Dryer years and brighter sky - the predictable simulation outcomes for Germany's warmer climate from the weather resampling model STARS

Abstract

The supposed statistical regional climate model STARS, which is based on a resampling approach, was evaluated. Results from a theoretical analysis, its recent application for Germany, and a model intercomparison experiment for Germany were used. The study shows that the conditioning of the resampling process by a prescribed temperature trend as realized in STARS leads to changes in its covariables such as precipitation and global radiation that are dominated by the (immediate) interannual correlations between temperature and those variables, and not by the correlation between their temporal trends. Only in the latter case, the climatic past would be extrapolated into a scenario future. Instead, the typical linkages between warmer summers, higher global radiation and lower precipitation in historical climate trigger corresponding changes in the resamples if conditioned for warmer summers. For the winter months, the association of past warmer winters with higher precipitation establishes a preference of the corresponding resamples for future wetter winters. Owing to the dominance of summer over winter correlations at the sub-annual resampling levels, the resampled German climate becomes dryer and associated with brighter skies and higher global radiation levels. According to its concept, STARS-based climate projections turn short-term interannual variability between temperature and covariables into long-term climate trends. However, these projections are conceptually invalid, empirically often disconnected from past climate trends, implausible taken the extent of possible future changes projected by global circulation models (GCM), and diverge from GCM projections with an identical sample basis. Therefore, recent conclusions for adaptation drawn from the STARS-based warming scenarios might be misleading and should be reconsidered as vulnerable illustrations for a selected climate change.