Simulation of extreme wind climate for landscape modelling of a forested complex-terrain area

Extreme wind events are among the most damaging weather-related hazards in mid-latitudes. In forested areas, the windstorms cause frequent disturbances to the vegetation cover, which disrupts the landscape dynamics. In order to successfully run a landscape model dealing with such effects, spatial distribution of extreme wind speed statistics is needed. Suggested method employs the non-hydrostatic model WRF to dynamically downscale selected extreme wind events.

The events to be simulated were selected using classification of ERA Interim reanalysis pressure-level data at the center of the domain. The situations were clustered using the wind speed and direction at 850 hPa level and thermal stability expressed as temperature gradient between 850 hPa and 925 hPa levels. One representative event for each cluster was downscaled with WRF. Several parameterization schemes were tested for calculation of radiation and surface layer to properly address orographic effects in the domain. Finally, the results of simulation were statistically processed to obtain parameters of extreme value distribution in each grid point of the domain. In the landscape model, probability of windthrow was calculated from randomly generated annual maxima. The corresponding function was calibrated using data of real forest damage due to the extreme wind speed.

The method is demonstrated on the area of Bohemian Forest that represents one of largest and most compact forested mountains in Central Europe. The selected domain contains a professional meteorological station with suitable history of wind speed measurements. The observations were cleaned of inhomogeneity and classified to convective and non-convective cases using the index CAPE. The resulting dataset allowed validation and calibration of previously downscaled data. The simulation was also done for the storm Kyrill to be compared to the reported forest damage.