Comment on "rupture Process of the 1987 Superstition Hills Earthquake from the Inversion of Strong-motion Data" by Wald

This comment addresses a discrepancy between the results obtained by Wald et al. (1990) and Frankel and Wennerberg (1989) concerning seismic slip during the 1987 Superstition Hills earthquake (M s 6.6) in southern California (Fig. I). Both studies used strong-motion records to locate subevents that produced the high-frequency (__> 0.5 Hz) strong ground motions for this earthquake. Wald et al. (1990) found that most of the seismic slip for subevent 3 of this earthquake occurred about Ii to 13 km southeast of the hypocenter, along the southern section of the Superstition Hills fault (Fig. 2). In contrast, Frankel and Wennerberg (1989) determined that slip from subevent 3 was located 0 to 8 km southeast of the hypocenter, so that the southern part of the fault did not generate strong ground motions at frequencies >__ 0.5 Hz. Both studies found that subevents 1 and 2 were located near the hypocenter. These papers used different inversion methods on somewhat different sets of strong-motion velocity records. The resolution of the discrepancy between these two papers has important implications for understanding the links between high-frequency (__> 0.5 Hz) strong ground motions, low-frequency (_<_ 0.I Hz) seismic energy recorded at teleseismic distances, and co-seismic surface offset on a fault. Unraveling how and where high-frequency strong ground motion is produced along a fault has major consequences to seismic hazard assessment. The t iming of the arrivals from subevent 3 on the strong motion records for stations at various azimuths presents the clearest evidence supporting the conclusions of Frankel and Wennerberg (1989). Figure 3 taken from that paper shows that the t ime difference between the arrivals of subevents 2 and 3 is essentially identical for stations Parachute Test Site (PTS), Plaster City (PLA), and Calipatria (CAL). These stations have azimuths of 123 °, 181 °, and 68 °, respectively, from the hypocenter (see Fig. 1). The observation that the time separation between arrivals from subevents 2 and 3 is independent of azimuth is strong evidence that the two subevents are located close to each other, clearly within 8 km. This places subevent 3 along the northern portion of the fault. However, the results of the inversion from Wald et al. (1990) indicate that the principal slip zones of subevents 2 and 3 are at least 11 km apart. Calculations based on a half-space with shear-wave velocity of 3.0 km/sec show that such a l l k m difference in the locations of these two subevents (assuming each has a depth of 9 km) would produce a 1.4-sec difference in the time separation of arrivals from these subevents between PTS and PLA and a 1.7-sec difference in the time separation between PTS and CAL. Such an azimuthal difference in time separation for arrivals from subevents 2 and 3 is not observed (Fig. 3). Note that Wald et al. (1990) did not use PLA in their inversion, whereas Frankel and Wennerberg (1989) did. This recording is very important for constraining the rupture geometry because of its azimuth and the impulsiveness of the arrival for subevent 3. The synthetics determined by Wald et al. (1990) from the results of their inversion do not match the t iming of the arrivals in the data. Figure 4 is taken