Cluster Analysis of Marine Terraces and Quantitative Seismotectonic Interpretation of the Boso Peninsula, Central Japan

Abstract

AbstractSubjectivity frequently arises in the classification of marine terraces using conventional methods, impacting the robustness of geodynamic inferences made based on those terraces. This paper proposes a new method for classifying marine terraces using a digital elevation model (DEM) and a statistical clustering analysis which provides us quantitative data sets of prehistorical crustal deformations along the subduction zones. We focus on the characteristics of marine terraces with paleoshoreline angles (the bases of abandoned erosional cliffs) that are approximately parallel to each other in an elevation view and adopt modified K‐means clustering to group them. This method can simultaneously provide observational data on various geomorphological parameters of marine terraces, such as the width of the terrace platforms, the curvature of cliffs, and accurate elevation distributions of the paleo–sea levels. We applied this method to the Holocene marine terraces developed on the Boso Peninsula, central Japan, called Numa terraces, which are formed by recurrent interplate earthquakes. Four Numa terrace levels are successfully distinguished by this method solely based on objective processing. We evaluated the degree of weathering of the terrace landform and found it matches with terrace age. By quantifying elevations of Numa terraces and their spatial variation, we see the cumulativeness of tilting of the terrace elevations, dipping up toward the trough axis, and that the relative terrace heights are similar despite different time intervals between terraces. These findings are inconsistent with the prediction of a previously proposed dislocation model and provide essential new constraints for the slip source of the paleo‐earthquakes.