Science Service System

Summary of Proposal OCE3027

TitleStudy on marine oil spill classification using dual polarimetricTerraSAR-X images
Investigator Yu, LI - The Chinese University of Hong Kong, Institute of Space and Earth Information Science
Team Member
Professor Hongsheng, Zhang - The Chinese University of Hong Kong, Institute of Space and Earth Information Science
Professor Yuanzhi, Zhang - Nanjing University of Information Science and Technology, Department of Oceanography
SummaryThis study specifically aims to investigate oil spill detection and classification using TerraSAR-X single and dual polarimetric SAR images.Firstly, scattering mechanisms of microwave on clean and oil covered sea surface will be investigated based on two-scale BPM model. Then optimized polarimetric features for discriminating oil spill and look-alikes will be developed based on statistical analysis. Moreover, the potential of dual polarimetric SAR modes for marine oil spill detection will also be investigated. The additional information provided by coherent dual polarimetric SAR system will be quantitatively analyzed. The proposed study plans to use TerraSAR-X single and dual polarimetric SAR data that obtained during major oil spill accidents and field experiments. For example the data that acquired during Deepwater Horizon oil spill accident at Gulf of Mexico in 2010 and oil-on-water experiment carried out by Norwegian institutes in 2011 and 2012. Due to its high orbit and radiation accuracy, TerraSAR has been proven its usefulness in a variety of applications such as deformation and subsidence monitoring, sea surface winds and current retrieval. Most of these applications are based on single polarimetric SAR images. However, the high quality of polarimetric calibration of TerraSAR system makes is possible to provide very useful polarimetric information of ground targets through its dual and experimental quad-pol modes. Sometimes oil slicks and look-alikes can be well distinguished using single polarimetric SAR characteristics, however these characteristics of oceanic phenomenon in SAR images may vary according to specific conditions,which may results in misclassification. For example in some cases biogenic films holds very similar shape and texture as mineral oil. So the largest challenge of automatic oil spill detection facing is to reduce the false alarm rate. Polarimetric SAR capabilities of new SAR sensors will help to reduce false targets in SAR imagery. The feasibility of polarimetric SAR oil spill classification relies on the fact that the polarimetric mechanisms for oil-free and oil-covered sea surface are largely different. Before the availability of polarimetric observation, mineral oil and look-alikes such as biogenic slicks are difficult to be distinguished because they damp short gravity-capillary waves with almost the same strength. Based on different polarimetric scattering behaviors, mineral oil and biogenic slicks can be better distinguished: for oil-covered area, Bragg scattering mechanism is largely suppressed and high polarimetric entropy can be witnessed, while in case of a biogenic slick, Bragg scattering is still dominant, but with a lower intensity. So similar polarimetric behaviors as those of oil-free area is expected for biogenic films. This proposed study aims to investigate the contribution of phase information provided by dual-pol TerraSAR-X SAR images,and develop algorithms that take advantage of this polarimetric information to improve the performance of oil spills detection. The study will also compare the performance of dual-pol and currently largely studied quad-pol and compact polarimetric SAR paradigms, to provide guidance for the selection of operation modes or how to jointly use multi-source polarimetric SAR images, for the task of oil spills detection.

Back to list of proposals

DLR 2004-2016