SciELO - Scientific Electronic Library Online

Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Gayana (Concepción)

versión impresa ISSN 0717-652Xversión On-line ISSN 0717-6538

Gayana (Concepc.) v.68 n.2 supl.TIProc Concepción  2004 


Gayana 68(2) supl. t.I. Proc. : 297-304, 2004 ISSN 0717-652X



Andrei Yu. Ivanov1 & Ivan S. Ermoshkin2

1. P.P.Shirshov Institute of Oceanology, Russian Academy of Sciences Nakhimovsky prospect 36, Moscow, Russia,
2. Data + Bolshaya Gruzinskaya St. 10, Moscow, Russia,


This application study highlights the potential of the synthetic aperture radar (SAR) imagery for regional oil pollution mapping. The paper presents an approach and methodology for oil spill mapping in the Caspian Sea based on analysis of the ERS-1/ERS-2 SAR image quick-looks; this image seta has been collected in May 1996. By using expert evaluation all oil spill candidates on the SAR images have been detected, analyzed and finally introduced into geographical information system (GIS). GIS consisting of layers (databases) including coastal line, oil production infrastructure, hydrography, currents, bathymetry, allowed, first, to create oil spill distribution map and, second, to define oil spill accumulations and outline risk areas. Detected oil slicks were mostly located in the southern part of the Caspian Sea and were related to the oil exploitation and production areas, oil rigs, terminals, river runoff and bottom seepages. Results indicate that periodically refreshing oil spill distribution maps are a value source of information on oil spill distribution, statistics and sources. Moreover, GIS-based oil spill maps together with ground truth data is considered to be an important element in establishing of oil spill monitoring and risk management system.



The problem of pollution of the oceanic waters by crude oil is presently considered as one of the most vital issue. Every day, some new quantities of crude oil are released into the ocean and oil spills continue to occur on the sea surface, mainly in coastal zones and inland basins. The inland Caspian Sea, bordering by Russia, Azerbaijan, Iran, Turkmenistan and Kazakhstan, is one of the major sources of crude oil. The Caspian Sea has unique ecosystems with rich flora and fauna, including unique populations of the Caspian sprat, sturgeon and seal. Oil production activity, started on the Caspian Sea in the end of 19th century, has resulted in significant contamination of the sea and surrounding areas. Basically, the present problem in Caspian Sea is an outcome from unreasoned approach to environmental protection during development of marine oil fields in USSR. On the other hand, after the USSR's collapse the Caspian Sea became a center of action (since the beginning of 90's) of regional governments of new independent states and international oil companies for oil exploration to obtain strategic and geopolitical advantages.

Therefore, the oil activity in the Caspian Sea is increasing and its associated pollution is also happening now. A lot of worries for new oil production projects in the sea and its impact on ecological catastrophe of regional scale were pronounced in press (Targulyan, 2002).

The main environmental impact of oil spills is assumed to be well established (Kasymov and Askerov, 2001). It has been reported that tanker and oil platform accidents significantly contribute to total marine oil pollution, while illegal ship discharges are also important component. Berkelieva (2001) showed that main sources of oil pollution in the Caspian Sea are: 1) oil exploration and production in the marine oil fields in the Russian, Azerbaijan, Turkmenistan and Kazakhstan sectors; 2) oil transportation by tankers and pipelines; 3) secondary oil pollution, associated with old oil wells and plants; 4) river run-offs, and 5) seepages on the sea bottom (natural oil pollution). Berkelieva (2001) concluded that monitoring of oil spills in the sea by using conventional or traditional methods is rather problematic.

Synthetic aperture radar (SAR) is an active remote sensing tool, in which a satellite antenna transmits microwave signals toward to the ocean surface and receives signal after its interaction with the sea surface. It's well known, that crude oil and oils form films of various thickness on the sea surface. Oil films floating on the sea surface locally dampen the short surface waves and change the sea surface roughness. For this reason oil spills look as dark patches on the SAR images (Scott, 1986).

SAR has become an important tool in monitoring oil spills mainly due to its wide coverage, all-weather and day-n-night ability. Modern SARs, including wide-swath Radarsat and Envisat, providing high resolution, wide swath and high revisit time coverage are the most suitable tools for systematic and regular monitoring of oil spills in the sea (Ivanov et al., 2004). Since the launch of the first SAR many cases of oil spill detection have been documented (Bern et al., 1992; Masuko et al., 1995; Ivanov, 2000) and a number of regional monitoring campaigns have been carried out (Pavlakis et al., 1996; Gade and Alpers, 1999; Lu et al., 2000; Ivanov et al., 2002).

Within this project the Caspian Sea has been chosen as a test basin in order to detected various type of oil pollution. Reason is that the SAR images of the Caspian Sea acquired since 1991 is not still collected and analyzed. SAR images acquired in May 1996 have been collected and analyzed with respect to radar signatures of oil spills. In this paper we describe an approach and methodology how using SAR images and Geographical Information System (GIS) to produce an oil spill distribution map for the Caspian Sea and summarize the results of brief analysis.


1. SAR Image Quick-Looks

In this analysis the ERS-1/ERS-2 SAR low resolution images, so-called quick-looks, have been used as primary data. Gade and Alpers (1999) concluded that quick-look images have an advantage due to the fact that the dark signatures of oil spills can still be delineated, but data files of the small size make it easier to quickly processing of a large amount of SAR images. All images were downloaded from the ESA EOLI ODISSEO Catalogue and collected in JPEG format with pixel size 200 m (resolution ~400 m); each image covers a marine area of 100x100 km. As a first step, every SAR image has been cataloged and visually analyzed with respect to occurrence of dark patches (oil slicks).

Figure 1 shows examples of quick-looks of the SAR images acquired on May 13 (ERS-2, left), and May 28 (ERS-1, right) 1996 over the Caspian Sea eastward of Apsheron Peninsula (Azerbaijan) at wind speed 3-5 m/s. The large dark patches in the figure have been attributed to oil pollution, because of its isolated occurrence, shape and location close to the oil production site (Neftyanie Kamni). The shape irregularity of this spill seems to be caused by the action of the wind and local prevailing currents.

The best coverage of the Caspian Sea in period from 1991 to 2002 is found to be in 1996, when the ERS-1 and ERS-2 satellites were flown in Tandem Mission. General coverage of the Caspian Sea by the ERS-1 and ERS-2-SAR images in May 1996, produced in the ESA' DESCW software is shown in Figure 2.

Figure 1: Examples of the ERS-1and ERS-2 SAR image quick-looks showing large oil spills in the Caspian Sea. © ESA.

Figure 2: Coverage of the Caspian Sea by ERS-1 (right) è ERS-2 (left) SAR images in May 1996.

2. Approach and Methodology

Detectability of oil spills in SAR images depend on both the wind and oil parameters (wind speed, spill age, oil type etc.) (Bern et al., 1992). In SAR images oil spills appear as dark patches among the bright (rough) sea surface. In spite of development of automatic and semi-automatic methods of SAR image analysis, which also requires operator supervision, visual methods of analysis still dominates. In general approach, developed by Espedal et al. (1998) both direct (shape, size/length, location, orientation, type of edge, dB-contrast and texture) and contextual (wind/current/rain history, offshore and onshore sources etc.) analysis have to be used to discriminate oil slicks in the ocean. The same approach with some variations has been used by Gade and Alpers (1999), Lu et al. (2000) and Ivanov et al. (2002). In Ivanov et al. (2002) and Ivanov et al. (2004) an approach and methodology of oil spill mapping using wide-swath SAR imagery was proposed.

In our analysis we followed by general approach including: (1) SAR image processing, (2) Visual detection of dark features, (3) Interactive dark feature classification on the basis of geometrical and textural properties, contextual information, and (4) Discrimination among other look-alikes.

Further analysis of the SAR images for oil slick detection has been performed in several steps. When the oil spill candidates in each SAR image strip have been identified, the image strips were georeferenced and mosaic layer created (Figure 3). Then all identified medium and large oil slicks (> 1-2 km2) were contoured (Figure 4). The contoured areas on the SAR images as vector layers were separated, removed from image and put on to the map background. By such way the oil spill distribution map of the Caspian Sea, shown in Figure 6, has been creat

Figure 3: SAR image stripline of 5 quick-looks (ERS-2) (left) and SAR image mosaic (right) © ESA.

Figure 4: Contouring of an oil spill (creation of a vector layer) on the ERS-2 SAR image.

3. Geographical Information System

Simple GIS of the Caspian Sea was created on the base of Digital Chart of the World (DCW; ESRI, 1993) and a set of digital maps by the Caspian Environmental Program (CEP); they include:

1. Coastline (source: CEP). Vector layer was built in 1997 based on the nautical chart with scale 1:1,000,000 (1987) and updated with use of the Russian «Resurs» satellite images (Figure 5a),

2. Hydrography (rivers and lakes) (source: DCW) updated by CEP on the base of the map with scale 1:1,000,000) (Figure 5a),

3. Settlements (source: DCW). Update based on the map with scale 1:1,000,000 (source: GUGK, 1986),

4. Political boundaries (source: DCW) (Figure 5; in red).

Additional layers reflected specific tasks of the project (special content) include:

a. Marine oil fields and oil production infrastructure (oil rigs, drilling platforms, oil terminals, refineries, etc.) (sources: CEP, DCW); additional information was captured from Osadchii (2002) (Figure 5b),
b. Regional oil pipelines (sources: DCW, CEP, Kaztransoil) (Figure 5b),
c. Bathymetry (source: CEP) (Figure 5c),
d. Surface currents (source: Lopatin (1968)) (Figure 5d).

All these layers were integrated into GIS. Finally SAR image mosaics have been processed using various functions of GIS techniques including vectorization, data analysis, overlaying etc. to yield the oil spill distribution map as a thematic layer.

Figure 5: Layers of the Caspian Sea GIS: (a) hydrography,(b) oil production infrastructure (oil rigs, refineries, pipelines, newly developed oil fields), (c) bathymetry, and (d) surface currents


Figure 6 shows the distribution map of oil spills of the Caspian Sea in May 1996. SAR images covered most part of the Caspian Sea (with the exception of the North Caspian) within one month. In total 111 SAR image quick-looks were analyzed in order to obtain oil spill map. All oil spill candidates have been detected in a comparatively narrow range of wind speed, i.e. between 3 and 8 m/s, and among other surface manifestations (slicks) associated with oceanic and atmospheric processes. The shape of majority of the detected oil spills was patchy.

Information about wind speed, wind direction, local currents, coastline, political borders, hydrography and bathymetry, location of oil rigs, terminals, pipelines and seepages have been used to support our analysis. High, medium or low probability was assigned to each oil spill candidates. Of course, low resolution of SAR images doesn't allow extracting major oil spill parameters for accurate discrimination between oil slicks and look-alikes. No any verification of detected oil slicks was applied. For these reasons a part of large oil slicks were detected with a probability of 50 %. Nevertheless SAR image quick-looks can be used for oil spill monitoring in case, when full resolution images are not available. Advantage of quick-looks is that they have sufficient resolution for medium and large slick detection and they can be easier/quickly transferred, processed and analyzed.

Analysis of Figure 6 reveals that the majority of the oil spills were detected in the southern part of the Caspian Sea. The sea was most polluted near the marine exploitation areas, oil rigs, close to seashore refinery plants, river mouths as well over the bottom oil seeps. Among main reasons of oil pollution in the Caspian Sea there are old and life-expired equipment, technology and human factors, etc. Gul (2001). The main sources of oil pollution there are both point sources from rigs, terminals or seeps and distributed ones like slicks related to river runoff. The first ones have small scale and widespread, but the effect can be cumulative. In general, results of our analysis didn't prove observations by Gade and Alpers, (1999), Lu et al. (2000) and Ivanov et al. (2002) that most oil pollution in seas occurs along main ship routes. It's concluded, therefore, that oil pollution for the Caspian Sea is multi-source.

For example, we found that the highest occurrence of oil pollution is in the southern Caspian Sea, off the Iran (Figure 6) that maybe associated with river run-off from the coast (A. Knizhnikov, private communication). Moreover, there are a lot of small-scale slicks associated with bottom oil seeps. It is noteworthy that oil slicks as larger as 200 km2 were found there. These results indicate that the Caspian Sea and its coastal zones have a very high risk of oil pollution with a long-term effect on the sea and coastal marine resources.

Analysis of oil spill distribution map in the GIS together with current and bathymetric layers allows making a number of conclusions concerning oil spill spreading and accumulation areas. They are shown in Figure 6 by the color ellipses. Identification of such areas is an important preliminary step in developing the monitoring scenario for the Caspian Sea based on SAR imagery. Moreover, additional data, measurements and information integrated with GIS as supplemental layers can significantly improve classification procedure.

Figure 6. Summary map of distribution of the slicks related to crude oil, oil products and other man-made pollution on the surface of the Caspian Sea (May 1996). Ellipses show areas of oil slick accumulations over the oil production fields (blue), bottom seepages (violet) and river run-off (red)


This study demonstrates how imagery of SAR-equipped satellites can be used for regional monitoring of oil spills. The Caspian Sea has been chosen as test basin to work out an oil spill mapping technology. Available data set of the ERS-1/ERS-2 SAR image quick-looks acquired over the Caspian Sea in May 1996 have been collected and analyzed.

SAR image set allowed us to prepare oil spill distribution map for the Caspian Sea and conduct its analysis using simple GIS, built on the base of digital maps. Areas of oil slick accumulations have been identified, georeferenced and mapped using GIS technologies. The results indicate that such periodically refreshing oil spill maps are a value source of information on oil spill distribution, statistics and sources. Therefore, such GIS-based maps together with ground truth data is considered to be an important element in establishing of oil spill monitoring and risk management system.

Such a system is extremely needed because the Caspian Sea is an inland sea without water exchange and refreshing with the ocean, and the sea has a unique ecosystem with population of Caspian sturgeon, seal and food fishes. Oil production and, in turn, oil pollution, of the Caspian Sea is very high and its scale is unknown right now. After the USSR collapse three new independent states have appeared and delimitation of responsibility and damage liability are also needed.

Below most important findings of the study are summarized:

1. SAR image quick-looks can be useful data set for oil spill mapping in case, when full resolution data are not available;

2. Combination of SAR imagery and GIS provides an operative system for analysis, decision-making and risk management. On this base a prototype of oil spill analyzing and managing system for the Caspian Sea is proposed;

3. Oil slicks covered large marine area especially in the south part of the Caspian Sea and had multi-source nature;

4. Among main sources of oil spills in the Caspian Sea there are marine exploitation areas, oil rigs, onshore refinery and terminals, river runoff, bottom seeps, while main reasons of oil pollution in the Caspian Sea are considered to be old and life-expired equipment, uncontrolled technology and human factors.


Authors are very grateful to European Space Agency for a possibility to use ERS-2 SAR images. Furthermore, we would like to thank the direction "Data+" and Caspian Environmental Program for the possibility of using image GIS software and digital maps. This work has been supported by the Russian Foundation for Basic Research under grant #03-02-16763.


Berkelieva, L.K. 2001. Contamination of the Caspian Sea. (in Russian). [         [ Links ]1]

Bern, T.-I., et al. 1992. Oil spill detection using satellite based SAR: Experience from a field experiment. Proc. 1st ERS-1 Symposium, Cannes, France, 4-6 November 1992, 829-834. [         [ Links ]2]

Espedal, H.A., et al. 1998. COSWATCH'95 ERS 1/2 SAR detection of natural film on the ocean surface. J. Geophys. Res., 92, 24,969-24,982. [         [ Links ]3]

Gade, M., & W. Alpers, 1999. Using ERS-2 SAR images for routine observation of marine pollution in European coastal waters. In: The Science of the Total Environment 237/238. Elsevier Science B.V., London, 441-448. [         [ Links ]4]

Gul, A.K. 2002. Comments to wreck of the "Mercury" Ferry. Caspian Sea Bulletin, 6(38) (in Russian). [         [ Links ]5]

Ivanov, A.Yu. 2000. Oil pollution of the sea on Kosmos-1870 and Almaz-1 radar imagery. Earth Obs. Rem. Sensing, 15(6), 949-966. [         [ Links ]6]

Ivanov, A., M.-X. He, &M. Fang, 2002. Oil spill detection with the Radarsat SAR in the waters of the Yellow and East China Sea: A case study. Proc. 23rd Asian Conference on Remote Sensing, 25-29 November 2002, Kathmandu, Nepal ( [         [ Links ]7]

Ivanov, A., M.-X. He, & M. Fang, 2004. An experience of using ERS-1/2, Envisat and Radarsat SAR images for oil spills mapping in the waters of the Caspian, Yellow and East China Sea. Envisat Symposium Programme and Abstract Book. Envisat & ERS Symposium, 6-10 September 2004, Salzburg, Austria. [         [ Links ]8]

Kasymov, A., & F. Askerov, 2001. Biodiversity: Oil and Bioresources of the Caspian Sea. Print Studio, Baku (in Russian). [         [ Links ]9]

Lopatin, L.I. (ed.) 1968. The Caspian Sea. Moscow State University, Moscow (in Russian). [         [ Links ]10]

Lu, J., et al. 2000. Mapping oil pollution from space, Backscatter, February, 23-26. [         [ Links ]11]

Masuko, H., et al. 1995. Observation of artificial slicks with SIR-C/X-SAR around Japan. Proc. IGARSS'95, Florence, Italy, 14-18 July 1995, 227-229. [         [ Links ]12]

Osadchii, A.V. 2002. Large oil of the Caspian Sea. Nauka i Zhizn, 12 (in Russian). [         [ Links ]13]

Pavlakis, P., A. Sieber, & S. Alexandry, 1996. Monitoring Oil-Spill Pollution in the Mediterranean with ERS SAR. Earth Observation Quarterly, 52, 13-16. [         [ Links ]14]

Scott, J.C. 1986. Surface films in oceanography. ONRL Workshop Rep. C-11-86. Office of Nav. Res., London, 19-34. [         [ Links ]15]

Targulyan, O.Yu. 2002. Dark pages of the «dark gold». Environmental aspects of activity of oil production companies in Russia. Greenpeace, Moscow (in Russian). [         [ Links ]16]


Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons