Groundwater level dynamics under freeze-thaw conditions remain unclear. We use interpretable deep learning to simulate water table changes and identify seasonal drivers in seasonally frozen regions. During freeze-thaw, changes in soil water potential cause two-way exchange between soil water and groundwater, while rainfall, runoff, and irrigation dominate in other periods. These insights inform groundwater modeling and management in cold regions.

Water balance data are affected by various errors (bias and noise). To reduce these errors, this study presents a water balance data fusion approach that combines multi-scale data (from satellites and in-situ sensors) for each water balance variable and jointly calibrates them, resulting in consistent, bias-corrected and noise-filtered, water balance estimates, along with uncertainty bands. These estimates are useful for constraining process-based models and informing water management decisions.

Our research presents an improved method to enhance the understanding and prediction of water flows in rivers and streams, focusing on key runoff components: surface flow, baseflow, and total runoff. Using a streamlined model, the MPS model, we analyzed over 600 catchments in China and the US, demonstrating its accuracy in capturing the spatial and temporal variability of these components. This model offers a practical tool for water resource management.

This study presents the first detailed hydrodynamic assessment of the May 2024 floods in southern Brazil. The model accurately reproduced flood levels and extent, validated with in situ and satellite observations. Results indicate limited effectiveness of proposed hydraulic interventions, highlighting constraints of structural solutions. The findings improve understanding of complex river–estuary–lagoon dynamics and support better planning for future extreme floods under climate change.

This study presents a novel approach aimed at using biogeochemical data from surface water to decompose streamwater flow into spatial and vertical contributions. A selection of tracers was used in a mixing model to estimate contributions at the outlet of a peri-urban catchment. Results provided a better understanding of hydrological processes in the catchment and demonstrated the potential of biogeochemical data to discriminate spatial contributions according to land use.

This study evaluates hybrid hydrological models combining physics-based and data-driven components, using Information Theory to measure their relative contributions. When testing conceptual models with Long Short-Term Memory (LSTM) networks that adjust parameters over time, we found performance primarily comes from the data-driven component, with physics constraints adding minimal value. We propose a quantitative tool to analyse this behaviour and suggest a workflow for diagnosing hybrid models.

We developed a literature-based approach for projecting future snow depths, which was applied to four measurement stations in Switzerland under a +2 °C temperature scenario, revealing significant declines in snow depths. Validation against published data shows that the approach captures key trends in snow loss. This resource-efficient method provides a practical tool for estimating climate change related snow depth declines, which are lacking highly resolved climate projections.

Focusing on Drought-Flood Abrupt Alternation (DFAA) under climate change, this study investigates the mitigating role of reservoirs on DFAA in Lancang-Mekong River Basin. The drought-to-flood (DTF) probability is higher than the flood-to-drought (FTD) at all intensities. Reservoirs markedly reduce wet season’s FTD and year-round DTF, effectively shorten DFAA’s monthly span. The reservoir's mitigating effect on the total risk of DFAA is positively correlated with their total storage.

Long-term overuse of groundwater has lowered water levels and increased pollution in parts of northern China. We studied whether adding river water back into the ground can restore groundwater and reduce nitrate pollution. Using computer simulations based on field data, we found that recharge can significantly raise water levels and lower nitrate levels, mainly by dilution rather than natural removal processes. 

Intermittent streams are vital to ecosystems and water supply, but are hard to monitor and increasingly affected by climate change. To address this, we used field camera images from 2017 to 2023 at a stream in northern California to train a machine learning model that classifies streamflow as dry, low, or high. This low-cost method enables monitoring of changing intermittent stream conditions and supports water management in data-scarce regions.

Floods threaten lives and property and require rapid mapping. We compared two artificial intelligence approaches on aerial imagery: a fine‑tuned Segment Anything Model (SAM) guided by point or bounding box prompts, and a U‑Net network with ResNet‑50 and ResNet‑101 backbones. The point‑based SAM was the most accurate with precise boundaries. Faster and more reliable flood maps help rescue teams, insurers, and planners to act quickly.

Predicting river flow accurately is crucial for managing water resources, especially in a changing climate. This study used deep learning to improve streamflow predictions across Australia. By either enhancing existing models or working independently with climate data, the deep learning approaches provided more reliable results than traditional methods. These findings can help water managers better plan for floods, droughts, and long-term water availability.

The extent to which streamflow varies in response to variability in precipitation and potential evaporation is essential for understanding climate change impacts on water resources. This so-called streamflow sensitivity is often estimated directly from observational data, but the robustness of these estimates remains unclear. Through systematic examination of existing approaches, we highlight uncertainties inherent in all approaches and discuss their origins.

This study presents a low-cost, automated system for monitoring river water levels using cameras and AI. By combining AI-based image analysis with photogrammetry, it accurately measures water levels in real-time, even in challenging conditions. Tested over 2.5 years at four sites, it achieved high accuracy (errors of 1.0–2.3 cm) and processed over 219 000 images. Its resilience makes it ideal for flood detection and water management in remote areas.

Flood impacts can be enhanced when they occur after droughts, yet the effectiveness of hydrological models in simulating these events remains unclear. Here, we calibrated four conceptual hydrological models across 63 catchments in Chile and Switzerland to assess their ability to detect streamflow extremes and their transitions. We show that drought-to-flood transitions are generally poorly captured, especially in semi-arid high-mountain catchments than in humid low-elevation ones.

We use drone surveys to map river intermittency in reaches and classify them into "Wet", "Transition", "Dry" or "Not Determined". We train Random Forest models with 40 candidate predictors, and select altitude, drainage area, distance from dams and dynamic predictors. We separate different models based on dynamic predictors: satellite indices (a) and (b); or (c) antecedent precipitation index (30 days). Model (a) is the most successful in simulating intermittency both temporally and spatially.

This paper explores  how recharge strategies and the reactive barrier affect water infiltration, retention, and biofilm dinamics in a pilot-scale Soil Aquifer Treatment (SAT) system. We combine Electrical resistivity tomography (ERT) imaging with field measurements (water content, temperature, oxygen) in the unsaturated zone, to link to biofilm and plant dynamics. This integrated approach provides insights into balancing microbial treatment benefits with hydrogeological impacts and system longevity.

Master Recession Curves have long been used to estimate the decline in streamflows following rainfall. Although the approach has wide practical application in water management, it can be overly simple and not match all observations. We propose an approach that represents the changes in streamflow behaviour due to variations in climate and geology. The approach matches observed behaviour well, is consistent with field observations, and is well suited to uncertainty analysis.