Change point detection in ERA5 ground temperature time series

Abstract: We analyze the ERA5 reanalysis 2-meter temperature time series on all land grid points using change point analysis. We fit two linear slopes to the data with the constraint that they merge at the point in time where the slope changes. We compare such fits to a standard linear regression in two ways: We use Akaike's and the Bayesian information criteria for model selection, and we test against the null hypothesis of no change of the trend value. For those grid points where the dual linear fit is superior, we construct maps of the time when the trend changes, and of the warming trends in both time intervals. In doing so, we indentify areas where warming speeds up, but find as well areas where warming slows down. We thereby contribute to the characterization of local effects of climate change. We find that many grid points exhibit a change to a much stronger warming trend around the 1980s. This raises the question of whether the climate system has already passed some tipping point.