Adaptive Ensemble Deep Learning for Short-Term Precipitation Prediction in a Tropical Fluvial Basin under Hydroclimatic Variability

This dataset consists of hydrometeorological time series collected for the La Dorada region (Colombia), used for the development and validation of predictive models for precipitation in a tropical environment characterized by high climatic variability. Two versions of the dataset were used: 2001–2025 period: employed for initial model training during the early development phase. 2002–2026 period: an updated version including more recent data, used for model retraining, validation, and comparative performance assessment in the current phase. Both datasets include key climatic variables such as precipitation, temperature, and other relevant atmospheric parameters, structured as uniformly distributed time series.