The intensification of extreme heat events is a potent thermal disturbance that can trigger abrupt permafrost degradation. However, a systematic understanding of their spatiotemporal variation across the permafrost region of the Northern Hemisphere (PRONH) is lacking, hindering predictions of regional-scale responses and climate feedback tipping points. In this study, six indices are systematically employed to analyse the historical spatiotemporal variations (1991–2020) of extreme heat events in the PRONH and to project their future changes (2021–2100) under Shared Socioeconomic Pathway (SSP) scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5), particularly across four permafrost types. The results indicate that the increasing trends of warm day (TA95p; 2.95 ± 0.56 d per decade), warmer day (TX95p; 3.20 ± 0.59 d per decade), warmest day (TXx; 0.35 ± 0.12 ℃ per decade), heatwave intensity (2.67 ± 0.96 ℃ per decade), heatwave frequency (0.23 ± 0.05 events per decade) and heatwave duration (0.54 ± 0.44 d per decade) were significant (p < 0.05) during 1991–2020. Under SSP5-8.5, the Arctic and Tibetan Plateau are projected to experience 150–200 d TX95p annually, making such events routine by the late 21st century (2076–2100). Under the same scenario, continuous permafrost regions are projected to face the most severe exposure, with TX95p reaching 148 ± 24 d annually, whereas the discontinuous (134 ± 23 d), sporadic (130 ± 22 d) and isolated (109 ± 19 d) permafrost regions are expected to experience fewer extreme heat events. However, their fragmented distributions render them particularly vulnerable and heighten the risk of degradation. This study underscores the urgent need to integrate these extreme heat events into permafrost vulnerability assessments and climate adaptation strategies.

Temperature inversions are frequently observed in the boundary layer and lower troposphere of polar regions. Future variations of the low-level temperature inversions in these regions, especially the Antarctic, are still poorly understood. Due to the scarcity of observations in the Antarctic, reanalysis data and numerical simulations are often used in the study of Antarctic climate change. Based on ERA-Interim, ERA5, JRA-55, and NCEP–NCAR reanalysis products, this study examines temporal and spatial variations of Antarctic inversion depth in austral autumn and winter during 1979–2020. Deeper inversions are found to occur over the high plateau areas of the Antarctic continent. Based on the Mann–Kendall test, ERA-Interim and ERA5 data reveal that the Antarctic inversion depth in austral autumn and winter increased during 1992–2007, roughly maintained afterwards, and then significantly decreased since around 2016. The decrease trend is more obvious in the last two months of winter. Overall, JRA-55 better represents the spatial distribution of inversion depth, and ERA-Interim has better interannual variability. The Community Earth System Model Large Ensemble (CESM-LE) 30-member simulations in 1979–2005 were first verified against JRA-55, showing reasonable consistency, and were then used to project the future changes of Antarctic low-level inversion depth over 2031–2050 under RCP8.5. The CESM-LE projection results reveal that the temperature inversion will shallow in the Antarctic at the end of the 21st century, and the decrease in depth in autumn will be more pronounced than that in winter. In particular, the temperature inversion will weaken over the ice-free ocean, while it will remain stable over the ice sheet, showing certain spatial heterogeneity and seasonal dependence on the underlying cryospheric surface conditions. In addition, the decrease of inversion depth is found closely linked with the reduction in sea ice, suggesting the strong effect of global warming on the thermal structure change of the Antarctic.