Measurement and analysis procedures for remote identification of oil spills using a laser fluorosensor

Abstract

Abstract A laser fluorosensing procedure has been developed for remotely detecting and identifying crude oils and oil based products from an airborne platform. Selected spectral and temporal characteristics of laser induced fluorescence in an unknown oil are subjected to principal component analysis followed by linear discriminant analysis, thereby facilitating an effective compression of required identification parameters and an enhanced separation of possible groups to which the unknown oil belongs. Identification relies on the availability of a trained data set, comprising principal components with large associated variance and discriminant components with large spread, of all oils likely to be involved in oil spillage at sea. Identification of the unknown oil then involves the application of a classification procedure, (such as K-nearest neighbour or SIMCA), which uses the principal components and discriminant components of the unknown oil and the trained data set.