Numerical simulation of seismic waves in a model realistically accommodating arbitrary shaped topography is important in exploration geophysics, earthquake disaster mitigation, environmental/civil engineering applications, etc. Finite difference method (FDM) can be applied to fairly complex model with simple manner using rotated staggered grid (RSG) technique. However, RSG scheme requires many grid points per a minimum wavelength to assure the accuracy sufficient for the simulation of surface waves propagating on an arbitrary-shaped ground surface. On the other hand, a Hamiltonian particle method (HPM) has been applied to simulate surface wave propagation taking advantage of the simplicity in the accommodation of free-surfaces. Although the particle method successfully simulated surface wave propagation, the accuracy of the method for modeling Rayleigh waves has not been verified quantitatively.
In the present study, we investigate the accuracy of a Hamiltonian particle method, which is one of the particle methods, for the simulation of Rayleigh wave propagation. We calculate seismograms of the Lamb's problem using HPM and FDM with RSG (FDM-RSG), and compare them with those from the analytical approach. FDM-RSG solutions have the highest accuracy for amodel with a simple planar free surface aligned with the grid structure, however, suffer from the effect of inclined slope that may not be aligned with grids while those of HPM keep the accuracy enough to be applied in case of the inclined slope. We also investigate the accuracy of HPM for a range of the source frequencies and the offset distances. The results show that HPM can simulate surface wave propagation with smaller number of particles than FDM-RSG. Our numerical results indicate that HPM has some advantages over FDM-RSG in terms of the accuracy for modeling surface waves on arbitrary topography.
View full abstract