Sulaimaniyah 2015 Earthquake Swarm Analysis Using Matrix Profile TechniqueI

Abstract:

In this study, we use the relatively new Matrix Profile algorithm of time series data mining to analyze a large dataset of seismic waveforms from northeastern Iraq to identify large numbers of individual events that comprise a complex earthquake swarm. The seismic dataset consists of high-frequency records from the NISN network and KSIRS array that contain earthquake swarm that occurred during January to February of 2015 near and around southwest of Sulaimaniyah, Iraq. It is unlikely that this earthquake swarm is related to magmatic or hydrothermal activity, but rather is probably caused by stress triggering. All seismic events associated with the swarm were quite shallow, less than 15 km into the crust, and quite small, with mb magnitudes less than 2. The earthquake swarm activity was not felt by the local population. The high degree of similarity among the waveforms suggests a common tectonic source. The seismic swarm consisting of 110 events was initially identified by Array Information Technology using the traditional method of LocSAT. However, we then applied the Matrix Profile method to assist in time series analysis and mining. The Matrix Profile algorithm is a measure of similarity between subsequences within a time series, and in this case we used a SCAlable Matrix Profile (SCAMP), which computes an index matrix profile correlation that is assigned to incoming subsequences. This provided correlation peaks that allowed us to increase the number of identified swarm events by 14%.