The evolution of Arctic permafrost over the last 3 centuries from ensemble simulations with the CryoGridLite permafrost model

Abstract

Abstract. Understanding the future evolution of permafrost requires a better understanding of its climatological past. This requires permafrost models to efficiently simulate the thermal dynamics of permafrost over the past centuries to millennia, taking into account highly uncertain soil and snow properties. In this study, we present a computationally efficient numerical permafrost model which satisfactorily reproduces the current ground temperatures and active layer thicknesses of permafrost in the Arctic and their trends over recent centuries. The performed simulations provide insights into the evolution of permafrost since the 18th century and show that permafrost on the North American continent is subject to early degradation, while permafrost on the Eurasian continent is relatively stable over the investigated 300-year period. Permafrost warming since industrialization has occurred primarily in three “hotspot” regions in northeastern Canada, northern Alaska, and, to a lesser extent, western Siberia. We find that the extent of areas with a high probability (p3 m>0.9) of near-surface permafrost (i.e., 3 m of permafrost within the upper 10 m of the subsurface) has declined substantially since the early 19th century, with loss accelerating during the last 50 years. Our simulations further indicate that short-term climate cooling due to large volcanic eruptions in the Northern Hemisphere in some cases favors permafrost aggradation within the uppermost 10 m of the ground, but the effect only lasts for a relatively short period of a few decades. Despite some limitations, e.g., with respect to the representation of vegetation, the presented model shows great potential for further investigation of the climatological past of permafrost, especially in conjunction with paleoclimate modeling.

Robust reconstruction of historical climate change from permafrost boreholes

“…Throughout this study, we assume saturated conditions, so θ tot is taken to be equal to the natural porosity of the soil volume. For further details on the parameterizations of the thermal properties, see Groenke et al (2023); Langer et al (2024).…”

“…The spatial discretization has a grid cell spacing of 10 cm in the upper 2.5 m of the ground, where the freeze-thaw dynamics play the largest role, which is then increased non-uniformly to 10 m in the bottom 900 m of the domain. We augment the numerical scheme of Langer et al (2024) to allow for more realistic forms of θ w (T )…”

“…The historical air temperature forcing data used in the synthetic experiments can be downloaded from Zenodo (Langer et al, 2022a). The model outputs from Nitzbon et al (2023) and Langer et al (2024) used in Figure 1 are also available on Zenodo (Langer et al, 2022b). Air temperature data from Tiksi can be obtained directly from the NOAA Climate Data Center (NOAA National Centers of Environmental Information, 1999).…”

“…Recent studies have established from ground temperature records that permafrost is warming globally over the last two decades (Biskaborn et al, 2019). Long-term historical simulations of land surface heat exchange (Langer et al, 2024) as well as ice core reconstructions (Opel et al, 2013) suggest that this warming trend likely dates back to the conclusion of the Little Ice Age in the early-to mid-19th century. However, paleoclimate reconstructions from proxy data carry with them significant uncertainties (Hernández et al, 2020) and thus may not always be representative for terrestrial permafrost regions.…”

“…In this work, we propose a new method for inverting GST from permafrost boreholes where latent heat effects play a dominant role due to both seasonal and long-term thawing of ground ice. Our method builds upon the thermal modeling scheme of Cryo-GridLite (Langer et al, 2024), which allows for efficient, high-fidelity simulation of twophase heat conduction over large time scales. Similar to Hopcroft et al (2007), we employ a Bayesian formulation of the inverse problem that provides a probabilistic interpretation of uncertainty in the reconstructed GST histories and permits the inclusion of prior information where available.…”