The aim of this course is to show participants the EMOS (ensemble mean output statistics) methodology to improve the short and medium term forecasts of the European Centre for Medium-Range Weather Forecasts (ECMWF) IFS (integrated forecasting system). To do this, participants learn how to download ECMWF forecasts and generate a local forecast by calibrating them with observed data from weather stations. This methodology also allows incorporating multiple models, for example local models at national level, and also generating spatial forecasts thanks to the incorporation of topographic descriptors (global EMOS).
Participants will learn how to access IFS forecasts and will be able to obtain an ECMWF license provided by MeteoSwiss. This license allows participants to use the ECMWF api via Python to access all available data for ECMWF member states. The license will be available to participants also after the course. Participants will learn the theoretical foundations of the EMOS method, as well as the global EMOS, and will participate in an exercise that they will later be able to adapt and use on data from their own country. In a second part, the course will include two presentations on machine learning (ML) at MeteoSwiss.
For more information about this course please contact rmezher@smn.gob.ar