New insights into the environmental factors controlling the ground thermal regime across the Northern Hemisphere:a comparison between permafrost and non-permafrost areas

Abstract The thermal state of permafrost affects Earth surface systems and human activity in the Arctic and has implications for global climate. Improved understanding of the local-scale variability in the global ground thermal regime is required to account for its sensitivity to changing climatic and geoecological conditions. Here, we statistically related observations of mean annual ground temperature (MAGT) and active-layer thickness (ALT) to high-resolution (∼1 km²) geospatial data of climatic and local environmental conditions across the Northern Hemisphere. The aim was to characterize the relative importance of key environmental factors and the magnitude and shape of their effects on MAGT and ALT. The multivariate models fitted well to both response variables with average R² values being ∼0.94 and 0.78. Corresponding predictive performances in terms of root-mean-square error were ∼1.31 °C and 87 cm. Freezing (FDD) and thawing (TDD) degree days were key factors for MAGT inside and outside the permafrost domain with average effect sizes of 6.7 and 13.6 °C, respectively. Soil properties had marginal effects on MAGT (effect size = 0.4–0.7 °C). For ALT, rainfall (effect size = 181 cm) and solar radiation (161 cm) were most influential. Analysis of variable importance further underlined the dominance of climate for MAGT and highlighted the role of solar radiation for ALT. Most response shapes for \(\mathrm{MAGT{ ≤ 0°C}}\) and ALT were non-linear and indicated thresholds for covariation. Most importantly, permafrost temperatures had a more complex relationship with air temperatures than non-frozen ground. Moreover, the observed warming effect of rainfall on \(\mathrm{MAGT_{ ≤ 0°C}}\) reverted after reaching an optimum at ∼250 mm, and that of snowfall started to level off at ∼300–400 mm. It is suggested that the factors of large global variation (i.e. climate) suppressed the effects of local-scale factors (i.e. soil properties and vegetation) owing to the extensive study area and limited representation of soil organic matter. Our new insights into the factors affecting the ground thermal regime at a 1 km scale should improve future hemispheric-scale studies