A stochastic model for oil spill detection in marine environment with SAR data

Abstract

The catastrophic explosion which destroyed the DeepWater Horizon (DWH) oil rig in the Gulf of Mexico, in April 2010, caused, in a period of three months, the discharge of some 4.9 million barrels of oil. The DWH remains the largest accidental marine oil spill in the history of petroleum industry. In order to detect the oil slick, and to measure its extent and geo-location, we present a methodology based on the use of SAR images and of stochastic process theory. The task of scene interpretation makes use of pixel potential functions supported by the Ising model. The scheme is applied to an ASAR image of the Gulf of Mexico, at the time when the DWH oil spill had already completely developed. The result shows both the labelled field of the image elements and the extent of the oil spill itself. The MRF model and the parameters of the stochastic optimization procedure are fully described.