Seminar, Hosted by World Meteorological Organization
7th WMO AI Webinar - good practices of ECMWF and SMHI
Maria Luisa Taccari and Yiheng Du
ECMWF; Swedish Meteorological and Hydrological Institute (SMHI)
Abstract
Two forecasting teams examine how machine learning can improve streamflow prediction at different scales. Maria Luisa Taccari explains how ECMWF’s AIFL forecasts global streamflow up to 15 days ahead and how regional services can adapt it without retraining the full model. Yiheng Du describes how SMHI combines AI with the process-based E-HYPE model across local hydrological conditions and what remains to bring the approach into operational flood warning.