24-May-2012

EO Data Access
		How to Apply

		How to Access


3rd ERS SYMPOSIUM Florence 97 - Abstracts and Papers

Operational use of wind data for slick detection

OPERATIONAL USE OF WIND DATA FOR SLICK DETECTION

Services

Site Map

Frequently asked questions

Glossary

Credits

Search
Advanced Search

OPERATIONAL USE OF WIND DATA FOR SLICK DETECTION

T. Hamre		Nansen Environmental and Remote Sensing Center, Edvard Griegsvei 3A, N-5037 Solheimsviken, Norway Torill.Hamre nrsc.no http://www.nrsc.no:8001/~torill/

H. A. Espedal		Nansen Environmental and Remote Sensing Center, Edvard Griegsvei 3A, N-5037 Solheimsviken, Norway Heidi.Espedal nrsc.no http://www.nrsc.no:8001/~heidi/

Abstract

The objective of the reported work has been to provide an operational tool for use of temporal wind (model) data jointly with SAR image sequences for slick detection and verification. To meet this objective, we focused on: (1) the development of procedures for use of temporal wind data in a satellite SAR oil spill detection system, and (2) the collection and discussion of examples of SAR imagery where wind data may be used in an operational context, and of examples where wind data alone is not sufficient to explain the slick configuration seen in the images. In addition, a preliminary study on the feasibility of using surface current fields in slick detection and verification was conducted. The developed methodology and an example of analysis of ERS SAR imagery will be presented, including a discussion of available wind history. Finally, conclusions on the benefits of using a combination of satellite radar and meteorological data for detection of potential oil spills are presented.

Keywords: slick detection; wind history data; surface current fields

1 Introduction

Oil spills can cause severe and long-lasting damage to the marine and coastal environment. Radar images, have been used to detect oil spills both in scientific projects (e.g. Hovland et al., 1994; Bern, 1993; Alpers et al., 1991), and in an operational context (e.g. Wahl et al., 1996; Pedersen et al., 1995). The ability of the SAR sensor to image the sea surface independently of light and cloud conditions, and its capability to handle varying wind conditions makes it a useful data source for oil spill monitoring. Radar sensors are mounted on several operational platforms today, e.g. ERS-2, RADARSAT and JERS-1, and are also planned on future satellite missions, such as ENVISAT. Thus, the amount of new data should be stable or even increasing in the years to come.

However, a SAR image alone does not always provide sufficient information for identification of slicks caused by oil or oil-based material. There are several types of "look-alikes" (Hovland-Espedal et al., 1994; Johannessen et al., 1994), e.g. natural film, that can give rise to SAR signatures similar to slicks caused by oil spills. To aid in the analysis of SAR images with respect to slick detection and discrimination, additional data, such as wind and current fields, have proven to be useful (Espedal et al., 1995). Instantaneous wind (i.e. wind closest to the image in time and space), is useful to eliminate natural film, which will disperse at wind speeds higher than 6-7 m/s (Scott, 1986). In other cases, analysing the wind history in conjunction with the radar image(s) will help in determining the slick type.

Thus, in order to improve the classification of slicks in SAR images, an algorithm for operational use of wind data has been developed (Section 2). Using this algorithm, a set of ERS SAR images has been analysed, and some results are shown in Section 3, along with a brief discussion of when wind and currents data may be useful. Finally, conclusions are presented in Section 4.

2 Methodology

2.1 Pollution from point sources

In this study, a simple oil drift model has been used (Bern, 1993). This model assumes that oil will drift with 3% of the wind speed with 15 degrees to the right of the wind direction. It is further assumed that the release of oil is continuous from a fixed point source (e.g an oil platform), with sufficient amount to keep the slick connected. Assuming that wind is the only physical factor influencing the oil spill, there will be a linear relationship between the age of a slick, its length and the wind speed that has stretched it. For instance, for a wind speed of 5 m/s, a spill that has lasted for 2 hours would be about 1.1 km long. However, since this linear model includes only wind, it should be used only as a guideline. Ocean currents can also influence the spread and configuration of an oil spill, and should be taken into account in a more elaborate oil drift model.

For moving sources (ships), the direction of the vessel will be the most important factor for how the oil spreads. Thus, such a spill will mainly follow a straight line along the ship's heading, provided that the instantaneous wind or current don't immediately break it up. Only after some time will the slick be more heavily influenced by wind and ocean currents. Then it will gradually be realigned by the changing wind and current fields.

2.2 Demands for temporal wind data

Based on the current study (Hamre et al., 1997) and earlier work (e.g. Espedal et al., 1995), the following needs for wind and ocean current data have been established: Wind/current fields should be available for every third hour, or more often in cases where an oil spill has occurred. The preferred spatial resolution of these fields is 25-30 km, to ensure several wind/current vectors within and around each ERS SAR scene. A system for delivery of such meteorological data in near-real time is currently being implemented by DNMI (the Norwegian Meteorological Institute) and TSS (Tromsø Satellite Station).

2.3 Procedure for use of temporal wind data

The main steps in the algorithm for using temporal wind data in slick detection and verification, are briefly discussed below.

2.3.1 Step I: Slick detection

The instantaneous wind is very important in slick detection. If the wind speed is known to be > 7 m/s at the time when the image was taken, any dark features in the image should be considered "suspicious". For lower instantaneous winds, the decision on whether a slick is a potential oil spill or not, is determined by analysing it with respect to shape, size, configuration and texture.

2.3.2 Step II: Slick verification

Once a slick has been classified as suspicious, the wind history becomes a valuable tool for determining slick type. In brief, the algorithm tries to rule out slick caused by natural phenomena by analysing the instantaneous wind, the wind history and the slick configuration. If this elimination fails, the slick will be classified as either "potential oil spill" or "undetermined slick type".

2.3.3 Step III: Estimating the age of a slick

The procedure for estimating the age of a slick is shown in Figure 1 and Figure 2, which show the steps necessary to determine slick source and the matching of wind history and current slick configuration, respectively. This procedure will only be carried out when the operator has judged the slick to be a potential oil spill.

*Figure 1: Procedure for estimating a slick's age*

*Figure 2: Procedure for matching wind history and slick configuration*

3 Results / Discussion

A set of SAR scenes were analysed in the current study (Hamre et al., 1997), and one example is included here to demonstrate the developed algorithm. On 29-30 April 1996, an ERS tandem pair were obtained in North Sea near 57.7 degrees N, 4.5 degrees E. The ERS-2 scene showed an angular slick connected to a bright spot (oil platform), while the image the day before had no slicks. Weather maps from DNMI were obtained for every third hour, as modelled wind fields were not available. This yielded a poorer spatial resolution than desired, and the wind history had to be taken from a point (B) about 150 km away from the slick.

At point B the wind at 18:00 GMT on 29 April matched that of part a of the slick in Figure 3, and the wind at 06:00 GMT on 30 April matched that of part b. With very low winds in between (below 2.5 m/s) we concluded that the wind history from point B matched the necessary wind history. Ideally, the slick shape should have been as shown to the right at the middle of Figure 3, with the slick being diffused by the lower wind speeds in the beginning of its lifetime. Even if the slick seen in the image has a slightly different shape as the diffusion of part b is closest to the source in stead of closest to the bend, we considered the wind histories to be matching. By comparing the slick lengths (part b is approximately twice as long as part a) and the wind speeds in the time intervals, we estimated the slick to be about 17-18 hours old at the time when the ERS-2 image was taken.

Figure 3: ERS-2 SAR image from the North Sea on 30 April 1996 (10:47 GMT), covering about 10x10 km. The necessary wind history (to generate the seen slick configuration), sketches of the slick shape when exposed to low winds causing the slick to diffuse, and the available wind history from a selected point (below) are included

In the above example, the type and age of the slick could be found with the developed algorithm. In other cases, it was not possible to conclude whether a slick was likely to contain oil or determine its age (Hamre et al., 1997). Ocean current fields may have helped in some of these cases. Especially when the wind was low during the expected lifetime of the slick, and when the slick configuration was circular/winding. However, current fields were not available for this preliminary study.

Without coincident ocean current data, we focused on user requirements and recommendations for the use of such data in future oil spill monitoring.

The ocean currents data should be on approximately the same spatial and temporal scale as the wind data that are used in the analysis, to simplify the combined analysis.
The grid cell size of the ocean currents data should not be coarser than 20-30 km, to ensure that there will be several values within and around an ERS SAR scene.
The ocean currents data should also cover the same area as the wind fields to be obtained in near real-time from DNMI (or other meteorological organisations).
A set of SAR images and near-coincident ocean currents fields should be gathered for a future study, with the aim to use these data to develop a procedure for their use in an operational oil spill monitoring service. For these examples, it is necessary to obtain wind data (wind fields) for the same time intervals.

4 Conclusions

An algorithm for the use of temporal wind data (wind history) to support the analysis of slicks in SAR images, has been developed. A set of single ERS scenes and some tandem pairs and their associated wind data have been analysed using this algorithm, and one example was included here for demonstration. Based on these first analyses we consider the algorithm to be useful both for scientific and operational purposes in oil spill monitoring.

More work is needed, however, before the algorithm can be put into daily use in an operational oil spill monitoring service, like the one conducted by TSS in Norway. In particular, the analysis of a larger data set, including wind and current fields for a period prior to the image where slicks are found, will be needed to fully validate the algorithm. Also some work will be required to incorporate surface currents fields in the slick detection and discrimination algorithm.

Acknowledgement

The work reported here was supported by TSS under contract TSS-OLJ-1288/JPP-96. TSS also supplied sample SAR images as well as wind data.

References

Alpers, W., Wismann, V., Theis, R., Huhnerfuss, H., Bartsch, N., Moreira, J., Lyden, J., 1991: The damping of ocean surface waves by monomolecular sea slicks measured by airborne multi-frequency radars during the saxon-fpn experiment. Proc. IGARSS'91.
Bern, T., 1993: Norwegian Slick Study, Vol.1: Oil slick study, Oceanor, Trondheim, Norway.
Espedal, H. A., Hamre, T., Hamre, Wahl, T., Sandven, S., 1995: Oil Spill detection using satellite based SAR, Pre-operational phase (A), NERSC Technical report no. 102.
Hamre, T., Espedal, H., Samuel, P., Sandven, S., 1997: Operational use of wind data for slick detection and verification in ERS SAR images, NERSC Special report no. 47.
Hovland, A. A., Johannessen, J. A., Digranes, G., 1994: Norwegian surface slick report, NERSC Technical report no. 81.
Hovland-Espedal, H. A., Johannessen, J. A., Digranes, G., 1994: Slick detection in SAR images, Proc. IGARSS'94.
Johannessen, J. A., Digranes, G., Espedal, H., Johannessen, O. M., Samuel, P., Browne, D., Vachon, P., 1994: SAR ocean feature catalogue, esa-SP-1174, ISBN 92-9092-133-1.
Pedersen, J.P., Seljelv, L.G., Dahle Strøm, G., Follum, O. A., Andersen, J.H., Wahl, T., Skøelv, Å., 1995: Oil spill detection by use of ERS SAR data, from R&D towards pre-operational early warning detection service. Presented at 2nd ERS User Workshop in London 6-8 December 1995.
Scott, J. C., 1986: Surface Films in Oceanography, ONRL Workshop Report, C-11-86, 19-34.
Wahl, T., Skøelv, Å.;., Pedersen, J. P., Seljelv, L-G., Andersen, J. H., Follum, O. A., Anderssen, T., Dahle Strøm, G., Bern T-I., Hovland Espedal, H., Hamnes, H., Solberg, R., 1996: Radar satellites: A new tool for pollution monitoring in coastal waters, Coastal Management, 24, 61-71.

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry