A weather generator based on the European 'Grosswetterlagen'

Abstract

The investigation of impacts of a changed climate on social and natural systems usually requires daily meteorological data on a spatial scale of about 1 to 100 km2. Even the strongest tools for the estimation of the changes in future climate, the Global Climate Models (GCMs), are not yet able to provide regional scenarios of climate change at this resolution. Presently, GCMs operate at spatial resolutions of several hundred kilometers far too coarse for most regional analyses. Therefore, 'downscaling' approaches are needed in order to transform GCM climate change scenarios to appropriate scales for climate impact investigations. A statistical approach is developed based on the relationships between variations in regional climate and changes in large-scale circulation over the North Atlantic Ocean and Europe. Using the well-known classification scheme of circulation types, the European 'Grosswetterlagen', a conditional climatology for different large-scale air flow patterns was calculated using long daily meteorological data sets at 3 weather stations in Germany. The coupling of flow direction and climate at a site of interest varies with season and with c~rculation pattern. Even a single weather type produces relationships that are highly variable. A possible explanation could be the interaction of changes in mean duration time, preceding patterns or ~ h d r d ~ t e r i ~ t i ~ weather conditions caused by the weather type. To account for this natural variabil~ty an approach has been developed of stochastically modelling the linkage between daily weather type and regional daily weather Using long observed time series of daily surface air temperatures and precipitation, the model was calibrated and validated. Comparison of observed and simulated data demonstrated the ability of the model to adequately explain the statistical structure of the daily time series at 3 German sites. Possible applications of this model include sensitivity studies, downscaling of GCM output and investigation of the frequency and duration of extreme events.