Toward an operational streamflow forecasting system in Mexico: a pioneering ensemble approach with WRF-Hydro in a tropical mountain basin

Operational ensemble streamflow forecasting is widely applied internationally but remains rare in Mexico. This study develops an ensemble-based framework for the La Sierra River basins in southeastern Mexico, using WRF-Hydro coupled with the WRF atmospheric model. Ensemble members are generated by varying cumulus parameterizations to represent atmospheric uncertainty, providing a probabilistic basis for risk-informed decision making. Performance is assessed with standard metrics (bias, Kling-Gupta efficiency, Nash-Sutcliffe efficiency, continuous ranked probability score, and skill score) against traditional deterministic methods. Results show that ensemble forecasts perform well when precipitation spread captures observed events, particularly in larger basins, but skill declines in small, steep basins or when storms are poorly represented. This first application of ensemble forecasting in Latin America, based on a high-resolution and physically-based framework for quantifying forecast uncertainty, demonstrates its potential and establishes a foundation for future advances through data assimilation and bias correction.