Comparison of Aftershock Relocations and Velocity Models Through the Vicinities of the 1971 San Fernando and 1994 Northridge Earthquakes, California: Results from Combining LARSE99 Reflection and CISN/SCSN/TriNet Seismicity Data

We use relocated hypocenters and reflection data from LARSE99 to infer the fault surfaces that were ruptured by the 1971 San Fernando and 1994 Northridge earthquakes. Different relocation techniques and different velocity models give somewhat different results. For example, relocation of the San Fernando aftershocks using the technique and velocity model of Mori et al. (1995) gives an alignment of aftershocks at the north end of the aftershock region that projects to the surface at or near the Sierra Madre fault. Relocation of these same aftershocks using the technique and velocity model of Hauksson (2000) moves this alignment not as far south, so that it projects to the surface at or near the San Gabriel fault. Comparison of Mori's and Hauksson's velocity models with that determined independently from LARSE99 indicates that Mori's model is too slow. The relocations of Hauksson are thus, preferred. Mainshock ruptures for the San Fernando earthquake include two echelon planes dipping northward at 45-52 degrees, according to Heaton (1992). There is a hint of these two planes in the relocated aftershocks, but the base of the aftershocks dips northward at a much shallower angle (~25 degrees), passing near the bottoms of both of these inferred planes, and projecting into a prominent north-dipping reflective zone imaged in LARSE99. This reflective zone extends to ~25 km depth at the San Andreas fault. The base of the aftershock zone projects updip to the northwest of the mainshock hypocenter at 25 degrees to the deep projection of the Northridge Hills fault, in the northern San Fernando Valley, as inferred from industry reflection data. Most of these aftershocks to the northwest are along a linear zone that coincides with a lateral ramp in the overlying Santa Susana fault and may, alternatively, reflect a complexity in the San Fernando fault zone. Southeast of the mainshock hypocenter, the base of the aftershocks is steeper than to the northwest. Mainshock rupture for the Northridge earthquake, as modeled by Wald et al. (1996), was apparently simpler than for the San Fernando earthquake, extending from ~16- to 5-km depth along a plane dipping southward at ~35 degrees. This plane is located at or near the base of the relocated aftershocks and projects downward into a southdipping reflective zone imaged in LARSE99. The dip of the base of the aftershocks, and perhaps also the dip of the main rupture plane, becomes steeper than 35 degrees to the west of the mainshock hypocenter. Both the San Fernando and Northridge aftershock zones extend about 5 km beyond the updip limits of the modeled mainshock ruptures and interpenetrate one another. Since the modeled rupture plane of Wald et al. (1996) stops at the base of the San Fernando aftershock zone, it might be inferred that the San Fernando rupture truncates the Northridge rupture. However alignments of Northridge hypocenters are observed above and north of this inferred truncation, suggesting that at least secondary Northridge ruptures penetrate the hanging wall of the San Fernando rupture. This model construction exercise and uncertainties in the final models point toward some of the challenges of using seismicity and imaging data to constrain the three-dimensional structure of faults at depth.