In light of recent natural disasters, such as the devastation in Mexico, could new technology to forecast and warn about future earthquakes help lessen the catastrophic damage we’ve seen around the world?
Rolling global news, such as the tremendous earthquake in Mexico and back-to-back hurricanes around the Caribbean can leave the impression that the frequency and ferocity of geophysical hazards has increased in recent years as natural disasters have brought devastation around the world.
Earthquakes are the biggest natural nasties in terms of direct impacts on humankind and outbursts of seismic energy now pose a threat to around one third of the planet’s population.The threat is on the increase due to unrelenting population encroachment upon seismically-unstable lands.
The 2017 Atlas of the Human Planet - the European Commission’s index of Global Exposure to Natural Hazards - reports the number of people who live in areas prone to destructive earthquakes increased by 93 per cent in the 40 years between 1975 and 2015 (from 1.4bn to 2.7bn). The disastrous impact of tsunami-triggering quakes has added to the breadth of the peril.
In addition to loss of life, the financial turmoil that earthquakes cause is also escalating. This would, of course, run into many billions when a seismic event strikes a developed country, but the impact on developing nations that have been striving to establish first-generation critical infrastructure to take their economies forward is doubly damaging. It is widely recognised that a significant degree of the damage earthquakes cause could be avoided given better forecasting and warning.
Most conventional earthquake public warning systems - delivering their alerts via a range of media (including radio flashes, SMS text messages, emails and public sirens) - give best endeavours notice period of around 60 to 90 seconds, depending on the distance of the earthquake hypocentre to the strike zone, travel speed and distance between the P-waves and S-waves, and effectiveness of the type of relay technology used (assuming the intended recipient catches the alert).
The potential benefits of automated earthquake warning alerts that connect to, and rapidly de-activate, industrial processes - machinery such as that found in factories and plants - as well as automated infrastructure like transport, is generating heightened interest in California, as the state waits for the next predicted seismic convulsion. More of its built environment acquires smart management systems either as part of their innate design or as retrofitted to computer-controlled mechanical systems, such as lifts and automated doors. Widespread automated earthquake warning response is in its early stages, but could prove as effective in saving lines as alarm systems that urge building evacuation, says Joshua Bashioum, manager and principal investigator at Early Warning Labs.
“It is really important to automate response for earthquake early warning systems,” says Bashioum. “Fire engines get stuck inside a station when the earthquake shake jams the bay doors shut.” Basic measures to ensure that doors open automatically when the first tremors are felt could mean emergency services are able to respond faster, rather than having to waste time trying to release their equipment from its base.
Earthquake-savvy ‘smart sensors’ that can be installed directly on the machinery itself and can close-down production lines and other potentially hazardous equipment when seismic vibration is detected – such as the Omron D7S Seismic Sensor – have come onto the market. Launched earlier this year, the D7S can also help determine post-quake damage by mapping seismic intensity and providing information on risk of building collapse.
“The D7S sensor has algorithm technology that gives high-precision measurement of spectral intensity, allowing the sensor to reject impulse vibration noise and respond only to genuine seismic activity,” says Gabriele Fulco, product manager at Omron. “If a plant happens to be located next to a train line, the sensor uses its three-axis accelerometer and Omron’s spectral-intensity value-calculation algorithms, to distinguish between seismic activity and other movement. It has the same accuracy [as] a seismograph.”
Each D7S features internal memory and I2C (I²C – inter-integrated circuit) interface designed to enable it to be integrated with IoT devices and smart meters, as well as features of the built environment such as bridges and flyovers that might take a hit from earthquake shocks. D7Ss are already used in Japanese buildings to monitor seismic activity and help prevent secondary disasters.
For years, seismological establishment researchers have been at work trying to develop more reliable methods of calculating when an earthquake will happen. Although some of their forecasts have achieved degrees of accuracy, there has been no winning breakthrough technique that some observers feel the discipline should have delivered by now. Localised warning systems, though available in some vulnerable regions, have not acquired the installed base or build-out scale that would enable them to markedly increase likelihood of lives saved in a quake striking a densely populated district.
Alternative approaches to earthquake forecasting and warning - as two separate but related activities - are reliant on developments in information and communications technologies, enabling an area of study that used to be the preserve of seismological researchers to be enfranchised, in theory at least, by other scientific disciplines, along with technology start-ups who see opportunities to use existing communications infrastructure to make earthquake forecast and warning faster and more accurate.
In some cases, it is a matter of using high-speed digital communications to overtake earthquake S-waves (big tremors that cause the worst damage) and get warnings to vulnerable areas before shocks pass through. These would-be solutions are at various points in their journeys to market, but demonstrate the scope of expertise that earthquake forecasting and warning is attracting.
Earthquake alarms that can be fitted to premises to provide some warning of an impending strike, so occupants evacuate to safer surroundings or find cover inside the building, have been available since the 1990s. Some work by detecting earthquake P-waves directly using sensors installed in the ground near a premises, others can make use of signals from P-wave-detectors linked to broadcast warning services or always-on broadband connections. Yet even in seismically-high-risk urban areas, earthquake alarms have to be purchased, and are nowhere near as commonly installed as smoke alarms.
The domestic smoke alarm ideal was the model initially adopted by Mexican earthquake warning device start-up, Grillo. Its prototype device, developed in 2014, aimed to tap into signals emitted by Mexico’s semi-national SASMEX (Seismic Alert System of Mexico) earthquake warning network that had been building-out field sensors across the country since the 2000s, with more than 160 located along the coast and inland. SASMEX alerts are relayed via radio signals through broadcast media, and to purpose-designed receiver units, branded Sarmex. However, these devices were allegedly off-putting in price and could prove be difficult to procure.
Grillo saw an opportunity for a similar, but simpler device that would share attributes of a smoke alarm, and would receive its signal directly from the SASMEX field sensor radio transmission. A combination of technical snags and funding limits caused the company to reconsider its core proposition however, and rethink the concepts around how earthquake data is detected and collated. The Grillo version two ‘sensor station’ combines both P-wave sensor (in the form of an accelerator connected to a microcontroller which relays vibration data) and local alarm in a compact unit, and is intended to be installed in premises that have high-quality broadband connectivity that can be used to transmit warning alerts to other Grillo devices.
Grillo is reportedly embarked on an installation programme across Mexico’s medical centres and educational facilities. The sensor stations send data using Internet of Things technologies, the company says, and this is hosted and processed in cloud services using proprietary algorithms; other information about the how the solution work remains company confidential.
Grillo has also made available an earthquake warning smartphone app freely available, and the potential of the mobile connected devices is another phenomenon that has attracted interest in the development of seismological monitoring and warning systems. Researchers have been exploring possibilities to make smartphones frontline devices in earthquake detection and warning.
The Quake-Catcher Network (QCN), for example, is operated by the Southern California Earthquake Centre and the Incorporated Research Institutions for Seismology. QCN is a collaborative initiative for developing an extended, low-cost strong-motion seismic network by using sensors inside, and attached to, internet-connected computers in local homes and offices. It is based on a distributed computing infrastructure that links volunteer-hosted computers into real-time motion-sensing systems.
The volunteer computers monitor motion with MEMS (micro-electro-mechanical systems) accelerometers, and digitally transmit ‘triggers’ to QCN servers when strong motions are detected. QCN servers analyse signals to determine which represent actual earthquakes and which emanate from other sources of vibrational noise.
Meanwhile, a free app for Google Android smartphones, MyShake, from the Berkeley Seismology Lab, is designed to detect earthquake shaking using accelerometer sensors in smartphones. When shaking detected matches the vibrational profile of a seismic event, the app sends information (anonymised) to a central processing centre that confirms location and magnitude of the possible quake. MyShake also provides users with information about recent earthquakes around the world, along with pertinent global historical earthquakes.
MyShake’s backend system uses past earthquake information to develop a classifier algorithm to identify earthquake shaking on a single phone handset and then communicate with a centralised processing centre, says Richard Allen, director at Berkeley Seismological Laboratory.
At the processing centre, a network detection algorithm confirms that an earthquake is under way. In the lab, cloud-based software constantly reviews all incoming data and, if at least four smartphones have signalled an earthquake-like event, and this represents more than 60 per cent of all phones within a 10-kilometre radius of the epicentre, the program confirms earthquake detection.
Location, origin, time and magnitude of the earthquake are then determined based on multiple triggers from other smartphones on the MyShake network. This information can be used to estimate shaking intensity, and also remaining time until dreaded S-waves reach a given target location. Researchers further cross-check information with the California Integrated Seismic Network, which monitors earth movement all over the state using underground seismometers.
MyShake’s inventors claim it can record magnitude 5.0 earthquakes at distances of 10km or less, and develop an on-phone detection capability to separate earthquakes from other everyday shakes. The system apparently needs at least 300 smartphones within a 110-kilometre-square area in order to have a reasonable estimate of earthquake location, magnitude and origin time – the denser the network, the sooner earthquake can be detected. With a dense enough network, detection, analysis and warning can take ‘less than a second’ according to MyShake’s inventors.
About 50 new users per day download and install the app, which as of July 2017 had 236,912 registered users, says Richard Allen at Berkeley Seismological Laboratory. He adds that “MyShake cannot replace traditional seismic networks, like those run by the US Geological Survey, but we think [it] can make earthquake early warning faster and more accurate in areas that have a traditional seismic network and can provide life-saving early warning in countries that have no seismic [monitoring and warning] networks.”
Also based in California, Early Warning Labs (EWL) is an earthquake early warning technology developer and integrator. It is working geographic information system mapping software and spatial data analytics provider Esri, the US Geological Survey, and other research partners, toward test-completion, integration and delivery of a commercial earthquake early warning solution to consumers, government entities and commercial users for roll-out in 2018.
EWL is cagey about divulging specifics of its earthquake warning system, saying it is developing “a robust cloud server environment to handle low-cost mass distribution of these warnings”, according to Joshua Bashioum at EWL. In addition, it is developing automated response standards and systems that allow public and private sector entities to take ‘pre-defined automated actions to protect lives and assets’ – such as opening bay doors of fire stations located in impact zones when P-waves are picked up.
“We have the capability to deliver earthquake early warnings to the entire US west coast – Washington, Oregon and California,” Bashioum adds. “There are already enterprise partners using EWL’s live [pilot] platform operationally. If there is a large earthquake in California, billions of dollars of infrastructure and millions of lives would be protected by our platform under its pilot deployment.
“The EWL platform has already been deployed as a pilot program under an official partnership with the US Geological Survey. The platform is nearing its final stages of testing, and is being prepared for roll-out.”
EWL is also tight-lipped about how its system will operate in detail. Given the partnership with Esri, and that EWL engineers have been working in collaboration with regional telcos and smartphone manufacturers, it’s reasonable to suppose its EEW solution will be using mobile phone network technology. Bashioum predicts the system will be up to “delivering millions of messages in seconds” using a proprietary distribution platform.
Meanwhile, the controversial GeoCosmo initiative describes itself as an ‘alliance of scientists, corporations, aid organisations, governments and concerned citizens’ united in a common aim: to support development of a new global earthquake forecast system. According to GeoCosmo’s expert team, thought and action around earthquake forecasting has ‘been at a standstill’ for decades: “By primarily relying on classical mechanics to identify foreshocks, the vast majority of earthquake forecasting initiatives [by many traditional seismologists who take a mechanistic approach] have proven ineffective”, they add.
Although full details of the GeoCosmo initiative will be revealed later in 2017, GeoCosmo’s backers say it uses a combination of ‘semiconductor physics’, satellite-based technologies, artificial intelligence, Big Data, and other technologies to forecast ‘statistically likely’ earthquakes. In this context, ‘semiconductor physics’ refers not to the physics of microprocessors, but to ‘proton conductivity’ – physics of protons carrying electric currents as they relate to certain chemical processes.
The GeoCosmo proposition is based on the acceptance that rocks contain dormant ‘electronic charge carriers’ which can be activated by geological stress. According to this theory, when Earth applies high levels of stress, the rocks can undergo rupture that leads to an earthquake. Electronic charge carriers that are activated in the rocks – so-called ‘positive holes’ – travel at a speed of about 200 metres a second, through many kilometres of rock, and start flowing out of stressed rocks up to several weeks before an earthquake.
When they arrive at Earth’s surface, positive holes have key properties for earthquake forecasting, argues Friedemann Freund, chairman and chief scientist at GeoCosmo. Positive hole currents produce identifiable signals, or ‘precursors’: ultra-low frequency electromagnetic waves, air ionisation, total electron content anomalies, thermal infrared anomalies, ozone formation, carbon monoxide release, ground potential changes, and groundwater chemistry changes. GeoCosmo uses detection and analysis of these phenomena to inform its earthquake forecasts.
It does this via aggregation of precursor data gathered from satellite-based monitors, ground-based sensors, and mobile devices. This massive data set becomes the input for GeoCosmo’s earthquake forecasting algorithms. Earthquake forecasts are generated from outputs of these algorithms using artificial intelligence and natural language generation technologies.
Traditional seismologists have disregarded such “clearly-explainable phenomena” that precedes earthquakes, and continue to focus research on mechanics of rock movements, which can only give you a prediction over a very wide geographic area with an uncertainty of 30 years, says GeoCosmo’s Freund.
“If you look at a magnitude 9.0 earthquake [for instance], the total energy released during the earthquake is equivalent to the simultaneous explosion of two million atomic bombs of the  class,” Freund says. “From the physics perspective, I find it incomprehensible that a process that releases so much energy at the moment of rupture will not express itself in some recognisable way before the rupture occurs.” Freund believes these ‘recognisable ways’ are key to determining if an earthquake is en route.
There are many earthquake precursory signals we can recognise and understand, Freund adds, and “after the physics of these processes has been unravelled [and] deciphered, we can go to the next step and [ask] how can we use these different signals to produce a global earthquake forecast system that will provide early warning – days before a major earthquake happens.”
Ahmad Wani isn’t your typical Silicon Valley entrepreneur. Instead of inventing his company in a college dorm room or a Palo Alto garage, he conceived of his big idea while trapped in monsoon floods that rocked Pakistan and northern India.
As Wani and his family spent a week waiting to be rescued in 2014, the then-Stanford University engineering student began to wonder whether first responders have enough information to decide where to focus their resources after a devastating event. Returning to the U.S. to another natural disaster — this time an earthquake in Napa Valley just north of San Fransisco — reinforced the question. “I realized that there’s no science behind response,” he says. “There are thousands of 911 calls and response is not necessarily directed to places that need it.”
Out of that realization came a plan to reshape disaster management using big data. Just a few months later, Wani worked with two fellow Stanford students to create a platform to predict the toll of natural disasters. The concept is simple but also revolutionary. The One Concern software pulls geological and structural data from a variety of public and private sources and uses machine learning to predict the impact of an earthquake down to individual city blocks and buildings. Real-time information input during an earthquake improves how the system responds. And earthquakes represent just the start for the company, which plans to launch a similar program for floods and eventually other natural disasters.
Now, two years later, the company employs two dozen people, including top emergency management professionals, and counts the earthquake-prone cities of San Fransisco and Los Angeles as clients. “Their ability to predict damage at a really microlevel is really game changing,” says Michael Dayton, deputy director of the San Francisco Department of Emergency Management. “It was night and day from other technology.”
Previous software might identify a general area where responders could expect damage, but it would appear as a “big red blob” that wasn’t helpful when deciding exactly where to send resources, Dayton says. The technology also integrates information from many sources and makes it easy to parse in an emergency situation when every moment matters. The instant damage evaluations mean fast and actionable information, so first responders can prioritize search and rescue in areas most likely to be worst-hit, rather than responding to 911 calls in the order they are received.
One Concern is not the only company that sees an opportunity to use data to rethink disaster response. The mapping company Esri has built rapid-response software that shows expected damage from disasters like earthquakes, wildfires and hurricanes. And the U.S. government has invested in programs to use data to shape disaster response at agencies like the National Oceanic and Atmospheric Administration (NOAA).
A demonstration of the One Concern technology during a recent visit to the company’s Palo Alto office shows how users can map out the most at-risk parts of the city. The screen can display the effects of specific events — such as a magnitude 6.0 earthquake originating offshore — or look at the average of a large sample of potential events.
Different users will find different features most relevant. A team tasked with preparing the city for an earthquake can look at which schools are in the worst shape. The urban search and rescue team can predict which buildings are most likely to collapse completely depending on where the earthquake originates. Planners might look at which bridges are in greatest need of a retrofit.
The technology comes as earthquakes continue to rattle highly populated urban centers across the globe. A series of earthquakes in Mexico this year have killed hundreds, and those events pale in comparison to the deadliest events, like the 2010 earthquake in Haiti that killed tens of thousands of people.
The cities that have adopted One Concern have not experienced such devastating disasters since launching the technology, but the software has already reshaped how they prepare. San Fransisco has adjusted the location of relief staging areas and potential shelters to avoid hazards, while making it easier to access the most vulnerable, says Dayton.
“Our mission is to save lives,” Wani says. “How do we make the best decisions to save the most lives?”