MODELING OF RAINFALL TIME-SERIES USING THE BARTLETT-LEWIS MODEL

The use of a stochastic model for point rainfall, the Random Parameter Bartlett-Lewis Rectangular Pulse Model, is examined in this paper by fitting it to hourly data from three different rain-gauges in England, These correspond to three regions of the country with different annua l rainfall statistics, After examination of the rainfall data statisti cs at different time-scales, the model is calibrated for the three dat a sets, and it is shown that the model parameters reflect the local ra infall characteristics. Fine tuning of the model is achieved by an opt imization method which, however, is not applicable to some of the wint er data because of its more complex clustering structure, Seasonal fea tures, the depth distribution and the intermittent structure of the ra infall are well reproduced by the calibrated model, as well as the ext reme depths over a range of return periods, The model thus provides a valuable tool for the generation of rainfall data at time-scales from the hourly to the monthly.