Evaluation of a dynamic-diagnostic modelling approach to generate highly resolved wind fields in the Alpine region

Abstract

For generating highly resolved wind fields in the Alpine region the presented work focuses on the evaluation of a hybrid dynamic-diagnostic downscaling procedure. The diagnostic model CALMET is driven by the dynamic model MM5, which is nested into ECMWF's re-analysis ERA-40. Near surface winds are downscaled via multiple nesting from ∼120 km to 200 m grid spacing without ingestion of any further observational data for the test period between 7 September to 15 November 1999 within a mountainous study area (140 km × 70 km) located in the eastern Alps (Hohe Tauern). Two MM5 grid spacings for driving the diagnostic model were evaluated, 5 km and 10 km, respectively. Detailed error statistics based on observations from surface stations show drastic improvements of modelled air flows compared to the grossly deviating driving data (ERA-40). The overall bias relative to the station-averaged observed mean wind speed of 5.8 m/s is systematically reduced from −4.1 m/s to −0.7 m/s. In general, low wind speeds are slightly overestimated, while higher wind speeds are increasingly underestimated. A reduction of the finest horizontal grid resolution of the dynamic model from 5 km to 10 km, which would be computationally favourable, induces additional errors with respect to unresolved air flows, which the diagnostic model is unable to correct in most cases. Important wind climatologic characteristics (e.g., bimodal frequency distributions) disappear, which accentuates the importance of high-quality, high-resolution initial wind fields for diagnostic models operating in complex terrains.