Das Projekt "Regionalisierung" wird vom Umweltbundesamt gefördert und von Max-Planck-Institut für Meteorologie durchgeführt. The work will concentrate on the development and testing of nested high resolution limited area models (LAMs) and the testing of their skill in reproducing the present climate. A series of LAM simulations with 'perfect' boundary conditions (from analyses of observed data) will be carried out. The simulations will be validated against observed data using a number of standard and non-standard techniques. Experiments will be carried out with different parametrization schemes for the modelled physics, different treatments of the lateral boundaries, and different spatial resolutions. These experiments will give upper bounds for the accuracy to be expected for applications of the models in a climate change context. Two of the LAMs will be used to 'regionalize' climate simulations from atmospheric models (AGCM's at NCAR and at MPIM). The lateral boundaries will be taken from outputs of these AGCM's covering 10-20 years. The models are forced by prescribed (analysed) sea-surface temperatures (SSTs) and sea-ice fields. The simulations (both from the LAMs and the AGCMs) will be validated against observations and analyses for the period in question, and with climatology. A special study will investigate whether ocean SST (in casu the Atlantic SST's off the Portuguese coast) can be simulated realistically using an ocean limited area model (OLAM). Very high resolution climate simulations will be investigated by two specific studies. The first will use a statistical dynamical methodology to give very high resolution of the climate of the (greater) Alpine area. The methodology combines the use of a mesoscale dynamical model with a statistical classification scheme for the large scale flow regime. The second high resolution study is concentrating on the Iberian peninsula, and will investigate the further nesting into a (already nested) LAM of a higher resolution model.