A review of three sugarcane simulation models with respect to their prediction of sucrose yield

A review of the agronomic and physiological concepts of three sugarcane simulation models was conducted with the view of highlighting their published strengths and limitations with respect to the simulation of sucrose yield. A brief history and description of each model is presented with an examination of their performance and a suggested way forward to improve their accuracy and utility. The models examined were the Australian APSIM-Sugarcane model, the South African CANEGRO model and another Australian model; QCANE. Despite limited published performance data, ail the models have performed reasonably well, but the prediction of sucrose yields were not the same. Mean errors of prediction (root mean square of residuals) for sucrose yield for APSIM-Sugarcane were 4.12 Mg ha(-1) for CANEGRO 6.07 Mg ha(-1) and for QCANE 2.51 Mg ha(-1). Improvements for each of the models lie in better understanding (1) the effects of stress (water, nitrogen and temperature) on the partitioning of photosynthate to stored sucrose, (2) the response of different cultivars to stress, and (3) the differences between plant and ratoon crops in respect of radiation-use efficiency and transpiration efficiency. A new approach employing a source-sink concept is suggested, but includes the volume of stalks as a state variable to define the sink size. A central feature is to include a new state variable; reducing sugars to allow the hydrolysis and re-synthesis of sucrose for the construction of the structural stalk carbon (fibre) and to supply stalk maintenance carbon (CO2). Such an approach should offer more mechanistic and explanatory investigations into the growth and management of sugarcane with respect to its sucrose yield and purity, particularly in respect of various ripening strategies. (C) 2000 Elsevier Science B.V. All rights reserved.