Quake threats

Much is known but scientists have so much more to find out about the quake threat to central New Zealand, writes Michael Daly. There’s no room for complacency, but the news is not all doom and gloom.

Science papers are usually fairly dry documents, but one that discusses the days immediately after the November 2016 Kaiko¯ura earthquake reads a little like a thriller.

Within 11 days of the magnitude 7.8 event tearing through the northeast corner of the South Island, GPS data revealed three slow slip events (SSE) were under way on the Hikurangi margin subduction zone.

SSEs are similar to earthquakes in that two pieces of the Earth’s crust move along a fault. Unlike earthquakes, where the movement happens in seconds, in SSEs it can take weeks or months.

The Hikurangi subduction zone stretches from the seafloor off Marlborough all the way to the seafloor north of East Cape. Along that zone the gigantic forces at work on the Earth are driving the seafloor of the Pacific tectonic plate beneath the Australian Plate, on which the North Island sits.

The Hikurangi margin is New Zealand’s largest fault, and under the lower North Island it’s locked. The plates aren’t slipping past each other; instead stress is building up – possibly to be relieved in a future massive earthquake.

Numerous research projects are adding to scientists’ knowledge about the Hikurangi margin, but there’s a lot they still don’t know. At the extreme end, scientists consider it’s possible the margin could rupture in a gigantic magnitude-9.0 megathrust event.

That’s the same size as the 2011 To¯ hoku-Oki earthquake in Japan, which caused a tsunami that produced waves up to 40 metres high. By some estimates the two plates in that earth-shattering event moved relative to each other by 80m. The seafloor was lifted by as much as 5m over an area of 15,000sq km, and the coast of Japan was shifted up to 5m eastward.

When the three SSEs started along the Hikurangi margin – at East Cape, Hawke Bay and Ka¯ piti – following the Kaiko¯ ura earthquake, it was assumed they were likely partly triggered by the passing main shock energy of the Kaiko¯ ura earthquake, despite being up to 600km from the quake’s epicentre.

The SSEs had been monitored for 20 years and it was the first time the three had been observed happening simultaneously. The Ka¯ piti SSE also appeared to be slipping faster than usual.

Seven scientists from GNS Science, led by seismologist Dr Matt Gerstenberger, explain in a paper on the National Hazards Research Platform, why those events raised concerns about the probability of more big earthquakes.

One thing was that the locked segment of the Hikurangi megathrust, beneath the lower North Island, had not been seen to slip since GPS observations started in the mid-1990s. That segment was largely surrounded by the SSE, and the SSE impact on the locked patch was not known.

The second reason was the apparent and likely triggering of the SSE by the Kaiko¯ ura main shock, and the likely loading of portions of the locked segment by the main shock.

Thirdly, there was a magnitude6.1 earthquake on the Hikurangi megathrust at the point where it transitions from the Hawke Bay SSE to the locked segment.

‘‘On November 25, the Ministry of Civil Defence and Emergency Management (MCDEM) was notified of the concerns,’’ the paper said. Information about the SSE was also published on the GeoNet website.

Shortly after, work got under way on producing more formal advice from GNS Science on the likelihood of future magnitude-7.8 or greater events in central New Zealand. That included work on any potential SSE impact.

Up to that time, there were no well-established and tested models explicitly accounting for the role of SSE in earthquake forecasts. The paper explains how scientists came up with probabilities for future large earthquakes in central New Zealand.

The work involved developing earthquake forecast models that were SSE specific. The models, along with observations, were discussed at two workshops of scientists from Japan, Taiwan, the US and New Zealand – in September and November 2017.

Many factors, including datasets and model outputs, were taken into account. Some of the data included observations of earthquake rates during SSE in New Zealand, earthquake clustering, and evidence of past earthquakes before the historic record.

‘‘The models were developed within a very short time-frame, but have laid the groundwork for future, more sophisticated models that will hopefully be able to better constrain forecast probabilities with reduced uncertainty,’’ the paper said.

Among key priorities identified was the need to set up an international working group focused on understanding the role of SSE in earthquakes.

The result of the work was the earthquake forecasts now on the GeoNet website.

For a magnitude-7.8 or greater earthquake: there is a 0.3-3 per cent forecast probability, with a 1 per cent best estimate, within the next year, and 2-20 per cent, with a 7 per cent best estimate, within the next decade. For a magnitude-7.0 or greater: 2-14 per cent, with a 6 per cent best estimate, within the next year, and 10-60 per cent, with a 30 per cent best estimate – considered ‘‘unlikely’’ – within the next decade.

GeoNet said the larger the range of probabilities, the less certain the experts were.

The one-year best estimates were 20 per cent higher for a magnitude7.0 event, and 100 per cent higher for the larger earthquake, compared to the long-term estimates in the National Seismic Hazard Model (NSHM), from before the Kaiko¯ ura earthquake.

In simple terms, the NSHM estimates the probabilities of ground shaking for the next 50 years. Data used include a fault model of the country, with more than 500 active faults – although there are others, some of which are unknown. The potential for earthquakes to occur on unknown faults is modelled using recordings of earthquakes in New Zealand from the last 200 years.

Other information includes the expected magnitude of earthquakes on the faults and how often they rupture.

The 2002 version of the NSHM was used in the New Zealand building design standards.

The upper bounds for both of the one-year magnitude range estimates on the GeoNet site are significantly higher than those in the NSHM. Similarly, the upper bounds on the decade estimates are a significant increase over the long-term estimates.

But the estimates for November 2017 were well down on those immediately following the Kaiko¯ ura earthquake. In December 2016, the chance of a M7.8+ event in the coming year was put at 5 per cent.

‘‘The general consensus across all of the models and data examined was that the impact of the SSE on the probability of future earthquakes was most significant in the first year following the Kaiko¯ura earthquake,’’ the paper said.

‘‘One year following the earthquake, the Kaiko¯ ura-triggered SSE have mostly stopped and their influence on future earthquakes is likely to be largely reduced.’’

The Hikurangi is the largest of the monsters lurking in the depths below the central North Island, but there are others also capable of unleashing mayhem.

A GNS Science report published in 2013 estimated damage and casualties in the Wellington region for seven large earthquake scenarios. ‘‘Four of the earthquakes, involving rupture of the Wellington, Wairarapa and O¯ ha¯ riu Faults, and the Hikurangi subduction zone, are considered to be the most costly and deadly earthquakes likely to impact the Wellington region.

‘‘In the modelling they generated shaking losses ranging from $9 billion to $16 billion.’’

A Wellington fault earthquake during the daytime was estimated to cause more than 1600 deaths, with a further 250 deaths from a tsunami.

Depending on how people react, a daytime subduction zone rupture could be the deadliest event. There would be an estimated 350 earthquake deaths, and 3200 tsunamirelated deaths. But that assumes a worst-case situation with no-one evacuating. But it’s not been all gathering gloom.

Computer modelling has suggested the 8.2-magnitude 1855 Wairarapa earthquake has reduced stress on both the Wellington fault and the nearby O¯ ha¯ riu fault.

A GNS Science study, as part of the It’s Our Fault project, said the estimated probability of rupture of the Wellington-Hutt Valley segment of the Wellington fault during the next 100 years was about 11 per cent.

The chance of rupture in the next 50 years was about 5 per cent. That was a decline of 50 per cent or more in the previous estimated probability of rupture. The Wellington fault last ruptured 300-500 years ago with a 7.6-magnitude event.

Of the other major faults in the region: the O¯ ha¯ riu fault ruptured 1100-1200 years ago and has a recurrence interval of 1500-5000 years, the Wairau fault last ruptured more than 800 years ago and has a recurrence interval of 1000-2300 years, and the Shepherds Gully fault last ruptured about 1200 years ago and has a recurrence interval of 2500-5000 years. The Wairarapa fault has a recurrence interval of 1150-1200 years. In the 1855 earthquake, it moved about 15m sideways and about 6m vertically.

The whole Wellington region was severely shaken, uplifted and tilted to the west. The land and sea floor near the harbour rose about 1-1.5m.

The tsunami produced had a runup – height of the water onshore – of 10m about 40km east of Wellington. Between 300 and 500km of coastline was affected.

Runups of 4-5m were widespread around Wellington and on the northern Marlborough coast, while the Rongotai isthmus was reportedly overtopped.

A2014 article in the journal Oceanography by four scientists said a tsunami generated by a magnitude9.0 earthquake on the Hikurangi margin would inundate much of the east coast of the North Island.

Tsunami models based on the rupture of the whole Hikurangi margin – from offshore Marlborough to offshore East Cape – showed predicted water levels of more than 10m above sea level at many places along the coastline.

Runup heights could be roughly double that of water level heights in some places. Models demonstrated inundation of Napier by up to 7km, and also significant inundation of Gisborne. But there are many unknowns that could affect what would happen.

A possible candidate for a major subduction interface earthquake recorded in Ma¯ ori oral history was the Hao-whenua earthquake, thought to have happened around 1460, the article said.

Geological evidence and Ma¯ ori oral history suggested it caused uplift in the Wellington region, including closure of a sea channel previously used to cross the Rongotai isthmus by canoe.

Evidence points to an earthquake and tsunami 7100 years ago, with sites from the south coast of the North Island to Pakarae, north of Gisborne, showing evidence of uplift or subsidence and tsunami inundation. ‘‘This is our strongest candidate for occurrence of a great Hikurangi subduction earthquake,’’ the article said. There was also evidence, but not as strong, at multiple sites for events about 5600, 4500, 3000, 1600 and 600 years ago.’’

Assumptions were involved, but a picture was emerging of great earthquake recurrence every 1000 to 1500 years.