Fast Distributed and Parallel Pre-processing on Massive Satellite Data Using Grid Computing

Abstract

The rest of portion is under distributed processing. To ensure independence of distributed processing, each processing program has own working directory. After verification, distributed process environment for pre-processing performance can be improved up to 560.65% from single processing system. Also, file that is stored in NAS storage is moved to local storage during the processing of satellite image data and the processed results are saved into NAS storage. This method is compared with the baseline. Through the results, Conversion of Level-1B file to Level-2 file is CPU bound task, and local storage shows the greater Extensibility.