부산대학교 도서관

Ocean pollution by oil slicks is a major environmental hazard highlighted by the spectacular accidents of Exxon Valdez, Erika and Prestige. But, these oil tanker accidents only account for 5% of total oil pollution worldwide, 95% coming from wild discharges. In order to monitor and therefore to detect surface pollution, an effective tool has to be found. Satellite detection is well adapted to this problem, and especially synthetic aperture radar (SAR) can fill the lack of pollution survey which affects seas and coasts. Indeed SARs image oil slicks allowing estimation of the pollution risk in coastal areas. On satellite or aircraft, SAR is an interesting tool because, on the one hand, SAR images can be acquired through clouds unlike infrared or optical images. On the other hand, it is well suited to the detection of slicks because slicks modify seawater viscosity, producing a strong impact on short waves measured by SAR. The backscatter is attenuated and oil spills appear as a dark patch on the SAR image. Studies had shown that it is not easy to always detect oil slicks as backscatter depends strongly on wind and sea surface conditions among other restrictive factors. A comprehensive synthesis of previous studies shows the suitable SAR acquisition modes to detect slicks, function of the influence of SAR parameters such as polarization (VV), wavelength (C-band), incidence angle (20 to 45/spl deg/), satellite or aircraft flight direction, waves and wind directions. We present here cases of easy and ambiguous detection to show the possibilities offered by SAR to automatically detect slicks, in order to improve the pollution detection and the drift forecasting in case of accidents for example. For this, ERS SAR and ENVISAT ASAR ocean images examples are shown, applying segmentation algorithms and a new approach with an original method based on ocean surface characterization.