Hayward Fault Subsystem

1868 earthquake

The Hayward fault (H1 and H2 in Figure 2 ) had at least one major historical earthquake in 1868. Lawson [1908] stated that the surface rupture in 1868 extended at least from Agua Caliente Creek (AC in Figure 5) northerly to San Leandro (SL), and less certainly to Mills College (MC). Recent trenching investigation [Lienkaemper and others, 1995] and analysis of 19th-century triangulation data [Yu and Segall, 1996] suggest that sizable slip occurred as far north as northern Oakland (MT, Montclair trench; BT, BART tunnel). We assume that the location of the southern termination of the 1868 subsurface rupture, at the base of the seismically active crust, coincides with the point where most slip is transferred from the Calaveras to Hayward fault. High creep rate (9-10 mm/yr) occurs on the Calaveras fault north of Halls Valley (HV). North of Calaveras Reservoir (CR) creep on the Calaveras fault is 3-6 mm/yr. We deduce that the high creep rates (9 mm/yr) that occur on the Hayward fault from Agua Caliente Creek (AC) northward [Lienkaemper and others, 1991] indicate that the main subsurface connection of creep between the Calaveras and Hayward faults lies north of the line between Halls Valley and Agua Caliente Creek. Because the microearthquakes (Figure 5) that connect the southern Hayward fault to the Calaveras fault have dominantly strike-slip focal mechanisms, the connection of deep slip probably occurs along a path associated with these small earthquakes [Ellsworth and others, 1982; Wong and Hemphill-Haley, 1993].

Click for high-resolution image

Figure 5. Segmentation of Calaveras and Hayward faults. Yellow circles show M>=2 recent seismicity 1989-95; stars, historic events M>=5.7 [Ellsworth, 1990] by year and magnitude. Open blue squares, trench sites (CA, Camille Ave.; LC, Leyden Creek; MH, Masonic Home; MT, Montclair; TP, Tule Pond; WC, Welch Creek); closed green squares, trilateration arrays (VA, Veras; HV, Grant Ranch in Halls Valley); red triangle SF19, alinement array. Locations: AC; Agua Caliente Creek; BT, BART tunnel; CR, Calaveras Reservoir; DC, Dublin Canyon; ER, Ellworthy Ranch; MC, Mills College; SL, San Leandro

Probably the most accurate measurement of the size of the 1868 earthquake comes from modeling 19th-century triangulation data by Yu and Segall [1996]. Their best-fitting model suggests 1.9 ± 0.4 m slip and M_w 7.0. Trenching evidence [Lienkaemper and others, 1995] at Montclair (MT) in north Oakland suggests that much larger slip occurred there in the previous earthquake (pre-1776; Toppozada and Borchardt, 1996) than in 1868, however this apparent conflict with the triangulation results could be easily explained if the large pre-1776 slip here was mainly shallow and the large 1868 slip was deeper (i.e., underlies the pre-1776 patch in the presumed overlap) as we show in a cartoon, Figure 6. The patch areas shown in Figure 6 yield M_w 7.0 for 1868 and 6.9 for the pre-1776 northern event assuming 1.9 m average slip over 12-km-deep ruptures for both events. Using the methodology of Savage and Lisowski [1993] and the regional slip rates of the NCEP database, the effect of strain reduction by creep in the upper 5 km of the fault zone (~7%) is not likely to be important to these calculations (i.e., fault zone is 93% locked and M_w is reduced only 0.02 by the surficial creep). These assumptions lead to a 210-yr recurrence time for major earthquakes on both segments which is the same as the best estimate of recurrence of Williams [1993] for the southern Hayward Fault based on trenching data at Tule Pond [TP] in Fremont. We do not know how the magnitude and length of the 1868 event compares to earlier events, because the paleoseismic record is not adequate to compare the size of events. If the entire 86-km length ruptured with 1.9 m slip it would be M_w 7.1. Adding the southeastern extension of the fault (H1a) to produce a 112-km rupture would still only produce a M_w 7.2 because its slip rate is much lower, ~3 ± 2 mm/yr.

Figure 6. Hayward fault, historic earthquakes. Cartoon showing assumed locking patches at depth. See text for further discussion and explanation. Abbreviations as in Figure 5

The actual segmentation and recurrence history of the Hayward fault remains highly uncertain. Previously, it was widely assumed that the last major event associated with rupture on the northernmost Hayward fault occurred in 1836, but this association apparently was mistaken [Toppozada and Borchardt, 1997] . Because we do not know which segmentation models are likely to be correct, our database divides the entire fault into two equal parts of 43-km length. The northern extent of the fault under San Pablo Bay was truncated to reflect the right-stepover and transfer of slip to the Rodgers Creek fault, thus leaving no overlap and duplication of seismic moment release. Coseismic slip of 1.9 m and 12 km depth of rupture [Oppenheimer and others, 1993] produces M_w 6.9 events for each hypothetical segment. We cannot currently conclude that the two segment model is more likely than a scenario involving the entire length, so we rank them equally likely. Petersen and others [1996a, b] chose to retain the 1.5 m slip and 167 yr recurrence time adopted by WGCEP [1990] for two Hayward fault segments, that also yields M_w 6.9 and an identical rate of moment release, but they felt it was a more cautious assumption given the great uncertainties in recurrence time.

Rodgers Creek fault

The Rodgers Creek fault (H3; Figure 2a and Figure 8) is believed to be entirely locked (i.e., no recognized creep, <2 mm/yr; Galehouse, 1995; USGS trilateration data, 1978-88) and has not had a major historical earthquake. Segmentation is fairly straightforward because the fault terminates both northward at H4 and southward at H2 in distinct right stepovers of a few kilometers width. In consultation with CDMG we chose to simplify segmentation by eliminating the overlaps of H3 with H4 and H2, so that seismic moment would not be duplicated in the hazard analysis. Actual ruptures would be expected to taper into the stepover overlap region in some complex way. We have adopted the 230 ± 130 yr paleoseismic recurrence time based on 3 events and characteristic coseismic slip of ~2 m from Schwartz and others [1993]. This slip yields M_w of 7.0. Slip rate of 8.4 ± 2 mm/yr by Schwartz and others [1993] suggests that the 9 mm/yr used on the Hayward fault be adopted for the Hayward fault subsystem north of San Pablo Bay as well.