| dc.description.abstract |
Rainfall is an important variable in the wheat production areas of Australia. This analysis examines, firstly, the pattern of rainfall over 2.3 million ha of a high-quality wheat-producing region, and secondly, develops regression equations for rainfall prediction over this region. Most of the variation in rainfall pattern across the region is accounted for by differences in October-to-March (summer) rainfall and in April-to-September (winter) rainfall. The summer rainfall differences account for over two thirds of the variation. Based on these two rainfall periods, a partitioning of the study area reveals five distinct regions. The second part of the analysis uses multiple regression to provide a set of equations for rainfall prediction at any location in the region, for a number of rainfall periods. These equations use altitude, longitude and latitude as predictors. Nearly all of the equations explain between 80% and 94% of the variation in rainfall. Differences between regions are accounted for in the analysis, making the equations widely applicable. The validity of the mean rainfall equations was tested on three further sites: the mean prediction error was 6.9%. This approach may be applicable where large land masses with similar geographical features occur. |
|