Accurate non-destructive evaluation of pavement stiffness profile based on Monte Carlo inversion of surface-wave dispersion spectra

Abstract

Surface wave methods provide significant advantages for non-destructive pavement evaluation, particularly in estimating layer thickness and stiffness. Inversion based on fitting the full surface-wave dispersion spectrum leverages complete phase information, accounts for survey configuration, and eliminates the need for mode identification and dispersion curve picking. This study proposes a dispersion-spectrum inversion scheme for shear-wave velocity (V_S) profiling, integrating an efficient full-wavefield forward model of layered media with the Markov chain Monte Carlo (McMC) simulations. Both layer thickness and V_S are treated as inversion variables to enable accurate stratification. The non-linearity of the problem is addressed using a global search via McMC, which also quantifies uncertainty in the inversion results. Synthetic tests show that McMC dispersion-spectrum inversion reduces uncertainty in V_S estimates by 10–75 % compared to effective-mode inversion. Incorporating higher-frequency content further decreases inversion uncertainty for both shallow and deep layers. For instance, expanding the frequency range from 2 kHz to 5 kHz results in a reduction of estimation uncertainty by more than 60 % for layer thicknesses and over 30 % for V_S. A field experiment was conducted to validate the robustness of the proposed dispersion-spectrum inversion approach under practical conditions.