Acoustic and Microseismic Characterization in Steep Bedrock Permafrost on Matterhorn (CH)

Abstract

Understanding of processes and factors influencing slope stability is essential for assessing the stability of potentially hazardous slopes. Passive monitoring of acoustic emissions and microseismology provides subsurface information on fracturing (timing and identification of the mechanism) and therefore complement surface displacement data. This study investigates for the first time acoustic and microseismic signals generated in steep, fractured bedrock permafrost, covering the broad frequency range of 1 − 105 Hz. The analysis of artificial forcing experiments using a rebound hammer in a controlled setting led to two major findings: First, statistically insignificant cross correlation between signals indicates that waveforms change strongly with propagation distance. Second, a significant amplification is found in the frequency band 33–67 Hz. This finding is strongly supported by evidence from artificial rockfall events and more importantly by naturally occurring fracture events identified in fracture displacement data. Thus, filtering this frequency band enables enhanced detection of microseismic events relevant for slope stability assessment. The analysis of 2‐year time series in this frequency band further suggests that the detected energy rate baseline of all automatically triggered events using the STA/LTA algorithm is not sensitive to temperature forcing, an observation of primary importance for long‐term data collection, analysis, and interpretation. The event detection in the established frequency band is not only improved but also not affected by the short‐ and long‐term temperature changes in such a rapidly changing environment.

Systematic Identification of External Influences in Multi-Year Micro-Seismic Recordings Using Convolutional Neural Networks

“…The subset is created by collecting continuous data from the sensors and additional data, which helps to characterize the external influences. In the presented case the additional data is non-continuous and consist of local observations, pre-processed STA/LTA triggers from (Weber et al, 2018), accommodation occupancy of a nearby hut and a non-exhaustive list of helicopter flight data from a duration of approximately 7 weeks provided by a local helicopter company. Before an annotated subset can be created the collected data must be characterized for its 20 usefulness in the annotations process, i.e.…”

“…The test set (2017) is chosen since wind data is not available for the whole training set due to a malfunction of the weather station in 2016. The experiment from (Weber et al, 2018) provides STA/LTA event triggers for 2016 and 2017. Table 1 shows statistics for several categories, which are 3 external influences and one category where none 5 of the 3 external influences are annotated (declared as "None").…”

“…The data used in this paper originate from a multi-sensor and multi-year experiment (Weber et al, 2018) focusing on slope 25 stability in high-alpine permafrost rock walls and understanding the underlying processes. Specifically, the sensor data is collected at the site of the 2003 rockfall event on the Matterhorn Hörnligrat, (Zermatt, Switzerland) at 3500 m a.s.l.…”

“…The micro-seismic segment's start time is set to 1 minute before the image timestamp. Event-linked segments are extracted based on the STA/LTA triggers from (Weber et al, 2018). For each trigger 10 seconds following the timestamp of the trigger are extracted from the micro-seismic signal.…”

“…In (Weber et al, 2018) a seismometer is used to assess the seismic activity by using an STA/LTA event detector. The seismometer is chosen as the primary sensor and STA/LTA triggering is used as the reference method to assess seismic activity.…”