Surface-wave Analysis In Complicated Near-surface Elastic Media

Surface-wave analysis is nowadays widely adopted for building shear(S)wave velocity profiles at a multiple scales—global seismology,exploration geophysics,nearsurface geophysics.All of these applications share the same principles: they use the dispersive characteristic of surface waves to infer the properties of the medium by identifying the model parameters.In near-surface applications,the multichannel analysis of surface wave(MASW)and multichannel analysis of Love wave(MALW)methods have been given increasingly attention and widely used to determine near-surface S-wave velocities during the past two decades.The multichannel analysis of surface wave(MASW)method is a non-invasive geophysical technique that uses Rayleigh-wave dispersion to estimate S wave velocities.This technique consists of acquisition of broad-band high-frequency Rayleigh waves using a multichannel recording system,extraction of dispersion curves from Rayleigh waves,and inversion of dispersion curves to obtain near-surface S-wave velocity(Vs)profiles.A pseudo-2D S-wave velocity section is constructed by aligning 1D models at the midpoint of each receiver spread and using a spatial interpolation scheme.It is nondestructive,non-invasive,low cost,and relatively highly accurate.Based on these benefits,the MASW method has become one of the main seismic methods in determining near-surface S-wave velocities for applications of geotechnical and environmental engineering.The multichannel analysis of Love wave(MALW)method is also receiving much more attention because of some advantages of Love waves.S-wave velocities determined by MASW and MALW have been reliably and consistently correlated with borehole data.The MASW method has been applied to a wide range of problems in geotechnical and environmental engineering,such as mapping bedrock,delineating the collapse feature,delineating the shallow fault zone,detecting voids,assessing landslide stability and so on.The traditional surface-wave analysis method is a 1D approach because the inverted S-wave velocity profiles from surface waves are based on the assumption of horizontally layered earth model,which usually neglects the presence of lateral variations in 2D environments.The resulting model is a simplified description of the site because the surface-wave path crosses different materials.Artifacts may be introduced in spatially 2D S-wave imaging when not accounting for the effects of lateral heterogeneity.In real world of MASW applications,however,the 1D approach is still adopted to investigate lateral variations for processing and inversion,and 1D velocity profiles are eventually merged to reconstruct 2D velocity structures to display lateral variations.In other words,data are processed and inverted,disregarding the effect of lateral variations,but the lateral variations are then retrieved and considered in the final interpretation.In this context,it is very important to assess the errors that could be introduced because of the presence of unknown lateral variations.One of key steps of using surface-wave methods to obtain S-wave velocities is to pick correct phase velocities in dispersive images,especially when higher modes are included in the inversion.It is essential to identify different modes in experimental data and it is necessary to compare the experimental-curve branches with specific theoretical modes.Hence,in most approaches to surface waves,the phase velocities need to be attributed to a specific propagation mode.This task is not straightforward because some modes may not be present in the experimental data and very smooth changes from one mode to another may occur.The misidentification of modes may produce significant errors.In real world applications,the complexity of energy distribution on a dispersion image is exacerbated due to the complicated near-surface earth models.Previous studies about a low velocity layer(LVL)among a layered earth model found that dispersive energy of such model “jumps” from the fundamental mode to higher modes and may not return to the fundamental mode at higher frequencies,which brings a pitfall of modemisidentification and produces incorrect inversion results.Although several papers have reported issues in picking and then inverting surface waves in the presence of lowvelocity layers,it still remains unclear why the energy of each mode “jumps” or disappears at higher frequencies for an LVL model.Previous studies on the surface waves of an LVL model were based on computing the theoretical dispersion curves and their corresponding eigenfunctions with efficient algorithms.Few researchers focused on the dispersion energy characteristics based on surface-wave wavefield modeling.Commonly Rayleigh waves are obtained only from the vertical component of P-SV waves.However,the multimodal Rayleigh waves estimated from the horizontal(radial)component data have a different energy distribution compared with those estimated from the vertical component data.The combined use of the horizontal component data with the vertical component data would contribute to preventing mode misidentification and improving S-wave velocity estimations.In order to develop the surface-wave analysis technique in complicated near-surface media,I completed the following studies.(1)I present a case study using both of Rayleigh and Love waves to estimate 3D near-surface S-wave-velocity structures.The S-wave velocities(Vs)at the Boise Hydrogeophysical Research Site(BHRS)were estimated by multichannel analysis of Rayleigh(MASW)and Love waves(MALW).I reconstructed the 3D velocity structures by merging all the 2D Vs sections at different survey lines.Vs structures obtained by both of MASW and MALW were compared with the results of borehole measurements,which showed great consistence.I used the 3D Vs distribution to identify a relatively lowvelocity anomaly(i.e.,the sand channel)and its boundaries,which had been identified by the borehole and ground-penetrating radar(GPR)measurements.The lateral variability in position and shape of the boundaries agreed well with the borehole and GPR results.Vs structures and anomaly boundaries were delineated at the meter scale by the MASW and MALW methods.The Vs differences between MASW,MALW and borehole measurements were discussed and showed the near-surface anisotropy information at the BHRS.(2)I analyzed the horizontal resolution of the MASW method.According to different influencing factors of the horizontal resolution,I established different laterally heterogeneous models and observation systems and then simulated a multiple number of synthetic multichannel records with a finite-difference method along a linear survey line using the roll-along acquisition mode.After the extraction of dispersion curves of Rayleigh waves and inversion for S-wave velocity profiles for each synthetic shot gather,a pseudo-2D S-wave velocity section can be generated by aligning 1D S-wave velocity models.Ultimately,I evaluated the horizontal resolution capability of the MASW method on pseudo-2D Vs maps.My numerical investigation results and field data analysis indicate that Vs values on the maps are not the same as the true Vs values for structures whose lateral dimension is shorter than a receiver spread length and that anomalous bodies which are larger and have high velocity contrast are easier to be distinguished on the Vs maps with a shorter receiver spread length.The horizontal resolution decreases with the increasing depth and is approximately one half of the shortest Rayleigh wavelength that can penetrate to the depth.(3)I analyzed the dispersion energy characteristics of Rayleigh and Love waves in the presence of low-velocity layers.I introduce the guided waves generated in an LVL(LVL-guided waves,a trapped wave mode)to clarify the complexity of the dispersion energy.I confirm the LVL-guided waves by analyzing the snapshots of SH and P-SV wavefield and comparing the dispersive energy with theoretical values of phase velocities.Results demonstrate that LVL-guided waves possess energy on dispersive images,which can interfere with the normal dispersion energy of Rayleigh or Love waves.Each mode of LVL-guided waves being lack of energy at the free surface in some high-frequency range causes the discontinuity of dispersive energy on dispersive images,which is because shorter wavelengths(generally with lower phase velocities and higher frequencies)of LVL-guided waves cannot penetrate to the free surface.If the S-wave velocity of the LVL is higher than that of the surface layer,the energy of LVL-guided waves only contaminates higher mode energy of surface waves and there is no interlacement with the fundamental mode of surface waves.While if the S-wave velocity of the LVL is lower than that of the surface layer,the energy of LVL-guided waves may interlace with the fundamental mode of surface waves.Both of the interlacements with the fundamental mode or higher mode energy may cause misidentification for the dispersion curves of surface waves.(4)I conducted multimodal inversion of Rayleigh and Love waves for investigating LVLs with the determinant misfit function,which can avoid the necessity of numbering different modes.The Monte Carlo algorithm was applied for the inversion problem.Results of two synthetic examples demonstrated that an LVL can be efficiently and accurately investigated by inversion of Rayleigh waves with appropriate initial models and the inversion of Love waves is more independent of initial models on account of the precious property of the determinant misfit.Two field examples of Rayleigh-wave data,which have been verified by comparison with other in-situ measurements,reveal the great potential of applications in geotechnical characterization for investigating LVLs.(5)I implemented a joint inversion method to estimate both of the near-surface Swave velocity and radial anisotropy from Rayleigh and Love wave dispersion data.The proposed joint inversion method can not only combine the strengths of the multichannel analysis of Rayleigh and Love wave methods,but also allow the existence of radial anisotropy(the vertically polarized shear(SV)wave velocities(Vsv)differ from horizontally polarized shear(SH)wave velocities(Vsh)).A parameter carrying the radial anisotropy information is introduced,which provides connection and constraints between Vsv and Vsh.The inversion algorithm was tested on synthetic data and then applied to a field case.Results were compared with those of individual inversions and joint inversion without considering radial anisotropy,and also with velocity profiles retrieved by downhole seismic surveys.The synthetic and field examples demonstrate the proposed joint inversion method can provide better constrained and more reliable S-wave velocity and radial anisotropy information in the shallow subsurface.(6)I expanded the single-station passive horizontal-to-vertical spectral ratio(HVSR)technique to active data.I proposed to calculate the HVSR of Rayleigh waves from active seismic records.I found that the existence of other wave types(e.g.,body waves)and interference of different Rayleigh-wave modes cause errors in the calculation of active HVSR.With synthetic data,I verified that it is necessary to separate different modes of Rayleigh waves from active seismic data and then calculate the HVSR with a single mode(basically the fundamental mode).The mode separation is implemented in the frequency-phase velocity(f-v)domain by the high-resolution linear Radon transformation(LRT),which requires multiple channel records to generate specific dispersion energy of Rayleigh waves.I tested the proposed method in the presence of high velocity contrasts and lateral variations.Synthetic data and a field data example demonstrate the validity of calculating active Rayleigh-wave HVSR by mode separation.The HVSR calculated from a separated Rayleigh-wave mode contains the information of lateral variations.HVSR peak and trough frequencies are very sensitive to parameters of the near-surface model,which possess great potential to investigate and reconstruct shallow subsurface 2D structures.