Australasian Science: Australia's authority on science since 1938

Cloudy with a Chance of Earthquakes

Credit: Petrovich12/Adobe

Credit: Petrovich12/Adobe

By Simon Lamb

The rupture of a megafault beneath New Zealand in 2016 has revealed a periodicity to earthquakes that may enable geologists to forecast seismic events based on satellite monitoring of the Earth’s movements.

Living in New Zealand, I have felt many earthquakes. Even though I am a geophysicist who studies their causes and effects, I still experience deep uneasiness whenever they occur. The idea of a solid and fixed bedrock to our world is deeply embedded in the human psyche, so it’s truly terrifying when this fundamental assumption is turned on its head during an earthquake.

During the few seconds of an earthquake – stretching sometimes to minutes in a big one – the bedrock is flung up and down, with accelerations that can be greater than those experienced by a rugby ball when it’s dropped or kicked. However, once the earthquake and aftermath of numerous aftershocks have receded into the past, the ground seems fixed and solid again and we quickly revert to our old assumptions of bedrock stability.

So why do certain parts of the Earth switch between long periods of stability and short bursts of violent seismic shaking? I think the answer to this question is central to finding new ways to forecast earthquakes in the future. The key to this is understanding what happens between big earthquakes.

Since satellite-based monitoring of the Earth’s surface enabled precise surveys of the landscape, it has been clear that the solid Earth is not fixed at all, but is instead in constant motion. The movements are very slow – not more than a few centimetres per year – so we are completely unaware of this motion as we go about our daily lives.

These movements are part of the bigger picture of plate tectonics, with the surface of the Earth made up of restless tectonic plates. The most earthquake-prone parts of the Earth lie along the boundaries of these plates (Fig. 1): New Zealand straddles the boundary between the Pacific Plate and the Australian Plate; California also lies on the edge of the Pacific Plate where it moves past the North American Plate. At these boundaries the landscape is cut by numerous active faults where the rocks slide past each other over geological time.

Geologists believe that each large active fault has its own in-built pacemaker determining the interval between earthquakes. The theory is that if you go deep enough, the rocks are sufficiently hot that they can freely slip past each other, moving day by day, year by year, driven by the same forces that shift the tectonic plates. However, the surface part of the fault is stuck or locked, and here the landscape is being twisted or distorted. There is only so much of this that the rocks can take before they break, giving rise to an earthquake as the surface rocks suddenly catch up with the deeper part of the fault slipping beneath it.

If this is the case, earthquakes on each fault would occur at approximately regular intervals and have a characteristic magnitude. Thus if geologists could work out the earthquake history for enough faults, they would have the basis for forecasting future earthquakes. Much of the current effort in earthquake forecasting is attempting to do this.

In 2016, the Kaikoura earthquake shook central New Zealand with a magnitude of 7.8. There were very few casualties, but from a scientific point of view it offered a rare opportunity to study the build-up in the landscape of the forces that cause a big earthquake. Richard Arnold of Victoria University of Wellington and James Moore of Nanyang Technical University in Singapore joined me to analyse ground movements over the previous two decades, a period when we had a sufficiently precise record of satellite-based monitoring. These movements could then be compared with what actually happened during the earthquake itself.

Our results, which have been published in Nature Geoscience (, were surprising. Rather than finding evidence for gradual slipping beneath each individual major fault in the region, we found that the distortion of the landscape was better explained if we ignored these faults altogether.

There was only one fault that mattered, and that was a fault called the megathrust. Huge in size, the megathrust lies beneath much of New Zealand at the junction between the Pacific Plate and the Australian Plate. It only reaches the surface offshore, on the seabed.

We found that beneath much of central New Zealand, the megathrust is freely slipping where it is at depths of about 30 km or more. At shallower depths, however, the megathrust ias stuck or locked. Here the overlying Australian Plate is slowly changing shape like a giant piece of elastic (Fig. 2).

This type of fault is not a peculiarity of New Zealand. Megathrusts underlie much of the Pacific Rim in a vast arc stretching for tens of thousands of kilometres from New Zealand, Tonga and on to the Philippines, Alaska, Central America and southern Chile. Another major megathrust underlies parts of South-East Asia, stretching from Timor through Indonesia to Myanmar. From an earthquake point of view, these are the most deadly parts of the planet (Fig. 1).

Our results matter, changing the way we think about future earthquakes on faults, because we have shown, in New Zealand at least, that each fault doesn’t have its own independent driving mechanism that ensures a unique pattern of semi-regular earthquakes. Instead, our results show that the driving mechanism is rather simple, and is not related to the faults we can see but is just due to the slipping of the underlying megathrust.

We think that the 2016 Kaikoura earthquake provides another vital clue that we are right, because the rupture was not confined to any particular fault. Rather, it was a broader zone of shattering that involved many faults, running parallel to the junction between the freely slipping and locked part of the underlying megathrust (Figs 1,2).

Think about bending a sheet of glass until it smashes. Beneath New Zealand, the movement along the megathrust is causing the bending and, like the smashed sheet of glass, the resultant pattern of cracks will be very difficult to predict. However, we might be able to say something about when and where the next shattering event will occur.

I think we are on the cusp of a revolution in our ability to forecast natural disasters such as earthquakes and volcanic eruptions, making much better use of the torrent of satellite monitoring data recording the behaviour of the Earth’s surface. We can use this information to understand in much greater detail, year by year, the forces driving earthquakes, potentially calculating what might happen in the future based on fundamental physical processes rather than relying on the past earthquake history.

Already, geophysicists are producing detailed computer models of the solid Earth on timescales of seconds to decades and longer, revealing what happens deep below the surface before and during big earthquakes. These models are beginning to provide the insights we need to interpret the satellite measurements in the build-up to the next big earthquake.

I see modern weather forecasting as an analogy for how this might work. In the early days of meteorology, weather forecasts were unreliable, based on sparse information about atmospheric conditions and past experience of weather systems. Remember the folk wisdom: red sky at night, shepherd’s delight, red sky in the morning, shepherd’s warning? This is a wonderful example of a traditional empirical approach to natural phenomena based on an observed association without a rigorous theoretical basis.

Today, the weather is forecast through intensive supercomputer models of the atmosphere, based on constantly updated satellite and surface information about the wind, temperature, humidity and pressure together with the fundamental physics of how gases in the atmosphere behave. These physics-based weather forecasts are getting better and better.

I believe it’s time to develop physics-based approaches to earthquake forecasting. If we put our knowledge of how the landscape is moving into computer models that reproduce the forces in the Earth with the known properties of rock, we could vastly improve our ability to monitor what is happening in the build-up to an earthquake.

We might never be able to say exactly when and where an individual fault will rupture, but we might be able to identify dangerous pressure points in the outer part of the Earth that are most prone to the next tremor or quake. This will help those of us who live with earthquakes to be better prepared for the future.

Our best protection, however, will always be resilience, whatever our forecasts tell us.

Simon Lamb is an an Associate Professor of Geophysics at Victoria University of Wellington.