Investigation of Long-lasting Damage Due to Hagibis Flood Using SAR Polarimetry and ALOS-2 Data in Japan
Abstract
The flooding events and the associated risk in urban areas are an increasingly crucial global issue. The risk and loss due to flood in an urban area are huge compared to a rural area. As a result, the rebuilding capacity of an area that faces frequent floods must be investigated. The task of damage mapping during post-flood is complex because of waterlogging and sediment deposition. Flood debris act as different types of scatterers along with partial waterlogging after a flood. Due to the side-looking nature of SAR, scatterers' height and aspect angle play a significant role when polarised radar signal is used in urban areas. The area surrounding the Watarase pond of Saitama prefecture, Japan, was flooded heavily due to typhoon Hagibis. Two ALOS-2 fully polarized data for non-flood and post-flood dates with large temporal baseline are acquired to account for long-lasting hazards. This study investigated the characteristics and mechanisms of the radar return changes induced by urban flood using polarimetric model-based decomposition. As per methodology, the decrease in scattering powers is identified for the post-flood situation. Following the preliminary results, a change in radar return from double bounce to volume is observed widely due to debris in place of buildings. On the contrary, some areas show a change in double-bounce scattering to odd-bounce scattering post-flood. It can be due to the presence of water in place of buildings that were washed away in flood. The change in the occurrence of double bounce in place of other scatterings has been mainly observed. The scattering class transition is calculated using the categorical image analyst tool. A change in pixel values is calculated based on the spectral Euclidean distance method. This study proposes an unsupervised damage index to identify long-lasting hazardous damage that does not hold enough capacity to rebuild after a disastrous flood.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.H41B..04A