Calibración de los pronósticos probabilísticos de precipitación derivados de un ensamble multi-modelo utilizando diferentes metodologias

probabilistic forecasts constitute a way to introduce the uncertainty in climate-weather forecasts. when probabilistic quantitative precipitation forecasts are derived from dynamically generated ensembles, a statistical post-processing or calibration should be done in order to improve forecasts reliability. the lack of forecast reliability is mainly due to systematic errors associated with each different ensemble member and from errors in the ensemble formulation. in this work, probabilistic forecasts derived from the university of san pablo multi model ensemble system have been calibrated and verified. several calibration strategies have been implemented and tested, including some that take into account systematic errors associated with individual ensemble members, as the bayesian model averaging technique (bma). results show that all the calibration strategies improve forecasts reliability. however, the simpler approach based on the ensemble mean shows, in general, the best results. an annual cycle has been found in pqpf skill over the region with higher skill scores during winter. during this same time of the year the impact of the pqpf calibration was smaller than in summer.