Computer Crop Model Applications : Developments in Swaziland

need to estimate transport requirements, a ripening and cutComputer simulation models such as CANEGRO have proting programme, and an operations budget. Any improvement in the accuracy of the estimating process would aid gressed to the point where they are being used with some these decisions. confidence to address strategic production issues in the Swaziland sugar industry, e.g. benchmarking yields and crop water requirements. Models of this nature have generally not featured as estate or farm management tools due to the onerous data requirements and a lack of skill required to run the models adequately. Two model application developments are discussed, one at the mill area level intended to improve mill estimates and the other at an estate level to aid field management decisions. Improving mill estimates Accuracy of yield estimates impacts enormously on decisions made at estate and mill levels, and logistic and marketing strategies at an industry level (Wood, 1995). In Swaziland the first official estimates of sugar production are called for in March, prior to the start of the crushing season. These yield estimates are made at an estate or farm level, based on average production adjusted for crop age. At a mill level these estimates are used to finalise mill opening dates, calculate delivery schedules and assign rateables. At the industry level, decisions regarding logistics (bulk storage, transport and shipping) and marketing are made at this stage. However, a large number of management decisions requiring yield estimates are made during November and December prior to the crushing season. At mill level, millers need to estimate season length and start-up date to plan a maintenancc programme, whereas at an estate level, growers In Swaziland all sugarcane is fully irrigated and potential sugarcane production is limited by radiation and temperature. The impact of seasonal variations in radiation and temperature on cane yield can be evaluated best by using a systems or modelling approach. In this exercise the CANEGRO crop model (Inman-Bamber, 199 1 ; Inman-Bamber et al., 1993) was used to integrate the effects of climate on yield. To test this climatelyield relationship, mean seasonal cane yields harvested at each of the three miller-cum-planter estates were compared with potential cane yields estimated using the model for the period 1980-1998 (Figure 1). Seasons during which severe water restrictions were imposed on the estates were excluded from the comparisons. At all three sites there was good agreement between actual and potential yields, suggesting that a large amount of the variability in cane production could be accounted for by climate. Mid to long range climate forecasts are becoming more reliable (Lumsden et al., 1998) and serve as the best indication of future climatic conditions. In the absence of reliable radiation and temperature forecasts, rainfall was used as a surrogate (seasons of high rainfall are usually associated with low radiation and temperatures). Modelling groups such as the South African Weather Bureau (SAWB) and the Climate Impact Prediction Centre (CIP) publish updated rainfall forecasts regularly. These are available on the internet at the following addresses: 140 Sinnlye Sugar Estate 130 110 81 120 -