Opening Pandora's Box

With each new stride in hazard research and science comes the ability to better calculate and differentiate risk 

Efforts by RMS scientists and engineers to better understand liquefaction vulnerability is shedding new light on the secondary earthquake hazard. However, this also makes it more likely that, unless they can charge for the risk, (re)insurance appetite will diminish for some locations while also increasing in other areas. A more differentiated approach to underwriting and pricing is an inevitable consequence of investment in academic research.

Once something has been learned, it cannot be unlearned, explains Robert Muir-Wood, chief research officer at RMS. “In the old days, everybody paid the same for insurance because no one had the means to actually determine how risk varied from location to location, but once you learn how to differentiate risk well, there’s just no going back. It’s like Pandora’s box has been opened.

“There are two general types of liquefaction that are just so severe that no one should build on them”

Tim Ancheta

RMS

“At RMS we are neutral on risk,” he adds. “It’s our job to work for all parties and provide the best neutral science-based perspective on risk, whether that’s around climate change in California or earthquake risk in New Zealand. And we and our clients believe that by having the best science-based assessment of risk they can make effective decisions about their risk management.”

Spotting a Gap in the Science

On September 28, 2018, a large and shallow M7.5 earthquake struck Central Sulawesi, Indonesia, triggering a tsunami over 2 meters in height. The shaking and tsunami caused widespread devastation in and around the provincial capital Palu, but according to a report published by the GEER Association, it was liquefaction and landslides that caused thousands of buildings to collapse in a catastrophe that claimed over 4,000 lives. It was the latest example of a major earthquake that showed that liquefaction — where the ground moves and behaves as if it is a liquid — can be a much bigger driver of loss than previously thought.

The Tōhoku Earthquake in Japan during 2011 and the New Zealand earthquakes in Christchurch in 2010 and 2011 were other high-profile examples. The earthquakes in New Zealand caused a combined insurance industry loss of US$22.8-US$26.2 billion, with widespread liquefaction undermining the structural integrity of hundreds of buildings. Liquefaction has been identified by a local engineer as causing 50 percent of the loss.

Now, research carried out by RMS scientists is helping insurers and other stakeholders to better understand the impact that liquefaction can have on earthquake-related losses. It is also helping to pinpoint other parts of the world that are highly vulnerable to liquefaction following earthquake.

“Before Christchurch we had not appreciated that you could have a situation where a midrise building may be completely undamaged by the earthquake shaking, but the liquefaction means that the building has suffered differential settlement leaving the floors with a slight tilt, sufficient to be declared a 100 percent loss,” explains Muir-Wood.

“We realized for the first time that you actually have to model the damage separately,” he continues. “Liquefaction is completely separate to the damage caused by shaking. But in the past we treated them as much of the same. Separating out the hazards has big implications for how we go about modeling the risk, or identifying other situations where you are likely to have extreme liquefaction at some point in the future.”

The Missing Link

Tim Ancheta, a risk modeler for RMS based in Newark, California, is responsible for developing much of the understanding about the interaction between groundwater depth and liquefaction. Using data from the 2011 earthquake in Christchurch and boring data from numerous sites across California to calculate groundwater depth, he has been able to identify sites that are particularly prone to liquefaction.

“I was hired specifically for evaluating liquefaction and trying to develop a model,” he explains. “That was one of the key goals for my position. Before I joined RMS about seven years back, I was a post-doctoral researcher at PEER — the Pacific Earthquake Engineering Research Center at Berkeley — working on ground motion research. And my doctoral thesis was on the spatial variability of ground motions.”

Joining RMS soon after the earthquakes in Christchurch had occurred meant that Ancheta had access to a wealth of new data on the behavior of liquefaction. For the first time, it showed the significance of ground- water depth in determining where the hazard was likely to occur. Research, funded by the New Zealand government, included a survey of liquefaction observations, satellite imagery, a time series of groundwater levels as well as the building responses. It also included data collected from around 30,000 borings.

“All that had never existed on such a scale before,” says Ancheta. “And the critical factor here was they investigated both liquefaction sites and non-liquefaction sites — prior surveys had only focused on the liquefaction sites.”

Whereas the influence of soil type on liquefaction had been reasonably well understood prior to his research, previous studies had not adequately incorporated groundwater depth. “The key finding was that if you don’t have a clear understanding of where the groundwater is shallow or where it is deep, or the transition — which is important — where you go from a shallow to deep groundwater depth, you can’t turn on and off the liquefaction properly when an earthquake happens,” reveals Ancheta.

Ancheta and his team have gone on to collect and digitize groundwater data, geology and boring data in California, Japan, Taiwan and India with a view to gaining a granular understanding of where liquefaction is most likely to occur. “Many researchers have said that liquefaction properties are not regionally dependent, so that if you know the geologic age or types of soils, then you know approximately how susceptible soils can be to liquefaction. So an important step for us is to validate that claim,” he explains.

The ability to use groundwater depth has been one of the factors in predicting potential losses that has significantly reduced uncertainty within the RMS suite of earthquake models, concentrating the losses in smaller areas rather than spreading them over an entire region. This has clear implications for (re)insurers and policymakers, particularly as they seek to determine whether there are any “no-go” areas within cities.

“There are two general types of liquefaction that are just so severe that no one should build on them,” says Ancheta. “One is lateral spreading where the extensional strains are just too much for buildings. In New Zealand, lateral spreading was observed at numerous locations along the Avon River, for instance.”

California is altogether more challenging, he explains. “If you think about all the rivers that flow through Los Angeles or the San Francisco Bay Area, you can try and model them in the same way as we did with the Avon River in Christchurch. We discovered that not all rivers have a similar lateral spreading on either side of the riverbank. Where the river courses have been reworked with armored slopes or concrete linings — essentially reinforcement — it can actually mitigate liquefaction-related displacements.”

The second type of severe liquefaction is called “flow slides” triggered by liquefaction, which is where the soil behaves almost like a landslide. This was the type of liquefaction that occurred in Central Sulawesi when the village of Balaroa was entirely destroyed by rivers of soil, claiming entire neighborhoods.

“It’s a type of liquefaction that is extremely rare,” he adds. “but they can cause tens to hundreds of meters of displacement, which is why they are so devastating. But it’s much harder to predict the soils that are going to be susceptible to them as well as you can for other types of liquefaction surface expressions.”

Ancheta is cognizant of the fact that a no-build zone in a major urban area is likely to be highly contentious from the perspective of homeowners, insurers and policymakers, but insists that now the understanding is there, it should be acted upon.

“The Pandora’s box for us in the Canterbury Earthquake Sequence was the fact that the research told us where the lateral spreading would occur,” he says. “We have five earthquakes that produced lateral spreading so we knew with some certainty where the lateral spreading would occur and where it wouldn’t occur. With severe lateral spreading you just have to demolish the buildings affected because they have been extended so much.”