Estimating rainfall distributions at high temporal resolutions using a multifractal model

Abstract. Rainfall data from 18 stations in the vicinity of Tokyo city, measured to a precision of 1 mm, were analysed for multifractal properties. A multifractal model based on the scaling properties of temporal distribution of rainfall intensities was formulated to investigate the intensity distribution relationships in the available scaling regime. Although conventional analysis did not provide encouraging results with these measurements, an alternative approach that could be applied to rainfall data of widely variable quality and duration was used to establish a scaling relationship between daily and hourly rainfall intensities. Using a discrete cascade algorithm based on the log-Lèvy generator, synthetic hourly rainfall series were generated from the multifractal statistics of daily-accumulated rainfall. Several properties of rainfall time series that are relevant to the use of rainfall data in surface hydrological studies were used to determine, statistically, the degree of agreement between the synthetic hourly series and observed hourly rainfall.

Keywords: rainfall modelling, cascades, multifractal, downscaling