Sensitivity evaluation of the Kudryavtsev permafrost model

Area of Science:

• Geosciences
• Climate Science
• Permafrost Science

Background:

• Permafrost dynamics are crucial for Arctic regions, yet data scarcity challenges accurate modeling.
• Kudryavtsev's active layer model, a physics-based analytical tool, is widely used due to minimal data requirements.
• The model's integration into a component modeling toolbox enables coupled permafrost and geomorphic process simulations over geological timescales.

Purpose of the Study:

• To systematically assess the influence of controlling parameters on permafrost temperature and active layer thickness predictions using Kudryavtsev's model.
• To investigate the model's sensitivity through Monte Carlo simulations and compare predictions with in-situ observations across Alaska.

Main Methods:

• Utilized Monte Carlo simulations to generate probability distributions for input parameters of Kudryavtsev's active layer model.
• Compared model predictions with a comprehensive benchmark dataset of in-situ permafrost observations throughout Alaska.
• Analyzed the sensitivity of permafrost temperature and active layer thickness to various input parameters.

Main Results:

• Permafrost surface temperature predictions strongly correlate with mean annual air temperature (r=0.78), annual temperature amplitude (-0.41), and winter snow thickness (0.30).
• Model uncertainty for permafrost temperature is low (RMSE=1°C) when air temperature and snow depth are well-constrained.
• Active layer thickness (ALT) predictions show RMSE of ~0.08m compared to observations, but soil water content bias can lead to significant ALT errors (RMSE=0.1m or 40% of observed ALT).

Conclusions:

• Kudryavtsev's active layer model provides relatively accurate permafrost temperature predictions when key parameters like air temperature and snow depth are well-defined.
• Soil water content is a critical parameter influencing active layer thickness predictions; significant bias in this parameter can undermine model accuracy despite improvements in other inputs.
• Accurate soil water content data is essential for reliable permafrost active layer modeling, particularly in data-poor Arctic regions.