ECSS2023-103
https://doi.org/10.5194/ecss2023-103
11th European Conference on Severe Storms
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Forecasting Large Hail using Additive Logistic Regression Models and the ECMWF Reforecasts

Francesco Battaglioli1,2, Pieter Groenemeijer1, Ivan Tsonevsky3, and Tomáš Púčik1
Francesco Battaglioli et al.
  • 1European Severe Storms Laboratory, Wessling, Germany (francesco.battaglioli@essl.org)
  • 2Institut für Meteorologie, Freie Universität Berlin, Berlin, Germany
  • 3European Centre for Medium Range Weather Forecasts, Reading, United Kingdom

An Additive Logistic Regression model for large hail (ARhail) was developed using convective parameters from the ERA5 reanalysis, hail reports from the European Severe Weather Database (ESWD) and lightning observations from the Met Office Arrival Time Difference network (ATDnet). This model was shown to accurately reproduce the climatological distribution and the seasonal cycle of observed hail events in Europe. To explore the value of this approach to medium-range forecasting, a similar four-dimensional model was developed using predictor parameters retrieved from the ECMWF reforecasts: Mixed Layer CAPE, Deep Layer Shear, Mixed Layer Mixing Ratio and the Wet Bulb Zero Height. This model was applied to ECMWF reforecasts to compute probabilistic large hail forecasts for all available 11 ensemble members, from 2008 to 2019 and for lead times up to 228 hours. First, we compared the hail ensemble forecasts for different lead times with observed hail occurrence from the ESWD focusing on a recent hail outbreak. Secondly, we systematically evaluated the model’s predictive skill depending on the forecast lead time using the Area under the ROC Curve (AUC) as a validation score. This analysis showed that ARhail has a very high predictive skill (AUC > 0.95) for forecasts up to 60 hours lead time. Although the performance scores progressively decrease with increasing lead time, ARhail retains a high predictive skill even for extended forecasts (AUC = 0.86 at 180 hours lead time) showing that it can provide useful guidance in hail forecasting well in advance. Finally, the performance of the four-dimensional model was compared with that of composite parameters such as the Significant Hail Parameter (SHP) and the product of CAPE and Deep Layer Shear (CAPESHEAR). Results show that ARhail outperforms CAPESHEAR (at all lead times) and SHP (especially at short lead times). This suggest that the developed Additive Logistic Regression model can improve hail forecasting compared to currently used composite indices in Europe.

How to cite: Battaglioli, F., Groenemeijer, P., Tsonevsky, I., and Púčik, T.: Forecasting Large Hail using Additive Logistic Regression Models and the ECMWF Reforecasts, 11th European Conference on Severe Storms, Bucharest, Romania, 8–12 May 2023, ECSS2023-103, https://doi.org/10.5194/ecss2023-103, 2023.