Attenuation and excitation of three-component ground motion in southern California

Citation
M. Raoof et al., Attenuation and excitation of three-component ground motion in southern California, B SEIS S AM, 89(4), 1999, pp. 888-902
Citations number
23
Categorie Soggetti
Earth Sciences
Journal title
BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
ISSN journal
00371106 → ACNP
Volume
89
Issue
4
Year of publication
1999
Pages
888 - 902
Database
ISI
SICI code
0037-1106(199908)89:4<888:AAEOTG>2.0.ZU;2-X
Abstract
Ground motion attenuation with distance and the variation of excitation wit h magnitude are parameterized using three-component, 0.25 to 5.0-Hz earthqu ake ground motions recorded in the distance range of 15-500 km for southern California to define a consistent model that describes both peak ground mo tion and Fourier spectra observations. The data set consists of 820 three-c omponent TERRAscope recordings from 140 earthquakes, recorded at 17 station s, with moment magnitudes between 3.1 and 6.7. Regression analysis uses a s imple model to relate the logarithm of measured ground motion to excitation , site, and propagation effects. The peak motions are Fourier velocity spec tra and peak velocities in selected narrow bandpass-filtered frequency rang es. Regression results for Fourier amplitude spectra and peak velocities ar e used to define a piecewise continuous geometrical spreading function, fre quency dependent Q(f), and a distance dependent duration that can be used w ith random vibration theory (RVT) or stochastic simulations to predict othe r characteristics of the ground motion. The duration results indicate that both the variation of the duration data with distance and its scattering decrease with increasing frequency. The ra tio of horizontal to vertical component site terms is about root 2 for all frequencies. However, this ratio is near unity for rock sites and is larger for soil sites. Simple modeling indicates that the Fourier velocity spectra are best fit by bilinear geometrical spreading of r(-1) for r < 40 km and r(-1/2) for r > 40 km. The frequency-dependent quality factor is Q(f) = 180f(0.45) for each of the three components and also for the combined three-component data set s. The T5%-75% duration window provides good agreement between observed and RVT predicted peak values.