Dynamical Downscaling of Wind Speed in Complex Terrain Prone To Bora-Type Flows

Abstract

he results of numerically modeled wind speed climate, a primary component of wind energy resource assessment in the complex terrain of Croatia, are given. For that purpose, dynamical downscaling of 10 yr (1992?2001) of the 40-yr ECMWF Re-Analysis (ERA-40) was performed to 8-km horizontal grid spacing with the use of a spectral, prognostic full-physics model Aire Limitée Adaptation Dynamique Développement International (ALADIN; the ?ALHR? version). Then modeled data with a 60-min frequency were refined to 2-km horizontal grid spacing with a simplified and cost-effective model version, the so-called dynamical adaptation (DADA). The statistical verification of ERA-40-, ALHR-, and DADA-modeled wind speed on the basis of data from measurement stations representing different regions of Croatia suggests that downscaling was successful and that model accuracy generally improves as horizontal resolution is increased. The areas of the highest mean wind speeds correspond well to locations of frequent and strong bora flow as well as to the prominent mountain peaks. The best results are achieved with DADA and contain bias of 1% of the mean wind speed for eastern Croatia while reaching 10% for complex coastal terrain, mainly because of underestimation of the strongest winds. Root-mean-square errors for DADA are significantly smaller for flat terrain than for complex terrain, with relative values close to 12% of the mean wind speed regardless of the station location. Spectral analyses suggest that the shape of the kinetic energy spectra generally relaxes from k?3 at the upper troposphere to the shape of orographic spectra near the surface and shows no seasonal variability. Apart from the buildup of energy on smaller scales of motions, it is shown that mesoscale simulations contain a considerable amount of energy related to near-surface and mostly divergent meso-?-scale (20?200 km) motions. Spectral decomposition of measured and modeled data in temporal space indicates a reasonable performance of all model datasets in simulating the primary maximum of spectral power related to synoptic and larger-than-diurnal mesoscale motions, with somewhat increased accuracy of mesoscale model data. The primary improvement of dynamical adaptation was achieved for cross-mountain winds, whereas mixed results were found for along-mountain wind directions. Secondary diurnal and tertiary semidiurnal maxima are significantly better simulated with the mesoscale model for coastal stations but are somewhat more erroneous for the continental station. The mesoscale model data underestimate the spectral power of motions with less-than-semidiurnal periods.