Understanding earthquakes is a challenging problemâ€”not only because they are potentially dangerous but also because they are complicated phenomena that are difficult to study. Interpreting the massive, often convoluted data sets that are recorded by earthquake monitoring networks is a herculean task for seismologists, but the effort involved in producing accurate analyses could significantly improve the development of reliable earthquake early-warning systems.Â
A promising new collaboration between Caltech seismologists and computer scientists using artificial intelligence (AI)â€”computer systems capable of learning and performing tasks that previously required humansâ€”aims to improve the automated processes that identify earthquake waves and assess the strength, speed, and direction of shaking in real time. The collaboration includes researchers from the divisions of Geological and Planetary Sciences and Engineering and Applied Science, and is part of Caltech's AI4Science Initiative to apply AI to the big-data problems faced by scientists throughout the Institute. Powered by advanced hardware and machine-learning algorithms, modern AI has the potential to revolutionize seismological data tools and make all of us a little safer from earthquakes.Â
Recently, Caltech's Yisong Yue, an assistant professor of computing and mathematical sciences, sat down with his collaborators, Research Professor of Geophysics Egill Hauksson, Postdoctoral Scholar in Geophysics Zachary Ross, and Associate Staff Seismologist Men-Andrin Meier, to discuss the new project and future of AI and earthquake science.Â
Meier:Â One of the things that I work on is earthquake early warning. Early warning requires us to try to detect earthquakes very rapidly and predict the shaking that they will produce later so that you can get a few seconds to maybe tens of seconds of warning before the shaking starts.Â
Hauksson:Â It has to be done very quicklyâ€”that's the game. The earthquake waves will hit the closest monitoring station first, and if we can recognize them immediately, then we can send out an alert before the waves travel farther.Â
Meier:Â You only have a few seconds of seismogram to decide whether it is an earthquake, which would mean sending out an alert, or if it is instead a nuisance signalâ€”a truck driving by one of our seismometers or something like that. We have too many false classifications, too many false alerts, and people don't like that. This is a classic machine-learning problem: you have some data and you need to make a realistic and accurate classification. So, we reached out to Caltech's computing and mathematical science (CMS) department and started working on it with them.
Yue:Â The reasons why AI can be a good tool have to do with scale and complexity coupled with an abundant amount of data. Earthquake monitoring systems generate massive data sets that need to be processed in order to provide useful information to scientists. AI can do that faster and more accurately than humans can, and even find patterns that would otherwise escape the human eye. Furthermore, the patterns we hope to extract are hard for rule-based systems to adequately capture, and so the advanced pattern-matching abilities of modern deep learning can offer superior performance than existing automated earthquake monitoring algorithms.
Ross:Â In a big aftershock sequence, for example, you could have events that are spaced every 10 seconds, rapid fire, all day long. We use maybe 400 stations in Southern California to monitor earthquakes, and the waves caused by each different earthquake will hit them all at different times.Â
Yue:Â When you have multiple earthquakes, and the sensors are all firing at different locations, you want to be able to unscramble which data belong to which earthquake. Cleaning up and analyzing the data takes time. But once you train a machine-learning algorithmâ€”a computer program that learns by studying examples as opposed to through explicit programingâ€”to do this, it could make an assessment really quickly. That's the value.
Yue:Â We are not just interested in the occasional very big earthquake that happens every few years or so. We are interested in the earthquakes of all sizes that happen every day. AI has the potential to identify small earthquakes that are currently indistinguishable from background noise.
Ross:Â On average we see about 50 or so earthquakes each day in Southern California, and we have a mandate from the U.S. Geological Survey to monitor each one. There are many more, but they're just too small for us to detect with existing technology. And the smaller they are, the more often they occur. What we are trying to do is monitor, locate, detect, and characterize each and every one of those events to build "earthquake catalogs." All of this analysis is starting to reveal the very intricate details of the physical processes that drive earthquakes. Those details were not really visible before.
Ross:Â Only in the last year or two has seismology started to seriously consider AI technology. Part of it has to do with the dramatic increase in computer processing power that we have seen just within the past decade.
Meier:Â Ultimately, we want to build an algorithm that mimics what human experts do. A human seismologist can feel an earthquake or see a seismogram and immediately tell a lot of things about that earthquake just from experience. It was really difficult to teach that to a computer. With artificial intelligence, we can get much closer to how a human expert would treat the problem. We are getting much closer to creating a "virtual seismologist."
Yue:Â Fundamentally both in seismology and beyond, the reason that you want to do this kind of thing is scale and complexity. If you can train an AI that learns, then you can take a specialized skill set and make it available to anyone. The other issue is complexity. You could have a human look at detailed seismic data for a long time and uncover small earthquakes. Or you could just have an algorithm learn to pick out the patterns that matter much faster.
Meier:Â The detailed information that we're gathering helps us figure out the physics of earthquakesâ€”why they fizzle out along certain faults and trigger big quakes along others, and how often they occur.Â
Ross:Â Having talked to a range of students, I can say with fairly high confidence that most of them don't want to do cataloguing work. [Laughs.] They would rather be doing more exciting work.
Yue:Â Imagine that you're a musician and before you can become a musician, first you have to build your own piano. So you spend five years building your piano, and then you become a musician. Now we have an automated way of building pianosâ€”are we going to destroy musicians' jobs? No, we are actually empowering a new generation of musicians. We have other problems that they could be working on.
Government officials and Caltech scientists gathered at theÂ Caltech Seismological LaboratoryÂ on October 17 to declare ShakeAlertâ€”an earthquake early warning system for the three states along the West Coastâ€”"open for business."
Caltech president Thomas Rosenbaum, Sonja and William Davidow Presidential Chair and professor of physics, led the midmorning press conference, which included U.S. Representative Adam Schiff; U.S. Representative Judy Chu; Tom Heaton, professor of engineering seismology; Lucy Jones, research associate in geophysics at Caltech and founder of the Dr. Lucy Jones Center for Science and Society; Doug Given, earthquake early warning coordinator for the United States Geological Survey (USGS); and Ryan Arba, seismic hazards branch chief at the California Office of Emergency Services.
"Caltech has worked for nearly 100 years with colleagues in government and other academic institutions to leverage the insights and tools of seismology against the risks of earthquakes," Rosenbaum said, announcing a new stage in the development of an earthquake early warning system for the West Coast. "Partner institutions can now use ShakeAlert to automatically slow trains; warn industrial sites to shut off gas lines; and warn personnel to drop, cover, and hold on."Â
Given added: "Today is important because we're making a large change from a production prototype in pilot mode to an open-for-business operational mode. Now, the system is not yet finished, it's not yet complete; there is a lot of work to be done. However, there is a lot of capability in the system as it exists today to the point that it can definitely be used."
Earthquake early warning systems like ShakeAlert consist of a network of sensors near faults that transmit signals to data-processing centers when shaking occurs. These data-processing centers use algorithms to rapidly determine the earthquake's location, magnitude, and the fault rupture lengthâ€”determining the intensity of an earthquake and sending out an alert that can provide seconds or even minutes of warning. Paired with automated responsesÂ that will shut off gas before shaking starts, ShakeAlert could be instrumental in preventing the fires that typically damage cities after a major earthquake, Jones said.Â
Earthquake early warning systems do not predict earthquakes before they happen. Rather, they transmit a heads-up that an earthquake is happening; a heads-up that can arrive ahead of the seismic waves generated in the quake, potentially providing crucial time to allow individuals to take cover and for infrastructure to prepare for the quake (for example, for trains to halt operation). These warnings operate on the principle that seismic waves travel at just a few miles per second, but messages can be transmitted almost instantly. During an earthquake, several types of seismic waves radiate out from the quake's epicenter, including compressional waves (or P-waves), transverse waves (or S-waves), and surface waves. The weak P-waves move faster than the damaging S-waves and surface waves. With an earthquake early warning system in place, those P-waves will trigger sensors that can send out a warning ahead of the arrival of the S-waves and surface waves.Â
Though only half of the sensor network that ShakeAlert will need has been built out so farâ€”primarily around major metropolitan areasâ€”the state of California and the federal government have allocated funding that should allow the rest of California's portion of the network to be constructed over the next two years, Given said. In addition, an upgrade to the software that processes data from the sensor networks was deployed on September 28. This new software should reduce the number of mistakes and missed alerts, making ShakeAlert more reliable, Given said.
A key step now is for companies and institutions to help find ways to take advantage of these alerts to save lives, he said.
"This is a wonderful milestone," Schiff said. "We can now see the end, I hope, in two or three years where the system is fully built out and funded and in operation. And once people come to see the benefit, then the future of the system will be even brighter. Getting that kind of advance notice is going to be so meaningful in terms of making sure people get to a safe place."
Future iterations of the system will be able to send warnings to cell phones as well, Schiff said. Such alerts will need to be rolled out with public education to explain to individuals what to do when they receive such alertsâ€”not to panicâ€”and know that there could be false alarms.
Chu, whose district includes Caltech, said, "One of the reasons that I am so proud to be a representative from this area is our science. In our district, amazing advances are happening every day that will take us to Mars or bring us a better understanding of our environment. And the ShakeAlert that we are announcing today belongs in that pantheon of history-making innovations to come out of Caltech."Â
For Heaton, one of the fathers of ShakeAlert and a scientist who has been interested in earthquake early warning since 1979, this is a day that was a long time in coming.Â
"In those days, I could see that we could technically do it. But what I didn't really understand was what was involved to get 40 million people on the West Coast to get together to try and make this system a reality. What it really takes is leadership to do that," Heaton said.Â
Earthquake early warning systems already exist in Mexico and Japan, which have experienced recent and devastating earthquakes. But it has been difficult to find the political will to spend millions of dollars developing a system for the U.S. West Coast, which is long overdue for a serious earthquake.
ShakeAlert has been in development since 2006. In 2011, Caltech, along with UC Berkeley and the University of Washington received $6 million from the Gordon and Betty Moore Foundation for the research and development of the system; and in 2015, the USGS announced approximately $4 million in awards to Caltech, UC Berkeley, the University of Washington, and the University of Oregon for ShakeAlert's expansion and improvement.
Currently, ShakeAlert's infrastructures consist of the California Integrated Seismic Network (400 ground-motion sensors operated by Caltech in partnership with UC Berkeley, the USGS, and the State of California), and the Pacific Northwest Seismic Network (a similar regional network operated by the USGS, University of Washington, and the University of Oregon).
Over the past few years, ShakeAlert has detected thousands of earthquakes, including two that caused damage. It began sending alerts within four seconds of the beginning of the magnitude 5.1 La Habra earthquake in 2014, and gave users in Berkeley five seconds of warning before seismological waves arrived during the magnitude 6.0 South Napa earthquake, also in 2014. Beta-test users received these alerts as a pop-up on their computers; the pop-up displayed a map of the affected region as well as the amount of time until shaking would begin, the estimated magnitude of the quake, and other data.Â
Researchers with theÂ Advanced Rapid Imaging and AnalysisÂ (ARIA) project, a NASA mission led by Caltech scientists, used new satellite data to produce a map of ground deformation on the resort island of Lombok, Indonesia, following a deadly 6.9-magnitude earthquake on August 5.
The false-color map shows the amount of permanent surface movement that occurred, almost entirely due to the quake, over a six-day period between satellite images taken on July 30 and August 5.
From the pattern of deformation in the map, scientists have determined that the earthquake fault slip was on a fault beneath the northwestern part of Lombok Island, and it caused as much as 10 inches (25 centimeters) of uplift of the ground surface. White areas in the image are places where the radar measurement was not possible, largely due to dense forests in the middle of the islands.
Through these maps, NASA and its partners are contributing important observations and expertise that can assist with response to earthquakes and other natural or human-produced hazards.
In work that offers insight into the magnitude of the hazards posed by earthquake faults in general, seismologists have developed a model to determine the size of an earthquake that could be triggered by the underground injection of fluids produced as a by-product of hydraulic fracturing.
Hydraulic fracturing, or "fracking," is a petroleum-extraction procedure in which millions of gallons of water (as well as sand and chemicals) are injected deep into underground shale beds to crack the rock and release natural gas and oil. According to the United States Geological Survey, fracking itself does not typically trigger earthquakes. Instead, the increased risk for seismicity is more strongly linked with the subsequent injection of the wastewater from fracking and other oil-extraction processes into massive disposal wells that are thousands of feet underground.
Previous attempts to model the relationship between injection of wastewater and the triggering of earthquakes suggested that the maximum magnitude of the seismic activity induced in this way would be proportional to the volume of the fluids injected. However, this interpretation fails to account for the fact that earthquakes can grow beyond the area impacted by fluid pressure, says Jean Paul Ampuero, professor of seismology at Caltech and co-author of a new study on the topic that appears in the journal Science Advances on December 20.
Combining theory and computer simulations of dynamic earthquake ruptures, Ampuero and his colleagues developed a model that explains how the size of injection-induced earthquakes depends on not only the volume of fluid being injected but also the energy stored on nearby faults. The result is a model that quantifies the distance that an earthquake can propagate beyond an injection siteâ€”which in turn predicts the maximum magnitude of an induced seismic event.
"Earthquakes induced by human activities involving underground injection of fluids or gas are a growing concern, a hazard that needs to be controlled in order to develop a safer and cleaner energy future," Ampuero says.
This induced seismicity has been the subject of significant research in recent years and is also attracting researchers who, like Ampuero, are primarily interested in unraveling the physics of natural earthquakes. "This may be the closest researchers will ever get to a large-scale controlled earthquake experiment," Ampuero says. For the new work, Ampuero teamed up with Martin Galis, postdoctoral researcher at King Abdullah University of Science and Technology (KAUST) in Saudi Arabia.
It is important to note that the new model only predicts the maximum possible magnitude of an earthquake rather than what the earthquake magnitude will actually be, the researchers say. It defines upper limits based on the amount of pent-up energy in the earth's crust prior to fluid injection.
The new model offers insight into natural earthquakes, creating a framework for understanding what causes earthquakes to stop shaking. Earthquakes can be triggered by the pressure and disturbance caused by fluid injection, but they may grow beyond the zone immediately impacted by the wastewater injection by tapping into tectonic energy that is already stored nearby. As is the case for induced seismicity, natural earthquakes can start in small areas of the earth's crust where that energy is concentrated. How large they grow is determined by the amount of energy in surrounding regions.
The paper is titled "Induced seismicity provides insight into why earthquake ruptures stop." Ampuero and Galis's co-authors include Paul Martin Mai of KAUST and FrĂ©dĂ©ric Cappa of the UniversitĂ© CĂ´te d'Azur in Nice and Institut Universitaire de France in Paris. Funding came from the National Science Foundation, KAUST, and the Agence Nationale de la RecherchĂ© in France.
This is the second study this month from Ampuero that offers new insight into earthquake science. On December 1, Ampuero and colleagues from Centre national de la recherchĂ© scientifique in Paris found that it is possible to observe disturbances in the earth's gravitational field almost instantly after an earthquake, raising the potential for the use of these disturbances as part of an early-warning system. (These disturbances travel at the speed of light, while the fastest seismic waves of an earthquake propagate at several kilometers per second, which means that monitoring the disturbances could potentially improve existing early-warning systems by seconds or even minutes.)
Ampuero and his colleagues found that seismometers in China and South Korea picked up perturbations in the earth's gravitational field during the 9.1 Tohoku earthquake in Japan in 2011 via signals that appeared as tiny accelerations on seismometers more than a minute before the ground beneath the seismometers started to shake.
The 2011 Japanese earthquake was a defining moment for Mark Simons. The devastating 9.0-magnitude quake and its subsequent tsunami, which took nearly 16,000 lives, spurred efforts around the globe that will shape how nations predict and prepare for future natural disasters and motivated new approaches to basic earthquake science that are applicable to seismic events large and small.
Read more on the Break Through campaign website.
By simulating earthquakes in a lab, engineers at Caltech have documented the evolution of friction during an earthquakeâ€”measuring what could once only be inferred, and shedding light on one of the biggest unknowns in earthquake modeling.
Before an earthquake, static friction helps hold the two sides of a fault immobile and pressed against each other. During the passage of an earthquake rupture, that friction becomes dynamic as the two sides of the fault grind past one another. Dynamic friction evolves throughout an earthquake, affecting how much and how fast the ground will shake and thus, most importantly, the destructiveness of the earthquake.
"Friction plays a key role in how ruptures unzip faults in the earth's crust," says Vito Rubino, research scientist at Caltech's Division of Engineering and Applied Science (EAS). "Assumptions about dynamic friction affect a wide range of earthquake science predictions, including how fast ruptures will occur, the nature of ground shaking, and residual stress levels on faults. Yet the precise nature of dynamic friction remains one of the biggest unknowns in earthquake science."
Previously, it commonly had been believed that the evolution of dynamic friction was mainly governed by how far the fault slipped at each point as a rupture went byâ€”that is, by the relative distance one side of a fault slides past the other during dynamic sliding. Analyzing earthquakes that were simulated in a lab, the team instead found that sliding history is important but the key long-term factor is actually the slip velocityâ€”not just how far the fault slips, but how fast.
Rubino is the lead author on a paper on the team's findings that was published in Nature Communications on June 29. He collaborated with Caltech's Ares Rosakis, the Theodore von KĂˇrmĂˇn Professor of Aeronautics and Mechanical Engineering at EAS, and Nadia Lapusta, professor of mechanical engineering and geophysics, who has joint appointments with EAS and the Caltech Division of Geological and Planetary Sciences.
The team conducted the research at a Caltech facility, directed by Rosakis, that has been unofficially dubbed the "seismological wind tunnel." At the facility, researchers use advanced high-speed optical diagnostics and other techniques to study how earthquake ruptures occur.
"Our unique facility allows us to study dynamic friction laws by following individual, fast-moving shear ruptures and recording friction along their sliding faces in real time," Rosakis says. "This allows us for the first time to study friction point-wise and without having to assume that sliding occurs uniformly, as is done in classical friction studies," Rosakis adds.
To simulate an earthquake in the lab, the researchers first cut in half a transparent block of a type of plastic known as homalite, which has similar mechanical properties to rock. They then put the two pieces together under pressure, simulating the static friction that builds up along a fault line. Next, they placed a small nickel-chromium wire fuse at the location where they wanted the epicenter of the quake to be. Triggering the fuse produced a local pressure release, which reduced friction at that location, and allowed a very fast rupture to propagate up the miniature fault.
In this study, the team recorded these simulated earthquakes using a new diagnostic method that combines high-speed photography (at 2 million frames per second) with a technique called digital image correlation, in which individual frames are compared and contrasted with one another and changes between those imagesâ€”indicating motionâ€”are tracked with sub-pixel accuracy.
"Some numerical models of earthquake rupture, including the ones developed in my group at Caltech, have used friction laws with slip-velocity dependence, based on a collection of rock mechanics experiments and theories. It is gratifying to see those formulations validated by the spontaneous mini-earthquake ruptures in our study, " Lapusta says.
In future work, the team plans to use its observations to improve the existing mathematical models about the nature of dynamic friction and to help create new ones that better represent the experimental observations; such new models would improve computer earthquake simulations.
The study is titled "Understanding dynamic friction through spontaneously evolving laboratory earthquakes."Â This research was supported by the National Science Foundation, the U.S. Geological Survey, and the Southern California Earthquake Center.
It is a common trope in disaster movies: an earthquake strikes, causing the ground to rip open and swallow people and cars whole. The gaping earth might make for cinematic drama, but earthquake scientists have long held that it does not happen.
Except, it can, according to new experimental research from Caltech.
The work, appearing in the journal Nature on May 1, shows how the earth can split openâ€”and then quickly close back upâ€”during earthquakes along thrust faults.
Thrust faults have been the site of some of the world's largest quakes, such as the 2011 Tohoku earthquake off the coast of Japan, which damaged the Fukushima nuclear power plant. They occur in weak areas of the earth's crust where one slab of rock compresses against another, sliding up and over it during an earthquake.
A team of engineers and scientists from Caltech and Ă‰cole normale supĂ©rieure (ENS) in Paris have discovered that fast ruptures propagating up toward the earth's surface along a thrust fault can cause one side of a fault to twist away from the other, opening up a gap of up to a few meters that then snaps shut.
Thrust fault earthquakes generally occur when two slabs of rock press against one another, and pressure overcomes the friction holding them in place. It has long been assumed that, at shallow depths, the plates would just slide against one another for a short distance, without opening.
However, researchers investigating the Tohoku earthquake found that not only did the fault slip at shallow depths, it did so by up to 50 meters in some places. That huge motion, which occurred just offshore, triggered a tsunami that caused damage to facilities along the coast of Japan, including at the Fukushima Daiichi Nuclear Power Plant.
In the Nature paper, the team hypothesizes that the Tohoku earthquake rupture propagated up the fault andâ€”once it neared the surfaceâ€”caused one slab of rock to twist away from another, opening a gap and momentarily removing any friction between the two walls. This allowed the fault to slip 50 meters.
That opening of the fault was supposed to be impossible.
"This is actually built into most computer models of earthquakes right now. The models have been programed in a way that dictates that the walls of the fault cannot separate from one another," says Ares Rosakis, Theodore von KĂˇrmĂˇn Professor of Aeronautics and Mechanical Engineering at Caltech and one of the senior authors of the Nature paper. "The findings demonstrate the value of experimentation and observation. Computer models can only be as realistic as their built-in assumptions allow them to be."
The international team discovered the twisting phenomenon by simulating an earthquake in a Caltech facility that has been unofficially dubbed the "Seismological Wind Tunnel." The facility started as a collaboration between Rosakis, an engineer studying how materials fail, and Hiroo Kanamori, a seismologist exploring the physics of earthquakes and a coauthor of the Nature study. "The Caltech research environment helped us a great deal to have close collaboration across different scientific disciplines," Kanamori said. "We seismologists have benefited a great deal from collaboration with Professor Rosakis's group, because it is often very difficult to perform experiments to test our ideas in seismology."
At the facility, researchers use advanced high-speed optical diagnostics to study how earthquake ruptures occur. To simulate a thrust fault earthquake in the lab, the researchers first cut in half a transparent block of plastic that has mechanical properties similar to that of rock. They then put the broken pieces back together under pressure, simulating the tectonic load of a fault line. Next, they place a small nickel-chromium wire fuse at the location where they want the epicenter of the quake to be. When they set off the fuse, the friction at the fuse's location is reduced, allowing a very fast rupture to propagate up the miniature fault. The material is photoelastic, meaning that it visually showsâ€”through light interference as it travelsÂ in the clear materialâ€”the propagation of stress waves. The simulated quake is recorded using high-speed cameras and the resulting motion is captured by laser velocimeters (particle speed sensors).
"This is a great example of collaboration between seismologists, tectonisists and engineers. And, not to put too fine a point on it, US/French collaboration," says Harsha Bhat, coauthor of the paper and a research scientist at ENS. Bhat was previously a postdoctoral researcher at Caltech.
The team was surprised to see that, as the rupture hit the surface, the fault twisted open and then snapped shut. Subsequent computer simulationsâ€”with models that were modified to remove the artificial rules against the fault openingâ€”confirmed what the team observed experimentally: one slab can twist violently away from the other. This can happen both on land and on underwater thrust faults, meaning that this mechanism has the potential to change our understanding of how tsunamis are generated.
The paper is titled "Experimental evidence that thrust earthquake ruptures might open faults." The lead author is Vahe Gabuchian (MS '08, PhD '15), a former PhD student at Caltech's Graduate Aerospace Laboratories (GALCIT), and coauthors include RaĂşl Madariaga of ENS. This research was funded by the National Science Foundation.
On February 18, 2015, an explosion rattled the ExxonMobil refinery in Torrance, causing ground shaking equivalent to that of a magnitude-2.0 earthquake and blasting out an air pressure wave similar to a sonic boom.
Traveling at 343 meters per secondâ€”about the speed of soundâ€”the air pressure wave reached a 52-story high-rise in downtown Los Angeles 66 seconds after the blast.
The building's occupants probably did not notice a thing; the building shifted at most three-hundredths of a millimeter in response. But the building's seismometersâ€”one is installed on every floor, as well as on the basement levelsâ€”noted and recorded the motion of each individual floor.
Those sensors are part of the Community Seismic Network (CSN), a project launched at Caltech in 2011 to seed the Los Angeles area with relatively inexpensive seismometers aimed at providing a high level of detail of how an earthquake shakes the Southern California region, as well as how individual buildings respond. That level of detail has the potential to provide critical and immediate information about whether the building is structurally compromised in the wake of an earthquake, says Caltech's Monica Kohler, research assistant professor in the Division of Engineering and Applied Science.
For example, if building inspectors know that inter-story driftâ€”the displacement of each floor relative to the floors immediately below and above itâ€”has exceeded certain limits based on the building's size and construction, then it is a safe bet that the building has suffered damage in a quake. Alternately, if inspectors know that a building has experienced shaking well within its tolerances, it could potentially be reoccupied soonerâ€”helping an earthquake-struck city to more quickly get back to normal.
"We want first responders, structural engineers, and facilities engineers to be able to make decisions based on what the data say," says Kohler, the lead author of a paper detailing the high-rise's response that recently appeared in the journal Earthquake Spectra.
The keys to the CSN's success are affordability and ease of installation of its seismic detectors. Standard, high-quality seismic detectors can cost tens of thousands of dollars and need special vaults to house and protect them that can easily double the price. By contrast, the CSN detectors use $40 accelerometers and other off-the-shelf hardware, cost roughly $300 to build, and require minimal training to install. Approximately 700 of the devices have been installed so far, mostly in Los Angeles.
However, the CSN sensors are roughly 250 times less sensitive than their more expensive counterparts, which is why the ability to successfully detect and quantify the downtown building's response to the ExxonMobil explosion was such an important proof-of-concept.
"It's a validation of our approach," says CSN's project manager, Richard Guy.
Sonic booms have been noted by seismic networks dozens of times before, beginning in the 1980s with the first detections of seismic shaking caused by space-shuttle reentries. The sonic booms, found Hiroo Kanamori and colleagues at Caltech and the United States Geological Survey, rattled buildings that, in turn, shook the ground around them.
"Seismologists try to understand what is happening in the earth and how that affects buildings by looking at everything we see on seismograms," says Kanamori, Caltech's John E. and Hazel S. Smits Professor of Geophysics, Emeritus, and coauthor of the Earthquake Spectra paper. "In most cases, signals come from the interior of the earth, but nothing prevents us from studying signals from the air. Though rare, the signals from the air provide a new dimension in the field of seismology."
The earlier sonic boom detections were made using single-channel devices, which typically record motion in one direction only. While this information is useful for understanding ground shaking, a three-dimensional record of the floor-by-floor motion of a building can reveal how much a building is rocking, swaying, and shifting; two or more sensors installed per floor can show the twisting of the structure.
"The more sensors you have in a small area, the more detail you're going to see. If there are things happening on a small scale, you'll never see it until you have sensors deployed on that scale," Kohler says.
Kohler and her colleagues found that the air pressure wave from the explosion had about the same impact on the high-rise as an 8 mile-per-hour gust of wind. A pressure wave about 100 times larger would have been required to have broken windows in the building; a wave 1,000 times larger would have been necessary to cause significant damage to the building.
The ExxonMobil blast was not the first shaking recorded by the building's seismometers. A number of earthquakesâ€”including a magnitude-4.2 quake on January 4, 2015, with an epicenter in Castaic Lake, about 40 miles northwest of downtown Los Angelesâ€”also were registered by the seismic detectors on nearly every floor of the building. But the refinery explosion-induced shaking was an important test of the sensitivity of the instruments, and of the ability of researchers to separate earthquake signals from other sources of shaking.
Other authors of the Earthquake Spectra paper, "Downtown Los Angeles 52-Story High-Rise and Free-Field Response to an Oil Refinery Explosion," include Caltech's Anthony Massari, Thomas Heaton, Egill Hauksson, Robert Clayton, Julian Bunn, and K. M. Chandy. Funding for the CSN came from the Gordon and Betty Moore Foundation, the Terrestrial Hazard Observation and Reporting Center at Caltech, and the Divisions of Geological and Planetary Sciences and Engineering and Applied Science at Caltech.Â
Since the late 1970s, Caltech seismologist Tom Heaton, professor of engineering seismology, has been working to develop earthquake early-warning (EEW) systemsâ€”networks of ground-based sensors that can send data to users when the earth begins to tremble nearby, giving them seconds to potentially minutes to prepare before the shaking reaches them. In fact, Heaton wrote the first paper published on the concept in 1985. EEW systems have been implemented in countries like Japan, Mexico, and Turkey. However, the Unites States has been slow to regard EEW systems as a priority for the West Coast.
But on February 2, 2016, the White House held the Earthquake Resilience Summit, signaling a new focus on earthquake safety and EEW systems. There, stakeholdersâ€”including Caltech's Heaton and Egill Hauksson, research professor in geophysics; and U.S. Geological Survey (USGS) seismologist Lucy Jones, a visiting associate in geophysics at Caltech and seismic risk advisor to the mayor of Los Angelesâ€”discussed the need for earthquake early warning and explored steps that can be taken to make such systems a reality.Â
At the summit, the Gordon and Betty Moore Foundation announced $3.6 million in grants to advance a West Coast EEW system called ShakeAlert, which received an initial $6 million in funding from foundation in 2011. The new grants will go to researchers working on the system at Caltech, the USGS, UC Berkeley,Â and the University of Washington.
"We have been successfully operating a demonstration system for several years, and we know that it works for the events that have happened in the test period," says Heaton. "However, there is still significant development that is required to ensure that the system will work reliably in very large earthquakes similar to the great 1906 San Francisco earthquake. This new funding allows us to accelerate the rate at which we develop this critical system."
In addition, the Obama Administration outlined new federal commitments to support greater earthquake safety including an executive order to ensure that new construction of federal buildings is up to code and that federal assets are available for recovery efforts after a large earthquake.
The commitments follow a December announcement from Congressman Adam Schiff (D-Burbank) that Congress has included $8.2 million in the fiscal year 2016 funding bill specifically designated for a West Coast earthquake early warning system.
"By increasing the funding for the West Coast earthquake early-warning system, Congress is sending a message to the Western states that it supports this life-saving system. But the federal government cannot do it alone and will need local stakeholders, both public and private, to get behind the effort with their own resources," said Schiff, in a press release. "The early warning system will give us critical time for trains to be slowed and surgeries to be stopped before shaking hitsâ€”saving lives and protecting infrastructure. This early warning system is an investmentÂ we need to make now, not after the 'big one' hits."
ShakeAlert utilizes a network of seismometersâ€”instruments that measure ground motionâ€”widely scattered across the Western states. In California, that network of sensors is called the California Integrated Seismic Network (CISN) and is made up of computerized seismometers that send ground-motion data back to research centers like the Seismological Laboratory at Caltech.
Here's how the current ShakeAlert works: a user display opens in a pop-up window on a recipient's computer as soon as a significant earthquake occurs in California. The screen lists the quake's estimated location and magnitude based on the sensor data received to that point, along with an estimate of how much time will pass before the shaking reaches the user's location. The program also gives an approximation of how intense that shaking will be. Since ShakeAlert uses information from a seismic event in progress, people living near the epicenter do not get muchâ€”if anyâ€”warning, but those farther away could have seconds or even tens of seconds' notice.
The goal is an improved version of ShakeAlert that will eventually give schools, utilities, industries, and the general public a heads-up in the event of a major temblor.
Read more about how ShakeAlert works and about Caltech's development of EEW systems in a feature that ran in the Summer 2013 issue of E&S magazine called Can We Predict Earthquakes?
In the event of a major earthquake in Los Angeles, first responders ideally would immediately have a map of the most intense shaking around the cityâ€”allowing them to send help to the hardest-hit areas first.
A new collaboration between Caltech researchers and schools of the Los Angeles Unified School District (LAUSD) provides a crucial step in the creation of such damage maps by vastly broadening the scope of a dense network of seismic sensors in the Los Angeles Basin.
To create an accurate shaking-intensity map, seismologists need to measure ground motionâ€”which can vary from kilometer to kilometer because of differences in soil and earth structureâ€”at many locations across the region. In 2011, Professor of Geophysics Rob Clayton and his colleagues, Professor of Engineering Seismology Tom Heaton and Simon Ramo Professor of Computer Science, Emeritus, K. Mani Chandy, began creating a web of such sensors via the Community Seismic Network (CSN), a program funded by the Gordon and Betty Moore Foundation.
The CSN consists of hundreds of small, inexpensive accelerometersâ€”instruments that detect ground movements before, during, and after a seismic eventâ€”installed initially in the homes of volunteers in the greater Pasadena area. Since 2011, each device has been actively collecting and feeding seismic information to the CSN via its host house's Internet connection, allowing Clayton and his colleagues to create high-resolution maps of seismic activity in the western San Gabriel Valley. But the Caltech team wanted to find a way to expand the reach of the network throughout the earthquake-prone broader Los Angeles Basin. Their inventive solution? Integrate accelerometers into the infrastructure of L.A.'s public schools.
Through the efforts of Richard Guy, CSN project manager, sensors already have been installed in 100 LAUSD schools, covering an area ranging from northeast Los Angeles to downtown. CSN is now working to expand the project to include all of the district's more than 1,000 schools.
The new collaboration has the potential to help millions of people in Southern California when a big quake strikes. For example, data from the new network could be incorporated into the ShakeAlert early-warning system that is currently under development. Although no sensor can predict an earthquake, the accelerometers can detect an earthquake in one area of the L.A. Basin so quickly that an alert or warning could be sent to people in adjacent areas of the LA Basin before strong shaking arrivesâ€”potentially giving them enough time to find a safe spot.
The new dense network of sensors will also provide an improved map of shaking intensity for the whole region. The U.S. Geological Survey already provides a similar service called ShakeMap, which relies on sensors that are located several miles from one another and hence cannot provide a block-by-block resolution of shaking and possible damage. The new dense network of sensors has the potential to provide ShakeMap with a more accurate assessment of damage for response and recovery efforts.
"You can imagine a fire chief stepping out and saying, 'Wow. That was a big one. Now where do I go to help the community?' Obviously they want their focus to be where the maximum damage and danger is. They have other things to worry about too, but the best proxy for damage that we have is the level of shakingâ€”and our dense network of sensors can provide that information," Clayton says.
But the new sensors do more than feed information into the networkâ€”they also provide valuable information to individual schools. "Principals have a particularly difficult problem in the event of an earthquake," Clayton says. "The first thing during a quake, of course, is to tell everyone to get under their desk. When the shaking stops, all of the kids are evacuated out of the school and into the schoolyard. And then what do principals do? At that point, they have to decide if it's safe enough to go back into the school, or if they should just send the kids home. But they do not know how badly the school is damaged."
The new school sensors could help inform this judgment call, Clayton says. Although they work in much the same way as those that were previously placed in volunteers' homesâ€”recording ground accelerations and transmitting those data back to the researchers via an Internet connectionâ€”the sensors also contain an onboard computer that compares the event to a so-called fragility curve. Fragility curves provide predictions of the damage that a particular building would sustain under the shaking measured.
"Coupled with the fragility curve, the sensors could allow a school official to decide whether or not it is safe to reenter the school," Clayton says.
The Community Seismic Network's LAUSD collaboration was funded by the Gordon and Betty Moore Foundation. The network is a collaboration between Caltech's seismology, earthquake-engineering, and computer-science departments.