A new landslide prediction model from Dino Bellugi, Taylor Perron, and Paul O’Gorman could help communities prepare for disaster in the face of changing climate.
Taylor Perron has seen the aftermath of landslides firsthand. As a graduate student at the University of California at Berkeley, the MIT professor surveyed resulting debris flows and damaged infrastructure in California. “It really made me sit up and take notice,” he said. “We’re used to thinking about many geological processes as being gradual, but to be able to understand something that happens that quickly and has such destructive power was something I knew I needed to do.”
Many landslides are triggered by extreme precipitation events—large amounts of rain falling over short periods of time—which are increasing in both intensity and frequency. Predicting where landslides are more likely to happen is paramount for vulnerable communities adjusting to a changing climate, but the task isn’t easy.
“The challenge of predicting landslides is somewhat like the challenge of predicting earthquakes,” said Perron, now Associate Professor of Geology in MIT’s Department of Earth, Atmospheric and Planetary Sciences. “We know enough to be able to map out the susceptibility of different areas, but knowing where and when an individual event will strike is a different matter.” Thanks to a new model developed by MIT researcher Dino Bellugi, landslide prediction may no longer be a far-off goal.
The model predicts individual landslide locations and sizes, and the researchers are taking the innovation a step further. Perron and Bellugi have teamed up with EAPS Associate Professor of Atmospheric Science Paul O’Gorman to integrate this model with models of extreme rainfall to see how landslides might shift under different climate scenarios.
“This is a real challenge to construct and use this kind of landslide model and figure out how to integrate that with climate science,” Perron said. “If we succeed, the output is going to be something that really matters on a timescale of years to decades.”
First, the researchers coupled a landslide search algorithm with a model that uses soil depth, root strength, and pore water pressure to assess the ground stability of a defined region—in this case, a research site in the Pacific Northwest prone to landslides. “The research at this site gave us a natural experiment where we know the landslide pattern in recent history and have several important parameters constrained by the field research there,” said Bellugi. “It was a natural site to test our model.”
Bellugi and Perron mapped the resulting output and compared it to known landslides. They found that the predictions overlapped well with mapped observed landslides—not just in location, but in size as well. In the future, the model could be used to predict locations where landslides have yet to occur.
Next, O’Gorman and Bellugi combined the landslide model with a spatially resolved climate model that provided an estimate of regional precipitation changes near the end of this century. A statistical analysis of the resulting output revealed that a 10 to 20 percent increase in the intensity of extreme rainfall, and even larger increases in previously drier areas, was likely.
“A 20 percent increase in precipitation intensity, which is what we expect and models predict, may be enough to open up windows of opportunity for landslides to occur in places that aren’t currently considered hazardous,” said Bellugi. These areas won’t necessarily see an increase in average rainfall, but they will see more extreme precipitation events, which puts them at greater risk for landslides. “Risk is not just about hazard, it’s also about people’s exposure to that hazard,” Perron added. “If people are not used to being exposed to a hazard and then become exposed, the risk can increase dramatically.”
The next goal on the trio’s agenda is connecting their model to one that predicts how far debris flows—by far the most dangerous part of landslides—will go, and how close they will come to communities and infrastructure. “If you can tell people something more concrete—‘here you’ll have an increased risk, get prepared’—that may be something that can be easily translated to policy,” Bellugi said.
This article appears in the 2015-2016 issue of EAPS Scope.