Development and Application of a Building Group Destruction Probability Model Based on the Great East Japan Earthquake Tsunami Experience

Abstract

This study examined the regional spatial characteristics of tsunami flooding and building damage using geographic information systems. An analytical model that evaluates total building destruction risk was developed using building damage data from coastal areas in Iwate Prefecture. Building density characteristics in the study area were categorized into four types of environments: (1) many isolated buildings, (2) combination of isolated and neighboring buildings, (3) combination of neighboring and surrounding buildings, and (4) many neighboring buildings. Many isolated buildings were located along the narrow, low-lying areas facing the Pacific Ocean. In comparison, higher building densities were observed along the inner part of the enclosed gulf topography. Most buildings located near the shoreline collapsed. Farther inland, a higher percentage of buildings experienced half-collapse or no-damage. Closer inspection of the varying spatial distribution characteristics and resultant building damage among the 27 target areas led to the identification of several key indicators for predicted building damage including the structure, use, and density of the affected building and the extent of tsunami inundation. Based on the building damage characteristics following the 2011 tsunami, a building group destruction probability model was developed and verified. The proposed model successfully estimated building collapse ratios using the available data.

Change Detection in Heterogeneous Optical and SAR Remote Sensing Images Via Deep Homogeneous Feature Fusion

“…Therefore, we select the optical image as the target feature space for homogeneous transformation to reduce the loss of semantic content during image transformation. The ground truths of changed regions are provided by Yanagawa [36] (the first and the second datasets) and United Nations Institute for Training and Research [37] (the third dataset). To deal with different resolutions between the heterogeneous images, we use the bilinear interpolation [38] to equalize their resolutions for the homogeneous transformation.…”

“…5(a). The buildings near the coasts were severely damaged by the earthquake and tsunami [36]. The pre-event optical image was acquired with the size of 1250 × 1250 pixels in September 2009.…”