Science Center Objects
The most frequent and widespread damaging landslides in the U.S. are induced (started) by prolonged or heavy rainfall. The majority of rainfall-induced landslides are shallow (less than a few meters deep), small, and move rapidly. Many rainfall-induced landslides transform into debris flows (fast-moving slurries of water, soil, and rock) as they travel down steep slopes, especially those that enter stream channels where they may mix with additional water and sediment.
Research at the USGS to improve understanding and predictive tools for rainfall-induced landslides relies on a network of hillside hydrologic monitoring stations to provide data about response of soil moisture and water pressures in the shallow subsurface to rainfall and snowmelt. These data provide observations informing conceptual models as well as ground truth for testing numerical models of rainfall infiltration, subsurface flow, and slope instability.
The monitoring network also provides opportunities to evaluate approaches for landslide early warning. Landslide warning systems exist in several countries and a few areas of the U.S. The USGS collaborates with the National Weather Service on a prototype debris-flow warning system for post-fire debris flows in southern California.
Since the 1980’s, prediction of rainfall-induced landslides has relied on maps of landslide susceptibility and catalogs of landslide occurrence and corresponding rainfall amounts (rainfall thresholds). Due to their incomplete description of conditions needed to induce landslides, conventional rainfall thresholds have considerable uncertainty. For example, in the case of Seattle, Washington, when rainfall exceeds existing thresholds there is only a 10% to 70% chance of landslide occurrence. USGS research seeks to improve the reliability of landslide warning criteria through field monitoring efforts to better define the relationships between landslide timing, soil moisture or pore-water pressure conditions, rainfall intensity, and other factors that influence occurrence of landslides. Numerical modeling efforts also underway provide additional insight and are expected to help extend landslide warning tools to areas where detailed historical landslide information is unavailable.