On the night of 16 June 2023, about 1.2 million cubic meters of rock rumbled down into the valley near Brienz (GR). A team from the Swiss Federal Institute for Forest, Snow and Landscape Research WSL and ETH Zurich tracked the event using an unusual method: they detected the shock waves on underground internet fiber optic cables.

Ground waves lead to extremely small strains and compressions in the optical fibers. Using a method called Distributed Acoustic Sensing, or DAS for short, researchers can measure these deformations in real time and even determine their origin in the fiber with an accuracy of a few meters. As optical fibers are often many kilometers long, the method is extremely interesting for monitoring natural hazards from a distance.

Using existing infrastructure

To do this, the researchers need a dark fiber, i.e., an unused fiber in a telecoms cable. They connect a device to it, the interrogator, which sends laser pulses through the dark fiber. If the fiber is minimally deformed somewhere, the pulses come back altered.

The method can be used wherever fiber optic cables for communication are buried in the ground—which is the case in many places in Switzerland, for example along railway lines. The WSL-ETH team led by Fabian Walter, a seismologist at WSL, has already successfully recorded avalanches along the Flüela Pass road using this method.

The rock movements in Brienz now provided a unique opportunity to test the method for rockfalls: the mountain was meticulously monitored with radar equipment and seismometers before and during the rockfall, and the company Swisscom Broadcast AG gave Fabian Walter access to a fiber optic cable that runs between Tiefencastel and Filisur.

The researchers sent laser pulses through this cable for 45 days using an interrogator from the Seismology and Wave Physics Group at ETH Zurich until the landslide began on the night of 15 to 16 June. "The measurements exceeded our expectations," says Walter. "We were able to measure hundreds of small rockfalls before the big event, and the big fall anyway."

The most difficult thing about fiber optic detection is filtering out the signals they are looking for from the countless other vibrations caused by trains, traffic or rivers. To this end, WSL doctoral student Jiahui Kang used artificial intelligence to develop an algorithm that automatically recognizes the signals.

It was able to correctly identify 95% of the rock movements, the researchers report in the journal Geophysical Research Letters. The radar measurements of the company Geoprevent, which monitored the Brienz rockslide area, were used for comparison.

Vast amounts of data

So what stands in the way of widespread use of the fiber optic method if it works so well? According to Fabian Walter, there are several hurdles: The computational models to filter out the desired signals from the noise are only at the research stage. The measurements also accumulate enormous amounts of data—which can be several terabytes per day. "You first have to learn how to analyze the data in real time," says Walter.

Last but not least, the interrogators are still very expensive at more than CHF 100,000 (~ 115,000 USD), but the trend is downwards. Walter expects these problems to be solved in the next few years, stating, "This information is too valuable not to use it."

The fiber optic method can potentially monitor rockfalls, avalanches, earthquakes and debris flows with local precision and over long distances. All that is needed is a fiber optic communication network—and this continues to grow worldwide.