Jimeno‐Sáez et al. (2026) Coupling SWAT+ with machine learning-enhanced ECMWF seasonal forecasts for predicting inflows to Entrepeñas and Buendía reservoirs

Identification

• Journal: Environmental Modelling & Software
• Year: 2026
• Date: 2026-06-26
• Authors: Patricia Jimeno‐Sáez, Sara Asadi, Nerea Bilbao-Barrenetxea, Gerardo Castellanos-Osorio, Adrián López-Ballesteros, Javier Senent‐Aparicio
• DOI: 10.1016/j.envsoft.2026.107094

Research Groups

• Department of Civil Engineering, Catholic University of San Antonio
• Centro de Investigaciones sobre Desertificaci´on (CIDE), CSIC-UV-GVA
• Faculty of Engineering of Bilbao, University of the Basque Country (UPV/EHU)
• Department of Computer Engineering, University Polytechnic of Valencia

Short Summary

This study develops a seasonal hydrological forecasting framework using ECMWF's SEAS5 ensemble climate predictions. Bias-corrected forecasts with machine learning generate daily precipitation and temperatures up to seven months ahead, driving a calibrated SWAT+ model to simulate reservoir inflows.

Objective

• Develop a reliable seasonal hydrological forecasting system for predicting inflows to Entrepe˜nas and Buendía reservoirs in the Upper Tagus River Basin (UTRB).

Study Configuration

• Spatial Scale: The study area covers an area of 7418 km2, including the UTRB.
• Temporal Scale: The analysis spans from 1981 to 2019.

Methodology and Data

• Models used: SWAT+ hydrological model, LSTM-based bias correction for seasonal forecasts.
• Data sources: ECMWF SEAS5 (C3SV.5.1) seasonal forecast model, observed climate data from the Spanish National Meteorological Agency (AEMET), and observed inflows of the Entrepe˜nas and Buendía reservoirs.

Main Results

• The study demonstrates the effectiveness of integrating seasonal climate forecasts with a hydrological model for predicting reservoir inflows.
• Bias-corrected forecasts show improved performance compared to raw forecasts, with significant reductions in bias across all variables and sub-basins.
• The CRPSS values indicate that the bias-corrected forecasts have superior predictive skill compared to the reference system.

Contributions

• This study contributes to the development of reliable seasonal hydrological forecasting systems for water resource management in drought-prone basins.
• The integration of LSTM-based bias correction with a hydrological model demonstrates improved performance and accuracy in predicting reservoir inflows.

Funding

• This research was funded by [project 1], [project 2], and [reference code].

Citation

@article{JimenoSáez2026Coupling,
  author = {Jimeno‐Sáez, Patricia and Asadi, Sara and Bilbao-Barrenetxea, Nerea and Castellanos-Osorio, Gerardo and López-Ballesteros, Adrián and Senent‐Aparicio, Javier},
  title = {Coupling SWAT+ with machine learning-enhanced ECMWF seasonal forecasts for predicting inflows to Entrepeñas and Buendía reservoirs},
  journal = {Environmental Modelling & Software},
  year = {2026},
  doi = {10.1016/j.envsoft.2026.107094},
  url = {https://doi.org/10.1016/j.envsoft.2026.107094}
}