Evaluación de estimaciones satelitales de precipitación en la cuenca del río Jubones

Abstract

This study evaluates the ability of satellite precipitation products (CHIRPS, GPM-IMERG, TRMM, and reanalyses such as ERA5) to represent the spatial and temporal variability of precipitation in the Jubones River basin (southwest coast of Ecuador). Using observational series from local rain gauge stations (1981–2019/2022), verification metrics were applied: Pearson correlation coefficient, mean absolute error (MAE), RMSE, percentage bias, and skill tests to detect extreme events (rainy day, p95/p99). In addition, an ENSO dependence analysis and spatial VALIDATION by subbasin and by altitudinal range were performed. The results show that CHIRPS generally provides better estimates in monthly accumulated series and in regions with few stations, while IMERG/IMERG-Final shows better performance in daily detection and volumes in intense events; its effectiveness varies with the temporal scale and local orography. The implications for hydrological applications (modeling with SWAT or HBV), the need for local bias correction, and gauge-satellite fusion strategies to reduce uncertainties in simulated streamflows are discussed. Finally, reproducible validation protocols and criteria for selecting satellite products based on the application (climate monitoring vs. operational hydrological modeling) are proposed.