Scholte-wave Adjoint Tomography for Building Low-frequency, Offshore Shear-wave Velocity Models

Field observations have shown that low-frequency (sub-1 Hz) Scholte waves retrieved from ocean-bottom node (OBN) data are strongly influenced by large-scale velocity heterogeneities such as salt bodies, underscoring their potential for offshore model building. However, procedures for translating this sensitivity into reliable subsur- face velocity models remains poorly understood. Motivated by these observations, we investigate the feasibility of using low-frequency Scholte-wave travel-time adjoint tomography to construct long-wavelength offshore shear-wave velocity (VS ) models suitable as starting models for elastic full-waveform inversion (E-FWI). Using a 2-D synthetic model that includes an ocean layer and a realistic, laterally variable salt body, we simulate empirical Green’s functions recorded by dense OBN arrays and compute travel-time misfit kernels for both background and true VP –ρ parameteriza- tions, starting from a background VS model with large (∼50%) perturbations relative to the true model. Fr´echet-kernel analyses confirm that Scholte waves are dominantly sensitive to VS variations, with only weak indirect dependence on VP and ρ. The to- mographic inversion employs a multimodal strategy that combines fundamental and higher-order Scholte modes within a hierarchical frequency-stepping scheme. Even when all parameters are initialized from the background, the inversion recovers the overall salt geometry and VS structure reasonably well, demonstrating that large-scale VS recovery is achievable without the true VP –ρ contrasts. Incorporating the correct VP and ρ distributions further improves depth focusing and reduces parameter trade- offs, yielding enhanced vertical resolution. The recovered VS models reproduce the long-wavelength salt geometry, with intermediate-depth velocities estimated within ∼10% of the true values, although structures deeper than ∼6 km and shallower than ∼2 km remain underconstrained within the tested 0.10–0.45 Hz band. These results demonstrate that low-frequency Scholte-wave adjoint tomography provides a robust, practical pathway for constructing reliable long-wavelength VS models and a physi- cally consistent foundation for initializing E-FWI in complex offshore environments.