Around the globe, thousands of seismometers record ground motion caused by earthquakes and other sources such as underground equipment, volcanic eruptions, and nuclear weapon tests. While this network of instruments is effective for observing significant tremblors, thousands more seismometers—located much closer together—would be required to measure the lower vibrations of everyday ground motion that can gradually damage roads, bridges, and other infrastructure. A research team in Lawrence Livermore’s Atmospheric, Earth, and Energy Science Division (AEED) has tested the capability to meet this need by creating virtual seismometers collecting seismic data every few meters with an instrument called an interrogator and access to existing fiber-optic telecommunications cables.
An interrogator sends a pulse of light into an optical fiber and analyzes reflected and scattered signals to determine how a segment of buried fiber-optic cable moves over time. These changes in light—optical signals—are translated into electrical, digital signals to quantify data using interferometry, a technique of splitting and recombining a laser beam to enable very small changes in distance—less than the wavelength of the laser light itself—to be measured. These small changes in distance can be the result of temperature, pressure, strain, vibration, and similar parameters affecting an underground cable. In the case of seismic data, an interrogator measures the motion and strain at the nanometer level as a fiber-optic cable is momentarily deformed by tremors. This approach, known as distributed acoustic sensing (DAS), was first applied to monitor ground motion for the oil and natural gas exploration industry. Scientists seeking the same capability for other research fields had to build or lease interrogators for years. When interrogators became available off the shelf several years ago, the AEED research team saw numerous possibilities for advancing ground motion detection research. “We brainstormed use cases for the interrogator, properties we could measure in the built, urban environment,” says seismologist Eric Matzel. “Of those ideas, we asked ourselves what would be of the most interest to the Laboratory.” Matzel along with fellow seismologists Gene Ichinose, Bill Walter, and Rengin Gok, landed on detecting seismic hazards to infrastructure such as bridges, buildings, tunnels, and dams.
Rolling Data
From February through April 2025, the team, along with Livermore computer scientist Tim Brandt, introduced the newly obtained interrogator into an unused, 80-kilometer fiber-optic cable network running from the Moscone Center in San Francisco to Sunnyvale, California. Project collaborators included fiber-optic testing companies Luna Technologies and Viavi Solutions and a California Institute of Technology researcher representing the Optical Fiber Communications Society. “In this evolving field, fiber-optic companies and interrogator companies are interested in new applications for their technologies and want to share information,” says Walter.
At first, the DAS system observed ground motion along a regional train operator’s rail lines as well as a small earthquake near Gilroy, California (about 70 kilometers southeast of Sunnyvale). Then, on March 17, 2025, the DAS system registered a magnitude 3.9 earthquake in the hills west of Dublin, California—a city 55 kilometers west from San Francisco and about the same distance northeast from Sunnyvale. An earthquake this size creates noticeable shaking or rolling but rarely causes lasting damage in an area with good construction quality. When Ichinose felt the first rumblings of the earthquake, he immediately checked into the monitoring system and watched every detail of the seismic wave field propagating across the region.
“The test demonstrated that signals can be separated from ground motion with a resolution of approximately 10 meters along the fiber-optic cable,” says Ichinose. “That level of data resolution is equivalent to operating 1,000 seismometers for every 10 kilometers of cable. Comparatively, data from a public source such as the USGS (United States Geological Survey) would provide readings from fewer than 10 seismometers and require interpolation of data between sensor readings.” Tapping into more, and regionally concentrated, seismic data enables researchers to measure the entire wave field from a seismic event. A larger distributed network yields insights about small-scale fault line motion that may provide an early earthquake warning system.
As the experiment continued, collected data confirmed the value of DAS for studying ground motion in urban areas. The interrogator highlighted subsurface variations from 10 kilometers below ground, tidal variations in San Francisco Bay, and ground motion caused by people as well as shifting buildings and soil. Measuring motion in infrastructure such as roads, bridges, dams as well as skyscrapers and other buildings, could offer advance warning of issues or help plan maintenance in a way that balances resources. “The key goal is looking for something that is about to break, to know where it is weak,” says Ichinose. Adds Matzel, “Earthquakes produce low-frequency waves, around 5 hertz. We need a higher frequency to study smaller changes in the built environment, to study the small cracks in a structure or infiltration of water and potential for liquefaction, for example. The urban environment is filled with signals rich in these high frequencies, such as motors, construction equipment, automobile and train traffic, which we can use at no additional cost.” DAS could also help locate smaller fault lines and unidentified fractures in areas where seismometers could not be installed in the past because of cost or construction restrictions.
Potential testing partners emerged following early press coverage of the Laboratory’s findings. For example, the city of Fort Bragg, California, which had received a grant to install fiber-optic cables to provide broadband-for-all, seeks to interrogate their network and collect data to identify earthquake-prone areas and monitor bridge safety. The city is located near the San Andreas fault, but little seismic data is available given the limited access to install traditional seismometers.
Greater Reach
AEED’s interest in seismic data extends beyond infrastructure safety and earthquake prediction to support the Laboratory’s mission. Seismometers collect data to characterize nuclear testing from other ground motion and inform explosives monitoring programs for treaty compliance. (See S&TR, January 2021, Seismic Sleuths Set Off the Source Physics Experiment.) “Earthquakes are a useful signal to calibrate detection technologies for explosives testing,” says Walter. “Our group looks five or more years out for technologies to improve seismic detection down to the background noise of the Earth. Getting measurements in a new way adds a new set of data.”
The Global Seismographic Network (GSN), a cooperative effort of the National Science Foundation and the USGS, operates approximately 150 monitoring stations worldwide. GSN estimates that roughly 26,000 additional stations exist (about 3,000 in the United States), although some are thought to be inactive. However, the number of installed instruments is small compared to the potential to operate thousands of virtual seismometers over a region just a few hundred kilometers in area. DAS requires a smaller investment than seismometers as well. The price of one seismometer can reach $10,000 plus the additional effort to either bury it on a bare rock surface or install it within a custom-made structure to accurately record ground movement. “Prices and installation challenges are the reasons we don’t see that many seismometers in the world,” says Walter.
The other element required for DAS—fiber-optic cable—is abundantly available. While the fiber-optic cables accessed for the 2025 test were not in use, telecommunications cables can be interrogated without disrupting their communications purpose. “Fiber-optic cables are everywhere, opening up a real possibility of continental seismic data,” says Gok. Where cable networks don’t exist, such as areas of seismic activity off the California coast, fiber-optics are easier to install than seismometers. “We can access locations only served with a few seismometers or places even more difficult to reach,” says Walter.
Shaking Out the System
The team has considered factors inherent to DAS that can make the technology less precise than physical seismometers. The soil where fiber-optic cables are installed naturally packs and unpacks, and its temperature varies, potentially creating false ground motion readings. The echoes of ground motion, called the coda, scatter near any fiber-optic cable location. Yet such a diversified signal enables in situ calibration of the data close to where and when the ground motion originally deformed the cable. Gok recently demonstrated that this calibration yields a more accurate reconstruction of the original ground motion along the length of fiber-optic cable.
Another potentially limited factor is that seismometers typically record all three Cartesian coordinates—up–down, east–west, and north–south—providing a 3D motion map at a single location. Fiber channels record strain along the fiber axis only. However, with fiber-optic cable enabling thousands of channels at meter-scale spacing, DAS provides a much higher detail of the full wavefield.
Finally, as the use of fiber-optic technology expands, researchers must contend with the flood of data collected. “An earthquake puts a lot of energy across the fiber at once creating a burst of recorded data. We may record as much as 50 terabytes (TB) over a month for our test bed system,” says Ichinose. “In the case of the Laboratory’s tested DAS system, analyzing data from 8,000 virtual seismometers generating new data points every 30 nanoseconds quickly reaches the edge of many institutions’ computing capabilities. The demand will increase as more fiber-optic cables are accessed.” Edge computing—processing raw data in the field next to the interrogator and transmitting only intermediate or final products for analysis—reduces data storage demands. The team’s testing revealed that a sufficient sample size can be collected in 12 TB of data.
The team continues to study seismology within urban settings. “We can see signals from foot traffic, cars, and trains,” says Matzel. “How much of that signal could be used to identify a risk such as undetected cracks in a bridge? How fine a detail can we get to monitor and ultimately differentiate signals indicating potential damage?” The team has reached out to engineers at cities and universities to develop applications and enable testing for different parameters. For example, fiber-optic cable could be installed to study the effects of seismic activity or other sources of motion, such as wind, on specific bridges or buildings. Accessing a utility’s data for steel pipe or concrete would offer insights into the effects of vibrations on critical infrastructure. As more data is collected across the region, more correlations between ground motion and events such as weather or construction could be made. Says Gok, “We’re developing a tool, a powerful yet flexible processing method to assess hazards.”
—Suzanne Storar
For further information contact Gene Ichinose (925) 423-8489 (ichinose1 [at] llnl.gov (ichinose1[at]llnl[dot]gov)).
