Subsurface mechanical characterization is a crucial issue in many fields of both Earth Sciences and Geotechnical Engineering. Besides the use of direct investigations, common alternative methods comprise the surface waves methods (SWM) that have been widely used for near surface characterization. This includes the original Spectral Analysis of Surface Waves (Nazarian et al. 1984; Stokoe et al. 1994), multi-channel methods (Park et al. 1999) and passive techniques (Louie, 2001). Recently, soil damping ratios were estimated as well (Rix et al. 2001). SWM assumes Rayleigh waves propagating through a stack of horizontal soil layers. Such methods are well established and computationally efficient. However they only capture the vertical variation of elastic properties. Further, the flat-layered model is only an approximation and Rayleigh waves cannot describe the scattered wave-field when strong lateral variations occur. Observations and modeling of earthquakes confirm, for example, that geometry dramatically affects the seismic wave propagation. There have been several attempts to include lateral variations by combining successive 1D inversions via the so-called "pseudo-2D" (Luo et al. 2008). However this approach still retains the 1D approximation and can only capture weak variations. SWM has been thus limited by the lack of strategies accounting more realistic soil representations. Recently, for the crustal scale such limitations have been overcome by the full waveform inversion (FWI) which exploits a finite differences (FD) forward modeling (Virieux and Operto 2009). Unfortunately, when FD is used, the subsurface is finely discretized and the number of parameters associated largely exceeds the number of measurements available leading to severely ill-posed problems that can suffer from slow convergence and instability. Here, we propose an alternative inversion formulation based on the boundary element method (BEM) that overcomes the above mentioned limitations of both FWIs and SWMs (Bignardi et al, 2011).
Two-dimensional seismic wave modeling and inversion using the boundary element method
BIGNARDI, Samuel;SANTARATO, Giovanni
2011
Abstract
Subsurface mechanical characterization is a crucial issue in many fields of both Earth Sciences and Geotechnical Engineering. Besides the use of direct investigations, common alternative methods comprise the surface waves methods (SWM) that have been widely used for near surface characterization. This includes the original Spectral Analysis of Surface Waves (Nazarian et al. 1984; Stokoe et al. 1994), multi-channel methods (Park et al. 1999) and passive techniques (Louie, 2001). Recently, soil damping ratios were estimated as well (Rix et al. 2001). SWM assumes Rayleigh waves propagating through a stack of horizontal soil layers. Such methods are well established and computationally efficient. However they only capture the vertical variation of elastic properties. Further, the flat-layered model is only an approximation and Rayleigh waves cannot describe the scattered wave-field when strong lateral variations occur. Observations and modeling of earthquakes confirm, for example, that geometry dramatically affects the seismic wave propagation. There have been several attempts to include lateral variations by combining successive 1D inversions via the so-called "pseudo-2D" (Luo et al. 2008). However this approach still retains the 1D approximation and can only capture weak variations. SWM has been thus limited by the lack of strategies accounting more realistic soil representations. Recently, for the crustal scale such limitations have been overcome by the full waveform inversion (FWI) which exploits a finite differences (FD) forward modeling (Virieux and Operto 2009). Unfortunately, when FD is used, the subsurface is finely discretized and the number of parameters associated largely exceeds the number of measurements available leading to severely ill-posed problems that can suffer from slow convergence and instability. Here, we propose an alternative inversion formulation based on the boundary element method (BEM) that overcomes the above mentioned limitations of both FWIs and SWMs (Bignardi et al, 2011).I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
