Predicting paddy yield at spatial scale using optical and Synthetic Aperture Radar (SAR) based satellite data in conjunction with field-based Crop Cutting Experiment (CCE) data

The accurate, reliable, and robust information on crop yield has great importance in food security measures. In this study, both optical (Moderate Resolution Imaging Spectroradiometer (MODIS)-derived Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI)) and Synthetic Aperture Radar (SAR) (Sentinel-1A) satellite datasets were used for predicting paddy yield at a spatial scale in conjunction with Crop Cutting Experiment (CCE) data in Sahibganj district of the Jharkhand state (India) during the monsoon season 2017. The yield prediction models were developed from linear (LR) and multiple regression (MR) analysis. The AquaCrop model was also employed to simulate the paddy yield. The key findings showed that the MR-based yield model developed from SAR (Vertical transmission and Vertical reception (VV) + Vertical transmission and Horizontal reception (VH) polarizations) data was more accurate, reasonable, and reliable as compared to the optical-based (NDVI + EVI) MR yield model. The SAR-based MR yield model showed better accuracy between predicted and observed yield (CCE) as evaluated using Nash-Sutcliffe Efficiency (NSE = 0.68). However, the optical-based MR yield model showed relatively lower accuracy (NSE = 0.62). The relative deviation (RD) of the predicted yield from the SAR and optical-based MR model was nearly 3% and 4%, respectively. Using the AquaCrop model, the simulated yield was underestimated by approximately 4%. We conclude that the SAR-based MR-yield model outperformed the optical-based model to predict paddy yield at a spatial scale, and the adopted methodology can beneficial for decision-makers withing agriculture monitoring. Hence, satellite imagery has always been a reliable source for yield forecasting and crop assessment at a regional and national scale.