Visualizing Science in Action
A surgeon can repair a human heart, but like all living organs, the heart presents a host of problems to scientists who want to understand it in detail. X-rays, probes and scans show only a partial picturea snapshotwhile often what is needed is a motion picture of how all the parts interact. In 1993 scientists at New York University came up with a solution. Working at the NSF-funded Pittsburgh Supercomputing Center, they created the first three-dimensional, animated model of a beating heart.
That first successful "heartbeat" required nearly a week of computing time and represented fifteen years of work by mathematicians Charles Peskin and David McQueen. Subsequently, the work has had broad influence in studies of biological and anatomical fluid flow.
Other simulations demonstrate the dynamics of much larger events, such as tornadoes. Scientists at the University of Illinois have traced air motion within and around tornadoes by introducing thousands of weightless particles into the flow. With that information, and with the computing power at the NSF-supported National Center for Supercomputing Applications (NCSA), also at the University of Illinois, they created a model that provides a closer look at updrafts, downdrafts, and strong horizontal changes in wind speed.
Robert Wilhelmson, an atmospheric computer scientist, began modeling storms almost thirty years ago in hopes of better predicting severe occurrences. One of the founders of NCSA, he wanted from the beginning to model how storms evolve. Wilhelmson and his research group have pushed storm visualizations from static two-dimensional images to three-dimensional animations at ever-greater resolution and over longer time spans.
All the sciences use the visual arts in some form or another, depicting everything from molecules to galaxies. Engineering, too, relies on detailed renderings. Over the years, NSF has funded the development of computer visualizations in many fields, while at the same time challenging computer specialists to go back to the basics and learn how to make these visualizations more accurate and more useful as scientific predictors.
Ensuring this accuracy, according to Cornell University's Don Greenberg, a long-time NSF grantee, entails making sure that computer-generated visualizations obey the laws of physics. Cornell, one of five institutions in the NSF Science and Technology Center for Computer Graphics and Visualization, addresses this issue by developing ways to incorporate the physics of how light behaves and how our eyes perceive it. Their renderings look like architectural photosstudies of light and space. And their research has been used by General Electric Aircraft Engines, Battelle Avionics, Eastman Kodak, and others.
Moving computer visualizations from what is acceptable to what is most useful from a scientific standpoint has taken a lot of work, says Greenberg. "It's easy to make visualizations believable; Jurassic Park and Star Wars did a fine job of that," he says. "But they weren't very accurate. It's much harder to make them accurate."