ABSTRACT
Atmospheric data are seldom apposite for the use of the traditional tools of statistics. In addition to the more dramatic effects that occur on short time scales, the data abound with diurnal, seasonal, seasonal, annual, and multi-year patterns over a broad range of magnitudes. Replicate experiments are often impossible or unfeasible. These properties suggest that innovative treatment of the data (termed exploratory data analysis or EDA) may bring substantial rewards. In this chapter we hope to demonstrate that such a statistical approach is well within the feasibility of modern computational systems and that considerable useful information can result from taking such an approach to the analysis of atmospheric data.
