Potential and Actual Sugarcane Yields in Southern Brazil as a Function of Climate Conditions and Crop Management

The use of crop simulation models with geographical information systems (GIS) has become an important tool to evaluate the sugarcane yield potential in different regions. Therefore, the objectives of the present study was to calibrate and test the FAO crop yield model to estimate the potential and actual sugarcane yields in the State of So Paulo to map these variables and define maximum yield increase considering a full irrigation system. The sugarcane crop yield was estimated by the FAO model with simulations for a period between 1974 and 2003 for 178 locations, using 10-day period weather data. The soil water holding capacity for each location was determined according to the predominant soil type in the region. The leaf area index and crop phenology parameters were adjusted in order to fit estimated and observed yields. We also inserted a frost factor into the model to improve its performance. The simulations considered the cycles of plant and ratoon crops, which were weighted to compose the annual potential and actual yields. The performance of the model was evaluated by the following statistical coefficients: R-2, d, C and RMSE. The maps of sugarcane Y-p and Y-a were generated with linear models and their respective deviations in a GIS platform. The model performed with reasonable precision and good accuracy, as proved by the statistical coefficients: R-2 = 0.58; d = 0.83 and C = 0.63. The RMSE was 5.0 Mg ha(-1) and the mean error was 2 Mg ha(-1).