Near-surface site characterization is of great significance in the fields of geotechnical engineering and resource exploration. In this paper, we propose a near-surface site characterization method based on the joint iterative analysis of first-arrival and surface-wave data (JIAFS). The proposed method combines the advantages of first-arrival traveltime tomography (FATT) and multichannel analysis of surface waves (MASW). First, the 1D S-wave velocity ( $$v_{{\text{S}}}$$vS ) models obtained by MASW are interpolated to construct the pseudo-2D $$v_{{\text{S}}}$$vS model. According to the available geological survey information and borehole data, the initial Poisson’s ratio ( $$\sigma$$σ ) model is estimated. Based on the estimated $$\sigma$$σ model, the pseudo-2D $$v_{{\text{S}}}$$vS model is converted to a referenced P-wave velocity ( $$v_{{\text{P}}}$$vP ) model which is utilized to constrain the progress of FATT. This helps FATT overcome the inherent defect that it cannot effectively identify velocity-inversion interfaces and low-velocity zones. On the other hand, the $$v_{{\text{P}}}$$vP model obtained by FATT can provide a favorable priori information to improve the reliability of the results of MASW. Then, the $$v_{{\text{P}}}$$vP and $$v_{{\text{S}}}$$vS models obtained by constrained FATT and MASW are used to update the $$\sigma$$σ model. In addition, the $$v_{{\text{P}}}$$vP and $$v_{{\text{S}}}$$vS models are also used as initial models in the next iterative analysis. Finally, through the iteration of this process, the two inversion methods can make use of their own advantages to improve each other, so we can establish accurate near-surface $$v_{{\text{P}}}$$vP , $$v_{{\text{S}}}$$vS and $$\sigma$$σ models under complex geological conditions. A velocity model including low-velocity zone is established for synthetic model test to analyze and verify the advantage of JIAFS. The $$v_{{\text{P}}}$$vP , $$v_{{\text{S}}}$$vS and $$\sigma$$σ models obtained by JIAFS can accurately identify the low-velocity zone and match the true models well. In addition, the proposed method is applied to the field seismic data acquired for oil and gas exploration in Northwest China. Compared with the results of individual inversions and borehole data, JIAFS can establish more reliable 3D $$v_{{\text{P}}}$$vP , $$v_{{\text{S}}}$$vS and $$\sigma$$σ models by interpolating the 2D inversion results, which reveals further details and enhances the geological interpretation significantly.

Full waveform inversion is a methodology that determines high-resolution parameters. The widely used L₂-norm misfit function has local minima if the low wavenumber components are not accurate. Suffering from cycle skipping problem, the solution of waveform inversion will be trapped in the local minima. Dynamic time warping aims to find an optimal alignment between two signals, which is a more robust measure to avoid cycle-skipping challenges. However, the discontinuity makes the conventional dynamic time warping distance unsuitable for waveform inversion. We introduce a soft dynamic time warping distance as the misfit function, which is differentiable that inverted solution can converge to the accurate global minimum. We compare the convexity of the L₂-norm and soft dynamic time warping distance and show that the soft dynamic time warping distance has a wider convexity range with different time shift and amplitudes. It can alleviate the half-wavelength limitation of the conventional L₂-norm. We calculate the gradient using the automatic differentiation technique and the minibatch strategy and then analyze the alignment paths of different smooth parameters. A significant smooth parameter γ makes the Soft-DTW distance tending to the L₂-norm, which generates new local minima. We recommend a small smooth parameter to ensure the convexity of the Soft-DTW distance. Numerical examples show that the soft dynamic time warping can effectively reconstruct the deep velocity parameters of the BG Compass and Marmousi models with noise robustness and lower dependence on the initial model.

Compared with surface waves, guided waves are rarely applied in near-surface investigation. The main reason may lie in the complexity of their dispersion curves. Besides, the study and understanding of guided wave dispersion characteristics are now also inadequate and not deep enough. In this paper, we derived the complete theoretical dispersion curves of P–SV-wave and pure P-wave systems in layered media based on the transmission matrix method and obtained the relative displacement amplitude coefficients at the free surface as a function of frequency and phase velocity for both surface and guided waves. By assigning the value of relative displacement amplitude coefficient to the corresponding point (f,v) on dispersion curve, we got a multi-information diagram called relative amplitude dispersion map (RADM). As a unified description of surface and guided waves, RADM not only shows the velocity–frequency relationship but also represents the polarized energy ratio at the free surface by display colours. The accuracy of RADM was proved by synthetic seismic records, in which RADMs fit well with the corresponding dispersion energy of surface and guided waves. In addition, we designed six models with different Poisson's ratio (PR) and different layer numbers for comparison. It shows that the dispersive vertical-to-horizontal amplitude ratio of guided waves is complex and discontinuous in RADM, which brings great difficulty for mode identification and even affects the subsequent inversion. Tests also show that for high PR layers, the trends of guided P–SV-wave dispersion curves are basically consistent with those of pure P wave. With the decrease of PR, dispersion curves of guided P–SV wave gradually deviate from those of pure P wave. However, RADMs can be greatly consistent with the dispersion energy in either case. This is of great significance for the inversion of near-surface P and S velocities by using dispersion relationships of multimode surface and guided waves.