Early detection of earthquakes could be vastly improved by tapping into the world's internet network with a groundbreaking new algorithm, researchers say.
Fibre optic cables used for cable television, telephone systems and the global web matrix now have the potential to help measure seismic rumblings thanks to recent technological advances, but harnessing this breakthrough has proved problematic.
A new paper published today in Geophysical Journal International seeks to address these challenges by adapting a simple physics-based algorithm to include fibre optic data that can then be used hand-in-hand with traditional seismometer measurements.
Not only could this "exciting" advancement be integrated into existing earthquake early warning systems, it may also help detect seismic activity associated with erupting volcanoes, geothermal boreholes and glacier icequakes.
"The ability to turn fibre optic cables into thousands of seismic sensors has inspired many approaches to use fibre for earthquake detection. However, fibre optic earthquake detection is not an easy challenge to solve," said lead researcher Dr Thomas Hudson, a senior research scientist at ETH Zurich.
"Here, we lean on combining the benefit of thousands of sensors with a simple physics-based approach to detect earthquakes using any fibre optic cable, anywhere.
"Excitingly, our method can combine fibre optic and traditional seismometer measurements, allowing fibre optic sensing to be included in existing earthquake early warning systems."
Distributed acoustic sensing (DAS) is a nascent technology that uses fibre optic cables to detect acoustic signals and vibrations. It can be used to monitor a variety of things, including pipelines, railways or the subsurface.
It therefore has the potential to turn fibre optic networks – which carry data super fast – into measurements of seismic activity that can be used to detect earthquakes.
This is tantalising because fibre optic networks are ubiquitous in populated regions and even cross oceans, providing the possibility of far more detailed and effective seismic monitoring networks than those that currently exist.
Turning this potential into reality, however, is a much trickier proposition.
Real-world fibre network geometries are often complex – and seismologists have no control over the geometry. On top of this, fibre optic cables are often located in noisy urban environments, making it difficult to differentiate between earthquake activity and other sources in the way traditional seismometers do.
Another challenge is that DAS measurements are only sensitive to strain in the axis of the fibre, whereas seismometers measure 3D ground motion. This makes surface fibre optic cables far more sensitive to slower S-waves (which travel only through solids and are the second waves to arrive during an earthquake) than faster P-waves (which travel through liquids and solids), meaning it is tougher to detect earthquakes and locate them.
One solution is to combine information from both traditional seismometers and fibre optic cables to detect earthquakes, but this isn't easy because of the different instrument sensitivities and measurement units.
The other issue is that turning a fibre optic cable into thousands of sensors generates a lot of data. Processing this data in real time is essential for earthquake monitoring, so efficient data processing algorithms are required.
The new algorithm works by taking the energy observed at receivers – either fibre optic cable channels and/or seismometers – and migrating that energy back through space and time to find a coherent peak in energy corresponding to a potential earthquake.
This approach was also found to be effective in detecting earthquakes at erupting volcanoes, geothermal boreholes and glacier icequakes.
"A key strength of this physics-based approach is that it works well even in noisy environments, since noise is generally less coherent than an earthquake signal," said Dr Hudson.
"It can also be applied out-of-the-box to any fibre network."
He added: "Although we don't claim to have completely solved the large data volume issue, we present pragmatic ways to deal with this and our algorithm runs in real time for the datasets tested.
"The method is provided open-source, so that the wider seismology community can immediately benefit."
ENDS