Automatic Estimation of Oil Seep Locations in Synthetic Aperture Radar Images

A framework for the automatic detection of natural oil slicks and estimation of their associated oil seeps using synthetic aperture radar (SAR) images is presented, and the methodology used has been explained in detail. The designed detection system is the first automatic oil seep estimation system known to exist. The system detects oil slicks in individual SAR images and estimates their origins on the sea surface. Spatial clustering of temporally recurrent slick origins is conducted in order to estimate the locations of the associated oil seeps on the sea floor. The system is implemented in the programming language Python and a direct rule-based approach is employed for the classification unit. A data set of 178 images of the Black Sea acquired by ENVISAT's Advanced Synthetic Aperture Radar was used to test the algorithm. In this paper, the methodology used to design the algorithm and the automatically estimated oil seep locations are reported. The efficiency of the system with respect to manual detection is discussed.