LESSONS LEARNED FROM FOUR DECADES OF IN SITU MONITORING OF MINOR CREEK LANDSLIDE

The dynamics of slowly moving landslides can change in response to variations in forcing, and many landslide investigations have durations too brief to reveal typical behavior. This limitation is highlighted by results from a 40-year study of Minor Creek landslide in the Franciscan terrane of Redwood Creek basin, Humboldt County, California. In 1973 the USGS established benchmarks and about 200 theodolite survey stakes to measure surface displacements along five transects crossing the 800-m-long, 100-m wide landslide. Soon thereafter, two rain gages, two extensometers, and two gully-flow stage recorders were installed. Then, from 1978 to 1982, numerous boreholes were drilled and fitted with 67 piezometers and 15 inclinometer casings. Drilling also yielded core samples that were subjected to geotechnical tests. Most subsurface monitoring was discontinued in 1997 due to progressive failure of borehole casings. Most surface monitoring continued until 2000, when a network of 19 GPS monuments was installed to replace deteriorating survey stakes. Campaign GPS monitoring still continues.

From 1973 through 2012, a central 600-m long segment of the landslide mass moved downslope 11 to 16 m, and more than half of this total displacement occurred during just three water years: 1974, 1984, and 1998. By contrast, motion of the uppermost part of the landslide (within about 100 m of the slowly receding headscarp) was quite consistent from year to year. Relatively chaotic motion occurred within 100 m of the landslide toe, where behavior was strongly influenced by the shifting morphology of adjacent Minor Creek.

The three water years in which most landslide motion occurred were wetter than average, but not by wide margins. Water-table heights gauged by piezometers < 3 m deep were virtually indistinguishable for wet years and dry years. Periods with the highest pore-water pressures at landslide-base depths of 5 to 6 m (determined by inclinometry) coincided with periods of fastest landslide motion, but this relationship became clear only when data from many spatially distributed piezometers were averaged. The data show that landslide motion is extremely sensitive to subtle differences in wet-season groundwater flow fields -- behavior that might be inscrutable if detailed, long-term monitoring were lacking.