Enhancing Envisat ASAR WSM Segments to Detect Oil Spills in the Framework of the < 20 Years of Oil Routes > Project
Riazanoff, Serge; Gross, Kévin; Cottray, Arnaud
VisioTerra, FRANCE

A project is under way to retrieve the routes of oil tankers from the spills they release across the world seas. This project called "20 years of oil routes" requires the processing a large amount of ERS SAR and Envisat ASAR data acquired from 1991. A huge bibliography demonstrates that the automatic detection of oil slicks is not an easy task. Many factors are influencing the readability of the images and many black bodies that often are not oils may be improperly detected. This paper presents a rigorous way to preprocess the Envisat ASAR radar segments acquired in the wide swath mode (WSM). Three models of radar backscattering over oceans (CMOD4, CMOD5 and CMOD5.N) are compared to the mean of the cross-sections. The best model CMOS5 is used to normalize the column distribution to a predefined mean m₀ and a predefined standard deviation σ₀ (see fig. 1). The ScanSAR mode WSM may produce visible boundaries between the 5 sub-swaths. Analysing precisely the statistics around the supposed locations of the 4 junctions enables to detect and correct a possible discontinuity (see fig. 2).