Due to the coarse spatial resolution, the results from global climate models (GCM) are not capable to provide detailed regional estimations of future climate conditions. The 10-25 km horizontal resolution of the regional climate models (RCM) nested in GCMs is expected to improve the regional climate scenarios for European subregions. Expected regional climate change in the Carpathian basin (located in Central/Eastern Europe) is modelled by four different RCMs. Two of them (RegCM and PRECIS) are run by the Department of Meteorology, Eotvos Lorand University, Budapest, the other two RCMs are run by the Hungarian Meteorological Service: ALADIN (developed by the Meteo-France) and REMO (developed by the Max Planck Institute, Hamburg).

In this paper, control runs of RegCM and PRECIS are compared for the 1961-1990 period. For the validation, monthly data sets of the Climate Research Unit (CRU) of the University of East Anglia are used. Model PRECIS was developed at the UK Met Office, Hadley Centre, and it uses the boundary conditions from the European Centre for Medium-Range Weather Forecast (ECMWF) reanalysis datasets (ERA-40) and the HadCM3 GCM. The horizontal resolution of PRECIS is 25 km and it uses 19 vertical levels. Model RegCM3 was developed by Giorgi et al. and it is available from the ICTP (International Centre for Theoretical Physics). We used the ERA-40 Reanalysis data as boundary conditions. The horizontal resolution of RegCM is 10 km and it runs with 23 vertical levels. Both RCMs are 3-dimensional, sigma-coordinate, primitive equation models. The accomplished model experiments cover the entire Central/Eastern European region with special focus on the Carpathian basin. On the base of control runs, temperature and precipitation results are discussed. Furthermore, future climate of the Carpathian basin is also simulated by both RCMs: (1) PRECIS experiments are accomplished for 2071-2100 using A2 and B2 global scenarios, (2) RegCM experiments are accomplished for 2021-2050 and 2071-2100 using A1B global scenario.