Local populations excavating the site of a landslide in Yambali, Papua New Guinea, in
Xinhua / Alamy
When a dangerous landslide strikes, the rescuers are blurring to help to know exactly where to look for – and now a new technique for analyzing seismic data can do exactly that, tracing the source of the disaster a few kilometers in seconds.
Existing methods can only refine the locations less than tens of kilometers, explains Stefania Ursica to the Helmholtz Center for Geosciences in Germany. In remote areas, this can cause delays if the rescuers are sent to the wrong place. “This loss of time can be crucial,” Ursica told a press conference at a meeting of the European Union of Geosciences in Vienna last week.
Many countries have a network of seismic monitors to record the activity linked to earthquakes and volcanoes. This data can also be used to detect events such as landslides – the risk of which increases due to climate change – but data from this type of event is much more disorderly and more difficult to analyze than those of earthquakes, explains Urica.
There are two new aspects as she approaches her team, she said. The first is to analyze five different aspects of the seismic wave form to choose in the noise exactly when an event occurred.
This information is then transformed to a dozen mathematical agents who seek the location of the event, such as the initial rock fall leading to a landslide. They do this by considering what wave form would have been produced if the event had occurred in a particular place and comparing it to the registered wave shape. If that doesn’t match, they try a different place.
Each agent “moves” in a scheme inspired by the behavior of animals, from the spiral of a hawk to the long migrations of elephants, until they have collected at the most likely location of the event. The entire process only takes about 10 seconds and is much more precise than previous approaches. “We essentially have an order of magnitude [of] Improvement, ”explains Ursica.
In addition to helping rescue efforts, this will help researchers locate events in distant areas where satellite data is not clear or unavailable, she says: “We can identify the events that we could otherwise see.”
The team plans to publish details soon and put the code available for others.
Subjects:
