An artificial trend in District Average Rainfall in the Snowy Mountains

A strong decline is apparent in annual District Average Rainfall (DAR) for District 71, located in the Snowy Mountains of New South Wales (Fig. 1). There has been some concern that this decline of about 200 mm, i.e. about 20 per cent, since 1913 could be the result of human interference with the environment, and that a continuing decline in rainfall could threaten ecosystems and irrigation. The DAR was originally used by the Bureau of Meteorology to provide a general overview of the rainfall across Australia (Chappel 1995). The original division of Australia into what became meteorological districts is described in Hunt (1910). The primary divisions were along State borders, with secondary divisions based on knowledge of physiographical factors (Chambers 2001). The DAR is the arithmetic mean of a sample of stations within a district, where stations were chosen using local (State) knowledge of their reliability and representativeness (Chappel 1995; Jones and Beard 1998). As stations were closed, and new stations opened, the stations used to calculate the DAR would have changed. Documentation of the stations used each year since 1913 is incomplete. Jones and Beard (1998) demonstrated that for much of Australia the DAR provides a reasonable indication of rainfall, both in the mean and for individual months. However, in areas with large spatial variations in rainfall, and few stations, they found that the conventional method of calculating the DAR led to errors. Jones and Beard identified District 71 as one such district. Chappel (1995), and Jones and Beard (1998), showed that changes in stations used to calculate the DAR could produce artificial, long-term variations in DAR. In this study, an attempt is made to identify the stations most likely to have been used in the calculation of the DAR for District 71 since 1913. The cause of the apparent decline in rainfall in this district is investigated, using these stations. An alternative strategy would have been to use only ‘high-quality’ rainfall stations to estimate trends in rainfall in this region. However, Lavery et al. (1992, 1997) could not identify any high-quality stations in District 71.