A Hybrid Damage Detection Approach Based on Multi-Temporal Coherence and Amplitude Analysis for Disaster Response

The coherence-based and amplitude-based methods have been used to detect damaged area caused by natural disasters. However, the coherence-based method becomes less reliable for pixels having low coherence and amplitude-based method is in general less sensitive to the damage than coherence-based method. This paper presents an approach to damage proxy maps by jointly utilizing the interferometric coherence and amplitude information acquired from multi-temporal synthetic aperture radar images to complement their weakness. This technique is evaluated for the landslides triggered by the M6.7 Hokkaido earthquake on September, 6, 2018 in Japan. The experimental results demonstrate the benefit of the hybrid technique showing better performance than individual approaches.