Estimation of extreme spatial rainfall intensities

Abstract

Rainfall intensity occurring over a catchment is typically estimated from point rainfall in current flood estimation practice. This point rainfall is converted to a spatial rainfall intensity using areal reduction factors (ARF) that assume point rainfall intensity is greater than the average rainfall intensity over an area. This paper evaluates the direct incorporation of area into an intensity-frequency-duration (IFD) relationship, producing an IFDA, to account for spatial variation in rainfall intensity. A case study using gridded daily rainfall from the Australian Water Availability Project (AWAP) for regions in New South Wales and Queensland was undertaken. The case study focuses on coordinate based daily IFDA relationships. These coordinate based IFDA relationships are constructed for a set of areas surrounding each grid square within the subject region for a range of durations. This direct incorporation of area enables the direct estimation of spatial rainfall without the need for an ARF and shows there is considerable variability in the ARF that may need to be considered. These IFDAs are compared with statistical ARFs calculated for each coordinate, area, duration and average recurrence interval (ARI) to evaluate whether the use of a fixed set of ARF values for a region provide conservative spatial rainfall estimates. Coordinate based IFDAs were observed to vary significantly with location. Accounting for the spatial variation in IFDA curves over a region allows for more accurate and conservative spatial rainfall intensity estimates for extreme flood estimation. ARF sets calculated for each coordinate demonstrate that ARF values are spatially heterogeneous and are frequently greater than 1, demonstrating that spatial rainfall can be greater than the point rainfall sampled. The application of traditional ARFs in this context will result in an underestimate of rainfall intensity. It is concluded that the coordinate IFDA methodology provides an efficient means of estimating extreme spatial rainfall intensities which is also capable of considering partial storm catchment coverage. The importance of high quality spatial rainfall data sources to the construction of IFDA curves is emphasised.