A preliminary analysis of the atmosphere and ionosphere over Japan in March reveals infrared and electron anomalies coincident with the Tōhoku earthquake, researchers in the US and Russia claim. The anomalies are the latest evidence for a possible link between seismic activity and changes in the atmosphere or ionosphere, although sceptics believe they are unrelated.

Seismologists have searched for early-warning signals of earthquakes for more than a century. These range from small tremors in the ground, to aurora-like lights in the atmosphere and even to bizarre animal behaviour. But despite a few records of such incidents coming before quakes – usually noted retrospectively – there has never been any consistent method to accurately predict when a major shock is going to happen.

Tell-tale signs?

Many scientists still monitor various parameters around quake-prone regions in the hope that they will improve forecasting, or perhaps open up avenues towards prediction. These parameters include infrared emissions in the upper atmosphere and the total electron content (TEC) of the ionosphere – the part of the Earth's atmosphere between altitudes of 80 and 1000 kilometres that is made up of electrons and ions. Changes in both infrared emissions and TEC are known to occur for non-seismic reasons: the infrared varies with cloud cover, for instance, while TEC gets a boost during heightened solar activity. Yet researchers have still claimed that they can pick out anomalous behaviour in the infrared and TEC that coincided with various past quakes, such as the 2008 Sichuan earthquake in China and the 2010 Haiti earthquake.

Now Dimitar Ouzounov of Chapman University in Orange, California, and colleagues claim to have evidence of anomalous infrared and TEC signals shortly before the magnitude 9.0 earthquake that struck off the coast of Tōhoku, Japan, on 11 March this year. The researchers believe that the apparent anomalies could be evidence that major seismic activity is preceded by a release of radon gas that ionizes and heats the surrounding air.

Ouzounov's group retrospectively analysed four parameters: the Earth's outgoing infrared radiation, using satellite imaging; the ionosphere's TEC, calculated from global positioning satellite signals; the cross-section or "tomography" of the ionosphere, using data from low-Earth-orbit satellites; and the density of upper-ionosphere electrons, calculated from signals taken at four Japanese ground-based ionosonde stations. The infrared data was analysed for the month of March over a period of eight years – from 2004 to 2011 – while the ionospheric data was analysed only for around the time of the Tōhoku quake.

The researchers found what they say is the first indication of an infrared anomaly on 8 March 2011, three days before the quake. By 11 March, the day of the quake, the location of the maximum infrared emission apparently fell exactly over the quake's epicentre. Meanwhile, they also found an increase in electron density, reaching a maximum on 8 March. This day also showed an abnormal variation in TEC over the epicentre, according to the findings. On 3–11 March the ionosondes recorded a "large increase" in electron density.

Supporters and sceptics

"The results are interesting for me even though the physical mechanism is not clear," says Katsumi Hattori, a geoscientist at Chiba University in Japan. "My opinion is that their approach is one of the hopeful ways to forecast seismic activity. I think the prediction – when, where and what magnitude – is difficult, but the monitoring of [infrared emissions] and TEC can provide information for seismic activity. They are just the same as the parameters for a weather forecast."

Yet many seismologists are sceptical about the benefits of such analyses, believing that it is easy to find correlations when data is taken selectively. Ian Main, a seismologist at the University of Edinburgh in the UK, says signals in the atmosphere and ionosphere "fluctuate all the time, and it would be surprising is some fluctuation did not occur around the time of the earthquake". He adds, "One of the things you can predict about earthquakes is that following the event there will be claims of precursory behaviour identified in retrospect."

Thomas Heaton, a seismologist at California Institute of Technology in the US, is also sceptical of prediction. "Through the years I have seen dozens of reported anomalous geophysical signals," he says. However, we have yet to discover a precursor to an earthquake that reliably produces a significant signal before it occurs. "In fact, the more we look, the more it seems as though a large earthquake starts similarly to a small earthquake," he adds, explaining that due to the similarities, even an advance signal would not help to judge the intensity of an upcoming earthquake.

Still, Ouzounov and his group are hopeful that their work will help both forecasting and prediction. Ouzounov told physicsworld.com that they have listed more than 100 earthquakes during the last decade and have discovered a "systematic appearance of atmospheric and ionospheric signals in the same time frame we have shown for the Tōhoku earthquake".

The preliminary results are available on the arXiv preprint server.