The Anomalies in Temperature Anomalies

Abstract

It is standard procedure in the study of global warming to deseasonalize the temperature time series monthly means into monthly “temperature anomalies” by subtracting the corresponding monthly mean temperatures in a reference period. This procedure contains the assumption that the seasonal cycle is fixed and unchanging. This assumption is found to be invalid because the source data from measurement stations show that long term trends in the twelve calendar months are not uniform. An alternative procedure is proposed for the study of warming trends net of the seasonal cycle that involves the study of the calendar months separately using a “trend profile” tool to investigate trends within the full span. The proposed methodology is robust to the changes in the seasonal cycle.

Keywords: global warming, climate change, temperature anomalies, time series analysis, deseasonalized analysis, statistics, applied statistics