Earthquakes have claimed approximately 8 million lives over the last 2,000 years (Dunbar, Lockridge and others, 1992) and fatality rates are likely to continue to rise with increased population and urbanizations of global settlements especially in developing countries. More than 75% of earthquake-related human casualties are caused by the collapse of buildings or structures (Coburn and Spence, 2002). It is disheartening to note that large fractions of the world's population still reside in informal, poorly-constructed & non-engineered dwellings which have high susceptibility to collapse during earthquakes. Moreover, with increasing urbanization half of world's population now lives in urban areas (United Nations, 2001), and half of these urban centers are located in earthquake-prone regions (Bilham, 2004). The poor performance of most building stocks during earthquakes remains a primary societal concern. However, despite this dark history and bleaker future trends, there are no comprehensive global building inventories of sufficient quality and coverage to adequately address and characterize future earthquake losses. Such an inventory is vital both for earthquake loss mitigation and for earthquake disaster response purposes. While the latter purpose is the motivation of this work, we hope that the global building inventory database described herein will find widespread use for other mitigation efforts as well.
For a real-time earthquake impact alert system, such as U.S. Geological Survey's (USGS) Prompt Assessment of Global Earthquakes for Response (PAGER), (Wald, Earle and others, 2006), we seek to rapidly evaluate potential casualties associated with earthquake ground shaking for any region of the world. The casualty estimation is based primarily on (1) rapid estimation of the ground shaking hazard, (2) aggregating the population exposure within different building types, and (3) estimating the casualties from the collapse of vulnerable buildings. Thus, the contribution of building stock, its relative vulnerability, and distribution are vital components for determining the extent of casualties during an earthquake.
It is evident from large deadly historical earthquakes that the distribution of vulnerable structures and their occupancy level during an earthquake control the severity of human losses. For example, though the number of strong earthquakes in California is comparable to that of Iran, the total earthquake-related casualties in California during the last 100 years are dramatically lower than the casualties from several individual Iranian earthquakes. The relatively low casualties count in California is attributed mainly to the fact that more than 90 percent of the building stock in California is made of wood and is designed to withstand moderate to large earthquakes (Kircher, Seligson and others, 2006). In contrast, the 80 percent adobe and or non-engineered masonry building stock with poor lateral load resisting systems in Iran succumbs even for moderate levels of ground shaking. Consequently, the heavy death toll for the 2003 Bam, Iran earthquake, which claimed 31,828 lives (Ghafory-Ashtiany and Mousavi, 2005), is directly attributable to such poorly resistant construction, and future events will produce comparable losses unless practices change. Similarly, multistory, precast-concrete framed buildings caused heavy casualties in the 1988 Spitak, Armenia earthquake (Bertero, 1989); weaker masonry and reinforced-concrete framed construction designed for gravity loads with soft first stories dominated losses in the Bhuj, India earthquake of 2001 (Madabhushi and Haigh, 2005); and adobe and weak masonry dwellings in Peru controlled the death toll in the Peru earthquake of 2007 (Taucer, J. and others, 2007). Spence (2007) after conducting a brief survey of most lethal earthquakes since 1960 found that building collapses remains a major cause of earthquake mortality and unreinforced masonry buildings are one of the mos