Divining earthquakes with water
Changes in groundwater could predict temblors, Cathal O'Connell reports.
Japan’s 1995 Kobe earthquake killed more than 6,000 people. No one saw it coming. But now a team of Icelandic scientists may have come up with an earthquake early-warning system based on changes in the composition of groundwater, Nature Geoscience reported this September.
Until now geologists could only make coarse predictions of when and where a quake might strike. Earthquakes arise deep in the earth’s crust when the sections jammed together along a fault suddenly give, releasing strain that has built up over time. GPS can map tiny movements along the fault lines – as little as 5 mm – to predict the sites most likely to snap, but as far as predicting when that happens it could be tomorrow, a hundred years, or more, explains Brian Kennett, a seismologist at the Australian National University. Not very useful information for a major city.
The researchers detected groundwater changes six months before each of two magnitude-five quakes that rocked the scenic Hafralækur region near the north coast of Iceland. Alasdair Skelton from Stockholm University and his team detected spikes in five separate chemicals: hydrogen, oxygen, sodium, silicon and calcium.
It’s not the first time people have thought of checking groundwater chemistry as an earthquake predictor, notes Kennett. After the Kobe earthquake seismologists found data had been inadvertently collected for them. In the months before the quake a Kobe factory had been bottling spring water with the collection dates meticulously recorded on the bottles. Analysing the water provided hints that the concentrations of chloride and other ions had changed in the lead-up to the quake. But only hints.
So Skelton and his colleagues set up a sampling program to monitor groundwater chemistry in Iceland, an earthquake hotspot. Starting in 2008 the team took regular water samples from a 100-metre-deep borehole.
“The important feature of the new results from Iceland is that they track through two successive medium-size earthquakes,” says Kennett. This reduces the chance that the groundwater changes were merely coincidence to less than one in 100,000, Skelton calculates.
So what’s the explanation? “We are probably detecting mixing between different groundwater sources,” says Skelton. His theory is that accumulating strain in the rocky plates opens up tiny cracks allowing separate bodies of water, each with their own chemical profiles, to mix.
Skelton admits that the specific groundwater changes observed are probably unique to the local area. He makes no claim about being able to predict earthquakes yet but hopes the new results are compelling enough that other scientists will start monitoring groundwater chemical changes in earthquake-prone zones. Kennett has no doubt that they will. “The paper will encourage examination of long duration records in groundwater in other environments, and we will then see if the results are transferable to other regions,” he says.