The objective of the proposed program is to investigate and develop visualization tools that address the special requirements of large-scale computing. Efficient analysis of large, multi-dimensional databases consisting of seismic data requires the interactive use of high performance graphic tools. Processing seismic data to produce graphic images produces a quantity of information on the order of megabyte per image, while large-scale databases are on the order of a gigabyte. This several order of magnitude difference produces a severe bottleneck when using existing graphics workstations to generate images. This problem is worsened when time-dependent calculations or data are visualized, since in these cases large amounts of data must be transferred and rendered repetitively. Rendering various types of data, such as volumetric data, can also pose major difficulties since current algorithms are computationally intensive, requires a large amount of memory and are beyond the current level of performance of graphic workstations. High performance rendering tools consisting of thoroughly integrated software and hardware need to be developed. These tools are then incorporated into a complete VisualizationEnvironment (VE) consisting of thoroughly integrated supercomputer and/or special purpose graphics hardware and software. This VE will provide the required advanced visualization capabilities, minimize data transfer bandwidth, and achieve a properly balanced computation load among the various systems involved. The proposed program is aimed at developing high performance rendering tools for scientific visualization of seismic data. These tools require implementation on supercomputers and/or special purpose graphics boards. The proposed tools will provide researchers, designers and other users an efficient means to interactively analyze large multi-dimensional seismic databases. Anticipated benefits/potential commercial applications - earthquake detection and prediction, oil explorations, nuclear testing detection.
