Mapping mango area using multi-temporal feature extraction from Sentinel 1A SAR data in Dharmapuri, Krishnagiri and Salem districts of Tamil Nadu

摘要

Judicious and reliable information on crop area and production for tactical and strategic decision making is the need of the hour by all stakeholders in agriculture, such as producers, processors, resource managers, marketing, finance, and the government. The lacunae in conventional methods of crop delineation and area estimation can be overcome by the scientific method of estimation using remote sensing and GIS techniques. Sentinel 1A SAR data pertaining to the 2018 year acquired at 12 days intervals were downloaded and processed for intensity in Mapscape software. Sentinel 1A is anactive SAR microwave data, which can capture crop characteristics irrespective of weather and illumination condition. Ground truth observations collected during a surveyfor mango/ non-mango was used to derive mango signature from the processed satellite images. The dB values extracted as signature were then subjected to the Multi-Temporal feature extraction method to delineate the mango growing areas. Around 8015 ha and 31118 ha was mapped as mango growing areas in Dharmapuri and Krishnagiri districts, respectively. Accuracy assessment using the confusion matrix techniquewas done with the 40 per cent of the ground truth datedataand a kappa coefficient value of 0.71 was obtained which showed a good accuracy of estimation.