This study was designed to calibrate and test an agrometeorological model over 18 growing seasons in three soybean production areas in Brazil: Passo Fundo (Rio Grande do Sul State), Londrina (Paraná State), and Dourados (Mato Grosso do Sul State). The soybean potential yield (Yp) was determined by two methods: estimated using the FAO Agroecological Zone Model or based on the maximum yield published by the Brazilian Institute of Geography and Statistics (IBGE), increased by 10, 20 and 30%. The estimate of actual yield (Ya) was calculated by correcting Yp for the relative water deficit at different growth stages. The results showed that the best performance was obtained when the Yp was represented by the maximum yield increased by a certain percentage. The model showed satisfactory Ya estimates for the three locations, generating R² values of 0.64, 0.46 and 0.70, with mean absolute errors (MAE) of 303, 289 and 259 kg ha-1 for Passo Fundo, Londrina and Dourados, respectively. In a global analysis, the performance of the model was satisfactory, with an R², agreement modified index (d1), confidence index (C) and MAE of 0.64, 0.52, 0.71 and 300 kg ha-1, respectively.
Glycine max; potential yield; actual yield; water deficit; simulation models
