Empirical downscaling method as a tool for development of regional climate change scenario

Abstract

Results of simulations with general circulation models (GCM) are the basis for the future climate change and impact studies. In our case, results of simulations with four GCM (CSIRO/Mk2, UKMO/HadCM3, DOE-NCAR/PCM and MPI-DMI/ECHAM4-OPYC3), based on SRES A2 and B2 marker emission scenarios, were used. The mean monthly values of near ground air temperature and sea level pressure were selected for the period 1951-2100 and used as predictor values. The results of GCM were projected to near ground air temperature and precipitation amount at five locations in Slovenia (Ljubljana, Novo mesto, Murska Sobota, Rateče, and Bilje) by using empirical downscaling. Different regression techniques were used for the development of empirical downscaling models (EM) that relate selected large-scale predictors with local-scale predictands. The EM were developed by means of data from ARSO archive and NCEP/NCAR reanalysis data for the period 1951-2002. Derived EM explained a great part of variability of local air temperature at all five locations. This was not the case for the local precipitation, where the quality of EM is acceptable only for the months of the cold half of the year. Local projections of GCM results by EM were additionally scaled to other (A1Fl, A1B, A1T, and B1) marker SRES scenarios and used as a base for the development of regional climate change scenarios.