Scientists use distant sensors to monitor American Samoa earthquake flocks

From late July to October 2022, residents of the Manua Islands in Samoa, USA, had their planet shaking several times a day, causing concerns about an imminent volcanic eruption or tsunami.

Nothing appeared in the area’s earthquake catalogues as the islands lacked earthquake monitoring networks that could assist seismologists who sway in searching for sources of earthquake swarms.

However, residents of Tau, Ou, and the Orosega Islands needed answers, so Clara Yun and her colleagues from the US Geological Survey found another way to fill the seismic gap. They used machine learning and another technique called template matching with shaking data recorded from a single seismic sensor 250 kilometres away from the American Samoa herd.

In earthquake records, Yoon and his colleagues explain how to track flocks using these single station data until a permanent local earthquake station was installed in Samoa, USA in August and September 2022. I’m sharing it.

The unblocking volcanic earthquake swarm began in July 2022 about 15 kilometres off the coast of Koh Tao. As Pacific tectonic plates move over hotspots in the South Pacific, the Samoa Volcanic Islands arise.

Reports of frequent shaking, occurring multiple times a day for a few seconds at a time, initially the only information about the herd.

“When the earthquake began, Samoa, USA had no geophysical monitoring of equipment, so even basic information about the source of shaking with impacts on emergency decision-making and public safety was not present. “Yun said.

To remedy this, researchers turned to remote earthquake stations at UPOLU in Samoa, part of the global seismography network. It has data that can be downloaded at almost realistic times from the EarthScope data center.

Although earthquake signals from the American Samoa earthquake flocks were difficult to detect at far stations, Yun and his colleagues chose to select these small earthquakes from Noisy, along with a technique called template matching, and eqtransformer We used a deep learning model called Earthquake background.

“Eqtransformer discovered many earthquakes in places that coincide with Samoa in East America, and its largest region coincided with the era of Felt’s report,” Yoon explained. “These felt reports donated to the National Weather Service by local residents in Samoa, USA, are important sources of earthquake data, and the events detected by EQTransformer are the same earthquakes that locals actually felt. It gave me confidence.”

This new seismic catalogue of events allowed researchers to characterize the onset and peak of herd activity. The portable and inexpensive raspberry shake sensor deployed in August 2022 helped us quickly find the area of ​​the herd.

The herd ended in October 2022 without erupting, but it is likely linked to the volcano magmatic movement, the researchers concluded.

Yoon points out that approaches like single station technology may be useful elsewhere around the world where permanent earthquake monitoring is sparse and earthquake risks are less understood. did.

She adds that the largest earthquake in the American Samoa herd is 4.5 in size, making it unlikely that it was detected by the global seismic network.

“If no one lived nearby to report that he was shaking frequently, this American Samoan herd might have been completely unaware,” Yun said. “We are waiting for many unknown seismic sources and phenomena to be discovered through the future, large-scale, comprehensive application of deep learning approaches in seismology.”