How similar were the 1934 and 1966 Parkfield Earthquakes

Bakun and McEvilly (1984) showed that surface waves from the 1934 and 1966 Parkfield earthquakes are remarkably similar. This observation, together with other seismic and geologic data, led to the forecast of a repeat of a “characteristic” Parkfield earthquake by 1993. In this paper we model near-field geodetic measurements collected before and after the 1934 and 1966 Parkfield earthquakes. The slip in 1934 on the northwestern half of the rupture is poorly constrained due to the limited station coverage. In this part of the rupture, near the hypocenter, the geodetic measurements permit the two earthquakes to have the same slip distribution. On the other hand, the data clearly show that the 1966 earthquake ruptured through the right step in the surface trace of the fault, while the 1934 earthquake stopped northwest of the step. If we assume only that the slip distribution is to some degree smooth, we find that the centroid of the 1934 slip distribution could be as much as 10 km northwest of the centroid of the 1966 slip. The data are consistent with, but do not demand, the largest slip in 1934 occurring in a region of low slip in 1966. The geodetically determined moments of the two earthquakes from smooth inversions are indistinguishable (M0 = 4.4 ± 0.4 × 1018 N m). We also examine inversions in which we assume that the two earthquakes start identically and estimate how far down the fault the two earthquakes could have remained the same. We find that the 1934 event could have had the same slip as in 1966 up to the fault step. The 1934 earthquake must have stopped at the step, whereas the 1966 event slipped southeast of the step. In this “quasi-characteristic” model, M0 of the 1934 quake is only 60% of that in 1966, the difference presumably occurring as afterslip in 1966. If the 1934 geodetic moment was indeed less than in 1966 and the loading rate was constant, this observation could explain the “early” occurrence of the 1934 earthquake.