Enhancing the accuracy of near-real-time seismic loss estimation using post-earthquake aerial images

Regional seismic economic loss of buildings is an important decision variable. Field investigation of seismic loss often takes several weeks to complete. In comparison, the earthquake-induced loss can be estimated within hours... [ view full abstract ]

Regional seismic economic loss of buildings is an important decision variable. Field investigation of seismic loss often takes several weeks to complete. In comparison, the earthquake-induced loss can be estimated within hours using seismic damage simulation method based on nonlinear time-history analysis (THA). However, the accuracy of such method depends on the rationality of the ground motion input. With the development of remote sensing technology, plenty of aerial images of the disaster area can be available within a few hours after an earthquake. Collapsed buildings can be identified from those images. In contrast, the damage within the building is difficult to determine, resulting in underestimated seismic loss. In this work, a framework for near-real-time estimation of regional seismic loss of buildings is proposed, which improves the accuracy of THA-based loss estimations by taking advantages of the identified building collapse scene of the disaster area. In particular, a series of THA are performed, generating a large number of simulation results. Among the results, those most similar to the identified collapse scene are treated as the optimal simulation results, and then be used to estimate the seismic loss. Through the case of the 2014 Ludian earthquake in China, the proposed method is illustrated and validated. The results show that due to the use of the building collapse scene, the economic loss estimation is close to the actual loss even without a recorded ground motion input.