Trend Profiles of Atmospheric Temperature Time Series

Abstract

Time series of lower atmospheric and surface temperatures tend to be chaotic, non-linear, and volatile with known violations of OLS assumptions. Their OLS trends differ significantly depending on the choice of sample periods and the endpoints they imply. In this brief note we show that the trend properties of a temperature time series may be explored more fully by constructing what can be called a trend profile. For a time series of monthly mean temperature anomalies longer than 30 years, we demonstrate a trend profile procedure in which OLS trends for all possible 30-year intervals in the study period are plotted against the month in the middle of each interval. The shape and structure of these curves are more robust and provide greater insight into the trend behavior of the temperature time series than a single OLS trend line across the study period bounded by arbitrary end points of convenience.

Keywords: Climate Change, Global Warming, UAH, Hadcrut, Surface Temperature, Troposphere, Time Series, Chaotic, Trend Analysis, Pause in Global Warming, Global Warming Hiatus, Applied Statistics, OLS Trend, Trend Profile