Publication Date:
2021-12-06
Description:
This study analyzes the ability of polarimetric Synthetic Aperture
Radar (PolSAR) measurements to quantify post-earthquake
damages. To achieve this goal, a twofold task is addressed: on
one side a processing chain, which exploits multi-polarization SAR
features, and a decision-tree classifier is proposed to quantify the
levels of damage in earthquake-affected urbanized areas using
dual-polarimetric (DP) SAR imagery. On the other side, a new
damage index is developed that allows a fair spatial intercomparison
of building-by-building information, collected via
ground surveys on the damaged areas, and SAR-derived damage
maps. The proposed rationale is showcased using measurements
related to the Central-Italy Earthquake occurred in 2016 where both
Sentinel-1 DP imagery and ground-based information are available.
Experimental results demonstrate the soundness of the proposed
approach. The main outcomes can be summarized as follows: a) DP
features perform better than single-polarization ones; b) DP features
exhibit a larger sensitivity to lower damage grades if compared
to the single polarization (SP) feature; c) the accuracy of the
estimated damage levels depends on the requested granularity in
the damage maps; d) the accuracy obtained using DP features
spans from 52% up to 71% when five and two damage classes
are considered, respectively.
Description:
Published
Description:
5971–5986
Description:
3T. Fisica dei terremoti e Sorgente Sismica
Description:
JCR Journal
Keywords:
remote sensing
;
earthquake damage
;
SAR polarimetry
Repository Name:
Istituto Nazionale di Geofisica e Vulcanologia (INGV)
Type:
article