Practical generation of synthetic rainfall event time series in a semi-arid climatic zone

Abstract

An event-based approach adapted to semi-arid climatic conditions and applicable in those cases when only unreliable observation data are available is developed. Emphasis is placed on ease of modeling and computation. The case study of a 20,000 km 2 area in Central Tanzania where rainfall is mostly from convective-type storms is used to illustrate the approach. A phenomenological description of both convective-storm and frontal system precipitation is provided by a regional study of point rainfall events defined as an uninterrupted succession of days with more than 5 mm of precipitation, and dry events defined as a succession of dry days. The rainfall events are defined by depth and duration, which are found to be correlated. An analysis of the depth per event is thus performed, conditioned on the event duration. The log Pearson type III distribution appears to provide a good overall fit for the depth per event, especially for events lasting more than one day, while the one-day long events fit the negative binomial distribution. The event duration is found to fit a geometric distribution. The dry events during the rainy season seem to fit a negative binomial distribution. The climatic cycle length is fitted to a log Pearson type III distribution. A step-by-step procedure provided herein to simulate regional rainfall can easily be implemented on a microcomputer or even a programmable hand calculator.