Scripps Institution of Oceanography

Seeing the Big Picture

By Frank Lacombe

Computer scientist Atul Nayak and seismologist Debi Kilb view a theoretical model of a deformation along the San Andreas fault on the iCluster display wall at the Scripps Institution of Oceanography Visualization Center.

“Here at the Visualization Center, we’re not interested in building something just to see if we can do it,” says Dr. Graham Kent, research geophysicist and director of the Visualization Center at the Scripps Institution of Oceanography (SIO). “A successful technology project for us is one that yields immediate and tangible results for our scientists. We support the scientists at SIO by providing the tools they need to do better collaborative science among themselves and with scientists elsewhere, and to share scientific information with the larger community.”

The most recent addition to SIO’s visualization infrastructure is iCluster, a 50-megapixel “wall of information” comprising 12 Apple Cinema Displays driven by six Power Mac G5 systems running open source graphics middleware developed at the University of Illinois at Chicago’s Electronic Visualization Laboratory (EVL). The iCluster project is funded by the National Science Foundation (NSF) EarthScope program’s USArray Network Facility (USANF) and the Real-time Observatories, Applications, and Data management Network (ROADNet, also NSF-funded) at SIO/UCSD, with the collaboration of UCSD’s California Institute for Telecommunications and Information Technology (Calit2).

When Disaster Strikes

In the wake of earthquakes, tsunamis, and other natural disasters, news organizations depend on the Earth scientists at SIO to get the facts. That’s when the institution’s cutting edge work in visualization moves into the public eye. These days, the news crews are often pointing their cameras at the array of Apple Cinema Displays that form the iCluster’s eye-catching frontend.

“iCluster’s Apple Cinema Displays create a much sharper and brighter image that comes across very well on camera. The news crews love it.”

Atul Nayak, a computer scientist at SIO, worked with Kent, SIO seismologist Dr. Debi Kilb, and others to create iCluster. “When the news channels come here after an earthquake, they want our scientists to explain the seismic forces that caused the event and give the TV audience an idea of what to expect in terms of aftereffects,” Nayak says. “Until recently, we solely used our Panoram Technologies projection-based display system to present visualizations to the reporters. But the iCluster’s Apple Cinema Displays create a much sharper and brighter image that comes across very well on camera. The news crews love it.”

Overcoming Trade-Offs

There’s more to the new interest in tiled flat-screen displays among visualization experts than news camera-friendliness. Scientific visualization allows many different scientists to look at the same data, at the same time, in the same place, and at full resolution. They notice different aspects of the data and call each other’s attention to them. In his role as the Visualization Center’s director, Kent pays close attention to how scientists do their collaborative work and is especially sensitive to emerging trends.

“One of my goals as director of the Visualization Center is to have as many visualization options as possible,” says Kent. “Our 10’x30’ curved-screen Panoram stereo projection system is amazingly effective when collaborating with 40 scientists or running outreach programs for dozens of kids — and at one time it represented the cutting edge in scientific visualization tools. Today, we need more resolution than standard projection-based solutions can realistically provide.”

Even when the cost of immersive projection environments like Panoram was many times higher than they are today, plenty of customers thought the investment was worthwhile. Oil and gas exploration companies, for example, have been consistent early adopters of visualization technologies because the astronomical cost of developing a prospective extraction site makes bad decisions ruinously expensive.

Today’s scientists generate huge amounts of field data through a variety of sensors and networking devices. They perform large-scale computations to understand the science better and analyze the data to make predictions. These activities tend to generate enormous data sets, which continue to grow rapidly. Scientists need visualization technologies that help them understand the data sets they work with now while laying the groundwork for even larger and more interactive data sets, applications, and media streams on the horizon.

In an ideal circumstance, an entire data set fits on the display at full resolution. Where this is not possible, the choices are to down-sample the image to fit on the display (sacrificing the details) or to display only a portion of the image at a time (sacrificing the big picture). In either case, insufficient resolution leads to painful trade-offs.

“There’s a lot of value tied up in getting the best possible interpretation of the available data,” Kent observes. “Lots of people — including us here at SIO — are willing to invest heavily to make sure that happens. But at 3.2-megapixel total resolution, the projection approach to visualization is not keeping pace with the data sets scientists are working with today. The iCluster project is our attempt to get out ahead of that curve.”

Looking at the available options, the iCluster team decided the best way to accomplish that goal and to complement its existing environment was to construct a tiled array of flat-screen displays with a target resolution of 50 megapixels, built on a graphics middleware foundation that could accommodate arbitrarily large display resolutions.

1 2 3 Next Page >