I am working on the detection of oil spill within the ocean. My criteria of classification consists of a thresholding of SAR reflectivity. So, the low value of reflectivities may correspond to the oil spill while the high values correspond to the Bragg sea surface.
But, by applying this criteria, I obtain a lot of fake estimation of oil: Some large low-reflectivity areas are detected as oil Spill while they are not.