The importance of locality in the visualization of large datasets

Many scientific phenomena in large high-resolution datasets such as the U.K. Ocean Circulation and Advanced Modelling (OCCAM) ocean model are better discovered through visualization than by algorithmic analysis: it is often more straightforward to see a feature than it is to characterize it numerically. Using traditional rendering techniques, the size of modern datasets presents a challenge for even high-end graphical supercomputers, and the cost of such hardware limits its availability for day-to-day analysis. We present an architecture that brings visual analysis to the desktop by exploiting consumer-grade graphics hardware in order to provide initial interactive exploration and Web services to enable finer-grained analysis and interoperability with traditional visualization tools. Copyright (c) 2006 John Wiley & Sons, Ltd.