The Arctic climate system is very sensitive to external perturbations, which results in more rapid surface air temperature (SAT) changes in the Arctic compared to lower latitudes. This study aims at assessing the performance of global climate models from the phases 5 and 6 of the Coupled Model Intercomparison Project (CMIP5 and CMIP6) in reproducing the periods of Arctic warming and cooling known from the observations. Two experiments are considered for 9 CMIP5 models, and one experiment is used for the 108 ensemble members from 32 CMIP6 models. The ensemble mean of unforced Arctic amplification variability in the CMIP5 pre-industrial control experiment suggests a crucial role of external forcings in the present Arctic amplification. The ensemble mean of the same models in the historical experiment shows a century-long upward trend and clearly reproduces the present-day Arctic amplification. However, there is no consistent response of increasing Arctic amplification in the early twentieth century warming period during the 1920s–1940s in the CMIP5 models. This indicates the 1920s–1940s Arctic warming was unlikely to happen due to external forcings. Using the large CMIP6 ensemble in the historical experiment we show that the multi-model ensemble mean in the new generation of high-resolution CMIP6 models does not reproduce that warming either, thus posing questions how well the models capture internal climate variability and distinguish it from natural and anthropogenic forcings. This issue ultimately affects the reliability of future Arctic climate projections.

北极气候系统对外部扰动非常敏感，这导致与低纬度地区相比，北极的地表气温 (SAT) 变化更快。本研究旨在评估耦合模型比对项目（CMIP5 和 CMIP6）第 5 和第 6 阶段的全球气候模型在再现观测中已知的北极变暖和变冷时期的性能。9 个 CMIP5 模型考虑了两个实验，一个实验用于 32 个 CMIP6 模型的 108 个集合成员。CMIP5 工业化前控制实验中非受迫北极放大变异的集合平均值表明外部强迫在当前北极放大中的关键作用。历史实验中同一模型的集合平均值呈现出长达一个世纪的上升趋势，清楚地再现了当今的北极放大现象。然而，在 CMIP5 模型中，在 1920 年代至 1940 年代的 20 世纪初期变暖期，北极放大率增加没有一致的响应。这表明 1920 年代至 1940 年代北极变暖不太可能因外部强迫而发生。在历史实验中使用大型 CMIP6 集合，我们表明新一代高分辨率 CMIP6 模型中的多模式集合均值也不能再现这种变暖，因此提出了模型如何很好地捕捉内部气候变率并将其与自然和人为强迫。这个问题最终会影响未来北极气候预测的可靠性。