Mapping of mango cropping area using machine learning techniques in Tamil Nadu

Mango area mapping is crucial for forecasting production prior to harvest and developing policies that ensure food and nutritional security. This study focuses on the integration of remote sensing and machine learning algorithms for real time prediction of mango cropped area in Krishnagiri and Dharmapuri districts of Tamil Nadu state in India. The cloud free images of Sentinel 2 were acquired corresponding to the fruit setting stage of the crop during the main crop season (January to June 2024) and pre-processed in ArcGIS 10.8 software. Supervised classification was carried out using eCognition Developer 10.3 software by combining object-based image analysis (OBIA) and Random Forest algorithm. Based on the analysis, the area under mango cultivation in Krishnagiri district was 31824.22 hectares, showing a 1.5 % deviation from the Department of Economics and Statistics data. The overall classification accuracy was 0.85, with a kappa index of 0.70. In Dharmapuri district, the mango area was slightly overestimated (14950.24 ha) compared to the government data (14589 ha), with a percent deviation of -2.5 %. The overall classification accuracy was 0.88, with a kappa index was 0.76. Accurate mango area mapping supports precision agriculture, efficient resource management and yield optimization, thereby aiding farmers, scholars and policymakers in informed decision-making.