Universal kriging of hail impact energy in Greece

Abstract

In this study the spatial variability of hail impact energy was examined using universal kriging. The hailstorm of June 26th, 1988 in northern Greece was used for the development of methodology and the hailstorm of June 19th, 1988 was used for the analysis and validation. The emphasis was placed on the computation of empirical semivariogram, the identification of the appropriate optimal interpolation model and the optimal design of the hailpad network. During the hailstorm 42 hailpads were damaged and hail reports received from 15 communities sited in and around the area. These 42 values were used to compute the empirical semivariogrant of the observed hail impact energy. Due to the large variation of the data, the various theoretical semivariogram models used were tested against the experimental semivariogram of logarithmically transformed data. From all the tested models the one finally selected is the anisotropic monomial model with a quadratic equation for the computation of drift. Furthermore, analysis of the experimental isotropic semivariogram suggested that the optimal spacing of the hailpads should be 3-3.5 km, which is smaller than the usual spacing of the 4.5 km and larger than the dense spacing of 1.5 km. The experimental semivariograms in the W-E and SW-WE suggested a hailpad spacing of 2 km. These findings indicated that the hailpad network should be redesigned. Application of universal kriging with the anisotropic monomial model for the June 19th, 1988 hailstorm validated the above results. (C) 2002 Elsevier Science Ltd. All rights reserved.