Answers are grounded in this paper and cite their sources.
Accuracy and uncertainty of the probability estimates associated with extreme events of long return periods in a typical event attribution context are not well understood, making it difficult to interpret the meaning of event attribution results. This study evaluates the effectiveness of approaches used in event attribution studies for estimating the return periods of hot extremes based on large samples from large ensembles of climate model simulations. We found that, even with large sample sizes, annual maxima of temperature extremes do not fit well to a generalized extreme value (GEV) distribution in the far right tail, leading to an overestimation of return period. However, the maxima of multi‐year blocks generally fit well to a GEV distribution, improving the accuracy of exceedance probability estimates for very rare events. For events with shorter return periods, such as those spanning tens of years or less, fitting annual maxima to a GEV distribution can generally still provide robust estimates. Additionally, estimates based on data samples of similar lengths to observational records are less reliable due to limited sample sizes. These findings highlight the need for caution when interpreting event attribution results for events with long return periods.
No takes yet. Share an insight, caveat, or question.
Liang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: