US scientists have linked tidal cycles to Mt Ruapehu's surprise eruption in 2007 - and even suggest the apparent correlation could offer a new way to predict future blows at volcanoes.

But New Zealand scientists have cast doubt over that idea and questioned the methodology used for the study, which was recently published in a top international scientific journal.

The study indicated how, just before Ruapehu's eruption on September 25, 2007, seismic tremors near its crater became tightly correlated with twice-monthly changes in the strength of tidal forces.

"Looking at data for this volcano spanning about 12 years, we found that this correlation between the amplitude of seismic tremor and tidal cycles developed only in the three months before this eruption," said Tarsilo Girona, a NASA postdoctoral fellow at the Jet Propulsion Laboratory who led the work while based at Brown University.


"What that suggests is that the tides could provide a probe for telling us whether or not a volcano has entered a critical state."

Earth's tides rose and fell daily due to the gravitational tug of the moon as the Earth rotated.

In full and new moons, the lunar gravitational pull lined up with that of the sun, which made the daily tidal bulges a little larger during those moon phases.

During the first and third-quarter moons, the daily tidal bulge was slightly smaller, and this twice-monthly change in tidal amplitude was sometimes referred to as the fortnightly tide.

While tides were generally thought of as rising and falling waters, gravitational stresses could also affect the planet's solid crust.

"A lot of research has been focused on whether or not tidal forces can trigger eruptions, and there's no definitive evidence whatsoever that they do," said study co-author and Brown University professor Christian Huber.

"We wanted to take a different angle with this study and look at whether there's some detectable signal associated with tidal forces that can tell us something about a volcano's criticality."

The researchers chose to study Ruapehu partly because its activity has been closely monitored for years by GNS Science, and were particularly interested in data from seismic sensors located near the crater.


The team combed through 12 years of seismic data, looking for any period when the seismicity was correlated with lunar cycles.

They found that for most of that period, there was no correlation between tremor and lunar cycles, except the few months before 2007's steam-driven eruption.

While the fluctuations in seismic amplitude were subtle, the strength of the correlation to the tidal cycle was as strong as 5 sigma, the researchers say, meaning that the probability that pattern arose by chance was about one in 3.5 million.

To understand how tidal forces may have been affecting Ruapehu during those three months, the researchers used a model of seismic tremor that they had developed previously.

Volcanoes like Ruapehu had a vertical conduit through which lava rises, and a solid rock plug at the top.

The Dome Shelter, near the crater lake, was bombarded with mud and boulders during the eruption. Mountaineer William Pike was inside at the time and lost his right leg. Photo / File
The Dome Shelter, near the crater lake, was bombarded with mud and boulders during the eruption. Mountaineer William Pike was inside at the time and lost his right leg. Photo / File

Gases released from the lava formed a pocket between the rocky plug and the lava pool, and could resonate against the plug, which creates seismic tremor.

Their model suggested that when the pressure of the gas pocket reaches a critical level — a level at which a steam eruption was possible — the differing stresses associated with changing tidal forces were enough to change the amplitude of tremor.

"That's what we think was happening in 2007," Huber said.

"When the pressure in the system became critical, it became sensitive to the tides. We were able to show that the signal is detectable."

Huber said the team would like to collect more data from other eruptions and other volcanoes to see if the tidal signal showed up elsewhere.

"Then we can start to think about using it as a potential means of predicting future eruptions of this kind."

The study, funded by the US National Science Foundation, was published in the journal Scientific Reports.

New Zealand experts approached by the Herald held some concerns.

Professor Mark Bebbington, a geostatistician at Massey University-based Volcanic Risk Solutions, said the study was "interesting" but fell short of what he considered a robust forecasting method.

Many choices - including the volcano used and the type of waveform analysed - had been made in assembling the model, each of which could have affected the calculated significance levels.

"The 'standard procedure' for obtaining the daily median seismic amplitude has a lot of specific steps, to which the result may or may not be sensitive," he said.

"The paper notes that it is sensitive to the length of the window used."

The method also looked to have been tailored to the data, Bebbington said.

"It next needs to be verified by application to an independent data set, without any alterations, and show similar results.

"I should emphasise that this is a standard development path – first try and find something by exploratory analysis, then demonstrate that it is real.

"This is still a step short of proving that the method can actually forecast eruptions, as it is not yet specified in a usable form."

Professor James White, head of Otago University's Department of Geology, said the inferred effect seemed limited to "a very narrow set of circumstances".

An underlying challenge, he added, was that the explosion mechanism for steam-driven or "phreatic" eruptions remained, at best, poorly described.

"The paper considers the mechanism to be one of gas trapping driven by changes in rock permeability, but doesn't directly address the strength of this rock."

White said if it was simple to keep the model running to see if there was a period when the pattern of Ruapehu's shallow seismicity correlated with model predictions, it could be worth doing.

But, White said, with only two events addressed by the paper - and one of them not "predicted" by the model - he wasn't confident that it would have any predictive value.