Crop Classification Using Sentinel1 Data and Google Earth Engine

Crop classification in small farmlands is a complex task. The present study combines the Radar vegetation index (RVI) and cross ratio (CR) with sentinell VV and VH data for better crop classification in the Yavatmal district of Maharashtra, India. The performance of following four classification scenarios for crop classification were compared 1) Only VH and VV band 2) VV, VH and CR 3) VV, VH and RVI 4) VV, VH, CR and RVI. For 2022, crop-type mapping was generated with overall accuracies varying between 74% to 40% and a kappa coefficient between 0.57 and 0.25 using a Support vector machine and Random Forest classifier on Google Earth Engine. The results indicated that RVI outperformed VV, VH and CR for crop classification and fusion of sentinell data with the RVI increased the accuracy by 31%. This work demonstrates the use of sentinell data and vegetation indices to classify wheat, chickpea, other crops, water and urban areas.