TUESDAY, 22 JUNE 2021It is a common misconception that seismologists can predict earthquakes. In the Italian town of L'Aquila in 2009, following a magnitude 4 earthquake, a group of government-appointed scientists claimed that there was a low probability of a larger earthquake occurring. Miscommunication with local politicians led to newspaper headlines stating that there was ‘no danger’. Only a week later, a magnitude 6 earthquake occurred in the middle of the night, killing 29 people. Public outcry led to seven of the scientists being convicted of manslaughter in 2012. The justification: that they communicated the earthquake hazard ineffectively and didn’t tell residents to leave their houses. An appeal was successful in 2014 after scientists across the world highlighted the limits of science's ability to predict earthquakes. This case study demonstrates the importance of clear scientific communication, which ensures that citizens are well-informed of any potential seismic hazard, but are also aware of the uncertainties.
Due to our limited knowledge of previous events and plate motions, the exact time and place of earthquakes cannot be predicted. Scientists can only make educated guesses as to where the next big one will be. This is because it is not possible to directly look into the Earth to measure forces building up, nor measure frictional properties of the fault planes where earthquakes are hosted. During my MSci research in the Cambridge Department of Earth Sciences, I have explored how remote sensing techniques can elucidate the nature of earthquakes.
How Does Remote Sensing Improve Seismic Hazard?
In general, the largest and most common earthquakes occur where two major plates grind past each other. In 2011, the magnitude 9 Tohoku-Oki earthquake in Japan, well-known for the resulting tsunami and Fukushima nuclear disaster, occurred as the Pacific Plate subducts under Japan. Whereas, intraplate earthquakes occur on weak fault planes within the large plates, away from major boundaries. Stresses slowly build-up on these fault planes over time, due to immense forces at plate boundaries. These are then released in smaller magnitude earthquakes, when a critical breaking stress is reached.
Technological advances mean that orbiting satellites can now measure deformations of the surface of the Earth with unprecedented detail. Synthetic Aperture Radar Interferometry (InSAR) can show how a fault plane ruptured at the surface with millimetre-scale accuracy, as well as estimate how the fault moved at depth. In addition, the Global Positioning System (GPS) shows how a large region is deforming in response to tectonic forces, including estimating the strain accumulating on individual faults. High resolution Digital Elevation Models (DEMs) and satellite images allow large faults to be remotely located in the landscape. These data can be interpreted to create better forecasts of where the deadliest earthquakes are likely to occur in tectonic settings around the world. With this information, scientists can create more accurate seismic hazard maps, inform policymakers on where the strictest building regulations should be, and how the public should take precautions for a large earthquake.
Investigating the Mochiyama Fault From the Other Side of the World
On a smaller scale, my research investigates fault behaviour. I focus on the intraplate Mochiyama Fault in Japan, which intriguingly ruptured twice in an extremely short period of time: firstly a magnitude 6 earthquake in March 2011, and then another magnitude 6 event in December 2016. I estimated the maximum stress that may have built up on the Mochiyama Fault from 2011 to 2016 using InSAR, GPS and microseismicity data, and modelling simple fault plane processes. I found that the stress built up over this time was less than half the magnitude of that which caused the initial earthquake. This makes it unlikely that the fault reached the same critical breaking stress as before, suggesting that the fault somehow became weaker over time.
‘I think the most important outcome from your work is that the yield strength of the Mochiyama fault may have decreased between the two earthquakes. This would be one of the first inferences that a fault has changed strength over the time-scale of a few years,’ Dr Sam Wimpenny, my MSci project supervisor, writes to me. Many seismic hazard models assume that strain accumulates slowly on faults, and that the repeat time between them is on the order of hundreds of years. My unique case study suggests an alternative; that faults can rupture again only a few years after a previous major event. My research demonstrates how increased accuracy of remote sensing techniques increases our understanding of fundamental processes occurring on fault planes.
Limits of Remote Sensing in Reducing Seismic Risk
Space-based monitoring is biased towards seismic activity which occurs on land. This is primarily due to the difficulties of installing permanent seismic and GPS stations on the seafloor, and transmitting signals through the dense medium of water. This issue is starting to be resolved with new technologies, such as underwater GPS. A group of scientists in New Zealand have been using modern equipment to characterise how the large offshore faults move in earthquakes, which is key to understanding how tsunamis form. This is particularly important following the 2011 Tohoku earthquake, because the tsunami defences along the coast of Japan were insufficient, as demonstrated by the Fukushima nuclear disaster. With better characterisation of how these faults move, the immense potential for loss and damage in future massive tsunamis across the world can be greatly reduced through preparation strategies.
The fastest deforming regions, where surface deformation is easiest to measure with satellites, are usually the areas which experience large earthquakes anyway. This leads to a well-prepared population, such as in Japan or California. In contrast, the Alpine-Himalayan mountain range is a widely-deforming region, and can host large earthquakes across a broad area. These earthquakes occur much less often, meaning the population is often less prepared. An example of this is the devastating magnitude 6.5 earthquake in 2003, in the historic city of Bam, Iran. Historic mud-brick buildings collapsed, causing over 25,000 deaths as the city slept. The earthquake occurred on an unknown fault with no obvious expression in the landscape. This is because the repeat time of earthquakes on this fault is likely to be much longer than erosional processes occurring in the region. In addition, the GPS strain measured in this area would likely be very small, in comparison to Japan, and therefore it would be difficult to suggest whether enough stress has accumulated on a fault plane in order for it to rupture.
What Can We Do to Reduce Seismic Risk?
A great deal of work is put into finding which faults have the greatest probability of breaking in devastating earthquakes. That research is often well-communicated through seismic hazard maps. Armed with this knowledge, local governments and residents should take necessary precautions to protect themselves and prevent widespread damage. Investment should be made to ensure all buildings are prepared for substantial shaking. My research into the short recurrence time between the two earthquakes in Japan has demonstrated the unpredictability of large earthquakes. There is only so much that earthquake science, through both remote sensing and field-based techniques, can tell us about how to reduce seismic risk. The impact of earthquakes is well-known, and focus should always be on preparation rather than short-term prediction.
Natalie Forrest is a fourth year undergraduate student at Corpus Christi College studying Earth Sciences. Artwork by Rianna Man.