In-scene techniques for atmospheric correction of remotely sensed multispectral imagery have been collected and applied to a series of image sets. The methods used rely on image derived statistics and a minimal use of ancillary data. Five different schemes for estimating upwelled radiance are presented. Upwelled radiance predictions are compared to estimates derived from reflectance panel calibrations and the radiative transfer model LOWTRAN. Inversion to reflectance space is performed with an in-scene technique that draws upon knowledge of surface material reflectance spectra and the upwelled radiance estimates generated by accompanying techniques. This research attempts to answer the following questions: What features must be present for robust implementation of each routine? In which spectral bands does each routine provide good estimates of upwelled radiance? What are the limits in spectral and spatial resolution for each routine? Drawing on the conclusions of this research a modus operandi is suggested for in-scene atmospheric correction.
Library of Congress Subject Headings
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Barnes, Paul, "In-scene atmospheric correction for multispectral imagery" (1997). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus