Implications of the Tonopah magnitude-6.5 earthquake for worldwide seismic hazard assessment

By: Ross S. Stein, Ph.D. (Temblor, Inc., CEO)
Shinji Toda, Ph.D. (International Research Institute for Disaster Science, Tohoku University)
Volkan Sevilgen, M.Sc. (Temblor, Inc., CTO)
 

The magnitude-6.5 quake provides yet another indication of the shortcomings of fault-driven hazard models, and the strength of models that use strain instead, particularly for global consistency. To assess what could happen next, we calculate how the rupture transferred stress to nearby faults; the newly stressed faults should now be monitored.
 

Citation: Stein, R. S., Toda, S., Sevilgen, V., 2020, Implications of the Tonopah magnitude-6.5 earthquake for worldwide seismic hazard assessment, Temblor, http://doi.org/10.32858/temblor.091
 

With no vegetation to get in the way, all faults are laid bare, right? Apparently not. Credit: Google
With no vegetation to get in the way, all faults are laid bare, right? Apparently not. Credit: Google

 

A magnitude-6.5 quake struck approximately 35 miles (56 kilometers) west of Tonopah, Nevada early in the morning on May 15th, 2020. The quake was felt throughout the western United States and as far as the San Francisco Bay Area. No major damage occurred but reports emerged showing surface cracking along roadways and other landforms. Scientists are now working out the details about which fault ruptured during the event.

With the closest seismic station 20 miles (35 kilometers) away, the shaking intensity was not well recorded. At that distance the peak was about 10% the force of gravity, well below that typically associated with damage. Near the epicenter, the shaking was likely much higher.

 

It packed a wallop but left barely a trace

To the surprise of many, the earthquake produced no continuous surface rupture. Rather, the cracks seen in roadbeds and dirt tracks seem to be caused by shaking, or as suggested by the work of Chris Milliner, a scientist at NASA/JPL, represent a mesh of short, widely distributed faults. The absence of a clear linear surface rupture is borne out by radar satellite imagery, which shows a series of distributed surface cracks, with no tears along a discrete fault zone, as would be seen if the fault had broken the surface.

So far, aftershocks have occurred along a 15.5 mile (25 kilometer) long band, striking WSW, suggesting that the fault that ruptured has roughly the same orientation and length at depth. There are some small north-striking and west-striking faults within 1.2–6.2 miles (2–10 kilometers) of the epicenter, but no continuous faults.

Strike-slip earthquakes around magnitude-6.5, like the one near Tonopah, typically produce at least 8 inches (20 centimeters) of slip at the surface, along a length of at least 9 miles (15 kilometers; Wells and Coppersmith, 1994). The July 4, 2019 magnitude-6.4 Ridgecrest quake produced up to 3.3 feet (1 meter) of slip over a length of 12.4 miles (20 kilometers). The Tonopah quake produced none. So, this event — which occurred in a similar tectonic environment — is an outlier.

 

The 15.5 mile (25 kilometer) long swath of aftershocks suggest the rupture lies along this orientation (top panel). There are some small north-striking and west-striking faults within 1.2–6.2 miles (2-10 kilometers) of the epicenter, but none are oriented similarly to the pattern of aftershocks (middle panel). The ground deformation from radar satellites shows few tears in the fringes, indicating that the rupture is largely ‘blind,’ meaning that the fault that slipped does not break the surface. A mesh of N-S and E-W (red) faults with very small offsets are also inferred from the satellite radar imagery (bottom panel).
The 15.5 mile (25 kilometer) long swath of aftershocks suggest the rupture lies along this orientation (top panel). There are some small north-striking and west-striking faults within 1.2–6.2 miles (2-10 kilometers) of the epicenter, but none are oriented similarly to the pattern of aftershocks (middle panel). The ground deformation from radar satellites shows few tears in the fringes, indicating that the rupture is largely ‘blind,’ meaning that the fault that slipped does not break the surface. A mesh of N-S and E-W (red) faults with very small offsets are also inferred from the satellite radar imagery (bottom panel).

We cannot depend on mapped faults to forecast hazard

Because of its aridity and accessibility, central Nevada is one of the best mapped regions on Earth. Generations of geologists have scoured the landscape and mapped a cornucopia of faults. And in the past few days, many dozens of geologists have searched for the one that ruptured in the magnitude 6.5 quake, to no avail.

Because this region is so well mapped and a continuous fault along the aftershock trend has not been identified, and because the earthquake did not generate a continuous surface rupture, it is clear that at least for quakes up to magnitude-6.5, we cannot depend on fault mapping anywhere to accurately forecast earthquake hazard. Compounding this, in so many regions, vegetation or urbanization hides faults.

But do we care about earthquakes of this size, or is damage driven by much larger shocks, say magntitude-7.5?

Because there will generally be ten M 6.5 shocks for every M 7.5 shock, the small shocks do matter because there are so many more of them, and near their epicenters, shaking will be strong and damage likely. In fact, a magnitude 6.5 in an urban area can be catastrophic, as occurred in Christchurch, New Zealand, in 2010.

This in no way is meant to diminish the critical and demanding work of fault mapping. Each fault mapped is a seismic source better understood. But we must recognize that worldwide fault inventories are incomplete, and will likely remain so for decades, and so we must build testable models that acknowledge and accommodate this lack of knowledge.

 

Temblor’s PUSH model performs well in these situations

As are most earthquake models, the USGS hazard model (Petersen et al, 2014) is fault-based. This means that it uses the major active faults in a region to make an assessment of the future seismic hazard in the area. Here, the USGS model includes the Coaldale Fault, which is inferred to lie 12.4 miles (20 kilometers) south of the Tonopah epicenter. Between faults, these models include ‘area sources’ that fill in the potential for smaller shocks, such as a magnitude-6.5.

 

Maps of seismic hazard prior to the Tonopah earthquake generated from Temblor’s PUSH (Probabilistic Uniform Seismic Hazard) at top and the USGS 2014 NSHMP (National Seismic Hazard Mapping Project) at bottom for comparison. PUSH does not use faults as input whereas the USGS does. Despite this, the hazard is very comparable, if not slightly higher for PUSH than NSHMP.
Maps of seismic hazard prior to the Tonopah earthquake generated from Temblor’s PUSH (Probabilistic Uniform Seismic Hazard) at top and the USGS 2014 NSHMP (National Seismic Hazard Mapping Project) at bottom for comparison. PUSH does not use faults as input whereas the USGS does. Despite this, the hazard is very comparable, if not slightly higher for PUSH than NSHMP.

 

In contrast, Temblor’s worldwide PUSH (Probabilistic Uniform Seismic Hazard) model is not based on faults. Instead, it uses a blend of past quakes and geodetically measured surface strain, because faults strain the surface, and the strain is measured by GPS receivers. So, strain is a stand-in for faults. For the Tonopah quake, the shaking likelihood is similar, or a bit higher, in PUSH than in the USGS model. So, in our judgment, one does not need faults to forecast hazard. Because in most places around the world, the fault inventory is vastly inferior to that of Nevada, and so a fault-driven model can suffer.

 

In a strain rate map calculated from GPS velocities, the site of the Tonopah earthquake is quite high (Kreemer et al., 2009). PUSH blends strain with the past 45 years of earthquakes to make its forecast.
In a strain rate map calculated from GPS velocities, the site of the Tonopah earthquake is quite high (Kreemer et al., 2009). PUSH blends strain with the past 45 years of earthquakes to make its forecast.

 

Could the Tonopah event trigger a larger earthquake?

There is always a small chance that a mainshock will produce an aftershock larger than itself. So, it is important to determine which faults and areas have been brought closer to failure, meaning an earthquake is more imminent, and which have been inhibited by Friday’s magnitude-6.5 quake. For this, we performed a Coulomb analysis, which calculates how surrounding faults are sheared and unclamped by the mainshock, which promote failure.

 

Coulomb stress changes indicate where the hazard has risen as a result of the magnitude-6.5 mainshock (red ‘stress trigger zones’) and where it has dropped (blue ‘stress shadows’). Since there are both strike-slip and tensional (‘normal’) faults in the Basin and Range province, we show two calculations. The left image shows the relative hazard for normal faults in the region, whereas the right image shows that of strike slip faults of the region.
Coulomb stress changes indicate where the hazard has risen as a result of the magnitude-6.5 mainshock (red ‘stress trigger zones’) and where it has dropped (blue ‘stress shadows’). Since there are both strike-slip and tensional (‘normal’) faults in the Basin and Range province, we show two calculations. The left image shows the relative hazard for normal faults in the region, whereas the right image shows that of strike slip faults of the region.

 

Strike-slip faults located to the north, south, east and west and tensional (‘normal’) faults to the southwest and northeast have been brought closer to failure. There are some long, major faults in the red lobes, and so these are the strongest candidates for triggered events. But fortunately, none are in populated areas.

 

Bottom Line

If we can’t find all the faults capable rupturing strong shocks in Nevada, arguably the most well-mapped region in the world, then we are woefully uninformed about fault locations in most areas of the globe. So, models that blend past earthquakes with current strain give us a powerful, globally consistent alternative.

 

Want to know your earthquake risk? Check it at Temblor.
 

Acknowledgements

We thank Kang Wang (U.C. Berkeley) and Chris Milliner (NASA JPL) for posting their preliminary findings on twitter, Ben Brooks (USGS) for reporting on field mapping, and Alka Tripathy-Lang for scouring the interferograms.

 

Further Reading

Kreemer, C., Blewitt, G., and Hammond, W.C., 2009, Geodetic constraints on contemporary deformation in the northern Walker Lane: 2. Velocity and strain rate tensor analysis, in Oldow, J.S., and Cashman, P.H., eds., Late Cenozoic Structure and Evolution of the Great Basin–Sierra Nevada Transition: Geological Society of America Special Paper 447, p. 17–31, doi: 10.1130/2009.2447(02).

Petersen, M. D., Moschetti, M. P., Powers, M. P., Mueller, C. S., Haller, K. M., Frankel, A. D, et al. (2015). The 2014 United States National Seismic Hazard Model, Earthquake Spectra, 31, S1-S30, doi: 10.1193/120814EQS210M.

Wells, D. L., and K. J. Coppersmith (1994). New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement, Bull. Seismol. Soc. Am. 84, 974–1002.

  • Joanne Verbeek

    Could there have been more than one fault involved in the quake or a fault laid hidden under another fault involved? Just throwing it out there.

  • Ross Stein

    Yes there could. The epicenter lies roughly at the junction of a N-S trend of right-lateral and normal faults and an E-W trend of left-lateral faults. Chris Milliner’s inferred surface ruptures would tend to support this inference.

  • Love your website. Love your blog posts. Very appreciative of the effort to push forward earthquake disaster communication and preparation. However, please please please stop correlating Magnitude and Damage. They’re nearly unrelated. Magnitude is only useful for bookkeeping. PGA and PSA are much more predictive of damage and are directly measurable, location specific, and translate better to a heat-map type visualization.

  • Patrick McClellan

    So, the unfaulted crust immediately overlying this “blind” strike-slip fault is now also strained and subject to rupturing in aftershocks?

  • Ross Stein

    Yes and no. It is probably subject to postseismic creep, but shallow earthquakes are less likely, as they would probably be too shallow to store stress. But repeated satellite deformation imagery will tell the tale.

  • Ross Stein

    Thank you. I don’t know how we gave the impression that it is magnitude, rather than shaking, that does damage. We did report that the closest shaking measurement was 25 km away, recording 10% PGA, fairly low, but near the epicenter it was probably much larger. We also showed both the Temblor and USGS shaking forecasts for the epicentral region at 10% in 50 years probability of exceedance, which is roughly equivalent to the peak shaking expected over the next 500 years. But, we’ll be clearer next time.

  • rsdadvocate4u

    i remember a guy we worked with at tahoe took a group of us down kingsbury back in the late 70s and showed us the Genoa Fault Line. you could most certainly see the shadow of the line for a good mile plus when the sun hit it at a certain time of the day. Was very interesting to learn about, much like this site!

  • Januka Attanayake

    Hi Ross, Can you please add a print button to Temblor articles so that we can download them as pdfs? What an excellent website you’ve developed! Thanks!

  • Dirtmover

    Thanks for a more detailed analysis. Best one I have found yet.
    This has a slight similarity to the Iceland “magma travel” of a few years ago.(Bárðarbunga)
    Previous to this swarm, there was another due West. and West of that is Long Valley caldera.
    Just how far of an outside chance can this be a volcanic event?
    A crazy ditch digger would like to know.

  • Ross Stein

    Great idea. We’ll pursue this.

  • Ross Stein

    It’s not impossible, and you can see that swarm 5-10 km east of Mono Lake in one of our maps. But there are no volcanic features near the Tonopah quake site.

  • Ross Stein

    From Prof. Kerry Sieh, Founder of Earth Observatory of Singapore: Hi Ross, This article seems unnecessarily antagonistic with respect to the value of using mapped faults for assessing seismic hazard. Is there some background politics lately that I’m unaware of?

    Reply from Ross: Thank you, Kerry. Few people have added more insights into faulting and records of prehistoric earthquakes than you. My coauthor, Shinji Toda, is a professional fault mapper and paleoseismologist. He and I both love and value fault investigations. The Temblor app shows active faults in 75 countries.

    But the seismic hazard community has built its universe around the concept that all the major faults have been mapped, that the maximum earthquake magnitude can be estimated from mapped fault lengths or bends and breaks along faults; and that ‘characteristic earthquakes’ can be defined for major faults, which have a much higher rate of occurrence than smaller quakes on those same faults. But all of those concepts have been violated repeatedly by recent large events. So, I believe we need to forecast hazard in a globally consistent way, with data that is uniformly available. That’s what the GEAR model does. And where we have good fault inventories (e.g., California), we can compare GEAR to the USGS/SCEC California model (UCERF3); they are very close to each other. Fault-based models, in my judgment, will be inadequate in so many parts of the world where faults have not been adequately assessed.

  • Dirtmover

    Thanks for the reply. I watched the previous swarm as it happened.
    Then this one started soon after. I haven’t got to check out seismograph info as of yet. Magmatic movement registers differently than everyday run of the mill quakes.(so I’m told) It will take someone plenty smarter than me to figure that out.
    And just noticing another 5.3 hitting within the last 24 hours.
    The real teller would be uplift. We shall see.
    Thanks again for the chat.

  • Dirtmover

    Things are still rocking. another 5+ in the last 24 hours.

  • Ross Stein

    Yes, there are many surface features, all with very small displacements, and many of those displacements inconsistent. The only good map we have seen is Chris Milliner’s, and I believe that supports the conclusion that this is a blind rupture, or one with a distributed set of small faults.

  • Alex H.

    As a field geologist who mapped the displacements referred to above as “inconsistent,” I simply disagree! This rupture is not blind. There are many such “good maps” of remote sensing data, and a map of field observations will be just as useful in complementary ways once all field observations are compiled. Stay tuned.

  • There are actually lots of through-going NS and NNW striking faults underlying that cluster of aftershocks based on decades worth of mapping in the area (most recent maps by Oldow and Cland, 2018).

  • Ross Stein

    Scott, If you would like to send us a figure, we would publish it. If by ‘through-going,’ you mean 25-km-long contiguous faults, I would be surprised if there were many rather than one.

  • Ross Stein

    We would be pleased to publish such a compiled map as soon as it is ready. Nevertheless, the InSAR data shows largely continuous fringes across the area, which can best—and perhaps only—be explained by a largely blind event.

  • I sent the paper in an email