Deep Learning for Post-Processing Ensemble Weather Forecasts

T. Hoefler, ETH Zurich (CH)

Abstract: Quantifying uncertainty in weather forecasts typically employs ensemble prediction systems, which consist of many perturbed trajectories run in parallel. These systems are associated with a high computational cost and often include statistical post-processing steps to inexpensively improve their raw prediction qualities. The talk proposed a mixed prediction and post-processing model based on a subset of the original trajectories. The model implemented deep learning methods to account for non-linear relationships that are not captured by current numerical models or other post-processing methods. Applied to global data, these mixed models achieve a relative improvement of the ensemble forecast skill of over 13%. The talk demonstrated that this is especially the case for extreme weather events on selected case studies, showing an improvement in predictions by up to 26%. It was also shown that, by using only half the trajectories, the computational costs of ensemble prediction systems can potentially be reduced, allowing weather forecasting pipelines to run higher resolution trajectories, and resulting in even more accurate raw ensemble forecasts.