11:00 〜 13:00
[SCG44-P23] 高サンプリングGPSを用いたCascadiaスロースリップの時空間発展の推定
キーワード：スロースリップ、SSE、GPS、キネマティックGPS、スロー地震
The short-term slow slip event (SSE), a class of slow earthquakes which has duration of a few to tens of days, is typically detected and modeled from daily static Global Positioning System (GPS) data. However, the daily GPS data cannot image the sub-daily SSE processes. Due to the instrumental characteristics, the subdaily time scale (i.e., hundreds of seconds to a day) has been a secret zone of the slow earthquake spectrum. Lack of the access to this time scale has left underlying mechanisms of SSEs and slow earthquakes still elusive. For example, recent studies indirectly infer the presence of subdaily slip rate changes and isolated small events within a day, but it has not been confirmed yet. Whether such events exist would provide an important constraint to the controversial scaling relationship between the duration and moment.
By processing the raw GPS observables in the kinematic analysis approach, we can obtain surface deformation field at the raw observation interval (i.e., 30 seconds or shorter), which has great potential to overcome the time resolution issue present in the daily static GPS data. Although the kinematic GPS coordinates are known much noisier (~ cm) than the daily static coordinates (~ a few mm), recent applications to postseismic deformation studies achieved identifying sub-cm deformation. Motivated by them, we for the first time applied the kinematic GPS coordinates to model the short-term SSE to investigate applicability and limitation of the subdaily coordinates for SSE studies.
We chose one Cascadia SSE in March – April 2017, which has been already reported from daily GNSS data. We performed the kinematic GPS analysis at a 30-second interval for observations during the event occurrence using the TRACK program of GAMIT/GLOBK. Although the obtained raw coordinate series were quite noisy, we were able to discern the transient motion of a few mm during the event after carefully removing non-tectonic position fluctuation such as multipath effects, common mode errors and outliers.
Then, we inverted the cleaned data at a 30-minute interval using a Kalman-filter based method to infer spatiotemporal evolution of slip. The obtained spatiotemporal slip distribution exhibits a multi-stage evolution consisting of an isotropic growth of SSE and subsequent along-strike migration and termination. The transition of the slip growth mode occurs when the slip area fills the rheologically permitted down-dip width for the SSE occurrence. As conceptualized by Gomberg et al. (2016, GRL), this is analogous to the rupture growth of regular great earthquakes, so it implies the presence of common mechanical factors behind the two distinct slip phenomena. More extensive investigation is necessary to confirm this idea. The inferred moment rate has two peaks, which are consistent with the daily tremor counts in this region.
We carried out another slip inversion using the daily static GPS data recorded during the same period and the same inversion method to investigate the performance and limitation of our kinematic GPS data. A moment rate inferred from the daily data has also two peaks, so our 30-minute inversion result has the comparable time resolution to that derived from the widely-used daily data. This is an astonishing result given the long-believed low signal-to-noise ratio of the kinematic GPS. During the subdaily data inversion, maximizing the log-likelihood to determine the spatiotemporal slip smoothness resulted in extremely oscillated spatiotemporal slip evolution, which compelled us to manually impose the stronger smoothness constraint. This result strongly highlights the importance of better understanding of the non-tectonic noise in the kinematic GPS analysis, which will further improve the temporal resolution of SSE.
By processing the raw GPS observables in the kinematic analysis approach, we can obtain surface deformation field at the raw observation interval (i.e., 30 seconds or shorter), which has great potential to overcome the time resolution issue present in the daily static GPS data. Although the kinematic GPS coordinates are known much noisier (~ cm) than the daily static coordinates (~ a few mm), recent applications to postseismic deformation studies achieved identifying sub-cm deformation. Motivated by them, we for the first time applied the kinematic GPS coordinates to model the short-term SSE to investigate applicability and limitation of the subdaily coordinates for SSE studies.
We chose one Cascadia SSE in March – April 2017, which has been already reported from daily GNSS data. We performed the kinematic GPS analysis at a 30-second interval for observations during the event occurrence using the TRACK program of GAMIT/GLOBK. Although the obtained raw coordinate series were quite noisy, we were able to discern the transient motion of a few mm during the event after carefully removing non-tectonic position fluctuation such as multipath effects, common mode errors and outliers.
Then, we inverted the cleaned data at a 30-minute interval using a Kalman-filter based method to infer spatiotemporal evolution of slip. The obtained spatiotemporal slip distribution exhibits a multi-stage evolution consisting of an isotropic growth of SSE and subsequent along-strike migration and termination. The transition of the slip growth mode occurs when the slip area fills the rheologically permitted down-dip width for the SSE occurrence. As conceptualized by Gomberg et al. (2016, GRL), this is analogous to the rupture growth of regular great earthquakes, so it implies the presence of common mechanical factors behind the two distinct slip phenomena. More extensive investigation is necessary to confirm this idea. The inferred moment rate has two peaks, which are consistent with the daily tremor counts in this region.
We carried out another slip inversion using the daily static GPS data recorded during the same period and the same inversion method to investigate the performance and limitation of our kinematic GPS data. A moment rate inferred from the daily data has also two peaks, so our 30-minute inversion result has the comparable time resolution to that derived from the widely-used daily data. This is an astonishing result given the long-believed low signal-to-noise ratio of the kinematic GPS. During the subdaily data inversion, maximizing the log-likelihood to determine the spatiotemporal slip smoothness resulted in extremely oscillated spatiotemporal slip evolution, which compelled us to manually impose the stronger smoothness constraint. This result strongly highlights the importance of better understanding of the non-tectonic noise in the kinematic GPS analysis, which will further improve the temporal resolution of SSE.