Electronic Supplement to
Long-Term Creep Rates on the Hayward Fault: Evidence for Controls on the Size and Frequency of Large Earthquakes

by James J. Lienkaemper, Forrest S. McFarland, Robert W. Simpson, Roger G. Bilham, David A. Ponce, John J. Boatwright, and S. John Caskey

1. Plots of 2007 creep event
2. Comparison of creepmeter and alinement array data
3. Maps and analysis of Rodgers Creek-Hayward fault connection
4. Depth of creep trials and tables of long-term creep observations

The explanatory text of this supplement follows the list of all tables and figures, and is composed of four sections: (A) the time histories of the 2007 creep event and models, (B) comparisons of colocated creepmeter and alinement array slip histories, (C) analysis of microearthquakes locations in the Rodgers Creek-Hayward fault connection, and (D) results of depth-of-creep trials based on best-fitting the long-term creep observations.

---

List of Tables and Figures

Table S1a. Best post 1989 Loma Prieta earthquake creep rates (following quiescence)

Table S1b. Smoothed best post-1989 Loma Prieta creep rates discretized to 1-km

Table S2. Calculated magnitudes for 161-yr accumulated strain along a 70-km-long Hayward fault for a range of depth (z) of loading and locking

Table S3. Calculated magnitudes for range of recurrence interval uncertainty: 161 +/- 61 yr (2 sd)

Figure S1. We model observations (in black) from alinement arrays (McFarland et al., 2009) and a creepmeter (Bilham et al., 2004) using the empirical afterslip modeling program, AFTER (green lines; Boatwright et al., 1989; Budding et al., 1989) resulting in estimates of final slip (Uf), duration (T) and power law exponent (C) [U(t) = Uf/(1+(T/t))^C]. Background creep rate was subtracted from observations to obtain solutions in AFTER; In plots above, background rate was added to those solutions for direct comparison to the data. AFTER solutions usually require time input with t0 as time of mainshock, but for this event we approximate t0 by the arrival time of the wave front of slip at the surface as calibrated by its arrival at CTM 12 days post-earthquake. All AFTER solutions appear consistent with this assumption.

Figure S2. Comparison of creep measured near the north end of the Hayward fault on alinement array (HCCC) and creepmeter (CPPP) with no adjustment to the creepmeter data. Ratio of creep rates CPP/HCC 1995-2009 is 99.7%. Alinement array data probably were seasonally affected by tree roots in 2003-2004 before being rebuilt in 2005.

Figure S3. Comparison of creep measured in north Oakland on alinement array (HTEM, 1973-present) and creepmeter (CPPP, 1997-present) includes a scaling adjustment made to the creepmeter data. The ratio of CTM/HTEM rates is 78% (1998-2009). A deflection array in the BART tunnel (1985-1999) 0.56 km north of HTEM indicates the same creep rate (4.05 mm/yr) for their period of overlap (1994-2000). An earlier deflection array using a laser beam for alinement suggested a possibly lower rate (1972-1993), but error estimates for these earlier readings were 2-3 times larger than the later deflection readings (1985-1999).

Figure S4. Comparison of creep measured in south Oakland on alinement array (HENC ) and creepmeter (CPPP) with no scaling adjustment to creepmeter data. The ratio of COZ/HENC rates is 117% (1996-2010).

Figure S5. Comparison of creep measured in Hayward on alinement arrays (HPAL, 1976-present; previous array, 1989-1992) and creepmeters (CPPP, 1994-present; HWP1, 1970-1993, Schulz, 1989, K. S. Breckenridge, written communication, 1993). Present alinement array (HPAL) established by City of Hayward in 1976, recovered in 1992 (Lienkaemper and Galehouse, 1997). No scaling adjustment to creepmeter data. Ratio of CHP/HPAL rates is 78% (1998-2009). Although colocated the 10-m-long creepmeter (on 10-m-deep screw piles) expresses creep at different times than the 174-m-long alinement array, the latter reflecting a more pronounced stick slip behavior on a slightly deeper part of the fault. Both the current and previous alinement arrays suggest quiescence following a possible triggered slip from the 1989 Loma Prieta event.

Figure S6. Timing of the 1996 creep event is deduced by comparison of creep measured in part of Fremont (km 63-65) on alinement arrays (HPIN, 1989-present; HUNI, 1993-present) and creepmeter (CFW, 1994-present). Present alinement array (HPIN) initial survey by Fremont in 1989, recovered in 1993 (Lienkaemper and Galehouse, 1997). These data encompass the northern limit of the major creep event of 9 Feb. 1996 causing rupture of a major water main near km 66 (exact time unknown). Slip reached the creepmeter 2.4 km to the north (CFW, km 63.6) by that evening (6 PM PST), but did not reach another array (HUNI,) 2.9 km north of the broken pipe, for several months, probably in early 1997 when CFW indicated an additional creep event occurred (0.5 km south of HUNI). Creepmeter data scale has not been adjusted. Ratio of CFW/HUNI rates (not collocated sites) is 105% (1994-2009). Thicker blue line interprets the timing of 1989 triggering and 2006 creep event (HPIN), and the green line, the delayed arrival of the 2006 creep event at HUNI.

Figure S7. Double-difference earthquake locations 1984-2008 (Waldhauser, 2008). A, map view shows shallow events (see legend below) on trend with Hayward fault for ~10 km into San Pablo Bay, then they appear to bend and steepen toward the Rodgers Creek fault into a linking fault, B, cross-section of events from km -20 to km 5, circles indicate the surface traces of the mapped faults.

Figure S8. Double-difference earthquake locations 1984-2008 (Waldhauser, 2008). A, map view shows shallow events (see legend below) on trend with Hayward fault for ~10 km into San Pablo Bay, then they appear to bend and steepen toward the Rodgers Creek fault into a linking fault; color gradient indicates gravity contours of Ponce (2001), B, cross-section of events from km -20 to km 5, circles indicate the surface traces of the mapped faults.

Figure S9. Double-difference earthquake locations 1984-2008 (Waldhauser, 2008). A, map view shows shallow events (see legend below) on trend with Hayward fault for ~10 km into San Pablo Bay, then they appear to bend and steepen toward the Rodgers Creek fault into a linking fault; color gradient indicates magnetic anomaly contours from Phelps et al. (2008) B, cross-section of events from km -20 to km 5, circles indicate the surface traces of the mapped faults.

Figure S10. Double-difference earthquake locations 1984-2008 (Waldhauser, 2008). Fault normal cross-sections referenced to grid used in Fig. C1.

---

A) 2007 creep event, observations and models from AFTER

We used the program AFTER of Boatwright et al. (1989; Budding et al., 1989) to model the creep transient observed on five alinement arrays and a creepmeter following the 2007 Oakland earthquake (Fig. S1).

---

B) Comparison of creepmeter and alinement array data

We present a plot for each of the Hayward fault creepmeters (Bilham et al., 2004) showing the accumulated creep at each site, also including data from nearby or collocated alinement arrays for comparison (Figures S2, S3, S4, S5, S6)

---

C) Connection between Rodgers Creek and Hayward faults

We include a map and cross-sections (Figs. S7, S10) showing microearthquakes relocated by Waldhauser and Schaff (2008) using the double-difference method of Waldhauser (2001) and Waldhauser and Ellsworth (2000, 2001). In addition, we include active traces of the Hayward fault (Lienkaemper, 2006) and Rodgers Creek fault (Qfaults06) and gravity data (Fig. S8) of Ponce (2001) and magnetic data (Fig. S9) from Phelps et al. (2008). Plotting of microseismicity as color-coded depth symbols and in cross-sections in the north end of San Pablo Bay suggests that the Hayward fault transitions into an 8-km-long link fault, which connects with the Rodgers Creek fault near the southeast end of its mapped surface trace.

---

D) Iterative solution for depth of creep along the Hayward fault

We present a sequence of plots showing stepwise improvements in fit using the method of Simpson et al. (2001) as a series of PNG image files displayed as a slideshow using Javascript. For each file the model values of surface creep rate are indicated by the hachured line, the dotted line shows the 1-km discretized values (Table S1b) of the observed mean long-term creep rates as determined from the data in Fig. 3 of the text (Table S1a). These data with error bars are also included in each plot to visualize goodness of fit. Below each plot is a cross section showing the resulting estimates of creep rate with depth for that iteration. Trial number 4 provides the best fit and it is the model shown in Fig. 4 of the text.

The following link lets you view the iterative solution for driving depth at 12 km [Creep Depth Trials, 12 km].

Alternative models may be viewed the same way. Assume driving depth at [10 km], [14 km], or tapered from driving depth of [14 km to base of locking at 12 km] or tapered from driving depth of [12 km to base of locking at 8 km] or tapered from [a 14 km base to locking above 10 km]. The magnitude of earthquakes possible for HF depend especially (±0.2) on the accumulation time since the previous earthquake, with an average time between events of 161 ± 61 yr. The second largest factor (±0.1) is what fraction of slip deficit held in the creep patches is released coseismically. The depth of driving slip and presence or absence of coseismic slip below the seismogenic zone is a relatively minor factor (±0.06) as shown in the summary tables (Tables S2 and S3).

[ Back ]