High-accuracy real-time microseismic analysis platform: case study based on the super-sauze mud-based landslide

Understanding the evolution of landslide and other subsurface processes via microseismic monitoring and analysis is of paramount importance in predicting or even avoiding an imminent slope failure (via an early warning system). Microseismic monitoring recordings are often continuous, noisy and consist of signals emitted by various sources. Automated analysis of landslide processes comprises detection, localization and classification of microseismic events (with magnitude <2 richter scale). Previous research has mainly focused on manually tuning signal processing methods for detecting and classifying microseismic signals based on the signal waveform and its spectrum, which is time-consuming especially for long-term monitoring and big datasets. This paper proposes an automatic analysis platform that performs event detection and classification, after suitable feature selection, in near realtime. The platform is evaluated using seismology data from the Super-Sauze mud-based landslide, which islocated in the southwestern French Alps, and features earthquake, slidequake and tremor type events.