UC Berkeley

We deployed seismic and infrasound sensors at a historically active cliff in Yosemite Valley for the purpose of detecting and locating rock falls at local (<1km) distances and demonstrate the potential for using these techniques for real-time rock fall monitoring. The project ran for two winters: the first deployment was a system feasibility study consisting of a single station with a geophone and an accelerometer; the second deployment was a network of seven stations at four different locations with the addition of infrasound sensors. We demonstrated that small (<20m³) rock falls are detectable at distances of several hundred meters, individual impacts can be identified, and seismic waves are generated prior to the first main impact for some rock falls. We also found that infrasound is viable and compliments seismic, especially for locating events. We correlated the data with environmental conditions and extracted information about the initiation, triggering, and dynamics of rock falls. A major part of the research effort was the development of a triggering algorithm and criteria for distinguishing rock falls from thousands of seismic triggers. Twelve rock falls were identified in the continuous seismic recording by searching for triggers and comparing them with known rock falls and other forms of seismic activity. Physical evidence or reports of rock falls exist for only eight of the twelve rock falls that we identified; thus, we have demonstrated that instrumented monitoring can significantly augment the detection of rock falls even in heavily-trafficked areas such as Yosemite Valley. Six of the rock falls appear to be related to each other as an ongoing instability, while the rest appear to be independently occurring events. After we identified the individual rock fall events, we focused on characterizing the seismic data in terms of timing, frequency, and P/S/Rayleigh wave phases in order to develop a set of characteristic parameters indicative of rock fall signals.