Title: Improving Extreme Streamflow Simulations in Arid Chile: Linking High-Resolution Ensemble Precipitation to Hydrological Response

Authors: Lourdes Mendoza-Fierro, Hsin-I Chang, Christopher L. Castro, Rodrigo Valdés-Pineda, Christoforus Bayu Risanto.

Abstract: Northern Chile in recent years, has experienced extreme precipitation events driven by convective systems, and enhanced by the complex terrain, causing flash flooding, landslides, and associated adverse impacts. Predicting streamflow response to these events in northern Chile remains challenging due to limited in-situ observations, complex terrain, and the convective nature of storms. Current operational streamflow forecasting systems on the region rely on limited gauge availability data and satellite observations as input. For this work, we aim to improve stormflow simulations by using the Convective-Permitting Weather Research and Forecasting (CP-WRF) model products at 4-km resolution with the GR4J hydrological model. We analyze how precipitation forecasts translate into streamflow at different basins for individual extreme streamflow events. Simulations for each extreme event are driven by 20 ensemble members from the Global Ensemble Forecast System (GEFS), each spanning 120 forecast hours to capture conditions before and after the peak event. CP-WRF results are then used to force the GR5J, the commonly utilized bucket runoff model, to assess streamflow timing, magnitude, and variability across ensemble members. Results show that CP-WRF improves total precipitation representation over GEFS, directly enhancing the signal of peak streamflow simulations in arid catchments. Streamflow simulations forced with CP-WRF showed improvements in timing and magnitude, highlighting the value of incorporating high-resolution precipitation data for basins in complex arid regions.