Basic Temperature-Based Solar Radiation Models to Simulate Soybean Yield Potential.

Several temperature-based models have been proposed for estimating daily solar radiation (Rs). However, it is interesting to access how estimated values impacts other crop process, such as evapotranspiration and crop yield. This study evaluated some simple temperature-based solar radiation models in the soybean yield potential in Triangulo Mineiro region of Brazil. The evaluated models were: Annandale (AN), Hargreaves (HA), Modified Hargreaves (HA-1), Hunt (HU), Bristow and Campbell (BC), Chen (CH), Donatelli and Campbell (DC) and De Jong and Stewart (JS). This research was made using historical data (2009-2014) from six sites of Triangulo Mineiro region of Brazil where measured Rs values were available. Daily recorded values of minimum and maximum air temperature and rainfall were used in the models. The dataset was separated into two sub-datasets, one for calibration (2009) and the other for evaluating their performance. The estimated Rs data were used in SoySim software to estimate potential soybean yield. Statistical indexes as: root mean square error (RMSE), relative root mean square error (RRMSE), coefficient of determination (R2) and mean error (ME) were used as performance indicators. Although the Rs estimated by the eight models has presented satisfactory performance for estimating Rs values when these data were used in the simulation of the potential soybean yield, the performances diverged considerably between models. All models showed a general tendency to overestimate the potential soybean yield, however, for the AN, HA, HA-1 and HU models this tendency was higher than the other models. The BC, CH, DC and JS models showed satisfactory performance in yield simulation with R2 and RRMSE varying from 0.76 to 0.80 and 3 to 4%, respectively. We are grateful to FAPEMIG for supporting our participation in this meeting.