Most deep learning studies in hydrology adopt single-task frameworks that address individual variables such as rainfall or streamflow independently, limiting opportunities for shared learning across related environmental processes. This study introduces a unified multi-task, multi-modal deep learning framework capable of performing both 24-hour horizon streamflow forecasting and rainfall temporal super-resolution within a shared architecture. The model employs a shared Transformer encoder with task-specific decoders to integrate temporal and spatial hydrological information within a single architecture. To assess the influence of joint optimization, the same model is also trained individually for each task, enabling direct comparison between single-task and multi-task configurations and performance. Results show that multi-task training improves streamflow forecasting accuracy while maintaining comparable rainfall reconstruction performance relative to individually trained counterparts and established baselines. The framework demonstrates stable streamflow forecasts and hydrologically consistent rainfall reconstructions, highlighting the potential of unified, process-aware architectures for representing multiple components of the hydrological cycle within one coherent learning system.
Climate change is causing significant changes in the hydrology of the Euphrates River Basin, which has traditionally been one of the most important water basins in the region, due to climate change, changing surface runoff patterns, increasing water demand, and changing hydroclimate patterns (i.e., rainfall) over time. The use of hydrological models, therefore, plays an essential role in providing information to support decision-making regarding the use of water resources, and also in understanding how climate and operational changes affect the hydrology of the Euphrates River Basin. In this regard, we used the SWAT (Soil and Water Assessment Tool) model within ArcGIS to simulate the hydrology of the Euphrates River Basin. This is the first time that this model has been applied at the scale of the Euphrates River. Our findings show that by using 16 key parameters to calibrate the model, we were able to attain a greater degree of accuracy in simulating monthly discharge from the Birecik Dam to the Haditha Dam on the Euphrates River. Evidence suggests that the parameters SOL_BD, RCHRG_DP, GWQMN, and ESCO were found to be relatively more sensitive when compared with the more commonly used CN2 parameter to simulate hydrology (contrary to earlier findings). Our results indicate that the model was able to achieve satisfactory performance indicators during both the calibration and validation stages of the project. The calibration results for R², NS, and RSR were 0.83, 0.77, and 0.46; whereas the validation results were 0.87, 0.83, and 0.23. The uncertainty indices were greater during the validation period (probability coefficient 81% and correlation coefficient 0.91) than during the calibration period (probability coefficient 74% and correlation coefficient 0.87), indicating an increase in the reliability of the hydrological forecasts during the validation period. Additional findings of this study were the accurate estimation of the average annual depth of spatial runoff over 29 sub-basins, including point-source discharges, which is a key aspect of advanced hydrological studies. The additional capabilities provided by ArcGIS 10.5 provided additional spatial interpretations of this important variable.
The Réal Collobrier hydrological observatory, located in south-eastern France and managed by INRAE (formerly Cemagref) since 1966, is a benchmark site for regional hydro-climatology. Created by the Ministry of Agriculture, its initial objective is to improve understanding of hydrological processes in Mediterranean regions underlain by metamorphic soils. The observatory's catchment, situated in the Maures massif near the Mediterranean coast, is densely instrumented. Flow measurements are collected at the outlets of ten small forested nested catchments (ranging from 1.57 to 70 km²), including four headwater streams. A dense network of 15 rain gauges records rainfall data at a fine temporal scale. The dataset also includes climatological data, water temperature and soil data. The vegetation is dominated by forest communities on crystalline substrates (maquis of heath, cork oak, maritime pine, and chestnut). The geological formations are predominantly crystalline, with metamorphism increasing from east to west (from gneiss to schists and phyllites) [1]. Direct human influence has been negligible over the past 60 years, with land use and land cover remaining almost unchanged, except for a wildfire in 1990 that affected one small sub-catchment. All data presented in this article are available in the INRAE hydrological observatory's open database (https://bdoh.inrae.fr/). The article describes the long-term dataset collected at the observatory and the validation procedures. The raw dataset underwent quality control, gap filling, and homogenization procedures to ensure temporal consistency and to improve data reliability. This rigorous quality control process results in a robust dataset used for research purposes. This well-documented hydro-climatic information can significantly advance understanding of hydrological processes [2-6], help validate and evaluate models in Mediterranean environments [7-9]. Given the non-perennial nature of rivers in this area, these data are particularly useful for studying the origin of intermittent flow, as well as the start and end dates of flow period [10]. The hydrological dataset now spans 58 years, offring the opportunity to evaluate long-term hydrometeorological trends [11,12]. Since 2019, observations of soil moisture at several depths have been added, providing valuable information on soil water availability and vegetation dynamics. The Réal Collobrier catchment area is part of the SOERE-RBV (Long-Term Observation and Experimentation System for Environmental Research - Mountain Basin Networks), which belongs to the OZCAR research infrastructure [13], certified by AllEnvi (National Research Alliance for the Environment) (http://www.ozcar-ri.org/real-collobrier/).
We present 42 years (1979-2021) of elemental and isotopic data (Cu, Zn, Cd, Pb) from oysters (Crassostrea gigas) in the Gironde Estuary (SW France), integrated with historical river discharge records to disentangle anthropogenic and environmental drivers of trace metal bioaccumulation. Cu and Zn concentrations in oyster tissues show pronounced interannual variability-up to twofold-closely tracking river flow. Their fluctuations likely reflect hydrological controls on the mobilization, speciation, and bioavailability of legacy-bound metals rather than new contaminant inputs. Cd concentrations decline more steadily, consistent with the attenuation of upstream industrial sources, while Pb displays a more complex trend shaped by legacy decline, hydrological variability, and possible changes in source composition. Stable isotope signatures (δ66Zn, δ114Cd, δ65Cu, 206Pb/207Pb) remained remarkably stable over four decades, indicating a constant source apportionment with a dominance of anthropogenic sources on the observed fingerprints. Nevertheless, metal-specific fractionation processes imprint distinct signatures: Zn is enriched in heavy isotopes due to kinetic adsorption onto suspended particles, whereas Cd is isotopically light, consistent with selective desorption via chloride complexation. Slightly heavier δ65Cu values likely reflect vineyard runoff coupled with complexation in the dissolved phase. Pb isotope ratios reveal a gradual shift toward radiogenic values, shaped by diminishing industrial inputs and increasing lithogenic contributions. Together, these findings demonstrate that integrating hydrological data with multi-isotope analysis can disentangle source persistence from estuarine reactivity, offering a framework for interpreting long-term contamination dynamics under changing hydrological regimes.
Bioretention is a widely adopted green infrastructure practice for urban stormwater management, yet its long-term hydrological performance remains poorly understood. Concerns exist that physical clogging of sediments will impair infiltration capacity over time, potentially leading to system failure. This review synthesizes results from over 150 peer-reviewed studies to evaluate how bioretention's hydrological performance evolves over operational timescales of 5-20+ years. Across field studies of systems aged 5-22 years, most well-maintained systems maintained or exceeded design hydraulic conductivity thresholds (typically 25-50 mm/h), with a minority of studies reporting declining performance attributable to maintenance deficiencies or design failures. We identify a key observation: accelerated laboratory column experiments consistently show declines in saturated hydraulic conductivity (Ksat), while field surveys of mature bioretention reveal that most well-maintained systems appear to maintain their infiltration capacity over time, though performance is context-dependent and conditioned on adequate design and maintenance. We attribute this pattern to biological processes (root growth, macrofauna, wetting-and-drying, pedogenesis) that appear to mitigate physical clogging in field systems but are absent from most laboratory experiments. The review addresses four interconnected topics: (1) hydraulic performance and clogging, (2) soil media evolution through pedogenesis and pollutant accumulation, (3) vegetation maturation and ecohydrological feedbacks, and (4) current models and their capacity to simulate long-term behavior. We propose a conceptual three-phase framework for the typical lifecycle of bioretention systems: establishment (0-3 years), maturation (3-10 years), and steady state (>10 years). The available evidence suggests that well-designed and maintained bioretention can function as a living ecosystem whose biological inhabitants help sustain hydrological function, rather than a static engineered filter that inevitably clogs; however, this outcome is not universal and depends critically on vegetation establishment, maintenance regime, and design quality. Critical knowledge gaps remain regarding performance beyond 20 years, transferability across climate zones, and quantitative representation of biological feedbacks in predictive models.
Increased frequency and intensity of extreme weather events, driven by climate change, are expected to alter hydrological and water quality processes in the Connecticut River watershed. This study aims to model the temporal and spatial impacts of climate change on the watershed's hydrology and nutrient dynamics. We used the Hydrologic and Water Quality System (HAWQS), which incorporates the Soil and Water Assessment Tool (SWAT), to establish a baseline scenario and assess two climate scenarios: Coupled Model Intercomparison Project (CMIP5-) Representative Concentration Pathways (RCP) 4.5 and RCP 8.5. The model was validated using observed data from the United States Geological Survey (USGS) gage sites. Our results show that both climate scenarios will cause significant changes in hydrological processes, including a shift in precipitation seasonality, with more rainfall expected during winter and early spring. These changes will affect nutrient loading, shifting the seasonal peaks of nitrogen and phosphorus. Notably, the nitrogen-to-phosphorus ratio is projected to decrease across the entire watershed under both climate scenarios. These findings suggest that the altered hydrological regime and nutrient dynamics could have cascading effects on aquatic ecosystems, impacting phytoplankton and algal growth, with important implications for future nutrient management strategies.
Natural rivers play a crucial role in regulating the global carbon cycle. However, as critical engineered waterways, inter-basin water transfer projects (IBWTs) substantially alter regional hydrological connectivity and biogeochemical processes, yet their contribution to the carbon cycle remains poorly understood. Here, we address this knowledge gap with a comprehensive field investigation of the South-to-North Water Diversion Middle Route Project (SNWD-MRP), the world's largest IBWT. The results show that carbon dynamics within the canal are dominated by autochthonous processes, primarily algal photosynthesis and microbial metabolism, which establish a seasonal pattern of "summer consumption and winter storage" for dissolved carbon species and are synergistically modulated by anthropogenic hydrological management. From summer to winter, the concentrations of dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) increased by 20.26% and 30.95%, respectively, whereas the partial pressure of carbon dioxide (pCO2) and the dissolved methane concentration (dCH4) decreased by 49.19% and 42.06%, respectively. Using 2024 as an example, we found that the canal exported 236.94 Gg C yr-1 laterally and emitted 33.77 Gg C yr-1 vertically as carbon dioxide (CO2) and methane (CH4), representing only about 2.57% of the total flux from large natural rivers. Notably, the decadal cumulative lateral DOC export (146.78 Gg C) rivals the annual DOC export of large river systems such as the Yellow River. Algal-fixed CO2 was microbially converted into potent greenhouse gases like CH4, further intensifying the greenhouse effect caused by emissions from the entire canal. Overall, our findings suggest that the large-scale IBWT can create a novel aquatic corridor that disrupt natural carbon boundaries and drive regional-scale carbon redistribution, thereby providing critical insights for coordinating water resource management strategies with carbon neutrality objectives.
Rivers are increasingly exposed to combined anthropogenic and climate-driven pressures, particularly along rural-to-urban gradients where land use, population density, and contaminant sources progressively intensify downstream. In parallel, prolonged droughts followed by intense rainfall events are becoming more frequent, potentially enhancing contaminant accumulation during dry periods and their rapid mobilization during subsequent first-flush events. However, the interactions between these hydrological extremes and spatial anthropogenic gradients in controlling river geochemistry remain insufficiently understood. This study investigates the spatial and seasonal evolution of river chemistry along the Greve River (Tuscany, Italy), a tributary of the Arno River, characterized by a marked rural-to-urban transition, during the exceptional 2022 European drought and subsequent heavy rainfall event. Six sampling campaigns were conducted between May and October 2022, combining high-resolution spatial monitoring with geochemical and isotopic tracers, including total dissolved solids (TDS), nutrients, and trace elements. The results reveal a progressive downstream enrichment in anthropogenic tracers (e.g., NO3- and K+, up to 9 mg/L and 6.6 mg/L, respectively) and pronounced evaporative isotope enrichment during the driest months (slope of the evaporation line ~ 4), indicating strong hydrological stress conditions. The first intense rainfall event triggered rapid remobilization of accumulated material and heterogeneous contaminant transport dynamics. Elevated nutrient concentrations (e.g., NH4+, up to 1.4 mg/L) in downstream sectors further reflect the combined influence of hydrological stress and urban inputs. These findings demonstrate how climate-driven hydrological extremes can amplify contaminant mobilization along urbanized river continuums. The integration of isotopic and geochemical tracers provides a transferable framework for assessing water-quality vulnerability in river basins increasingly exposed to hydroclimatic instability.
Floodplains play a critical role in mitigating nitrogen loads and nutrient pollution in aquatic ecosystems through denitrification. However, the effects of the impacts of dramatic water-level fluctuations on nitrogen dynamics and the associated microbial communities remains poorly understood. This study explored denitrification processes and alterations in functional microbial communities in the Poyang Lake wetlands, with particular focus on microbial responses to hydrological recession and flooding. Denitrification activity was negligible in soils dominated by Carex cinerascens and Cynodon dactylon during water recession, but potential denitrification rates increased significantly during flooding. Ordination regression analysis and Mantel tests showed that these denitrification changes correlated strongly with shifts in functional microbial communities. In particular, studies have shown that the α-diversity of the soil nirS bacterial community was significantly lower after the recession than during flooding, indicating inhibition of denitrifying communities under drought. During flooding, key species such as nirS-type Rhodocyclales, nirK-type Bradyrhizobium, and Nitrosospira became abundant. These denitrifiers and environmental factors (moisture, temperature, and substrate availability) played crucial roles in denitrification during the hydrological changes in floodplains.
Rising water levels in Qinghai Lake, a key hydrological disturbance, have submerged extensive lakeshore grasslands, triggering ecological changes in the newly inundated zones. This study investigated how this hydrological process alters soil microbial community succession and co-occurrence networks through inputs of dissolved organic matter (DOM) and nitrogen-phosphorus (N-P) dynamics in the lakeshore zone of Qinghai Lake. The results demonstrated that grassland inundation caused by rising water levels enriched the soil with specific DOM components (fulvic and humic acid), driving significant changes in N and P concentrations. High-throughput sequencing results indicated that following grassland inundation, α-diversity declined, whereas β-diversity of soil microorganisms initially rose and then decreased as the duration of flooding prolonged. Furthermore, microbial community assembly shifted over time: in the initial inundation stages, higher proportions of readily bioavailable DOM components and N-P nutrients coincided with the predominance of stochastic processes. Over time, the proportion of recalcitrant DOM components increased, leading to a gradual dominance of deterministic processes. These microbial successional patterns were closely linked to DOM quality and N-P availability. Simultaneously, DOM and N-P content shaped more stable and complex co-occurrence network structures by influencing the formation of core microorganisms. Collectively, these findings highlight the central role of soil microorganisms in mediating ecosystem responses to hydrological disturbances. Our findings elucidate how hydrological disturbances in lakeshore ecosystems reshape soil microbial assembly-transitioning from stochastic responses to deterministic selection-and enhance the stability of co-occurrence networks through interactions between DOM, N, P, and microbial communities. This framework, linking hydrological processes to coupled soil biogeochemical and microbial ecological responses, establishes a vital scientific basis for predicting and managing the ecological consequences of climate-driven lake expansion. It also provides critical insights for developing adaptive management strategies aimed at preserving ecosystem resilience and biogeochemical functioning in lakeshore zones experiencing dynamic hydrological regimes.
Climate variability fundamentally realigns the processes by which mining-derived contaminants are generated, transported, and transformed at major scales, from mineral surface to watershed hydrology. Synthesizing the mechanistic architecture of the climate-mining-water nexus, the review shows that non-stationary hydroclimatic forcing wherein historical precipitation return periods, temperature baselines, and streamflow statistics no longer reliably predict future conditions invalidates the steady-state conditions of traditional water quality predictions. Coupled thermal hydrological geochemical biological atmospheric interactions produce behaviors that emerge at thresholds, hysteresis, and contaminant pulses, which reductionist frameworks overlook. Predictive capacity, therefore, calls for bidirectional, not parallel, integrative approaches to monitoring and modelling for observatories that discriminate within competing conceptual models and reactive transport frameworks that integrate multi-platform data in the face of formal uncertainty quantification. Adaptive management will need to replace static designs with decision architectures robust to deep uncertainty for signpost-based triggers, flexible infrastructure, and iterative learning. Facing up to the nexus will be a transdisciplinary fusion of molecular mechanisms to watershed outcomes, which can sustain prudent stewardship of waters influenced by mining during the age of accelerating climate change.
Terrestrial and aquatic ecosystems are interconnected through runoff and hydrological networks that facilitate the transfer of microbial communities across landscapes. While microbial transport along surface waters is well documented, the role of subsurface hydrological paths in shaping microbial community composition remains poorly understood, particularly in complex karst systems. Here, we studied bacterial communities under stable hydrological conditions across a peri-alpine karst landscape, where mixed limestone-sandstone catchments drain via both surface and subsurface hydrological networks into Lake Thun (Switzerland). We profiled 16S rRNA gene sequences from soils, sediments, surface and subsurface waters, and distinct lake strata. All environments except the lake exhibited high microbial diversity. We observed a clear transitional gradient in bacterial communities along the terrestrial-aquatic interface, with environment type explaining 19% of total variation. Core microbiome analyses revealed both environment-specific and shared taxa, with the strongest overlap between surface and subsurface hydrological networks (63.8%-84.6% shared core taxa). Co-occurrence network analysis identified six major modules. Three of them represented distinct metabolic assemblages tightly associated with specific environment types: peat soils, lake strata, and the subsurface network, respectively. One recurrent module spanned multiple environments and was linked to redox-driven processes, including the oxidation of nitrogen compounds, metals, and methane. Two additional modules comprised aquatic copiotrophs associated with streams and soil heterotrophs prone to export and short-term persistence within the hydrological network. Overall, our results demonstrate that specific environmental settings and hydrological connectivity jointly contribute to selection of microbial species within the karst landscape.
Acid mine drainage (AMD) has emerged as a major global water-environment threat because legacy sulfide mining continuously releases acidity and toxic metals into surrounding water systems, causing long-term ecological degradation and threatening water-resource security. This study combined hydrochemical analysis, hydrogeochemical modeling, and a novel Carbonate-adjusted Modified Acid Mine Drainage Index (MAMDI_C) to investigate AMD evolution and iron mineral transformation processes across four representative mining areas (DBS, TLG, LHC, and JCG) in Southern Shaanxi, China. Results showed that mine adit drainage and waste-rock leachate were the primary pollution sources, characterized by extremely low pH and elevated concentrations of sulfate and dissolved metals (e.g., Fe, Mn, Zn). Hydrochemical facies progressively evolved from SO4-Ca-type waters in mining-affected zones to HCO3-rich waters downstream due to dilution and carbonate buffering. Precipitation significantly enhanced infiltration, discharge, and hydrological connectivity, promoting the flushing and downstream transport of stored oxidation products, particularly in the TLG mining area. The MAMDI_C index effectively differentiated high-risk adit drainage, transitional reaches, and relatively stable upstream waters by integrating pollution intensity with carbonate-buffering capacity. Hydrogeochemical modeling revealed a pH-dependent sequence of Fe transformation from dissolved Fe2+ and sulfate-complexed species under acidic conditions to secondary minerals such as jarosite-K, goethite, and hematite during downstream neutralization. These mineral transformations played a dominant role in controlling trace metal mobility and long-term immobilization. This study provides a process-sensitive framework for understanding AMD evolution, assessing AMD risks, and developing targeted remediation strategies in carbonate-buffered mining watersheds.
Remote sensing (RS) and Artificial intelligence (AI) are increasingly applied to monitor vegetation and hydrology in the Arctic and Antarctic, where logistical and environmental constraints make fieldwork difficult. These technologies offer new opportunities to track ecological change, but the extent, consistency, and methodological quality of current applications have not been systematically reviewed. This study presents the first PRISMA 2020 based systematic synthesis of AI enhanced RS, collectively termed GeoAI, applied to Arctic and Antarctic environments (2005-2025; 116 studies). Publication activity has expanded significantly since 2018, driven by the convergence of uncrewed aerial vehicle (UAV), multispectral imaging, satellite archives, and deep learning (DL). Bibliometric and conceptual-network analyses reveal a rapid shift from isolated ecological monitoring toward integrated, data-fusion frameworks linking vegetation, hydrology, and climate processes. Classical machine learning approaches remain foundational, while DL-based convolutional neural-network architectures are emerging as powerful tools for fine-scale segmentation and prediction. Most studies still operate at the landscape scale, with few achieving full UAV-to-satellite integration, exposing persistent spatial-resolution and validation gaps. Vegetation hydrology coupling is reported in most cases, though subsurface and process-based monitoring remain limited. Spectral-index analysis reveals a persistent reliance on greenness metrics, yet there is a growing shift toward pigment, moisture, and cryptogam-sensitive indices that more accurately capture plant physiological function and microclimatic interactions. This review establishes the empirical foundation for next-generation polar monitoring, emphasising hierarchical UAV-to-satellite fusion, open benchmark datasets, and explainable, ecologically grounded AI as essential pathways for scalable, climate-adaptive conservation of Earth's fastest-changing regions.
Vegetation restoration is widely regarded as a key measure for mitigating soil erosion in the middle reaches of the Yellow River. However, the regulation of flood sediment transport by forest-grass vegetation coverage (Ve) shows strong nonlinear characteristics and threshold effects. This makes it difficult for traditional physical models to accurately describe sediment production responses across different vegetation stages. To address this issue, this study proposes a stage-specific water-sediment simulation framework that integrates Ve threshold identification with machine learning, namely a threshold-aware modeling framework. Four typical basins in the middle reaches of the Yellow River, namely the Kuye, Wuding, Fen, and Wei River basins, were selected as the study areas. Based on multi-source data from 1979 to 2025, including flood events, rainfall, Ve, and land use, random forest was first used to identify the dominant factors controlling sediment production. The exponential function, piecewise linear regression, and Copula model were then combined to identify Ve threshold bands from three perspectives: functional form, structural breakpoint, and probability dependence. Based on the identified thresholds, threshold information was embedded into LSTM, XGBoost, and SVR models to construct overall and stage-specific simulation scenarios.The results show that: (1) Ve and rainfall are the key factors controlling sediment production; (2) clear Ve threshold bands exist in all basins, with recommended ranges of 31%∼35% for the Kuye River Basin, 27%∼32% for the Wuding River Basin, 23%∼28% for the Fen River Basin, and 20%∼23% for the Wei River Basin; (3) model performance improved significantly after introducing threshold information. Taking the Kuye River Basin as an example, the NSE values of the LSTM, XGBoost, and SVR models increased to 0.858, 0.861, and 0.830, respectively; and (4) vegetation exerted a strong regulatory effect on suspended sediment concentration in the pre-threshold stage, whereas the system gradually shifted to a sediment-supply-limited state in the post-threshold stage, with a markedly weakened marginal vegetation effect. This study reveals the stage-specific regulatory mechanism of vegetation restoration on water-sediment processes. It also provides new theoretical and methodological support for intelligent modeling of complex water-sediment systems and ecological management of river basins.
Natural wetlands deliver a range of ecosystem services like water, food and fibre provisioning, carbon sequestration, nutrient retention, and support for biodiversity. With respect to climate change, wetlands may act as a carbon sink or source, depending on management conditions. Despite their value, wetlands are disappearing at an alarming rate, and threatened by hydrological alteration, pollution and climate change. For an effective wetland policy there is a need to relate the state of wetlands to regional and global land-use and climate change projections, and to relate ecosystem services to wetland processes. Wetlands are, however, generally under-represented in global models and assessments. Here we present a model that estimates vegetation biomass production, carbon emissions, and water quality of freshwater wetlands on a global scale with different hydrological and climate conditions. The main hydro-ecological processes are described in a generic way, accounting for climate zones, water level fluctuations and main hydrological types: rain-/groundwater fed (ponded) wetlands and surface water-fed floodplain wetlands (flooded). The model is coupled to global hydrological (PCR-GLOBWB) and climate and land-use (IMAGE-GNM) models. It estimates the wetlands ecosystem services, in particular regulating ecosystem services like water availability, carbon sequestration/emission and nutrient retention that are difficult to quantify otherwise. The model was applied to several wetland types in widely varying climate regions (Sweden, Germany, Spain, subtropical China, tropical Brazil and Kenya). Results show that the model generates plausible results compared to measured data of greenhouse gas emissions and nutrient concentrations. Furthermore, the model can discriminate between wetlands with different environmental conditions, resulting in wetlands being either a sink or a source of carbon. A regionalized parameterization is in progress. Further potential applications of model outcomes include regional assessments of wetland ecosystem services, determining ecosystem services under alternative management, climate and land use scenarios, and link these to conditions for biodiversity.
Understanding of the transport and retention of perfluoroalkyl substances (PFAS) precursors in the unsaturated zone remains limited, particularly regarding the coupled effects of hydrological variability and microbial processes under variably saturated conditions. This study investigated the transport behavior of 6:2 fluorotelomer sulfonate (6:2 FTS), a representative PFAS precursor, through variably saturated columns packed with sand or soil. Quantification of retention was combined with solid-phase characterization and microbial community sequencing. An instrumented apparatus was used to collect in-situ samples and to continuously monitor water saturation (Sw) during the experiments. The results demonstrated that 6:2 FTS transport was co-regulated by porous media type, Sw, and microbial colonization. Retention was stronger in agricultural soil than in quartz sand and decreased with depth as increasing Sw reduced air-water interfacial area. Extracellular polymeric substances (EPS) introduced a rate-limited retention domain that operated alongside rapid air-water interfacial adsorption, which enhanced overall retention and produced nonideal transport behavior, including dual-plateau breakthrough curves. Greater EPS abundance and spatial heterogeneity in soil accounted for stronger and more prolonged retention despite higher Sw. Microbial community structure was shaped primarily by porous media type, followed by 6:2 FTS exposure. Co-occurrence network and biomarker analyses further identified media-specific and 6:2 FTS-tolerant genera. Overall, this study provides mechanistic evidence that microbial processes can substantially modulate hydrologically controlled PFAS transport in the vadose zone. Incorporating such coupled biophysical processes is therefore essential for accurate PFAS fate prediction and risk assessment in subsurface environments.
The timing of the initial spring water-level rise represents a key indicator of seasonal hydrological transition in snowmelt-dominated river systems of high-latitude regions. This study evaluates the capability of ensemble machine learning (ML) models to estimate the onset date of the spring water-level rise in Arctic-subarctic rivers of the Anadyr-Kolyma basin district in northeastern Russia using a station-year dataset for the period 2008-2022, combining hydrological observations with meteorological and basin-related predictors. Five regression algorithms were tested using grouped cross-validation by year. CatBoost achieved the highest predictive accuracy with an out-of-fold mean absolute error of 4.54 days, RMSE of 9.79 days, and [Formula: see text], slightly outperforming ExtraTrees (MAE 4.66 days) and RandomForest (MAE 4.70 days). Spatial analysis shows that most gauging stations exhibit prediction errors within 0.5-3 days, whereas errors exceeding 10 days occur mainly in small or topographically complex basins with limited observational coverage. Model interpretation using SHapley Additive exPlanations (SHAP) and partial dependence (PDP) analysis indicates that predictors describing thermal forcing during late winter and early spring dominate the model response, with positive degree days during March-April, the first thaw day, and indicators of rapid water-level rise providing the largest contributions. The onset of spring water-level rise in the studied Arctic-subarctic river systems is primarily associated with the interaction between temperature-driven snowmelt processes and the early hydrological response of the river network, whereas precipitation and spatial descriptors exhibit comparatively smaller contributions. These statistical relationships are conditioned on the 2008-2022 period and may vary under different climatic conditions or longer observational records, which should be considered when applying the model for prediction.
Bangladesh is a land of numerous fluvial waters bodies traversing across the whole landscape which has the complex waterbodies form the hydrological and ecological regime. Prediction of watershed change intrinsically imposes challenges due to complex hydro-climatic reasons and water quality processes. The Kaptai lake manifests fluctuations for its seasonal precipitation and changes in vegetation condition. The remote sensing-based watershed behavior analysis using Machine learning (ML) architecture particularly Stacked Regression (SR) for Kaptai boundary is seen to have a potential to address the variability that is absent in the current literature. This study has been conducted using the Kaptai watershed, which is one of the largest artificial lake of South-East Asia, with the objective for developing a data driven methodology for a sustainable forecasting of watershed behavior. Thus, water Quality Index, meteorological variables, integrative synthesis of remote sensing indices along with combination of Machine Learning Architecture have been applied. For capturing the spatial and temporal dynamics, spatial maps spanning the 1990-2025, a period of 35 years have been generated using ArcMap (ArcGIS) for salient biophysical and water quality parameters. The study reveals that the steep slope of the Chittagong Hill Tracts (CHT) is the root cause of high TDS and Turbidity (NDTI). The heavy rains and land-use changes, namely deforestation and shifting cultivation further assist in transport of large sediment load into the reservoir. The indication of a positive non-linear water quality dynamics, periodic eutrophication and turbidity peaks (NDCI up to 0.99) and spikes, and significant vegetation loss and significant increase in LST are seen to play the role. A LightGBM model for Time Series prediction and a Stacked Regression model comprising of XGBoost, Random Forest and Multi-Layer Perceptron has been integrated through RidgeCV meta-Learner. The derived analytical outcomes give an R2 = 0.977 from Stacked Regression modelling. Moreover, by utilizing Light GBM and ANN an R2 value of 0.9402 and 0.99 have been obtained, respectively. This framework shows substantive functionality for forecasting in a scalable approach by involving hydrological, and biophysical processes within the watershed under complex variability. The findings here are a pathway towards an empirically grounded water management and climate resilient behaviors of Kaptai lake, that in turn help in holistic decision making.
The rivers in the middle and lower reaches of the Yangtze River are important wintering and migration stopovers for waterbirds. The hydrological characteristics of rivers directly affect the habitats of overwintering waterbirds and thus lead to changes in the diversity of overwintering waterbirds. The construction of artificial low-head dams has altered the natural hydrological processes of rivers, and therefore, investigating their influence on the composition of wintering waterbird communities is of great significance for the conservation and management of waterbirds. This study was carried out in the Xin'anjiang River Basin from October 2021 to March 2022, with 11 low-head dams selected as the research sites. Utilizing the sampling method, it investigated the species and abundance of wintering waterbirds in both the catchment and tailwater zones of these dams. Subsequently, the diversity of overwintering waterbirds in the two aforementioned zones was calculated, and their inter-zonal differences were analyzed and compared. The results of the study indicate that there are significant differences between the catchment area and the tailwater area of the "ZSJC" Dam (Z = 1.945, p = 0.001), whereas no significant disparities are observed in the species count and abundance of wintering waterbirds using that particular area between the catchment and tailwater areas of other dams. Compared with the catchment areas, the tailwater areas of the dams exhibit a more concentrated and abundant distribution of overwintering waterbirds, while the distribution of overwintering waterbirds in the catchment areas is more uniform than that in the tailwater areas. The 11 dams under study all demonstrated spatial turnover advantages, suggesting that catchment areas and tailwater areas make comparable contributions to β diversity. Bivariate correlation analysis in SPSS detected a significant correlation between dam vertical length and β diversity. In summary, low-head dam construction significantly affects the alpha diversity, beta diversity, abundance, and community composition of wintering waterbirds by modifying hydrological conditions and habitat structure in the Xin'an River Basin. This study provides a scientific basis for waterbird protection and low-head dam management.