Objective Identification of Extreme-Most Daily Maximum/Minimum and Hour-to-Hour Historical Temperature Patterns Using Principal Components Analysis

• Activity Number: 255
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Charles Fisk*+
• Keywords: Principal Components Analysis ; Extreme Temperature Patterns

Abstract

A familiar means of depicting day-to-day max/min temperature data is the floating-bar or "hi-lo" chart. In viewing individual charts, the question may arise on how typical the patterns are relative to those of other years covering the same calendar segment, and to this end, it should be useful to have objective, pattern-oriented means of characterization and comparison. The same could apply to line-depicted day-to-day hourly temperature observations. Utilizing Downtown Los Angeles daily maximum/minimum (1921-2010) and Los Angeles International Airport ("LAX") hourly (1947 and 1939-2010) temperature data, the utility of Principal Components Analysis (Correlation and Covariance, each unrotated) is demonstrated for this purpose. First component correlation loadings characterize "shape", first covariance loadings, "spread". The highest and lowest correlation and covariance loadings identify the most anomalous patterns in terms of these attributes, and for a selection of calendar periods, graphs of the most extreme configurations are presented as illustrative examples.