Euler-pole clustering of GNSS velocities using unsupervised machine learning in the Southeastern Tibetan Plateau: Crustal block identification and the dominance of sinistral-slip faults

Abstract

Previous studies have constrained fault slip rates and crustal block geometries of the Southeastern Tibetan Plateau (SETP) with contradictory results due to complex geodynamics and deformation patterns as well as subjective choices of crustal block boundaries. In this work, we address the issue of uncertain crustal block geometries by employing an unsupervised machine learning Euler pole clustering algorithm that automatically resolves regions that behave as rigid blocks (clusters) rotating on a sphere using GNSS velocity vectors. Optimal clustering results, determined by F-test and Euler-vector (angular velocity vector) overlap analyses, indicate 4 elongated crustal blocks exist in the SETP that are approximately parallel and delineated by a set of arcuate sinistral-slip faults. Our clustering results redefine the first-order kinematics of the SETP region with new crustal block definitions that elucidate the dominance of sinistral-slip faults.