The Delayed Decrease of Seismicity In The Eastern Margin of The Japan Sea Due To The Megathrust Event In 2011 Along The Japan Trench

23 In the eastern margin of the Japan Sea, off the west coast of Tohoku district, the 24 seismicity increased right after the M9 megathrust event off the east coast of the Tohoku 25 district on March 11, 2011. Four months later, the seismicity decreased to the half level 26 of that before the M9 event. Such quantitative study was done by the point-process 27 model selection with AIC. The decrease lasted for eight years until an M6.7 event 28 occurred within the area in 2019. When we compare the seismicity change 29 between before and after the M9 event, with the post seismic change of the maximum 30 shear stress obtained by the viscoelastic simulation for a thousand years after the M9 31 event, we can estimate a loading rate of the shear stress in the area before the M9 as 24 32 kPa/y. For the term after the M9 event, the rate is a half of it; 12 kPa/y. When we 33 assume the whole dilatation change due to the M9 event had been canceled by the time 34 of the M6.7, the increasing rate of the mean stress after the M9 event is 21 kPa/y at 35 most. When we will be able to use JMA catalog for 2020 or later years, we can obtain 36 the seismicity level after the M6.7 quantitatively, and we will be able to narrow down 37 this estimation. 38 39


23
In the eastern margin of the Japan Sea, off the west coast of Tohoku district, the 24 seismicity increased right after the M9 megathrust event off the east coast of the Tohoku 25 district on March 11, 2011. Four months later, the seismicity decreased to the half level 26 of that before the M9 event. Such quantitative study was done by the point-process 27 model selection with AIC. The decrease lasted for eight years until an M6.7 event 28 occurred within the area in 2019. When we compare the seismicity change 29 between before and after the M9 event, with the post seismic change of the maximum 30 shear stress obtained by the viscoelastic simulation for a thousand years after the M9 31 event, we can estimate a loading rate of the shear stress in the area before the M9 as 24 32 kPa/y. For the term after the M9 event, the rate is a half of it; 12 kPa/y. When we 33 (Ogata, 1988) for the point process analysis in the present study. We excluded the 9% 103 data in the tail part, which is the half-year period after the largest event of M6.7 on June 104 18, 2019, from the analysis to eliminate aftershocks of this largest event. A half year is 105 too short to analyze the seismicity level after the M6.7 confidently. We examined 3844 106 events, which occurred in the 7930 days since Oct. 1 st , 1997, and focused on the base 107 rate comparison before and after the M9 event. 108 When ( ℎ , , ) is number of earthquakes of M≥Mth during the period from S 109 to E, the averaged occurrence rate of M≥Mth for that term; λ( ℎ , , ) is 110 . = ( ℎ , , ) = ( ℎ , , ) ( − ) .
(2) 116 When λ changed at t=Tc, λ for the term before Tc, and λ for the term after Tc are, 117 = ( ℎ , 0, ) = ( ℎ ,0, ) , and AIC of this model is 120 The penalty of introducing Tc in a model is not larger than 12 (Kumazawa et al., 2010). 124 When < , the model with the rate change is selected, and the Tc of the best 125 model minimizes . In order to avoid the effect of inhomogeneity in the long term 126 catalog, Mth of 2.5 and 3.0 are examined for model selections. Models with Tc within a few-year term after the M9 event give far smaller AIC ( Fig.  132 3(c)) than that of Eq. (2). It is confirmed that the occurrence-rate decrease delayed for a 133 few months from the occurrence time of the M9 event ( Fig. 3(d)). In order to detect the start of that delayed decrease correctly, data within the one month after the M9 event is 135 excluded in the next analysis (Fig. 4). When Mth=3.0, the similar delayed decrease of 136 AIC is also confirmed (Fig. 5). Tcs with smallest three AICs of each model are listed in 137 induced M6.4 event and its aftershocks are included, the delayed decrease is significant 141 (Fig. 3). The decrease rate sustained for more than four years, which made us notice it 142 even from the raw distribution maps of epicenters. 143 Due to the COVID-19, the JMA hypocenter catalog is not yet fixed for 2020 and 144 2021 at present. Since available catalog after the M6.7 covers only a half year, we 145 cannot obtain λ for the period after M6.7 by eliminating its aftershocks in this study 146 precisely. However, the most right part of Fig. 2 looks that λ after the M6.7 at the end of 147 2019 is larger than ℎ . This must be verified in the future study with a longer 148 catalog. In this study, we report that the seismicity decrease to the half level prevailed 149 for eight years after four-month immediate activation after the M9 event.  Table 3. The area we examined the seismicity is the 163 shallow part of the left end of their simulated region (400~550 km horizontally remoted 164 to the WNW direction from the trench axis on the A-A' line). 165 In Fig. 8, temporal changes of the mean stress and the maximum shear stress at the point shown in Fig. 6 (10km-depth and -422km from the trench axis on A-A') are 167 plotted. Since that point is in the elastic layer, the dilatation and the maximum shear 168 strain are proportional to the mean stress (the expanding direction is positive), and the 169 maximum shear stress, respectively. 170 The simple approximated expression of the isotropic stress σiso(t) of the point (- However, after Tc, the seismicity of the area is so stable for eight years. We can treat 220 the pore fluid pressure after Tc as a part of the background stress field, which should 221 have been changing very smoothly in the area, after the drastic change of pore fluid had 222 been settled before Tc. 223 Therefore, we assume that the gradually changing crustal stress after Tc was the 224 summation of the background tectonic stress and the coseismic and post seismic stress 225 changes due to the M9 event. The post-seismic increase of 96 kPa in the maximum 226 shear stress for eight years resulted in the post seismicity of the area. To sustain the 227 double level of seismicity for years before the M9 event, the double of the post 228 maximum shear stress change is necessary. It give us 24 kPa/y in the studied area as the 229 tectonic accumulation rate of shear stress before the M9 event. This rates give us a 230 recurrence time estimation of about two hundred years for a major event of this area 231 with 5 MPa stress drop. As the shear strain rate, it is equivalent to 4×10 -7 /y. 232 When we assume that the dilatation due to the M9 event, including both coseismic 233 and post seismic changes, had been almost canceled by the time of the M6.7 in the 234 studied area, the tectonic rate of the compressional stress in this area after the M9 event 235 becomes 21 kPa/y. It is also equivalent to the strain rate of 4×10 -7 /y. If we assume that 236 the recovery was half of the pre-M9 level at the time of M6.7, the rate is 10 kPa/y. In 237 order to determine the current mean stress status, we should examine the post M6.7 238 seismicity quantitatively in future study with a catalog with the extended period. 239 As the order estimation of the background loading rate in the eastern margin of the 240 Japan Sea, we get 24 kPa/y of shear rate for the term before the M9 event, and 12 kPa/y 241 of shear rate for the term after the M9 event. Since the seismicity decrease we found 242 was stable for eight years, ignoring the effect of changes in the pore fluid pressure of the 243 area does not affect our estimations seriously. We also ignore the effect of after slips 244 around the source area of the M9 event, because the studied area is more than 150 km 245 away from the regions slipping afterwards. As the background accumulation rate of compressional mean stress after the M9 event, we get 21 kPa/y or lower. With the 247 analysis of the seismicity after the M6.7 event, it will be better to use ΔEFS to examine 248 post-M9 mean-stress status further. 249 The rate of horizontal strain obtained from GPS data for the eastern margin of the the Niigata region as 6×10 -8 /y for pre-M9, and 3×10 -8 /y for post-M9 terms. Their strain 253 rates are two dimensional. Plane strain is usually smaller than three-dimensional strain. 254 It is considered that rates we obtained are within the permissible range of the GPS 255 observation. The shear strain rate for the post-M9 term we obtained is a half of the pre-256

Conclusion 258
We examined the delayed decrease of seismicity of the area in the eastern margin 259 of the Japan Sea with the point process model selection. The area remotes 300-400 km 260 horizontally away along the plate motion direction from the center of the mega-thrust 261 event in 2011. In this area, the seismicity increased right after the M9 event with the induced M6.4 strike-slip event. After about four months, the seismicity became the half 263 level of that before the M9 event. The decreased rate lasted for eight years. After 8.25 264 years, an M6.7 thrust-type event occurred in the area. Due to the aftershocks of this 265 event, we cannot determine the base occurrence rate at present yet with the fixed JMA 266 catalogue. However, the raw cumulative number curve suggests that the seismicity is 267 recovering after the M6.7. 268 Referring the viscoelastic simulation, the coseismic and post seismic dilatation 269 change for eight years due to the M9 was examined. It gives the 21 kPa/y as the 270 maximum tectonic increase rate of mean compressional stress on this area after the M9 271 event, when we assume the occurrence of the M6.7 represents the full recovery of the 272 compressional stress field in the eastern margin of the Japan Sea. When the JMA 273 catalogue will allow us to analyze seismicity after the M6.7 fully, the mean stress level 274 at present will be clearer. 275 From the post seismic maximum shear strain change and the half level of the 276 seismicity after the M9 event, the pre-M9 seismicity in the area suggests 24 kPa/y as the 277 rate of tectonic shear stress increase before the M9 event. The estimated loading rates of 278 this region are a several times larger than those obtained from the GPS data in the plane 279 strain framework. However, our finding that the pre-M9 term rate is the double of the 280 post-M9 term well agrees with that obtained for EW-deformation rate in Niigata 281 (Fukahata et al., 2020). For a long-term hazard estimation of the eastern margin of the 282 Japan Sea area, the rates we obtained will be another proxy values.    Since the point is in the elastic layer, dilatation and the maximum shear strain are 434 proportional to the mean stress, and the maximum shear stress, respectively. With the 435 bulk and shear moduli in Table 3 Underlined value is the smallest AIC for each data set. A numerical subscript for AIC 444 and λ represents Mth of each data set. "w/c" represents that the data without a month 445 after the M9 event is used. All λs are occurrence rates per one day.    Occurrence rates and AIC at each Tc for M≥2.5 without one month data after the M9 event. Tc in July 2011 gives the smallest AIC. See the caption of Fig. 3.  Temporal changes of the isotropic strain due to the M9 event. The vertical section of the isotropic strain changes along the line A-A' in Fig. 1 at t=0 (coseismic), 5, 10, 20, 30, 50, and 100 years after the M9 event are shown by color and contours. The red and blue color scales represent expansion, and contraction, respectively. Contours of 5×10-5, ±2×10-5, ±1×10-5, ±5×10-6, ±2×10-6 are also shown by thin lines. The thick solid line indicates the PAC plate interface shallower than 50km. The thin horizontal line at 40km-depth indicates the interface between elastic and viscoelastic layers. The analyzed area is shown by the blue dotted rectangle in the top gure. The green star in the top gure shows the position of the point whose stress change is plotted in Fig. 8. Deviatoric shear strain changes due to the M9 event. The vertical section of the deviatoric strain elds along the line A-A' in Fig. 1 at t=0 (coseismic), 5, 10, 20, 30, 50, and 100 years after the M9 event. The deviatoric maximum shear strain is shown by color, and contours of 5×10-5, 2×10-5, 1×10-5, 5×10-6, and, 2×10-6. The black and white bars show the extension and contraction directions with those magnitudes at each point, respectively. See the caption of Fig. 6.

Figure 8
Changes of the mean stress (extension positive), and the maximum shear stress at the point of (422kmwest from the trench axis, 10km-depth) shown in Fig. 6 due to the M9 event, for 50 years. Since the point is in the elastic layer, dilatation and the maximum shear strain are proportional to the mean stress, and the maximum shear stress, respectively. With the bulk and shear moduli in Table 3, these are converted to stresses. The values at t=0 are coseismic values.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. AdditionalFiles.pdf