Detection of building collapse from the shadow lengths in optical satellite images

ABSTRACT: Visual interpretation of optical satellite images is widely used to detect building damages after recent earthquakes in the world. However, since optical satellites images are acquired from the vertical direction, they can observe only roofs of buildings. Building damages such as mid-story collapse are often overlooked because the upper surfaces of buildings do not change so much. This paper proposes the method to detect this kind of building collapses from the change in building heights, estimated from the shadow lengths measured before and after an earthquake. Boumerdes city, Algeria, is selected as a case study site where many buildings were collapsed in the 2003 Boumerdes Earthquake. Two QuickBird images, taken before and two days after the earthquake, were employed for analysis. Base on the measurement of shadow lengths, some mid-story collapses were successfully detected, which could not be detected by visual interpretation. Although a limitation still remains in measuring shadow lengths in densely build-up or highly vegetated areas, the proposed method can help to reduce omission errors, which may be associated in visual damage interpretation of vertical satellite/aerial images