Extreme Events

The DAKI-FWS project highlights the critical role of bias correction and downscaling for accurate, localized assessments of extreme weather impacts. This approach is essential for generating reliable probabilistic warnings based on seasonal forecasts. Recent project phases have focused on refining data processing techniques to improve forecast precision.

The project implemented a robust AI-based architecture that significantly enhances spatial resolution, successfully increasing it up to 60 times for temperature, precipitation and other meteorological variables over Germany and adjacent areas. This improvement marks a major milestone in the field, achieving results beyond what is available in current scientific literature (November 2024).

The project’s downscaled meteorological variables maps closely align with reference data, maintaining a controlled bias of under 1 degree across most areas. This innovation marks a significant advancement in the precision of climate data modeling for Germany and beyond.