Multidisciplinary monitoring of progressive failure processes in brittle rock slopes - Concepts and system design

The evolutionary failure processes leading to large-scale mass movements in massive crystalline rock slopes are the subjects of a multidisciplinary research project in the Swiss Alps. Focus is directed towards detecting and analysing rockslide processes that involve the progressive development of a failure surface as opposed to sliding along a pre-existing one. In order to monitor the underlying mechanisms of progressive failure, several new and conventional instrumentation systems were combined with an existing in situ monitoring program at an active rockslide site in the Valais (Switzerland). Design of the instrumentation network is based on site investigations and preliminary geomechanical models of the acting rockslide processes with respect to the rate of displacements, position and orientation of geological features that delineate the unstable rockmass. The network set-up considers additional findings from borehole logging and testing. Parameters that will be measured include microseismicity, fracture patterns and the temporal and spatial evolution of 3-D displacement fields and fluid pressures. fields, microseismicity, fracture patterns and the temporal and spatial ev lution of fluid pressures. persistent discontinuities and frictional sliding with asperity interlocking (i.e. stick slip). To identify and discriminate between these different processes, several forms of surface and subsurface information are required. First, it is essential to properly investigate the 3-D geological structure of the monitored rock mass. Important geometrical parameters include discontinuity orientation, spacing, persistence and connectivity, and the location of discrete surface features (Einstein et al. 1983). This information can be attained through a combination of geological mapping, detailed discontinuity mapping, both at surface and in boreholes, and active geophysical testing. New developments in 3-D seismics, 3-D georadar, crosshole tomography and borehole to surface testing methods show promising trends with respect to improving the quality of geological models based on surface mapping data (e.g. Schepers et al. 2001). More specific problems related to identifying and characterizing progressive failure processes can be addressed through the combined monitoring of mass movement kinematics, pore pressures and microseismic activity. Each of these physical quantities must be captured in terms of their 3-D spatial distribution and time (i.e. 4-D) to provide sufficient data quality for a mechanism-directed analysis. The kinematics of a sliding rock mass generally manifests itself in terms of surface displacements and subsurface deformation. Surface displacements are commonly derived through geodetic measurements. Of great value are surveys providing information on displacement rates and directions. Continuous monitoring of surface displacements is less common, although new developments in GPS techniques are answering this need. In terms of monitoring subsurface deformation fields, several boreholebased methods are available to provide information that is continuous in time as well as continuous along the borehole axis (Kovari 1990). For example, the 3-D deformation field around a borehole can be measured using a combined inclinometerextensometer system (e.g. Solexperts’ TRIVEC or Interfels’ INCREX systems). However, these measurements are only taken periodically and must be supplemented by multipoint in-place inclinometers to provide continuous real-time monitoring. As such, continuous monitoring is restricted to multiple zones of interest where deformations are expected or previously measured using conventional inclinometer probe surveys. Another key to monitoring progressive rock slope destabilization is to focus in on the localized failure mechanisms underlying the development of fracture systems. Microseismic monitoring provides a key tool in this respect. Precise mapping of microearthquake swarms can resolve the geometry and extent of the developing structure (Fehler et al. 2001), Figure 1. a) Location and digital elevation model of project area (DEM provided by CREALP); b) cross-section showing 1991 rockslide (after Wagner 1991); c) existing early warning system and distribution of surface cracks. 1 European Conference on Landslides, 24-26 June, 2002. Prague, Czech Republic. pp. 477-483. whilst fault plane solutions of the microseisms themselves obtained through moment tensor inversion provides insight as to its microstructure by defining the local orientation of the failure plane and the nature of slip on it (Nolen-Hoeksema & Ruff 2001). Such techniques have been successfully applied in mining (Mendecki 1997), hydraulic fracture mapping (Phillips et al. 1998), and large-scale tectonic investigations (Maurer et al. 1997). Building on these developments, it is believed that microseismic monitoring will provide a key means to image rockslide dynamics. Furthermore, it is important to recognize that these techniques must be applied at depth, i.e. through the deployment of subsurface geophones, in order to provide sufficient data quality. The final key component to monitoring progressive rockslide processes is pore water pressures. In a fractured rock mass, water pressures act to drive fracture propagation (e.g. sub-critical fracture propagation), and during periods of significant increases, acts as an important triggering factor. To identify the influence of fracture water pressures on rock slope destabilization, piezometric conditions at depth must be recorded continuously. Temporal variations in measured water pressures can also provide additional information with respect to identifying active fracture processes, sliding mechanisms and/or dilatancy effects (e.g. Scholz 1990). By integrating these different systems into one monitoring network design, it is believed that the key elements required to meet the objectives of this study (focussing on the progressive development of brittle rock slope sliding surfaces) can be measured and analysed. 3 SITE SELECTION AND INVESTIGATION 3.1 Study site Randa Based on the preliminary investigation of several sites across the Swiss Alps (Eberhardt et al. 2001), an active rockslide site was selected near the village of Randa (Canton Valais, Switzerland) in the Matter Valley (Fig. 1a). This site was chosen based on the identification of several indicators suggesting that the destabilizing mechanisms relate to progressive brittle fracture processes. These included the massive crystalline nature of the rock mass, the absence of highly persistent discontinuities dipping out of the slope along which a pre-existing slide plane may be inferred, the presence of open fractures, relatively small displacement rates and observations relating to an earlier massive rockslide at the site (the 1991 Randa rockslide). An additional consideration with respect to the microseismic component of the planned instrumentation network was the relatively low background noise level (e.g. that arising from heavy vehicle traffic). The research area covers a 500 x 500 m area between elevations of 1800 and 2650 m above sea level. The area belongs to the Penninic SiviezMischabel nappe. Its lithology comprises polymetamorphic gneisses, schists and amphibolites (paragneisses) and metamorphosed Permian granite intrusions (orthogneiss, Randa Augengneiss); the metamorphosed Permian-Triassic sedimentary cover is not included in the project area. In terms of surface topography, the lower boundary of the research area is defined by the back scarp of an earlier rockslide, which occurred as two main slide events in April and May 1991 with an estimated total volume of 30 million m (Schindler et al. 1993). By reviewing the general geological situation and initial analysis of the 1991 slide, much can be inferred with respect to the present-day instability, especially with regards to the progressive nature of the failure mechanisms. As foliation is dipping into the slope, the most important discontinuities contributing to slope instability were identified by Wagner (1991) as persistent cross-cutting joint sets along which sliding was believed to have occurred (Fig. 1b). These persistent joints can be observed in the scarp of the earlier slide but are more limited in persistence when encountered in surface outcrops. In terms of triggering factors for the 1991 rockslide, analysis of climatic and regional seismic data showed no clear indications of a triggering event (Schindler et al. 1993). Permafrost distribution models of the Matter Valley by Gruber & Hoelzle (2001) likewise show that permafrost was not a contributing factor. Instead, Eberhardt et al. (2001) suggest timedependent mechanisms relating to strength degradation (e.g. through weathering, brittle fracturing, etc.) and progressive failure as causing the failure.