Abstract

When the dynamics of a system is too complex to be analytically modelled, it has been found useful to assume that expected values of explanatory variables generate expected values of the response variable, and hence, deviations from the expected value of the response variable can be modelled by a Linear Perturbation Model (LPM) of the explanatory variables. This method is used in this study to develop a technique to update crop forecasts where climate is a major factor in crop production. The study is important because modern cultivars, which are the result of genetic gains, are sensitive to climatic variability, and recent studies with general circulation models suggest that one of the consequences of an increase in greenhouse gases may be greater variability in the climate of a region.

The usefulness of the LPM technique in the study of agriculture–climate relationships is tested through application to the Fitzroy catchment in Central Queensland. Since no reported climatic change is yet occurring in the region, the expected values for climatic conditions are obtained through averaging. By contrast, the expected values of crop yield are obtained from trend analysis; such trends are mainly attributable to genetic gains in the recent past. Three crops (wheat, barley, and sunflower) have been studied. Deviations (or perturbations) in crop yields are related, in the framework of LPM, to deviations in minimum, maximum, and average values of rainfall, temperature, and humidity at planting, flowering, and harvesting time. The most significant climatic factors affecting deviations in crop yield are identified. Regression models are developed which are capable of filtering and updating crop forecasts due to any unexpected climatic conditions, assuming consistent genetic trends and management practices.