Segmentation geometry of strike-slip fault systems in slow-deforming regions: a proposed method and case study of the Yangsan Fault, South Korea

Fault location and geometry are the most fundamental input data in seismic hazard analysis, the ultimate aim of which is to mitigate damage from future large earthquakes. In regions prone to large earthquakes or where cumulative deformation by multiple earthquake events are well expressed in the landscape, fault models are constructed primarily by (1) identifying active fault traces, mapped mostly by the surface ruptures associated with large earthquakes; (2) simplifying fault traces while capturing their geometrical characteristics; and (3) segmenting the simplified geometry, given that a single earthquake does not always rupture the entire length of a fault system. In slowly deforming regions, however, the construction of fault models is challenging, even though geologic records of large earthquakes exist, because of the lack of clear active fault traces. Indeed, surface-rupturing earthquakes may not be part of the historical periods owing to their long recurrence time of thousands of years or more. Nevertheless, seismic hazard analysis is required for densely populated and industrial areas in slowly deforming regions, such as South Korea. On the basis of criteria established previously for determining segmentation geometry in fault models, here we propose a methodology for identifying the segmentation geometry of strike-slip fault systems in slowly deforming regions. In terms of the criteria used to identify segment boundaries, we examine along-fault variations not only in fault geometry but also in fault-surrounding lithology and fault-related geomorphic features. We test the methodology for assessing the fault segmentation geometry in a case study of the Yangsan Fault, which is one of the most active seismogenic strike-slip faults on the Korean Peninsula. Results show that the ∼200 km length of the Yangsan Fault on land consists of 12 to 15 distinct fault segments. We discuss how models of fault segmentation geometry are able to improve seismic hazard analysis in regions that have not experienced surface-faulting earthquakes in historical period.