Interactive Visualization Tools for Exploring Multivariate Atmospheric Science Data

A key principle in exploratory data analysis is to graphically examine the data as a first step in the analysis. This rule should apply to all data sets, except that with large volumes of data, the task can be quite cumbersome even with the aid of substantial computing power. Two- and three-dimensional static plots of multivariate data, though helpful, often do not fully reveal patterns of inter-relationships between variables. For high-dimensional data, interactive and dynamic approaches to graphical displays are most informative. These methods are becoming readily available and are highly applicable to large data sets. Visual inspection using these methods allows one to 1) uncover previously unknown features, 2) aids in reaching an understanding of the features, and 3) allows for the cross-validation of conclusions and statements about the data made from other methodologies. In this presentation, a demonstration of simple visual tools applied to the exploration of atmospheric science data sets will be given. XGobi software linked with ArcView will be used to demonstrate interactive visualization of multivariate spatially-referenced data. A discussion of interactive and dynamic methods for exploring atmospheric science data will be included in the presentation.