Thrust Belt Imaging
The Tools for Accurate Seismic Imaging
From left to right, the founders of TBI: Dale Schack, Jon Gittins, and Rob Vestrum.
Although much of the industry has moved toward a data-driven approach to seismic image analysis, largely limiting itself to oceanic exploration in which there’s more data density and less noise, TBI takes an interpretive approach to image the subsurface below rugged environments. To an analyst, interpretive results are only as accurate as the monitor displaying the imagery. TBI can display 5120-by-1600 seismic images on 30-inch Apple Cinema Displays, making them a key component in TBI’s decision to use Mac systems.
“The increase in productivity resulting from the size of the displays cannot be overstated. The larger displays allow me to look at the data in detail while retaining an understanding of how the details fit within its geologic context. I cannot imagine going back to smaller displays.”
“For our interpretive workstations, what we needed first and foremost were large displays — the larger the better,” Schack says. “Apple has these beautiful 30-inch displays; by combining a Power Mac G5 with a nice graphics card, you can drive two of them. With the amount of data we need to display, this seemed like a natural solution.”
Gittins adds: “The increase in productivity resulting from the size of the displays cannot be overstated. The larger displays allow me to look at the data in detail while retaining an understanding of how the details fit within its geologic context. I cannot imagine going back to smaller displays.”
TBI works hand in hand with its clients. They often sit together at TBI’s interpretive workstations to analyze the data, making decisions on velocity models and parameters to send back to the cluster for processing. A fast, user-friendly experience is crucial to a positive exploration process.
“When we sit down with a client, we can try a few different interpretations and send them off to the cluster,” Vestrum says. “In 10 minutes, we get back all the calculations we need. On a single system, this would have taken most of the day. While we’re waiting for results, we look at another dataset. We need really fast turnover so we can be experimental.”
At the end of an interpretive session, the geologist will have a new velocity model. TBI’s Python script is then executed, which creates a sequence of jobs designed to run one after another. Each job has approximately 40 tasks. Xgrid sends each task out to a different processor on the cluster and returns the results.
Working in a Stable Environment
The long-term benefits of using Apple’s hardware solutions and robust development environment are already beginning to appear, including the luxury of not needing a full-time system administrator to care for TBI’s compute environment. In fact, since moving the cluster into their machine room months ago, TBI’s developers and users need an excuse just to visit the hardware.
“We originally had the system running in our office space while the machine room was being built out,” Schack says. “We would watch the system whenever we submitted a job to the cluster, just to watch the processors light up. Since we moved the cluster into the machine room, we haven’t had to go in there once, aside from taking a picture for this article. I can control everything through Apple Remote Desktop and have never had a problem with the hardware.”
TBI uses Apple Remote Desktop, Apple’s desktop management solution, for software installations, package updates, and detailed reports of system status. Tasks can be completed in parallel, so large clusters can be administered in the same amount of time as a single desktop computer.
TBI is opening new areas to explore for oil and natural gas without traditionally high-risk drilling from uncertain image analysis. As energy companies realize the benefits garnered from making accurate corrections for anisotropic effects, the natural resources in rocky foothills such as those found near TBI’s home in Calgary will be used more efficiently, benefiting both TBI’s clientele and regional economies. As TBI’s project base grows, so will its Xserve cluster. Xserve is a proven supercomputing platform that has been scaled up to thousands of nodes in production and the research environment.
“It’s just rock-solid, steady, and hums through the jobs,” Schack says of TBI’s Xserve cluster. “It’s fantastic. I know I sound like a rabid Mac fan — which I wasn’t — but after the ease of this experience, I’m thinking of becoming one.”
