Abstract
Earthquake-induced rubble in urbanized areas must be mapped and characterized. Location, volume, weight and constituents are key information in order to support emergency activities and optimize rubble management. A procedure to work out the geometric characteristics of the rubble heaps has already been reported in a previous work, whereas here an original methodology for retrieving the rubble’s constituents by means of active and passive remote sensing techniques, based on airborne (LiDAR and RGB aero-photogrammetric) and satellite (WorldView-3) Very High Resolution (VHR) sensors, is presented. Due to the high spectral heterogeneity of seismic rubble, Spectral Mixture Analysis, through the Sequential Maximum Angle Convex Cone algorithm, was adopted to derive the linear mixed model distribution of remotely sensed spectral responses of pure materials (endmembers). These endmembers were then mapped on the hyperspectral signatures of various materials acquired on site, testing different machine learning classifiers in order to assess their relative abundances. The best results were provided by the C-Support Vector Machine, which allowed us to work out the characterization of the main rubble constituents with an accuracy up to 88.8% for less mixed pixels and the Random Forest, which was the only one able to detect the likely presence of asbestos.
Highlights
The enormous amount of rubble from partial or total collapse of buildings/structures caused by an earthquake hitting urbanized areas with vulnerable historical centres, like those of many Italian towns, must be rapidly mapped and characterized after catastrophic events
In addition to airborne surveys based on altimetric LiDAR (Laser Imaging Detection and Ranging) and photogrammetric data, High Resolution/Very High Resolution (HR/VHR) satellite imagery has been widely used for assessing damages and rubble detailed distributions linked to catastrophic events including earthquakes [7,8,9,10,11,12], since it makes it possible to exploit the intrinsic capacity for repetitive, multispectral and possibly stereoscopic acquisition, capable of supporting the monitoring of evolution of post-emergency scenarios
We proposed an original application based on EO (Earth Observation) in order to characterize the urban rubble heaps’ volume originating from buildings affected by partial or total collapse
Summary
The enormous amount of rubble from partial or total collapse of buildings/structures caused by an earthquake hitting urbanized areas with vulnerable historical centres, like those of many Italian towns, must be rapidly mapped and characterized after catastrophic events. The dramatically augmented availability of the VHR multi/hyperspectral and Synthetic Aperture Radar (SAR) data provided by increasing number of operative satellite remote sensing missions must be suitably coupled with advanced approaches based on data mining, machine learning and clustering scheme for properly monitoring and characterizing increasingly heterogeneous, complex and wide urban areas. In this context, the most recent machine learning algorithms based on Artificial Neural Network (ANN), Support Vector Machine (SVM), K-Nearest Neighbour (KNN) and Random Forest (RF) play a relevant role [41]. Once the in situ acquired spectral signatures had been resampled to the same sensor configuration, these algorithms were tested within a supervised learning scheme, for mapping (classifying) the endmembers provided by SMA into the rubble material of interest detected on site
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
