DEVELOPMENT AND EVALUATION OF A SORGHUM MODEL BASED ON CERES-MAIZE IN A SEMIARID TROPICAL ENVIRONMENT

This paper reports on the development and evaluation of a grain sorghum model (CERES-Sorghum (SAT) for use in the semi-arid tropics. The model was developed from a version of CERES-Maize, previously adapted for use in this climatic zone. Functions for phenology, leaf growth, leaf senescence, assimilate accumulation and grain growth were modified using a small subset of sorghum data and validated against a much larger field-data set. When tested with cultivar De Kalb DK55 at Katherine, Northern Territory, the model successfully predicted grain-yield with a root mean square deviation of 0.972 to ha-1 over a range of sowing dates and water regimes resulting in observed yields ranging from 1.56 to 6.28 t ha-1. Deviations of predicted from observed yields were no greater than those of maize predictions by the parent model. Prediction of components of yield and biomass were also satisfactory. Calibration required 28 changes to the CERES-Maize (SAT) model, of which 15 were changes to coefficients in equations rather than substantial changes to the model. Because of the ease of conversion and the time-use efficiency found in these analyses, the techniques used in this paper could have application where locally calibrated models are required. © 1990.