Statistical-dynamical downscaling of wind roses over the Czech Republic

Abstract

The present study describes a new method for statistical-dynamical downscaling that combines two different approaches, namely, a set of patterns simulated with a numerical flow model and a transformation function used to process both calculated data and measurements at a reference station. The combined method produces wind roses and wind speed histograms at an arbitrary location in the model domain. The inflow wind direction represented the key parameter to define a set of wind field simulations. The other two inflow parameters, namely, thermal stratification and geostrophic wind speed, were derived from corresponding averaged soundings. The results showed that in the Czech Republic, there are areas where wind roses are deformed by the surrounding terrain. The deformations occur in relatively shallow and wide valleys, and they are more sensitive to the inflow wind direction. Calculated wind roses are compared to corresponding observations at 22 synoptic stations. The most frequent wind direction sector in simulations agreed with measurements at 17 stations. The resulting error in frequency in that sector was under 5 % at 10 stations. In general, the main features of the wind roses are modelled well, even at a relatively large distance from the reference station. However, better performance was achieved for smaller distances between reference station and the site. In further studies, a more extensive set of flow patterns with reduced intervals of thermal stratification and wind speed will likely improve calculated wind roses.