Application of Multi-dated Sentinel-2 Imageries to Assess the Cropping System in Gosaba Island of Indian Sundarbans

Abstract

Application of satellite based remote sensing in agriculture has reached a new level with introduction of medium to high resolution earth observation satellites like Landsat series, SPOT, Sentinel-2, etc. This study assesses the cropping system and the spatio-temporal variability of crops and fallow land particularly during the post-monsoon season, at the Gosaba island of Indian Sundarbans using multidated Sentinel-2 data. Sentinel-2 data offers 10-20 m spatial resolution, 5-day revisit frequency, global coverage and compatibility to the Landsat missions and provides new opportunities for regional to global agriculture monitoring. Monitoring of crop conditions, soil properties and mapping tillage activities help to assess land use, predict harvests, monitor seasonal changes and assist in implementing policy for sustainable development. Indian Sundarbans is considered to be one of the most endangered regions in the world from climatological and biodiversity view point. The low cropping intensity of Sundarbans forces the people to unscientifically explore forest resources leading to degradation of natural biodiversity. Cropping system intensification is a promising and sustainable approach to support the livelihood of the people and to save the biodiversity of Sundarbans region. In this study, multi-dated Sentinel-2 data were classified by following supervised classification to generate thematic map for determination of the spatiotemporal variability of cropped and fallow land during the period of November, 2017 to March, 2018. The overall accuracy of the study was 72 to 85%. The periodical ground observations revealed five predominant cropping systems viz. rice-fallow, rice-fallow-rice, rice-grass pea-fallow, rice-fallow-chilli, rice-fallow-green gram in the area. Temporal profile of reflectance was generated for each cropping system using the multi-spectral response of crop surfaces in Visible and Near Infrared (VNIR) bands. Temporal profile of Normalized Difference Vegetation Index (NDVI) and Visible Atmospheric Resistant Index (VARI) were generated to determine the crop duration. NDVI reached its maximum during the peak growth stages of each crop. The present research is helpful to have a complete understanding of the in-season temporal changes in land use and land cover using Sentinel-2 imageries.