A stochastic cluster model for hourly precipitation data

Abstract

The objective of this research was to develop a stochastic model of hourly precipitation that preserves the pattern of occurrence of precipitation events throughout the year as well as several characteristics of the duration, amount, and intensity of precipitation within events. In the developed model, the pattern of occurrence of precipitation events is described by a Poisson cluster process. The duration of events and the time between events within clusters are described with identical Logarithmic Negative Mixture distributions. The hourly precipitation amounts within events are described with a nonstationary, first-order autoregression model. The results of Monte Carlo simulations for Salem and Pendleton, Oregon (stations from very different climatic regimes) showed that the model accurately reproduces the seasonal pattern of event occurrence and the marginal and conditional distributions of the magnitude, duration, and intensity of precipitation during events. Autocorrelation functions for the historical and simulated data were also similar.