5:15 PM - 6:45 PM
[STT35-P04] An attempt to detect landslides around Mt. Tsukuba using Stacking-InSAR method with ALOS-2
Keywords:InSAR, stacking, landslide, Mt.tsukuba
In Japan, many sediment disasters occur every year everywhere, and there were 1,468 sediment disasters occurred in 2023 (Ministry of Land, Infrastructure, Transport and Tourism). InSAR (Interferometric Synthetic Aperture Radar) is a well-known technology for landslide observation, which can measure the distribution of ground surface displacement through repeat-pass observation by microwave imaging technology. InSAR can detect high-resolution ground surface displacement over a wide area regardless of daylight and is not much affected by rainfall. InSAR also do not require any observation instruments on the ground. Based on previous studies, it has been shown that InSAR can accurately and frequently observe landslide in estimating active blocks (Une et al. 2008; Sato et al. 2012; Ishitsuka et al. 2017).
However, there are no research cases where InSAR observations of landslides have been conducted around the Mt. Tsukuba area. In this study, we tried to detect landslides around Mt. Tsukuba, including Mt. Amabiki and Mt. Tomiya in Ibaraki Prefecture, by performing Stacking-InSAR method using ALOS-2 data.
In this study, InSAR images were created based on a total of 25 SLC(Single Look Complex) images derived by ALOS-2 SM1-mode acquired from February 15, 2015 to May 15, 2022. The multi-look process for noise reduction was set to achieve an InSAR image resolution of 10 m × 10 m. Although the higher spatial resolution results in a noisier image, the analysis was performed at high spatial resolution to be enable of detecting small-spatial scale surface displacements. Stacking-InSAR was used because it is particularly effective in reducing random noise in temporal. After detection of ground surface displacement signal from Stacking-InSAR image, we evaluated whether the signal was a landslide or not from individual InSAR images at location where the surface displacement signal was detected. Those individual InSAR images were taken before application of the GW-filter and phase unwrapping process to check whether the signal exceeded the noise level. Then, the annual displacement rate obtained from Stacking-InSAR was compared with the LOS displacement estimated from the slope azimuth and inclination angle measured from DEM (Digital Elevation Model), the microwave incidence angle, and the satellite's azimuth direction.
As a result of Stacking-InSAR, an approximately 40 m × 50 m ground surface displacement signal was detected on a slope in Oizumi(Sakuragawa City, in Ibaraki Prefecture). The annual displacement rate estimated from Stacking-InSAR was 4 cm ~ 5 cm. However, some of the InSAR images used for Stacking-InSAR were found to have phase unwrapping errors. Therefore, the estimated annual displacement rate is not quantitatively reliable. A visual check of individual InSAR images (before application of the GW-filter and phase unwrapping process) showed that this surface displacement signal was continuous from February 15, 2015 to May 15, 2022 and coherence was stable for the slope where the signal was observed. The signal had a possibility of landslide based on the continuous signal and coherence stability. However, according to the geometric information from DEM, the slope azimuthal angle was approximately 250° (counterclockwise from east) and the inclination angle was 20°~25°. Under the assumption that the ground surface was displaced along the slope, the approximately 4 cm ~ 5 cm annual displacement rate in the LOS direction observed in a Stacking-InSAR could be explained by upslope 121 cm ~ 151 cm displacement. It is not realistic. Because the south-southwest facing slope was observed by ascending-right-looking satellite, it is possible that a large vertical component of displacement occurred.
However, there are no research cases where InSAR observations of landslides have been conducted around the Mt. Tsukuba area. In this study, we tried to detect landslides around Mt. Tsukuba, including Mt. Amabiki and Mt. Tomiya in Ibaraki Prefecture, by performing Stacking-InSAR method using ALOS-2 data.
In this study, InSAR images were created based on a total of 25 SLC(Single Look Complex) images derived by ALOS-2 SM1-mode acquired from February 15, 2015 to May 15, 2022. The multi-look process for noise reduction was set to achieve an InSAR image resolution of 10 m × 10 m. Although the higher spatial resolution results in a noisier image, the analysis was performed at high spatial resolution to be enable of detecting small-spatial scale surface displacements. Stacking-InSAR was used because it is particularly effective in reducing random noise in temporal. After detection of ground surface displacement signal from Stacking-InSAR image, we evaluated whether the signal was a landslide or not from individual InSAR images at location where the surface displacement signal was detected. Those individual InSAR images were taken before application of the GW-filter and phase unwrapping process to check whether the signal exceeded the noise level. Then, the annual displacement rate obtained from Stacking-InSAR was compared with the LOS displacement estimated from the slope azimuth and inclination angle measured from DEM (Digital Elevation Model), the microwave incidence angle, and the satellite's azimuth direction.
As a result of Stacking-InSAR, an approximately 40 m × 50 m ground surface displacement signal was detected on a slope in Oizumi(Sakuragawa City, in Ibaraki Prefecture). The annual displacement rate estimated from Stacking-InSAR was 4 cm ~ 5 cm. However, some of the InSAR images used for Stacking-InSAR were found to have phase unwrapping errors. Therefore, the estimated annual displacement rate is not quantitatively reliable. A visual check of individual InSAR images (before application of the GW-filter and phase unwrapping process) showed that this surface displacement signal was continuous from February 15, 2015 to May 15, 2022 and coherence was stable for the slope where the signal was observed. The signal had a possibility of landslide based on the continuous signal and coherence stability. However, according to the geometric information from DEM, the slope azimuthal angle was approximately 250° (counterclockwise from east) and the inclination angle was 20°~25°. Under the assumption that the ground surface was displaced along the slope, the approximately 4 cm ~ 5 cm annual displacement rate in the LOS direction observed in a Stacking-InSAR could be explained by upslope 121 cm ~ 151 cm displacement. It is not realistic. Because the south-southwest facing slope was observed by ascending-right-looking satellite, it is possible that a large vertical component of displacement occurred.