DOI: 10.3390/pr14193099 ISSN: 2227-9717

A Saturation-Constrained Four-Phase Acoustic Screening Model for Shallow Gas-Bearing Permafrost: Coupled Effects of Ice Cementation, Gas Saturation, and Pore Pressure

Long Cheng, Haoyu Pu, Chenxin Li, Zhiqiang Hu, Yi Zhu

Shallow free gas trapped within or beneath polar permafrost can be destabilized by drilling-induced thermal and pressure perturbations, but its acoustic response is difficult to separate from variations in lithology and freezing state. This study develops a saturation-constrained, Biot–Gassmann-informed semi-empirical screening model for a four-phase representative volume containing a mineral skeleton, pore ice, unfrozen water, and methane gas. The classical poroelastic equations establish the governing physical dependencies, whereas the reported P-wave velocity is calculated with a stress-sensitive corrected travel-time relation because permeability, tortuosity, ice morphology, and gas-patch dimensions are unavailable. A conditional freezing degree ensures that ice, water, and gas saturations remain non-negative and sum to unity. At 10 kHz, increasing gas saturation from 0 to 0.20 reduced velocity from 2893 to 1879 m/s and increased attenuation from 3.87 to 37.29 dB/m. At 5% gas saturation, increasing pore pressure from 0.1 to 5.5 MPa reduced velocity by 20.7% and attenuation by 89.3%. An 8000-realization analysis sampling 17 uncertain inputs identified gas saturation as the strongest negative correlate with velocity (rho = −0.659) and the dominant positive correlate with attenuation (rho = 0.851); freezing degree was the strongest positive velocity correlate (rho = 0.565). Ablation showed that phase substitution, the empirical ice-cementation correction, and their interaction contributed 908, 469, and 284 m/s, respectively, to the 1662 m/s thawed-to-frozen velocity increase. The attenuation formulation uses calibrated screening coefficients and a frequency-squared surrogate at 10 kHz; transfer to seismic or ultrasonic frequencies requires recalibration. The model is therefore intended for relative anomaly screening and predrilling risk ranking rather than stand-alone saturation inversion.