MODELING OF BRITISH RAINFALL USING A RANDOM PARAMETER BARTLETT-LEWIS RECTANGULAR PULSE MODEL

The performance of a stochastic model, in which Poisson cluster proces ses represent the arrival of rain cells within a rainstorm, is assesse d for hourly point rainfall data from Birmingham, UK. The Bartlett-Lew is Rectangular Pulse Model is extended to include random cell duration , and its performance is compared with that of the basic model analyse d previously. Both models perform well in reproducing statistics of th e rainfall depth distribution at different time-scales for individual months, but the randomised model also enables a good reproduction of t he proportions of dry periods of different durations. Performance is i mproved by an optimisation method which addresses issues of parameter identification. However, some deficiencies remain with respect to extr eme values. Finally, the seasonal variability of parameters is represe nted on a daily, rather than monthly basis, with comparable results.