A Study of the Numerical Data Compression for Large-Scale Flow Visualization

Abstract

Very large scale computational fluid dynamics (CFD) are frequently attempted to obtain high resolution solutions or to solve large scale problems and bulky numerical data are generated. The data must be visualize to grasp and clarify precise flow characteristics. In order to reduce computer loads of the large data transfer and I/O, compression and restoration schemes are studied. In our study, the size of the original data file are compressed approximately by 1/100 by making the data thin out properly and by changing data type. The compressed data are transferred to a graphic workstation and read in a visualization software with less computer efforts. Then, they can be restored by applying an interpolation technique with allowable losses of the original fluid information.