Processing Satellite Images on Tertiary Storage: A Study of the Impact of Tile Size on Performance

摘要

Before raw data from a satellite can be used by an Earth scientist, it must first undergo a number of processing steps including basic processing, cleansing, and geo-registration. Processing actually expands the volume of data collected by a factor of 2 or 3 and the original data is never deleted. Thus processing and storage requirements can exceed 2 terrabytes/day. Once processed data is ready for analysis, a series of algorithms (typically developed by the Earth scientists) is applied to a large number of images in a data set. The focus of this paper is how best to handle such images stored on tape using the following assumptions: (1) all images of interest to a scientist are stored on a single tape, (2) images are accessed and processed in the order that they are stored on tape, and (3) the analysis requires access to only a portion of each image and not the entire image.