Remote sensing monitoring of Gulf of Mexico oil spill using ENVISAT ASAR images

Abstract

Marine oil spill causes ecological pollutions that result in serious impacts to the quality of marine eco-environment. Effective oil spill detection and monitoring are the basis for the rapid response and play an important role. Due to its all-weather, day and night detection, wide coverage, and real-time monitoring capability, Synthetic Aperture Radar (SAR) is the most applicable sensors for marine oil spill monitoring and detection. Radar detection of target always uses back scattering of its reaction. After the oil spills, oil slicks spread to the sea surface dampen the Bragg waves and reduce radar backscattering coefficients; that is, smooth the sea surface. So oil spills appear as dark areas in the SAR images. The critical part of the oil spill detection is to distinguish oil spills from other natural phenomena. Low wind speed areas, internal waves, biogenic films and so on can lead to this phenomenon, which are called "look-alikes". Based on the principle of oil spill detection using SAR images, analysis accomplished over ENVISAT ASAR data gathered in the Gulf of Mexico and related to the Deepwater Horizon oil spill accident, which happened on April 20, 2010, is mentioned in this paper. The main areas of interest related to such disaster are the following: (1) to detect oil spills at sea and (2) analyze changes of oil slicks over the sea. In order to achieve the proposed goal, the method consists of three steps: preprocessing, detection of dark spots and analysis. Firstly, the SAR data are pre-processed, including histogram equalization, geometric correction and speckle filtering. Enhanced Lee filter is used to reduce the speckle noise due to the good capability in noise suppression and edge preserving. Secondly, SAR image classification and oil spill identification are carried out. Because of dark patches in SAR images cause by other natural phenomena, the purpose of the classifier is to distinguish oil spills from look-alikes. So the object-based classification is applied with support vector machine (SVM) owing to a better effect for oil spill identification. Last but not the least, based on the results of oil spill exaction and polluted areas, the development process of Deepwater Horizon oil spill is analyzed, such as the temporal dispersion scope, diffusion and drift of the oil slick. In addition, oil spill area is calculated for area variation tendency. At the early stage of this accident, shapes of oil slicks mainly appear as patches. As time passed by, quite a few parts of oil slicks showed as stripes. Oil slick shape and position were affected by winds, ocean currents and so on. During the middle or later period, a portion of oil slick eroded the coasts of Louisiana, Alabama and Florida. In the early days of the Deepwater Horizon oil spill, the size of oil slicks presented the tendency of increasing as a whole and reached a peak in June. However, after July, areas of oil slicks began to fall. Relevant reports such as weather and plugging are also attached, which indicate analysis and reports are in accordance.