AN ATTEMPT TO EVALUATE HAIL SUPPRESSION IN CROATIA

The number of hail days in the warm period of the year, June-September, at meteorological stations, is used for the evaluation of hail suppression activities in north-western Croatia. The mean number of hail days at the station is connected with the mean number of hail days in the experimental region and the mean area covered by individual hail falls in the region (Long, 1980). The assumptions of this analysis are that the regional frequency of hail days cannot be changed by hail suppression and that the change of the mean area covered by individual hail falls is a consequence of hail suppression. Multivariate linear regression has been chosen as the method of analysis. The frequency of hail days at stations is a dependent variable, and the year, the number of days with thunder, and the amounts of precipitation greater than 10, 20 and 50 mm, are independent variables. The assumption is that more stormy weather means more hail days in the region and that the chosen correlates can compensate the difference in the number of hail days between years in the region. The effect of hail suppression is represented by a step function, the value of which is one for years with hail suppression and zero otherwise. The regression coefficient of this function, according to analysis assumptions, reflects the effect of hail suppression. Data from seven meteorologicai stations in north-western Croatia. spread over an area of 11000 km2, have been used in the analysis. The operational hail suppression system started working in this part of Croatia in I971, and by 1974 it spread over the whole region, so that there were years during which some stations were under the influence of hail suppression while others were not. The number of meteorological stations in the region has changed over the years, too. The first station started working in 1862 and the most recent one in 1982. Analyses have been done with data from all years with at least one working station, and for the two periods, 1937-2000 and 1945-2000. The common characteristic of all three sets of data is a very low percentage of explained variance, 15-20%, and a negative regression coefficient of the step function representing the effect of hail suppression. According to the value of this regression coefficient, the yearly mean number of days taken at all stations was 22% lower for the years with hail suppression, but it was not possible to distinguish it from the long-term trend in all cases