Statistical-dynamical downscaling of wind climatologies

A statistical-dynamical downscaling procedure is applied for an investigation into the availability of wind power over a region of 80 x 87 km which covers flat and hilly terrain. The approach is based on the statistical coupling of a regionally representative wind climate with a numerical atmospheric mesoscale model. The large-scale wind climatology is calculated by a cluster-analysis of a time series of radiosonde data over 12 years. Any of the resulting 143 clusters represents a particular combination of geostrophic wind components and vertical temperature gradient. For each cluster, a highly resolved steady-state wind field is simulated with a non-hydrostatic mesoscale model. These wind fields are statistically evaluated by weighting them with the corresponding cluster frequency. The resulting three-dimensional wind field and the frequency distributions of windspeed and direction are compared with observations at synoptic stations.