Comparing the value of seasonal climate forecasting systems in managing cropping systems

Cropping systems involve sequences of crops and management actions designed for profitable production and maintenance of resource integrity. Decisions made on crop choice and management in one season have ramifications on the sequencing of crops in subsequent seasons and on potential for resource degradation. In the previous paper, Carberry et al, (2000) found that a seasonal forecast based on 2-month SOI phases could be used to improve management in the subsequent two years of a dryland cotton cropping system at Dalby, Old. Their study showed that the intensity of cropping could be increased by planting either sorghum or cotton in place of the fallow in specific season types identified by the forecasting system. The objective of this paper was to compare a range of forecasting systems in relation to their value in managing a cropping system in northern Australia, The same dryland cotton cropping system was used to compare the potential value of four seasonal climate forecasting systems. The four systems were the SOI phase system, which utilises SOI patterns over a 2-month period a system based on SOI patterns over a 9-month period a system based on Pacific Ocean sea surface temperatures over a 5-month period a system based on projected SOI patterns from GCM runs for a 7-month period All forecasting systems showed skill in shifting the median rainfall for the 6-month summer cropping period of October-March. For the 12- and 18-month rainfall totals commencing in October, the separation of medians among groups defined by each system was greatest for the SST-based system and least for the GCM-based system, The four All forecasting systems showed some value in improving management decisions over the 2-year period examined in the dryland cropping system, In all cases, this was associated with increasing the intensity of cropping in specific forecast year types, either by introducing a sorghum or a cotton crop to replace the fallow in the second year of the rotation. The best outcome was associated with the forecasting systems based on either the 5-month SST patterns or the 9-month SOI patterns, which gave a 20% increase in gross margin, a 30% decrease in soil erosion, but an increased risk of making a financial loss. The other two forecasting systems gave a gross margin increase of about 13%. There were some differences in relation to financial risk and erosion outcomes among forecasting systems. The superior performance associated with tactical use of the forecasting systems reflected their ability to select years to increase cropping intensity. This was likely related to both the magnitude of predicted shifts in climate and the associated variability. The analysis highlighted the potential for seasonal forecasts to have value at long lead times. While the results suggested that systems based on SST patterns or longer term SOI patterns may be more useful, it was not possible to reach a general conclusion. A far wider range of locations and management scenarios need to be tested. This paper presents an appropriate framework for approaching this task. It also provides a means to integrate design of a forecasting system with its effective application.