An operational system for producing objective daily national and regional-scale analyses of rainfall has been developed for use in the Australian Bureau of Meteorology. The system uses a modular Barnes successive correction scheme to produce the analyses, either in near-real-time or using historical datasets. The national analysis is produced on a 25 km grid with a correlation length scale of 80 km, which is appropriate for the data density across Australia. Smaller correlation lengths and finer grids can be used for regional analyses. The accuracy of the real-time system has been tested by comparing analyses obtained from the full climatological dataset collected by the National Climate Centre (NCC) with the analyses limited to the real-time data. The two datasets were also compared to obtain information about the reliability of real-time observations. These investigations have shown that over most of Australia and with few caveats, reasonably accurate analyses are obtained using real-time data. A particular problem is found to be caused by telegraphic rainfall stations that do not report in real time when they record no rain. The analysis system is used to illustrate seasonal and event-related climatological features of Australian rainfall.
Weymouth et al. (Wed,) studied this question.