Postseismicity of slow-slip doublets discerned on the outermost of the Nankai Trough subduction megathrust

Abstract

Despite dissimilar slip rates, slow earthquakes are faulting as ordinary earthquakes are. It is therefore physically natural that slow earthquakes also cause postseismic motions similarly to ordinary earthquakes, even though coseismic and postseismic slips remain undifferentiated for slow earthquakes. We pursue the slow-earthquake postseismicity based on the analysis of a fault slip beneath the Bungo Channel, the westernmost region of the Nankai Trough subduction zone in southwestern Japan. Its 2010 long-term slow slip event (SSE) was mispredicted by physics-based models, which concludes that the initial acceleration of this SSE was too abrupt for a slow variant of a fault rupture. We identify that a mispredicted GNSS signal evolves logarithmically in time, preceded by minor signals that evolve exponentially, lasting about two years west and about half a year east. By performing sparse inverse modeling on the GNSS, we have estimated that exponential slips occur at the same depth, bracketing a logarithmic slip that occurs beneath the channel. The regions of exponential slips match repeating slow-slip regions, and deep tremors synchronize exclusively with the logarithmic slip. This source complexity can be explained as a neighboring rupture doublet and its afterslip and aftershocks by the known mechanics of ordinary earthquakes. If slow earthquakes have a dual origin in exponentially nucleating slow rupture and logarithmically decelerating postseismic creep, it is possible to pick the slow earthquake nuclei that could accelerate into megathrust catastrophes.

Figures (5)

• Figure 1: Study area and data set. (a) Selected GNSS stations surrounding the Bungo Channel. 83 stations were selected from 86 stations located within 131--134$^{\circ}$E and 32--34$^{\circ}$N, excluding three outliers where annual variations exceed 0.5 cm. Three groups of distinctive trends are colored blue, purple, and khaki, while the others are shown in gray. Geographical names appearing in the text are also indicated. (b) Three characteristic signal trends of the 2010 Bungo Channel SSE. The horizontal components of GNSS displacements normal to the trough are plotted from July 2006 to June 2011 in light colors, with their stacked values in dark colors. The color scheme follows Fig. \ref{['fig:1']}a. Data were processed to remove signals of antenna replacements and linear trends fitted within a pre-event window from 2006.5 to 2008.5, filtered using a moving average over one month subsequent to the date on the horizontal axis.
• Figure 2: Time-domain decomposition and slip inversion of decomposed signals. Time-domain decomposition is based on least-square functional fitting, and slip inversion evaluates the most probable estimate using the prior constraint of the Elastic Net. Estimated slips are visualized by two different color scales, distinguished by whether the moment-magnitude estimate is below or above 6.7, assuming a rigidity of $30 \mathrm{~GPa}$; the contours of the preceding and subsequent events are drawn in blue and red, respectively. (a) Partitioning of coseismicwise and postseismicwise time domains based on functional fitting (a1: eq. 1; a2: eq. 2). The colors of the lines follow Fig. \ref{['fig:1']}b, and the fitted curves are shown in the corresponding light colors. Vertical solid lines indicate the estimated onset times of the coseismicwise and postseismicwise signals ($t_1$ and $t_2$), and vertical dash-dots indicate the estimated termination time of the coseismicwise signal $t_2^\prime$ set by eq. (2), which is the intersection of the sigmoidal growth and the logarithmic rise (a3, dash-dot curves). Horizontal lines in Fig. \ref{['fig:2']}a1 represent the cutoff time $t_c$ of the log slip. The dotted lines in Fig. \ref{['fig:2']}a1 indicate the fitting of eq. (1) for the data period before the end of October 2010, after which the eastern Shikoku SSE may contaminate the data. (b) Inversion of logarithmic signals and preceding signals, separated at time $t_2$, February 2010. (c) Inversion of GNSS data separated at the onset time of the coseismicwise signal $t_1$, July 2009, and its termination time $t_2^\prime$, May 2010.
• Figure 3: Comparison between the inferred 2010 SSE and slow-earthquake records, and a Bungo slow-earthquake synthesis. (a) Estimated slip zones of the 2010 event and documented slow earthquakes. The estimated slip follows the result shown in Fig. \ref{['fig:2']}b2, where the red contour outlines the exp slips that precede the log slip ($t<t_2$), while the color map represents the log slip ($t>t_2$). The slow-earthquake record consists of the estimated SSE areas [red rectangles takagi2019along] and the tremor events (gray circles) that occurred between February and October annoura2016total. For the synchronized events south of 33.4$^{\circ}$N (orange circles), the slip distance due to burst events exceeding the offset rate is calculated from the radiated energy. The number of tremors and its offset rate (straight line) are shown alongside the stacked pre-&co-seismicwise signals (gray, from Fig. \ref{['fig:1']}b purple). (b) Interpretations of the estimated episodes, fitted to the Bungo slow-earthquake cycle. Longitudinal slip averages for latitudinal bins of 10 km are calculated from slip inversions. Exponential growths and the primary parts of logarithmic slips are respectively interpreted as coseismic ruptures and postseismic creeps, namely afterslips. An early part of the log trend is interpreted as a compound of rupture and creep, considering repeater locations and source mechanics. Slip contours obtained in Fig. \ref{['fig:2']}b for $t<t_2$ and $t_2<t$ and Fig. \ref{['fig:2']}c2 for $t_1<t<t_2^\prime$ are drawn in the colors of the average slips for the time windows of the same ends.
• Figure S1: Slip estimates for $t_1<t<t_2$ and $t_2<t<t_2^\prime$. The most probable estimates using the prior constraint of the Elastic Net are evaluated for $t_1<t<t_2$ (top) and $t_2<t<t_2^\prime$ (bottom).
• Figure S2: Differential and cumulative slips calculated from slip estimates at different time windows. Longitudinal averages are calculated as in Fig. 3b but using the slip estimates for $t<t_1$ (Fig. 2c1), $t<t_2$ (Fig. 2b1), $t_2<t<t_2^\prime$ (Fig. S1, bottom), and $t>t_2^\prime$ (Fig. 2c3). (a) Differential slips. Longitudinal averages and contours are shown. Each slip estimate is plotted as is, but for $t_1<t<t_2$, where the slip is calculated from the difference in the estimated slip between $t<t_1$ and $t<t_2$. (b) Cumulative slips, calculated as the sums of differential slips shown in Fig. \ref{['fig:sup2']}a.