1. Estimates of design wet season rainfall depths across northern Australia.
- Author
-
Batchelor, Michael
- Subjects
RAINWATER ,CLIMATE change ,WATER supply ,HYDROLOGY ,FLOOD risk - Abstract
Water containment structures at resource operations in northern Australia are designed to contain wet season rainfall. Government regulations set design freeboards at annual exceedance probabilities (AEPs) between 5% and 0.1 % AEP. Studies of rainfall frequency typically focus on durations of interest for flood studies, and the Bureau of Meteorology's (BoM's) design rainfalls are available up to 7 days. For longer durations, designers derive design rainfalls by analysing long-term at-site records by fitting probability distributions to series of accumulated 2 to 4 month wet season rainfall. Over northern Australia, long continuous rainfall records are sparse. As a result, long-term timeseries derived from gridded rainfall estimates (such as those from the SILO Australian climate database) are often used for the analysis. In this paper, the Generalised Extreme Value (GEV) distribution was fitted (using LH-moments) to wet season rainfall maxima series extracted at each SILO grid point to yield maps of design wet season rainfall across northern Australia. This approach differs from BoM's derivation of design rainfalls, in that: • The parameters are derived from at-site rainfall datasets - as opposed to regionalisation of L-moment ratios across pools of gauges. All rainfall datasets are long and continuous - but comprise temporally and spatially interpolated records. In areas of sparse rainfall station coverage, the resultant distributions are based entirely on interpolation of distant daily rainfall records (as opposed to regionalisation of probability distribution parameters derived from shorter, more-distant rainfall records). The design rainfall estimates might therefore be susceptible to sample bias, with the presence or absence of individual large events disproportionately affecting the design rainfalls. The potential magnitude of these effects was estimated by comparing 7-day rainfall depth grids to the BoM's design rainfalls. Comparison maps show that in areas of higher rainfall station density, design rainfalls derived from gridded data are generally similar to BoM values, but where point data is sparse, the gridded data estimate can be significantly lower. While likely less susceptible to sample bias than 7-day rainfalls, wet-season totals derived from at-site analysis should be used with caution in regions where 7-day totals are signficantly less than design 7-day IFD rainfalls. This paper identifies regions where consideration should instead be given to estimating wet season rainfalls from at-site analysis of station data from, say, the nearest four rainfall stations. [ABSTRACT FROM AUTHOR]
- Published
- 2023