Nucleation process of the 2011 northern Nagano earthquake from nearby seismic observations

The 2011 magnitude (M) 9.0 Tohoku-oki earthquake was followed by seismicity activation in inland areas throughout Japan. An outstanding case is the M6.2 Northern Nagano earthquake, central Japan, occurred 13-h after the megathrust event, approximately 400 km away from its epicenter. The physical processes relating the occurrence of megathrust earthquakes and subsequent activation of relatively large inland earthquakes are not well understood. Here we use waveform data of a dense local seismic network to reveal with an unprecedented resolution the complex mechanisms leading to the occurrence of the M6.2 earthquake. We show that previously undetected small earthquakes initiated along the Nagano earthquake source fault at relatively short times after the Tohoku-oki megathrust earthquake, and the local seismicity continued intermittently until the occurrence of the M6.2 event, being likely ‘modulated’ by the arrival of surface waves from large, remote aftershocks off-shore Tohoku. About 1-h before the Nagano earthquake, there was an acceleration of micro-seismicity migrating towards its hypocenter. Migration speeds indicate potential localized slow-slip, culminating with the occurrence of the large inland earthquake, with fluids playing a seismicity-activation role at a regional scale.


Text S1
[Detection of Tohoku-oki aftershocks that could dynamically trigger local seismicity] Many large aftershocks occurred immediately after the M9.0 Tohoku-oki megathrust event, raising the question whether dynamic stress changes caused by the large aftershocks modulated local seismicity at places located at teleseismic distances from the megathrust's epicenter. Therefore, we have searched for Tohoku-oki aftershocks that possibly triggered local events in the northern Nagano region, using the procedure outlined below.
To identify aftershocks to include in this analysis, we first estimated a threshold magnitude for an event that could potentially trigger local seismicity by dynamic stresses. Van der Elst and Brodsky (2010), using a statistical approach, estimated that the minimum dynamic stresses that could trigger earthquakes are on the order of 0.1 kPa.
The peak dynamic stress is proportional to Gu'/vph (Jaeger and Cook, 1979), where G is the shear modulus (30 GPa), vph is the phase velocity (considered here as 3.5 km/s for Rayleigh waves and 4.1 km/s for Love waves, e.g., Peng et al., 2009, Shimojo et al., 2014, and u' is the peak particle velocity (determined directly from seismograms).
Thus, dynamic stress changes on the order of 0.1 kPa correspond to peak velocity amplitudes on the order of a few 10 4 of nm/s. A visual check of the low-frequency envelope of the vertical component waveform at station NZWH showed that most peak amplitudes excited by events larger than MJMA5.5 that occurred in the Tohoku-oki aftershock area (34.38° -40.11°N, 137.58° -144.79°E) exceeded a few 10 4 nm/s.
Hence, we considered 113 aftershocks of MJMA ≥ 5.5, which occurred during a 13-hour period after the Tohoku-oki mainshock, as potentially being able to trigger earthquakes in northern Nagano.
To detect remote aftershocks that triggered local seismicity, we first calculated the Love-wave arrival times for the 113 MJMA ≥ 5.5 Tohoku-oki aftershocks (referred to as large events) in each of the following northern Nagano areas: the eastern and western clusters referred in this paper, as well as the South and Zigokudani areas ( Figure S1), where triggered seismicity has been detected by Shimojo et al., (2014) through MFT analysis, using only Hi-net stations. We used the center of the area studied in this paper (37.0226°N, 138.6193°E -the center of the black rectangle in Figure 1) as a reference point for the seismicity in the western and eastern clusters and assumed a Love-wave  Figure S1).
We then generated 1000 synthetic catalogues by randomly shuffling the time difference between temporally adjacent local events listed in the real catalogue for each area. Then, we have applied the procedure described above to these randomized synthetic catalogues. Since we have used 113 Tohoku-oki aftershocks, there were 113  1000 = 113,000 values of a and ab obtained for each area. We calculated the cumulative frequency distributions of these 113,000 values for a and ab, for each area, and defined the upper 5% of the cumulative distribution as the threshold value for detecting remote aftershocks with "anomalously" large a or ab values.
Finally, using the thresholds obtained for each area and time window, we have identified the large remote events (i.e., the Tohoku-oki aftershocks) associated either with significant a values (interpreted as remarkable local seismicity) or ab values (interpreted as significant activation) and considered them as potential triggering events.
We have then visually checked the low-frequency waveforms at station NZWH and excluded from the list of triggering events those earthquakes for which surface wave trains could not be confirmed.  Table S2. In Tables S1 and S2, the threshold values for each event, area, time window, and range of magnitudes are shown at the bottom of each column.

We next calculated A and
As a result, in both the western and eastern clusters, the first local events occurred during the passage of the surface waves from Tohoku-oki aftershocks ( Figure S2 and Table S1). Figure S2 and Table S1 include only moderate or large aftershocks with (a -  Table S3 presents the estimated maximum dynamic stress changes for seven Tohoku-oki aftershocks that were followed either by a significant increase of local seismicity or a remarkable, increased local activity. The surface wave arrivals from the MJMA 6.5 and MJMA 7.4 aftershocks, which were followed by the earliest local events in the western cluster, about 5 km south of the M6.2 northern Nagano event, were associated with maximum dynamic stress changes on the order of 10 kPa. Similarly, the arrival of surface wave from the MJMA 7.5 event, which was followed by the two earliest local events in the eastern cluster, was also associated with a maximum dynamic change on the order of 10 kPa. The surface wave arrivals from the MJMA6.8 and MJMA6.1 aftershocks, which were followed by significant levels of local seismicity in the western cluster, were associated with maximum dynamic stress changes on the order of 10 kPa and 1 kPa, respectively. The surface waves arrivals from both the MJMA6.2 and MJMA6.0 events, which were followed by the first and third periods of activity in the western cluster ( Figure 4), respectively, were associated with maximum dynamic stress changes of about 1 kPa.
We also confirmed whether local seismicity in the western, eastern, South, and Zigokudani areas was not only dynamically triggered by individual aftershocks occurred in the Tohoku region, but was also modulated throughout the aftershock sequence. In the western, eastern, and South areas, aftershocks did not significantly increase seismicity in any time window or magnitude range (Table S2). In the Zigokudani area, however, aftershocks with magnitudes larger than MJMA6.0 clearly increased local seismicity significantly in most examined time windows. Shimojo et al.  The symbols have the same meaning as in Figure 1a. The area "Zigokudani" is a concentrated seismicity spot, characterized by high heat flow, where the occurrence of earthquakes proves to be sensitive to small (remote) stress perturbations.   Figure S1. Besides the "East" and "West" regions, discussed in this paper, we have also tested the triggering for the "South" and the "Zigokudani" areas, both known for geothermal activity. The "A" and "B" symbols represent the sum of individual "a" and "b" (Table S1) Table S3. List of dynamic stress changes associated with the 2011 Tohoku-oki earthquake and several of its aftershocks that have been discussed in the paper.
The columns list the event information (origin time, location and magnitude), the maximum dynamic stresses (kPa) corresponding to the passage of the Love and Rayleigh wave trains, the statistical significance for triggering (Table S1) and some important notes.  We estimated diffusivity values of ~10 m 2 /s for earthquake migration or expansion in both eastern and western clusters. We assumed that the permeability K is proportional to D C T (Noir et al., 1997), where D is diffusivity, is porosity, is viscosity, and C T is the isothermal compressibility coefficient. Using nominal values = 0.05, = 2.0×10 -4 Pa·s, and C T = 4.0×10 -10 Pa -1 for water under temperature and pressure conditions of around 5 km depth, we estimated the permeability to be 4.0×10 -14 − 4.0×10 -13 m 2 . This permeability is 0-5 orders of magnitude larger than expected at depths of 1-10 km in typical brittle crust (Ingebritsen and Manning, 2002) and is roughly consistent with the relatively high values sometimes associated with earthquake migration or expansion involving the flow of highly pressurized fluids (Ingebritsen and Manning, 2010). We speculate that the relatively high permeability values in this study indicate fluid diffusion from highly pressurized source areas to lower pressure areas, which may have resulted from temporarily enhanced permeability due to the influence of the Tohoku-oki earthquake.