Multi-scale seismic characterization of a landslide : the case of the Séchilienne deep-seated landslide (French Alps)

Landslides pose a serious threat to populations and infrastructure. Monitoring their activity makes it possible to detect episodes of reactivation and to identify their controlling factors, such as hydro-meteorological conditions and water infiltration. Multiparametric geophysical instrumentation helps to analyze landslide dynamics, detect precursors of destabilization or rupture, and better understand the influence of external triggers. It typically combines surface deformation measurements, hydrological sensors, meteorological data and seismic instrumentation.
This study focuses on the seismic signals recorded at the Séchilienne landslide, located about 25 km southeast of Grenoble, in the southern Belledonne massif (Fig. 1A). The slope has long been instrumented because it threatens the river and road below. Between 2006 and 2013, the landslide reactivated and its surface velocity increased strongly ; it has since slowed down, with displacement rates now lower than before 2006. This change in dynamics is thought to be linked to changes in the hydro-mechanical properties, although the mechanisms involved remain poorly understood. Owing to the high associated hazard, the site benefits from operational surveillance and, through the national observation service OMIV, is equipped with research instrumentation, providing more than 15 years of observations, notably seismological ones.
On this landslide, previous seismic analyses include the detection and characterization of landslide-related seismicity, which is typically composed of rockfalls and of microearthquakes caused by rock-mass fracturing [1,2]. In addition, ambient noise seismic interferometry, which analyzes temporal variations in surface-wave velocity (dV/V), provides a complementary way to monitor landslide properties [3]. It typically reveals seasonal fluctuations linked to hydrological and/or temperature changes, as well as rheological changes associated with landslide reactivation.
In this project, we propose to analyze 15 years of continuous seismic data using a network-based approach that exploits the coherence of signals recorded simultaneously by several receivers of the network. Analyzing the network covariance matrix may allow various types of seismic events to be detected and clustered into families. The intern will use the Python library CovSeisNet [4] to analyze the spatial coherence of signals recorded by the seismic network and find these families. This will allow the seismic content of the recordings to be explored without a priori assumptions, and potential new classes of signals to be detected. We hope to be able to separate, with better resolution than before, anthropogenic signals and local noise from the signals originating from the landslide.
The results will be compared with previous seismic analyses (seismic catalogues and dV/V variations, Fig. 1C) to assess the reliability of each approach. They will also be compared with deformation measurements (Fig. 1B) and environmental parameters, such as precipitation and water infiltration. This long-term analysis should provide new insight into the long-term dynamics of this landslide.

Figure 1 : (A) Map Of the landslide area, including the seismic instrumentation. (B) Landslide displacement rates over the last 30 years. Note the strong acceleration/deceleration around 2013. (C) Previous seismic analysis : Seismic noise monitoring (light grey curve, dV/V in the 6-9Hz range), and histograms of detected of microquakes (blue) and rockfalls (orange).

References :
[1] Helmstetter, A., et S. Garambois. « Seismic Monitoring of Séchilienne Rockslide (French Alps) : Analysis of Seismic Signals and Their Correlation with Rainfalls ». Journal of Geophysical Research : Earth Surface 115, no F3 (2010). https://doi.org/10.1029/2009JF001532.
[2] Lacroix, P., et A. Helmstetter. « Location of Seismic Signals Associated with Microearthquakes and Rockfalls on the Séchilienne Landslide, French Alps ». Bulletin of the Seismological Society of America 101, no 1 (2011) : 341-53. https://doi.org/10.1785/0120100110.
[3] Le Breton, M., N. Bontemps, A. Guillemot, L. Baillet, et É. Larose. « Landslide monitoring using seismic ambient noise correlation : challenges and applications ». Earth-Science Reviews 216 (mai 2021) : 103518. https://doi.org/10.1016/j.earscirev.2021.103518.
[4] Tong, F., Seydoux, L., Journeau, C., & Soubestre, J. (2024). Covseisnet Python package. covseisnet/covseisnet : Associate repository with a Zenodo DOI (1.0.0) [Software]. Zenodo. https://doi.org/10.5281/zenodo.10990032