A MULTICOMPONENT DECOMPOSITION OF SPATIAL RAINFALL FIELDS .2. SELF-SIMILARITY IN FLUCTUATIONS

In the first paper (Kumar and Foufoula-Georgiou, this issue) we developed a methodology for the segregation of large- and small-scale features (fluctuations) of spatial rainfall fields. In this paper we develop a framework for testing the presence and studying the nature of self-similarity in the fluctuations. It is found that rainfall fluctuations may be approximated by stable distributions and show scaling up to a certain scale. We define and estimate parameters that characterize the scaling and spatial dependence of the rainfall fluctuations and we use these parameters, estimated for several radar rainfall frames (in time), to relate to and identify the evolutionary nature of rainfall. Two radar depicted rainfall fields have been extensively analyzed: a severe spring time midlatitude squall line storm and a mild midlatitude winter type storm. The type of scaling in rainfall fluctuations shows significant variation from one rainfall field to another.