Atmospheric Correction for Long Slant Path Imagery

This proposal by Spectral Sciences, Inc. (SSI) addresses the need for innovative, automated, and fast software for accurate atmospheric correction of hyperspectral and multispectral imagery in highly off-nadir, long slant path (long standoff distance) remote sensing. In Phase I we assembled long slant path spectral imagery, performed atmospheric correction on the data, characterized and identified sources of error, and developed novel algorithms and strategies for improved correction, particularly of aerosol/haze. The new algorithms include a regression method for visibility determination that does not require the presence of specific materials in the scene and a prototype automated version of QUAC (QUick Atmospheric Correction), a semi-empirical method that is very useful with radiometrically uncalibrated data and highly promising for real-time data processing. In Phase II, the algorithms will be fully implemented and integrated into state-of-the-art AFRL/SSI atmospheric correction software. Upgrades will include an improved adjacency correction method designed to provide the best possible accuracy with long slant paths, and a fully automated algorithm for near-real-time atmospheric correction in operational image processing systems. Software validation will be conducted using newly acquired field data and high-fidelity spectral image simulations.