Atmospheric profile variability impact on the performance of hyperspectral remote sensing detection systems

Atmospheric profile variability impacts the effectiveness of hyperspectral remote sensing systems to identify geological materials from space. This variability alters the "truth" spectrum (e.g. reflective or emissive) by modifying its spectrum through such atmospheric factors as transmissivity, upwelling radiance, and downwelling radiance. Incomplete characterization of these effects can lead to incorrect target assignments and therefore can impact subsequent quantification estimates. It is thus necessary to understand the statistical variability of the atmosphere as seen from space and ultimately how it impacts uncertainty in the measured spectra of materials. This paper estimates atmospheric profile variability over the Hawaiian Islands for a period of one year. This variability is compiled over a daily interval and reported for each month. The material spectral variability resulting from atmospheric temporal variance is estimated using 6 selected geological igneous rock classes across the VIS/SWIR region. The effectiveness of the Spectral Angle Mapper (SAM) to effectively discriminate between rock classes with limited a priori knowledge of atmospheric conditions is also estimated. The ultimate goal of this effort is to assess to what level, if any, supplemental local meteorological data and/or in situ atmospheric retrieval methods are required to assure system discrimination/identification performance objectives