We present smash, an open-source framework for high-resolution hydrological modeling and data assimilation. It combines process-based models with neural networks for regionalization, enabling accurate simulations from the catchment scale to the country scale. With an efficient, differentiable solver, smash supports large-scale calibration and parallel computing. Tested on open datasets, it shows strong performance in river flow prediction, making it a valuable tool for research and operational use.

This paper emphasizes the need to account for operational constraints when developing discharge forecast models. Using an open access dataset (CAMELS-US) for hydrology, two established rainfall-runoff models (LSTM and SAC-SMA), and a multilayer perceptron for implementation, we evaluate the importance of data assimilation, the persistence and ensemble analysis under various scenario. Results show DA is crucial, and models performances can sharply drop from idealized to operational conditions.

This paper describes a dataset that includes input, output, and validation data for the simulation of flash flood hazards and three specific flash flood events in the French Mediterranean region. This dataset is particularly valuable as flood mapping methods often lack sufficient benchmark data. Additionally, we demonstrate how the hydraulic method we used, named Floodos, produces highly satisfactory results.

This work targets the improvement of a hydrologic model used for flash flood warnings. A gridded model is used to spatially describe the hydrological processes. We develop a method to estimate the best model setup based on scarce river flow observations. It uses a complex algorithm combined with geographical descriptors to generate gridded parameters that better capture catchment characteristics. Results are promising, improving the discharge estimations where no observations are available.