Forecast of Increased Earthquakes due to Slowing of Earth’s Rotation

Paul Somerville, Risk Frontiers.

In the past few weeks there have been sensational reports about a forecast accelerated rate of occurrence of large earthquakes in 2018.  Fortunately, one of the authors of the work that lies behind these reports has explained her calm view of the situation.  The following article, written by Sarah Kaplan, appeared in the Washington Post, last updated 22 November 2017.

Rebecca Bendick would like you to not panic. The University of Montana geophysicist knows you may have read the articles warning about “swarms of devastating earthquakes” that will allegedly rock the planet next year thanks to a slowdown of the Earth’s rotation. And she feels “very awful” if you’ve been alarmed. Those dire threats are based on Bendick’s research into patterns that might predict earthquakes – but claims of an impending “earthquake boom” are mostly sensationalism.

There is no way to predict an individual earthquake. Earthquakes occur when potential energy stored along cracks in the planet’s crust gets released, sending seismic waves through the Earth.  Since scientists know where those cracks exist, and how they are likely to convulse, they can develop forecasts of the general threat for an area. But the forces that contribute to this energy buildup and trigger its release are global and complex, and we still cannot sort out exactly how it might unfold.

In a paper published in August in the journal Geophysical Research Letters, Bendick and colleague Roger Bilham, a geophysicist at the University of Colorado, Boulder, did find a curious correlation between clusters of certain earthquakes and periodic fluctuations in the Earth’s rotation. By examining the historic earthquake record and monitoring those fluctuations, scientists might be able to forecast years when earthquakes are more likely to occur, they suggest.

“Something that people have always hoped to find . . . is some kind of a leading indicator for seismicity, because that gives us a warning about these events,” Bendick said. But that conclusion is by no means set in stone. It hasn’t been demonstrated in the lab or confirmed by follow-up studies. Several scientists have said they’re not yet convinced by Bendick’s and Bilham’s research. “The main thing I came away thinking was real old-fashioned scientific ‘let’s check this’ kind of thoughts,” research geophysicist Ken Hudnut told Popular Science. Hudnut, who works on earthquake-risk programs at the US Geological Survey, was not involved in the paper. And that reaction is okay with Bendick. That’s how these things are supposed to go: “Someone says something kind of marginally outlandish, and everyone checks their work and that’s how science progresses,” she said.

Historically, the field of earthquake forecasting has seen some particularly outlandish claims. People have tried to predict temblors based on the behaviour of animals, gas emissions from rocks, low-frequency electric signals rippling through the Earth – all without much success.  For that reason, Bendick said, “it’s a little bit scary to get into the game.” But getting a prediction right can mean the difference between life and death for countless people. The stakes are too high not to try.

For their recent paper, she and Bilham looked through the century-long global earthquake record to see if they could spot any signs that temblors around the world are linked. Initially, the data appeared completely random. But then Bendick and Bilham added a new number to their analysis: the “renewal interval,” or the amount of time a given earthquake zone requires to build up potential energy for a really big quake. “Basically you can think of earthquakes as something like a battery or a neuron; they have a certain amount of time they need to be charged up,” Bendick said.

A certain class of earthquakes – those with a magnitude of 7.0 or more, and a short renewal interval between 20 and 70 years – seemed to cluster in the historic record. Every three decades or so, the planet seemed to experience a bunch of them – as many as 20 per year, instead of the typical 8 to 10. It was as if something was causing the earthquakes to synchronise, even though they were happening in spots scattered around the globe. Contrary to some reports on the study, “it’s not exactly the case that every 32 years we have a bad patch,” Bendick said. “If it were that, people would have found [the pattern] ages ago. That would be super obvious in the record.” Instead, she explained, “events with that renewal interval happen together more often than they happen at random, and that pattern is statistically significant.” Sure, it’s a less flashy finding than, “we know when earthquakes will happen,” she acknowledged. But that’s geophysics for you. “We’re scientists, not magicians,” she said.

Next, Bendick and Bilham tried to figure out what mechanism might explain these earthquake clusters. They studied a wide range of global phenomena that unfold over the same time scales: sloshing of the molten rock in the mantle, ocean circulation changes, momentum transfer between the Earth’s core and the lithosphere (the planet’s solid, outermost shell).

The best fit were tiny, cyclical changes in the speed of the Earth’s rotation. The planet slows down infinitesimally every 30 years or so, and roughly five years later, a cluster of these severe, short-interval earthquakes appears. Russian geophysicists Boris Levin and Elena Sasorova have pointed out this correlation before, Bendick noted. So she and Bilham tried to take it a step further: They found a mechanism that might link the Earth’s rotation and clusters of quakes.

See, when the Earth’s rotation rate changes, its shape shifts. As the planet speeds up, mass moves toward the equator, much the way a dancer’s skirt flares out when she spins. When it slows, that mass shifts back toward the poles. The cumulative effect is tiny – a millimetre difference in the width of the globe. But if potential energy has already built up at a number of faults – “if they’re locked and loaded, as we’d say in Montana,” Bendick noted – “that tiny change is enough to kick some proportion of the faults over into their failure mode, which is earthquakes.”

Earth is currently at the end of a slowing period, Bendick pointed out, and the historic record would indicate another “cluster” may be on its way. She and Bilham hope the pattern might help scientists and public officials make some sense of the Earth’s unpredictable shaking. If disaster planners can say with some assurance that the planet is entering a period in which quakes are more likely, they might have an easier time making the case for preparedness measures.

But that doesn’t necessarily mean 2018 will be a particularly devastating year. For one thing, the kinds of temblors Bendick and Bilham analysed happen in areas that are already earthquake-prone – Japan, New Zealand, the west coast of the United States. For people who live in those regions, there is always a risk of a quake, and it is always good to be prepared.

Their study is about probabilities, not predictions, Bendick cautioned. Earth’s slowing does not mean that a quake will happen in the next year or so, just that the likelihood may have gone up. Moreover, this pattern of earthquake occurrence is definitely not the only factor influencing the Earth’s behaviour – if it were, scientists would have noticed the pattern a long time ago. There are doubtless other earthquake cycles on the planet, driven by phenomena not considered in the paper.

The research got a lot of attention after Bilham presented it at the October meeting of the Geological Society of America. Several critics noted that correlation is not causation – earthquake clusters and fluctuations of Earth’s rotation might happen on the same time scales, but that does not mean they are linked. Bendick acknowledged that there is less evidence for the proposed mechanism than for the pattern itself. But she’s confident the pattern is there. “I think this is likely to inspire many people to look at this pattern, and it’s possibly someone will come up with an even better explanation,” she said.

Notes by Paul Somerville

The following is excerpted from the abstract of Bilham and Bendick (2017).

On five occasions in the past century a 25-30% increase in annual numbers of Mw≥7 earthquakes has coincided with a slowing in the mean rotation velocity of the Earth, with a corresponding decrease at times when the length-of-day (LoD) is short. The correlation between Earth’s angular deceleration (d[LoD]/dt) and global seismic productivity is yet more striking, and can be shown to precede seismicity by 5-6 years, permitting societies at risk from earthquakes an unexpected glimpse of future seismic hazard.

The cause of Earth’s variable rotation is the exchange of angular momentum between the solid and fluid Earth (atmospheres, oceans and outer core). Maximum LoD is preceded by an angular deceleration of the Earth by 6-8 years. We show delayed (increase in) global seismic productivity is most pronounced at equatorial latitudes 10°N-30°S.

The observed relationship is unable to indicate precisely when and where these future earthquakes will occur, although we note that most of the additional Mw>7 earthquakes have historically occurred near the equator in the West and East Indies. A striking example is that since 1900 more than 80% of all M≥7 earthquakes on the eastern Caribbean plate boundary have occurred 5 years following a maximum deceleration (including the 2010 Haiti earthquake).

The 5-6 year advanced warning of increased seismic hazards afforded by the first derivative of the LoD is fortuitous, and has utility in disaster planning. The year 2017 marks six years following a deceleration episode that commenced in 2011, suggesting that the world has now entered a period of enhanced global seismic productivity with a duration of at least five years.

The correlation between the change in Earth’s rotation rate and the frequency of Mw>7 earthquakes from Bendick and Bilham (2017) is shown in Figure 1.  I have not seen the Bilham and Bendick (2017) presentation.

Figure 1. Changes in the length of the day correlate with decadal fluctuations in annual M ≥ 7 earthquakes, smoothed with 10 year running mean. Peak seismic activity and rotational acceleration occur at 15, 33, 60, and 88 year intervals. Source: Bendick and Bilham, 2017.


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 (2017). A five year forecast for increased global seismic hazard.  Invited presentation, Geological Society of America Meeting, Seattle, Washington.

Victoria on alert for worst floods in over 20 years

This article by Anna Prytz was published in today’s issue of The Age.

Heavy rain is forecast to arrive in Melbourne on Friday. Photo: AAP

Record-breaking rain is bearing down on Victoria, triggering warnings of dangerous flash flooding across the state.

After a scorching end to spring, Melbourne is set to get one month’s worth of rain in just two days, and possibly an entire summer’s worth of rain in the season’s first three days.

A severe weather warning has been issued for all of Victoria as the Bureau of Meteorology prepares for what could be the state’s most significant rain event in over 20 years.  Read more:

Changes in Earthquake Hazard Levels in the draft Geoscience Australia National Seismic Hazard Assessment (NSHA18)

Paul Somerville, Risk Frontiers


Geoscience Australia (GA) has embarked on a project to update the seismic hazard model for Australia through the National Seismic Hazard Assessment (NSHA18) project.  The following information is excerpted from Allen et al. (2017) and from discussions that took place at the Annual Conference of the Australian Earthquake Engineering Society (AEES) in Canberra, November 24-26, 2017 and a pre-conference workshop organised by GA on the NSHA18 project held on November 23.

The draft NSHA18 update yields many important advances on its predecessors, including:

  1. calculation in a full probabilistic framework using the Global Earthquake Model’s OpenQuake-engine;
  2. consistent expression of earthquake magnitudes in terms of moment magnitude, Mw;
  3. inclusion of epistemic uncertainty through the use of alternative source models;
  4. inclusion of a national fault-source model based on the Australian Neotectonic Features database;
  5. the use of modern ground-motion models; and
  6. inclusion of epistemic uncertainty on seismic source models, ground-motion models and fault occurrence and earthquake clustering models.

The draft NSHA18 seismic design ground motions are significantly lower than those in the current (1991-era) Standards Australia AS1170.4:2007 hazard map at the 1/500-year annual ground-motion exceedance probability (AEP) level. The large reduction in seismic hazard at the 1/500-year AEP level has led engineering design professionals to question whether the new draft design values will provide enough structural resilience to potential seismic loads from rare large earthquakes. These professionals are planning to use a seismic design factor of 0.08g as a minimum design level for the revised AS1170.4 standard, due to be released in 2018, and are discussing the idea of transitioning to a 1/2475-year AEP in the longer term, consistent with the trend in other countries including Canada and the United States.

The primary reason for the significant drop in seismic hazard is due to adjustments to earthquake catalogue magnitudes. Firstly, prior to the early 1990’s, most Australian seismic observatories relied on the Richter (1935) local magnitude (ML) formula developed for southern California. At regional distances (where many earthquakes are recorded), the Richter scale will tend to overestimate ML relative to modern Australian magnitude formulae. Because of the likely overestimation of local magnitudes for Australian earthquakes recorded at regional distances, there is a need to account for pre-1990 magnitude estimates due to the use of inappropriate Californian magnitude formulae. A process was employed that systematically corrected local magnitudes using the difference between the original (inappropriate) magnitude formula (e.g., Richter, 1935) and the Australian-specific correction curves (e.g., Michael-Leiba and Malafant, 1992) at a distance determined by the nearest recording station likely to have recorded a specific earthquake (Allen, 2010).

Another important factor determining the reduction in hazard is the conversion of catalogue magnitudes such that magnitudes are consistently expressed in terms of moment magnitude, MW. Moment magnitude is the preferred magnitude type for probabilistic seismic hazard analyses (PSHAs), and all modern ground-motion models (GMMs) are calibrated to this magnitude type. Relationships between MW and other magnitude types were developed for the NSHA18. The most important of these is the relationship between ML and MW because of the abundance of local magnitudes in the Australian earthquake catalogue. The preferred bi-linear relationship demonstrates that MW is approximately 0.3 magnitude units lower than ML for moderate-to-large earthquakes (4.0 < MW < 6.0). Together, the ML corrections and the subsequent conversions to MW effectively halve the number (and subsequently the annual rate) of earthquakes exceeding magnitude 4.0 and 5.0, respectively. This has downstream effects on hazard calculations when forecasting the rate of rare large earthquakes using Gutenberg-Richter magnitude-frequency distributions in PSHA.

The secondary effect of the ML and MW magnitude conversion is that it tends to increase the number of small and moderate-sized earthquakes relative to large earthquakes. This increases the Gutenberg–Richter b-value, which in turn further decreases the relative annual rates of larger potentially damaging earthquakes (Allen et al., 2017).

The final main factor driving the reduction of calculated seismic hazard in Australia is the use of modern ground motion models (GMMs). While seismologists in stable continental regions (SCRs) worldwide recognise the complexity in characterising the likely ground motions from rare large earthquakes, more abundant ground-motion datasets of moderate-magnitude earthquakes are emerging. The NSHA18 hazard values are based on modern GMMs with improved understanding of instrumental ground-motion source amplitudes and attenuation in Australia and analogue regions. The peak ground accelerations (PGAs) predicted by these modern models in general are up to a factor of two lower than the Gaull et al. (1990) peak ground velocity (PGV)-based relationships at distances of engineering significance (generally less than 100 km). At larger distances, the lower rates of attenuation of the Gaull et al. (1990) relationships yield ground-motion values up to factors of 10 higher than modern GMMs (Allen et al., 2017).

It is anticipated that the National Seismic Hazard Assessment (NSHA18) project will be complete in mid-2018, at which time Geoscience Australia has agreed in principle to provide a briefing on it in Sydney for the insurance Industry.    The updated version of AS1170.4 will be released in 2018.


Allen, T., J. Griffin, M. Leonard, D. Clark and H. Ghasemi (2017). An updated National Seismic Hazard Assessment for Australia: Are we designing for the right earthquakes? Proceedings of the Annual Conference of the Australian Earthquake Engineering Society in Canberra, November 24-26, 2017.

Standards Australia (2007). Structural Design Actions, Part 4 Earthquake Actions in Australia. AS1170.4:2007.

The Mw 7.1 Puebla, Mexico Earthquake of 19 September 2017 – the anniversary of the Mw 8.0 Michoacan earthquake of 1985.

By Paul Somerville, Risk Frontiers

As reported by the USGS, the September 19, 2017, Mw 7.1 Puebla earthquake in Central Mexico occurred as the result of faulting within the subducted Cocos plate at a depth of approximately 50 km and about 120 km southeast of Mexico City. At least 220 people were killed at Mexico City, 74 in Morelos, 45 in Puebla, 13 in Estado de Mexico, 6 in Guerrero and 4 in Oaxaca. At least 6,000 people were injured. At least 44 buildings collapsed and many others were damaged at Mexico City. Many other buildings were damaged or destroyed in the surrounding area. Significant damage occurred to the electrical grid in Estado de Mexico, Guerrero, Mexico City, Morelos, Oaxaca, Puebla and Tlaxcala.

This earthquake occurred on the anniversary of the devastating Mw 8.0 Michoacan earthquake of 19 September 1985, which caused extensive damage to Mexico City and the surrounding region. That event occurred as the result of thrust faulting on the plate interface between the Cocos and North America plates, about 450 km to the west of the September 19, 2017 earthquake.

Most of Mexico City is founded on a clay-filled lake. The clay has a resonant period of 1 to 2 seconds and has very unusual properties – it is very elastic (has low damping), which allows a very large resonance to build up due to the trapping of energy within this shallow sedimentary basin (Figures 1 and 2).  This resonance caused the collapse of buildings, especially ones having natural periods of 1 to 2 seconds, and generated a seiche in Lake Chapultepec (part of the original lake that has not been filled in) seen in a widely viewed video, in which the waves have a period of about 2 seconds.

The 1 to 2 second resonance of the lakebed can also be set up by marching soldiers.  This occurred exactly 33 years earlier to the day, when I was on holiday in Mexico City.  It was September 19, Independence Day, and the soldiers were marching down Reforma Avenue.  I was standing on the roof of my ten story hotel, which was swaying noticeably.  One year to the day later, at 07:17 am on 19 September 1985, the Mw 8.0 Michoacan earthquake occurred.  I doubt that my hotel  survived the earthquake.

After the 1985 earthquake I spoke with my colleague, Lloyd Cluff, who had been at a meeting with Mexican government officials on the day of the earthquake to discuss seismic issues for nuclear power plants.  The meeting was held on the edge of Mexico City outside the lakebed area (blue area of Figure 1).  After he returned to his hotel that evening he turned on the TV and saw photos of a disastrous earthquake.  It took him some time to recognise the scene of the disaster as Mexico City.  No one at the meeting had known that it had occurred early that morning in Mexico City, because the shaking outside the lakebed area had been so weak.

Figure 1. Response spectral accelerations at 1 second period in Mexico City from the Mw 7.1 Puebla, Mexico earthquake of 9 September 2017. The largest ground motions, shown in the red and yellow colours, occurred in the parts of the city founded on the lakebed. Source: UNAM.
Figure 2. Locations of damaged buildings in Mexico City from the 2017 Puebla earthquake showing correlation with the western part of the ground motion map in Figure 1.