Surface-Wave Dispersion Curve Extraction Using L q 1 – L q 2 Regularization-Based High-Resolution Linear Radon Transform

Obtaining surface wave dispersion curves is a critical step for the reconstruction of the near-surface shear-wave velocity models. Due to the inherently weak energy of higher-mode signals, they are easily masked by random noise under low signal-to-noise ratio (SNR) conditions. Furthermore, the dominant energy of the fundamental mode severely interferes with the accurate picking of these weak higher modes, thereby making the extraction of surface wave dispersion curves significantly challenging. To address the limitations of conventional methods under low SNR conditions, specifically the reduced extraction accuracy and the loss of effective dispersion point, this study proposes a multi-mode dispersion curves extraction method based on the L_q₁ - L_q₂ multi-regularized High-Resolution Linear Radon Transform. First, as a required preprocessing step, an adaptive signal separation mechanism based on the Non-subsampled Shearlet Transform (NSST) is applied to the original seismic records to coarsely separate the fundamental-mode and higher-mode signals. Subsequently, a sparsity-constrained, mode-dependent inversion model is then constructed in the Radon domain and solved using the Alternating Direction Method of Multipliers (ADMM), enabling high-quality multi-mode dispersion extraction. Tests on both simulated and field data demonstrate that, even under low SNR conditions, the proposed method robustly recovers the fundamental-mode dispersion curves while simultaneously extracting distinct higher-mode information. This study presents a novel framework for multi-mode surface wave dispersion extraction, which utilizes higher modes to strengthen the constraints on inversion, thereby further improving the reliability of the inverted shear-wave velocity structure.