Oil pollution at sea is one of the most critical and destructive consequences due to human activity in marine areas, with an impact on the environment that requires decades to be reabsorbed. Satellite based remote sensing systems could be implemented for a surveillance and monitoring network. At present, the SAR system is the most widely used sensor for this purpose as it offers day and night high resolution images and it is not influenced by the presence of cloud cover, dust or smoke over the scene. The operational capabilities of these kinds of sensors are limited by factors such as the low revisiting frequency over the scene, the data’s high cost, and finally the inability to make a certain assessment of the nature of the detected event without collecting data from complementary instruments. SAR sensor limitations could be complemented by optical sensor capabilities.
In particular, multispectral sensors like MODIS offer a high number of spectral bands to detect, identify, classify and describe an oil spill event, and guarantee daily image frequency. However, optical sensors are highly dependent on meteorological conditions over the study area, they offer only low and medium resolution images and, finally, dedicated algorithms for image processing do not exist at present. For these reasons, the optical sensors play only a subordinate role with reference to SAR sensors. This work shows the results achievable through the development of dedicated algorithms for automatic image processing from MODIS data, and a method to classify and describe oil spill events.