Nine different crop management strategies were represented in the crop yield simulations performed. These strategies consisted of all combinations of 3 different planting dates and 3 different plant populations. The choice of planting date and plant population for a season are important management decisions that can be altered relatively easily in response to a seasonal climate forecast to potentially improve yields for that season. The plant populations considered were categorized as either low (15 000 plants ha-1), medium (25 000 plants ha-1) or high (35 000 plants ha-1), and were applied to all catchments. The planting dates considered for the various catchments were categorized as either early, average or late. The average planting date for a catchment was set to be equivalent to the long-term climatically optimum planting date for that catchment, as determined by Schulze (2003). The respective early and late planting dates for the catchment were then set to be one month before and one month after the average date. The early, average and late planting dates for the various catchments are shown in Fig. 21.2.
The potential for crop yield forecasts to improve crop management decisions was assessed over the 1981/1982 to 1992/1993 seasons by comparing: (1) the yield that would have been obtained if crop management strategies were selected according to yield forecasts, i.e. the yield of the forecast selected strategy, versus (2) the yield that would have been obtained if management strategies were selected according to long-term yield performance, i.e. the yield of the long-term strategy. It was assumed that in the absence of a yield forecast, a farmer would have selected his crop management strategy based on the long-term yield performance of the different strategies. In a particular catchment, the long-term strategy was the same for each of the 12 seasons considered, while the forecast selected strategy varied for the different seasons. The forecast selected and long-term strategies were compared on a seasonal basis, with the yields being simulated by CERES-Maize using the observed daily rainfall record for the season concerned. If the forecast selected strategy outperformed the long-term strategy, then it is assumed the farmer would have benefited from crop yield forecasts, provided he/she heeded them.
Before the above comparisons could be made, the forecast selected and long-term strategies had to be identified for each season. Yield forecasts were produced using the appropriate analogue rainfall season records identified previously during the downscaling of the SAWS rainfall forecasts. To identify the forecast selected strategy for a season, the medians of the forecast yield distributions of the 9 crop management strategies simulated were compared, and the highest yielding strategy identified. Simi-
larly, to identify the long-term strategy, the medians of the long-term yield distributions (derived from historical yield simulations for the 1950/1951 to 1979/1980 seasons) of the 9 strategies, were compared, and the highest yielding strategy identified.
Was this article helpful?