AUTHOR=Merizalde María José , Muñoz Paul , Corzo Gerald , Muñoz David F. , Samaniego Esteban , Célleri Rolando TITLE=Integrating geographic data and the SCS-CN method with LSTM networks for enhanced runoff forecasting in a complex mountain basin JOURNAL=Frontiers in Water VOLUME=Volume 5 - 2023 YEAR=2023 URL=https://www.frontiersin.org/journals/water/articles/10.3389/frwa.2023.1233899 DOI=10.3389/frwa.2023.1233899 ISSN=2624-9375 ABSTRACT=Hydrological forecasting in complex mountain basins is a challenging task given the difficulty to characterize runoff generation processes in addition to data scarcity issues. The performance of deep learning forecasting models can be improved by leveraging available spatial rainfall datasets and incorporating process-based hydrological knowledge, both of them incorporating feature engineering (FE) strategies. In this study, we assessed the improvement of short-term runoff forecasting models using long short-term memory (LSTM) networks. The selected FE strategies were based on geographic data and the Soil Conservation Service Curve Number (SCS-CN) method. For this, we developed referential and specialized LSTM models for a 3390-km 2 basin using the GSMaP-NRT satellite precipitation product (SPP). We developed forecasting models for lead times of 1, 6, and 11 h to account for near-real-time forecasting, flash floods, and concentration time of the basin, respectively. Our results show that the proposed FE strategies improved the efficiencies of LSTM referential models for all lead times, with Nash-Sutcliffe efficiency values of 0.93 (1 h), 0.77 (6 h), and 0.67 (11 h). These results are comparable to other studies relying on ground-based precipitation information. The proposed methodology and insights derived from this study provide hydrologists with new tools for developing advanced data-driven runoff models that integrate available geographic information in other precipitation-ungauged hydrological systems.