|
Construction of the 3-D velocity models for the strong ground motion simulation in a complex tectonic environment, e.g. subduction zone, is a difficult problem. Popular ?gstand along?h methods, e.g. seismic tomography or the receiver function inversion, suffer from the inversion instabilities and require an intensive constraining of parameters. In this work, to construct the 3-D velocity model in a wide area, that includes both the inland areas, covered by seismometers, and the offshore areas, where seismic observations are scarce, we simply generalize different kinds of the available information in the studied region.
The 3-D crustal velocity model for the area of the Philippine Sea subduction zone in Japan, was constructed from 9 constant velocity layers: (1) oceanic sedimentary layer, OSL, (2) surface low-velocity layer, LV, (3) upper crust, UC, (4) lower crust, LC, (5) mantle wedge, MW, (6) oceanic upper crust layer, OUC, (7) oceanic lower crust layer, OLC, (8) Philippine sea slab, SLB, (9) upper mantle, UM. Next data were used: the off-shore seismic profiles (OSL, LV, OUC, OLC), the seismicity (UC and SLB), the receiver function inversion results (UC, LC, SLB), the 1-D velocity models for the hypocenter determination (LV, UC, LC and MW), the seismic tomography results (LC, SLB), and the 1-D waveform inversion (LV).
In order to validate the developed velocity model, we calculated the P-wave travel times and compared them with the observed times. Average standard deviation is small, around 0.35 sec; moreover residuals don?ft show any trend with distance. Additionally, using the finite-difference method, we synthesized low frequency, i.e. 0.1-0.5Hz, waveforms for several moderate earthquakes and compared them with the observed waveforms at the strong motion sites. Except for a few limited regions agreement is good and in some cases match is perfect. Therefore we can conclude that in combination with the basin model around a target site, the crustal model is acceptable for the simulation of the strong ground motions. Also it can be used as the initial model for farther tuning, using seismic tomography or waveform inversion for example.
|
