Oil spill detection in the Chinese Seas by spaceborne synthetic aperture radars: challenges and pitfalls (Project: 10705 OPAC ) Werner Alpers Institute of Oceanography, University of Hamburg, Hamburg, Germany Kan Zeng Ocean Remote Sensing Institute, Ocean University of China, Qingdao, China Danling Tang South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
The Chinese Seas are sea areas with heavy ship traffic (tankers, cargo ships, fishing boats ) oil platforms natural oil seeps aqua farming installations large inflow of land-based pollutants (from urban sewage plants and industrial plants ) Potentially, they all release surface active material into the sea that floats on the sea surface and damps there the short gravity waves that are responsible for the radar backscattering. These areas look black on SAR images.
But they are not the only cause for dark areas often visible on SAR images of the sea surface ( oil-spill look alikes ) Dark area can originate from: low winds which are often encountered in the lee of islands or coastal mountains cold upwelling water which are associated with high biological activity giving rise to biogenic surface films and also to a change in the stability of the air-sea interface divergent flow regimes associated, e.g., with internal waves or tidal flow over sand banks dry fallen sand banks turbulent water as encountered in ship wakes rain cells grease ice
The great challenge is to single out those dark patches visible on SAR images that originate from pollution by mineral (petroleum) oil. Essentially, five discrimination methods have been proposed : by the degree of the reduction of the normalized radar cross section (NRCS or ) relative to the background by the geometry/location of the dark patch by the texture of the dark patch by different radar responses at different radar frequencies (multi-frequency SARs) by different radar responses at different polarizations (multipolarization SARs).
My evaluation of the different methods: Difference in σσ 0 : Practically impossible. The reductions in can be of the same order of magnitude for (almost) all types of dark patches. Difference in geometry/ position: Often, but not always, possible. Difference in texture: Sometimes possible Difference in the radar frequency : Sometimes possible, but multi-frequency spaceborne SARs are not available. Difference in polarization: Possible only when the oil or oil/water emulsion layer is sufficiently thick (of the order of millimeters). Of little practical use because multipolarization images have a small swath width.
My evaluation of the different methods: Difference in σσ 0 : Practically impossible. The reductions in can be of the same order of magnitude for all types of dark patches Difference in geometry/ position: Often, but not always, possible Difference in texture: Sometimes possible Difference in the radar frequency : Sometimes possible, but multi-frequency spaceborne SARs are not available. Difference in polarization: Possible only when the oil or oil/water emulsion layer is sufficiently thick (of the order of millimeters). Of little practical use because multipolarization images have a small swath width.
ship Biogenic surface films ERS-1 SAR image acquired over the Baltic Sea (Pommerian Bight) on 16 April 1994 at 21:04 UTC ROS Workshop, Delmenhorst, 23vNov. 2009
Counter example for linear features not associated with oil spills from ships: Biogenic slicks entrained in the Gulf Stream front Source: Ben Holt, SAR Users Manual, Chapt. 2 http://www.sarusersmanual.com/manualpdf/noaasarmanual_ch02_pg025-080.pdf
Wind direction Pollution by mineral oil Oil polluted sea area off the coast of Malaysia (near Kuantan) which is a busy shipping lane. The wind blows from an easterly direction causing the feathering of oil trails. This ERS-2 SAR image was acquired on 4 April 1997 at 3:25 UTC over the South China Sea. Imaged area: 100 km x 100 km). ESA
But even experts can misinterpret black patches on SAR images! During an oil pollution detection exercise carried out over shipping lanes in the North Sea and Baltic Sea in April 2003 involving oil pollution surveillance planes from several countries, the crews reported 5 potential oil spills. But samples that were taken later from ships or helicopters showed that in 2 cases it was palm oil (from cleaning tankers, it is legal) in 1 case it was fish oil and in only 2 cases it was mineral oil.
Position of the slick Compare with the position of oil platforms marked on sea charts
Platforms ID longitude latitude Name 0 114.623600 20.845000 No Name 1 114.708400 20.851700 No Name 2 114.786900 20.853400 No Name 3 114.716800 21.351000 23-1 4 114.911900 21.379400 XiJiang24-3 5 114.955800 21.305700 No Name 6 114.976100 21.271400 30-2 7 115.083200 21.364400 HZ32-2 8 115.085100 21.177100 HZ32-2 9 115.172300 21.160600 HZ32-2 10 115.218400 21.161300 HZ32-3 11 115.265700 21.173500 8Zhi 12 115.422900 21.335100 Hui Zhou 21-1B 13 115.411100 21.360800 HZ21-1 14 115.652200 20.658900 No Name 15 115.702700 20.828900 No Name
Biogenic surface films encountered in regions of high biological activity (upwelling regions) SeaWiFS, Chl-a concentration, July 2005 ASAR, 25 July 2005 12 db
SAR polarimetry for oil pollution monitoring
From the physical point of view: There should be no difference in the radar backscattering mechanism (Bragg scattering) from clean sea surfaces or sea surfaces covered by thin mineral oil films or biogenic surface films. Thus, e.g., the polarimetric quantity entropy should be the same in all three cases. Different statements can be found in the literature. Very likely reason: The poor performances of the radars, i.e., to the low signal-to-noise ratio of Radarsat 2 for crosspolarization.
Proof: Skrunes et al.: Noise floor https://earth.esa.int/c/docum ent_library/get_file?folderid= 233928&name=DLFE-2192.pdf
This is confirmed by measurements carried out with the Jet Propulsion Laboratory s (JPL) Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) having a very low signal-tonoise ratio during the Gulf Oil Spill Campaigns in 2010-2012: Bragg scattering theory describes well radar backscattering from clean sea surfaces as well as from sea surfaces covered by thin surface films. The difference in radar backscattering from thin and thick oil films (oil/water emulsion) is due to the large difference in the dielectric constant between water and oil/emulsion. (Minchew et al.: POLARIMETRIC ANALYSIS OF BACKSCATTER FROM THE DEEPWATER HORIZON OIL SPILL, TGRS, 2012) 17
Dielectric Constant of Ocean and Oil -Emulsion Forms New Dieletric Layer Complex Permittivity ε = ε' iε'' Sea water ε sw = 80 i70 -High conductivity surface Crude oil ε O = 2.3 i0.02 -Low conductivity surface Ocean Surface (no oil) Ocean Surface +Thin Sheen ε SW = 80 i70 Sheen ε SW+Sheen ~~ 80 i70 -Frequency, temperature dependent -Reduced roughness -Sheen too thin to change ε sw Emulsion = Mixture of Oil Emulsion + Sea water ε Mixture = ε SW + ε O -New dielectric layer with ε mixture Alters scattering UAVSAR polarimetric signatures respond to volumetric fraction of emulsified oil as mixture of oil and seawater from Minchew et al. 2012 18
Polarized NRCS derived from multilooked UAVSAR data The damping ratio for the thick surface oil layer does not measure the ocean wave spectrum damping alone, but a combination of the wave spectral change and a dielectric change in the upper surface layer. from Minchew et al. 2012
Conclusion with respect to radar polarimetry: Multi-polarization SARs is only useful for discriminating between mineral oil films and natural surface films when the mineral oil film is sufficient thick such that the radar signal become sensitive to the dielectric constant of the oil or oil/water emulsion layer. General conclusion: Identifying mineral oil pollution on SAR images of the ocean still remains a challenging task. It requires the fusion of information from many different sources, like from additional sensors. But also meteorological, oceanographic, and geographic information is needed. 20
Sensors flying on the German Oil Pollution Surveillance Plane Wide Range Sensor (± 30 km): Side Looking Airborne Radar (SLAR) Short Range Sensors (± 250 m): Infrared/Ultraviolet Scanner (IR/UV) Microwave-Radiometer Laser-Fluoro Sensor Forward Looking Infrared Camera Satellite Communication Devices
Thank you for your attention!