by Lynda Lester, NCAR/SCD -- NCAR scientists have a long track record of creating scientific visualizations as an end product in atmospheric research, using them to illustrate presentations and publications. Visualization, however, has rarely been an integral part of the analysis process itself. Only after research is complete and data have been explored are results visualized. SCD software engineer John Clyne and a team from SCD's Visualization and Enabling Technologies Section are trying to change that. As part of a $2 million grant from the National Science Foundation's Information Technology Research (NSF ITR) program, they are developing new tools that will combine scientific visualization with traditional software for statistical data analysis. A growing problem "Computational datasets are getting too big to analyze using traditional methods," John says. The problem is especially acute in turbulence research, which is essential to understanding the dynamics of the atmosphere, the oceans, and the sun. High-resolution turbulence models have many time steps — that is, the data vary over time — and consequently, generate torrents of data output. "Tremendous resources go into running these large numerical simulations, so we'd like to think we can extract all the science out of the data," he explains. "But that's getting harder to do; a lot of the data remain unexplored. We're at the point where we can compute data that we can't completely analyze." In recognition of this gap in information technology, and in collaboration with investigators at the University of California-Davis, Ohio State University, NASA's Jet Propulsion Laboratory, and the University of Tokyo, John co-authored a successful proposal to develop new approaches to visualizing time-varying scientific data. Collaborative R&D The five-year ITR grant from NSF has two components: research and development. "It's a neat partnership," notes John. "Our university collaborators are leading the research effort, trying to find new ways of accessing and exploring these datasets. Their efforts are being driven by NCAR's scientific needs — we're giving them some interesting problems, and they're working on ways to solve them for us. In turn, we're taking the results of their research and deploying new tools for the user community." John heads the development effort, overseeing lead software architect Alan Norton, who joined SCD in May; Joey Mendoza, an SCD Vislab technician; and Hiroshi Akiba, a Ph.D. candidate from UC Davis who was an SCD student assistant for part of the summer. "The hope is that if you have the right tool, visualization can be a powerful advantage for scientific discovery — but only when combined with the more traditional analysis tools that scientists require," John observes. To that end, the SCD team is working closely with turbulence researchers in various NCAR divisions, hoping to have an initial release of the new software within a year. The software package, an integrated environment for data analysis, is called VAPoR. International steering committee The ITR grant, titled "Gleaning Insights in Large Time-Varying Scientific and Engineering Data," is part of NSF's ongoing mission for funding innovative research that benefits science. The project is guided by an international steering committee comprised of representatives from several NCAR divisions and programs (the High Altitude Observatory, the Climate and Global Dynamics Division, and the Advanced Studies Program), the University of Colorado, and l'Observatoire de la Côte d'Azur, a French research institute. Help for researchers "This is an opportunity to do something that will have a real impact on researchers struggling with hard problems," John says. "It's a chance to improve scientific productivity and further scientific advancement. And it's fun stuff. I'm excited about it." For more information For an exciting look at what SCD is doing with scientific visualization and its state-of-the-art Vislab, see the Visualization and Enabling Technologies Section's home page: http://www.vets.ucar.edu/index.shtml — Lynda Lester