11:30 AM - 11:45 AM
[STT35-04] Estimation of crustal deformation during the 2024 Noto Peninsula Earthquake detected by ALOS-2/PALSAR-2
Keywords:InSAR, 2024 Noto Peninsula Earthquake, Phase Unwrapping, 3-D analysis
On January 1, 2024, a magnitude 7.6 earthquake occurred in the Noto region of Ishikawa Prefecture, and as part of the analysis, interferometric SAR analysis using ALOS-2/PALSAR-2 images were conducted. Interferogram images were generated for each day (January 1, 2, and 3) immediately following the earthquake, and unwrapped images were created for three-dimensional analysis. The specific analysis methods are outlined below.
1) Improvement of Phase Difference Images
The three interferogram images used had low interferometric, especially the image on January 1, which exhibited very low interferometric. While results were obtained using speckle tracking and phase difference images, no report on unwrapped images was obtained. To increase unwrappable areas, adjustments were made to the number of looks and iterative processing with the Goldstein & Werner filter (GW filter) . As a result, for example, processing the image on January 1 with 16×16 looks, GW filter coefficient of 0.2, and 2 iterations increased the interferometric of the image and confirmed an increase in unwrappable areas. (The image on January 2 had 12×12 looks, GW filter coefficient of 0.2, and 2 iterations, while the image on January 3 had 8×8 looks, GW filter coefficient of 0.2, and 2 iterations.)
2)Creation of Mask Images to Address Unwrap Errors
Even with adjustments to the number of looks and the GW filter to increase unwrappable areas, it was challenging to remove unwrap errors caused by phase discontinuities in places with extremely large variations. For such locations, it is common to address unwrap errors visually or create mask images from coherence images or phase standard deviation images. However, coherence images and phase standard deviation images obtained from non-interfering images did not show sufficient correlation with images that increased unwrappable areas, making them insufficient for use as mask images in this analysis. Therefore, this analysis proposes a method for creating mask images for images with increased unwrappable areas. The creation method involves determining how far apart one pixel horizontally and one pixel below a certain pixel are in the phase difference image adjusted with the number of looks and the GW filter, using this information with a median filter to create the mask image. Additionally, in this analysis, the unwrap process is performed after applying mask processing to the phase difference image, reducing unwrap errors.
3)Three-Dimensional Analysis Efforts
In this analysis, images from three directions (Ascending-Left, Descending-Left, and Ascending-Right) were used for analysis. The three generated unwrapped images were decomposed into east-west, up-down, and north-south components, and three-dimensional analysis was conducted. The results of the three-dimensional analysis were compared with the electronic reference points of the Geospatial Information Authority of Japan for data verification. Furthermore, as images from Descending-Ring are also obtainable, efforts to decompose into east-west, up-down, and north-south components from four directions are also underway.
1) Improvement of Phase Difference Images
The three interferogram images used had low interferometric, especially the image on January 1, which exhibited very low interferometric. While results were obtained using speckle tracking and phase difference images, no report on unwrapped images was obtained. To increase unwrappable areas, adjustments were made to the number of looks and iterative processing with the Goldstein & Werner filter (GW filter) . As a result, for example, processing the image on January 1 with 16×16 looks, GW filter coefficient of 0.2, and 2 iterations increased the interferometric of the image and confirmed an increase in unwrappable areas. (The image on January 2 had 12×12 looks, GW filter coefficient of 0.2, and 2 iterations, while the image on January 3 had 8×8 looks, GW filter coefficient of 0.2, and 2 iterations.)
2)Creation of Mask Images to Address Unwrap Errors
Even with adjustments to the number of looks and the GW filter to increase unwrappable areas, it was challenging to remove unwrap errors caused by phase discontinuities in places with extremely large variations. For such locations, it is common to address unwrap errors visually or create mask images from coherence images or phase standard deviation images. However, coherence images and phase standard deviation images obtained from non-interfering images did not show sufficient correlation with images that increased unwrappable areas, making them insufficient for use as mask images in this analysis. Therefore, this analysis proposes a method for creating mask images for images with increased unwrappable areas. The creation method involves determining how far apart one pixel horizontally and one pixel below a certain pixel are in the phase difference image adjusted with the number of looks and the GW filter, using this information with a median filter to create the mask image. Additionally, in this analysis, the unwrap process is performed after applying mask processing to the phase difference image, reducing unwrap errors.
3)Three-Dimensional Analysis Efforts
In this analysis, images from three directions (Ascending-Left, Descending-Left, and Ascending-Right) were used for analysis. The three generated unwrapped images were decomposed into east-west, up-down, and north-south components, and three-dimensional analysis was conducted. The results of the three-dimensional analysis were compared with the electronic reference points of the Geospatial Information Authority of Japan for data verification. Furthermore, as images from Descending-Ring are also obtainable, efforts to decompose into east-west, up-down, and north-south components from four directions are also underway.