Oil-Slick Analysis With Polarimetric C-Band Sar Data: Some Aspects and Insights Over Coasts of Indian Ocean Region

Oil-spill pollution in marine environment is a major cause of hazard for the marine lives. Polarimetric SAR (Pol-SAR) imagery with its day-night and all-weather capability is an effective option, where ocean surface appears as Bragg scattering and Oil-spill as non-Bragg scattering (dark area). This study represents detection & analysis of oil-slick area, with an emphasis on coastal water in Indian Ocean region (demonstrated over three coastal sites) using EOS-4 and Sentinel-1 Pol-SAR data on different dates. The methodologies are categorized as: (a) Polarimetric Decomposition H-A-Alpha and Wishart Classification on polarimetric SLC data, (b) Sigma-naught backscattering and Texture based components with K-means classification on Ground Range Detected (GRD) data, (c) Machine Learning (Random Forest based Classification) approach using GRD data. The results are compared to check the efficiency of the methods for detection. All the algorithms are developed using Scikit-learn library in Python environment. This study will help to establish a monitoring system using open-source EOS-4 and Sentinel-1 Pol-SAR data (along with other ancillary datasets) in Indian Ocean regions.