Post-processing methods for calibrating the wind speed forecasts in central regions of Chile

Abstract

In this paper we propose some parametric and non-parametric post-processing methods for calibrating wind speed forecasts of nine Weather Research and Forecasting (WRF) models for locations around the cities of Valparaíso and Santiago de Chile (Chile). The WRF outputs are generated with different planetary boundary layers and land-surface model parametrizations and they are calibrated using observations from 37 monitoring stations. Statistical calibration is performed with the help of ensemble model output statistics and quantile regression forest (QRF) methods both with regional and semi-local approaches. The best performance is obtained by the QRF using a semilocal approach and considering some specific weather variables from WRF simulations. ∗This research was partially supported by the Interdisciplinary Center of Atmospheric and Astro-Statistical Studies, University of Valparaíso, Chile. Annales Mathematicae et Informaticae 53 (2021) pp. 93–108 doi: https://doi.org/10.33039/ami.2021.03.012 url: https://ami.uni-eszterhazy.hu