Natural Environment Monitoring 2007, No 8, 73-90
TEMPORAL VARIABILITY OF THE SPATIAL STRUCTURE OF MAXIMUM DAILY PRECIPITATION TOTALS
Extreme precipitation events are of great significance for the operation of natural systems and can do much economic damage. Most of the climate change forecasts predict as highly probable an increase in their frequency and intensity. So far, however, little thought has been given to a possible change in the spatial structure of extreme precipitation. In the present study, a 25-year measurement series of maximum monthly and annual daily precipitation totals (MDPTs) from the area of Poland was employed to test a hypothesis about the existence of seasonal and multi-year variability of their spatial structure. For each of 325 sets of data an isotropic empirical semivariogram of normalised data with a range of 212.5 km (85 intervals 2.5 km in width) was calculated. The analysis of the seasonal and multi-year variability of the spatial structure was performed on parameters of semivariance models. The models were complex: they contained from two to five components. The spatial structure of MDPTs shows great variability, both yearly and multi-annual, which makes it hard to establish seasonal cyclicity and possible multi-year trends. On the basis of the results of univariate analysis of variance, it was found that statistically significant seasonal variability was displayed by the nugget variance (C0), the range (A1; about 15 km on average) and the variance of the first structure (C1). Mean values of other parameters also change in a regular way, but the differences are not significant statistically. The three above-mentioned parameters (C0, A1 and C1) also show distinct multi-year trends. They take the form of cyclic variability with a period of some 19 years (C1 and A1) and an indeterminate period longer than 25 years (C0). The nugget variance is the final effect of measurement errors and short-distance variability (including spatial discontinuity and the nonsynchronic nature of MDPTs). The first component can be identified primarily with precipitation from single convective cells. Thus, the performed analysis indicates that temporal changes largely affect the spatial structure of local precipitation.