A non-linear optimal estimator for plume concentration retrieval, using airborne hyperspectral measurement

Anthropogenic sources, especially industrial, emit into the atmosphere gases and aerosols, which play an important role in atmospheric exchanges. The use of new hyperspectral airborne sensors in the infrared range at high spatial resolution opens the way for further improvements in the characterization of industrial plumes. Existing methods, based on linear or Bayesian approach, present two major limitations: the heterogeneous environment impact on retrievals and the spatial and vertical extents of the plume are not considered. In this paper, a new method for plume characterization is presented to overcome such limitations. This method uses an accurate non linear formalism of cloud gas radiative impact. It includes furthermore a ground classification of the scene, in order to take into account the soil heterogeneity and spectral behaviour. This strategy will be validated using synthetic scenes of industrial area, and real data from Telops Hyper-Cam airborne platform.