HAIPI (Hailstorm Analysis, Impact, and Prediction Initiative)

Hail is one of the hazards associated with extreme convective events. It is one of the most expensive atmospheric hazards, and recent events have demonstrated this repeatedly with costly damage to vehicles, buildings, and agriculture. Hail is still one of the biggest challenges in forecasting, which is mainly due to the insufficient amount and quality of available data, together with the short spatio-temporal process scales. HAIPI aims to improve this situation integrating novel data sources to develop a product that estimates expected hail stone sizes through state-of-the-art machine learning algorithms. The focus is on crowd-sourced hail reports from the DWD WarnWetter-App as well as new dual-pol radar products, and existing products used in the DWD seamless forecast chain, e.g. KONRAD3D. As a first step, a routine for plausibility testing and quality control for the crowd-sourced data will be developed. The radar products will then be systematically evaluated, and uncertainties quantified. Using comparative machine learning analysis, an approach will be developed to predict expected hail stone sizes based on preprocessed input data. If proven skilful, a hail climatology for Germany will be derived from the resulting data product and the potential for impact assessments be evaluated. The outcomes will significantly advance the systematic observation and thus the prediction and warning of hail.

contact person: Schröer K
Email: katharina.schroeer@geographie.uni-freiburg.de

Start of project: 2024
End of project: 2027

Albert-Ludwigs-University Freiburg
Schröer K

Actual Research Report

• DWD EMF (extramurale Forschung)