Recent advances in time series research facilitate the development of foundation models. While many state-of-the-art time series foundation models have been introduced, few studies examine their effectiveness in specific downstream applications in physical science. This work investigates the role of integrating domain knowledge into time series models for hydrological rainfall-runoff modeling. Using the CAMELS-US dataset, which includes rainfall and runoff data from 671 locations with six time series streams and 30 static features, we compare baseline and foundation models. Results demonstrate that models incorporating comprehensive known exogenous inputs outperform more limited approaches, including foundation models. Notably, incorporating natural annual periodic time series contribute the most significant improvements.

Streamflow prediction is one of the key challenges in the field of hydrology due to the complex interplay between multiple non-linear physical mechanisms behind streamflow generation. While physics based models are rooted in rich understanding of the physical processes, a significant performance gap still remains which can be potentially addressed by leveraging the recent advances in machine learning. The goal of this work is to incorporate our understanding of hydrological processes and constraints into machine learning algorithms to improve the predictive performance. Traditional ML models for this problem predict streamflow using weather drivers as input. However there are multiple intermediate processes that interact to generate streamflow from weather drivers. The key idea of the approach is to explicitly model these intermediate processes that connect weather drivers to streamflow using a multi-task learning framework. While our proposed approach requires data about intermediate processes during training, only weather drivers will be needed to predict the streamflow during testing phase. We assess the efficacy of the approach on a simulation dataset generated by the SWAT model for a catchment located in the South Branch of the Root River Watershed in southeast Minnesota. While the focus of this paper is on improving the performance given data from a single catchment, methodology presented here is applicable to ML-based approaches that use data from multiple catchments to improve performance of each individual catchment.

The reliability of routine health data in low and middle-income countries (LMICs) is often constrained by reporting delays and incomplete coverage, necessitating the exploration of novel data sources and analytics. Geospatial Foundation Models (GeoFMs) offer a promising avenue by synthesizing diverse spatial, temporal, and behavioral data into mathematical embeddings that can be efficiently used for downstream prediction tasks. This study evaluated the predictive performance of three GeoFM embedding sources - Google Population Dynamics Foundation Model (PDFM), Google AlphaEarth (derived from satellite imagery), and mobile phone call detail records (CDR) - for modeling 15 routine health programmatic outputs in Malawi, and compared their utility to traditional geospatial interpolation methods. We used XGBoost models on data from 552 health catchment areas (January 2021-May 2023), assessing performance with R2, and using an 80/20 training and test data split with 5-fold cross-validation used in training. While predictive performance was mixed, the embedding-based approaches improved upon baseline geostatistical methods in 13 of 15 (87%) indicators tested. A Multi-GeoFM model integrating all three embedding sources produced the most robust predictions, achieving average 5-fold cross validated R2 values for indicators like population density (0.63), new HIV cases (0.57), and child vaccinations (0.47) and test set R2 of 0.64, 0.68, and 0.55, respectively. Prediction was poor for prediction targets with low primary data availability, such as TB and malnutrition cases. These results demonstrate that GeoFM embeddings imbue a modest predictive improvement for select health and demographic outcomes in an LMIC context. We conclude that the integration of multiple GeoFM sources is an efficient and valuable tool for supplementing and strengthening constrained routine health information systems.

Reliable hydrologic and flood forecasting requires models that remain stable when input data are delayed, missing, or inconsistent. However, most advances in rainfall-runoff prediction have been evaluated under ideal data conditions, emphasizing accuracy rather than operational resilience. Here, we develop an operationally ready emulator of the Global Flood Awareness System (GloFAS) that couples long-and short-term memory networks with a relaxed water-balance constraint to preserve physical coherence. Five architectures span a continuum of information availability: from complete historical and forecast forcings to scenarios with data latency and outages, allowing systematic evaluation of robustness. Trained in minimally managed catchments across the United States and tested in more than 5,000 basins, including heavily regulated rivers in India, the emulator reproduces the hydrological core of GloFAS and degrades smoothly as information quality declines. The framework establishes operational robustness as a measurable property of hydrological machine learning and advances the design of reliable real-time forecasting systems. Catchment response to precipitation varies in space and time with climate, storage dynamics, and human regulation, making reliable prediction dependent on both data availability and model adaptability [3, 4]. Although advances in observations, reanalysis products, and computational methods have expanded predictive capability [5-9], translating this progress into forecasting systems that operate continuously and robustly in real time remains unresolved. Operational forecasting requires models that sustain accuracy and physical realism when input data are asynchronous, incomplete, or inconsistent with the conditions used for training, and that can do so with limited human intervention [10-12].

Conceptual rainfall-runoff models aid hydrologists and climate scientists in modelling streamflow to inform water management practices. Recent advances in deep learning have unravelled the potential for combining hydrological models with deep learning models for better interpretability and improved predictive performance. In our previous work, we introduced DeepGR4J, which enhanced the GR4J conceptual rainfall-runoff model using a deep learning model to serve as a surrogate for the routing component. DeepGR4J had an improved rainfall-runoff prediction accuracy, particularly in arid catchments. Quantile regression models have been extensively used for quantifying uncertainty while aiding extreme value forecasting. In this paper, we extend DeepGR4J using a quantile regression-based ensemble learning framework to quantify uncertainty in streamflow prediction. We also leverage the uncertainty bounds to identify extreme flow events potentially leading to flooding. We further extend the model to multi-step streamflow predictions for uncertainty bounds. We design experiments for a detailed evaluation of the proposed framework using the CAMELS-Aus dataset. The results show that our proposed Quantile DeepGR4J framework improves the predictive accuracy and uncertainty interval quality (interval score) compared to baseline deep learning models. Furthermore, we carry out flood risk evaluation using Quantile DeepGR4J, and the results demonstrate its suitability as an early warning system.

Combining attention with recurrence has shown to be valuable in sequence modeling, including hydrological predictions. Here, we explore the strength of Temporal Fusion Transformers (TFTs) over Long Short-Term Memory (LSTM) networks in rainfall-runoff modeling. We train ten randomly initialized models, TFT and LSTM, for 531 CAMELS catchments in the US. We repeat the experiment with five subsets of the Caravan dataset, each representing catchments in the US, Australia, Brazil, Great Britain, and Chile. Then, the performance of the models, their variability regarding the catchment attributes, and the difference according to the datasets are assessed. Our findings show that TFT slightly outperforms LSTM, especially in simulating the midsection and peak of hydrographs. Furthermore, we show the ability of TFT to handle longer sequences and why it can be a better candidate for higher or larger catchments. Being an explainable AI technique, TFT identifies the key dynamic and static variables, providing valuable scientific insights. However, both TFT and LSTM exhibit a considerable drop in performance with the Caravan dataset, indicating possible data quality issues. Overall, the study highlights the potential of TFT in improving hydrological modeling and understanding.

We present a dataset for rainfall streamflow modeling that is fully spatially resolved with the aim of taking neural network-driven hydrological modeling beyond lumped catchments. To this end, we compiled data covering five river basins in central Europe: upper Danube, Elbe, Oder, Rhine, and Weser. The dataset contains meteorological forcings, as well as ancillary information on soil, rock, land cover, and orography. The data is harmonized to a regular 9km times 9km grid and contains daily values that span from October 1981 to September 2011. We also provide code to further combine our dataset with publicly available river discharge data for end-to-end rainfall streamflow modeling.

The Caravan large-sample hydrology dataset (Kratzert et al., 2023) was created to standardize and harmonize streamflow data from various regional datasets, combined with globally available meteorological forcing and catchment attributes. This community-driven project also allows researchers to conveniently extend the dataset for additional basins, as done 6 times to date (see https://github.com/kratzert/Caravan/discussions/10). We present a novel extension to Caravan, focusing on enriching the meteorological forcing data. Our extension adds three precipitation nowcast products (CPC, IMERG v07 Early, and CHIRPS) and three weather forecast products (ECMWF IFS HRES, GraphCast, and CHIRPS-GEFS) to the existing ERA5-Land reanalysis data. The inclusion of diverse data sources, particularly weather forecasts, enables more robust evaluation and benchmarking of hydrological models, especially for real-time forecasting scenarios. To the best of our knowledge, this extension makes Caravan the first large-sample hydrology dataset to incorporate weather forecast data, significantly enhancing its capabilities and fostering advancements in hydrological research, benchmarking, and real-time hydrologic forecasting. The data is publicly available under a CC-BY-4.0 license on Zenodo in two parts (https://zenodo.org/records/14161235, https://zenodo.org/records/14161281) and on Google Cloud Platform (GCP) - see more under the Data Availability chapter.

The prediction of streamflows and other environmental variables in unmonitored basins is a grand challenge in hydrology. Recent machine learning (ML) models can harness vast datasets for accurate predictions at large spatial scales. However, there are open questions regarding model design and data needed for inputs and training to improve performance. This study explores these questions while demonstrating the ability of deep learning models to make accurate stream temperature predictions in unmonitored basins across the conterminous United States. First, we compare top-down models that utilize data from a large number of basins with bottom-up methods that transfer ML models built on local sites, reflecting traditional regionalization techniques. We also evaluate an intermediary grouped modeling approach that categorizes sites based on regional co-location or similarity of catchment characteristics. Second, we evaluate trade-offs between model complexity, prediction accuracy, and applicability for more target locations by systematically removing inputs. We then examine model performance when additional training data becomes available due to reductions in input requirements. Our results suggest that top-down models significantly outperform bottom-up and grouped models. Moreover, it is possible to get acceptable accuracy by reducing both dynamic and static inputs enabling predictions for more sites with lower model complexity and computational needs. From detailed error analysis, we determined that the models are more accurate for sites primarily controlled by air temperatures compared to locations impacted by groundwater and dams. By addressing these questions, this research offers a comprehensive perspective on optimizing ML model design for accurate predictions in unmonitored regions.

Long Short Term Memory networks (LSTMs) are used to build single models that predict river discharge across many catchments. These models offer greater accuracy than models trained on each catchment independently if using the same data. However, the same data is rarely available for all catchments. This prevents the use of variables available only in some catchments, such as historic river discharge or upstream discharge. The only existing method that allows for optional variables requires all variables to be considered in the initial training of the model, limiting its transferability to new catchments. To address this limitation, we develop the Hydra-LSTM. The Hydra-LSTM processes variables used across all catchments and variables used in only some catchments separately to allow general training and use of catchment-specific data in individual catchments. The bulk of the model can be shared across catchments, maintaining the benefits of multi-catchment models to generalise, while also benefitting from the advantages of using bespoke data. We apply this methodology to 1 day-ahead river discharge prediction in the Western US, as next-day river discharge prediction is the first step towards prediction across longer time scales. We obtain state-of-the-art performance, generating more accurate median and quantile predictions than Multi-Catchment and Single-Catchment LSTMs while allowing local forecasters to easily introduce and remove variables from their prediction set. We test the ability of the Hydra-LSTM to incorporate catchment-specific data by introducing historical river discharge as a catchment-specific input, outperforming state-of-the-art models without needing to train an entirely new model.