09:15 〜 09:30
[SEM15-07] Investigating the Role of Fluids in Triggering Slow Slip Events through Marine Magnetotelluric Monitoring on the Northern Hikurangi Subduction Zone
キーワード：Slow Slip Events, marine Magnetotelluric, monitoring, Hikurangi Subduction Zone
Over the past 25 years, the increased establishment of dense GPS monitoring stations has enabled the identification and analysis of slow slip events (SSEs), a newly defined fault slip behavior (Schwartz and Rokosky, 2007). SSEs often occur in the transition zone between the unstable locked/stick-slip and stable creep segments. This positioning suggests that SSEs may facilitate stress concentration on fault lock patches and indicate the nucleation process of large earthquakes at depth (Kato et al., 2012; Obara and Kato, 2016). Accurate modeling of SSEs is thus crucial for a better understanding of the earthquake cycle. However, the precise triggering mechanism for SSEs remains largely unclear, hindering the development of robust dynamic models for SSEs. Numerous laboratory and field experiments, along with geophysical explorations in SSE source regions, suggest that fluids may play a critical role in triggering SSEs at depth. Although the spatiotemporal migration pattern of SSEs can be effectively constrained through the inversion of displacement data collected on land or at the seafloor (Woods et al., 2024), no spatiotemporal fluid migration pattern has been established to thoroughly analyze the evolutionary relationship between these two phenomena.
To address the spatiotemporal fluid migration pattern in SSE source regions and their vicinities, three marine magnetotelluric (mMT) monitoring stations were installed above a significant SSE source region on the northern Hikurangi subduction zone, one of the most extensively studied SSEs regions globally due to its extremely shallow source. One year of monitoring data was recovered in October 2024. We preliminarily analyzed the time series and estimated the MT response curves with BIRRP (Chave and Thomson, 2004) and TRACMT (Usui et al., 2024), resulting in high-quality impedance tensors and tippers at three sites. These impedance tensors were compared with data collected ~5 years ago at the same locations by the Scripps Institution of Oceanography's OBEM (Chesley et al., 2021). This comparison (Figure 1) represents the first instance of analyzing data curves from two ocean-bottom instruments at nearly identical locations. At all three sites, the comparing plots indicate generally consistent curves in MT off-diagonal components, but considerable differences happened in the MT diagonal components. We suspect that these differences in diagonal components can be attributed to the different recording lengths of two measurements and the inability to position the two comparison sites at the exact same location on the seafloor. Additionally, local discrepancies in the YX component at Site C (Figure 1) can be identified at a period range of 300s to 700s. The distance between these two comparison sites is ~1.67 km. So far, the exact reason for this discrepancy is unknown and requires further investigation. Besides, by reviewing the time series, we identified interesting occasional variations with frequencies higher than 1 Hz, likely associated with vessels passing by on the sea surface. Finally, we will discuss our time series processing plan to achieve the temporal variation patterns of subsurface fluid migration. To reach this goal, achieving robust variations in the MT responses is essential. As the uncertainties of MT transfer function estimations increase with the decreasing length of calculation samples, we plan to involve the latest proposed multitaper spectral analysis (Chave, 2019; Comeau et al., 2024) and frequency-domain independent component analysis (Cui et al., 2013; Sato et al., 2021; Zhou et al., 2022) to suppress the variances in each frequency response and obtain feasible variation sequences of transfer functions with reasonable time resolution. Additionally, our seafloor magnetic field is recorded with a sample rate of 600s during the later recording period to save power for maintaining a high sample rate (8Hz) in the electric field channel. Therefore, magnetic data on land have to be used to estimate the MT transfer function at a period range of 10s-1000s. We will show preliminary processing results by using the land magnetic data and compare them with the curves derived from the local magnetic fields.
Figure 1. Comparison of curves derived from two mMT measurements conducted in 2024 with ERI’s OBEM (Baba et al., 2020) and in 2019 with SIO’s OBEM (Chesley et al., 2021) on the northern Hikurangi Subduction Zone. The data collected in 2019 was provided by Chesley et al., (2021) and was processed with EMTF (Egbert, 1998), and the data collected in 2024 was processed with BIRRP (Chave, 2004).
To address the spatiotemporal fluid migration pattern in SSE source regions and their vicinities, three marine magnetotelluric (mMT) monitoring stations were installed above a significant SSE source region on the northern Hikurangi subduction zone, one of the most extensively studied SSEs regions globally due to its extremely shallow source. One year of monitoring data was recovered in October 2024. We preliminarily analyzed the time series and estimated the MT response curves with BIRRP (Chave and Thomson, 2004) and TRACMT (Usui et al., 2024), resulting in high-quality impedance tensors and tippers at three sites. These impedance tensors were compared with data collected ~5 years ago at the same locations by the Scripps Institution of Oceanography's OBEM (Chesley et al., 2021). This comparison (Figure 1) represents the first instance of analyzing data curves from two ocean-bottom instruments at nearly identical locations. At all three sites, the comparing plots indicate generally consistent curves in MT off-diagonal components, but considerable differences happened in the MT diagonal components. We suspect that these differences in diagonal components can be attributed to the different recording lengths of two measurements and the inability to position the two comparison sites at the exact same location on the seafloor. Additionally, local discrepancies in the YX component at Site C (Figure 1) can be identified at a period range of 300s to 700s. The distance between these two comparison sites is ~1.67 km. So far, the exact reason for this discrepancy is unknown and requires further investigation. Besides, by reviewing the time series, we identified interesting occasional variations with frequencies higher than 1 Hz, likely associated with vessels passing by on the sea surface. Finally, we will discuss our time series processing plan to achieve the temporal variation patterns of subsurface fluid migration. To reach this goal, achieving robust variations in the MT responses is essential. As the uncertainties of MT transfer function estimations increase with the decreasing length of calculation samples, we plan to involve the latest proposed multitaper spectral analysis (Chave, 2019; Comeau et al., 2024) and frequency-domain independent component analysis (Cui et al., 2013; Sato et al., 2021; Zhou et al., 2022) to suppress the variances in each frequency response and obtain feasible variation sequences of transfer functions with reasonable time resolution. Additionally, our seafloor magnetic field is recorded with a sample rate of 600s during the later recording period to save power for maintaining a high sample rate (8Hz) in the electric field channel. Therefore, magnetic data on land have to be used to estimate the MT transfer function at a period range of 10s-1000s. We will show preliminary processing results by using the land magnetic data and compare them with the curves derived from the local magnetic fields.
Figure 1. Comparison of curves derived from two mMT measurements conducted in 2024 with ERI’s OBEM (Baba et al., 2020) and in 2019 with SIO’s OBEM (Chesley et al., 2021) on the northern Hikurangi Subduction Zone. The data collected in 2019 was provided by Chesley et al., (2021) and was processed with EMTF (Egbert, 1998), and the data collected in 2024 was processed with BIRRP (Chave, 2004).