Impact of seasonal ground freezing on ambient noise HVSR inversion: improving deep velocity artifacts via borehole-based constrained inversion

This study investigated the impact of seasonal ground freezing on the generation of shear-wave velocity ( V S ) profiles through ambient noise Horizontal-to-Vertical Spectral Ratio (HVSR) inversion from a numerical analysis perspective. To this end, the CRB seismic station located in Wonju, Republic of Korea, was selected as the study site. Ambient noise data acquired during winter and summer as well as borehole logs were utilized for the analysis. Comparative analyses were conducted by designing scenarios for unconstrained inversion and constrained inversion. The research results confirmed that in the unconstrained inversion process, stiffness changes caused by surface freezing led to numerical artifacts. This resulted in numerical artifacts where the velocity of the deep bedrock was calculated to be higher in summer than in winter. In contrast, performing constrained inversion by fixing the bedrock boundary successfully eliminated these deep artifacts. The numerical interpretation revealed that the dramatic stiffening of the surface layer, representing a more than four-fold increase in V S , caused the expansion of the effective resonance depth. This mechanism rationally explains the low-frequency shift of the peak frequency observed during the freezing period. The frost penetration depth of approximately 1.0 m showed a very high correlation with regional empirical formulas and field measurement results. Finally, this study suggests that setting constraints based on borehole information is an essential prerequisite for ensuring the physical reliability of V S profiles in seasonally frozen ground.