A Multi-Scale Poisson-Voronoi Inversion Framework with Joint Multi-Mode Surface Wave Dispersion for DAS-Based Near-Surface Imaging

Urbanization-driven demand for high-resolution near-surface imaging and monitoring has promoted the application of distributed acoustic sensing (DAS) technology. However, DAS-based surface-wave velocity imaging faces challenges from low signal-to-noise ratios, strong lateral heterogeneity, and scale-dependent surface-wave sensitivity. To address these issues, we propose a multi-scale surface-wave inversion framework tailored for DAS observations, which inverts multi-scale and multi-mode sub-array dispersion curves simultaneously for the two-dimensional (2D) shear-wave velocity model. The method integrates three core technical components: multi-scale overlapping sub-array selection, frequency-Bessel (F-J) transform-based dispersion extraction (enabling reliable capture of both fundamental and higher-mode surface-wave energy), and a Poisson-Voronoi (PV) tessellation inversion strategy for dimensionality reduction. By integrating dispersion information across multiple frequency bands and multiple modes via multi-scale sub-arrays, the framework achieves complementary sensitivity to shallow and deeper subsurface structures. The PV tessellation stabilizes the inversion and avoids artificial lateral velocity variations inherent in conventional 1D inversion approaches. Synthetic tests confirm the method’s ability to reliably recover low- and high-velocity anomalies with improved lateral continuity and depth resolution. Application to urban DAS ambient noise data from Hefei, China, yields a geologically plausible 2D shear-wave velocity model. This study provides a robust methodological foundation for high-resolution near-surface imaging in complex urban environments using DAS technology.