Deep Learning for Post-Processing Ensemble Weather Forecasts
This work addresses the high computational costs and accuracy limitations in weather forecasting, particularly for extreme events, offering a practical solution for meteorological applications.
The paper tackles the problem of reducing computational costs and improving accuracy in ensemble weather forecasts by proposing a mixed model that uses a subset of trajectories with deep neural network post-processing, achieving a 14% relative improvement in forecast skill (CRPS) and comparable results with fewer trajectories.
Quantifying uncertainty in weather forecasts is critical, especially for predicting extreme weather events. This is typically accomplished with ensemble prediction systems, which consist of many perturbed numerical weather simulations, or trajectories, run in parallel. These systems are associated with a high computational cost and often involve statistical post-processing steps to inexpensively improve their raw prediction qualities. We propose a mixed model that uses only a subset of the original weather trajectories combined with a post-processing step using deep neural networks. These enable the model to account for non-linear relationships that are not captured by current numerical models or post-processing methods. Applied to global data, our mixed models achieve a relative improvement in ensemble forecast skill (CRPS) of over 14%. Furthermore, we demonstrate that the improvement is larger for extreme weather events on select case studies. We also show that our post-processing can use fewer trajectories to achieve comparable results to the full ensemble. By using fewer trajectories, the computational costs of an ensemble prediction system can be reduced, allowing it to run at higher resolution and produce more accurate forecasts.