Comparison of ground temperature and permafrost conditions in the Arctic simulated by land surface process models of different complexity

Permafrost and ground freezing/thawing processes are physically and eco-climatologically important factors in the terrestrial cryosphere. The model reproducibility of frozen ground affects the certainty and reliability of simulated eco-climate conditions in cold regions as well as on a global scale. This study evaluated the variations and their attributes in the model performance developed and employed in the recent decade regarding the subsurface thermal state using outputs from Japanese and international model intercomparison projects and reanalysis data. The simulated surface and subsurface physical states were compared at four Arctic sites under different frozen ground conditions (Fairbanks, Kevo, Tiksi, and Yakutsk). The results showed that despite large variations in the modeled permafrost temperature, all the models, including the reanalysis data, successfully reproduced the permafrost conditions for the continuous permafrost sites. In contrast, some models failed to reproduce the presence of permafrost for the sites in the discontinuous to isolated permafrost zones. Evaluations of near-surface ground temperature variability revealed that the overall wellness of the simulated ground thermal states relied on winter reproducibility. The importance of snowpack metamorphosis for adequate thermal insulation was confirmed and demonstrated. The results at the coastal tundra site imply the importance of snow cover redistribution and wind crust formation owing to strong winds, the lack of which resulted in overestimations of thermal insulation and overcooled near-surface ground by most models.