Empirical seismic vulnerability assessment of Icelandic buildings affected by the 2000 sequence of earthquakes

In June 2000, two Mw6.5 earthquakes occurred within a 4-day interval in the largest agricultural region of Iceland causing substantial damage and no loss of life. The distance between the earthquake epicentres and the fault rupture was approximately 15 km. Nearly 5000 low-rise residential buildings were affected, some of which were located between the faults and exposed to strong ground motion from both events. The post-earthquakes damage and repair costs for every residential building in the epicentral region were assessed for insurance purposes. The database is detailed and complete for the whole region and represents one of the best quality post-earthquake vulnerability datasets used for seismic loss estimation. Nonetheless, the construction of vulnerability curves from this database is hampered by the fact that the loss values represent the cumulative damage from two sequential earthquakes in some areas, and single earthquakes in others. A novel methodology based on beta regression is proposed here in order to define the geographical limits on areas where buildings sustained cumulative damage and predict the seismic losses for future sequence of events in each area. The results show that the average building loss in areas affected by a single event is below 10% of the building replacement value, whilst this increases to an average of 25% in areas affected by the two earthquakes. The proposed methodology can be used to empirically assess the vulnerability in other areas which experienced sequence of events such as Emilia-Romagna (Italy) in 2012.