Detecting Oil Spill on Sentinel-1 Imagery based on Texture Analysis and Instantaneous Wind Speed Estimation

Abstract

Illegal oil discharge from ship routine operations frequently exposes marine ecosystem in serious danger. Oil pollution at sea has been increasingly reported as the marine traffic also increases. Therefore, there is a pressing need for oil spill detection at sea to reduce the oil pollution. Synthetic Aperture Radar (SAR) imagery has been widely used for oil spill detection. Nevertheless, oil spill detection on SAR image is limited to wind speed. Thus, ocean wind is highly required. This paper presents the utilisation of Sentinel-1 SAR imagery to detect oil spill at sea and to retrieve instantaneous ocean wind information. Oil spill detection is carried out using image texture analysis and Support Vector Machine (SVM) classification. Moreover, instantaneous ocean wind field is estimated based on Fast Fourier Transform (FFT) and CMOD5 geophysical model. This study shows satisfactory result of oil spill detection and classification using texture analysis and SVM classification. Agreeable result is also shown for instantaneous ocean wind speed over oil spill formation which ranges between 2.7 – 9.0 m/s.