A Joint Scheme For Imaging The Urban Basin Structure With High-resolution Based On The Seismic Ambient Noise At A Dense Array

Large cities are often located above sedimentary basins,which tend to capture and amplify seismic energy,resulting in greater damage in cities when earthquakes happen.The significance of constructing a sedimentary basin model for the metropolitan area is twofold.On the one hand,understanding the amplification effect of the basin requires obtaining the velocity structure and site amplification effect of the basin,which helps the seismic hazard assessment of the metropolitan area and the preparation of the new generation of seismic zoning maps.On the other hand,the unprecedented economic boom has driven the rapid development of large cities.Planning and building one or more urban sub-centers around the central city to relieve the pressure of the growing population in the central city is the main problem faced by many large cities at present.The high-resolution sedimentary basin velocity structure can provide basic information for urban planning and underground space utilization.The combination of dense array observation and seismic interferometric imaging makes array-based surface wave imaging possible,especially in today’s era of rapid development of artificial intelligence and big-data.Seismic interferometric imaging based on dense array provides seismologists with unprecedented imaging resolution for studying the Earth’s interior,the fine structure of fault zones and sedimentary basin model of metropolitan area,and is particularly suitable for work in study area with limited active sources in complex environments such as fault zone and urban area.In this paper,a beamforming technique for extracting multi-mode surface wave from ambient noise is proposed,and a joint imaging method combining the array-based high-mode surface wave inversion and the single station microtremor HV spectral ratio method,to construct a high-resolution 3D sedimentary basin model of the metropolitan area.The method is applied to data consisting of a dense array in the Tongzhou area,a sub-center of Beijing,with an average station spacing of 1 km and 914 stations,to construct a high-resolution sedimentary basin model of the Tongzhou area.The method starts with a moving window beamforming,which divides the dense array into multiple sub-arrays and applies the beamforming method to each sub-array to extract the multimode surface wave dispersion curves,thus obtaining the 2D phase velocity distribution below the entire array without pure path inversion.The multi-mode dispersion curves of each sub-array are in depth inversion to obtain the 3D S-wave velocity structure.Based on the same data,the peak frequency and peak amplitude of the HV spectral ratio curves are extracted using the microtremor HV spectral ratio method.Based on the inversion 3D S-wave velocity structure,the peak frequencies are converted into depth to outline the strong impedance interface.Thereby,the 3D S-wave velocity structure and the depth of strong impedance interface are combined with the geological geophysical data and drilling data of the study area to construct a high-resolution sedimentary basin model containing the thickness of sedimentary layers in Tongzhou area,and the corresponding tectonic implications are discussed.The main innovative results of the thesis include:1)A beamforming method for extracting multi-mode surface wave dispersion from ambient noise is proposed,i.e.,Weighted Cross-correlation Beamforming(WCBF)and Modified Cross-correlation Beamforming(MCBF),which can extract azimuthaveraged multi-mode surface wave dispersion images based on shorter continuous records with higher resolution,where MCBF can overcome the effect of station array geometry and extract azimuth-dependent phase velocities.Compared with the conventional Cross-correlation Beamforming(CBF),WCBF corrects the geometric diffusion factor of the column surface wave propagation and estimates the phase velocity of the structure below the array by calculating the phase coherence between the incident plane wave and the observed wave field in different azimuths.For isotropic medium,the observed wavefield through the array is theoretically equivalent to a columnar wave represented by a zero-order Bessel function J0(kr).The azimuthaveraged WCBF,is equivalent to a 2D Fourier transform of this radially symmetric spatial wavefield.For azimuthally anisotropic medium,only plane waves incident in the steady phase region are considered in the MCBF to overcome the effect of the irregular geometry of the array and to extract the azimuth-dependent phase velocities.If the azimuthal anisotropy of the medium is not considered,MCBF is able to extract azimuthal-averaged multi-mode phase velocity dispersion curves with a fairly high resolution.The effectiveness of the above method is verified based on synthetic data,regional and local scale dense array observations.(2)A scheme to overcome the phenomenon of aliasing in multi-mode dispersive image in the frequency-phase velocity domain is proposed.By using different forms of the cross-spectral density matrix(CSDM),the effect caused by negative wave number in array processing is overcome.It is also confirmed that the randomly distributed array helps to suppress the aliasing and period extension phenomena arising from finite sampling.Due to the finite sampling of the array,the array-based signal processing techniques inevitably produce aliasing,so that the final dispersion image contains the true wave number,the period extension of the true wave number,the negative wave number and the period extension of the negative wave number.In a uniform array,the period extension is determined by the station spacing.In the search along the azimuth,two directions are actually searched for waves propagating in both directions,i.e.,outward-propagating and inward-propagating columnar surface waves,corresponding to the true and negative wave number,respectively.The causal matrix allows the construction of a complete wavefield propagating outward,so that the aliasing generated by negative wave numbers can be removed from the results,but its period extension cannot be avoided.Based on synthetic data and actual observations,it is confirmed that using randomly distributed stations can suppress the effects of low resolution due to period extension.(3)A new technique to convert the microtremor HV spectral ratio peak frequency into the impedance interface depth based on a 3D S wave velocity model is developed.Although the theory of the microtremor HV spectral ratio is not clear,the peak frequency of the microtremor HV spectral ratio has a good correspondence with the S wave resonance frequency of the medium below the station.Therefore,the peak frequency of the microtremor HV spectral ratio curve can be used to quickly estimate the depth of the impedance interface.The traditional method of converting resonant frequencies to interface depths is to use an empirical relationship fitted to drilling data,which is not the same in different regions or on different geological units.Based on the 3D S wave velocity model,the relationship between resonant frequency and impedance interface depth can be established,thus eliminating the need for traditional empirical relationships to estimate sediment layer thickness and avoiding the calibration of empirical relationship coefficients based on drilling data.(4)Combining the newly developed beamforming method and microtremor HV spectral ratio method,a high-resolution sedimentary basin model of Tongzhou area in Beijing’s urban sub-center is constructed based on the data at a dense array in Tongzhou,and the corresponding tectonic implications are discussed,giving some evidence that the Daxing Fault continues to extend northeast after crossing the Niubaotun Fault.Based on the ambient noise data of the Beijing Tongzhou dense array,the Rayleigh wave dispersion curves of the fundamental and first-order mode of each sub-array are obtained by applying the moving window beamforming method.The 3D highresolution S wave velocity structure is obtained by inversion of each sub-array.The microtremor HV spectral ratio method is applied to extract to the two peaks of the HV spectral ratio curve and the corresponding peak frequencies.Using the high-resolution 3D S-wave velocity structure obtained by surface wave dispersion inversion,the peak frequencies are converted to depth to obtain the thickness of the overlying soft soil layer and the Quaternary sediment layer.Combining the sedimentary layer thickness distribution given by the S wave velocity equivalent surface and the microtremor HV spectral ratio,some evidences of northeast extension of the Daxing Fault are given.