Characteristics of rainstorms over a temperate region derived from multiple time series of weather radar images

Weather radars provide invaluable data to characterize rainstorms spatially and temporally. A correct description of rainfall in space and time contributes to improvements in hydrological modeling and design, and ultimately to a better water management. To provide a stochastic rainfall model with an accurate parameterization, frontal rain systems over Belgium (Western Europe) are analyzed. In this paper different structures within rainstorms and their relative spatial positions are studied, the movement of rainstorms is analyzed and distribution functions are constructed to characterize several features of rainstorms statistically. A correlation technique is applied to determine the direction and velocity of the translation of storms. Well developed rainstorms are isolated in radar images and their dimensions and shape are investigated. To describe the dimensions of the rainstorms three methods are proposed and compared. Statistical analyses provide probability distributions for the dimensions, the perimeter, the area, the velocity and the direction of the movement for rainstorms. Furthermore, the spatial distribution of clusters in a rainstorm is studied. It is shown that a simple Poisson process performs well in the representation of this spatial distribution. Two methods to calculate the single parameter in a Poisson process (in 1D and in 2D) are proposed and compared. (c) 2004 Elsevier B.V. All rights reserved.