EMPIRICAL-ANALYSIS OF STRONG GROUND MOTION FROM THE 1992 LANDERS, CALIFORNIA, EARTHQUAKE

A total of 167 ground-level accelerograms recorded during the 28 June 1992 M(S) 7.6 (M(W) 7.3) Landers, California, earthquake were used to study the dependence of peak horizontal acceleration and 5%-damped horizontal response spectra on distance, azimuth, and site geology. A comparison of these recordings with ground-shaking amplitudes predicted by contemporary attenuation relationships indicated that (1) relationships developed prior to the Landers earthquake did a reasonably good job at predicting the Landers ground motions within about 60 km of the fault, but underpredicted the ground motions at further distances by as much as a factor of 2 to 3; and (2) relationships developed after the Landers earthquake did a reasonably good job at predicting the Landers ground motions within the distance ranges for which they were applicable. An extrapolation of near-source attenuation rates based on the newer relationships was found to significantly underpredict the Landers ground motions at distances beyond about 60 km, indicating a need to develop attenuation relationships that can account for differing rates of attenuation at near-source and far-source distances. Peak accelerations and response spectra were found to be significantly affected by source radiation pattern, source directivity, and site geology, with higher ground motions observed in the direction of rupture propagation and lower ground motions observed on rock beyond distances of about 40 km. Regression analyses of the Landers recordings indicated an unusually low rate of attenuation, controlled primarily by recordings obtained at distances greater than about 60 km, which can be attributed to one or more of the following factors: (1) a change in geometric attenuation rate as a result of the dominance of surface waves; (2) amplification in the Los Angeles Basin, the San Fernando Valley, the San Bernardino Valley, and the Coachella Valley because of the effects of deep sediments and basin response; and (3) the arrival of critical reflections off the base of the crust (MOHO). The fault-normal components of response spectra at sites located closest to the fault were found to be systematically larger than the fault-parallel components at periods of 1 sec and greater. This effect was largest at stations located north of the epicenter in the direction of rupture propagation and indicates a need to account for this effect in the design of long-period structures.