This webinar is co-hosted together with ECMWF and will be longer than usual, approximately 1 hour. Otherwise the procedure is as usual.
Speaker: Tilmann Gneiting, University of Heidelberg
Date and time: Wednesday, February 11, 2015 15:30 (UTC)
Link to the recording here
Abstract: Statistical post-processing techniques serve to improve the quality of numerical weather forecasts, as they seek to generate calibrated and sharp predictive distributions of future weather quantities and events. I will review the state of the art in post-processing, with focus on ensemble forecasts and ongoing joint work between the ECMWF and the Computational Statistics group at the Heidelberg Institute for Theoretical Studies (HITS). Current and future challenges include the treatment of extreme events, and the calibration of probabilistic forecasts of combined events and spatio-temporal weather trajectories, for which discrete copula based techniques, such as ensemble copula coupling (ECC) and the Schaake shuffle, are attractive options.
About the speaker