A Robust Test for OLS Trends in Daily Temperature Data

Abstract

Trends in time series data estimated with OLS linear regression may be tested with a robust procedure that is less sensitive to influential observations and violations of regression assumptions. The test consists of comparing the average age of data tritiles. If the higher tritiles are newer a rising trend is indicated for the sample period. If the higher tritiles are older a declining trend is indicated. If neither of these conditions is met, no sustained trend in the sample period may be inferred from the data. Daily temperature data from selected USHCN and USCRN stations are used to demonstrate the utility of the proposed methodology.

Keywords: trend analysis, rising temperatures, global warming, climate change, Hurst exponent, robust statistics