Building Damages Detection of the 2023 Ms 6.2 Jishishan (Gansu, China) Earthquake Using Multi-temporal Dual Polarization ALOS-2 /PALSAR-2 Data

Objectives: On the early morning of December 18, 2023, an earthquake of magnitude Ms6.2 occurred in Jishi Mountain, Gansu Province, damaging 15,000 houses and severely affecting people's lives and property safety. Efficient and accurate detection can help to understand the distribution of building damage and provide crucial support for post-disaster emergency relief. Methods: This paper combines intensity change detection, coherence detection, and InSAR technology using intensity and phase information from multi-temporal dual-polarized ALOS- 2/PALSAR-2 images to identify building damage regions and co-seismic deformation fields. First, a likelihood ratio change detection method is used to detect change regions from multi-temporal dual-pol intensity images. Moreover, optimal coherence interferometric phase is extracted using different polarization mode combinations to obtain a more accurate co-seismic deformation field. In addition, coherence change detection is introduce to detect different degrees of surface change distribution using pre- and post-earthquake interferometric coherence based on dual-polarization information. Finally, we combine intensity and coherence change detection results with WSF 2019 to map building damage regions. Results: The Beijing III optical images with a 0.3m resolution is used to verify the effective of proposed method in building damage mapping. And the experimental results show that (1) intensity change detection results based on dual-polarization information are significantly better than single polarization, and the coherence estimation results based on dualpolarization information are also superior to single polarization interferometric pairs. Besides, coseismic deformation field using dual-polarization information avoids more phase unwrapping errors, and the deformation magnitude is similar to the Sentinel-1 results. (2) in the seven earthquakeaffected towns, using the intensity change detection with dual-polarization information can detect the building damage area about 0.4489 km² and using coherence change detection can detect the building damage area about 1.0047 km². (3) both temporal intensity and coherence change detection methods can extract building damage. In most case, damage regions with coherence-based change detection generally larger than intensity-based. Conclusions: (1) Dual-polarization information can significantly improve the accuracy of change detection and the co-seismic deformation field; (2) Both intensity-based and coherence-based change detection using dual-polarization information can detect damaged buildings in the disaster area; (3) Compared to intensity-based change detection, coherence-based detection is more sensitive and detects a larger damage regions; (4) Intensity-based and coherence-based techniques can complement and corroborate each other, and effectively extract damaged buildings in the earthquake area.