Stereo Cloud Heights From Multispectral IR Imagery via Region-of-Interest Segmentation

Multispectral thermal imagery acquired from low Earth orbit was used to develop a method of cloud-height determination that applies image brightness temperature histograms and region-of-interest (ROI) image segmentation as a processing step prior to stereo-height retrieval. The National Aeronautics and Space Administration´s Infrared Spectral Imaging Radiometer (ISIR) acquired all imagery during the August 1997 STS-85 mission of the Space Shuttle Discovery. ISIR, the first Earth-observing spectroradiometer to employ an uncooled large-format microbolometer-array focal plane, provided continuous coverage in four spectral bands along track in an 80-km-wide swath. ROI segmentation created binary cloud masks ranging over brightness temperatures in the imagery for which the parallax was determined by a two-dimensional correlation method, allowing stereo heights to be determined using the standard parallax equations. A subpixel parallax algorithm allowed stereo heights to be determined with a precision of roughly plusmn0.39 km for clouds in the observed altitude range of 0.5-10 km