Verification of seismic risk models using observed damages from past earthquake events

Abstract

The accuracy of a seismic risk model to predict future losses is a significant challenge as it depends on many parameters, each with important uncertainties, that are present in all components of the model. The aim of this study is to present a process that can be followed to validate the efficiency of a predictive risk model for seismic risk assessment at urban or regional scale, through comparison of estimated physical damages with those recorded at past seismic events, discussing at the same time the uncertainties involved in the different steps of this process. The proposed process is applied for the verification of the open-access uniform European Seismic Risk Model (ESRM20), considering two major past seismic events in Greece, i.e. the Athens 1999 Mw 5.9 and the Thessaloniki 1978 Mw 6.5 earthquakes, for which we possess relatively good damage reports for a representative portion of the two metropolitan agglomerations. Despite the numerous uncertainties and the unavoidable simplifications, a good comparison is found between the estimated and observed damage, especially for the city of Thessaloniki. Regarding the distribution of damage, it is anticipated that the most heavily damaged typologies for both case studies are found to be the low-to-medium-rise, non-ductile masonry buildings. The analyses also revealed the importance of the credibility of the adopted exposure model and of the adequate selection of ground motion models for seismic hazard assessment. Τhis challenging work and the positive results acquired so far aim to contribute to the efforts towards increasing the resilience of cities.