FORECASTING LANDSLIDE MOVEMENT: WHAT DOES 16 YEARS OF REAL-TIME MONITORING REVEAL?

Contrary to what many studies assume, our 16-year record of real-time monitoring and repeated mapping shows that landslide movement can vary greatly through time, both in rate and location. Following the catastrophic failure of a slide that blocked U.S. Highway 50 in the Sierra Nevada of California for 4 weeks in 1997, we began monitoring the nearby Cleveland Corral landslide (CCL) perched above the highway, with a goal of trying to forecast slide behavior. At the CCL, we have recorded precipitation, pore-water pressures, and surface displacements in real time over the 3-5 m deep, 450 m long slide, and have episodically mapped movement patterns using ultra-short baseline (<1 km) GPS surveys.

We found that the CCL only moves during 3-4 winter/spring months in years with above average annual precipitation. Different parts of the slide move in different wet years, and rarely is the entire slide active at once. In a typical sequence, the toe moves first, followed by the head 100’s of meters upslope, whereas middle sections may lag in response or not move at all. Elements can exhibit different deformation styles. When active, large parts of slide move slowly, at rates of about 1-2 cm/day. However, smaller pieces at the steep toe or lateral margins can fail quickly and mobilize into debris flows. Our real-time monitoring has detected rapid acceleration during several flow events. Field bulk-density sampling indicates that this material often has a density less than the critical-state density, providing one means of forecasting which areas are more likely to mobilize into flows when they fail.

Annual slide displacement correlates with the amount of time the CCL is active, not necessarily with yearly precipitation. The slide requires about 3-4 months of antecedent precipitation (~1000 mm) to sufficiently elevate pore-water pressures on the failure surface and initiate movement. Elevated pore pressures at shallower depths are insufficient to trigger motion. Once moving, large storms can accelerate slow-moving parts of the slide; velocities can double within 1-2 days following rainfall yet still remain slow. Intense rain is needed to trigger small debris flows. Our monitoring reveals movement patterns that can vary over decadal time scales, but forecasting is possible given a comprehensive understanding of slide behavior.