Uncertainty in Ranking the Hottest Years of U.S. Surface Temperatures

Ranking years based on statistical estimates of regional and temporal averages is subject to uncertainty. This uncertainty can in fact be quite substantial and can be described by the rank distribution of an ensemble of such averages. The authors develop a method for estimating it using simulation. The effect of temporal correlation is quite limited in the case studied in this paper: the contiguous United States' annual-mean temperature. The method also allows assessment of derived quantities such as the probability of a given year being one of the 10 warmest in the historical record.