Abstract:

Highresolution satellite data is of  powerful potentiality in applications such as surface object identification. However, the related quantitative remote sensing usually requires an accurate atmospheric correction, which is quite timeconsuming. The serial and parallel processing of atmospheric correction in the PCbased cluster system is studied. The atmospheric corrections of HJsatellite CCD data over North China Plain in July 2012 are shown in this paper, and the execution time of each step in the serial and parallel processing is also analyzed, which proves the high feasibility of atmospheric correction parallelization. To avoid unbalanced loads and frequent communications, we design a data division strategy based on pixel features and compare the performance of three different division strategies, which indicates that the proposed strategy has higher reliability and can achieve a higher speedup ratio.

Key words: remote sensing;atmospheric correction;aerosol optical depth;data division;balanced loads