Heterogeneity of Tectonic Regimes in Sulawesi Island: Applying Unsupervised Machine Learning to Focal Mechanism Data

Abstract

Abstract Sulawesi Island was formed in the tripel junction area between the Sunda plate, Australian plate, and Philippine Sea plate. The collision between the three plates resulted in Sulawesi island having a variety of fault mechanisms. To better understand the tectonic processes in Sulawesi, we clustered earthquake events in the region, which were then analyzed based on their focal mechanisms. The data we used is the distribution of focal mechanisms around the island of Sulawesi for 1390 events from 1967 to 2022 that we obtained from a combination of the ISC GEM, Global CMT, and BMKG earthquake catalogs. We applied one of the unsupervised machine learning technique that is Density-Based Spatial Clustering of Application with Noise (DBSCAN) to cluster these earthquake events. This method clusters earthquake events based on the density of the events using two parameters: the minimum number of events that occur in cluster (minPts), and distance radius for neighbouring points (𝓔). Futhermore, we analysed the density of the focal mechanism in each cluster to comprehend its fault mechanism heterogeneity. We obtained ten earthquake clusters, using the parameters minPts=15 and 𝓔=0.40, where these parameters explain the reliability of the tectonic setting in Sulawesi. Based on the clusterization, we infer that sevent clusters have compressional regimes located in the northern arm of Sulawesi, the Molucca Sea Double Subduction area, the West Sangihe trench, the North-Vergent thrust, and the Nusa Tenggara Back Arc thrust. Additionally, we found that the clusters associated with the Palu-Koro Fault and Matano Fault exhibit characteristics of a transform regime, while the cluster located between the Lalanga and Tongian ridges is characterized by an extensional setting.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint