摘要 : Since about ten years monthly mean values of climatologic parameters have been interpolated to a grid for Germany with a density of 1 x 1 km as a product for climate monitoring by the Deutscher Wetterdienst. The quality of these i... 展开 Since about ten years monthly mean values of climatologic parameters have been interpolated to a grid for Germany with a density of 1 x 1 km as a product for climate monitoring by the Deutscher Wetterdienst. The quality of these interpolated grid fields has not been checked sufficiently up to now. Therefore a cross validation program has been developed for testing the quality of these grid fields. It calculates the differences between measured point values and the corresponding grid values resulting from the interpolated grid without using the measured value. With this cross validation program grid fields and measurements of temperature and precipitation have been compared for different periods (1901– 1910, about 189 stations; 1981 – 1990, about 570 stations). The differences have been analysed to get an index for the quality regarding means, absolute mean values of difference, standard deviation, variance and frequency. These statistic parameters have been examined in dependence of the elevation of the stations, too. The method of validation has been applied to all 10 year periods of the 20th century. For both parameters the mean error decreases with the increase of station density. Seasonal influence on the mean error is foremost found for precipitation. Mean absolute errors less or equal 0.42 K for mean monthly temperature and less than 17 % for mean monthly precipitation have been found. For the periods since 1950 mean absolute error for temperature is 0.3 –0.4 K. For precipitation the value is 6 % in winter and 13 % in summer. The mean errors have been compared with those given by several GIS interpolation methods. Especially cokriging procedures show partly insufficient results. Using more stations in the surrounding of the interpolation point, the mean error decreases, but the resulting interpolation is smoothed very much. Therefore statistical parameters and interpolation results indicate to continue to apply the interpolation method used by the Deutscher Wetterdienst up to now. 收起