Figure 1. Wind speed effect on the radar backscattering contrast (in dB) of a crude oil spill (modeled by Pavlakis 1995, based on field data of Singh et al. 1986), and spectral regions sensed by radar sensors.
² European Commission, Joint Research Centre Institute for
Remote Sensing Applications,
Advanced Techniques Unit,
I-21020
Ispra (Va) Italy
Phone: +39-332-789089. Fax: +39-332-785469.
E-
mail: alois.sieber@jrc.it
An application study that highlights the potential of ERS SAR imagery for the regional monitoring of oil-spill pollution is presented. A set of ERS-1 SAR frames covering the Mediterranean coastal zone were interpreted for oil-spill occurrences. The results indicate that despite the MARPOL Convention, spills attributable to leakage or tank-washing discharges, which usually escape detection and abatement response, are still jeopardizing the region's marine environment.
The lethal effects of mineral oils on fragile marine and local coastal ecosystems are well recognised. As long as there is a global reliance on petroleum products, however, oil spills will continue to occur, making the strengthening of their monitoring very necessary.
Oil spills can occur both in the open sea and in the vicinity of coastal waters. There have already been several spectacular accidents, involving the loss of very large quantities of crude oil from disabled supertankers and offshore platforms. Most of these triggered an immediate response by the responsible authorities.
The size of the spill, however, does not necessarily tell one much about its potential to inflict damage. Even a small spill can wreak havoc in an ecologically sensitive environment. For example, near Norway in 1981, a small operational discharge of oily bilge washings from the tanker 'Stylis' killed an estimated 30 000 seabirds, because it impacted a location where they were seasonally abundant (Kornberg 1981). Local meteorological and oceanographic singularities can facilitate the continued concentration of oil fractions, such as the commonly known 'floating tarballs', the primary source of which is thought to be tank washings rather than the weathered residue of accidental spills. Moreover the associated environmental damage is especially accentuated in 'closed' seas with infrequent water renewal.
The Mediterranean basin, as demonstrated here through via ERS SAR surveillance, is just such a region, still prone to frequent oil spills of various types. The study reported here demonstrates the substantial help that current and future spaceborne SAR missions can provide in monitoring and policing the marine-pollution problem.
Detection At oblique incidence angles, the radar backscattering strength from the sea surface is dominated by the Bragg backscattering mechanism. This mechanism is directly related to the height of the wind-generated gravity-capillary sea-surface waves. The presence of spilled oil causes damping of these high-frequency wave modes (Alpers & Huhnerfuss 1988), resulting in a detectable signature in ERS SAR imagery. However, the sensitivity of these waves to the prevailing wind influences the detection capability.
Too low wind speeds, typically lower than 2-3 m/s (Donelan & Pierson 1987), will not produce sufficient surface roughness in the surrounding sea to contrast with the oil, while high winds will result in increased microwave backscattering from the spill area, reducing its contrast with the surrounding sea. Furthermore, at high wind speeds (typically above 15 m/s), the oil is washed down by the waves (Alpers 1993) and the spill disappears below the surface. Reported cases also (Bern & al 1992) indicate that the wind-speed range in which reliable detection is possible varies depending on the imaging radar parameters, as well as on the type of oil and the age of the spill.
A recent theoretical investigation (Pavlakis 1995) has shown that C, X and Ku-band radar imagery is capable of yielding a detectable contrast signal (i.e. higher than about 3 dB) for wind speeds up to about 9-10 m/s (Fig. 1). It also transpired from this study that phenomena related to non-linearities in sea- surface wave development enhance the sensitivity of discrete spectral regions of the sea-surface wave field in reacting to the presence of a spill. It may be of practical importance that the strongest of such nonlinear effects were found in the spectral region sensed by C-band radar sensors (very dark narrow bands in Fig. 1), and very close to the Bragg wave number span sensed by ERS-1 and 2 SAR antennas (i.e. 76-97 m for near and far swath). Moreover, it has been implied that the abnormal wave damping in these bands, which is related to an overshoot singularity of the sea-surface wave spectral form, becomes stronger at higher wind speeds. It therefore has to be expected that, due to their C-Band SAR antennas, the current ERS SAR missions possess an enhanced capability to yield a strong contrast signature of an oil spill within an operationally acceptable wind condition range (i.e. for wind speeds up to 13 m/s).
Figure 1. Wind speed effect on the radar backscattering contrast
(in dB) of a crude oil spill (modeled by Pavlakis 1995, based on
field data of Singh et al. 1986), and spectral regions sensed by
radar sensors.
The semi-enclosed Mediterranean sea covers about 2.5 million km², and its coastal zone is habitated by some 81 million people, expected to increase to as many as 170 million by 2025 (UN Statistical Office 1992-93). Due to the region's mild climate and historical background, the annual number of tourists is estimated to reach 260 million by 2025. Such numbers clearly imply strong financial and industrial activity, with an impact on the environment.
There are already about 40 oil-related sites (i.e. pipeline terminals, refineries, offshore platforms, etc.) distributed along the Mediterranean coastal zone, from and to which an estimated 0.55 and 0.15 billion metric tons, respectively, of crude oil and petroleum products are annually loaded, unloaded and transported by oil tankers.
Although the Mediterranean has been declared a 'special area' by the MARPOL Convention in which deliberate petroleum discharges from ships are banned, there is still ample evidence of numerous repeated offences. In 1972, it was estimated that the total amount of mineral oil released in this region was 300 000 metric tones. At present, estimates vary greatly, some being as high as 1 200 000 tons (Alpers 1993).
A set of 190 ERS-1 SAR frames acquired over the coastal zone between 14 April 1992 and 19 September 1993 were interpreted to trace such oil-spill occurrences (Fig. 2). The shapes of the majority of the detected spills, i.e. either elongated segments collocated with ship wakes or short zigzag stripes (Fig. 3), indicate leakages or bilge-water discharges. This conclusion is also supported by the ship accident record (Fig. 4), reported to REMPEC station in Malta (Regional Marine Pollution Response Centre for the Mediterranean Sea) (IMO/UNEP 1993, IMO/UNEP 1994). None of the spills shown on ERS-1 SAR imagery (Fig. 2) corresponds to a reported ship accident (Fig. 4), while a glance at Figures 2 and 4 is sufficient to see the considerably higher frequency of the 'deliberate' oil spillages, which usually escape detection and prosecution.
Figure 2. Frames of the interpreted ERS-1 SAR images and the
detected spills.
Figure 3. Typical shapes of detected spills on the corresponding
ERS-1 SAR orbits and frames.
Figure 4. Ship accidents reported from April 1992 to September
1993, which have caused pollution by oil and other harmful
substances.
Although this sample of the interpreted ERS-1 SAR frames can be considered small, some first conclusions can already be drawn about this potential threat to the Mediterranean environment, such as the generally higher abundance of spills along the northern coastal zone, and their local concentration close to sites associated with known oil activities, i.e. pipeline terminals, refineries, etc. (Fig. 5). Unfortunately, these data indicate that authoritative prevention measures are not being strictly applied. A typical example is shown in Figure 6.
Figure 5. Oil-related sites in the Mediterranean and detected
spills.
Figure 6. Oil spill (A) from a refinery detected in Isthmia,
Greece, on 28 November 1992.
The current results indicate that some areas in the Mediterranean coastal zone face a higher incidence of oil spills than others, and possible localised chronic pollution. Figure 2 shows two striking examples, one south of Sicily and the other north of Port Said (centred about the coordinates 36°N/14°E and 32°N/32°E, respectively). Here considerable spill concentrations were found within two successive ERS-1 SAR frames. Other 'eventful' but less critically affected regions appear to be the congested waterway approachess, such as the Straits of Gibraltar and Messina, and zones with major oil transit traffic such as the Ionian Adriatic route and the passage between the Mediterranean and the Black Sea, i.e. the Aegean route.
It is worth stressing here that the identification of such areas is an important preliminary step for planning intense monitoring schemes based on ERS SAR, which are attainable within the limitations of the current acquisition scenario. The thousands of ERS SAR data acquired and archived during the last five years over the Mediterranean region constitute a substantial resource for this purpose, not least because they can provide a true appraisal of the real scale of the problem of 'deliberate' oil spillage.
The present study was supported by the European Commission. The authors also wish to express their thanks to the European Space Agency for all of the kind support provided by its services.
W 1993, Potential application of remote-sensing techniques for monitoring marine pollution events in the Mediterranean Sea, STOA Project on Pollution of the Mediterranean Sea, Directorate General for Research, 10-11 September 1993.
Alpers W & H Huhnerfuss 1988, Radar signatures of oil films floating on the sea surface and the Marangoni effect, J. Geophys. Res., 93, 3642- 3648.
Bern T I, T Wahl, T Anderssen & R Olsen 1992, Oil-spill detection using satellite-based SAR: Experience from a field experiment, Proc. First Thematic Conf. 'Remote Sensing for Marine and Coastal Environments', 15-17 June 1992, New Orleans, USA, pp. 49-67.
Donelan M A & W J Pierson Jr. 1987, Radar scattering and equilibrium ranges in wind-generated waves with applications to scatterometry, J. Goephys. Res., 92, pp. 4971-5029.
IMO/UNEP 1993, Regional information system, Part C, Databanks, forecasting models and decision support systems, Section 4: List of alerts and accidents in the Mediterranean, REMPEC, March 1993.
IMO/UNEP 1994, Regional information system, Part C, Databanks, forecasting models and decision support systems, Section 4: List of alerts and accidents in the Mediterranean, REMPEC, April 1994.
Kornberg H 1981, Royal Commission on Environmental Pollution, 8th Rept. H. M. Stationery Office, London.
Pavlakis P 1995, Investigation of the potential of ERS-1/2 SAR for monitoring oil spills on the sea surface, European Commission JRC/IRSA/AT, Rept. EUR 16351 EN, Luxemburg.
Singh K P, A L Gray, R K Hawkins & R A O'Neil 1986, The influence of surface oil on C-band and Ku-band ocean backscatter, IEEE Trans. Geosci. Remote Sens., GE-24, pp. 738-744.
ESA EOQ 52