Automated seismic monitoring of mass movements in the Mont-Blanc massif

Mass movements in the Mont-Blanc massif (French Alps) are traditionally monitored by a network of human observers (AlpRisk and ObsAlp networks in the past and Regard d’Altitude currently) and annual LiDAR campaigns for some areas. These observations provide accurate locations and estimates of event size. However, those approaches have several limitations: observations are biased toward areas frequently visited by people and are potentially incomplete in remote regions. In addition, temporal accuracy is frequently poor (except during peak periods for mountaineers), as many observations are based on debris deposits rather than on the events themselves.

Seismic monitoring using permanent seismic stations installed in the area offers a promising complementary solution to these limitations. Rapid mass movements such as serac collapses and rockfalls generate particular seismic signals, providing excellent temporal resolution and continuous coverage, including in areas that are rarely observed directly. Their seismic signatures differ significantly from those of earthquakes, requiring dedicated methods for event localization and size estimation.

Based on field observations of mass movements and the Sismalp seismic event catalog, we compiled a reference catalog currently consisting of 107 seismic events associated with 91 field observations, including volume estimates for 55 events. This catalog was used to fine-tune and evaluate automated algorithms for the localization and size estimation of mass movements using seismic data.

Mass movement localization is performed using a combination of an amplitude decay method and the BackTrackBB algorithm based on signal coherence. We achieved a median location accuracy of 1.6 km and observed a significant improvement in localization accuracy with increased seismic station coverage. Event size was estimated using a simple linear model based on seismic energy, resulting in a median relative error of approximately 70 %.

Our results show that automated seismic monitoring of mass movements can be successfully applied in remote high-mountain environments. The performance of our method can be further improved by increasing the number of seismic stations and by improving data processing techniques.