Picture This: Visualization as a Tool for Data Analysis

August 8, 2007

Transformers are no longer just toys from the 1980s.

Bell-Boeing Ospreys are a real-life example of a transforming machine; they can begin flight as a helicopter, with rotors on top, and change mid-flight to operate like an aeroplane, with rotors in front.

A few years ago, however, after two Osprey accidents, the United States government grounded the fleet.

"The planes were having problems during the rotor transition," says Kelly Gaither. "Our job was to figure out why."

Gaither is a visualization expert at the Texas Advanced Computing Center. She and her team simulated, visualized and created a 3000-frame movie to show air movement around the Osprey during transition.

Where art meets science

"When creating this movie, I had a chance to be just as artistic as analytical. People are drawn in when they look at it," she says. "It's visually stunning and scientifically meaningful. We were able to take an abstract idea and translate it for a lay audience."

Gaither's job is to transform digital data into images and animations to help people better understand the data's message.

"The human visual system is very good at noticing patterns from visual input," she says. "These visualizations are much better than just looking at a column of numbers."

In a visual format several aspects of the data can be noticed simultaneously, says Gaither. Thus the principal of computer visualization is the same as that of the bar graph.

"We take something very complex and make it simple by turning it into an image, something the brain has a greater intuitive understanding of," says Mike Papka, a visualization expert at Argonne National Laboratory in Chicago, U.S.

Beyond your TV weather forecast

One very familiar example is a weather forecast: With just a glance we can gather information about temperature, pressure and precipitation across a geographic area.

As science becomes increasingly digitized and data-based, visualization is set to become more and more important. Using visualization, Gaither and Papka can help scientists from all fields: medical science, cosmology, computational fluid dynamics, architecture, geosciences and meteorology.

"You can put detectors or sensors on practically anything," says Papka.

The biggest challenges, according to both Papka and Gaither, are visualizations of very abstract systems. In such cases collaborating with scientists and learning about the research domain is crucial to creating useful visualizations.

Visualization follows some hard and fast rules: A visualization expert must understand the principals of color, must know how the human eye perceives images and must obey the physical laws of the system being visualized.

"Beyond those basic rules," Gaither says, "it's pretty creative."

Gaither and Papka draw on grid computing resources from the TeraGrid to create their computationally intensive images.