Accurate and timely estimation of crop area is essential for effective agricultural planning and policy-making, particularly in regions critical to food production such as the Cauvery Delta in Tamil Nadu, India. This study investigates the use of optical remote sensing data from Sentinel-2A and 2B satellites to delineate kharif paddy cultivation during the 2019 season. The high-resolution multispectral imagery, with a revisit time of five days and coverage across 13 spectral bands, was processed using supervised classification techniques. Ground truth data comprising 300 sample points were collected across Thanjavur, Thiruvarur, and Nagapattinam districts to train and validate the classification model. A False Color Composite (FCC) images and Maximum Likelihood Classifier (MLC) was employed to classify paddy and non-paddy classes across six categories: paddy, waterbodies, settlements, barren land, other crops, and miscellaneous. The classified output revealed a total kharif paddy area of 82,231.93 ha, with Nagapattinam recording the largest area (41,170.07 ha), followed by Thanjavur (22,246.01 ha) and Thiruvarur (18,815.92 ha). Accuracy assessment using an error matrix and Kappa statistics showed strong classification performance, with an overall accuracy of 91.7%, average reliability of 90.0%, and a Kappa index of 0.83. The results confirm the efficacy of Sentinel-2 optical data combined with supervised classification for large-scale paddy area estimation. This approach offers a robust, cost-effective, and scalable solution for crop monitoring and resource management, supporting national food security strategies and enabling data-driven agricultural decision-making.

Download this article as