Uncertainty Quantification for Earthquake and Tsunami Risk Analysis with Application to Valparaíso, Chile

In a probabilistic analysis, epistemic uncertainties associated with modeling choices and limited data are accounted for, and their effect on the resulting risk metrics is quantified. In this paper, we address the challenge of identifying, classifying, quantifying, and comparing different sources of uncertainty and their influence on the losses and risk metrics of spatially distributed systems under sequential hazards, such as an earthquake event followed by a tsunami or a tropical cyclone followed by a storm surge. Through the example of earthquake and tsunami risk of the residential-building stock of the communes of Valparaíso and Viña del Mar in Chile, we investigate and discuss the relation between aleatory and epistemic uncertainties and their effects on different risk metrics. We identify epistemic uncertainty factors and perform a sensitivity analysis to assess their influence on the annual average loss, value at risk, and expected shortfall. Additionally, we investigate the influence of the considered time period on the uncertainty of the risk metrics and demonstrate that epistemic uncertainty dominates the total variance of the risk for long time periods. In the specific application, we observe that the risk metrics are most sensitive to the uncertainty of the earthquake fragility parameters and the ground motion model.