2:45 PM - 3:00 PM
[STT39-10] 3D velocities and 2D strain rates in and around the Atotsugawa Fault zone estimated by InSAR and GNSS
Keywords:InSAR, GNSS, SSM, Inter-seismic deformation, Strain rate, Fault
It is easy to detect inter-seismic crustal deformation with GNSS. But the number of GNSS stations is very limited in rugged mountainous areas, which leads to the low spatial resolution of the estimated velocity fields. This can be an essential problem for disaster prevention and/or mitigation. In contrast, InSAR analysis can be used to obtain the velocity fields at extremely high spatial resolution. However, in steep mountainous regions such as Japan, the sensitivity is low due to the use of long wavelength (L-band) microwaves, and there have been limited successes (e.g., Takada et al., 2018). Furthermore, L-band SAR is strongly affected by ionospheric disturbances, and correction for these disturbances is extremely important for obtaining long-wavelength inter-seismic crustal deformation. In this presentation, we first focus on the correction of ionospheric disturbances in InSAR images and then use the corrected results to estimate the velocity field by InSAR time series analysis. Finally, we estimate the 3D velocities and the strain rates by combining InSAR and GNSS data, which has been developed in recent years (e.g., Franklin and Huang, 2023; Weiss et al., 2020). The analysis targets the area around the Atotsugawa Fault, which meets the following conditions: (1) located in a steep mountainous region, (2) consists of the Niigata-Kobe Tectonic Zone (Sagiya et al., 2000), and (3) relatively dense GNSS observation network.
First, InSAR time series analysis was performed using SM1 data acquired by ALOS-2 from 2014 to 2022. The effects of ionospheric and tropospheric disturbances were corrected according to the following flow. First, the Split Spectrum Method (SSM) (Gomba et al., 2016) is used to correct the effects of ionospheric disturbances in individual interferograms; the major challenges in SSM are outlier removal and signal-to-noise ratio improvement. In this study, (1) outliers are removed by thresholding using an original combination of empirical thresholds in addition to the theoretical one proposed by Gomba et al. (2016), (2) spatial smoothing is performed using a Gaussian filter with an originally improved window size while considering weighting by coherence. Next, the tropospheric delay was corrected using a meteorological model with the open software GACOS (Yu et al., 2018). Using these corrected images, InSAR time series analysis (Schmidt and Bürgmann, 2003) was used to reduce the short-period disturbances and estimate the mean velocity field in the satellite line-of-sight (LOS) direction. The estimated results were consistent with the GNSS data.
Next, the 3D velocities and the 2D strain rates were estimated by integrating the high-resolution velocity field estimated by InSAR and that by GNSS. First, we subsampled the velocity field obtained from InSAR to create the grid data. The three velocity components at GNSS stations were interpolated to all the grids, and the 3D velocity field was estimated for each grid using the weighted least squares method. Then, the horizontal component of the estimated velocity field was used to calculate the strain rate. When the difference is calculated between adjacent grids, the effect of noise in each grid is extremely amplified. Therefore, first, we collected data for grids located within a certain distance from each grid. Second, we approximated the collected grids by a plane to calculate the displacement gradient and strain tensor. The estimated strain rates (Figure) show the strain concentration along the Atotsugawa fault system and its eastern and western ends. Large strain rates are also estimated for a part of the Takayama-Oppara fault zone. The current issues are (1) introduction of interpolation method of GNSS data considering variograms, (2) reduction of gaps in LOS displacements between the adjacent SAR images, and (3) reduction of the influence of meteorological disturbances.
First, InSAR time series analysis was performed using SM1 data acquired by ALOS-2 from 2014 to 2022. The effects of ionospheric and tropospheric disturbances were corrected according to the following flow. First, the Split Spectrum Method (SSM) (Gomba et al., 2016) is used to correct the effects of ionospheric disturbances in individual interferograms; the major challenges in SSM are outlier removal and signal-to-noise ratio improvement. In this study, (1) outliers are removed by thresholding using an original combination of empirical thresholds in addition to the theoretical one proposed by Gomba et al. (2016), (2) spatial smoothing is performed using a Gaussian filter with an originally improved window size while considering weighting by coherence. Next, the tropospheric delay was corrected using a meteorological model with the open software GACOS (Yu et al., 2018). Using these corrected images, InSAR time series analysis (Schmidt and Bürgmann, 2003) was used to reduce the short-period disturbances and estimate the mean velocity field in the satellite line-of-sight (LOS) direction. The estimated results were consistent with the GNSS data.
Next, the 3D velocities and the 2D strain rates were estimated by integrating the high-resolution velocity field estimated by InSAR and that by GNSS. First, we subsampled the velocity field obtained from InSAR to create the grid data. The three velocity components at GNSS stations were interpolated to all the grids, and the 3D velocity field was estimated for each grid using the weighted least squares method. Then, the horizontal component of the estimated velocity field was used to calculate the strain rate. When the difference is calculated between adjacent grids, the effect of noise in each grid is extremely amplified. Therefore, first, we collected data for grids located within a certain distance from each grid. Second, we approximated the collected grids by a plane to calculate the displacement gradient and strain tensor. The estimated strain rates (Figure) show the strain concentration along the Atotsugawa fault system and its eastern and western ends. Large strain rates are also estimated for a part of the Takayama-Oppara fault zone. The current issues are (1) introduction of interpolation method of GNSS data considering variograms, (2) reduction of gaps in LOS displacements between the adjacent SAR images, and (3) reduction of the influence of meteorological disturbances.