Evaluation of the Effect of Soil Texture on the Accuracy of Rice Yield and Biomass Simulation Using the DSSAT Model at a Large Scale

Crop Growth Models have high simulation precision at a small level and field scale. However, their application at larger and regional levels causes a significant decrease in model precision. Few studies have investigated the reasons for this phenomenon. The aim of the present study is to investigate the effect of soil texture on the precision of the DSSAT model for predicting rice yield and biomass at a large scale. First, the DSSAT model was calibrated and validated using field data and then used to predict rice yield and biomass in 110 paddy fields in Soumeh-Sara County, Guilan Province. The required data was collected from the fields and entered into the model. Evaluation of the model output showed that the absolute error related to the model covers a relatively large range (up to 50%). However, the majority of this error occurs in the range of ±30%. Also, the model precision is acceptable for average regional yield values, but the model precision decreases for maximum or minimum yield ranges. The results showed that the precision of the model's simulation of yield and biomass was inversely related to the amount of sand in the soil. The lowest model error, or in other words, the highest precision, was observed in the 0-15 percent sand group, and the highest error and lowest simulation precision were observed in the sand percentage group above 30 percent.