Long-term variability and trends of the observed extreme warm and cold temperature events were analyzed in central and south-eastern Europe. A warming trend prevails in both the warm and cold events during the 20^th century, and changes in the frequency and persistence of atmospheric circulation patterns explain a substantial part of the variations. From the long-term perspective, the 1994 heat wave was found unprecedented in the Prague-Klementinum temperature series since the beginning of measurements (1775), and its return period was estimated to be on the order of hundred to thousand years. Objective circulation classification methodology was used in the study of the circulation patterns and the results were compared with those obtained with help of the subjective Hess-Brezowsky catalogue of Grosswetterlagen.

The investigation of Global Circulation Model (GCM) outputs acquired a large attention as the IPCC (Intergovernmental  Panel on Climate Change) established  the DDC (Data Distribution Centre) whose primary role was to provide timely information and data to the international climate research community. The data from several recent GCMs‘ runs and the gridded observed climate datasets for the land surface of the globe were downloaded via internet. A number of  studies of the GCMs‘ performance was implemented. The outputs of a few GCMs have been validated over the globe and in the region of the Czech Republic (in detail). A large part of the  procedure was focused on various temperature characteristics, including extreme warm and cold events in central Europe. In addition, the Koeppen classification was used as a diagnostic tool for studying GCMs‘ performance over the global scale. The classification allows to take simultaneously into account a combination of variables, their annual cycles and natural vegetation patterns.

In collaboration with Czech Hydrometeorological Institute and Charles University, the development of the Regional Climate Model, based on the numerical weather prediction model ALADIN, has commenced. The operational version of the ALADIN model was run for several month-long periods with only minor modifications necessary for the model to be run in a climatological mode. In the integrations, the model was nested in observations. The ALADIN model appears to be integrable beyond its predictability limits, capable of reproducing the real climate, and does not manifest any undesirable accumulation of systematic errors or increasing trends in random errors.

The construction of climate change scenarios for climate change impact assessments in selected experimental units in the Czech Republic was based on monthly series from  transient runs of seven GCMs. The outputs of the GCMs were validated with respect to annual cycles of global solar radiation, precipitation, daily average temperature and daily temperature range. The pattern scaling technique was used in the scenario construction: the scenario was defined by a product of the standardised scenario and the change of the global mean temperature. Changes in this temperature for a given emission scenario and climate sensitivity were calculated by  simple climate model MAGICC. Two groups of uncertainties in developing the climate change scenarios were studied. The first group deals with uncertainties in determining the standardised scenario, the second one concerns uncertainties in determining the change in the global mean temperature. The following recommendation has been formulated: a set of climate change scenarios representing the uncertainties of different types should be made use of  in the impact assessment. As a by-product of the impact studies implemented, the PERUN system has been developed. The system, created for the National Agency for Agricultural Research, is designed to provide the seasonal forecast of the crop yields. The probabilistic forecast is based on a crop growth model WOFOST and stochastic weather generator Met&Roll.

Figure: Uncertainties in changes of the daily mean temperature (TAVG), daily temperature range (DTR), precipitation (PREC) and solar radiation (SRAD) due to choice of site (site error) and model (GCM error), and the internal variability of HadCM model (HadCM error). The vertical bars represent avg±std range, where avg and std are calculated from: a) site error: the scenarios related to four exposure units (Prague, South Bohemia, South Moravia, Beskydy mountains), b) GCM error: seven GCM‑based scenarios averaged over the four sites (GCMs: CCSR/NIES, CGCM1, CSIRO-Mk2, ECHAM4/OPYC3, GFDL-R15-a, HadCM2, NCAR DOE-PCM), c) HadCM error: four integrations of the HadCM model. The changes displayed in the figure define the standardised scenario which relate changes in individual climatic characteristics with increase of the global mean temperature by 1 K. To obtain the scenario for a specific period and emission scenario, the changes must be multiplied by a prognosed change in a global mean temperature.