Seismic Data Visualization
PI: Arie Kaufman
In this project we are developing a visualization cluster system for assisting geophysicists with the task of locating potential oil wells utilizing seismic data in conjunction with other geophysical data. Visualization of terabytes of multiple volumes along with geometric models requires coordination of massive distributed memory on a large scale machine, such as a cluster. We configured and built the Stony Brook Visual Computing Cluster consisting of 66 dual-boot compute nodes connected with Gigabit Ethernet and InfiniBand connections. Each node in the cluster is an HP dualprocessor Pentium Xeon 2.4GHz with 2.5 GB memory. 34 nodes are also equipped with a VolumePro 1000 board with 1 GByte of memory and a GeforceFX 5800 Ultra GPUs with 256 Mbytes of texture memory, and the other 32 nodes are equipped with Quadro FX 4500 GPUs with 512 Mbytes of texture memory. Nine of the nodes have HP Sepia-2a compositing cards connected via a Seismic data is essentially an expensive sonogram of the earth with resolution measured in meters. Our goal is to apply extensive experience in biomedical visualization to seismic data. We have been able to use our Stony Brook Visual Computing Cluster to visualize one or more multichannel volumes mixed with geometric data on a multigigabyte scale using cluster technology, image composition, and hardware assisted volume rendering. Interactive input from the user includes viewpoint, cut planes, transfer functions, and other viewing parameters. To facilitate the work of a geophysicist, our system goal is to register and fuse multiple volumes from a variety of sources into an interactive visualization. (NSF, Terarecon, IBM, HP)