Geological Engineering Department, Middle East Technical University, Inonu Bulvari, 06531 Ankara, Turkey.
Environ Monit Assess. 2011 Dec;183(1-4):409-23. doi: 10.1007/s10661-011-1929-6. Epub 2011 Mar 8.
The aim of this study is to propose and test a multi-level methodology for detection of oil slicks in ENVISAT Advanced Synthetic Aperture Radar (ASAR) imagery, which can be used to support the identification of hydrocarbon seeps. We selected Andrusov Ridge in the Central Black Sea as the test study area where extensive hydrocarbon seepages were known to occur continuously. Hydrocarbon seepage from tectonic or stratigraphic origin at the sea floor causes oily gas plumes to rise up to the sea surface and form thin oil films called oil slicks. Microwave sensors like synthetic aperture radar (SAR) are very suitable for ocean remote sensing as they measure the backscattered radiation from the surface and show the roughness of the terrain. Oil slicks dampen the sea waves creating dark patches in the SAR image. The proposed and applied methodology includes three levels: visual interpretation, image filtering and object-based oil spill detection. Level I, after data preparation with visual interpretation, includes dark spots identification and subsets/scenes creation. After this process, the procedure continues with categorization of subsets/scenes into three cases based on contrast difference of dark spots to the surroundings. In level II, by image and morphological filtering, it includes preparation of subsets/scenes for segmentation. Level III includes segmentation and feature extraction which is followed by object-based classification. The object-based classification is applied with the fuzzy membership functions defined by extracted features of ASAR subsets/scenes, where the parameters of the detection algorithms are tuned specifically for each case group. As a result, oil slicks are discriminated from look-alikes with an overall classification accuracy of 83% for oil slicks and 77% for look-alikes obtained by averaging three different cases.
本研究旨在提出并测试一种用于检测 ENVISAT 高级合成孔径雷达 (ASAR) 图像中油膜的多层次方法，该方法可用于支持烃渗漏的识别。我们选择黑海中部的安德烈索夫海脊作为测试研究区，该地区已知存在广泛的烃渗漏。海底构造或地层来源的烃渗漏会导致油性气体上升到海面，并形成称为油膜的薄油膜。微波传感器（如合成孔径雷达（SAR））非常适合海洋遥感，因为它们测量来自表面的反向散射辐射，并显示地形的粗糙度。油膜会抑制海浪，在 SAR 图像中形成暗斑。所提出和应用的方法包括三个层次：目视解释、图像滤波和基于对象的溢油检测。在数据准备完成后（通过目视解释），第一级包括识别暗斑和创建子集/场景。在此过程之后，根据暗斑与周围环境的对比度差异，将子集/场景分类为三种情况。在第二级，通过图像和形态滤波，包括为分割准备子集/场景。第三级包括分割和特征提取，然后是基于对象的分类。基于对象的分类是应用于通过 ASAR 子集/场景提取的特征定义的模糊隶属函数，其中检测算法的参数是针对每个案例组专门调整的。结果，油膜与类似物区分开来，油膜的总体分类精度为 83%，类似物的总体分类精度为 77%，通过平均三个不同案例获得。
