A census-derived building aggregated exposure model (AEM) for Japan

Building exposure, that is, characterization of buildings, number of people occupying it and its replacement cost, for a given region or area, is a primary component when determining building risk due to a hazard or group of hazards. Building characterization includes parameters like construction material, number of stories, lateral load resisting components, etc. There exists a large spatial variation in building exposure, as seen with the development of past global and regional exposure databases (Jaiswal et al., 2008, Yepes-Estrada et al., 2017). Such efforts have also resulted in uniformity and collection of building-type definitions globally.We have developed an exposure model for Japan at the municipality level. This model is entirely derived from open building data (e.g., census) and resulting in an open inventory of uniformly defined building parameters available for risk assessments. The Japanese land and housing database (https://www.e-stat.go.jp/en) provides information about building types and their number, dwelling type, their number and size, tenure type, construction material type, year of construction, number of stories, persons per dwelling and size of dwelling (area in square meters). After interpreting this information, different combinations of building type, construction material, year of construction, and number of stories were mapped to building classes based on the taxonomy of the Global Earthquake Model (GEM). Furthermore, building parameters required for loss assessments (e.g., replacement cost per unit area of a dwelling, number of people per dwelling) were assigned to the identified building class at the municipality level and stored in a database open for public access.The quality of the model is greatly affected by the availability of building parameters and their survey quality in the census data. The commercial and industrial data from the census lacks information on number of stories and have to be derived from other sources for our model. Additionally, the district information defining the model has layers of complexity due to Japan being divided in multiple administrative units at the same administrative level. Such constraints and complexities, limit the flexibility and versatility of the model to be up-scaled or down-scaled from higher to lower resolution and vice versa, also to due to the large scale uncertainty of the survey methods of the data.Despite the limitations, the model gives a clear snapshot of the distribution of regional building classes and their possible count. It also, thus represents an open catalogue of rules for mapping Japanese building attributes to a GEM taxonomy-based building class and, in doing so, creates an opportunity for the scientific and non-scientific community to collaborate on the assessment. We welcome, corrections, and expansion of individual aspects of the aggregated exposure model and like tohopefully encourages the development of such open aggregated exposure models for other countries. References:Jaiswal K, Wald D, Porter K (2010) A global building inventory for earthquake loss estimation and riskmanagement. Earthq Spectra 26:731–748. Yepes-Estrada C, Silva V, Valcárcel J, Acevedo AB, Tarque N, Hube MA, Coronel G, María HS 2017 Modeling the Residential building inventory in South America for seismic risk assessment. Earthquake Spectra 33:299–322. https ://doi.org/10.1193/10191 5EQS1 55DP