Comparison of three daily rainfall-runoff models with multiple regression analysis for catchment parameters in Biobío, Chile

This study analyzes the predictive capacity of three conceptual rainfall-runoff models (GR4J, GR5J, GR6J) in ungauged catchments within the Biobío Region, Chile. The study proposes a multiple regression-based regionalization approach using morphometric variables from nearby gauged catchments. Two unmonitored catchments were simulated due to their agroclimatic importance. The GR4J model excelled at simulating daily runoff, yielding satisfactory NSE, KGE, and RMSE values in most validations. Although GR5J and GR6J offered benefits in complex scenarios, they struggled with extreme precipitation events. Parameter regionalization, involving catchment area, mean slope, and shape factor, effectively transferred information, with R2 values from 0.68 to 0.99. Supported by a validated agroclimatic classification, this integrated approach underscores the value of combining regionalization and hydrological modeling in data-scarce catchments. Findings strengthen sustainable water resources management strategies and highlight the need for continued methods refinement under climate change and exacerbating aridity.