Fine divided method of active tectonic blocks based on adaptive density peak clustering and improved Euler model, revealed by GNSS velocities in Southwest Japan

Abstract

The rational division of active tectonic blocks is an important premise for distinguishing active faults, discovering hidden faults, and an in-depth understanding of regional crustal deformation characteristics, seismicity, and tectonic dynamics. In recent years, clustering algorithms have been widely used in active block division owing to their high efficiency and reliability. However, the classical clustering algorithm sensitivity to the block boundary under different conditions is inconsistent. Additionally, the Euler model constraint (EMC) process involved in the classical clustering algorithm ignores the influence of the boundary points and gross error points of the active block. Therefore, in this study, a new active block division method based on adaptive density peak clustering (ADPC) and an improved Euler model constraint (IEMC) is proposed. The simulation test results show that compared with the traditional clustering algorithm, the mean error rate of the new method can be reduced by up to 40% in the simple single-fault model and by 10% in the complex three-segment model. The new method was further applied to an actual case of an active block division in southwest Japan. The results further show that the new method is less affected by the regional environmental conditions. Compared with previous results in southwest Japan, the proposed method has higher accuracy and higher reliability of the active block division results. The new method for active block division proposed in this study can be used to accurately explore the location and quantity of block boundaries.