Supercomputers Are Rehearsing New Zealand's Next Earthquakes

Weather forecasting was transformed by computer models. Earthquake science is now following the same path.

For decades, scientists estimated earthquake hazards largely by looking backwards, using observations from past earthquakes to predict how the ground might shake in the future.

Today, researchers are taking a different approach. Instead of relying primarily on historical records, they are building detailed three-dimensional models of the Earth and simulating hypothetical future earthquake scenarios.

The approach is increasingly complementing traditional empirical methods, helping scientists develop more accurate forecasts of how future earthquakes could affect different parts of New Zealand.

At the University of Canterbury, Professor Brendon Bradley and his QuakeCoRE team are among the researchers leading that transition.

"Earthquake science is going through the same transformation that weather forecasting experienced decades ago.

"We're moving from statistical models based largely on past observations towards numerical simulations that model how seismic waves will travel through the Earth," Bradley says.

Building a digital Earth

Computer simulation showing how local geology can produce very different levels of ground shaking across the Wellington region.

Every earthquake sends waves travelling through the Earth's crust.

Like ripples spreading across the surface of a pond, the waves move through rock, sediment and soil before reaching our towns and cities. But unlike water, the Earth is far from uniform. Just as neighbourhoods can experience different weather, different geological conditions mean the same earthquake can produce very different levels of shaking in Wellington, Christchurch and Kaikōura.

That complexity is one of earthquake science's greatest challenges.

Researchers are tackling it by building increasingly detailed digital representations of the Earth's subsurface, allowing them to simulate how earthquakes propagate through different geological conditions and estimate the shaking likely to be experienced at specific locations.

This means rather than simply asking what happened during previous earthquakes, scientists are exploring what may happen during future ones.

Better models. Better decisions.

For most New Zealanders, the value of this research isn't the models themselves. It's the decisions they help inform.

The 2022 National Seismic Hazard Model helps inform NZS 1170.5 and New Zealand's Building Code, shaping how buildings and infrastructure are designed across the country. Improving the model gives engineers and policymakers a clearer picture of seismic risk, helping target investment where it will have the greatest impact.

"We don't want to under-invest in places that are genuinely vulnerable. But equally, there's no value in spending large sums of money where it isn't needed. It's really about making the smartest decisions with the finite resources we have," Bradley says.

More accurate modelling means communities can make better-informed decisions about schools, hospitals, transport infrastructure and buildings, improving resilience while making the best use of public investment.

Billions of calculations

Professor Brendon Bradley from the University of Canterbury uses high-performance computing to simulate how earthquakes travel through the Earth, helping researchers build more accurate models of future ground shaking.

Creating these simulations requires enormous computing power.

Researchers divide the Earth into millions of tiny sections before calculating how seismic waves travel through each one. A single ground-motion simulation can involve billions of mathematical calculations, with researchers often running hundreds or even thousands of different earthquake scenarios.

"These problems are fundamentally computationally limited. Even with access to high-performance computing, we're constantly thinking about how to make the most efficient use of those resources”, Bradley says.

Researchers develop and validate their models using New Zealand's national high-performance computing capability before scaling up to some of the world's largest supercomputers for their most demanding simulations. In 2026, Bradley's team secured more than 20 million core hours on the MareNostrum 5 supercomputer in Barcelona, one of the largest allocations awarded in that funding round.

REANNZ provides the advanced digital infrastructure that enables researchers to access New Zealand's national HPC capability and connect with international computing resources as their research grows.

Preparing for tomorrow's earthquakes

Earthquakes will always remain difficult to predict. Unlike cyclones, we simply can’t see them coming.

But understanding how the ground beneath us is likely to shake, and how that shaking varies from one neighbourhood to the next, is becoming increasingly sophisticated.

Just as weather forecasting has steadily improved over the past several decades, earthquake science is evolving through better observations, more powerful computers and increasingly realistic simulations.

For REANNZ, supporting that work means helping researchers tackle one of New Zealand's most important scientific challenges.

Every simulation brings us that much closer to understanding how the next major earthquake could affect our communities, helping ensure our future decisions about buildings, infrastructure and resilience are based on the best science available.