Flood event dataset (discharge and precipitation) for 50 events in two contrasting basins: Upper Hei River and Luoma Lake

This dataset contains hydrometeorological observations for 50 independent flood events in two hydrologically contrasting basins: the Upper Hei River (mountainous semi-arid basin) and the Luoma Lake basin (lowland floodplain lake–river system). The dataset includes hourly precipitation and streamflow observations used for probabilistic flood forecasting and model evaluation in the study “A Physics-Constrained Bayesian Transformer for Probabilistic Flood Forecasting Across Contrasting River Basins”. Specifically, the dataset consists of:(1) precipitation time series,(2) streamflow records,(3) 50 event-based flood hydrographs for each basin. The data were used to train, validate, and test the proposed PCBT framework and benchmark models under 24-hour-ahead flood forecasting settings. This dataset supports both deterministic and probabilistic hydrological modeling and ensures reproducibility of all reported results in the associated manuscript. All data are provided in original time-series format for further research in physics-informed and uncertainty-aware hydrological modeling.