PNNL

Abstract

Near infrared hpyerspectral imaging is finding utility in remote sensing applications such as detection and quantification of chemical vapor effluents in stack plumes. Optimizing the sensing system or quantification algorithms is difficult since reference images are rarely well characterized. The present work uses a radiance model for a down looking scene and a detailed noise model for a dispersive and Fourier transform spectrometer to generate well-characterized synthetic data. These data were used in conjunction with a classical least squares based estimation procedure in an error analysis to provide estimates of different sources of concentration-pathlength quantification error in the remote sensing problem.