EXPLORING CLIMATE VARIABILITY WITH SELF-ORGANIZING MAPS AND GOOGLE EARTHTM

Self-organizing maps (SOMs) are a relatively new tool for analyzing variability in complex (e.g., large, multidimensional, multivariate) geophysical datasets. Google Earth^TM (GE) is an even newer tool that makes it easy to place a wide variety of data (e.g., images, 3-D models) in a global, interactive geographic context. Here we explore what value GE can bring to the visualization of SOM-based research.

SOMs provide a powerful, nonlinear technique to optimally summarize and visualize complex data using a user-selected number of "icons" or SOM states, allowing rapid identification of preferred patterns and many other important aspects of data variability. SOM analysis produces a discrete, nonlinear classification of the continuum of conditions within the input data set without prior specification (or knowledge) of the "correct" output. SOMs also robustly handle missing values.

The reasonable learning curve of the freely available GE software provides a relatively "low-cost" opportunity to explore visualization of the spatial and temporal data produced by a SOM-based analysis. Here we will be exploring the utility of GE in the context of two projects: relating North Atlantic atmospheric circulation to Greenland ice core-based accumulation records and a browser for regional-scale views of global-scale climate model projections of the future. In the ice core study, 45 years (1957-2002) of annual accumulation data were analyzed to study spatial and temporal patterns of snowfall across most of the Greenland ice sheet. Similarly, the atmospheric circulation analysis used 45 years of monthly winter (DJF) mean sea level pressure data to study the North Atlantic Oscillation and similar variability in this domain. Synthesizing the results from these two studies should improve our understanding of this complex problem. GE should prove to be an important visualization component in this synthesis effort. Similarly, it plays a significant role in plans to develop an interactive browser tool for viewing SOMs-based regional-scale climate change projections from global-scale climate models.