GNGTS 2017 - 36° Convegno Nazionale

GNGTS 2017 S essione 2.1 249 is also more time- and resource- intensive. Furthermore, using either of the scales entails exposing the surveyors to a perilous environment for several weeks, largely overlapping with the ongoing emergency management activities. In this contribution we describe a novel methodology for the rapid, probabilistic estimation of the Macroseismic Intensity in the epicentral area, according to the EMS-98 scale. The methodology includes the use of mobile mapping and a collaborative on-line platform for rapid post-earthquake reconnaissance, which allows to rapidly and safely collect building-by-building damage data in the hours and days immediately following a destructive event. A probabilistic estimation of the MI at the locality aggregation level is then implemented using a Bayesian updating scheme able to integrate direct damage observations and prior information, hence allowing the consideration of ancillary data (such as for instance the damage grading provided by the Copernicus Emergency Mapping Service) and expert judgment. A feasibility study has been carried out in the area affected by the 2016 Amatrice earthquake and validated with the data provided by the Istituto Nazionale di Geofisica e Vulcanologia (INGV) (Azzaro et al ., 2016). The results show that the proposed technique may provide a fast and reliable preliminary assessment of EMS-98 intensity, efficiently integrating (and further promoting) post-earthquake building damage surveys. Methodology. A customized mobile mapping system, originally designed for seismic exposure and vulnerability characterization (see, e.g., Pittore and Wieland 2012, Pittore et al. , 2015) has been employed to rapidly collect geo-referenced spherical images of the built environment. The visual data has been collected following a sampling scheme based on the information made available shortly after the August 24 earthquake by the Copernicus Emergency Service (Copernicus Emergency Mapping Service, 2016). The route was covered in around two days between September 26 and 28 2016, and 9900 spherical images have been selected for the survey. A dedicated web-based on line platform named RRDA has been also developed (Fig. 1). Throughout this interface, a pool of selected surveyors analyzed the captured omni-directional images and provided a precise assessment of the observable damage, as well as the most probable EMS-98 vulnerability class, for a set of specific buildings. The interface also allows to query for the closest omni-directional image available from the Google StreetView™ service for a direct comparison between the appearance of the building before and after the earthquake. A total of 313 buildings has been assigned damage and vulnerability. These observations, spread over the epicentral area, refer to around 20 different localities where MI assessment has been carried out (see Fig. 2). The estimation of the MI is carried out in a probabilistic framework with an incremental Bayesian updating mechanism. Therefore in each stage of the process, the MI is represented Fig. 1 - Overview of the considered region with the Voronoi tessellation based on the coordinates of the corresponding settlements. The star marks the location of the epicentre.