Xiaohang Ji, Ming Xiao, Eileen Martin, Tieyuan Zhu

The warming climate in high-latitude permafrost regions is leading to permafrost degradation. Estimating seismic velocity in permafrost can help predict the geomechanical properties of permafrost and provide information for planning and designing resilient civil infrastructure in cold regions. Seismic velocities of permafrost are mainly influenced by three components in permafrost: soil grains, water, and ice. Unfrozen water content reflects the variation of ice content and therefore is a key parameter in predicting seismic velocity. This paper statistically evaluates the performance of seven seismic velocity models in predicting seismic wave velocity of permafrost; these models are the time-average model, Zimmerman and King model, Minshull et al. model, weighted equation model, three-phase model, the Biot‐Gassmann theory modified by Lee (BGTL model), and Dou et al. model. The unfrozen water content used in these models is obtained from a modified Dall’Amico’s model that we propose and this new model is evaluated against six existing unfrozen water content models based on soil temperature. The data used in the evaluation are from published laboratory and in-situ data, including 369 data points for joint P- and S-wave velocities from 9 publications and 980 unfrozen water content data points from 13 publications. This study finds that permafrost of all soil types generally shares the same linear trends between P- and S-wave velocities, regardless of porosity, grain size, and temperature. Fitting all existing data, we derive an empirical linear relationship between P- and S-wave velocities. Among the seismic velocity models evaluated in this study, the Minshull et al. model and BGTL model are the most accurate in predicting seismic velocity of permafrost. The study also provides the applications of seismic velocity models for various permafrost soil types.