Scientists have edged closer to understanding the link between New Zealand's largest geological threat and a mysterious earthquake phenomenon detected only in the last two decades.
What are called silent or "slow-slip" earthquakes can last from days to years, and can produce up to tens of centimetres of displacements along faults without seismologists even realising it.
In the wake of Japan's devastating Tohuku earthquake and tsunami in 2011, scientists trawled through data from thousands of quakes in the region to find one of these deep-seated, almost imperceptible events may have played a part.
Other studies have tied these slow slip episodes to two big quakes in 2014: slow slip events were thought to precede the 8.1 Iquique earthquake in Chile, and a 7.2 shake off the coast of Mexico hit just two months after a slow slip started.
Earlier this month, researchers reported how the slowest earthquake ever recorded - lasting 32 years - eventually led to the catastrophic 1861 Sumatra earthquake in Indonesia.
When one of these slow-slip quakes is under way here in New Zealand, scientists are now paying close attention.
Over recent years, a major international research effort has focused on how they play out within the Hikurangi Subduction Zone, where they were first observed using GPS in 2002.
Part of a bigger 3500km Hikurangi-Kermadec-Tonga system that fans up into the Pacific, this largely-offshore margin is where the Pacific Plate dives – or subducts – westward beneath the North Island.
If we drained the ocean, it would appear as an immense mountain range, rising from the seabed, not far from the East Coast.
Like other subduction zones, it's been shown to have unleashed quakes greater than 8.0 in magnitude - and possibly as high as 9.0 - with the largest ones happening around every 550 to 1400 years.
Buried around the Marlborough's Wairau Lagoon area was evidence of two tsunamis: one that occurred between 800 and 900 years ago, and another that struck around 500 years ago.
In Hawke's Bay – bang in the middle of Hikurangi's margin – there were signs of nine quakes in 7000 years that were big enough to have caused the land to subside by a metre or more.
Now, in a study just published in Nature Geoscience, scientists combined supercomputers with seismic "CT scans" to test several theories surrounding local slow quakes and subduction zones.
"Subduction zones are the biggest earthquake and tsunami factories on the planet," said the study's co-author Dr Laura Wallace, of GNS Science and the University of Texas Institute for Geophysics (UTIG).
"With more research like this, we can really begin to understand the origin of different types of earthquake behaviour at subduction zones."
Wallace believed that understanding the timing and likely location of the next large subduction zone earthquake could only happen by first solving the mystery of slow-slip events.
"One of the theories we tested – called rate-state friction – failed to hold up as well as expected," she said.
"That means we can probably assume there are other processes involved in modulating slow-slip events, such as cycles of fluid pressurisation and release."
The study's lead author, Adrien Arnulf of the University of Texas (UT), said the seismic imaging data of the Earth's sub-surface used in the research was gathered off the southern Hawke's Bay coast in 2005 as part of a Government-funded survey.
Applying next-generation techniques to this data had provided a whole new view of parts of the Hikurangi subduction zone.
"It was turned into detailed images by using similar techniques to those used in medical imaging, so geoscientists could visualise the interface between the Pacific and Australian tectonic plates to figure out what's going on underground," he said.
Scientists then used a supercomputer at the Texas Advanced Computing Centre to look for patterns in the data.
The study, funded by UTIG and an MBIE Endeavour fund grant, showed how weak the fault had become and where pressure was being felt within the joints in the Earth's crust.
He worked with UT Jackson School of Geosciences graduate student, James Biemiller, who used Arnulf's parameters in a detailed simulation he had developed for modelling how faults move.
The simulation showed tectonic forces building in the crust then releasing through a series of slow-motion tremors, just like slow-slip earthquakes detected at Hikurangi over the past two decades.
According to the scientists, the real success of the research was not that the model worked but that it showed them where the gaps are in the physics.
"We don't necessarily have the nail-in-the-coffin of how exactly shallow slow-slip occurs," Biemiller said.
"But we tested one of the standard nails - rate-state friction - and found it doesn't work as well as you'd expect.
"That means we can probably assume there are other processes involved in modulating slow slip, like cycles of fluid pressurisation and release."
Wallace said the study had ultimately enabled scientists to narrow their focus on the physical conditions in the fault where slow-slip events occurred - providing an important test of a range of theories for why they happen.
Arnulf said if scientists ignored slow-slip events, there was a risk they would miscalculate how much energy was being stored and released as tectonic plates moved around the planet.
"They are an important part of the earthquake cycle because they occur in the same places as high magnitude earthquakes, and release tectonic energy slowly over weeks to months."
He added that the Hikurangi subduction zone was ideal for studying slow-slip quakes because they occurred shallow enough to be imaged at high resolution using seismic techniques.
This aspect had therefore attracted numerous international science agencies to New Zealand, bringing sophisticated technology and know-how to focus on our highly active plate boundary.