Identifying individual rain events from pluviograph records: a review with analysis of data from an Australian dryland site

Rainfall is routinely reported as falling in 'events' or 'storms' whose beginning and end are defined by rainless intervals of a nominated duration (minimum inter-event time. MIT) Rain events commonly exhibit fluctuations in rain rate as well as periods when rain ceases altogether. Event characteristics such as depth, mean rain rate, and the surface runoff volume generated, are defined in relation to the length of the rain event. These derived properties are dependent upon the value of MIT adopted to defined the event, and the literature reveals a wide range of MIT criteria. Surprisingly little attention has been paid to this dependency, which limits the inter-comparison of results in published work. The diversity in criteria also diminishes the usefullness of historical data on event durations, rain rates, etc., in attempts to document changes in the rainfall climate. This paper reviews the range of approaches used in the recognition of rain events, and a 5 years pluviograph record from an arid location is analyzed. Changing MIT from 15 to 24 h (lying within the range of published criteria) alters the number of rain events form 550 to 118. The mean rain rate declines from 2.04 mm h(-1) to 0394 mm h(-1), and the geometric mean event duration rises from 0.66 h to 3.98 h. This wide variation in the properties of rain events indicates the more attention needs to be paid to the selection and reporting of event criteria in studies that adopt event-based data analysis. The selection of a MIT criterion is shown to involve a compromise between the independence of widely-spaced events and their increasingly variable intra-event characteristics. Copyright (C) 2008 John Wiley & Sons. Ltd.