampling. We propose a cost-effective, real-time groundwater monitoring proof-of-concept machine learning (ML) framework using in-situ sensors—pH, dissolved oxygen, electrical conductivity, and redox potential—to detect benzene, ethylbenzene, and xylenes (BEX). We built upon the established correlations between hydrocarbon concentrations and in-situ water quality parameters (iWQPs). Due to limited field data, we validated the framework using datasets at virtual wells within a simulated aquifer from our previously developed reactive transport model. In this application, we detected the spreading of pollution downstream of the established pollution plume. The used framework is a binary classification system that flags contamination at virtual downstream wells. We compared five ML classifiers, i.e. Logistic Regression, Random Forest, XGBoost, Multi-layer Perceptron, and Support Vector Classifier, for early warning when BEX reached or exceeded the regulatory threshold of 5 μg/L. The models were trained on virtual wells at and near the source zone and predicted contamination before BEX reached the threshold at downstream virtual wells. This reflects the spatial variability in flow and reaction dynamics that altered BEX-iWQP relationships. Scenario analyses revealed the ML models' sensitivity to aquifer properties, i.e., hydraulic conductivity, electrical conductivity, and electron acceptor availability. We also assessed the impact of sensor noise and seasonal fluctuations on iWQPs. We found that even moderate levels of noise (10–20 %) can significantly affect model accuracy, particularly when the noise was introduced into the test data. Therefore, we recommended to combine hardware stabilization with adaptive smoothing techniques. With these approaches, our proposed framework remains promising for providing early warnings of plume migration toward sensitive receptors.

Attention on the use of transboundary aquifers (TBAs) and their cross-border impacts is growing as countries become increasingly concerned about their long-term water security. Cross-border impacts, in groundwater quality and quantity, tend to concentrate in specific parts of TBAs, as they largely depend on the transboundary flow dynamics where anthropogenic actions operate. Thus, there is a growing consensus that strategies intended to prevent or mitigate such impacts should be implemented in strategic zones rather than in the whole TBA. These transboundary groundwater management zones (TGMZs) are relatively recent but have become a prominent topic in TBA management. However, until now, limited effort has been put into exploring the concept of TGMZs and the methods for their delineation. This research aims to fill these gaps and provide a basis for the delineation of TGMZs, thus helping neighbouring countries meet international responsibilities regarding the right to use and enjoy groundwater in TBAs. By reviewing academic and grey literature accessible from public sources, we present an overview of the concept and terminology of TGMZs, the approaches proposed for their delineation, and current operating examples. Additionally, we build a conceptual framework for assessing cross-border groundwater impacts by identifying their typologies and causal factors. We then apply our framework to evaluate and compare three reported methods which identify and delineate TGMZs from distinct perspectives, thereby gaining insights into their principles, performances, and limitations. Finally, we provide recommendations for further research towards optimising methods for delineating TGMZs.

This study aimed to identify ocean- and land-based sources of nutrients to the coral reef communities surrounding the Southern Caribbean islands Aruba, Bonaire, and Curaçao (ABC islands). The composition of water masses around these islands were assessed to depths up to 300 m and three distinct overlying water masses were identified, separated by mixing zones. A fluctuating pycnocline separating surface from deeper (>∼50 m) water indicated the presence of internal waves. Nutrient profiles were typical of tropical waters with oligotrophic waters occurring above the pycnocline and a deep chlorophyll-a maximum (DCM) just below it (∼65 m). Concentrations of dissolved nutrients differed among islands. Inorganic nitrogen (DIN) and phosphate concentrations were respectively lowest around Bonaire and Curaçao. The spatial distribution of chlorophyll-a (indicative of phytoplankton biomass), rather than nutrient concentrations, suggested the presence of higher-than-average nutrient concentrations in islands with higher population densities and near urbanized/industrial areas.

Iron (Fe²⁺), manganese (Mn²⁺), and ammonium (NH₄⁺) oxidation processes were studied in three single media and three dual media full-scale rapid sand filters (RSFs) using reactive transport modelling (RTM) in PHREEQC and parameter estimation using PEST. Here, we present the insights gained into the spatial distribution of Fe and Mn mineral coatings in RSFs and its influence on the oxidation sequence and rates. Fe²⁺ and Mn²⁺ oxidation predominantly occurred simultaneously in the RSFs, contrary to the expected sequential oxidation based on Gibbs free energy calculations. During backwashing, RSF grains become fully mixed, which initiates heterogeneous Mn²⁺ oxidation on Mn-coated grains that end up in the top layer. The resulting grains have a mixed Fe/Mn mineral coating, which is limiting heterogeneous Mn²⁺ oxidation due to the limited Mn mineral surface available. Mixed coatings did not seem to affect Fe²⁺ oxidation rates, instead oxidation rates were increasing at lower pH. We found that RSFs can be designed to spatially separate Fe²⁺ and Mn²⁺ oxidation, which results in optimal conditions for Mn²⁺ oxidation. The RSF needs to consist of two layers with varying density to inhibit mixing and complete Fe²⁺ oxidation should occur in the top layer. The developed RTM can be used to estimate the depth at which Fe²⁺ oxidation is complete, and thus the ideal intersection depth of the two layers. A novel perspective is provided on how mineral coating distribution in single and dual media filters influence removal rates and the sequence of oxidation, which contributes to the design of more efficient groundwater filters.

Petroleum products have contaminated groundwater with harmful organic compounds, such as benzene, toluene, ethylbenzene, and xylenes (BTEX). Collecting and analyzing polluted groundwater samples is expensive and undertaken infrequently. However, quick remedial action in case of unexpected events requires continuous monitoring. In-situ water quality sensors (pH, EC, DO, ORP) may show correlations with the components of dissolved petroleum hydrocarbon (PHC) such as aromatics and non-volatile mobile fractions. Correlations are prerequisite to ultimately develop real-time prediction models. Since suitable field data sets are limited, we simulated the fate of hydrocarbons in groundwater under various realistic conditions using a reactive transport model as novel approach to explore when, where, and why correlations occur. A stationary oil source zone continuously dissolved at the top of a heterogeneous and shallow sandy aquifer over a two-dimensional cross-section. Our model considered transient conditions (fluctuating water table) and spatially uniform hydrogeochemical composition. We observed a strong correlation between PHCs and water quality sensors (rolling Spearman's correlation > |0.8|) at varying periods. These correlations are strongly affected by the location of observation wells, the aquifer's hydraulic conductivity, and the availability of calcite and oxide minerals, and other electron acceptors. DO and ORP are significant for the early detection of hydrocarbon contamination, whereas pH and EC are important features for the long-term monitoring of hydrocarbons. Our findings lay the foundation for the subsequent development of a data analysis model to detect and estimate in real time PHC levels in groundwater using in-situ water quality sensors.

Agricultural aquifer storage recovery and transfer (ASTR) stores excess fresh water for later reuse in irrigation. Moreover, water quality improves because chemical pollutants and pathogens will be removed by degradation and attachment to the aquifer material. The source water may contain the bacterial plant pathogen Ralstonia solanacearum which causes plant infections and high yield losses. We used quantitative microbial risk assessment (QMRA) to investigate the removal of R. solanacearum during ASTR to predict infection risks of potato plants after irrigation with the recovered water. Laboratory experiments analyzed the ASTR treatment by investigating the bacterial die-off in the water phase and the removal by attachment to the aquifer sediment. Die-off in the water phase depends on the residence time and ranged between 1.3 and 2.7 log₁₀ after 10 or 60 days water storage, respectively. A subpopulation of the bacteria persisted for a prolonged time at low concentrations which may pose a risk if the water is recovered too early. However, the natural aquifer sand filtration proofed to be highly effective in removing R. solanacearum by attachment which depends on the distance between injection and abstraction well. The high removal by attachment alone (18 log₁₀ after 1 m) would reduce bacterial concentrations to negligible numbers. Upscaling to longer soil passages is discussed in the paper. Infection risks of potato plants were calculated using a dose-response model and ASTR treatment resulted in negligible infection risks of a single plant, but also when simulating the irrigation of a 5 ha potato field. This is the first QMRA that analyzed an agricultural ASTR and the fate of a plant pathogen focusing on plant health. QMRA is a useful (water) management tool to evaluate the treatment steps of water reclamation technologies with the aim to provide safe irrigation water and reduce risks disseminating plant diseases.

Well clogging was studied at an aquifer storage transfer and recovery (ASTR) site used to secure freshwater supply for a flower bulb farm. Tile drainage water (TDW) was collected from a 10-ha parcel, stored in a sandy brackish coastal aquifer via well injection in wet periods, and reused during dry periods. This ASTR application has been susceptible to clogging, as the TDW composition largely exceeded most clogging mitigation guidelines. TDW pretreatment by sand filtration did not cause substantial clogging at a smaller ASR site (2 ha) at the same farm. In the current (10 ha) system, sand filtration was substituted by 40-μm disc filters to lower costs (by 10,000–30,000 Euro) and reduce space (by 50–100 m²). This measure treated TDW insufficiently and injection wells rapidly clogged. Chemical, biological, and physical clogging occurred, as observed from elemental, organic carbon, 16S rRNA, and grain-size distribution analyses of the clogging material. Physical clogging by particles was the main cause, based on the strong relation between injected turbidity load and normalized well injectivity. Periodical backflushing of injection wells improved operation, although the disc filters clogged when the turbidity increased (up to 165 NTU) during a severe rainfall event (44 mm in 3 days). Automated periodical backflushing, together with regulating the maximum turbidity (<20 NTU) of the TDW, protected ASTR operation, but reduced the injected TDW volume by ~20–25%. The studied clogging-prevention measures collectively are only viable as an alternative for sand filtration when the injected volume remains sufficient to secure the farmer’s needs for irrigation.

Hydrochemical data of karst springs provide valuable insights into the internal hydrodynamical functioning of karst systems and support model structure identification. However, the collection of high-frequency time series of major solute species is limited by analysis costs. In this study, we develop a method to retrieve the individual solute concentration time series and their uncertainty at high temporal resolution for karst springs by using continuous observations of electrical conductivity ((Figure presented.)) and low-frequency ionic measurements. Due to the large ion content and non-negligible concentrations of aqueous complexes in karst systems, the concentration of each solute species occurring as free ion and as part of aqueous complexes are computed separately. The concentration of species occurring as free ions are computed considering their contributions to the total (Figure presented.), whereas the concentration of the species as part of complexes are obtained from speciation calculations. The pivotal role of the complexation processes for the reconstruction of solute concentration time series starting from the (Figure presented.) signal is investigated in two karstic catchments with different geologies and temporal resolution of the available hydrochemical datasets, that is the Kerschbaum dolostone system in Austria and the Baget limestone system in France. The results show that complexation processes are significant and should be considered for the estimation of the total solute concentration in case of SO₄, Ca, Mg and HCO₃. The (Figure presented.) signal of a karst spring can be used to interpolate and quantify the dynamics of those solutes characterized by large contribution (approximately >6%) to the total (Figure presented.) and low relative variability, that is HCO₃, Ca and Mg. Moreover, the presented method can be used to estimate concentrations of solutes when applied to karst systems with stationary and hydrogeochemical homogeneous contributing area. On the contrary, the method is affected by large uncertainty in case of dynamic systems characterized by varying contributions of water from different geological areas. This study aims to contribute to the problem of hydrogeochemical data availability and to support future works on karst systems conceptualization.

Irrigation with surface water carrying plant pathogens poses a risk for agriculture. Managed aquifer recharge enhances fresh water availability while simultaneously it may reduce the risk of plant diseases by removal of pathogens during aquifer passage. We compared the transport of three plant pathogenic bacteria with Escherichia coli WR1 as reference strain in saturated laboratory column experiments filled with quartz sand, or sandy aquifer sediments. E. coli showed the highest removal, followed by Pectobacterium carotovorum, Dickeya solani and Ralstonia solanacearum. Bacterial and non-reactive tracer breakthrough curves were fitted with Hydrus-1D and compared with colloid filtration theory (CFT). Bacterial attachment to fine and medium aquifer sand under anoxic conditions was highest with attachment rates of max. k _att1 = 765 day ^-1 and 355 day ^-1, respectively. Attachment was the least to quartz sand under oxic conditions (k _att1 = 61 day ^-1). In CFT, sticking efficiencies were higher in aquifer than in quartz sand but there was no differentiation between fine and medium aquifer sand. Overall removal ranged between < 6.8 log ₁₀ m ⁻¹ in quartz and up to 40 log ₁₀ m ⁻¹ in fine aquifer sand. Oxygenation of the anoxic aquifer sediments for two weeks with oxic influent water decreased the removal. The results highlight the potential of natural sand filtration to sufficiently remove plant pathogenic bacteria during aquifer storage.

A field injection experiment was performed in an anoxic sandy aquifer over 6 days to assess sorption characteristics of 7 commonly applied pesticides in agriculture and 2 frequently detected metabolites. Pesticide use changed considerably in the last decades, and there is insufficient knowledge of the fate of currently used pesticides in aquifers. Injected water arrival was monitored at 6 depth intervals of 1 m ranging from 11.4 to 32.2 m-below surface level with varying organic carbon contents (0.057–0.91%d.w.) to examine intra-aquifer variations in sorption. Observed pesticide concentrations were fit using a non-linear least squares routine to an advection-dispersion equation, from which retardation factors (R) were obtained. Pesticide degradation did not significantly influence the simulated R during the experiment. We observed that bentazon and cycloxydim were most mobile with R < 1.1 at all depths. Desphenyl chloridazon, methyl desphenyl chloridazon, and imidacloprid were, on average, less mobile, with maximum R of 1.5. Boscalid, chloridazon, fluopyram, and flutolanil showed a larger range of R, and R > 2.0 were observed in the shallowest part of the aquifer. Largest R were observed at the top of the aquifer and decreased with depth. K_oc values varied similarly, which indicates that sorption is not only influenced by sedimentary organic matter (SOM) content but also by its sorption reactivity. Obtained sorption parameters were substantially lower than reported in a widely used pesticide sorption database, which suggests that sorption parameters are influenced by methodological differences and variations in the sorption reactivity of SOM. The large intra-aquifer variations in pesticide sorption highlights that aquifer heterogeneity should be considered in groundwater risk assessments.

Degradation of 7 common pesticides (bentazon, boscalid, chloridazon, fluopyram, flutolanil, imidacloprid, and methoxyfenozide) and 2 metabolites of chloridazon (desphenyl-chloridazon, and methyl-desphenyl-chloridazon) was studied in an anoxic and brackish sandy aquifer before and during Aquifer Storage Transfer and Recovery (ASTR) operation. Fresh tile drainage water was injected and stored for later re-use as irrigation water. We hypothesized that electron acceptors (O₂, NO₃), dissolved organic carbon (∼24.7 mg/L), nutrients (NO₃: ∼14.1 mg/L, NH₄: ∼0.13 mg/L, PO₄: ∼5.2 mg/L), and biodegrading bacteria in tile drainage water could stimulate degradation of the pesticides and metabolites (ranging between 0.013 and 10.8 μg/L) introduced in the aquifer. Pesticide degradation was studied at 6 depths in the aquifer using push-pull tests lasting ±18 days before the onset of ASTR operation. Degradation was too limited to quantify and/or could not be assessed because of the potential occurrence of pesticide retardation. Utilizing push-pull tests to obtain degradation constants should only be considered in future studies for non-retarding pesticides with relative low half-lives (here <20 days). During ASTR operation, pesticide degradation was studied at the same depths during 3 storage periods equally spread over 1.5 years of ASTR operation. Overall, trends of degradation were observed, although with relatively high half-lives of at least 53 days. Microbial adaptation of the aquifer and/or bioaugmentation by the injected biodegrading bacteria did not result in enhanced degradation during consecutive storage periods. Operational monitoring data over longer periods and distances yielded half-lives of at least 141 days. The slow degradation mostly agrees with previous studies. The injected tile drainage water composition did therefore not notably stimulate pesticide degradation. The relatively persistent behavior of the studied pesticides/metabolites implies that ASTR abstracted water will have generally high pesticide concentrations, and non-abstracted water may form a contamination risk for the surrounding native brackish groundwater.