Author(s): K. Sih; K. Fowle; P. Jordan; P. Hill; N. Nandakumar; E. Weinmann; R. Nathan
Linked Author(s):
Keywords: Design rainfall; Stationarity
Abstract: The majority of studies that have considered stationarity of daily rainfall totals have concentrated on central tendency statistics such as mean or median values. However, for derivation of design rainfalls using regional frequency analyses, it is the stationarity of the rainfalls in the upper tail of the distribution such as annual maxima that are of interest. The impact of record length and timing on rare rainfall estimates has been investigated using rainfall data from New South Wales, the Australian Capital Territory and surrounding areas of the neighbouring states of Queensland, Victoria and South Australia. Where statistically significant break points were identified, quantiles were derived for each of the data sets defined by the break point. No consistent differences in the quantile estimates were identified from this process and so the time series appear to be stationarity for the purposes of analysing the rainfall totals in the upper tail of the distribution. Therefore the full instrumental record can be used in regional frequency analysis, without adjustment for non-stationarity effects, to estimate design rainfall.
Year: 2011