Precise and real‐time rice yield estimation is essential for strategic agricultural planning and ensuring food security. This study presents an integrated approach combining multi‐source remote sensing data from Sentinel‐1 and Sentinel‐2 satellites with machine learning algorithms to estimate rice yields across the Cauvery Delta Region of Tamil Nadu during the Kharif season. Seasonal maximum values of vegetation indices and SAR backscatter coefficients were used as input features. Four machine learning models including Random Forest, K‐Nearest Neighbors, Gradient Boosting, and Decision Tree were trained using crop cutting experiment (CCE) yield data. Among these, KNN and RF exhibited the best performance with higher R 2 values of 0.87 and 0.84, and lower RMSE values of 318 and 399.7 kg/ha respectively. Spatial yield prediction over the study area revealed a mean yield of 4949 kg/ha with significant yield variability across districts, influenced by local agronomic practices and water availability. Validation using independent CCE plots confirmed the robustness of the models. The integration of SAR and optical data proved effective in mitigating cloud‐related data gaps and enhancing prediction accuracy. These findings highlight how remote sensing combined with machine learning can be leveraged for large‐scale, site‐specific crop yield prediction and effective food security planning.
Satheesh et al. (Mon,) studied this question.