No abstract

High‐frequency Rayleigh waves were generated in the frequency range of about 1–100 Hz by an electromagnetic vibrator in co‐operation with the Vibration Instrument Company, Japan, and the Ensol Corporation, North Carolina, at 36 sites in Anchorage. From phase delay times between two sensors, the phase velocity of the fundamental mode at each site was computed. These data were inverted in terms of shear wave velocity structure by a stochastic inversion scheme. Of the 36 sites, values of shear wave velocity (β) as a function of depth were available at seven sites from downhole measurements. At these sites, the comparison of the results obtained by others between surface and downhole measurements showed reasonable agreement. These results were therefore combined with those of the remaining 29 sites obtained in this study. The β structure of the 36 sites could be subdivided into four groups. On the basis of NEHRP provision and from the time‐averaged β structures for the uppermost 30 m corresponding to the four groups, it has been possible to identify the lateral extent of soil classes C and D in the Anchorage area. From statistical analysis of the time‐averaged β data, it is shown that the lithology of the area in soil class C is distinct from that in D. The same is true for the subunits, namely, glaciofluvial deposits in areas of C and D. Moreover, the areas in soil class D along the Knik Arm in west Anchorage with relatively low β‐values coincide with the areas of high ground failure susceptibility identified by others from observations following the Prince William Sound earthquake (Mw = 9.2) of 1964.

In the summer of 1995, surface measurements of shear wave velocity (β) was conducted at thirty six sites, approximately, in the 0-50 m depth range. Of these, at seven sites values of β, soil log and blow count (N) from borehole measurement were available from previous investigations by others. Using these seven sites for calibration, we compared the velocity profiles yielded by the surface and borehole measurements for these sites. The results show broad similarities. Using the soil logs and shear wave velocity variations at the seven sites, four site classes (SC-Ic, SC-II, SC-III and SC-IV) could be identified. The surface method corresponding to the mean value of β tends to underestimate β between about 1 and 18 percent for site classes SC-Ic, SC-II and SC-III compared to the downhole method. For SC-IV, β is overestimated by 11 percent using surface method. Moreover, the blow count (N) data for each site class shows a linear relationship with β obtained by the surface measurement.

Scattering of elastic waves by three-dimensional canyons embedded within an elastic half-space is investigated by using a wave function expansion technique. The geometry of the canyon is assumed to be non-axisymmetric. The canyon is subjected to incident plane Rayleigh waves and oblique incident SH, SV and P waves. The unknown scattered wavefield is expressed in terms of spherical wave functions which satisfy the equations of motion and radiation conditions at infinity, but they do not satisfy stress-free boundary conditions at the half-space surface. The boundary conditions are imposed locally in the least-squares sense at several points on the surface of the canyon and the half-space.Through a comparative study the validity and limitations of two-dimensional approximations (antiplane strain and plane strain models) have been examined. It is shown that scattering of waves by three-dimensional canyons may cause substantial change in the surface displacement patterns in comparison to the two-dimensional models. These results emphasize the need for three-dimensional modelling of realistic problems of interest in strong ground motion seismology and earthquake engineering.

SUMMARYTime-harmonic Green's functions for a triclinic anisotropic full-space are evaluated through the use of symbolic computation system. This procedure allows evaluation of the Green's functions for the most general anisotropic materials. The proposed computational algorithms are programmed in a MATLAB environment by incorporating symbolic calculations performed using Maple Computer Algebra System. Extensive testing of the numerical results has been performed for both displacement and stress ÿleds. The tests demonstrate the accuracy of the proposed algorithm in evaluating the Green's functions.