Landslide monitoring using seismic ambient noise correlation: challenges and applications

Abstract

Monitoring landslides is essential to understand their dynamics and to reduce the risk of human losses by raising warnings before a failure. A decade ago, a decrease of apparent seismic velocity was detected several days before the failure of a clayey landslide, that was monitored with the ambient noise correlation method. It revealed its potential to detect precursor signals before a landslide failure, which could improve early warning systems. To date, nine landslides have been monitored with this method, and its ability to reveal precursors before failure seems confirmed on clayey landslides. However three challenges remain for operational early-warning applications: to detect velocity changes both rapidly and with confidence, to account for seasonal and daily environmental influences, and to check for potential instabilities in measurements. The ability to detect a precursory velocity change requires to adapt the processing workflow to each landslide: the key factors are the filtering frequency, the correlation time window, and the choice of temporal resolution. Other optional processing steps are described, to better measure rapid velocity changes, improve signal-to-noise ratio, or estimate the measurement uncertainty. The velocity also fluctuates seasonally, by 1 to 6% on the reviewed landslide studies, due to environmental influences. This review reveals a linear trend between the amplitude of seasonal fluctuations and the filtering frequency over the 0.1-20 Hz range, encompassing both landslide and non-landslide studies. The environmental velocity fluctuations are caused mostly by groundwater levels and soil freezing/thawing, but could also be affected by snow height, air temperature and tide depending on the site. Daily fluctuations should also occur on landslides, and can be an issue when seeking to obtain a sub-daily resolution useful for early-warning systems. Finally, spurious fluctuations of apparent velocity-unrelated to the material dynamics-should be verified for. They can be caused by changes in noise sources (location or spectral content), in site response (change of scatterers, attenuation, or resonance frequency due to geometrical factors), or in inter-sensor distance. As a perspective, the observation of seismic velocity changes could contribute in as-2 sessing a landslide stability across time, both during the different creeping stages occurring before a potential failure, and during its reconsolidation after a failure.

Controls on Seasonal Variations of Crustal Seismic Velocity in Taiwan Using Single‐Station Cross‐Component Analysis of Ambient Noise Interferometry

“…The peak‐to‐peak amplitude of periodic dv/v variations ranges from 0.02% to 0.2% (Figure S6), which is roughly consistent with the dv/v range of 0.01%–0.1% for the frequency band of 0.1–1 Hz in Le Breton et al. (2021). The time series of dv/v and weather data consistently have periodic cycles like in the example shown by Station MASB (Figure 3).…”

Effect of snowfall on changes in relative seismic velocity measured by ambient noise correlation

“…Among environmental factors, we assume that snowpack changes play the major role, leading to surface wave velocity fluctuations consecutive to snowfalls or snowmelt events. For example, atmospheric pressure changes may probably influence measured dV /V , but we expect the amplitude of this effect to be negligible (less than 0.1 % for a variation of a few kilopascals) (Le Breton et al, 2021;Hotovec-Ellis et al, 2014).…”

“…3), which are known to be sensitive to elastic properties of the extended subsurface between sensors. In these time windows, we estimated the relative velocity change (dV /V ) and the corresponding correlation coefficient (CC) by using the stretching method (Hadziioannou et al, 2011;Le Breton et al, 2021). We thus have dV /V time series with four values per day during the entire data period, in different frequency bands ranging from 10 to 25 Hz with a bandwidth of 4 Hz.…”

Using Ground Microtremor Data in Advanced Rockfall Early Warning Systems and Predicting Spatiotemporal Characteristics of Rockfall Hazards

“…When a rock slope becomes dangerous, the measured seismic response exhibits strong directional ampli cation 23,24 . Compared to displacement analysis, these results show that vibration analysis signal characteristics can be used to improve early warning approaches of rock collapse 2,3 . However, seismic monitoring has temporal and spatial limitations because it can record signals after a seismic event in only a seismic zone.…”