Gadimova Sharafat 
  
2 SATELLITE SAR OBSERVATIONS IN SOUTHEAST ASIAN WATERS CENTRED UPON SINGAPORE 
For the period of September 1995 through June 1998, 1400 ERS-1/2 SAR scenes of the part of the Southeast Asian waters 
centred upon Singapore were used in this study. These data were freely available through the standard World Wide Web 
(WWW) catalogue-browse system of the Centre for Remote Imaging, Sensing and Processing (CRISP) of the National 
University of Singapore (NUS) at spatial resolution of about 400 m. From these low-resolution images assumed oil slicks 
were mapped by means of visual interpretation. In addition to ERS images RADARSAT data (a set of 40 images) has been 
used as well. This satellite has the capability to cover a larger swath (500 km) than the ERS satellite (100 km), and its SAR 
can also operate in various modes. This was specially useful for mapping oil slicks. Improvements in terms of temporal 
coverage was achieved through images obtained during the ERS-1 and ERS-2 Tandem Mission period. In order to check the 
performance of the oil spill detection method some full resolution ERS SAR Precision images (PRI) were also used from 
selected areas. The interpretation of SAR data was supported by meteorological and geographical information. 
In SAR images, the brightness of the sea surface is a measure of the sea surface roughness. Surface roughness is induced by 
wind and current but can be reduced by an alteration of the surface tension due to a top layer of very thin natural or spilled 
oil. Low wind speed, typically below 2-3 m/s (Hamre et al., 1996), will not produce sufficient roughness to be visible in 
SAR images, while in high wind the increased microwave backscattering from a spill area can be appreciated. If the wind 
speed is too high, typically above 12m/s (Johanessen et al., 1994), certain types of oil are quickly mixed or dispersed into 
the water column, becoming invisible for SAR, although it can re-appear at the surface during subsequent calm weather. 
Hence, oil spill monitoring using SAR data is optimal at moderate wind speed (3-10 m/s), but it is problematic at low and 
high wind speeds. 
The main problem in the detection procedure is to distinguish oil slicks (man-made) from natural slicks, because they have 
very similar radar backscatter signatures. Both types of slicks are seen as dark patches in the radar image and under certain 
conditions it can be very difficult, or impossible, to separate them. 
The analysis of SAR images for oil slick detection was performed in two steps in order to identify possible key sites for 
further investigations at higher spatial resolution. First, possible oil slicks in each SAR scene were visually identified and 
discriminated from others, so-called look-alikes. Second, the identified oil slicks were further classified into “spill like”, “oil 
spills” and “spilling ships” in order to provide a confidence level for each oil spill alarm to end-users. 
  
  
  
  
  
  
  
  
  
Number of slicks Percentage, % 
Total slicks 1107 100.0 
Spill-likes 641 57.9 
Slick types Oil spills 458 41.37 
Spilling ships 8 0.72 
« ] knf 162 14.63 
Slick area 1 — 5 kn? 259 23.39 
5 — 10 kn? 39 3.52 
> 10 kn?’ 11 0.99 
  
  
  
  
  
  
Table 1. Statistics of possible (oil) slicks in Southeast Asian waters centred upon Singapore 
The oil slicks statistics derived from the analysis of ERS-land 2 and RADARSAT quick-look images are summarized in 
Table 1. Natural slicks are very common in three areas (Strait of Malacca, Gulf of Thailand and Brunei Bay) and in most 
cases they can be discriminated from oil spills by their size and shape. The form of the majority of man-made spills was 
found to be either as elongated segments collocated with shape wakes, or short zigzag stripes. Obviously such appearances 
indicate, either leakage on the course of the ship, or bilge water discharging during tank cleaning operations. Nevertheless, 
the large number of spill-likes (57.9% of the total number of slicks) indicates that there was a problem to discriminate 
between natural sea surface slicks and oil spills. À total of 458 oil slicks were identified. However none of these identified 
oil spills have been confirmed by in situ observations, but the evidence of spills indicate that oil pollution is a significant 
impact an environmental problem in the Southeast Asian waters. 
  
296 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B1. Amsterdam 2000. 
  
pa 
sli 
Hc 
sea 
All 
agı 
del 
In 1 
im: 
  
Fig 
3.1