Influence of the spatial distribution of hazard and exposure models on urban seismic risk assessment

The large number of assets in urban centers often imposes computational limitations on the evaluation of the seismic risk, which are addressed by clustering the assets at given locations (exposure aggregation). Such aggregation implicitly involves modifications of the hazard calculations, as the site conditions are taken from the aggregation location. Previous studies that focused on the influence of exposure aggregation on risk loss estimates at different geographical levels indicate that the choice of aggregation can introduce bias into the results. This work presents complementary analyses to enhance the understanding of the influence of exposure aggregation on risk metrics at the urban level. Three municipalities of different sizes (6–133 km²) with similar seismic hazard at rock conditions were considered in the analysis. Block-level exposure models that were developed as part of the Seismic Risk Model for Colombia (MNRS by its initialism in Spanish) were used as reference. These models were combined with the seismic hazard and the fragility/vulnerability functions used in the MNRS. In addition to the economic losses that are usually considered in seismic risk assessments, two parameters related to emergency response were included in the analyses: the number of buildings with complete damage and the number of fatalities. The risk metrics considered are average annual values (absolute values and their corresponding spatial distributions) and exceedance curves. The findings suggest that more evenly distributed aggregation units are preferable to geographical aggregation and that site conditions have a greater influence on exposure aggregation. The results also indicate that the average annual values alone are insufficient as a comparative measure, as they do not capture the spatial distribution of errors or the dispersion for different return periods.