Wilke et al. (2025) Hail events in Germany: rare or frequent natural hazards?

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2025
• Date: 2025-09-09
• Authors: Tabea Wilke, Katharina Lengfeld, Markus Schultze
• DOI: 10.5194/nhess-25-3141-2025

Research Groups

Deutscher Wetterdienst (DWD), Offenbach, Germany

Short Summary

This study investigates hail characteristics across Germany using crowdsourced observations and C-band weather radar data to assess hail frequency, spatial distribution, and size variations. It reveals a north-south gradient in hail occurrence, with June being the peak month, and highlights the challenges and biases in human observations versus radar-derived estimates.

Objective

• To investigate hail characteristics (frequency, spatial distribution, and size variations) across Germany.
• To assess the reliability of size estimation in human-observed hail data.
• To determine the suitability of radar-based algorithms (Vertically Integrated Ice (VII) and Maximum Estimated Size of Hail (MESH)) for determining hail climatologies for Germany.
• To address whether hail events in Germany are rare or frequent natural hazards.

Study Configuration

• Spatial Scale: Germany and adjacent areas, covered by 17 C-band Doppler radars. Radar data analyzed at a horizontal resolution of 1° × 250 m, with VII analysis at 1 km × 1 km resolution.
• Temporal Scale:
  • Crowdsourced observations: 2000–2023 (ESWD, WarnWetter app, DWD staffed weather stations).
  • Weather radar data: 2018–2023, with a focus on the convective season (April to September).
  • Insurance data: 2013–2022, for the months of April to September.

Methodology and Data

• Models used: ICON-D2 (Icosahedral Nonhydrostatic) model for deriving vertical temperature profiles (e.g., melting layer height H0, -10 °C height H-10, -20 °C height H-20).
• Data sources:
  • Crowdsourced observations: WarnWetter app (DWD), European Severe Weather Database (ESWD), DWD staffed weather stations (total 50,719 reports).
  • Weather radar data: German C-band Doppler radar network (17 dual-polarization radars), providing 3D reflectivity data.
  • Insurance data: Hail damage claims from the German Insurance Association (GDV), including postal code, damage expense, and number of damages.
  • Perception study: Survey using 12 3D printed hailstones (0.5 cm, 1 cm, 2 cm, 3 cm, 5 cm, 7 cm diameter, round and oval shapes) with 149 participants.
  • Algorithms: Vertically Integrated Ice (VII) and Maximum Estimated Size of Hail (MESH), with custom calibration for the German C-band radar network.

Main Results

• Human Observation Reliability: Collective crowd estimates of hail size closely approximate actual measurements, but individual estimations can vary significantly and are more likely to underestimate size. Providing categorical options with reference objects improves accuracy. Crowdsourced data exhibits an urban reporting bias, with more reports from densely populated areas.
• Diurnal and Annual Cycles: Hail reports show a clear diurnal cycle, peaking around 15:00 UTC (17:00 CEST). The annual cycle for hail occurrence peaks in May and June for crowdsourced data, and predominantly in June for radar-derived VII data. Large to giant hail (exceeding 5 cm) is most likely to occur from May to August.
• Spatial Distribution: A clear north-south gradient in hail occurrence is observed, with southern Germany (e.g., Alpine foothills of Bavaria, Baden-Württemberg) experiencing substantially higher hail probabilities and frequency compared to northern regions. Mountainous areas also show increased hail frequency.
• Radar Algorithm Performance: MESH, even with custom calibration for German C-band radars, generally overestimates hail sizes compared to crowdsourced observations. VII performs better for hail sizes up to 3 cm, providing magnitudes similar to human observations. For larger hail sizes, MESH (75% percentile formulas) performs better due to its design to anticipate maximum possible hail diameter.
• Hail Frequency (VII, 2018–2023): 80.6% of Germany (within a 1 km × 1 km area) was hit by hail at least once during the 6-year period. The mean number of hail occurrences per year is approximately 0.33, with a maximum of 19 hail days in a single grid cell.
• Economic Impact (Insurance Data, 2013–2022): June and July are the months with the most damage reports and highest loss expenses, with June having a mean of 33,972 damage reports. This peak is slightly later than the peak in hail observations, likely due to larger hail sizes in summer causing more significant damage.

Contributions

• Provided a comprehensive, multi-source analysis of hail characteristics (frequency, size, spatial and temporal distribution) across Germany, integrating crowdsourced observations, C-band radar data, and insurance claims.
• Conducted a novel study using 3D printed hailstones to quantitatively assess the reliability and biases of human hail size estimations, highlighting the value of categorical reporting.
• Calibrated and validated radar-based hail detection algorithms (VII and MESH) specifically for the German C-band radar network, offering insights into their performance for climatological studies and demonstrating VII's better fit for smaller hail sizes.
• Confirmed and refined existing knowledge on hail climatology in Germany, including the pronounced north-south gradient, urban reporting bias, and distinct diurnal and annual cycles.
• Linked meteorological hail events with their economic impact using insurance data, providing a more holistic understanding of hail as a natural hazard in Germany.

Funding

• The German Insurance Association provided access to their hail damage data. No specific project or program funding codes were listed.

Citation

@article{Wilke2025Hail,
  author = {Wilke, Tabea and Lengfeld, Katharina and Schultze, Markus},
  title = {Hail events in Germany: rare or frequent natural hazards?},
  journal = {Natural hazards and earth system sciences},
  year = {2025},
  doi = {10.5194/nhess-25-3141-2025},
  url = {https://doi.org/10.5194/nhess-25-3141-2025}
}