USING RATE-DEPENDENT RELATIONSHIPS FROM INCLINOMETER DATA TO MONITOR LANDSLIDES AND SLOPE MOVEMENTS

The special relationship between the stress and strain states of a landslide mass during terminal stages can be a useful predictor of failure (Saito, 1988, Voight, 1989, Fukuzono, 1990, Hutchinson, 2001, Kilburn and Petley, 2003, and Petley, 2004). For example failure can be preceded by an acceleration trend that is linear when inverse velocity, 1/v is plotted against time, t. This is likely due to brittle deformation during crack propagation. In some landslides an asymptotic acceleration trend prior to failure suggests a ductile deformation at depth. However there is still little understanding of how these deformation mechanisms operate resulting in failure.

During the winter of 2005, borehole inclinometers installed in a slump on the southeastern Lake Michigan coast were carefully monitored. The slope is historically unstable with maximum movements occurring during the winter-spring cycle. The behavior of three of the active inclinometers from one borehole at depths of 25 ft (7.62 m), 15 ft (4.57 m) and 5 ft (1.52 m) was analyzed with respect to the linear acceleration model described above. Each inclinometer's acceleration trend was examined a few days prior to four significant slope movements. The deepest inclinometer correlated well with the first two slope movements, while the intermediate depth inclinometer correlated well with the third slope movement. Results from the intermediate depth inclinometer also showed bimodal data distribution indicating that the fourth slope movement was actually composed of two separate failure events. When separated, the data showed good correlation. The shallowest inclinometer did not correlate well with the linear model, implying that its movements merely reflect occurrences at depth, better represented by the other two inclinometers. The significance of these results is threefold. Firstly, the acceleration behavior of a landslide can be useful in filtering out false displacement peaks from inclinometer data. Secondly, a good understanding of strain patterns or behavior could lead to the development of accurate prediction models. Thirdly, these results demonstrate the feasibility of using the acceleration model to gain insights into the mechanics and dynamics of rupture propagation in the basal region of an active landslide.