Our Changing World: Building a nation-wide model of landslide risk

Follow Our Changing World on Apple, Spotify, iHeartRadio or wherever you listen to your podcasts

Pictures of landslides cover two walls of Dr Chris Massey's office at Earth Sciences New Zealand in Lower Hutt.

Russia, Ethiopia, Hong Kong, Bhutan - all places he's worked in, called in by officials or companies to assess causes or impacts of various landslides.

He points to an image just beside his desk. Massive boulders scattered on a road. "This is from the Port Hills," he says. After the Christchurch earthquakes many of Earth Sciences New Zealand geologists were busy mapping where different slopes had failed and the scale of the landslides' impacts. The information was then stored away in the landslides database.

These post-event investigations have been the general history of New Zealand landslide research to date, Chris explains.

"Landslide research really started at DSIR (Department of Scientific and Industrial Research - a precursor to Earth Sciences New Zealand) because of the Abbotsford landslide in 1979".

After months of smaller land movements of a subdivision in Abbotsford, just south of Dunedin, a huge portion of the hill gave way at 9pm on the evening of 8 August 1979. 69 homes were destroyed, luckily no-one was killed.

National landslide research continued on a national needs-basis with investigations into the landslide risk of the hydro dam development in Clyde, says Chris, and then, after DSIR became GNS, in response to natural disasters like the Christchurch and Kaikoura earthquakes, and heavy rainfall events in the north island, including Cyclone Gabrielle.

Alongside this they began some work into prediction - what might happen as a result of an earthquake in Wellington.

Now Earth Sciences New Zealand are running two large publicly-funded projects aimed at pushing forward this shift-change in landslide research in New Zealand: from post-event investigations to pre-event forecasting.

'Insidious landslides'

Chris is co-leading 'Landslide Watch Aotearoa', which focuses on slow-moving landslides.

"They're the ones that people don't tend to focus on because they're really hard to find. You know, they're kind of hidden in the landscape. We usually find them because we build on them or something."

The Utiku landslide near Taihapē is one of them, the movement of which threatens State Highway 1 and a railway line. The Tāhunanui landslide in Nelson is another - New Zealand's largest urban landslide which threatens homes in the Tāhunanui hills.

Sometimes these landslides move very little over long periods of time, sometimes they speed up in response to rainfall.

A number of the already-identified slow-moving landslides are carefully monitored by government bodies, local councils or private companies, because of the risk they pose to infrastructure. They are constantly collecting data about ground movement, rainfall and pore water levels on these landslides, and now, sharing this information with Chris and his team, and working with them to install further monitoring equipment.

Chris hopes to develop a model that can identify these landslides and monitor their movement using satellite radar imagery. A time series of the radar data, which is updated every 12 days, allows them to measure ground movement from space to within millimetres per year. By identifying the rainfall patterns and groundwater conditions that trigger faster movement, he hopes to better forecast how these landslides will respond to future weather and climate conditions.

Forecasting rapid landslides

A few doors down from Chris' office Dr Saskia de Vilder pulls up an early prototype of one of the risk maps she and the Hōretireti Whenua / Sliding Lands team are developing.

This one is focused on an area of Wellington, with a colour-coded projection of which suburbs would be at risk of landslide if the 1855 earthquake were to happen today.

The idea is to create maps and models that would be useful for a whole range of stakeholders, says Saskia. They've been speaking to people who might be interested in the kind of landslide risk modelling and forecasting they are hoping to do - council planners, big developers, emergency management personnel, risk communicators - they have a list of ten groups they are keeping in mind as they develop the models.

To create them they've broken New Zealand down into different 'landscape domains' - regions that are similar in geology, tectonics and climate. For each region they then train the models on existing datasets, which includes all the information from all the past landslide mapping they've done. On to this they then layer other datasets about the triggering factors for the landslides - so past earthquake data, or rainfall events.

This 1855 earthquake map might be helpful for council planning, she suggests, looking at a long-term risk of landslides if a similar sized earthquake might occur.

There's also the option to create maps more helpful for short-term emergency management, she says. By inputting a rainfall forecast into their model, they can predict which areas are likely to be more susceptible to landslides happening - what suburbs or infrastructure might be at risk based on the data about the incoming weather event.

It's a five-year programme, due to finish in 2028, but the team are already getting some of the early models into the hands of the stakeholders, says Dr Chris Massey.

"What we've tried to do is go, hey, people said they wanted it early, this is what we've done now, it's got a few warts in there… but it's better than what we've had which was… well we didn't have anything really."

Sign up to the Our Changing World monthly newsletter for episode backstories, science analysis and more.