The Rodgers Creek and Hayward Faults are revealed to be one fault, capable of a Magnitude=7.4 earthquake

By Ross Stein, Temblor

Check your seismic hazard rank

A new finding by Janet Watt and her USGS coauthors appeared last month in Science Advances , heralding a “Missing link between the Hayward and Rodgers Creek faults.” From fine-scale seismic imaging and other geophysical sounding, they argued that the Rodgers Creek and Hayward faults are continuous, connecting through a curved fault section hidden beneath San Pablo Bay. This means that earthquakes as large as M=7.4 are possible on both sides of San Francisco Bay, rather than just on the San Andreas, as we had thought before.

The new San Pablo Bay link between the Rodgers Creek Fault (which extends out of the image to the upper left to Santa Rosa) and the Hayward fault (which joins the Calaveras Fault out of the image to the lower right) is orange. The beautiful base image is by Robert E. Crippen (NASA), and is freely available from


What does this discovery foretell for seismic hazards in the East Bay?

I put this question to USGS colleague and Watt et al. coauthor, Tom Parsons. Tom replied, “I think Janet’s observations show an easier connection than I thought was there. The fault also shows a sharp discontinuity right at the surface, which is very surprising in recently deposited Bay mud. It suggests that either the fault is creeping there, or there has been recent seismic slip.” In fact, according to Watt et al., even the most recent prehistoric earthquakes along the Rodgers Creek fault and Hayward faults, which struck sometime between 1715 and 1776, could have been a single combined event given dating uncertainties. This means that it will be extremely difficult to prove whether such a combined rupture has happened in the past.

I asked if this changes the USGS earthquake probabilities for M=7.4 East Bay shocks. Parsons replied that in the current USGS model (Field et al., 2015), if a rupture is deemed capable of jumping from one fault to the other, then the combined event is just as likely as if there were no jump (or ‘step-over’) at all, and this was the case for the Rodgers Creek and Hayward faults. So, the Watt et al. paper does not change the probability of a combined rupture, which stands at about 4% in the next 30 years, or one chance in 25.


The faults as they now appear in Temblor. The combined Rodgers Creek-Hayward Fault extends from Santa Rosa to San Jose (both outside the frame), two cities that were also badly damaged in the 1906 earthquake.


But a M=7.4 could be worse than a lone Rodgers Creek or Hayward Fault rupture

Whenever ruptures speed up or slow down, they shed seismic energy—which we feel as strong shaking. The curve in San Pablo Bay would likely be a ‘speed-bump’ to the rupture front as it unzips at about 5,000 km/hr (3,000 mph) down the fault, causing a spike in shaking, most strongly affecting the surrounding cities of Richmond, Vallejo, Novato, and Napa.

A M=7 shock on either fault would rupture for up to 100 km (60 mi), which would take up to 50 seconds. A combined rupture could double that duration to up to 100 seconds. Compare that to the M=6.9 Loma Pieta shock, which took a mere 20 seconds. The longer the shaking continues, the more buildings, bridges, pipelines, and dams would fatigue. Think of a thin stainless steel strap: Bend it back and forth a few times and it weakens; do this repeatedly and it snaps.

USGS earthquake engineer, Erol Kalkan, commented to me that “Long-duration shaking may weaken structures due to ‘low-cycle fatigue,’ and aftershocks could escalate failures of structures already weakened by the mainshock.” He also pointed out that most seismic design codes, including in the U.S., are based on ground-shaking intensity, but do not explicitly take into account the duration. A long shaking duration may also increase the liquefaction hazard.” This much is clear: the stronger the building, the better it would fare.

Rather than concealing the link, San Pablo Bay is its offspring

You would be forgiven for thinking that the reason it has taken so long to reveal the fault link is that it’s hidden under San Pablo Bay. But it’s actually cheaper and easier to capture high quality seismic reflection images like the one below with a single ship than with a fleet of trucks, and marine sediments preserve the delicate stratification of layers better than on land, particularly urban land.

The linking fault, denoted by the arrow, is unmistakable in this cross-section from Watt et al. (2016), looking like a bad slice through a wedding cake, or a French millefeuille pastry. Yes, I am dreaming about dessert—in Paris.


Instead, repeated fault slip through the curved section has caused the area to subside, creating the trough in the landscape that captures the drainage from the Great Valley that we know as the Delta. So, perhaps if the fault had not thrown us a curve, San Francisco would not have been the gateway to the Gold Rush, sending countless steamships across the ‘missing link,’ in San Pablo Bay, ferrying would-be miners to Sacramento and the Sierran gold fields.

The USGS’ Janet Watt being interviewed on CBS ‘This Morning’ in November.


Janet commented to me, “The ability to measure centimeter-scale deformation on the Earth’s surface through satellite radar (InSAR), topographic (LiDAR), and seismic (chirp) measurements has illuminated heretofore unknown fault connectivity. The 2016 M=7.8 Kaikoura earthquake in New Zealand is another case where these measurements could reveal whether the seemingly independent ruptures are also connected.”

If it’s one fault, it should have one name

If the Watt et al. paper stands, and I see no reason why it will not, it’s time to retire the name ‘Rodgers Creek Fault, and rechristen the entire 175 km (120 mi) long feature the ‘Hayward Fault.’ This is something that the USGS and California Geological Survey could do in concert, and it would be a fitting reminder of this fault’s unrivaled ability to damage a major U.S. urban corridor.



Janet Watt, David Ponce, Tom Parsons, Patrick Hart (2016), Missing link between the Hayward and Rodgers Creek faults, Science Advances, 2: e1601441, DOI: 10.1126/sciadv.1601441.

Edward H. Field, Glenn P. Biasi, Peter Bird, Timothy E. Dawson, Karen R. Felzer, David D. Jackson, Kaj M. Johnson, Thomas H. Jordan, Christopher Madden, Andrew J. Michael, Kevin R. Milner, Morgan T. Page, Tom Parsons, Peter M. Powers, Bruce E. Shaw, Wayne R. Thatcher, Ray J. Weldon II, and Yuehua Zeng (2015), Long-Term Time-Dependent Probabilities for the Third Uniform California Earthquake Rupture Forecast (UCERF3), Bull. Seismol. Soc. Amer., 105, 511–543, doi: 10.1785/0120140093

  • Bay Area Guy

    Curious how this might affect the Hazard Rank (from:

    • Ross Stein

      Good question. I believe it will slightly increase the Rank. The reason why the effect will be muted is that such large, rare shocks will soak up some of the seismic moment, slightly lowering the rate of smaller but more frequent shocks. In the next California rupture model (UCERF4?), we will see.

  • Suzanne Hecker

    Nice summary, Ross. To further investigate how earthquake ruptures may link up across San Pablo Bay, the U.S. Geological Survey is planning to trench across a southern splay of the Rodgers Creek fault (mapped in detail using airborne lidar imagery) that is on land and projects toward Watt’s newly discovered fault section beneath the bay. Trenching should expose the recent history of surface-rupturing (moderate to large) earthquakes. If we obtain clearance in time, we’re hoping to trench next summer.
    -Suzanne Hecker, USGS Menlo Park

    • Ross Stein

      Suzanne is a leading ‘paeloseismologist’ (the study of pre-historic quakes) at the USGS in Menlo Park. We forget how hard it is to get permission to do urban geology, despite the fact that this is where geology most counts.

  • Michael Blanpied

    The possibility of a link between the Rodgers Creek and Hayward faults has been the subject of speculation, calculation and study for many years. Working Groups on California earthquake probabilities have included the potential for joint rupture in their analyses as far back as 1999, but those groups have been hard pressed to assign a probability to that happening because whether and how the faults were linked was unknown. Watt and others’ results are thus a big step forward in our knowledge of Bay Area faults and the hazards they pose. Thanks, Ross and Temblor, for a nice summary of that work and its implications.

    • Ross Stein

      I could not agree more that Watt et al. (2016) is an important advance, and will accelerate the pace of Hayward Fault research that Patrick Williams, Suzanne Hecker, Jim Lienkaemper and others are carrying out—squarely in the public interest.

  • Tracey McTague

    I wondered about that…

    • Ross Stein

      The big picture Tracey, is that the Hayward fault has about half to a third of the slip rate of the San Andreas, and is one-fifth the San Andreas’s length, so it should be a bit player. Despite this, the mean inter-event times for the San Andreas (~250 yr) and Hayward Fault (~150 yr) are similar. I think that’s because the Hayward is simply a captive of the San Andreas. Stress thrown onto the nearby Hayward by the San Andreas dominates the Hayward’s behavior.

  • Rick Harlan

    Given the recent work by Chaussard et al suggesting a direct surface and subsurface link between the central Calaveras and Hayward faults, does anyone know if consideration is being given to the idea of the central Calaveras participating in a multi segment Hayward-Rodgers Creek rupture? If such a longer cascading rupture were considered plausible, could it raise the MCE above the level of M7.4 suggested for the Hayward-Rodgers Creek scenario? Granted the central Calaveras historically shows a high surface creep rate, but previous trenching along the southern part of the central Calaveras has suggested possibility of M7 sized surface ruptures having occurred prehistorically. So – my question is – is the central Calaveras capable of adding enough seismic moment to contribute to/participate in a multi-segment Hayward-Rodgers Creek rupture, such that the MCE could be greater than M7.4? Is that idea even remotely plausible?

    • Ross Stein

      The Hayward and Calaveras Faults merge east of San Jose and Gilroy, and so it probably would be easy for a rupture to jump from one to the other. (Faults are dumber than dogs–they don’t know their own names.) I believe that the California rupture model (UCERF3) already includes this possibility. Maximum Credible Earthquake (MCE) is another matter; it depends on how common such a cascading rupture could be.

  • Thank you for your explanation of the formation of the Delta! Let’s see if I get it. The curved section of the fault is convex in relation to the Delta. When it slips, it pulls toward the west away from the Delta, causing that land to subside. Is this what you are saying?
    Question: If the Delta subsides, something else must build up, perhaps to the West. What would that be? There’s no bulge that I can think of west of San Pablo Bay.

    • Ross Stein

      Great question. Because of the fault curvature, earthquake slip results in subsidence just east of the fault in San Pablo Bay; uplift on the northwest side is more subtle. So, the Bay will slowly change shape with time. There is a calculation and map of this in the Watt et al. paper, Fig. 6B.

  • Tom Regan

    Please pardon me for my ignorance. I’m not a seismologist, but in looking at the fault map presented above, it appears the Hayward fault would more naturally align better with the unnamed fault nearly parallel to and west of the Rodgers Creek fault. Since we can’t see the locations of the geophysical transect shown above it is hard to know where or why the jog is necessary. In my simple mindedness I could see the Rodgers Creek and Hayward faults not necessarily connected, but overlapping like many faults within the broader San Andreas fault zone and Pacific/American plate margin. We see this same thing in the Coachella Valley.

    Tom Regan, PG, CEG, CHg

    • Ross Stein

      Good point. The fault just west of the Rodgers Creek Fault north of San Pablo Bay is called the Tolay Fault. It is believed to be an ancestral northwest extension of the Hayward fault, and perhaps has not been active in the past 20,000 years and perhaps not even in the past several hundred thousand years. So, Rodgers Creek appears to have retired the Tolay. This kind of succession is common as fault systems grow and link up; the big ones decommission the wanna-be’s and has-been’s.

  • Patrick Williams

    Paleoseismic evidence of Hecker and colleagues suggested the interaction of these fault segments. Several ages found for the penultimate southern Hayward fault rupture are quite similar to those of the most recent Rodgers Creek event. Williams, Baldwin and Lienkaemper (NEHRP 2013) found that at San Pablo along the northern Hayward fault the most recent event also closely fits the age of the latest Rodgers Creek event (MRE). Disruption associated with six events is bimodal. Either strongly expressed as in E1 (MRE) and E5 (by a large fissure and shattered ground), or fairly weakly expressed as in E2, E3, E4, and E6. Because the San Pablo section of the fault creeps rapidly it was suggested separately by Burgmann, Lienkaemper and Simpson that it contributes little to seismogenic rupture. Dramatic surface disruption in only two of the six latest ruptures supports the aseismic suggestion and indicates that the San Pablo section of the fault may only produce substantial surface rupture in combined Hayward-Rodgers Creek events.

    • Ross Stein

      Thank you, Patrick. A little translation: MRE is the Most Recent Event (not Meals Ready to Eat), E1 is earthquake Event 1, with the numbers increasing with age. By ‘bimodal,’ he means that some prehistoric events have a large surface ruptures, others do not, without many modest ruptures.