On the Use of Variational Mode Decomposition for Seismic Event Detection

In this paper, we investigate the use of Variational Mode Decomposition (VMD) for the analysis of seismic signals obtained from the Cotopaxi Volcano in Ecuador. The VMD method is proposed here as a method for noise attenuation to improve the event detection and the identification of the starting and end points of seismic events. The main advantage of the VMD method over previous studied methods is its robustness for reducing noise and the number of features necessary to distinguish amongst the signals. Preliminary analysis shows that seismic events such as volcano-tectonic (VT) earthquakes, and long-period (LP) events, can be identified after applying the VMD to the seismic signal due to the fact that the modes obtained are considerably different between these types of seismic events, therefore this decomposition could also be used to extract features for an automatic classifier. Further observations show that the same process used for obtaining the modes of the signal can also be applied to detect the presence of events using a fixed-size window and an amplitude threshold, the numerical results show a 99.26% accuracy for obtaining the events onset and ending points.