Transformative Agricultural Monitoring: Enhancing Crop Classification in Jhansi Region With Machine Learning

In this study, we integrated multi-temporal satellite data from Sentinel-1 and Sentinel-2 to classify Rabi crops, specifically wheat, pea, mustard, and chickpea/lentil, in the Jhansi region of Uttar Pradesh, India. A Random Forest classifier was applied to an enriched feature set including Red Edge Normalized Difference Vegetation Index (ReNDVI), VH polarization composites, and entropy measures. The model achieved an overall accuracy of 84% and a kappa coefficient of 76%. Incorporating VH polarization and entropy features significantly improved the model’s ability to differentiate crops based on structural and textural characteristics. This research highlights the importance of integrating radar and optical data for more accurate crop mapping and suggests a pathway for improving agricultural monitoring practices using advanced machine learning techniques.