Abstract: Karst is a typical geological hazard in urban environments, and its development can affect the stability and safety of the infrastructures. Compared with the surrounding bedrock, the S-wave velocity of karst is relatively lower, and appear as low-velocity anomalies. Surface waves within ambient noise are highly sensitive to the S-wave velocity of underground media, with the higher-mode surface waves demonstrating even greater sensitivity to such low-velocity anomalies. However, the conventional staged inversion schemes for multimode dispersion curves rely on global optimization algorithms, resulting in substantial computational costs. To address this limitation, we propose an improved inversion workflow. In the first stage, the preconditioned steepest descent method, a local optimization algorithm, is employed to invert the identified fundamental-mode dispersion curve. In the second stage, the pattern search algorithm is utilized to invert all available dispersion data. During the iteration, the Kuhn-Munkres algorithm is implemented to dynamically assign the mode-orders to unlabeled dispersion points. The efficacy of the proposed method in characterizing shallow low-velocity interlayers is validated via synthetic tests based on a four-layer geological model containing a soft interlayer. Furthermore, the method is applied to the field passive-source data acquired in the karst development area of Longgang District, Shenzhen, to evaluate its practical utility. The results demonstrate that the higher-mode dispersion data serve as observational constraints independent of the fundamental-mode data, effectively enhancing the accuracy and vertical resolution of the inversion, and improving the delineation of soft interlayers. Additionally, the proposed approach successfully identified the underlying karst development area along the survey line. Notably, energy loss and modal aliasing of higher-mode surface waves dispersion are common phenomena in shallow surface wave data. Therefore, the dynamic mode-order identification is essential for robust multimode surface wave inversion.

Key words: passive surface-wave, karst development area, staged inversion, multi-order inversion, S-wave velocity