Sadegh Karimpouli published a paper in Geophysical Journal International entitled “Unsupervised clustering of catalog-driven features for characterizing temporal evolution of labquake stress”. The paper is supported by funding from DT-GEO project. In the paper we use 47 physics-based temporal features derived from acoustic emission data recorded during triaxial stick-slip experiments performed on the rough faults to classify the fault state. We reveal three different stages (clusters, in the sense of K-means approach applied) of fault: a) the stable stage corresponding to the initial stable deformation of the sample, b) transitional stage, and the c) unstable stage in which the large slip (large laboratory earthquake) can occur anytime. The latter two stages occurring transiently at high level of stress, highlighting the evolution of cm-scale asperities on the rough fault surface. Our findings hold promise for advancing the earthquake prediction methodologies based on laboratory experiments and catalog-driven features.
Reference: