Abstract
Ocean oil spills cause serious damage to the marine environment, especially around coastal waters. Synthetic aperture radar (SAR) has been proven to be a useful tool for oil spill detection under low to moderate wind conditions. SAR operates in the microwave band and the data is not affected by the cloud cover and day/night conditions. However, the operational application of SAR for oil spill detection in the ocean is limited by false alarm targets or lookalike phenomena such as low wind speed, natural films, etc. In this study, we develop analysis of variance (ANOVA) to extract the features based on their characteristic geometry, grey level and texture features in the SAR images. We further analysed a fuzzy logic algorithm to separate oil spills features from lookalikes. We trained this algorithm using 38 SAR images (11 ENVISAT-Advanced Synthetic Aperture Radar (ASAR) and 27 European Remote Sensing (ERS)-2 SAR images) with 120 known oil spills and 80 lookalikes to generate an oil spill probability of a dark pixel in a SAR image. An independent set of 26 SAR images were used to validate the algorithm and it was found that 80.9% of the oil spills were correctly classified, and 20.0% of the lookalikes were wrongly classified as oil spills. The complete algorithmic procedure was coded in Matlab7.0 using its Fuzzy Logic Toolbox.
Acknowledgements
SAR images were provided by ESA through ESA-NRSCC Dragon Cooperation Program, project 2566 and the ESA AO Envisat project 226, 431 and 6133. The views, opinions, and findings contained in this report are those of the authors and should not be construed as an official NOAA or US Government position, policy, or decision.