Remote sensing for seismic building vulnerability assessment

Abstract

We quantitatively evaluate the suitability of multi-sensor remote sensing to assess the seismic vulnerability of buildings. Therefore, features are derived from remote sensing data to characterize the urban environment. The derived features are combined with in situ observations. Machine learning approaches are deployed subsequently to identify meaningful sets of features that are suitable to predict seismic vulnerability levels of buildings. Supervised regression techniques are used to estimate the actual vulnerability level. Empirical evidence with respect to the viability of the approach is revealed for the city of Padang (Indonesia). When assessing the vulnerability level according to a scoring method, the overall mean absolute percentage error is 10.6%, if using a supervised Support Vector Regression approach. This study uncovers potential for a rapid screening assessment of large areas which should be further explored in the future.