Abstract. Regional monitoring of rock slope failures by the seismic technique is rarely studied due to significant source location errors, and it still lacks the signal features needed for understanding events of this type because of the complex mass movement involved. To better understand events of this type, ten known events along highways in Taiwan were analyzed. First, a hybrid approach (GeoLoc) composed of cross-correlation-based and amplitude-attenuation-based approaches was applied, and it produced a location error of maximum 3.19 km for the ten events. Then, we analyzed the ratio of local magnitude (ML) and duration magnitude (MD) and found that a threshold of 0.85 yields successful classification between rock slope failure and earthquake. Further, the GeoLoc can retrieve the seismic parameters, such as signal amplitude at the source (A0) and ML of events, which are crucial for constructing scaling law with source volume (V). Indeed, Log(V) = 1.12 ML + 3.08 and V = 77,290 A00.44 derived in this study provide the lower bound of volume estimation, since the seismic parameters based on peak amplitudes cannot represent the full process of mass loss. Second, while video records correspond with seismic signals, the processes of toppling and sliding present column- and V-shaped spectrograms, respectively. The impacts of rockfall directly link directly to the pulses of seismic signals. Here, all spectrogram features of events can be identified by event volumes larger than 2,000 m3, corresponding to the farthest epicenter distance ~2.5 km. The previous results were obtained using the GeoLoc scheme for providing the government rapid reports for reference. Finally, a recent event on 12th June 2020 was used to examine the GeoLoc scheme’s feasibility. We estimated the event's volume by the two scalings: 3,838 m3 and 3,019 m3, which was roughly consistent with the volume estimation of 5,142 m3 from the digital elevation model. The physical processes, including rockfall, toppling, and complex motion behaviors of rock interacting with slope inferred from the spectrogram features were comprehensively supported by the video record and field investigation. We also demonstrated that the GeoLoc scheme, which has been implemented in Sinwulu catchment, Taiwan, can provide fast reports, including the location, volume, and physical process of events of this type to the public soon after they occur.
Supplementary material to "Locating rock slope failures along highways and understanding their
physical processes using seismic signals"
