The so-calculated residuals are correlated with a positive scalar factor signifying the focusing potential at each observation point. The focusing potential is determined on the basis of the geometrical relation of the station relative to the rupture front on the fault plane.
The second data set consists of the acceleration directions derived from the windowed-time histories of the horizontal ground acceleration across the El Centro Differential Array (ECDA). The horizontal peak velocity residuals and the low-pass particle acceleration directions across ECDA require the fault rupture to propagate northwestward. The horizontal peak ground acceleration residuals and the high-frequency particle acceleration directions, however, are either inconclusive or suggest an opposite direction for rupture propagation. The inconsistency can best be explained to have resulted from the incoherence of the high-frequency radiation which contributes most effectively to the registration of PGA.
A test for the sensitivity of the correlation procedure to the souce location is conducted by ascribing the observed strong ground shaking to a single asperity located 12 km northwest of the hypocenter. The resulting inconsistency between the peak acceleration and velocity observations in relation to the focusing potential is accentuated. The particle velocity of Delta Station, Mexico, in either case appears abnormally high and disagrees with other observations near the southeastern end of the fault trace. From the observation of a nearly continuous counterclockwise rotation of the plane of P-wave particle motion at ECDA, the average rupture velocity during the first several seconds of source activation is estimated to be 2.0 to 3.0 km/sec. A 3 km upper bound estimate of barrier dimensions is tentatively made on the basis of the observed quasiperiodic variation of the polarization angles.