Optical quantification of oil emulsions in multi-band coarse-resolution imagery using a lab-derived HSV model

Oil emulsions can harm marine and coastal environments for extended periods. Timely identification and quantification of oil emulsions are essential for oil spill response. Although SAR is the most commonly used technique in detecting oil presence, it has limits in oil quantification. In contrast, optical remote sensing can fill this gap with more spectral bands. Hyperspectral remote sensing is capable of achieving this purpose. However, it is challenging to use multi-band coarse-resolution imagery due to the fewer bands and mixed pixel effect. Through laboratory measurements, numerical simulation, and Hue-Saturation-Value (HSV) model, we illuminate the multispectral mixed characteristics of oil emulsions and demonstrate Hue's role in characterizing the mixture features and oil concentration trends. Hue-based oil emulsion classification and oil concentration segmentation (OCS) methods are proposed and applied to Landsat-5 images under quantified uncertainties. This approach is expected to expand its application in multispectral remote sensing.