Skip to content
"Not rocket science [but] it does take special training”

As the world quieted down in 2020, Raspberry Shakes listened

Humble Raspberry Pis helped scientists track the seismic noise people stopped making in 2020.

Alka Tripathy-Lang | 29
The Raspberry Shake, a simple seismograph based on Raspberry Pi hardware. Credit: Mike Hotchkiss, Raspberry Shake
The Raspberry Shake, a simple seismograph based on Raspberry Pi hardware. Credit: Mike Hotchkiss, Raspberry Shake

“It’s the trains!” Ryan Hollister yelled to his wife Laura as he burst into their home in Turlock, California. For two weeks in 2017, they’d been staring at data from their newly installed Raspberry Shake, a Raspberry Pi-powered instrument that detects how the ground moves at a specific location. Expecting to see the tell-tale wiggles of distant earthquakes, they instead saw peculiar cigar-shaped waveforms at regular intervals. “The biggest challenge,” says Laura Hollister, “was the noise.”

“I thought it was the toilet flushing or the washing machine,” says Ryan Hollister, but simple tests of going to the restroom or doing the laundry proved him wrong. While stuck in his car watching a train rattle through Turlock, he realized the three tracks that criss-cross this small California town could be causing this mystery seismic noise. As soon as he got home, he pulled up the Raspberry Shake’s data. Sure enough, each weirdly intense caterpillar of seismic waves corresponded to a train, with the highest-amplitude waves correlating with the nearest track’s schedule, only a half mile from home.

The Hollisters’ Raspberry Shake continues to record the cigar-shaped signatures of the trains in Turlock, California, while also capturing far-off earthquakes, like this magnitude-7.4 event from New Zealand on June 18, 2020. The New Zealand quake is highlighted in yellow in the background and enlarged in the inset.
The Hollisters’ Raspberry Shake continues to record the cigar-shaped signatures of the trains in Turlock, California, while also capturing far-off earthquakes, like this magnitude-7.4 event from New Zealand on June 18, 2020. The New Zealand quake is highlighted in yellow in the background and enlarged in the inset. Credit: Ryan Hollister

It wasn’t the last time that their seismic listening device picked up signs of human activity. As COVID-19 engulfed our world, the Hollisters, a husband-wife team of Earth science educators, noticed that their Raspberry Shake registered much lower levels of activity than usual. The drop was pronounced at times when their street, a main artery to the local high school, should have been pulsing with teenagers.

That change was far from limited to Turlock. Thomas Lecocq, a seismologist who pays particular attention to Earth’s ubiquitous vibrations, discerned a marked decrease in high-frequency noise on a permanent seismic station under his purview at the Royal Observatory of Belgium. This peculiar hush was quieter and longer than the one he’d seen during the subdued days between Christmas and New Year and coincided with his country’s lockdown.

In the following months, Lecocq and 76 coauthors from around the world combed through data from seismic stations spanning more than 70 countries using Python code Lecocq wrote specifically for this purpose. A total of 268 stations had usable data, and 185 of them saw high-frequency seismic noise plummet by up to 50 percent in urban regions. The changes came in lockstep with each country’s closure in response to COVID-19. As the signals from driving, construction, and even walking fell away, Ian Nesbitt, one of Lecocq’s coauthors says, “We may be able to investigate [geologic] signals that we previously couldn’t see because it was masked by that noise.”

Many of the stations were high-end research instruments installed by university or government scientists. But 65 were tiny Raspberry Shakes, sitting in the homes and offices of scientists and hobbyists alike. It turns out that when humans make a lot of noise, seismically speaking, anyone with a spare Raspberry Pi and a few hundred dollars for a Raspberry Shake circuit board and some sensors can see it.

Build your own seismic station

The basic recipe for a seismic station requires four ingredients: sensors to measure Earth’s motion, a means to record the measurements, a long-term storage solution (either local or elsewhere), and a power source, says Emily Wolin, Seismic Network Manager for the U.S. Geological Survey (USGS) Albuquerque Seismological Laboratory.

State-of-the-art seismic stations boast numerous sensors that detect an immense range of frequencies, capturing Earth’s movement in three directions—up-down, east-west, and north-south. Digitizers and data loggers precisely record and time stamp the data. To power the equipment, the most remote stations may use solar panels, with power requirements varying dramatically based on communication needs, says Wolin.

Comparison between Raspberry Shake 4D, on the left, and typical U.S. Geological Survey seismic equipment deployed after an earthquake to monitor aftershocks.
Comparison between Raspberry Shake 4D, on the left, and typical U.S. Geological Survey seismic equipment deployed after an earthquake to monitor aftershocks. Credit: Anthony et al., 2018 (ple

To add a new seismic station to an earthquake monitoring network, Wolin says scientists must scout locations that take into account regional geology and possible noise sources—like railroads (the Hollisters' home would have never made the cut). With a list of candidate sites, they then identify and contact landowners for permission and secure access for construction, installation, and subsequent maintenance.

Wolin explains that sometimes, preparation may involve “hiring a drill rig to bore hundreds of meters into solid rock.” In some instances, thermally sealed and waterproof seismic vaults must be carefully constructed to house equipment so sensitive that they would otherwise pick up minuscule changes in pressure and temperature. Vaults also help minimize pesky anthropogenic noise. To install the sensor and electronics, “it’s not rocket science,” says Sue Hough, a USGS seismologist, but “it does take special training.”

Each layer of complexity adds another line to the bill. According to Hough, top-tier versions of a seismic station can cost well over $10,000, excluding installation costs. Branden Christensen, CEO of Raspberry Shake, says that when those costs are included, installing a single seismic station could cost upwards of $100,000. Those prices are exclusively affordable to government agencies, research institutions, and industry.

Raspberry Shakes, on the other hand, have basic versions of the same components at a fraction of the price. A Raspberry Shake circuit board costs as little as $100, and it plugs into almost any ethernet or wireless-enabled Raspberry Pi. “We thought that people would have [Raspberry Pis] sitting around in their drawers,” says Christensen, “and we [designed Raspberry Shakes to] support them all.”

A seismic sensor, like a geophone, plugs into the Raspberry Shake board, which serves as an amplifier and digitizer. The sensor’s output comes in the form of voltage differences that must be amplified and converted into a known voltage per velocity. This conversion, called a gain, leaves the output in voltage units, according to Nesbitt, who is also Raspberry Shake’s former chief scientist.

The Raspberry Shake digitizes this information and pipes it to the Raspberry Pi for further processing and archiving. An 8 gigabyte microSD card, which Nesbitt describes as the hard drive of the Raspberry Pi, ships with every Raspberry Shake and comes pre-loaded with all of the Shake software. The Raspberry Pi houses the SD card and provides power for the entire seismic station. “[The Raspberry Pi] is the computer underlying everything,” says Nesbitt.

With a Raspberry Shake board, building your own seismic station from scratch becomes as simple as adding a sensor and plugging the Raspberry Pi into your wall socket, although Christensen recommends crafting an enclosure (you can use Lego bricks!) to protect it from the bumps of the denizens of your household.

If you’d rather not assemble your own from scratch, Raspberry Shake makes several turnkey options based on the number and type of sensors you want. Turnkey options, Hough says, pack all these components into a compact plexiglass box.

The Hollisters chose the turnkey Raspberry Shake 4D, which can be had for under $400. To install, Ryan Hollister says all they needed to do was, “level it and point the axes in the right direction so it’s oriented properly, and plug it in.” Easy as, well… pi(e).

The install process, not so bad after all.

Sensing shakes

“If you imagine the whole gamut of earthquakes from the really small local ones that happen beneath your feet to the big ones that might happen half a world away,” Christensen says, “[all these earthquakes] are losing their high frequency energy as they move away from the source.” If your instrument is limited to high frequencies, by the time the waves of an earthquake arrive, you might not be able to see them. “Really expensive instruments look at all frequencies,” he continues, “and can see everything from local to regional to worldwide earthquakes.”

Raspberry Shakes focus on those higher frequency waves that fade away at great distances, which makes them ideal for detecting local and regional seismic sources, says Nesbitt. However, Shakers (as owners call themselves) can still see waves from earthquakes 10,000 kilometers away, so long as they are at least a magnitude 6.0 event. In fact, Shakers can look at the mobile app, released during lockdown and watch an earthquake roll in, complete with a countdown to the waves’ arrivals.

A countdown as an earthquake rolls in to the location of a Raspberry Shake. Credit: Mike Hotchkiss, Raspberry Shake

Every Raspberry Shake, including the most popular Raspberry Shake 1D, comes equipped with a geophone that measures the vertical component of seismic waves. According to Nesbitt, the frequency range detected by these sensors, between 0.5 to 50Hz, neatly brackets anthropogenic noise sources, which run between 5 to 20Hz. Raspberry Shakes, he says, are “perfect for the seismic noise study,” not only because of their ideal frequency range, but also because their owners often live where data from permanent stations may not be available to non-governmental scientists, such as India.

Screenshot of ShakeNet, showing global distribution of Raspberry Shakes
Screenshot of ShakeNet, showing global distribution of Raspberry Shakes.
Screenshot of ShakeNet, showing global distribution of Raspberry Shakes.

For people who care less about the cultural noise and more about big local earthquakes, Wolin cautions that geophones will saturate in the event of shaking strong enough to feel—the waves will be larger than the detector can register. “You can actually hear this,” says Wolin. “If you pick up your Raspberry Shake and shake it up and down [simulating the jolts of a large earthquake], you can hear the geophone hitting the top and bottom of the case”—it can’t move any farther than that.

To measure violent shakes that can throw you into the air, Nesbitt says that accelerometers, like the sensors that rotate the screen of your smartphone, can stay on scale during shaking up to several times Earth’s gravitational acceleration. Unfortunately, accelerometers fail to measure small ground motions because they produce a high amount of background noise; they only measure what people can feel, says Nesbitt.

Raspberry Shake solves this problem by combining a vertical-component geophone with accelerometers in the Raspberry Shake 4D. For people who live in earthquake country, “you cover all the bases with the same little box,” Hough says.

A seismic revolution?

Considering Raspberry Shake’s price-point, how do these petite instruments compare to their more expensive counterparts? The answer—surprisingly well.

In 2019, scientists at the USGS Albuquerque Seismological Laboratory determined that Raspberry Shakes, though designed for hobbyists, are among low-cost sensors that can enhance existing seismic networks. This may be especially useful in seismically active countries with limited budgets. The ability to strengthen an existing network without needing to spend $10,000 and up per station could be the beginning of a revolution in seismology, says Hough. Adding more stations to an existing network directly enhances Earthquake Early Warning efforts, where denser networks ensure rapid detection of earthquakes.

Wolin, who coauthored the USGS study, says their testing confirmed that the instruments meet manufacturers’ specifications and that the geophones record regional earthquakes well enough. However, she says, “Scientific users need to be conscious of [Raspberry Shakes’] limitations, and not expect them to perform to the same specifications as typical research-grade seismic equipment.”

“There will always be a place for stations that can record everything the ground is doing,” says Hough. Nevertheless, she says, “though a low-cost sensor can’t do everything… they can do an awful lot, especially for practical applications.”

Comparison between seismic data from high end seismometers versus a Raspberry Shake-4D. The Raspberry Shake appears to lag by about 10 milliseconds, a trivial amount to hobbyists, but significant for researchers.
Comparison between seismic data from high-end seismometers versus a Raspberry Shake-4D. The Raspberry Shake appears to lag by about 10 milliseconds, a trivial amount to hobbyists but significant for researchers. Credit: Anthony et al., 2018

Free software

To guarantee rapid, seamless integration of Raspberry Shakes into existing seismic networks, all of them produce output in seismology’s standard format, miniSEED, a stripped-down version of SEED, the Standard for the Exchange of Earthquake Data. This step happens in real time on the Raspberry Pi. According to Nesbitt, “The miniSEED files are broken up by day… but all that data is encoded to miniSEED on the fly.” This data is archived on the microSD card in real time and, if the hardware is connected to the Internet, can be shared via Raspberry Shake’s data center. Collectively, Shakers have made more than 30 terabytes of data from more than 1,000 Raspberry Shakes around the world available to anyone who knows what to do with it. The thoughtful choice in data format ensures that Raspberry Shake data can be processed in the same way as that from academia and industry, opening up a world of options for viewing and analyzing.

Almost anything you can think of to do with seismic data, says Nesbitt, you can do with free software. For example, Nesbitt says, “the USGS has a great piece of free software called Swarm.” Swarm users can see waveforms, create spectrograms, and even triangulate earthquake locations using multiple seismic stations, he says.

Another piece of free software, jAmaSeis, connects students in primary and secondary schools to seismic stations all over the world, including the more than 1,100 Raspberry Shakes that stream in real time. This software “lets students evaluate the data to answer fundamental questions about earthquakes and Earth’s structure,” says Wendy Bohon, an education specialist at the Incorporated Research Institutions for Seismology. Getting seismology in the classroom is a particularly important goal for Raspberry Shake, says Christensen. “I’ve had the joy of going into the classroom and seeing kids using Raspberry Shake… and it is an incredibly gratifying experience," he says.

While Swarm and jAmaSeis have some processing capabilities, Nesbitt says they’re built around visualization. For software focussed on processing, many seismologists—professionals and hobbyists alike—turn to ObsPy, a seismology-specific Python toolbox. ObsPy allows users to pull data from Raspberry Shake’s data center using standard tools with no special installation and filter out unwanted waveforms. Such waveforms could include trains, explosions, or even earthquakes, depending upon the question at hand.

A vibrant community

For many hobbyists, using ObsPy can seem daunting at first, especially without knowledge of Python.

Enter Twitter.

“Shakers form a vibrant, active group on Twitter,” says Christensen, with many scientists helping hobbyists examine their data with ObsPy.

Nowhere has this free flow of information and expertise in seismology been more pronounced than with the anthropogenic noise study, where Lecocq’s initial tweets on Brussels’ seismic lull kicked off a massive collaborative effort. “I thought it was neat to see it start on Twitter with people noticing this phenomenon, and then releasing code,” says Wolin.

The Hollisters plan to explore their Raspberry Shake data in more detail than straightforward visualizations with Swarm. “Like most things, the more you learn, you realize there’s more to it,” says Laura, “and you want to get deeper into it.” Ryan Hollister says he wants to learn ObsPy to eliminate the cigar-shaped train signals. As active “geotweeps” on Twitter, the Hollisters will be able to turn to their fellow seismology enthusiasts for help as they learn.

“Seismologists with a range of experience” respond to questions on Twitter, says Christensen. “When you look at all the stories about cultural noise, and how… they’re seeing signal where they saw noise before, that was born from observations made both by professional scientists and citizen scientists.” He pauses before saying, “They’re exchanging ideas back and forth, and I don’t think that existed before.”

Alka Tripathy-Lang is a freelance science writer with a Ph.D. in geology. She writes about earthquakes, volcanoes, and the inner workings of our planet.

Listing image: Mike Hotchkiss, Raspberry Shake

29 Comments