A MULTICOMPONENT DECOMPOSITION OF SPATIAL RAINFALL FIELDS .1. SEGREGATION OF LARGE-SCALE AND SMALL-SCALE FEATURES USING WAVELET TRANSFORMS

Issues of scaling characteristics in spatial rainfall have attracted i ncreasing attention over the last decade. Several methods based on sim ple and multiscaling and multifractal ideas have been proposed and par ameter estimation techniques developed for the hypothesized models. Si mulations based on these models have realistic resemblance to ''generi c rainfall fields.'' In this research we analyze rainfall data for sca ling characteristics without an a priori assumed model. We look at the behavior of rainfall fluctuations obtained at several scales, via ort hogonal wavelet transform of the data, to infer the precise nature of scaling exhibited by spatial rainfall. The essential idea behind the a nalysis is to segregate large-scale (long wavelength) features from sm all-scale features and study each of them independently. The hypothesi s is set forward that rainfall might exhibit scaling in small-scale fl uctuations, if at all, and at large scale this behavior will break dow n to accommodate the effects of external factors affecting the particu lar rain-producing mechanism. The validity of this hypothesis is exami ned. In the first of these papers we develop the methodology for the s egregation of large- and small-scale features and apply it to a severe spring time midlatitude squall line storm. The second paper (Kumar an d Foufoula-Georgiou, this issue) develops a framework for testing the presence and studying the nature of self-similarity in the fluctuation s.