Detection of building damage caused by Van Earthquake using image and Digital Surface Model (DSM) difference

Abstract

ABSTRACTAn earthquake occurred at Van City on 23 October 2011 at 13:41 local time with the local magnitude (ML) 6.7 and moment magnitude (Mw) 7.0. Approximately seventeen thousand buildings collapsed or were damaged, and 644 people died because of the main earthquake and its aftershocks. One hundred and fifty-two aerial images of the earthquake area covering 1,296 km2 were taken with an UltraCam X large format digital aerial camera with 30 cm ground sample distance (GSD) just one day after the earthquake on 24 October 2011. This study attempted to detect damaged buildings automatically with the help of pre and post-earthquake aerial images. With the image and digital surface model (DSM), different methods are used for damage detection, but it was seen that these methods are not satisfactory. So, a novel approach that uses the geometric characteristics of buildings, i.e. area and area to perimeter ratios, was introduced to improve the results. The results show that ‘Area/Perimeter’ approach improves the damage detection accuracy considerably.