Investigation the effect of observated and estimated dry matter from satellite imagery on the accuracy of Hashemi rice yield simulation using SWAP model

Predicting crop yields before harvest using simulation models and remote sensing technology is very important in sustainable agricultural management. The objective of this study was to use Sentinel 2 satellite images to improve the accuracy of SWAP model to simulate Hashemi rice yield in one of the research farms at the Rice Research Institute of Iran. For this purpose, LAI, NDVI and SAVI indices were derived from satellite images for growth period of plant. Also, leaf area index was measured directly and weekly at the field level and compared with indices derived from satellite images to provide the best equation for estimating the amount of dry matter. Then, using multivariate linear regression equation, the amount of dry matter was estimated from satellite images and given as the input parameter in the SWAP model. Finally, the measured yield was compared with the simulated values by the SWAP model with and without using satellite imagery. The results showed, when measured dry matter was as the model input, the error of model was 3.93% equal to 396.88 kg/ha and the model accuracy was acceptable (R2=0.95). but when the estimated dry matter from the satellite imagery was modeled as input, the model simulated the yield value of paddy with R2 equal to 0.99 and an error of 2.04% equal to 241.20 kg/ha. Finally, based on the obtained results, updating the SWAP model using satellite imagery improved the simulation accuracy and the model was able to simulate rice yield with higher accuracy.