TITLE:
Temporal Profile Analysis of Sentinel-1 Image to Identify T. Aman Rice Varieties in the Coastal Region of Bangladesh
AUTHORS:
Mst. Shetara Yesmin, Dipanwita Haldar, Debjit Roy, Md. Belal Hossain, Priya Lal Chandra Paul, Md. Anisur Rahman, Arafat Shahriar
KEYWORDS:
Sentinel-1 Synthetic Aperture Radar, Rice Classification, Multi-Temporal Analysis, Decision Tree Classification, Agricultural Monitoring
JOURNAL NAME:
Journal of Geographic Information System,
Vol.17 No.6,
December
10,
2025
ABSTRACT: Monitoring of seasonal crop area, type and variety is essential for national agricultural planning and ensuring food security. Remote sensing and GIS-based satellite image analysis offer an effective and cost-efficient way to assess temporal land cover changes. This study utilized multi-temporal Sentinel-1 SAR (Synthetic Aperture Radar) time series data that have less cloud coverage and capture pigment-based properties, for proposing an effective method of mapping T. Aman rice varieties during the wet season (July-December) of 2019 in the south-west coastal region of Bangladesh. First, the temporal variation of Sentinel-1 SAR backscattering coefficients (σ˚) over agricultural plots was analyzed to identify the most effective metrics for detecting and distinguishing rice fields. Sentinel-1 SAR data from five dates of transplanting at approximately 15-day intervals were downloaded from the Alaska Satellite Facility (ASF) for the period between 12 July 2019 and 22 September 2019. Following preprocessing by Sentinel Application Platform (SNAP) software, generating a temporal profile for the rice versus rice and other features were extracted with the help of QGIS land use land cover (LULC) mapping. An attempt was made to discriminate rice fields by analyzing SAR data and by a knowledge-based decision tree classification through Environment for Visualizing Images (ENVI) software. The accuracy of the mangrove forest, homestead, river, and waterlogged categories was found to be more than 90%, and the accuracy of all rice types was found to be more than 80%. The overall accuracy achieved was 76.3%. Among the analyzed varieties, BRRI dhan49 showed the highest VH-polarized backscatter (approximately −17.5 dB) during the vegetative to maturity transition, clearly distinguishing it from the other T. Aman varieties. The distinct SAR backscatter signatures demonstrated the potential of multi-temporal Sentinel-1 data for identifying T. Aman rice varieties in the coastal region of Bangladesh.