Quantification and practical solution for bottom boundary effects on long-term permafrost models

Permafrost models are commonly used to simulate future ground temperatures under the influence of climate change and/or proposed infrastructure. Most, if not all model input decisions are made based on limited subsurface knowledge. Shallow model domains offer more efficient run-times especially with gridded one-dimensional schemes as well as two-dimensional and three-dimensional simulations over longer time periods. Recent focus has been on the development of surface boundaries; however, less attention is given to the bottom boundary condition. In this paper, first we quantify the effect of model domain depth and bottom boundary condition type on long-term ground temperature evolution in transient model simulations. Model domains less than 100 m deep with the bottom boundary condition set to both geothermal gradient and a perfectly insulated base ( q = 0) show significant overwarming and overestimate thawing in cold permafrost. Guidance is provided to interrogate model results to avoid bottom boundary condition effects. For cases where further models are needed, a practical solution to this challenge is developed to obtain model depth-independent results. This solution allows for reduced computational requirements while maintaining consistent results for century-scale transient simulations considering climate change effects on thawing permafrost.