The dangers of rolling boulders

Field work in lonely and spectacular places is a privilege. Though it can be great, boredom sometimes sets in, which is hard for the lone geologist. Today, I guess a cell phone would help, especially in high places where the signal is good. That means of communication and entertainment only emerged in the 1980s and did not reach wild places until well into the 90s. Pre-cellnet boredom could be relieved by what remains a dark secret: lone geologists once rolled large boulders down mountains and valley sides, shouting ‘Below!’ as a warning to others. Their excuse to themselves for this unique thrill (bounding boulders reach speeds of up to 40 m s-1) was vaguely scientific: sooner or later a precarious rock would fall anyway. This week it emerged that Andrin Caviezel of the Institute for Snow and Avalanche Research in Davos, Switzerland, an Alpine geoscientist, rolls boulders for a living (Caviezel, A. 2022. The gravity of rockfalls. Where I work, Nature, v. 607, p. 838; DOI: 10.1038/d41586-022-02044-9). He finds that ‘…flinging giant objects down a mountain is still super fun’. The serious part of his job attempts to model how rockfalls actually move downslope, as an aid to risk assessment (Caviezel, A. and 23 others 2021. The relevance of rock shape over mass – implications for rockfall hazard assessments. Nature Communications, v. 12, article 5546; DOI: 10.1038/s41467-021-25794-y)

Caviezel’s team (@teamcaviezel) don’t use actual rocks but garishly painted, symmetrical blocks of reinforced concrete weighing up to 3 tonnes, which are more durable than most outcropping rock and can be re-used. A Super Puma helicopter shifts a block to the top of a slope, from which it is levered over the edge (watch video). The team deploys two types of block, one equant and resembling a giant garnet crystal, the other wheel-shaped with facets. The first represents boulders of rock types with uniform properties throughout, such as granite. The wheel type mimics boulders formed from rocks that are bedded or foliated, which are usually plate-like or spindly.

Vertical aerial photograph of a uniform, south-facing slope in the Swiss Alps used to roll concrete ‘boulders’. The red X marks the release point; the blue symbols show the points of rest of equant ‘boulders, the sizes of which are shown in the inset, the wheel-shaped ones are magenta. Coloured circles with crosses show the mean rest position of each category (the lighter the colour the smaller the set of ‘boulders’). The coloured ellipses indicate the standard deviation for each category. (Credit: Caviezel et al., Fig 2)

Unlike other gravity-driven hazards, such as avalanches and mudflows, the directions that rockfalls may follow by are impossible to predict. Rather than hugging the surface, boulders interact with it, bouncing and being deflected, and they spin rapidly. To follow each experiment’s trajectory a block contains a motion sensor, measuring speed and acceleration, and a gyroscope that shows rotation, wobbling and motion direction, while filming records jump heights – up to 11 m in the experiments. Despite the similarity of the blocks, the same release point for each roll and a uniform mountainside slope, with one cliff line, the final resting places are widely spread. That hazard zone of rockfalls is distinctly wider than that of snow avalanches; observing a boulder once it starts to move gives a potential victim little means of knowing a safe place to shelter.

The most important conclusion from the experiments is that the widest spread of tumbling ‘boulders’ is shown by the wheel-shaped ones. So, slopes made from bedded or foliated sedimentary and metamorphic rocks may pose wider hazards from rockfalls than do those underpinned by uniform rocks. However, plate-like or spindly boulders are more stable at rest than are equant ones. Yet boulders rarely fall as a result of being pushed (except in avalanches). On moderate slopes they are undermined by erosion, and on steep slopes or cliffs winter ice wedges open joints allowing blocks to fall during a thaw.

The Sendai great earthquake in close retrospect

Sendai Airport
Tsunami debris at Sendai airport

Media coverage of the disasters following the magnitude 9.0 earthquake of 11 March 2011  that devastated the  north-eastern coast of Honshu, Japan around the city of Sendai is now (early May) fitful and dominated by the aftermath of the tsunamis’ effect on the Fukushima Daiichi nuclear power station. For those who escaped the tsunamis the experience is irredeemably seared on their memory. Unlike the great waves that killed 10 times more people around the Indian Ocean on 26 December 2004, it will also be unforgettable for those of us far from the event who witnessed the lengthy, high-definition footage captured during the black-water torrents that swept all before them far inland. But that is no longer ‘news’…

Only 6 to 7 weeks later lessons are being learned that probably should have been anticipated long before. Japan has the world’s best disaster preparedness systems. They are centred on civil engineering that was proven to resist great earthquakes by that of 11 March; the terrifying tremors resulted in far fewer casualties than would have been the case anywhere else under such conditions. The tragedy lay with the magnitude of the tsunamis – as high as 30 m in some areas – that reached the coast within an hour of the seismic event. As well as the devastation and loss of life along the coast and up fertile low-lying valleys, waves of this size swept over defences of the coastal Fukushima Daiichi nuclear power plant cutting off emergency power supplies: the world’s largest tsunami barriers proved inadequate to the task and near-meltdown ensued.

Despite the densest network of seismometers anywhere and in-place earthquake early-warning  and risk-assessment systems, the events were not forecast and the only warning was that of the earthquake itself which alerted a well-versed population to the imminence of tsunamis to follow. Public education and preparedness proved to be the major life saver, except of course for those tragically killed or lost without trace. So what went wrong?

The risk assessment and warning systems produced results that bore little relation to the actual seismic shaking; the warning was for the immediate vicinity of Sendai city to experience the highest intensities (5-6), most of the rest of Honshu, including Tokyo, having expected intensities in the 2-4 range. For Fukushima Daiichi a maximum magnitude of 7.2 in its vicinity was predicted to have less than 10% chance of occurring over the next 50 years. In reality seismometers across the whole eastern part of the Honshu north of Tokyo recorded intensities between 5-7, demonstrated graphically by numerous CCT recordings in shops and offices. The emerging opinion is that the theory and historic data used for risk and warning systems are flawed or inadequate. For instance the earthquake ripped along 400 km of the Japan Trench subduction zone rather than being a point source – a lesson also from the Sumatra earthquake of 26 December 2004, when ocean-floor thrusting extended 1200 km northwards to the Andaman Islands. Great earthquakes are far too infrequent for sufficient modern-style seismic data to have been collected for previous cases in the 20th century, but it seems clear since 2004 that: (1) stresses accumulate to unexpectedly high values where opposed plates are coupled or stuck together; (2) the ‘point-source’ model for earthquakes, which the use of seismic focuses and epicentres pinpointed by the world-wide seismic network encourages, is far from reality, the more so for the biggest stress accumulations; (3) existing approaches will fail for events with magnitudes greater than 8.0.

Sendai tsunami model - NOAA via Google Maps
NOAA's Prediction of 11 March tsunami wave heights across Pacific Ocean. Image by cogito ergo imago via Flickr

Part of the problem is the sparse record of great earthquakes and the likelihood that, if they do have cyclicity, it may be of the order of hundreds to thousands of years. Historical sources record a large earthquake and tsunamis affecting Sendai district in 869 CE (Common Era), confirmed recently by geologists having located a typical tsunami deposit extending 3-4 km up the Sendai Plain, compared with more than 5 km in March 2011. The survey team claimed at the time that their discovery might indicate far higher risk now in the area than modelled ‘officially’. Sadly, evaluating the prediction was incomplete when disaster did strike. Geoscientists can map faults, infer the length of their activity and work out the mechanisms whereby they fail, but apart from historical data – often sketchy – pinpointing and quantifying past events is beyond us, Looking at more widespread secondary effects, tsunami deposits in particular that often contain dateable organic debris, seems a fruitful way forward for coastal areas likely to bear the brunt of both shaking and huge inundations and the powerful ebbing of their flood waters. That is a topic in its infancy, but likely now to burgeon.

Ominously, because great earthquakes are so rare along any plate boundary, for seven greater than magnitude 8 to occur worldwide in a matter of 6 years (Sumatra, 2004, 9.1, 2005, 8.8, and three with magnitude >7 in 2010; Kuril Islands, 2006, 8.3, 2007, 8.1; Sichuan, 2008, 8.0; Chile, 2010, 8.8; Japan, 2011, 9.0) raises the questions, do they occur in time clusters, and if so, why? Although the numbers are small enough to strain statistics, comparing the last six years with the previous century or so of seismometer recordings shows that great earthquakes have never occurred so frequently. Is there a domino effect so that, say, energy from the Sumatran earthquake of late 2004 has somehow been transmitted throughout the interconnected subduction-zone system to destabilise other highly stressed areas? It is widely acknowledged that in one subduction system there is evidence of clustering, and this may extend to the two great earthquakes (2006 and 2007) in the Kuril Islands on the same boundary as the Sendai event, and two off Sumatra (2004 and 2005) with three more with magnitude >7 in 2010 on what previously had been regarded as a relatively quiescent subduction zone. Analysing all recorded seismic events greater than magnitude 5 to improve the statistics suggests that clustering does not extend to global scales, yet great earthquakes buck other trends shown by lesser ones. Their motions both vertical and lateral could conceivably cause widespread destabilisation, yet worryingly the only test of the idea is the occurrence of yet more in the next few years.

Sources: Normile, D. et al. 2011. Devastating earthquake defied expectations. Science, v. 331, p. 1375-1376; Brahic, C. et al. 2011. Megaquake aftermath. New Scientist, v. 209 (19 March 2011), p. 6-8; Cyranoski, D. Japan faces up to failure of its earthquake preparations. Nature, v. 471, p. 556-557; Normile, D. 2011. Scientific consensus on great quake came too late. Science, v. 332, p. 22-23.

See also: Geller, R.J. 2011. Shake-up time for Japanese seismology. Nature, v. 472, p. 407-409; Kerr, R.A. 2011. New Work reinforces megaquake’s harsh lessons in geoscience. Science, v. 332, p. 911

YouTube video: