Extreme temperature events in Australia

Abstract

A high-quality set of historical daily temperature data has been developed for Australia. This data set includes 103 stations, most of which have data from the period between 1957 and 1996, and some for longer periods. A new technique, involving the matching of frequency distributions, is presented for the adjustment of temperature records for inhomogeneities at the daily timescale, and this technique is used in the development of the data set. A number of additional findings are presented on the impact of changing times of observation and accumulation of observations over periods longer than one day on the Australian temperature record.

This data set was used for an extensive study of extreme temperature events in Australia. Widespread changes in the frequency of extreme temperature events in Australia were found over the 1957-1996 period. These changes were found both by an analysis or trends at individual stations and by analysis of spatial averages of indices of extreme temperature. In general, increases were found in the frequency of high maximum and high minimum temperatures, and decreases in the frequency of low maximum and low minimum temperatures. The changes were greatest for low minimum temperatures and least for high maximum temperatures, and were generally greatest in winter. The greatest decreases in the frequency of extreme low minima were found in Queensland. The trends were not universal, with trends opposite to those for Australia as a whole being found in some regions in some seasons.

It was found, after examination of several possible models, that the frequency distribution of Australian daily maximum and minimum temperatures was best represented by a composite of two or three Gaussian distributions with different parameters. Using this model, it was found that the observed changes in temperature primarily resulted from changes in the means of the component distributions, indicating that the changes resulted principally from overall warming of the atmosphere rather than changes in circulation or air-mass incidence.

The relationship between the frequency of extreme temperatures and the Southern Oscillation Index (SOl) was examined, with strong relationships being found in some seasons in many parts of Australia for most extreme variables, particularly high maximum temperatures. The weakest relationships were found for low minimum temperatures. Many of these relationships, except in winter, were as strong (or stronger) with the value of the SOl one season previously as they were with the SOl of the current season, indicating potential useful skill in the forecasting of seasonal frequencies of extreme temperatures in many cases.