ENVISYS - A Remote Sensing System for Detection of Oil Spills in the Mediterranean

The Mediterranean Sea is a fragile ecological and economic area with with frequent oil pollution, both intentional and accidental. An EU financed project has undertaken the task to develop a demonstrator for a remote sensing system to detect and monitor oil spills and possibly other large-scale environmental emergency situations. The system will include automatic screening of SAR imagery for oil spill detection building on an algorithm developed previously to the project. The algorithm has three main parts: (i) detection of dark spots; (ii) feature extraction; and (iii) dark spot classification. If a dark spot has been classified as an oil slick, an alarm sounds and an operator has to inspect the probable oil slick manually. The initial results from SAR image screening experiments have shown that the concept of automatic detection is very promising, however, it has to be refined. This will be one of task undertaken by the ENVISYS project. Presented at the 16th EARSeL Symposium, Integrated Applications for Risk Assessment and Disaster Prevention for the Mediterranean, Malta, 20-23 May 1996 operating such surveillance platforms in a monitoring role. In comparison, the use of satellite data has long been recognized, as in many monitoring activities, as a source of regular extensive geographic coverage. The use of ships or aircrafts for sea monitoring is not fully adequate, however, as they present two basic problems: the limited coverage and the operational costs. Recently, with the advent of remote sensing with Synthetic Aperture Radar (SAR), interest is devoted to their use for oil spill monitoring and detection, especially in the North Sea. Although there is not any ideal remote sensing instrument for the operational detection and monitoring of oil spills, satellite-borne SAR presents a number of advantages over other systems: worldwide regular coverage, day-night imaging capability, independence of cloud coverage and ability to detect both oil spills and ships. Three operational satellite systems (ERS-1 and 2 and RADARSAT) have the capability of oil spill detection, and more are coming soon (e.g. ENVISAT). Norway got interest in oil-spill detection using SAR images more than ten years ago, and has been in the forefront of developing this technique towards operational application in the North Sea (Wahl et al. 1996). In response to ESA’s science Announcement of Opportunity for ERS-1 in 1986, a Norwegian oil spill detection project was proposed. The project is chaired by the Norwegian Space Centre (NSC) and is supervised by an international steering committee with representatives from NSC, Norwegian Pollution Control Authority (SFT), European Space Agency (ESA), Marine Spill Response Corporation, Esso Norge A.S. and Statoil. The project had the following main phases: • 1990-91: Literature surveys, prelaunch preparations, planning of field experiment. • 1991: A dedicated oil spill experiment (DOSE’91) was carried out at Haltenbanken shortly after the launch of ERS-1. • 1992-93: Digital SAR images were manually inspected for oil slicks at the Norwegian Defense Research Establishment (NDRE) shortly after acquisition. Alarms were reported to SFT. • 1994: Tromsoe Satellite Station (TSS) took over the part of inspecting the SAR images as a preoperational near real-time service. Other countries in North Europe also got interest in oil spill detection by remote sensing within this period. The North Sea countries have, under the North Sea Treaty, established collaboration regarding data interpretation and follow-up routines for oil spills. The techniques have also been demonstrated in the waters of Britain and The Netherlands. Within some years, there will be several remote sensing satellites with SAR in orbit resulting in about daily coverage. Manual inspection of all these SAR images will be very work demanding. Therefore, the Norwegian Computing Center (NR) has focused some of its research on automatic oil-slick detection. A pathfinder project was carried out in 1992 (Weisteen et al. 1993). The study involved a very limited data set, but promising results were achieved. This study was continued in 1995 under a contract with TSS, this time with a larger data set (Solberg et al. 1996). The study refined the algorithms from 1992. High detection rates were achieved, however, more research is required to develop algorithms that are reliable for operational use. Some other research have also been carried out for oil spill detection, however, this research has not covered the important problem of discrimination between real oil spill and look-alikes. Barni et al. (1995) used fuzzy sets to detect oil spills in SAR images, and NDRE incorporated an algorithm for automatic slick detection in their ship detection system (Lindberg et al. 1995, Wahl et al. 1993). 1.3 The ENVISYS project An EU financed project within the Telematics Programme with partners from Greece, Spain and Norway has undertaken the challenge to develop a remote sensing system to detect and monitor oil spills and possibly other large-scale environment emergency situations. The project will develop an demonstrator of a near real-time system, ENVISYS, processing remote sensing imagery. The system will be demonstrated in the Mediterranean, and will, hopefully, result in an operational system covering the whole area within 3-4 years. The project started in February 1996, and is currently investigating user requirements and is developing the system design. The technical solutions and algorithms presented in the following sections are, therefore, only preliminary and based on early studies and the general experience from the automatic oil slick detection research carried out at the Norwegian Computing Center. 2 EMERGENCY DETECTION WITH ENVISYS ENVISYS is intended to be a tool for detection and management of environmental emergency situations. This means it will not only cover the oil spill application, but also other emergency situations like forest fires and floods. This may seem surprising since these situations are so different with respect to remote sensing techniques. The intention is that ENVISYS will be a building block system. Therefore, it will be possible to substitute the oil slick detection module in the system with a detection module for other emergency situations. Other important tools in ENVISYS, building on database and GIS functionality, will be about the same for all emergency situations. ENVISYS is not only intended to be a detection system, it will also be a system for management and support when the emergency situation lasts for several days. The three following sections comments on the three emergency situations in focus oil spill, forest fires and floods: