2002 Denver Annual Meeting (October 27-30, 2002)

Paper No. 6
Presentation Time: 2:40 PM


BAUM, Rex L.1, HARP, Edwin L.1, MCMULLEN, Steven R.2, MCKENNA, Jonathan P.1 and KIBLER, John D.1, (1)U.S. Geological Survey, Denver, CO 80225-0046, (2)Shannon and Wilson, Inc, P.O. Box 300303, Seattle, WA 98103, jmckenna@usgs.gov

Landslides on coastal bluffs of Puget Sound between Seattle and Everett, Washington have frequently closed the tracks along an approximately 25-mile-long segment of the Burlington Northern Santa Fe Railway (BNSF) that lies at the foot of these bluffs. In October 2001 the U.S. Geological Survey (USGS), in cooperation with BNSF and its consultant, Shannon and Wilson, Inc., instrumented two sites on unstable bluffs near Edmonds and Everett, Washington. The instrumentation supports USGS research on rainfall-induced landslide initiation and a BNSF experiment in the use of site-specific precipitation data for anticipating landslide activity in high-hazard areas.

Shallow landslides commonly occur in weathered glacial deposits and colluvium on the Puget Sound bluffs after periods of relatively heavy rainfall or snowmelt. The coastal bluffs at these sites are underlain by subhorizontally bedded glacial and interglacial sediments, which include sandy outwash overlying glaciolacustrine silt deposits. Observations indicate that water enters the slopes by direct infiltration and by lateral flow through sandy layers that rest on less permeable layers of silt or clay.

Volumetric water-content measurements made to depths of 0.7 m during the winter rainy season of 2001-2002 detected downward propagating wetting fronts that varied with rainfall intensity, duration, and antecedent moisture. Near-surface water content increased several percent following several days of rain at the end of October 2001. Several shallow landslides also occurred then. The soil remained relatively wet throughout the winter, with water content varying by a few percent. Intense winter rainfall produced sharp wetting fronts that rapidly raised the water content by 2-3 percent. In contrast, low-intensity rainfall gradually raised water content 1-2 percent. After the rain stopped, water content gradually declined over several days to pre-storm levels.

Methods for anticipating landslide activity using the near-real-time data are under development but appear promising. Our instrumental observations are beginning to define the operating limits of rainfall thresholds and infiltration models that could be applied to forecasting landslide activity. Preliminary data can be viewed at http://landslides.usgs.gov/pugetrt/index.html.