In the early hours of March 5, 2021, a few seconds before the shaking started, Graeme Atkins woke to yodels from outside—wild pheasants calling in the darkness. Soon his home was shaking so hard that it was difficult to stand up. He imagined Rūaumoko rattling the land. The god of earthquakes was angry.
Once the shaking settled, Atkins grabbed a torch, jumped into his ute and headed down a gravel road to check on each whare along his stretch of the East Cape, between Tikitiki and Ruatoria. The houses had stood up well, apart from some broken crockery and spilled shelves, and Atkins, a Civil Defence responder, knew these communities were far enough inland to be safe from a tsunami.
But as people went back to bed, his two-way radio crackled with messages from emergency responders at Te Araroa on the coast. There, locals were heading for higher ground. No tsunami alert had been issued, but people knew the Civil Defence message about earthquakes: “Long or strong? Get gone.”
In Wellington, Laura Wallace had been shaken awake by the jolt. Moments later, her phone pinged with a text message from duty staff at the 24/7 National Geohazards Monitoring Centre, summoning her to an online meeting. Wallace, a geophysicist, threw on a sweatshirt and joined other tsunami and earthquake experts to advise the National Emergency Management Agency (NEMA) on whether to issue a tsunami warning. The information they had to go on: earthquake wave patterns gathered GeoNet’s network of seismometers on land, and wave-height data from tsunami-monitoring buoys stationed along the massive plate boundary off the coast of Gisborne.
At 3.28am, an hour after the earthquake, NEMA issued a tsunami threat warning for coastal beaches and some areas further inland. Eight minutes later, mobile-phone alerts went out to people along the East Cape, urging those who hadn’t already left to evacuate immediately. At 5am, the threat was downgraded to a warning of “strong and unusual currents and unpredictable surges” at the shore.
By 6am, people were told the threat had passed and they could return to their homes—but not for long. It was still dark at 6.51am, when the Pacific Tsunami Warning Center in Hawai’i detected a magnitude-7.4 earthquake near the Kermadec Islands, followed at 8.28am by a magnitude-8.1 rupture.
By that time, NEMA’s emergency team had gathered in the bunker under the Beehive, New Zealand’s parliament buildings, and another evacuation notice was issued. People on the west coast of Northland—from Cape Reinga to Ahipara—and all the way down the east coast to the Bay of Plenty and the East Cape were told to head back into the hills, or further inland.
This time, evacuees had to wait most of the day before receiving the all-clear. At 3.45pm, NEMA cancelled all tsunami warnings. “It’s hard not to feel our country is having a run of bad luck,” Prime Minister Jacinda Ardern said later that afternoon, while emergency management minister Kiritapu Allan thanked people for acting quickly. If a tsunami had been triggered, there wouldn’t have been time to wait for the official warning.
New Zealand’s tsunami risk is at least as high as our earthquake risk. About 80 per cent of tsunami follow an earthquake, while others are triggered by underwater volcanic eruptions or landslides. Since 1840, about ten tsunami with waves higher than five metres have struck New Zealand. No part of the coastline is out of danger, but the North Island’s east coast faces the largest risk—and this is because of a long faultline that runs beneath the ocean floor.
Here, two massive tectonic plates—the Australian plate from the west and the Pacific plate from the north and east—bulldoze into each other. As they shove and jostle, the Australian plate pushes on top of the Pacific plate, which in turn dives down, at a gentle angle at first, then it drops off more steeply into the Earth’s mantle. This is the Hikurangi subduction zone, and it runs all the way from the East Cape down to Marlborough. To the north of the East Cape, the subduction zone continues to the Kermadec Islands and Tonga.
Less than a week before the March 5 quakes, I had joined Wallace on a research voyage to the Hikurangi subduction zone to deploy pressure-monitoring instruments to the seafloor off Hawke’s Bay. Soon after the quakes, Wallace set off again, this time aboard the research ship Tangaroa, with a remote-controlled submersible from Canada. It would retrieve similar instruments and download data from two undersea observatories monitoring changes in the Earth’s crust close to the fault.
The Hikurangi subduction zone is New Zealand’s biggest, most active, fastest-moving fault. It has the potential of rupturing in megathrust earthquakes of magnitude 8 or more, where the seafloor lifts instantaneously, setting off waves that carry the kinetic energy outward in all directions. These waves could hit the coast in minutes.
Off the East Cape, the plates grind past each other in what amounts in geological terms to a sprint—about six centimetres a year. At the southern end of the subduction zone, off the coast of the Wairarapa, the plates slow down to about half the speed: two or three centimetres per year. This is similar to the rate of movement along the Alpine Fault in the South Island, where the same plates collide head-on rather than subduct.
“Having that really big surface area combined with the very high rates of movement that are being accommodated on the plate boundary, you have more potential for generating large earthquakes,” says Wallace.
Plate subduction zones like Hikurangi fringe the entire Pacific Ocean. They’ve produced the world’s largest earthquakes—including the magnitude-9.1 earthquake that struck off the north-eastern coast of Japan on March 11, 2011. More than 18,000 people lost their lives in that quake or in the tsunami waves that followed. In some places, 40-metre walls of water travelled ten kilometres inland, devastating cities and triggering a meltdown at the Fukushima nuclear power plant.
The Hikurangi subduction zone has the potential to produce earthquakes of a similar magnitude. “But it’s a source that we don’t understand very well,” says Wallace. “We’re trying to change that.”
Wallace and her colleagues at GNS Science lead a $7 million Endeavour Fund programme to get a better understanding of the risk. What they already know is that the subduction zone behaves differently at each end. Under Wellington and the Wairarapa, the plates are stuck, accumulating stress that geoscientists expect will eventually be released in a large earthquake of magnitude 8 or more. Further north, off the coast of Hawke’s Bay and Tairāwhiti, the plates appear to be creeping past each other without accumulating as much stress. Here, slow-motion earthquakes release the same energy over weeks or months as a high-magnitude quake would in minutes. One of the research team’s biggest questions is why we see two such extreme fault behaviours along this and other subduction zones.
“This represents an opportunity to figure out why some subduction zones lock up and slip in big earthquakes, while others just tend to creep steadily,” says Wallace. “This could be one of the key places in the world to unlock that question—which matters for everyone around the Pacific Rim. My big question is to figure out why subduction zones slip in major earthquakes, and which places we need to be the most worried about.”
Hikurangi is one of the world’s most accessible subduction zones, drawing scientists from around the world. Elsewhere, subduction zones are much further out in the ocean or buried underneath continents. That’s why Hikurangi is one of just three places that have been closely scrutinised for the last decade as part of a US National Science Foundation programme. (The other places are Cascadia, off the west coast of North America, and the Aleutian subduction zone off Alaska.)
Slow-slip earthquakes were first discovered at the Cascadia subduction zone, which runs underneath marine geologist Evan Solomon’s home town of Seattle. There, scientists can use only remote-sensing technology to get a better idea of what that plate boundary looks like. “We can’t sample it or do near-field experiments trying to better understand why slow slip occurs,” Solomon says.
The Tangaroa voyages were his third expedition to the Hikurangi subduction zone. The first time he visited was in 2018, on board the United States research vessel JOIDES Resolution, which was brought to New Zealand waters to drill holes hundreds of metres deep into the seafloor. Two permanent observatories were embedded within the boreholes, nvolving different packages of equipment positioned at each at a specific depth to measure changes in strain, chemistry, temperature and the movement of water through the fault zone.
“The observatories are giving us far more information than we’ve ever had before on how these events evolve, their timing and migration along the offshore plate boundary, and how this relates to earthquakes occurring during the slow-slip events.”
One observatory is near the Hikurangi Trench, where the two plates meet on the seafloor, 2500 metres under the water. There, it runs through the active Pāpaku Fault, one of a number of splay faults that run off the boundary. The other observatory is closer to land, at 1000 metres deep, above a region with frequent slow-slip earthquakes.
Solomon is particularly interested in how water contributes to the build-up of tension along faults. Plate movement at a seismic fault is controlled mostly by friction, but the pressure of liquid in tiny spaces, or pores, in the rock is also important. Imagine a game of air hockey, he says. If the air is off, the puck is hard to move. But if the air flow is on, it slides easily. “That’s the same thing with the fault zone. And one of the hypotheses about what controls both slow-slip events and earthquakes is the change in pore pressure.”
Pressure builds up, and the fault becomes slippery and slides. This causes the water to drain out of the fault, which means the pressure is lost, and the fault locks shut again.
The underwater observatories are studded with instruments that continuously take measurements along the depth of the borehole: pore pressure in the rocks, temperature, and the flow of water. Based on her analysis of data downloaded from the undersea observatories, Wallace has found that the March earthquakes triggered a slow-slip event near the trench that the land-based GeoNet network could not detect.
“The part that I’m the most interested in is how the crust deforms, or distorts, during these slow-slip events,” she says. “When they happen, you can imagine some parts of the crust get squeezed, in which case you’d measure an increase in pressure, and some parts get pulled apart.”
Before this, Wallace had to track slow-motion tremors through millimetre-level movements of GeoNet sensors on land. Now, these ocean-floor observatories put her right on the scene to untangle how and why this type of gradual earthquake happens—and, more importantly, whether there is a link between these slow, silent Earth shivers and large, sudden jolts.
At the end of summer in 1947, Don Tunnicliffe and his wife, Novena Tunnicliffe, were on their honeymoon at Tatapouri Point, north of Gisborne, staying with friends.
Around 8.30am on March 26, an earthquake struck about 50 kilometres offshore on the Hikurangi subduction fault. It registered at magnitude 7.1, but the shaking didn’t cause any damage—until half an hour later. Don heard a great rumbling sound, and went outside to see what was happening. He watched a towering wave roll towards the coast. “Approaching the shore, and us, at breakneck speed and roaring like an express train was a wall of dirty coloured water towering a good 30 feet, boiling and curling as it picked up acres of beach sand on its way to engulf us,” recalled Don in an oral history collected by Hawke’s Bay Civil Defence.
He watched as the wave picked up a young man who shot past “like a spinning top”, then the water reached Don. “I also disappeared under the swirling, rolling and now seething mass of water, sand and seaweed.”
Two successive waves drove through the house and ran hundreds of metres up into the hills, only to retreat back towards the ocean loaded with debris. Don “could only gaze in stupefied wonder as sheds were picked up as though by a giant hand and smashed down on to several feet of water”. The young man who Don had seen was slammed against a bank, while Don “became entangled in the top strands of the barbed-wire fence, a good six feet under in a world of blackness”. Everyone survived.
On May 17, less than two months later, another earthquake triggered a six-metre tsunami that hit the coast between Gisborne and Tolaga Bay.
The Gisborne tsunami waves of 1947 were among the biggest recorded in New Zealand, but geologists know that the Hikurangi subduction zone has produced much stronger ruptures in the past. These megathrusts, and the tsunami they generated, have left their mark on the land.
In a large storage room at the back of GNS Science in the Hutt Valley, palaeoseismologists Charlotte Pizer and Kate Clark show me a series of mud-filled tubes sliced in half. These are one-metre cores extracted from several sites close to the shore of Lake Grassmere, near Blenheim, where the Hikurangi subduction zone changes gradually into the complex Marlborough fault system. If you drive past this shallow coastal lake today, it shimmers pink as warm north-westerlies evaporate seawater to produce salt. In the past, however, it was a partially enclosed lagoon.
As I look at the long strands of mud, I can see the clear segregation between silty layers and narrow bands of almost completely intact shells. “Something has come in quite rapidly and… deposited [the shells] here quite quickly,” Pizer says. “We know that because the shells have been preserved really well—they are all intact and some of them are articulated. The shells were closed and look like they’ve been picked up from a range of places, then deposited rapidly.”
The shells have been thrown together from different ecological habitats—another indication that this was the work of a large tsunami. None of the shells show up in the silt layers above or below. “They’ve been washed in from elsewhere.”
Pizer found the same tsunami signature further inland. She could determine how far the wave travelled by seeing how far the shell deposits stretched. “You can literally put down the gouge auger and pull it back up and the same deposits are replicable across 1.7 kilometres.”
Pizer was able to date the deposits—one band of shells originated about 2000 years ago, another about 1400 years ago—and she says the most likely tsunami triggers would have been ruptures along the southern end of the Hikurangi subduction zone.
The Lake Grassmere cores complement samples Clark collected a little further north at Wairau Bar. Together, these samples tell the stories of four large southern subduction zone thrusts—at about 500, 850, 1400 and 2000 years ago. Clark’s earlier research along the entire subduction zone suggests it produced ten large ruptures in the past 7000 years, likely followed by massive tsunami. Pizer and Clark estimate that the lower section of the Hikurangi subduction zone ruptures once every 500 years. They reckon there’s a 26 per cent chance of a big quake occurring in the next 50 years.
Next, Pizer shows me another set of longer cores with even more distinct layers, these ones cut from the Pakuratahi Valley in Hawke’s Bay. They show a clear sequence, spanning about 5000 years, starting with silt layers deposited in a tidal lagoon that then change abruptly to a peaty forest—and back again. Twice. “Something huge has happened here to cause the land to subside to take it from a forest back down to estuarine tidal flats,” says Pizer.
Using three main clues—coastline subsidence in lagoons, uplift in marine terraces and tsunami deposits—Clark and Pizer are able to decipher a landscape’s memories of past geological upheaval. Then, by comparing the ages of deposits such as those from Lake Grassmere with information about the timing of earthquakes on other faults, they can tease out whether or not the quakes ripped along the Hikurangi subduction zone. By modelling what it would have taken to produce a tsunami that inundated Lake Grassmere, they determined it had to have been a subduction zone quake. “Nothing else shifts the volume of water that’s required to get that tsunami inland that far,” Clark says.
The land’s physical traces of past earthquakes and tsunami are remnants from deep geological time. In the same way, memories live on in people’s stories, reaching across generations in the Māori oral tradition of pūrakau.
People in the Wairoa district don’t tend to talk about the March 1947 earthquake, but they do remember the tsunami it triggered, says Nigel How, chairman of Ngāti Kahungunu Wairoa Taiwhenua. “On the Mahanga side of the Mahia isthmus the water withdrew out to sea, then flooded back over the land,” he says. “A common, shared memory from people of that time was, after the water had receded, thousands of fish were left stranded. Local people mobilised to collect and preserve the fish for future use.”
The region’s pūrakau include a legend about a tohunga and seven whales, preserving traditional knowledge about forces that shaped the land. “There are seven large hills near Wairoa,” says Low. “Local hapū recount that these hills were once whales. The valley where the whales once resided is called Waiatai—tidal waters—and informs us that this valley, now many miles from the coast, was once a saltwater area. It also informs us that these hills were once islands. This is an ancient indicator of land lift out of the sea.”
As part of a wider project to raise awareness and understanding of Hikurangi subduction zone earthquakes, the Wairoa district reached out to 37 marae to hold wānanga to discuss historical narratives and knowledge about seismic events. The stories were recorded and, along with Civil Defence information and other resources, returned to the marae for future use.
Iwi and hapū throughout Aotearoa hold pūrakau about taniwha that periodically unleash massive destructive waves, or rival chiefs who summon coastal deluges in retaliation or grief. Such mātauranga taiao is an important part of the New Zealand Palaeotsunami Database, which cross-checks the stories with geological or archaeological evidence.
The database was inspired by collaborative research between tsunami researchers James Goff, now semi-retired (and COVID-locked) in Wales, and Darren King at NIWA. King says oral histories about past catastrophic hazards, as well as many place names that mark areas of high risk, provide valuable, if often neglected, information that can help coastal communities to be better prepared. Such information could also benefit science as a way of corroborating interpretations—or pointing to sites of geological interest.
The Wairoa district’s wānanga lead to ongoing discussions, particularly among rural communities close to the sea. “Many are now aware of how far they have to evacuate inland,” says How. “Some communities took their discussions further with plans to install Civil Defence bases on high ground, which will act not only as a place to evacuate to but also contain emergency supplies. One rural community already has this in place.”
He says communities further inland are now aware that people living closer to the coast may one day need their help to provide shelter and support.
Across Hawke’s Bay, similar initiatives are underway. Lisa Pearse has worked in Civil Defence for more than two decades and has seen an increase in awareness, particularly since the 2011 tsunami in Japan. “Because Hawke’s Bay has the history of the 1931 Napier earthquake, people consider earthquakes and floods as the biggest hazards,” she says. “But tsunami went from about a 30 per cent awareness right up to 60 to 70 per cent recognition as a significant risk.”
Pearse chairs the East Coast Lab, which invites scientists studying the Hikurangi subduction zone to share their work with the community. It organises annual tsunami hīkoi for schools to practise their evacuation plans. In October, NEMA added tsunami hīkoi to its ShakeOut earthquake drill in coastal areas.
While the aftershocks from the magnitude-7.3 earthquake on March 5 have returned to normal, Graeme Atkins still remembers walking down to the beach the following day, finding driftwood piles reassembled and laced with seaweed that could only have come from the deepest ocean. Had the tsunami waves reaches further inland, they would have left clues to be buried in the landscape, for future scientists to uncover the traces of upheaval.