Mudslides: Forecasting Risk

May 7, 2009

Aftermath of 1995 La Conchita landslide, in southern California.
Photograph by Robert Schuster, USGS

Media Inquiries

Kim Martineau

Landslides kill at least a thousand people each year but because they are often triggered by earthquakes or heavy rains, the danger remains poorly understood.

"In densely populated areas, landslides take no prisoners,” said Art Lerner-Lam, a scientist at Columbia University’s Lamont-Doherty Earth Observatory.  “They’ll wipe out an entire village at once. Even a small landslide can kill hundreds of people."

Scientists know less about landslides than they do other natural hazards such as earthquakes, which last year killed 88,000 people.

But they’ve learned to recognize several warning signs. Stripping away trees and natural ground cover for development is one way to increase landslide risk. Steep slopes, heavy rains and wet soil can also make an area more susceptible.

To be able to predict landslides, scientists have developed slope stability models to analyze the risk locally. More recently, NASA has created a preliminary algorithm to map landslide hazards globally using satellite measurements of rainfall, land cover and other surface variables. But before this model can be useful to planners and relief workers, its accuracy needs to improve.

Dalia Bach Kirschbaum, a doctoral candidate at Lamont-Doherty, decided to take on this challenge for her thesis.  To test the performance of NASA’s landslide hazard algorithm, she compared the algorithm forecasts against an inventory of recent rain-triggered landslides worldwide, assembled mostly from media reports. Her analysis found that the model had predictive value but requires higher-resolution measurements and a more complete catalog of real-life events to improve its forecast skill.

One challenge in compiling a global landslide catalog is that so many landslides go unreported, especially in less developed regions, she said. Another problem is that some might have been reported in the foreign language press, which she did not include in her study.

Still, from looking at the catalog a big picture emerged. The mountainous regions of South East Asia, for example, appear to be most susceptible to landslides globally, especially during the summer monsoon rains.  India, Nepal, China and Japan showed the largest number of landslide reports and deaths in all three years that Kirschbaum studied—2003, 2007 and 2008. Landslide reports and fatalities also increase during the summer hurricane season in Central America and the Caribbean. Her work is described in a newly published paper in the journal, Natural Hazards and Earth System Sciences.

“The catalog provides a foundation to do much more advanced studies,” she said. “With more comprehensive evaluations we can do landslide risk assessment as well as look at climate trends. For example, does an El Nino year bring more landslides?”