An improved Landsat Image Mosaic of Antarctica

A revised Landsat Image Mosaic of Antarctica （LIMA） is presented, using the 1073 multi-band scenes of the original Land- sat-7 ETM＋ LIMA image collection available at the United States Geological Survey （USGS： http：//lima.usgs.gov/）. Three improvements have been applied during the data processing： （1） DN saturation is adjusted by adopting a linear regression, which has a lower root mean square error than the ratio regression used by LIMA; （2） solar elevation angle is calculated using pixel-level latitude/longitude and the acquisition time and date of the central pixel of the scene, improving slightly upon the bi- linear interpolation of the solar elevation angles of scene comers applied in LIMA; and （3） two additional image bands, Band 5 and Band 7, are sharpened using the panchromatic band （Band 8） and a Gram-Schmidt Spectral Sharpening algorithm to more easily distinguish snow, cloud and exposed rocks. The final planetary reflectance product is stored in 16-bit bands to preserve the full radiometric content of the scenes. A comparative statistical analysis among 12 sample regions indicates that the new mosaic has enhanced visual qualities, information entropy, and information content for land cover classification relative to LIMA.

A revised Landsat Image Mosaic of Antarctica (LIMA) is presented, using the 1073 multi-band scenes of the original Landsat-7 ETM+ LIMA image collection available at the United States Geological Survey (USGS: http://lima.usgs.gov/). Three improvements have been applied during the data processing: (1) DN saturation is adjusted by adopting a linear regression, which has a lower root mean square error than the ratio regression used by LIMA; (2) solar elevation angle is calculated using pixel-level latitude/longitude and the acquisition time and date of the central pixel of the scene, improving slightly upon the bilinear interpolation of the solar elevation angles of scene corners applied in LIMA; and (3) two additional image bands, Band 5 and Band 7, are sharpened using the panchromatic band (Band 8) and a Gram-Schmidt Spectral Sharpening algorithm to more easily distinguish snow, cloud and exposed rocks. The final planetary reflectance product is stored in 16-bit bands to preserve the full radiometric content of the scenes. A comparative statistical analysis among 12 sample regions indicates that the new mosaic has enhanced visual qualities, information entropy, and information content for land cover classification relative to LIMA.