The Top Five Problems that Motivated My Work

heresa-Marie Rhyne has lately been organizing efforts for a number of researchers to produce lists of top visualization problems, and I am flattered to be included. A recent Visualization Viewpoints column (see the July/August 2004 IEEE CG&A) featured Chris Johnson's excellent list, and now it's my turn. Since the May 1999 VisFiles column 1 already describes my list of the top ten problems that will drive future visualization work, this Visualization Viewpoints column is a look back at the top five problems that drove my own work developing Vis5D, Cave5D, and VisAD (see the " Vis5D, Cave5D, and VisAD " sidebar). Some of these problems are high-minded and some are grubby and gritty. This is the mix of problems that I learned from Verner Suomi, the founder of the Space Science and Engineering Center where I do my visualization work. Suomi was literally the inventor of weather satellites. Among many other technical accomplishments, he held the patent on the spin-scan radiometer, which was the basis for the geostationary satellites that until recently produced the cloud animations we see on TV weather shows. He also lobbied the US congress for the funds to develop these satellites and was heavily involved in building the prototypes. He was a visionary, a politician, and what my father would have called a " dirty-fingered engineer. " I hope these problems reflect this combination of concerns that lead to useful change. I start at number five, and count down to the most important problem. Scientists have too much data to understand it as printed numbers, or even as simple 2D graphs. Every major weather modeling center has a large room whose walls are covered with hundreds of printed maps and charts showing the observations used to initialize their model and the data grids that come out of their model. When I started working on visualization in the late 1970s, some meteorologists were building 3D Plexiglas models depicting weather scenarios. Clearly they were anxious for a way to combine all their different vertical levels and different atmospheric fields into unified 3D pictures. During the 1980s, visualization researchers strove to meet this need by developing techniques and software for making 3D depictions of data. But single 3D images lacked a vital feature of the 3D Plexiglas models: the ability to interactively rotate the viewpoint. In 1988, commercially available workstations appeared that could interactively rotate shaded 3D graphics. This …