The development of AI models is increasing at a rapid rate. However, when are they ready to be deployed in real-world operational settings? In this paper, we introduce a framework to support such assessments and apply it to Google’s recently released AI-based flood prediction system, which is claimed to achieve “reliability in predicting extreme riverine events” and provide “accurate and timely warnings” that are available “earlier and over larger and more impactful events in ungauged basins”. The system has been integrated into an operational early-warning platform producing open, real-time forecasts in more than 80 countries. While this development promises to usher in a new and exciting age in global flood forecasting, the supporting evidence relies heavily on several subjective choices, the implications of which have not been acknowledged or assessed. Here, we evaluate the consequences of these choices on claims of operational deployment readiness across four dimensions: predictive accuracy, forecast timeliness, the characterization of extreme events, and benchmarking against state-of-the-art models. Our assessment reveals that the system’s actual predictive accuracy is likely to be substantially lower than reported—particularly for extreme events—raising concerns about responsible practices across modelling and publicity in high-stakes applications. The deployment of the Google AI model therefore risks misinforming those who depend on its outputs for evacuation and preparedness decisions, particularly in less-developed countries such as those targeted by the enterprise, given its alarmingly high (>90%) rates of false positives and false negatives. Beyond the immediate operational consequences, if left unaddressed, these outcomes may erode public trust in AI within hydrological sciences. We conclude by calling for greater transparency, accountability, and methodological rigor in the integration of AI into flood forecasting.

Preferential flow paths (e.g., macropores or subsurface pipe networks) in hydrological systems facilitate the rapid transmission of precipitation and solutes to streams, resulting in streamflow responses characterized by the release of younger water (i.e., recent precipitation) from the catchment and correspondingly short transit times (on the order of days). While preferential flow paths are documented in both the unsaturated zone and groundwater aquifers, it remains uncertain whether catchment-scale isotope-based transport models can adequately represent preferential flow using tracer measurements in streamflow. In this study, we hypothesize that the preferential release of young water from both the unsaturated zone and groundwater aquifers can be isolated from the streamflow tracer signal. This can be studied with StorAge Selection (SAS) functions, which describe how young or old water leaves a storage. We systematically compared multiple parameterizations of SAS functions describing how water of different ages is released from the unsaturated zone and groundwater aquifer within a single catchment-scale transport model using long-term measurements of hydrogen isotopes in water ( ^δ2H) from two headwater catchments (the Hydrological Open Air Laboratory (HOAL) in Austria and the Wüstebach catchment in Germany). The results show that ^δ2H measurements in streamflow exhibited sufficient variability to isolate the preferential release of younger water through preferential flow paths in the unsaturated zone. In contrast, the variability of ^δ2H in streamflow was insufficient to isolate the preferential release of younger water from the groundwater aquifer, as any seasonal variations in pore water ^δ2H were largely damped by substantial passive groundwater storage (water that mixes with the tracer signal of the active groundwater volume). Consistent with this interpretation, the degree of attenuation in the simulated streamflow isotope signal increased with increasing passive groundwater storage volumes and became pronounced when passive storage was orders of magnitude larger than active groundwater storage. The size of passive groundwater storage, in combination with groundwater SAS function parametrizations, regulated the long tails (100<T<1000 d) of transit time distributions, resulting in considerable uncertainty (± 20 % for HOAL and ± 23 % for Wüstebach) in the fraction of streamflow older than 100 d. The findings demonstrate that stable water isotope measurements from streamflow outlets is insufficient to constrain preferential groundwater flow in the two study catchments and plausibly in similar catchments characterized by large passive groundwater storage. The variability in streamflow TTD estimates arising from different groundwater storage SAS function parametrizations is considerable. Reducing uncertainty in groundwater transit time estimates and preferential flow contributions to streamflow requires complementary data sources, including multiple tracers, high-frequency tracer analysis, and groundwater-level observations, to improve catchment-scale transit time modelling.

Although large-sample hydrology data sets are increasingly used to advance predictions in ungauged basins, the influence of landscape data quality on model regionalization remains insufficiently explored. This study investigates whether geological catchment attributes derived from maps of increasing detail—global, continental, and regional—improve parameter transfer and model regionalization. To ensure robustness across model approaches, we applied both a semi-distributed process-based hydrological model using hydrological response units (HRUs) and a data-driven Long Short-Term Memory (LSTM) model. The analysis covered a total of 130 catchments in the Moselle (Central Europe) and Garonne (southwestern France) basins. We conducted five model experiments differing only in the representation of geological information: a benchmark without geology, a benchmark with random geology classes, and configurations based on the global-, continental-, and regional-scale geological maps. Model performance was evaluated using a modified Nash-Sutcliffe (NSE) metric for daily streamflow, as well as Pearson correlation and relative bias for three streamflow signatures: baseflow index, slope of the flow duration curve, and half-flow date. Across both basins and modeling frameworks, increasing geological detail consistently improved predictive performance under space–time evaluation. While differences in NSE were modest, improvements were pronounced for streamflow signatures: only models using the more detailed geological information, especially the regional map, consistently reproduced spatial variability in baseflow and flow regime characteristics. These findings highlight the importance of integrating high-quality geological data into hydrological modeling, particularly for improving predictions in ungauged basins through more reliable parameter transfer and regionalization.

Adaptation of ecosystems’ root zones to climate change critically affects drought resilience and vegetation productivity. However, a global quantitative assessment of this mechanism is missing. In this study, we analyzed high-quality observation-based data to find that the global average root zone water storage capacity (S_R) increased by 11%, from 182 to 202 mm in 1982–2020. The total increase of S_R equals to 1652 billion m³ over the past four decades. S_R increased in 9 out of 12 land cover types, while three relatively dry types experienced decreasing trends, potentially suggesting the crossing of ecosystems’ tipping points. Our results underscore the importance of accounting for root zone dynamics under climate change to assess drought impacts.

The rainfall-runoff transformation in catchments usually follows a variety of slower and faster flow paths, leading to a mixture of "younger"and "older"water in streamflow. Previous studies have investigated the time-variable distribution of water ages in streamflow (transit time distribution, TTD) using stable isotopes of water (δ 18O, δ 2H) together with transport models based on Storage Selection (SAS) functions. These functions are traditionally formulated based on soil moisture to mimic the preferential release of younger water as the system becomes wetter. In this study, we hypothesized that, in a heterogeneous catchment with a significant fast-runoff response component, precipitation intensity, in addition to soil moisture, plays a critical role in the preferential release of younger water. To test this hypothesis, we used high-resolution δ 18O data (weekly and event-based streamflow δ18O samples) in a 66 ha agricultural catchment. We tested two scenarios of the SAS function parameterization for the preferential-flow age selection: one as a function of soil moisture only and one as a function of both soil moisture and precipitation intensity. The results showed that accounting for both soil moisture and precipitation intensity to define the shape of SAS functions for preferential flow improved the tracer simulation in streamflow (increasing the Nash-Sutcliffe efficiency from 0.31 to 0.51). This also led to a higher percentage of streamflow (an increase from 2.87 % to 4.38 %) with shorter transit times (TTs younger than 7 d), with the largest differences occurring during the summer and autumn months. This was due to the fact that incorporating both soil wetness and precipitation intensity in the SAS formulation accounts for rapid flow pathways such as infiltration excess overland flow, preferential flow through macropores, and tile drain flow - allowing precipitation water to bypass much of the soil matrix and to reach the stream with minimal storage or mixing, even under dry soil conditions. We showed for the agricultural study catchment that a significant portion of event water bypasses the soil matrix through fast-flow paths, resulting in younger water reaching the stream for both low- and high-intensity precipitation. Thus, in catchments where preferential flows and overland flow are the dominant flow processes, soil-wetness-dependent and precipitation-intensity-conditional SAS functions may be required to better describe the timescale of solute transport in modelling, which has implications for stream water quality and agricultural management practices such as the timing of fertilizer application.

Large-sample hydrology datasets have advanced hydrological research, yet the impact of landscape map details on identifying dominant streamflow generation processes remains underexplored. This study investigates the role of geology using maps of increasing detail – global, continental, and regional – each reclassified into four permeability classes. These geological attributes were used along with topography, soil, land use, and climate attributes to identify dominant controls on streamflow signatures across 4469 European catchments. To distinguish landscape influences from the otherwise dominant influence of climate, we conducted separate analyses on nested basins. Three scales were considered to assess scale-dependent patterns: large (63 nested basins), intermediate (the Moselle nested basin), and small (five nested catchments within the Moselle). The large-scale study used geology information from global and continental maps, while the others also incorporated regional maps. At the large scale, dominant controls varied widely between nested basins, but landscape generally outweighed climate, highlighting the value of our nested basin design. At this scale, continental and global geology maps produced different correlation patterns, with neither consistently superior. At the intermediate scale, increased geological detail led geology to shift from the least to the most correlated variable for certain streamflow signatures. The small-scale experiment reinforced these findings, as the regional map highlighted controls more consistent with process understanding. This study underscores the benefit of integrating detailed, region-specific geological data into large sample hydrology studies, and demonstrates the utility of a nested basins design. These findings have important implications for hydrological regionalization and streamflow prediction in ungauged basins.

While measured streamflow is commonly used for hydrological model evaluation and calibration, an increasing amount of data on additional hydrological variables is available. These data have the potential to improve process consistency in hydrological modeling and consequently for predictions under change, as well as in data-scarce or ungauged regions. Here, we show how these hydrological data beyond streamflow are currently used for model evaluation and calibration. We consider storage and flux variables, namely snow, soil moisture, groundwater level, terrestrial water storage, evapotranspiration, and altimetric water level. We aim at summarizing the state-of-the-art and providing guidance for the use of additional hydrological variables for model evaluation and calibration. Based on a review of the current literature, we summarize observation methods and uncertainties of currently available data sets, challenges regarding their implementation, and benefits for model consistency. The focus is on catchment modeling studies with study areas ranging from a few km² to ~500,000 km². We discuss challenges for implementing alternative variables that are related to differences in the spatio-temporal resolution of observations and models, as well as to variable-specific features, for example, discrepancy between observed and simulated variables. We further discuss advancements required to deal with uncertainties of the hydrological data and to integrate multiple, potentially inconsistent datasets. The increased model consistency and improvement shown by most reviewed studies regarding the additional variables often come at the cost of a slight decrease in streamflow model performance.

The consistency of hydrological models, i.e. their ability to reproduce observed system dynamics, needs to be improved to increase their predictive power. As using streamflow data alone to calibrate models is not sufficient to constrain them and render them consistent, other strategies must be considered, in particular using additional types of data. The aim of this study was to test whether simultaneous calibration of dissolved organic carbon (DOC) and nitrate (NO3-) concentrations along with streamflow improved the hydrological consistency of a parsimonious solute-transport model. A multi-objective approach with four calibration scenarios was used to evaluate the model's predictions for an intensive agricultural headwater catchment. After calibration, the model reasonably simultaneously reproduced the dynamics of discharge and DOC and NO3- concentrations in the stream of the headwater catchment from 2008-2016. Evaluation using independent datasets indicated that the model usually reproduced dynamics of groundwater level and soil moisture in upslope and riparian zones correctly for all calibration scenarios. Using daily stream concentrations of DOC and NO3- along with streamflow to calibrate the model did not improve its ability to predict streamflow for calibration or evaluation periods. The approach significantly improved the representation of groundwater storage and to a lesser extent soil moisture in the upslope zone but not in the riparian zone. Parameter uncertainty decreased when the model was calibrated using solute concentrations, except for parameters related to fast and slow reservoir flow. This study shows the added value of using multiple types of data along with streamflow, in particular DOC and NO3- concentrations, to constrain hydrological models to improve representation of internal hydrological states and flows. With the increasing availability of solute data from catchment monitoring, this approach provides an objective way to improve the consistency of hydrological models that can be used with confidence to evaluate scenarios.

The new scientific decade (2023-2032) of the International Association of Hydrological Sciences (IAHS) aims at searching for sustainable solutions to undesired water conditions–whether it be too little, too much or too polluted. Many of the current issues originate from global change, while solutions to problems must embrace local understanding and context. The decade will explore the current water crises by searching for actionable knowledge within three themes: global and local interactions, sustainable solutions and innovative cross-cutting methods. We capitalise on previous IAHS Scientific Decades shaping a trilogy; from Hydrological Predictions (PUB) to Change and Interdisciplinarity (Panta Rhei) to Solutions (HELPING). The vision is to solve fundamental water-related environmental and societal problems by engaging with other disciplines and local stakeholders. The decade endorses mutual learning and co-creation to progress towards UN sustainable development goals. Hence, HELPING is a vehicle for putting science in action, driven by scientists working on local hydrology in coordination with local, regional, and global processes.

Coastal flooding events pose a critical risk in delta areas, since they are characterized by population growth and urban expansion. A better understanding of Extreme Water Levels (EWLs), the mechanisms generating them, and their components, i.e., astronomical tide and storm surge is of great importance as they drive the maintenance and design of flood protection systems. Therefore, a statistical investigation of them can provide new insights for more reliable flood risk mitigation infrastructures. In this study, we analyse these components and compare different probabilistic methods i.e., univariate extreme value analysis, copula functions, and Joint Probability Method (JPM) for the better estimation of EWLs. We use Hoek van Holland (NL) as a representative case study, since the dynamic conditions of this deltaic environment with man-made infrastructures render the area of strategic importance. The results indicate that a more accurate estimate of the declustering time between extreme events can be achieved using correlation of high surges and high wind speeds, taking into consideration also the wind direction. In the Southwest Delta this time estimated to be around 4 days. Furthermore, the EWLs components, i.e., surge and astronomical tide, show negative dependence. From the comparison between statistical approaches to model EWLs, results show that EWLs estimated via EVA and JPM do not vary significantly, while copulas’ seems to outperform the other methods. However, the selection of the proper copula to show the dependence is critical. As a conclusion, the analysis of the dependence between tides and storm surges can lead to more robust inferences of EWLs.

Large-sample hydrology datasets have become increasingly available, contributing to significant scientific advances. However, in Europe, only a few such datasets have been published, capturing only a fraction of the wealth of information from national data providers in terms of available spatial density and temporal extent. We present “EStreams”, an extensive dataset of hydro-climatic variables and landscape descriptors and a catalogue of openly available stream records for 17,130 European catchments. Spanning up to 120 years, the dataset includes streamflow indices, catchment-aggregated hydro-climatic signatures and landscape attributes (topography, soils, geology, vegetation and landcover). The catalogue provides detailed descriptions that allow users to directly access streamflow data sources, overcoming challenges related to data redistribution policies, language barriers and varied data portal structures. EStreams also provides Python scripts for data retrieval, aggregation and processing, making it dynamic in contrast to static datasets. This approach enables users to update their data as new records become available. Our goal is to extend current large-sample datasets and further integrate hydro-climatic and landscape data across Europe.