Automated Damage Indication for Rapid Geospatial Reporting

A methodology for automated extraction of damage indication from very high spatial resolution satellite imagery is presented for the Haitian towns of Carrefour and Leogane following the January 2010 earthquake. Damaged buildings are identified by changes to their shadows between pre- and post-event data. The approach makes use of object-based image analysis concepts to extract relevant information on damage distribution. The methodology selected requires pre- and post-disaster images with similar sun angles and will only be suitable for the detection of collapsed (or partly collapsed) structures. The new methodology proved to be effective for the Carrefour area despite using satellite images of different qualities with inadequate image (co-)registration, producing positively validated results within an acceptable processing time, and was, to the best of our knowledge, the only automated damage assessment method to have delivered appropriate results to requesting relief organizations within a few days of the Haiti earthquake.