Exploring the impact of epistemic uncertainty on a regional probabilistic seismic risk assessment model

Abstract

Probabilistic earthquake loss models are widely used in the (re)insurance industry to assess the seismic risk of portfolios of assets and to inform pricing mechanisms for (re)insurance contracts, as well as by international and national organizations with the remit to assess and reduce disaster risk. Such models include components characterizing the seismicity of the region, the ground motion intensity, the building inventory, and the vulnerability of the assets exposed to ground shaking. Each component is characterized by a large uncertainty, which can be classified as aleatory or epistemic. Modern seismic risk assessment models often neglect some sources of uncertainty, which can lead to biased loss estimates or to an underestimation of the existing uncertainty. This study focuses on exploring and quantifying the impact of a number of sources of uncertainties from each component of an earthquake loss model to the loss estimates. To this end, the residential exposure of Guatemala and Guatemala City were used as case studies. Moreover, a comparison of the predicted losses for an insured portfolio in the country between OpenQuake-engine and a vendor catastrophe platform was performed, assessing the potential application of OpenQuake in the (re)insurance industry. The findings from this study suggest that the uncertainty in the hazard component has the most significant effect on the loss estimates.