Revisiting the CMIP6 simulations of the most-concerned surface air temperature changes at the Earth's three poles

A unified framework for the systematic and simultaneous evaluation of CMIP6 models in reproducing warming responses across the Arctic, Tibetan Plateau, and Antarctic remains lacking. Most previous studies have focused on individual regions rather than cross-polar comparisons, restricting the quantification of model discrepancies and our comprehensive understanding of polar simulation capabilities. This study comprehensively assesses CMIP6 models across the three poles, with a focus on Arctic amplification (AA), elevation-dependent warming (EDW), and West Antarctica warming (WAW). We present an integrated analysis that highlights substantial regional disparities in climate model performance. Among the 22 CMIP6 models assessed, 77% of the models capture Arctic surface air temperature trends, AA intensity, spatial structure, and its temporal strengthening. However, only 23% of the models effectively reproduce the Tibetan Plateau's intensified warming and winter EDW pattern. Over Antarctic, 50% of the models exhibit moderate skill partially capturing the temporal evolution of WAW in spring, though most underestimate its peak intensity. Notably, while four models (ACCESS-ESM1-5, GFDL-ESM4, KIOST-ESM and INM-CM5-0) perform well in both the Arctic and Antarctic, only one (MPI-ESM1-2-LR) performs well in both the Antarctic and Tibetan Plateau. No model demonstrates high skill across both the Arctic and Tibetan Plateau—let alone all three regions. These findings underscore the urgent need for targeted climate model improvements through single-region optimization and enhanced representation of inter-regional teleconnections, to better represent the heterogeneous warming characteristics across Earth's three poles.