First results from the NORSE2015 oil spill remote sensing campaign

Satellite synthetic aperture radar (SAR) data is used operationally for continuous monitoring of ocean areas and for detection of accidental and illegal oil spills. However, much research is still needed in order to develop more reliable methods for discrimination between actual oil spills and other, natural, phenomena, and for extraction of information of a known oil spill. Acquiring data for this type of research can be challenging, as it is not known when and where an oil spill will occur. Much research also focuses on the use of multipolarization data, which measures on more than one polarization channel, but these data are not used operationally. Data collections over controlled, experimental oil spills are therefore very valuable for research purposes.

In June 2015, a dedicated oil spill remote sensing experiment, the NOrwegian Radar oil Spill Experiment (NORSE2015), was carried out in the North Sea. Four different oil types were released close in time, and remote sensing data from several SAR sensors, both satellite borne and airborne, were collected over the slicks for several hours after release. The campaign took place during the annual oil-on-water exercise conducted by the Norwegian Clean Seas Association for Operating Companies (NOFO). It was a collaboration between UiT, JPL/NASA and NOFO, and was partly funded by CIRFA. The resulting data set allows for many interesting studies, e.g., on oil slick spreading and drift, SAR detectability in different frequencies, polarizations and wind speed regimes, effect of slick aging and weathering on SAR signatures, and model verification for estimation of oil volume content in emulsions.

The first two journal papers describing findings from the NORSE2015 campaign have just been accepted for publication.
Brekke et al. (2016) evaluates data from two satellite SAR sensors, i.e., Radarsat-2 and RISAT-1. The latter is a relatively new sensor, launched in 2012, and is one of the first satellite sensors offering data acquisition in the so-called hybrid-polarity mode. The hybrid-polarity mode combines the advantages of the more commonly available and used modes, namely the high polarimetric information content of the quad-polarization mode and the large spatial coverage of the single- and dual-polarization modes. The hybrid-polarity mode is therefore very interesting for oil spill detection purposes, but very few studies using actual hybrid-polarity data exist. The RISAT-1 scene collected during NORSE2015 is one of the first scenes of this type containing confirmed, known oil spills. Brekke et al. (2016) evaluates features related to the cross-correlation between different polarization channels, which has been found useful for quad-polarization data in previous papers. A hybrid-polarity version of this feature is derived in this paper, and a comparison of the Radarsat-2 and RISAT-1 data is performed.

In Skrunes et al. (2016), a comparison of SAR scenes from three different sensors, collected less than six minutes apart, is presented. The data includes satellite SAR data from Radarsat-2 and TerraSAR-X and airborne data from the NASA Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR). The data is compared in terms of oil spill detectability, data quality and multipolarization characteristics. In addition, the data set is collected under high wind conditions of about 12 m/s, and the results are compared to previous studies for data collected under low-medium wind speeds. It is found that all three sensors detect the four slicks, with better detectability in TerraSAR-X and UAVSAR than in Radarsat-2. With the small sizes of the releases and the rough ocean surface, the limit of what can be detected with Radarsat-2 is being approached. The high wind condition results in lower slick-sea contrasts, i.e., lower detectability, than what has been found in previous low wind studies. In addition, SAR characteristics that have previously been applied for discrimination between different types of oils, i.e., mineral oil vs. simulated natural slicks, are not found useful for that purpose under these conditions.

The NORSE2015 data set is unique, and shows the value of this kind of dedicated campaigns for research purposes. There is currently a number of studies on-going based on this data, which are likely to result in many interesting publications in the future.

(text by Stine Skrunes)

 

References:

Brekke, C., C. E. Jones, S. Skrunes, B. Holt, M. Espeseth and T. Eltoft (2016): Cross-Correlation Between Polarization Channels in SAR Imagery Over Oceanographic Features, IEEE Geoscience and Remote Sensing Letters, pp. 5, 2016 (accepted). [intranett]

Skrunes, S., C. Brekke, C. Jones, and B. Holt (2016): A Multisensor Comparison of Experimental Oil Spills in Polarimetric SAR for High Wind Conditions, Journal of Selected Topics in Applied Earth Observation and Remote Sensing, 2016 (accepted). [intranett]