Can changes in Earth’s rotation be used to forecast earthquakes?

Preface by Ross Stein, Ph.D., Temblor

Roger Bilham and Rebecca Bendick’s presentation at the annual meeting of the Geological Society of America in Seattle last month fueled a firestorm of press coverage about future earthquakes, so we invited them to post an article for a wide audience. The talk extended the observations discussed by Bendick and Bilham in Geophysical Research Letters in August and is the subject of a new article describing a causal mechanism and its consequences, which is as yet unpublished. It is in review awaiting an editorial decision. So, in fairness to the journal reviewers and to the authors, Temblor has suspended its Comment/Reply forum for this article until their paper is published. In an effort to balance the needs of public access and peer review, we hope our readers will understand this editorial decision.

 

by Rebecca Bendick, Ph.D., University of Montana and Roger Bilham, Ph.D., University of Colorado Boulder

 

Recent media reports have described a proposed relationship between earth rotation and the number of large (M>7) earthquakes. In contrast to these headlines and their exaggerated claims, the conclusions they report are uncontroversial and are based on a straightforward analysis of publically available data.

We started by thinking about earthquakes as repeating systems of a particular type, integrate-and-fire oscillators. Other researchers (for example, John Rundle and Chris Scholz and even the UCERF3-TD team) have argued that this is a reasonable approximation, with fault patches requiring some time to accumulate elastic energy, after which they can fail at any time. If quakes do have this property, then we should expect sets of events with similar renewal interval (the time required to accumulate slip potential of the observed amount), to occur closer together in time than would otherwise be expected, because over many earthquake cycles, they nudge one another into phase alignment through very weak interactions. Papers by Chris Scholz and Charlie Sammis & Stewart Smith describe the effect for California and synthetic events, respectively. The forces required to synchronize events can be very small. For pendulum clocks in an otherwise empty room, even the sound of their ticking can nudge them into alignment over many cycles.

When we searched for clusters of events in time, based on renewal intervals, we found some intriguing patterns (GRL 10.1002/2017GL074934). First, events do appear to weakly cluster in this way. Second, numbers of Mw≥7 earthquakes in the past hundred years (15±5 earthquakes a year) fluctuate at 25±10 year intervals. Although these fluctuations appear somewhat random when considered in isolation, they correspond to similar fluctuations in decadal changes in the length of the day. Boris Levin and Elena Sasarova in Russia, and Peter Varga and colleagues in Hungary have also noted these relationships and have written extensively about the power spectrum of the seismic catalog and its relationship to earth rotation. In a recent invited talk at the Geological Society of America meeting in Seattle we noted that Earth’s angular deceleration (the time derivative of the length of the day) which precedes a slowing in Earth’s rotation by 5 or 6 years, can be used to forecast periods of slow angular rotational velocity (long length of day), and in principle can be used to forecast periods during which increased global numbers of Mw≥7 earthquakes occur (see figure).

During prolonged periods of slow rotational velocity (long length of the day) we find that the number of Mw≥7 earthquakes increase to about 15% above the long term average. This amounts to only a handful of additional earthquakes, statistically only 2-3 using a 5 year average, but sometimes more than 8. The earthquake catalogs (Centennial, PagerCat, and ISC/GEM are to various degrees incomplete for the past century of smaller magnitude earthquakes but in a global search the relationship appears to persist to Mw≥6, indicating that timing of dozens of potentially damaging Mw≥6 earthquakes are also influenced by Earth’s rotation.

 

Correspondence between global seismicity and Earth's angular acceleration.  Deceleration is plotted with a 5.5-year phase lead. Though imperfect, the correspondence permits global seismicity rates to be anticipated 5 years in advance.  We suppress in each time series periods longer than 50 years and shorter than 5 years.
Correspondence between global seismicity and Earth’s angular acceleration. Deceleration is plotted with a 5.5-year phase lead. Though imperfect, the correspondence permits global seismicity rates to be anticipated 5 years in advance. We suppress in each time series periods longer than 50 years and shorter than 5 years.

 

We envisage a mechanism in which, because many fault patches have similar renewal intervals of 20-60 years, even in the presence of very weak and noisy synchronization, many faults will be near failure at the same time. When the earth slows, the Earth’s mantle shrinks in radius, reducing its equatorial circumference by a few millimeters. Since the Earth’s lithospheric plates are unable to shrink to match this reduced circumference they preferentially increment strain at plate boundaries where earthquakes are already poised to occur. This advances the timing of these earthquakes, which is recorded as a decadal peak in seismic productivity. The flurry of earthquakes is followed by years with fewer events than average, because once the peak has passed, not only are there now fewer fault patches sufficiently mature to rupture, rotation-induced compression at plate boundaries is relaxed by the increase in Earth’s circumference attending faster rotation.

We emphasize that Earth’s rotation does not cause earthquakes, it merely advances or retards the timing of some 15% of the world’s plate boundary inventory of earthquakes poised to occur. Also, we consider numbers of earthquakes only, not their energy release. In our analysis, a Mw=9 earthquake is the same as a Mw=7.0 earthquake. As it happens, most of the world’s most damaging earthquakes have magnitudes less than M=8 and are responsible for 50% of the world’s earthquake related fatalities. They are moreover the most abundant.

Because many fault systems globally have a wide range of renewal intervals (tens to tens of thousands of years) and the alignment of their timing is also affected by direct triggering, strain partitioning, and other local interactions, large earthquakes occur all the time, and the catalog looks pretty random. We still have a long way to go in forecasting seismicity, but decadal intervals of enhanced probability for large events should encourage local authorities to implement seismic retrofits they may be considering sooner than later. A predictable decadal increase in damaging earthquakes (and its corollary) is also of significant utility to the earthquake insurance industry.

 

Some background reading

 

Anderson, D.L., (1974) Earthquakes and the Rotation of the Earth,  Science 186 (4158), 49-50. [doi: 10.1126/science.186.4158.49]
Bendick, R., and R. Bilham (2017), Do weak global stresses synchronize earthquakes?, Geophys. Res. Lett., 44, 8320–8327, doi:10.1002/2017GL074934.
Bilham, R and R Bendick (2018) A 5 year warning of increased global seismic hazard from Earth’s variable rotation, Geophysical Review Letters (in Review)
Kanamori, H., (1977), The energy release in great earthquakes, J. Geophys. Res., 82, 2981-
Levin B. and E. Sasarova, (2017).  The Earth’s entry into a new phase of reduction of its angular velocity and an increase in its seismic activity,   Geophys. Res. Abstr. 19, EGU2017-2933.
Levin B. W. and E. V. Sasarova (2015a), Dynamics of seismic activity during the last 120 years, Doklady Earth Sciences, 461, 254-259. original Russian Text Doklady Akademii Nauk, 461(1), 82-87. ISSN 1028-334X
Levin B. W. and E. V. Sasarova (2015b), Relationship between variations in the rotation velocity of the Earth and its seismic activity, Doklady Earth Sciences, 464, 987-981, original Russian Text Doklady Akademii Nauk, 461(1), 351-355.
Levin, B.W., E.V. Sasarova, G. M. Steblov, A. V. Domanski, A. S. Prytov and E. N. Tsyba (2017),  Variations of the Earth’s rotation rate and cyclic processes in Geodynamics, Geodesy and Geodynamics, 8, 2016-2012  dx.doi.org/10.1016/j.geog.2017.03.007.
Sammis, C., and S. Smith (2013), Triggered tremor, phase-locking, and the global clustering of great earthquakes, Tectonophysics, 589, 167–171.
Scholz, C. (2010), Large earthquake triggering, clustering, and the synchronization of faults, Bull. Seismol. Soc. Am., 100, 901–909, doi:10.1785/0120090309.
Shanker, D., N. Kapur, and V. Singh (2001) On the spatio-temporal distribution of global seismicity and rotation of the Earth—a review, Acta Geod. Geoph. Hung. 36, 175-187.
Varga, P, D. Gambis, Z. Bus and Ch. Bizouard (2004)   The relationship between the global seismicity  and the rotation of the Earth. Journées 2004 – systèmes de référence spatio-temporels. Fundamental astronomy: new concepts and models for high accuracy observations, Paris, 20-22 September 2004, edited by N. Capitaine, Paris: Observatoire de Paris, ISBN 2-901057-51-9, 2005, p. 115 – 120.

  • KenW

    So, the graph only goes through current time. Is the rotational deceleration predictable and increasing in the near future (extend the graph, please)? Even though it may be a weak correlation, it appears to be one additional factor to already high strain faults possibly rupturing.

  • Federica Riguzzi

    I suggest as additional background reading
    Riguzzi, F., Panza, G., Varga, P., Doglioni, C., 2010. Can Earth’s rotation and tidal despinning
    drive plate tectonics? Tectonophysics 484 (1–4), 60–73.
    We have already shown In 2010 that the time series of LOD and N (#large earthquakes) have significant peaks centered at 34 yr and the correlation coherence between the two is significant.

  • Federica Riguzzi

    I suggest as additional background reading
    Riguzzi, F., Panza, G., Varga, P., Doglioni, C., 2010. Can Earth’s rotation and tidal despinning
    drive plate tectonics? Tectonophysics 484 (1–4), 60–73.
    in which we have already shown the correlation and the spectral coherence between the time series of large earthquakes and the LOD (length of day) with significant peak at 34yr .

  • Roger Bilham

    The graph ends in the year 2020 and plots smoothed seismicity and deceleration data to 1 January 2017. The five year moving average means that the last two points are extrapolated. For this reason phase-advanced deceleration data beyond 2020 are omitted.

  • Dal Stanley

    That is a great post-publication discussion of the concept. I didnt really buy the concept while looking at abstract and previous discussions of the paper, but the attached figure to this post (I guess from the GRL article) is pretty convincing alone. And the paper mentioned by RIguzzi below is the real statement and explanation of the process. It is fundamentally the definition of the relationship and the figures are outstanding.
    Riguzzi, F., Panza, G., Varga, P., Doglioni, C., 2010. Can Earth’s rotation and tidal despinning
    drive plate tectonics? Tectonophysics 484 (1–4), 60–73.

  • Armen Kazarian

    This is was very interesting reading. And I have a question to authors. Why you do emphasize that Earth’s rotation does not cause earthquakes when you actually presenting very strong connection between earth rotation speed and the amount of earthquake? What is holding you from assumption that it does?

  • Sharwan Kumar

    Could you please send me a copy of your said paper ?

  • Sharwan Kumar

    Mr Roger , could you please send me a copy of your said paper ?

  • Angela Stallone

    This is an interesting work and I think it can promotes an exciting discussion. The first comment I have is: earthquakes are complex phenomena resulting from many different processes, hence I find it difficult to apply the concept of “synchronization” to their occurrence. There are so many factors playing a role into the earthquake occurrence – of much higher orders of importance than Earth’s rotation – which set the problem too far away from the pendulum clock example. Moreover, you write: “In our analysis, a Mw=9 earthquake is the same as a Mw=7.0 earthquake.”. I wonder whether this is valid for M<7 earthquakes as well and, if not, what would be the reason.

  • Roger Bilham

    Armen,
    Imagine a grocer weighing 1 kg of coffee beans on an old-fashioned balance scale. The bag weighs 0.99 kg and it tilts clockwise because the 1 kg mass is heavier than the bag of coffee. He now adds one more bean to the bag and the scale tips anticlockwise. The single bean causes the balance to shift. So it is with earthquakes. A tiny strain increment can influence the balance between stability and rupture, but it takes a whole lot of strain to rupture a plate boundary.
    RB