CAPTURING LANDSLIDE DYNAMICS USING NEAR-REAL-TIME MONITORING

Most landslide investigations are post-event autopsies. These do not capture time-dependent behavior, such as initiation and acceleration, that may signal the onset of rapid failure. For a decade, we have used near-real-time monitoring of active landslides with the goals of understanding initiation and forecasting rapid movement. Automated remote data collection and web-accessible graphs have provided immediate notification of landslide activity even in stormy weather, useful for public safety assessments. These systems help ensure continuous operation and more complete data, enabling the development of potentially better warning systems and remedial designs.

The 450 m long Cleveland Corral landslide, CA, with its toe perched about 50 m upslope of well-traveled U.S. Highway 50, is located within several km of two other landslides that catastrophically blocked the highway for months. To capture the conditions leading to activity of this slide, we have measured precipitation, groundwater pressures, and movement, using extensometers, since 1997 (http://landslides.usgs.gov/monitoring/hwy50). Our data show that 800-900 mm of seasonal precipitation triggers slow (0.5 to 5 cm/day) movement of the main slide. In contrast, small blocks fail rapidly off the active toe and transform into debris flows during intense storms. Over the 12 hours prior to rapid flow, these blocks accelerate to velocities exceeding 100 cm/day.

To better capture 3-D displacements over time, we have developed a low-cost, L1-only GPS receiver system. Using very short baseline, differential, automated processing, we can detect sub-cm displacement. During spring 2005, the large Ferguson rockslide reactivated and blocked CA Highway 140, one of the main routes into Yosemite National Park. Using helicopter sling loads, we deployed three of our GPS units, mounted on metal tripods, called spiders, on large blocks of the main rockslide. Over the relatively dry 2006-2007 winter, one of the outer blocks moved more than 2 m eastward and 2 m downward. This block was also very sensitive to rainfall, doubling its velocity (up to 2 cm/day) within 3 to 4 days after rainy periods. By monitoring the dynamics of these slides, we have been able to capture conditions triggering movement and deformation rates preceding rapid motion.