|
Abstract:
|
The proliferation of data in the computer age has made the search for methods of compressing higher dimensions onto a flat, two-dimensional graphical display more difficult and important. Here, we illustrate a graphical, exploratory approach using a generalization of Edward Tufte's 'sparklines,' or compact time series, that can be used to represent data of continuous variables distributed in space and time. Using this and other graphical tools, we identify certain peculiarities of seven four-year monthly time series of climate data on a 24 by 24 raster covering an area of South and Central America. Our methods of graphical display reveal variable truncation and multiple covariation, which must be addressed before applying predicative autoregressive models.
|
