Detecting Tsunami-Induced Damage on Building Side Walls Using X-Band of Post-Event Airborne Full Polarimetric Sar Data

Abstract eng:
A new method for detecting tsunami-induced damage on a building side wall is proposed using X-band of post-event full polarimetric airborne synthetic aperture radar (Pi-SAR2) data. Buildings with destroyed walls are distinguished based on the ratio of the double-bounce scattering power to the total scattering power in the building layover areas. A building side wall shows bright backscattering induced by layover and double-bounce scattering. If the lower part of a building side wall is destroyed by a tsunami, double-bounce scattering, which is dominant in the lower part of the wall, may be absent and its scattering power may decrease. Considering the radar properties in terms of building side walls, a method for distinguishing destroyed building side walls was developed using post-event full polarimetric SAR (PolSAR) data. A four-component decomposition model and the Yamaguchi four-component decomposition scheme were applied to compute the doublebounce scattering power using polarimetric SAR data. The ratio of the double-bounce scattering power to the total scattering power in the layover area was estimated and correlated with the surveyed data. A classifier for damage detection was created using this data set. To create the classifier, a decision tree application of a machine learning algorithm was applied, which discriminates between “destroyed” and “undestroyed” building side walls. The developed classifier was validated using 12 buildings with 48 side walls and demonstrated satisfactory performance with 77.1 % accuracy and a kappa statistic of 0.48.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket