Passive multi-spectral and hyper-spectral optical sensors offer great potential for remote-sensing in general and detection of low-concentration toxic species in particular. Correction for propagation through the intervening atmosphere (with molecular, aerosol and hydrosol constituents) is often the performance-limiting factor, a step which is unavoidably imprecise given uncertainties in the physical description of the atmospheric column. We propose to describe a general approach for mitigating the effects of these uncertainties using spectral sub-space projection, and to apply the technique to the difficult problem of trace-gas detection and quantification for environmental monitoring. While the approach may be applied to data collected by ground, aircraft or space-based sensors, we will illustrate it with hyperspectral off-nadir simulated imagery appropriate to a high-altitude aircraft. In addition to a description of the approach and simulated HSI data, detection/quantification performance using matched-filters for a number of common pollutants with and without sub-space projection will be presented. Use of matched-filters for detection has the additional benefit of quantitatively ranking the spectral weight of measurements for a given trace gas with the specified conditions and uncertainties. Such results bear directly on sensor design issues --spectral extent and resolution--and suggest that data volumes could be dramatically reduced by projection onto the target sub-space (s) prior to down-link.

Date Published: 20 January 2005
PDF: 12 pages
Proc. SPIE 5655, Multispectral and Hyperspectral Remote Sensing Instruments and Applications II, (20 January 2005); doi: 10.1117/12.578952

Published in SPIE Proceedings Vol. 5655:
Multispectral and Hyperspectral Remote Sensing Instruments and Applications II
Allen M. Larar; Makoto Suzuki; Qingxi Tong, Editor(s)