Environmental Pollution ( IF 7.6 ) Pub Date : 2021-01-29 , DOI: 10.1016/j.envpol.2021.116618 Sankaran Rajendran , Ponnumony Vethamony , Fadhil N. Sadooni , Hamad Al-Saad Al-Kuwari , Jassim A. Al-Khayat , Vashist O. Seegobin , Himanshu Govil , Sobhi Nasir
Oil spill incidents contaminate water bodies, and damage the coastal and marine environment including coral reefs and mangroves, and therefore, monitoring the oil spills is highly important. This study discriminates the Wakashio oil spill, which occurred off Mauritius, located in the Indian Ocean on August 06, 2020 using the Sentinel-1 and 2 data acquired before, during and after the spill to understand the spreading of the spill and assess its impact on the coastal environment. The interpretation of VV polarization images of Synthetic-Aperture Radar (SAR) C-band (5.404 GHz) of Sentinel-1 acquired between July 5 and September 3, 2020 showed the occurrence and distribution of oil spill as dark warped patches. The images of band ratios (5+6)/7, (3+4)/2, (11+12)/8 and 3/2, (3+4)/2, (6+7)/5 of the Sentinel-2 data detected the oil spill. The images of decorrelated spectral bands 4, 3 and 2 distinguished the very thick, thick and thin oil spills in a different tone and showed clearly their distribution over the lagoon and offshore, and the accumulation of spilled oil on the coral reefs and along the coast. The distribution of post-oil spill along the coast was interpreted using the images acquired after 21 August 2020. The accuracy of oil spill mapping was assessed by classifying the SAR-C data and decorrelated images of the MultiSpectral Instrument (MSI) data using the Parallelepiped supervised algorithm and confusion matrix. The results showed that the overall accuracy is on an average 91.72 and 98.77 %, and Kappa coefficient 0.84 and 0.96, respectively. The satellite-derived results were validated with field studies. The MSI results showed the occurrence and spread of oil spill having different thicknesses, and supported the results of SAR. This study demonstrated the capability of Sentinel sensors and the potential of image processing methods to detect, monitor and assess oil spill impact on environment.
中文翻译:
利用Sentinel-1和2数据检测毛里求斯的Wakashio漏油事件:传感器功能,图像转换方法和地图绘制
漏油事件污染了水体,破坏了包括珊瑚礁和红树林在内的沿海和海洋环境,因此,监控漏油非常重要。本研究使用泄漏前,泄漏中和泄漏后获得的Sentinel-1和2数据来区分发生在2020年8月6日位于印度洋毛里求斯附近的Wakashio漏油事件,以了解漏油事件的蔓延并评估其影响在沿海环境上。对2020年7月5日至9月3日之间获得的Sentinel-1合成孔径雷达(SAR)C波段(5.404 GHz)的VV极化图像的解释显示,漏油的发生和分布是深色的翘曲斑块。图像的带宽比(5 + 6)/ 7,(3 + 4)/ 2,(11 + 12)/ 8和3/2,(3 + 4)/ 2,(6 + 7)/ 5 Sentinel-2数据检测到漏油。与解相关的光谱带4、3和2的图像以不同的色调区分了非常厚,厚,稀的溢油,清楚地表明了溢油在泻湖和近海的分布,以及溢油在珊瑚礁和沿海地区的积累。使用2020年8月21日之后获得的图像来解释沿海漏油事件的分布。通过使用平行六面体对SAR-C数据和多光谱仪器(MSI)数据的去相关图像进行分类,评估了漏油绘图的准确性监督算法和混淆矩阵。结果表明,总体准确度平均为91.72%和98.77%,卡伯系数分别为0.84和0.96。通过实地研究验证了卫星得出的结果。MSI结果表明不同厚度溢油的发生和扩散,支持了SAR结果。这项研究证明了Sentinel传感器的功能以及图像处理方法在检测,监测和评估溢油对环境的影响方面的潜力。