A mixing model for monitoring trace oil at sea

Oil spill has become one of the most serious marine pollution. Oil films gradually disperse into trace oil after accidents, still harmful to marine environment. It can cause various forms of direct harm to marine organisms, which is manifested in chronic toxicity. Accordingly, the amount of oil spilled into sea is of vital importance to environment protection. Airborne and spaceborne hyperspectral sensors are widely used for detecting and monitoring information such as human activities and environmental changes. Due to the limitation of spatial resolution, the spectral information is recorded as reflectance of a mixture of materials. Nonlinear interactions occur in the scene containing water. In order to extract the abundances of oil spills at sea, nonlinear mixture models have been developed. For trace oil, fractional abundance is at such a low order of magnitude that accurate description of the complex effect between oil and seawater remains a challenge. In this paper, a new mixing model is proposed to approximate the complex nonlinear interactions. It delivers a better understanding of the mixture in terms of trace oil as in-depth element in the seawater. Synthetic data is used to test the model. Experiment results show that unmixing with the proposed model leads to an accurate and stable fractional abundance distribution, even when oil abundance is at 0 ⁶ level.