An optical model for the interpretation of remotely sensed multispectral images of oil spill

This work is focused on the interpretation of multispectral images of oil spills, by introducing an optical model of spectral signature for oil-covered sea surface. Oil spill detection and oil type identification can potentially be achieved using data from multispectral optical sensors. However, multispectral images interpretation is challenging, because the spectral signature depends not only on oil optical properties and film thickness, but also on the optical properties of the water column, the incident light distribution and the instrument viewing geometry. In this work a simulator has been developed, starting from an optical model for both clean and polluted surfaces, which makes it possible to analyze variability in the optical signal from an oil-covered water surface. Several simulations have been performed varying the water optical properties within a range typical of Case I waters, and considering different crude and refined oils. Incident light distributions and viewing configurations have been chosen according to a typical viewing geometry of the MERIS sensor over a particularly interesting Mediterranean area: the marine ecosystem of the Tuscan Archipelago. The results, shown in terms of both upwelling radiance and oil-water optical contrast, provide some general rules that may aid interpretation of MERIS data. In particular, the detectability of an oil slick has been shown to depend on oil type and film thickness: very thin oil films (sheens) are more easily detected at viewing directions near the sun-glint zone, while very thick films are more likely to be detected at viewing angles away from the sun. For films of intermediate thickness the detectability depends mainly on the oil's specific optical properties