10:45 AM - 12:15 PM
[STT39-P08] Extraction of crustal deformation in the Marlborough region during the 2016 Kaikoura earthquake
Keywords:ALOS-2/PALSAR-2, 2016 Kaikoura earthquake, InSAR, 2.5-D analysis
In 2016 Kaikoura earthquake, The M7.8 earthquake that occurred near Kaikoura, New Zealand on November 13, 2016 (local time 00:02 on the 14th) had aftershocks exceeding M6 even after the main shock. And a large crustal movement was confirmed around the epicenter. In this study, I estimate the surface deformation of the 2016 Kaikoura earthquake using ALOS-2/PALSAR-2 data. Also, by merging Stripmap-mode images, the measurement range is expanded.
The research procedure consists of image combination, InSAR processing, unwrapping, 2.5-D analysis, and comparison with GNSS points.
First, we tried to merge Ascending 5 and 3 scenes and Descending 6 scenes in the Canterbury and Marlborough regions. Noise was not confirmed in Ascending when merging 5 scenes or 3 scenes, but noise was confirmed in Descending when merging 3 or more scenes.
Next, when the unwrapping process was performed on the images that had been correctly merged, it was confirmed that there were places where the unwrapping process had not been completed. When the unwrap process was examined, it was found that the unwrap process was not performed correctly when the regularity of the interference fringes was lost. The regularity of interference fringes tends to disappear when approaching the epicenter or a place thought to be a plate interface, and the extremely large fluctuations seen during earthquakes produce large phase changes in phase-contrast images. It was thought that fluctuations that could not be captured even at long wavelengths in the L band would occur.
Based on these results, we analyzed Ascending 2 scenes and Descending 2 scenes and limited it to the Marlborough region.
Finally, 2.5-D analysis was performed to compare the quasi-vertical and quasi-east-west components of the Marlborough region with the GNSS data (MAHA and WITH points) included in the image. As a result, the difference between the upper and lower fluctuations of the MAHA point was -4.6 cm, which is close, and the difference between the upper and lower fluctuations of the WITH point was -22.0 cm, which is a small difference. On the other hand, the difference in the amount of movement between east and west was about -70.0 cm, which is a large difference.
In this study, we were able to measure a wide range by merging images in Stripmap-mode (resolution 10m). In addition, it was possible to measure ground surface displacement with higher accuracy than the image of ScanSAR-mode (resolution 100m), and the data harmonized with the result of pixel offset of previous research was obtained. However, there was a difference from the results of GNSS, which is more accurate than InSAR, and the error was particularly large in the east-west variation. In addition, we confirmed the advantages and problems of merging images, and confirmed that if multiple images of Descending can be merged, we can extract crustal movements with high precision over a wider range.
The research procedure consists of image combination, InSAR processing, unwrapping, 2.5-D analysis, and comparison with GNSS points.
First, we tried to merge Ascending 5 and 3 scenes and Descending 6 scenes in the Canterbury and Marlborough regions. Noise was not confirmed in Ascending when merging 5 scenes or 3 scenes, but noise was confirmed in Descending when merging 3 or more scenes.
Next, when the unwrapping process was performed on the images that had been correctly merged, it was confirmed that there were places where the unwrapping process had not been completed. When the unwrap process was examined, it was found that the unwrap process was not performed correctly when the regularity of the interference fringes was lost. The regularity of interference fringes tends to disappear when approaching the epicenter or a place thought to be a plate interface, and the extremely large fluctuations seen during earthquakes produce large phase changes in phase-contrast images. It was thought that fluctuations that could not be captured even at long wavelengths in the L band would occur.
Based on these results, we analyzed Ascending 2 scenes and Descending 2 scenes and limited it to the Marlborough region.
Finally, 2.5-D analysis was performed to compare the quasi-vertical and quasi-east-west components of the Marlborough region with the GNSS data (MAHA and WITH points) included in the image. As a result, the difference between the upper and lower fluctuations of the MAHA point was -4.6 cm, which is close, and the difference between the upper and lower fluctuations of the WITH point was -22.0 cm, which is a small difference. On the other hand, the difference in the amount of movement between east and west was about -70.0 cm, which is a large difference.
In this study, we were able to measure a wide range by merging images in Stripmap-mode (resolution 10m). In addition, it was possible to measure ground surface displacement with higher accuracy than the image of ScanSAR-mode (resolution 100m), and the data harmonized with the result of pixel offset of previous research was obtained. However, there was a difference from the results of GNSS, which is more accurate than InSAR, and the error was particularly large in the east-west variation. In addition, we confirmed the advantages and problems of merging images, and confirmed that if multiple images of Descending can be merged, we can extract crustal movements with high precision over a wider range.