EMPIRICAL FRAGILITY CURVES BASED ON RANDOM GROUND SHAKING FIELDS: EMPLOYING COPERNICUS-EMS DAMAGE GRADING MAPS FOR 2016 AMATRICE SEQUENCE

Significant structural and non-structural damage suffered by residential buildings due to re-cent earthquakes world-wide provide a sad testimony to the vulnerability of the existing build-ing stock. This evidence stresses the need to develop accurate seismic vulnerability assessment tools in the service of strategic planning and applicable at a territorial scale. The empirical fragility curves have been used quite often in order to characterize the vulnerability of prescribed buildings classes as opposed to individual buildings. In this approach, the fra-gility curve represents both the building-to-building variability within the class and the uncer-tainty in the building class response to future earthquakes. Development of empirical fragility curves for classes of buildings depends to a large extent on the pairs of seismic ground shak-ing intensity and observed damage data gathered from spatially distributed buildings. The seismic intensity experienced at the site of each building is usually estimated by employing maps of ground shaking for the earthquake of interest (a.k.a., shakeMaps). The accuracy in the estimation of the ground shaking intensity is arguably as important as that of the evalua-tion of the incurred damage. The former involves explicit characterization of the uncertainties in the ground-shaking intensity estimation. This can be achieved through application of ground motion prediction equations and the corresponding spatial correlation structure in the residuals (of the predictive equation), resulting in generation of random fields of ground shaking. Each field represents a plausible distribution of ground motion intensity for a given earthquake scenario. Then, the random ground shaking field can be updated based on availa-ble observations of ground shaking intensity from nearby stations. The generated random fields can be further modified to consider the local stratigraphic and topographic amplifica-tion. For a given damage pattern observed for a prescribed building class, the generation of random ground motion fields leads to a set of plausible empirical fragility curves (corre-sponding to exceedance of different damage states). The fragility curves and their 16th, 50thand 84th percentiles, obtained based on the generated random ground shaking fields and ob-served damage in the aftermath of Amatrice 2016 Earthquake (based on Copernicus EMS damage maps) show reasonable agreement with the empirical fragility curves available in the literature for similar buildings classes