Using highly resolved maximum gust speed as predictor for forest storm damage caused by the high-impact winter storm Lothar in Southwest Germany

Abstract

Results from a newly available empirical maximum gust speed model were evaluated for their predictive power of forest storm damage caused by the high-impact winter storm ‘Lothar’ in the German federal state of Baden-Wuerttemberg. In this state, Lothar was the most severe storm event of the last decades, causing nearly 30 million m3 of damaged timber. By applying a least squares boosting procedure, daily maximum gust speed values measured at 28 meteorological stations were used to empirically model highly resolved (50 × 50 m) near-surface gust speed fields associated with Lothar. Gust speed was modelled using terrain- and roughness-related variables as predictors. The modelled gust speed fields were then used as input to an empirical forest storm damage model. To build the damage model, the machine learning method random forests was applied. Results from this study demonstrate that the empirically modelled maximum gust speed field associated with Lothar was the most important predictor variable for forest storm damage at the landscape scale. Modelled maximum gust speed was nearly twice as important as all other available predictor variables. The medians of classified proportions of storm-damaged timber quasi-linearly increased with increasing maximum gust speed.