Fast algorithms for removing atmospheric effects from satellite images

Abstract

Remote-sensing satellites collect vast amounts of earth data, but correcting for the atmospheric effects that contaminate the imagery is computationally intensive. Using previous codes, high-accuracy atmospheric correction of a standard regional image area would have taken years on a workstation. An improved algorithm is described that takes 13 minutes and parallelizes well. An implementation of the algorithm is shown to be scalable in terms of number of processors, internal memory size, and number of I/O nodes. A case study shows it to be quite effective.