The faults that broke in the Kaikōura earthquake

Just after midnight on 14 November 2016, a magnitude 7.8 earthquake tore through more than 20 faults between North Canterbury and Cook Strait, in what GNS Science researchers have described as among the most complex earthquake ruptures ever recorded anywhere in the world. This November marks ten years since that night, and nearly a decade of research has turned the Kaikōura earthquake into one of the most closely studied events in the history of seismology.

The anniversary coverage will rightly focus on the recovery. This piece is about the geology. Which faults broke, why the rupture behaved in ways the textbooks said it should not, what it did to the coastline, and what the region’s earthquake activity looks like now. You can watch anything new as it happens on our live earthquake map.

Newswire also searched the GeoNet earthquake catalogue for this article, and found something the retrospectives will not tell you. Measured by shallow quakes of magnitude 3 and above, the rupture corridor has been quieter over the past four years than it was in the years before 2016. The full numbers, and the caveats that belong with them, are charted below.

• Magnitude 7.8, at 12:02am on 14 November 2016, starting near Waiau in North Canterbury
• Two people died, and the rupture ran about 170 kilometres north east in roughly two minutes
• More than 20 faults broke, producing about 180 kilometres of surface ruptures
• The coastline rose by up to 6.5 metres in places and dropped by 2.5 metres in others
• Ground acceleration of about 3 g at Waiau, the strongest ever recorded in New Zealand
• Shallow magnitude 3 quakes in the corridor are now fewer than before the earthquake

What happened on 14 November 2016

The earthquake began at 12:02am on a fault few had heard of, The Humps Fault, in the farmland between Culverden and Waiau. Over roughly the next two minutes the rupture worked its way about 170 kilometres north east, breaking fault after fault through the Kaikōura coast and inland ranges before dying out offshore beyond Cape Campbell. GeoNet’s event page records the depth at 15 kilometres, and most of the country felt the shaking, which lasted a minute or more in many places.

Two people died that night, one in the collapse of a farm homestead south of Kaikōura and one at Mt Lyford. That the toll was not far higher owes a great deal to timing. The ground motion near the epicentre was ferocious. Instruments at Waiau recorded a peak acceleration of about 3 g, meaning the ground briefly pushed on everything attached to it with three times the force of gravity, the strongest ground shaking ever recorded in New Zealand, though as at Heathcote Valley in 2011, later analysis suggested the instrument’s foundation exaggerated the sharpest spikes. The previous record was the 2.2 g measured in the February 2011 earthquake, which we covered in our companion piece on the faults that caused the Christchurch earthquakes. At midnight, almost everyone was at home in low, timber-framed houses, the most forgiving building type there is, and rural North Canterbury has few of anything else.

The catalogue records the day’s biggest aftershocks, most arriving while the region was still in the dark.

Time (NZDT)	Magnitude	Where
12:02am	7.8	Near Waiau, the mainshock
12:05am	6.0	Inland Marlborough, 95 km north east of the epicentre
12:12am	6.0	Just east of Kaikōura
12:32am	6.2	North of Kaikōura
12:52am	6.1	North of Kaikōura
2:31am	6.0	North of Kaikōura, inland
1:34pm	6.7	East of Hanmer Springs, the largest aftershock

Two housekeeping notes on the numbers. GeoNet’s catalogue lists the mainshock at magnitude 7.82, while the published scientific value is moment magnitude 7.8, a difference of method rather than a revision. The largest aftershock is the reverse case, reported as magnitude 6.3 on the day and later revised upward to 6.7 in the catalogue, so you will see both figures in older coverage.

What caused the Kaikōura earthquake

The short answer is that the rupture cascaded through a tangle of faults in the Marlborough Fault System, the zone where New Zealand’s plate boundary changes character. To the north, the Pacific Plate dives beneath the North Island along the Hikurangi subduction zone. To the south west, the plates grind past each other along the Alpine Fault. Marlborough and North Canterbury are the messy handover between the two, threaded with faults that both slide sideways and squeeze together, and that in-between geometry is a large part of why the earthquake was so complicated.

A tangle of more than 20 faults

The first scientific paper, published in Science within months, counted at least 12 faults in the rupture. Detailed field mapping eventually raised the count to more than 20, with about 180 kilometres of surface ruptures, several on faults that appeared on no map, an uncomfortable echo of Canterbury’s hidden fault problem. The biggest movement was on the Kekerengu Fault, where the ground surface shifted by up to 12 metres, among the largest surface displacements ever measured in an earthquake, and some models suggest slip at depth may have been twice that.

Part of the reason we know all this is that Kaikōura was among the first great earthquakes fully captured by modern radar satellites, which measure how the ground surface moved between passes, centimetre by centimetre, across the whole region. The deformation maps those satellites produced are some of the most detailed ever made of any earthquake.

Rupture jumps between faults

Before 2016, leading seismic hazard models assumed a rupture could not jump a gap of much more than five kilometres between faults. Kaikōura’s rupture transferred between faults separated by up to 15 to 20 kilometres, which forced hazard modellers worldwide to reconsider how large and how connected earthquakes can be. Whether those gaps were true jumps or connections through unmapped structures at depth is still being studied, and either answer changes how the next generation of models treats fault networks.

The fault that stayed quiet

The strangest detail is the fault that mostly did not break. The Hope Fault is the fastest-moving fault in the region and on paper the likeliest source of a large North Canterbury earthquake, and the rupture ran practically past its front door, yet its main segment stayed locked, with only minor slip near the coast. Computer modelling of the rupture suggests the earthquake’s own stresses, at the moment the rupture arrived, pushed the Hope Fault away from failure rather than towards it. The lesson, uncomfortable for anyone who likes tidy forecasts, is that a rupture chooses its path dynamically, in seconds, not just according to which fault is most overdue.

The subduction interface question

One question is still genuinely open. Beneath all those crustal faults lies the southern end of the Hikurangi subduction interface, the boundary where the Pacific Plate begins its dive beneath the North Island. Some models of the earthquake require metres of slip on that interface to explain the satellite and tsunami data, while others explain everything with the crustal faults alone. The answer matters, because it bears on whether the megathrust beneath Marlborough can join in future South Island earthquakes, and nearly ten years on the specialists still disagree.

Uplift along the coast

No New Zealand earthquake in modern times has rearranged a coastline like this one. Surveys published in Earth and Planetary Science Letters measured vertical movement along about 110 kilometres of coast ranging from 2.5 metres of subsidence to 6.5 metres of uplift, the most variable coastal deformation documented in any modern earthquake. At Waipapa Bay, the block of land between two strands of the Papatea Fault rose more than six metres, the greatest coastal uplift measured anywhere in the earthquake, lifting a new rocky platform out of the sea, seaweed, pāua and all.

The uplift rewrote the local marine ecosystem in a night. The pāua fishery closed along the entire affected coast and stayed closed for five years, reopening in December 2021 once the shellfish beds had recovered enough to sustain a short season. Offshore, the shaking flushed an enormous underwater avalanche of sediment down the Kaikōura Canyon, temporarily stripping life from one of the richest deep-sea habitats in the country, which has itself become a natural experiment in how quickly such ecosystems rebound.

Landslides, a tsunami and a cut-off town

The shaking brought the hills down. GNS Science’s first landslide inventory mapped more than 10,000 landslides across the region, with the completed count eventually running to roughly three times that, and identified 196 landslide dams blocking rivers and streams. The two big road slips either side of Kaikōura sealed the town off by land, and State Highway 1 north of the town did not reopen to daytime traffic until 15 December 2017, thirteen months later. Restoring the road and rail corridor cost about $1.2 billion.

There was a tsunami too, something many people outside the region never registered because it arrived in the small hours on a lightly populated coast. The sea reached about 7 metres above sea level at Goose Bay, south of Kaikōura. Coastal residents largely did the right thing that night, moving to high ground on the strength of the shaking alone, the “long or strong, get gone” rule working exactly as intended. Current tsunami advice and warning status is always on our tsunami warning page.

Effects across New Zealand

The earthquake physically moved the top of the South Island. The land between Kaikōura and Cape Campbell shifted north east by metres, with the Cape Campbell area moving more than two metres closer to Wellington. In the capital itself, 60 kilometres from the nearest rupture, the long, rolling shaking damaged mid-rise buildings on reclaimed land, partially collapsed floor sections in Statistics House, and led to several central city buildings being demolished.

Stranger still was what the earthquake set off silently. Its passing seismic waves triggered slow slip, a kind of slow-motion earthquake that releases energy over days to weeks without any shaking, along large stretches of the Hikurangi subduction zone off the North Island’s east coast, hundreds of kilometres away, with a separate deep episode under the Kāpiti coast. It was the largest and most widespread slow slip observed in New Zealand since monitoring began in 2002. A South Island earthquake had reached out and quietly moved the North Island’s plate boundary, a connection nobody had watched happen before.

Nearly ten years of aftershocks

Newswire searched GeoNet’s earthquake catalogue for the rupture corridor, from the plains south of Waiau to the far side of Cook Strait’s Cape Campbell corner. Since the mainshock it holds 30,779 located earthquakes of all sizes. The map shows the shallow magnitude 3 and above events, and you can see the sequence hugging the line of the rupture, densest at its northern end where the largest faults broke.

Aftershock sequences fade along a well-known curve, steeply at first and then ever more slowly, and Kaikōura’s has been no exception. What surprised us is where the curve has landed. For the past four years the corridor has produced 34 to 38 shallow magnitude 3 quakes a year, below its own average of about 60 a year across 2005 to 2012, before either the Cook Strait quakes of 2013 or the Kaikōura earthquake itself.

What does the quiet mean? Honestly, seismologists would offer you two readings and no promise. One is that the magnitude 7.8 released a large share of the strain stored on the region’s faults, buying the corridor time. The other is that some neighbouring faults now sit in a stress shadow, temporarily relaxed by the big rupture, with their turn deferred rather than cancelled. Either way, quiet is an observation about the past four years, not a forecast, and the earthquake itself is the standing proof that this region can produce decades of calm and then two minutes of chaos.

Could it happen again?

GeoNet still publishes a rolling forecast for the Kaikōura region, built from statistical models of how aftershock sequences decay. The most recent update was issued on 21 February 2025, covering the year that followed, and a fresh one had not been published as of this article. It put the chance of one or more magnitude 5.0 to 5.9 earthquakes in that year at 88 percent, magnitude 6.0 to 6.9 at 29 percent, and magnitude 7 or above at 3 percent.

“For example, there is a 29% chance of one or more M6.0-M6.9 within the next year.”

GeoNet, Kaikōura earthquake forecast, issued 21 February 2025

Those numbers describe a plate boundary region, not an aftershock problem. Marlborough and North Canterbury carry some of the fastest-moving faults in the country, the Hope Fault among them, still loaded and still overdue by its own average. The 2022 revision of the National Seismic Hazard Model raised estimated shaking hazard by about 50 percent on average, varying by location, and one of its Kaikōura lessons is baked in explicitly, because the model now allows the long, multi-fault ruptures that 2016 proved possible. The Alpine Fault, with a roughly 75 percent chance of rupturing within 50 years, remains the South Island’s biggest scheduled event.

More than 15,000 people filed felt reports for the mainshock, and every new quake in the region lands on our live earthquake map within minutes. For the geology of the country’s other recent surprises, see our explainers on the faults beneath Christchurch and the Taumarunui earthquake.

Nearly ten years on, the Kaikōura earthquake has settled into its place as the event that taught earthquake science how connected fault networks really are, and taught the rest of us that the map of known faults is a floor, not a ceiling. If you felt that midnight in 2016, or have questions about the faults in your own part of the country, share them in the comments below.

This article was written by AI, briefed to report the facts, hopefully without some of the bias people bring to the job 🙂