5:15 PM - 6:30 PM
[SCG39-P23] Variations of source characteristics of shallow tectonic tremor, Nankai trough, inferred from changes of the event size distribution
Keywords:DONET, Nankai trough, earthquake size distributions, slow earthquakes
The event size distribution is one of the scaling relationships that characterize the source processes of seismic phenomena. In our previous study of shallow tectonic tremor along the Nankai trough (Nakano et al., 2019), we found that the event size distribution follows a tapered Gutenberg-Richter (TGR) distribution (Kagan, 2002), which is a GR-like distribution with an exponential taper for events larger than a cut-off magnitude (Mc). For ordinary earthquakes, the slope (b-value) of the GR distribution has been related to the stress level in the medium, and the cut-off magnitude (Mc) of TGR may be related to the fault dimension. For Nankai trough slow earthquakes, both b-value and Mc differ in activities during different time periods in the same source region, implying that the source characteristics of these events change with time, although the cause is poorly understood.
In this study, we analyzed the size distributions of shallow tectonic tremor along the Nankai trough during tremor episodes that occurred off Kumano in 2016 and off the Kii Channel in 2018, both activities lasted at least one month. By fitting the data with the TGR distribution, we found that the b-value increased as Mc decreased in the later part of the episode as the rate of events decreased. Because the tremor sources migrated during each episode, the changes in these parameters may represent spatial or temporal changes in the source characteristics.
We qualitatively interpreted the controlling factors of the size distribution of slow earthquakes using the 2D probabilistic cell automaton (PCA) model proposed by Ide and Yabe (2019). The 2D PCA model has three probabilistic parameters as pb, pl, and pv, where pb is a probability of interactions between adjacent cells, pl is a probability of the random ignition of slip in a cell, and pv is a probability related to energy dissipation during slip. We found that the b-value increase while Mc decrease as pl decrease or pv increase, and the observed changes of the b-value and Mc of shallow tremor can be qualitatively explained by a decrease in pl or an increase in pv.
In the following, we discuss the causes that change these probabilistic parameters. Accumulated stress on the causative fault drives the spontaneous activity of regular and slow earthquakes. Accumulated stress that is initially high is released gradually in slow earthquakes, which may decrease the event ignition probability pl that is consistent with the observations. Because the slip history of previous slow earthquakes may result in a heterogeneous distribution of accumulated stress, the migration of tremor sources may also affect the tremor size distributions. The degree of coupling along the plate interface may affect the stress accumulation rate and accordingly the slip ignition probability, and the spatial variations of the coupling ratio may also be related to the variations of the event size distributions.
Frictional resistance depends on the pore-fluid pressure on the fault interface. A decrease in pore-fluid pressure increases frictional resistance during slip and therefore increases the energy dissipation probability pv or decreases the slip ignition probability pl in the 2D PCA model. The observed b-value increase and Mc decrease can be explained by a decrease of pore-fluid pressure as tremor activity decreases in the later part of tremor episodes. Spatially heterogeneous distribution of pore-fluid pressure and frictional properties on the fault interface may change the event size distribution of tremor as it migrates.
In summary, observed changes of the tremor event size distribution can be attributed to the release of accumulated stress by slow earthquakes or spatial-temporal variations of pore-fluid pressures during slow earthquakes.
Acknowledgments: This research was supported by JSPS KAKENHI Grant Number JP18K13639, JP19K04050, and 16H06477 in Scientific Research on Innovative Areas “Science of Slow Earthquakes”
In this study, we analyzed the size distributions of shallow tectonic tremor along the Nankai trough during tremor episodes that occurred off Kumano in 2016 and off the Kii Channel in 2018, both activities lasted at least one month. By fitting the data with the TGR distribution, we found that the b-value increased as Mc decreased in the later part of the episode as the rate of events decreased. Because the tremor sources migrated during each episode, the changes in these parameters may represent spatial or temporal changes in the source characteristics.
We qualitatively interpreted the controlling factors of the size distribution of slow earthquakes using the 2D probabilistic cell automaton (PCA) model proposed by Ide and Yabe (2019). The 2D PCA model has three probabilistic parameters as pb, pl, and pv, where pb is a probability of interactions between adjacent cells, pl is a probability of the random ignition of slip in a cell, and pv is a probability related to energy dissipation during slip. We found that the b-value increase while Mc decrease as pl decrease or pv increase, and the observed changes of the b-value and Mc of shallow tremor can be qualitatively explained by a decrease in pl or an increase in pv.
In the following, we discuss the causes that change these probabilistic parameters. Accumulated stress on the causative fault drives the spontaneous activity of regular and slow earthquakes. Accumulated stress that is initially high is released gradually in slow earthquakes, which may decrease the event ignition probability pl that is consistent with the observations. Because the slip history of previous slow earthquakes may result in a heterogeneous distribution of accumulated stress, the migration of tremor sources may also affect the tremor size distributions. The degree of coupling along the plate interface may affect the stress accumulation rate and accordingly the slip ignition probability, and the spatial variations of the coupling ratio may also be related to the variations of the event size distributions.
Frictional resistance depends on the pore-fluid pressure on the fault interface. A decrease in pore-fluid pressure increases frictional resistance during slip and therefore increases the energy dissipation probability pv or decreases the slip ignition probability pl in the 2D PCA model. The observed b-value increase and Mc decrease can be explained by a decrease of pore-fluid pressure as tremor activity decreases in the later part of tremor episodes. Spatially heterogeneous distribution of pore-fluid pressure and frictional properties on the fault interface may change the event size distribution of tremor as it migrates.
In summary, observed changes of the tremor event size distribution can be attributed to the release of accumulated stress by slow earthquakes or spatial-temporal variations of pore-fluid pressures during slow earthquakes.
Acknowledgments: This research was supported by JSPS KAKENHI Grant Number JP18K13639, JP19K04050, and 16H06477 in Scientific Research on Innovative Areas “Science of Slow Earthquakes”