Regional Rainfall Regimes Affect the Sensitivity of the Huff Quartile Classification to the Method of Event Delineation

Abstract

The widely used Huff quartile approach classifies rainfall events according to which quarter of their duration contains the largest rainfall depth. The rainfall events themselves are often delineated by specifying a minimum rainless interevent time (MIT) that must precede and follow a period of rainfall for it to be identified as a separate event. However, there is no standard or universally applicable value of this MIT criterion. Some studies have stipulated as little as 15 rainless minutes to mark the start of a new event, whilst others have required 24 h or more. The present work investigates how the adoption of different values of the MIT criterion, based, for instance, on the response time of a catchment or the drying time of a vegetation canopy, affects the Huff quartile classification. To date, this has not been explored. To address this issue, the Huff classification is herein applied to data from two Australian ground observing stations, one arid continental and one wet tropical. For each location, rainfall events were delineated using values of the MIT criterion ranging from 30 min to 24 h. In comparison with the 6 h MIT adopted by Huff (1967) as being appropriate for locations in the eastern USA, results show that, for instance, the proportion of events classified as 4Q ranges from 70.9% larger when events are delineated with a MIT = 30 min to 50.8% smaller when events are delineated using a MIT = 24 h. Moreover, the changes in the Huff classification that result from the use of different MIT values were not the same in wet tropical and arid locations. It is argued here that these findings reflect complexity in the arrival of rainfall, including differences in event duration and intraevent intermittency, that cannot be captured in the Huff classification system. Relevant rainfall regime characteristics such as these are likely to vary geographically, and the differences shown between the eastern USA and the two Australian locations are only examples of what is likely a more general effect. The results show that there is no single and generally applicable ‘Huff classification’ process, and that rather than a 6 h MIT being applicable everywhere, different MIT durations are needed in locations having differing rainfall regimes.