Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Conjugate faulting and structural complexity on the young fault system associated with the 2000 Tottori earthquake

Abstract

Young faults display unique complexity associated with their evolution, but how this relates to earthquake occurrence is unclear. Unravelling the fine-scale complexity in these systems could lead to a greater understanding of ongoing strain localization in young fault zones. Here we present high-spatial-resolution images of seismic sources and structural properties along a young fault zone that hosted the Tottori earthquake (Mw 6.8) in southwest Japan in 2000, based on data from a hyperdense network of ~1,000 seismic stations. Our precise micro-earthquake catalog reveals conjugate faulting over multiple length scales. These conjugate faults are well developed in zones of low seismic velocity. A vertically dipping seismic cluster of about 200 m length occurs within a width of about 10 m. Earthquake migrations in this cluster have a speed of about 30 m per day, which suggests that fluid diffusion plays a role. We suggest that fine structural complexities influence the pattern of seismicity in a developing fault system.

Introduction

Fault systems develop with an increase in total fault displacement. During the initial stage of development, tectonic strain is accommodated by pervasive brittle deformation that can be associated with discontinuities such as bends, stepovers, and conjugate faults1. However, as the total displacement increases, slip gradually localizes into one major smooth fault2, eventually resulting to a release of tectonic strain along it. In a young fault system, fault geometries, and structural properties are likely to change in a complex manner. Complicate interactions between discontinuous faults and lateral structural properties may promote or inhibit fault nucleation and growth, and produce fault linkage and coalescence1. In addition, these interactions can control spatio-temporal pattern of seismic activity, slip distribution, and the location of large earthquake rupture areas, which are important in seismic hazard assessment3,4. Recent seismic and geodetic observations of fault networks have revealed that multiple ruptures along conjugate and subparallel faults can occur simultaneously or in quick succession4,5,6,7,8. Crustal fluids spreading through a fault network can also enhance fault-slip9,10,11,12. However, fault complexity along young fault systems is poorly understood owing to a lack of high-spatial resolution images of seismic sources and structural properties at seismogenic depths.

To address this issue, we deployed a spatially dense seismic array of 1000 vertical short-period sensors in and around the source area of the 2000 Tottori earthquake (Japan Meteorological Agency (JMA) magnitude Mj 7.3; Mw 6.8), and obtained continuous waveform data from April 2017 to April 2018 (13 months) (Fig. 1). The Tottori earthquake took place along a left-lateral strike-slip fault within the San-in active shear zone, which is characterized by a high seismicity rate and a large shear strain rate of 1.0 × 10−7/year13,14,15. Along the San-in shear zone, a conjugate seismic lineation with WSW–ENE and NNW–SSE trends can be recognized from the JMA catalog (Fig. 1a). A Global Navigation Satellite System network has revealed that the shear deformation zone is as wide as 50 km along the seismic belt15. However, no major active faults have been identified at the surface in the San-in region (Fig. 1a). A detailed geomorphological study has clarified that conjugate active faults with WSW–ENE and NNW–SSE trends are distributed within the San-in shear zone, but that they are young growing faults with cumulative offsets of less than a few hundreds of meters16. Indeed, no obvious surface traces of active faults linked to the 2000 Tottori earthquake have been identified (Fig. 2a), indicating that a major fault structure has not developed in the study area. These observations suggest that the San-in shear zone is a young fault system with many faults that lie parallel and oblique to the shear zone15, but there is little constraint on the small-scale structures within the shear zone.

Fig. 1: Tectonic setting and hyperdense seismic network.
figure1

a Map of seismicity and surface traces of active faults (gray lines) in and around the San-in shear zone, SW Japan. The shear zone is characterized by a high seismicity rate and a large right-lateral shear strain rate. Blue rectangle denotes the location of (b). b Locations of seismic stations (red squares) in the hyperdense network and the relocated earthquakes (blue circles). Red star denotes the epicenter of the 2000 Tottori earthquake determined by JMA with CMT solution determined by the National Research Institute for Earth Science and Disaster Resilience (NIED). Black line is the surface trace of an active fault. c Plot of cumulative number of analyzed earthquakes as a function of magnitude. d Map of the grid used in the tomography analysis (crosses) and the relocated earthquakes (blue circles), overlaid on top of the terrain. The horizontal grid spacing near the mainshock fault plane is 0.5 km.

Fig. 2: Map and cross-sectional views of persisting aftershocks.
figure2

a Epicentral map of the relocated hypocenters, which are shown as circles that are scaled to earthquake magnitude and colored to hypocentral depth. Blue star denotes the epicenter of the 2000 Tottori earthquake. Each blue arrow indicates conjugate vertical lineation. A small red rectangle around the slice 19 denotes a tiny seismic cluster in Fig. 4. Purple arrows in the inset show the orientation of the maximum compressive stress (σ1)14. Red line is the surface trace of an active fault. b Depth sections of the relocated hypocenters along the profiles shown in (a). c Seismicity cross-section for events within 2.5 km of a profile along the mainshock fault plane (blue circles). Background color denotes the coseismic-slip distribution of the 2000 Tottori earthquake inverted using strong motion waveforms and GPS data27.

The average spatial interval of the hyperdense seismic network used in this study is ~1 km, covering the entire aftershock area within a radius of ~15 km (Fig. 1b). High-density seismic arrays have been deployed in other tectonic environments, but the network used in this study is the first high-density seismic array to be installed for more than 1 year, and above the aftershock area of a magnitude ~7 earthquake17,18, to the best of our knowledge. Using the waveform data retrieved from the hyperdense network, we obtained arrival time data for 4033 micro-earthquakes with magnitudes greater than the completeness magnitude of ML−0.6 (Fig. 1c). Dense, well-covered ray-paths from the many sources provide valuable opportunities to (1) precisely map the aftershock distribution following the 2000 Tottori earthquake (Fig. 1d), (2) delineate fine-scale P-wave velocity structures with a spatial resolution down to 500 m, (3) explore the spatial and temporal evolution of a small seismic cluster, and (4) understand the multiscale complexity of this young fault system at depth.

Results

Geometric complexity of faults and velocity structures

Nearly 20 years have passed since the 2000 Tottori earthquake, but the relocated hypocenters clearly reflect complex fault structures that extend along a dominantly NW–SE trend in the source area (Fig. 2a). The relocated seismicity shows that the seismogenic zone extends down to 15 km depth, gradually decreasing to ~5 km depth toward the tips of the fault (Fig. 2b, c). This pattern of detected seismicity is similar to that of aftershocks that immediately followed the mainshock rupture14,19. This means that micro-seismicity (down to negative magnitudes) can be used to reconstruct geometric complexities produced by the mainshock rupture, even after ~20 years. The JMA catalog shows a similar spatial pattern of seismicity (Supplementary Fig. S1), but the geometric complexity (the cross-section numbers from 15 to 23 in Fig. 2a) is poorly defined due to the sparse coverage of the nationwide online seismic network.

Near the initiation point of the mainshock rupture, most events are well-aligned along a steeply dipping fault that strikes N35W–S35E (Fig. 2b). Near the mainshock epicenter, the major ruptured area is characterized as a high-velocity body (Fig. 3 and Supplementary Figs. S2S6). Within the major ruptured area, aftershocks are notably less common (Figs. 2c and 3). This is a common feature in the large-slip regions of many other earthquakes20,21.

Fig. 3: P-wave velocity perturbation along the 2000 Tottori earthquake at each depth.
figure3

P‐wave velocity perturbation rotated along the major aftershock alignment at each depth (ai). White circles denote the relocated events within 0.5 km of each depth slice. White star denotes the epicenter of the 2000 Tottori earthquake determined by JMA. White arrows indicate a thin, low-velocity band that lies subparallel to the alignment of the majority of the aftershocks. j Horizontal grid used in the tomography (0.5 km spacing).

In contrast, many relocated events on the northwest side of the source area have conjugate vertical alignments (blue arrows in Fig. 2a). The data show that the fault zone exhibits conjugate faulting over multiple length scales. The most prominent conjugate fault plane extends toward W20S from the main trend, with a horizontal length of ~5 km along strike. This prominent aftershock lineation occurs close to the distinct boundary between a low-velocity body and a high-velocity body, on the northwestern side of the mainshock epicenter (Fig. 3 and Supplementary Figs. S2S6). The boundary between the low- and high-velocity bodies trends W20S–E20N, and the observed seismicity rate is higher along this boundary than in other areas. The magnitude of the velocity contrast between the NNW and SSE sides of this boundary increases from deep to shallow depths (Fig. 3). As discussed in previous studies19,22, the shallow remarkable low-velocity body might be composed primarily of non-alkali volcanic and pyroclastic rocks of the early to middle Miocene. However, the present model shows that the velocity contrast extends down to greater depths. The high-velocity body might be interpreted to be plutonic and metamorphic rocks from surface geological map.

Toward the southeastern edge of the fault zone, the relocated aftershocks begin to change dip slightly with depth. The aftershock orientations become listric, transitioning to a dip of ~75° toward W35S at a depth of around 8 km (slices 04–10 in Fig. 2b). This dipping direction is opposite to the aftershock orientations dipping ~75° toward E35N observed on the northwest side of the source area (slices 17–21 in Fig. 2b). We also recognized a weak bifurcation geometry in the aftershocks at depths between 4 and 6 km (broken lines along the slice 02 in Fig. 2b). In terms of velocity structure, the southeast end of the aftershock area is located in a relatively low-velocity body for each of the studied depth intervals, except for the depths of 3, 4, and 10 km (Fig. 3 and Supplementary Figs. S2S6).

In the central fault section there is a thin, low-velocity band that lies subparallel to the alignment of the majority of the aftershocks. This low-velocity band is offset by ~3 km to the west, as indicated by the white arrows in Fig. 3. A similar low-velocity band can still be recognized at a spatial grid spacing of 1 km (Supplementary Fig. S3), albeit less clearly than the band in Fig. 3. The horizontal and vertical dimensions of this low-velocity band are ~5 km and ~3 km, respectively. No aftershocks have been detected along the low-velocity band, indicating that it may be located within a stress shadow zone produced by the 2000 Tottori earthquake23. The spatial resolution of the tomographic image is limited to 500 m, so it is difficult to accurately image the width of the buried low-velocity band, but the width of the low-velocity band appears to be comparable to or less than 500 m.

Spatial and temporal evolution of a tiny seismic cluster

The hyperdense seismic network of this study makes it possible to relocate micro-earthquakes accurately and to depict numerous small seismic clusters (Figs. 2 and 3). Here, we focus on a tiny seismic cluster (boosted by a ML2.2 event) near the edge of the prominent conjugate fault on the NNW side of the fault zone (a small red rectangle around the slice 19 in Fig. 2a). It has a left-lateral strike-slip focal mechanism and displays a clear lineation (Fig. 4a). Most of the relocated events at ~5.65 km depth are aligned along a vertically dipping plane that is subparallel to the mainshock fault plane (Fig. 4b). The tiny delineated fault is ~200 m long in the direction of strike and 150 m long in the direction of dip (Fig. 4c). The lineation localizes to a width of around 10 m (Inset in Fig. 4a), which is larger than the horizontal uncertainty (6.5 m: 95% confidence limit) of the relative hypocenter relocation (see “Methods”).

Fig. 4: Geometry and spatio-temporal evolution of a tiny seismic cluster.
figure4

a Map view of the relocated events boosted by a ML2.2 event. The relocated hypocenters are shown as circles that are scaled to earthquake magnitude and colored to hypocentral depth. Blue beach-balls denote focal mechanisms determined by P-wave polarity data for events with ML greater than 1.3 (Supplementary Fig. S7). Gray arrows show a possible Riedel shear structure. The inset shows a frequency distribution of the hypocenters as a function of the distance perpendicular to the fault-strike. b, c Depth sections of the relocated events perpendicular and parallel to the fault-strike, respectively. df Space–time diagram and an MT diagram of the relocated events associated with the ML2.2 event, respectively.

The tiny seismic lineation consists of several small sub-clusters that are present in the central to northern parts of the structure (gray arrows in Fig. 4a). The fault-length of each sub-cluster is ~30 m and the sub-clusters are separated by ~15 m in the horizontal direction. High-accuracy focal mechanism solutions show that one of the nodal planes (NNW–SSE trend) is parallel to the direction of the major alignment (Supplementary Fig. S7). The earthquake alignment and focal mechanisms clearly indicate that this seismic sequence was macroscopically produced by left-lateral strike-slip movement.

The spatio-temporal evolution of this tiny fault zone can be used to define two sequences (Fig. 4d–f): one characterized by aftershock activity immediately after the largest ML2.2 event (mainshock), and another characterized by swarm-like seismicity that initiated several days after the first mainshock–aftershock sequence. Based on the distribution of the immediate aftershocks, the ML2.2 event is inferred to have initiated near the center (Fig. 4d) and down-dip (Fig. 4e) of the rupture area. Interestingly, the second swarm-like seismicity sequence migrated along the strike of the fault toward its southeast edge, as well as in the down-dip direction. The migration distance was around 100 m over 4 days. Thus, the migration speed of the second sequence was ~30 m/d.

Discussion

The ultra-high density of the seismic network shows that conjugate vertical lineations of relocated seismic events could be defined using very small-magnitude earthquakes (down to ML−0.6). Sub-vertical alignments can be identified in a cross-section along the fault-strike of the 2000 Tottori earthquake (Fig. 2c). These are interpreted to be conjugate faults related to the mainshock fault plane. Conjugate vertical faults were also observed during the 2019 Ridgecrest earthquake sequence in California8 and the 2012 Indian Ocean earthquake24, in strike-slip stress domains. A mixed rupture of conjugate strike-slip and thrust faults was depicted during the 2018 northern Osaka earthquake, Japan25.

Most studies on the source process of the 2000 Tottori earthquake26,27 have suggested that the largest coseismic-slip occurred around the central section (near the rupture initiation point), with a small amount of coseismic slip at both ends of the aftershock area (Fig. 2c). The central fault section at depths shallower than 6 km corresponds to the high-velocity body, with no obvious low-velocity fault zone visible along the aftershock lineation at 500 m spatial resolution (−3 <  Y < 6 km in Fig. 3), showing relatively weak aftershock activity (Fig. 2c). Thus, the 2000 Tottori mainshock must have ruptured an immature fault zone (close to intact rock), causing the largest coseismic slip to occur in the high-velocity body at depths shallower than 6 km. Previous tomography studies have also investigated structural heterogeneities on fault planes, revealing that high-velocity features with low aftershock activity correspond to areas of large coseismic slip20,28,29. Laboratory measurements of an exhumed fault zone30 reveal that microfractures in the fluid-alteration zone are sealed leading to relatively high P-wave velocities compared with the surrounding wall rock. Sealing of microfractures in fault zone after the dynamic rupture might contribute to postseismic recovery of seismic wave velocity, resulting to vanishing of a low-velocity damaged zone associated with the 2000 Tottori earthquake.

Even after ~20 years since the mainshock, the aftershock activity in the large-slip area has been persistently low (Fig. 2c), suggesting the potential use of micro-seismic activity as an indicator of large-slip area of historical earthquakes.

High-resolution seismic tomography has revealed a thin, low-velocity band (the length is ~5 km and the down-dip height is ~3 km) that is subparallel to the mainshock fault and is offset by ~3 km (white arrows in Fig. 3). Low velocities can be caused by fractures filled with crustal fluids. Thus, the thin, low-velocity band is interpreted to be a blind fault zone within the brittle crust, where intense and pervasive fracturing has accumulated from previous events. This localized low-velocity feature (<0.5 km wide) matches with fault zone property imaged by fault-trapped waves along several active faults, including those associated with the 1992 Landers earthquake and the 1999 Hector Mine earthquake within the eastern California shear zone in USA31,32. In contrast, the Calico fault located midway between the Landers and Hector Mine ruptures has a 1.5-km-wide low-velocity zone33,34. Differences in the inferred widths of low-velocity zones may be due to intrinsic variations in the width of damaged zones around different faults and/or different spatial resolutions of geophysical explorations33.

The distinct low-velocity anomaly is present at the northwest side of the aftershock area. The aftershock distributions are also more complex than in the central section (Fig. 3 and Supplementary Fig. S2). Multiple conjugate faults are developed on the northwest side of the Tottori mainshock rupture area, where intense aftershocks followed the mainshock event35. The low-velocity zone could represent developing fracture networks at the fault edge that are filled with crustal fluids. The mainshock rupture initiated in the high-velocity body at the central section, propagated to the northwest side27, and caused shear deformation in the soft volcanic rocks (low-velocity zone), resulting to the development of the conjugate faults.

We revealed the spatio-temporal migration of earthquakes in the tiny cluster both along fault-strike and down dip (Fig. 4). The migration occurred at a speed of ~30 m/d, down to the deep part of the ML2.2 source area. During the migration, each seismic sub-cluster was activated after the propagation front had passed. Earthquake migration is believed to be due to fluid diffusion, aseismic slow-slip transients, or a combination of the two36,37,38,39. The earthquake migration speed within the tiny seismic cluster is much slower than the typical migration speed of slow-slip events along subduction zones40. Therefore, the earthquake migration could reflect infiltration of crustal fluid into the down-dip extension of the coseismic rupture area of the ML2.2 event.

Brittle deformation associated with the ML2.2 event could have induced a local stress concentration in the deep parts of the fault, facilitating the development of fractures. Crustal fluid might have infiltrated these fractures and triggered the migrating seismicity41. In addition, cascading ruptures for which small earthquakes trigger each other by stress transfer could play a role to induce the earthquake migration42. However, cascading ruptures may be less important as far as we assume a plausible range of static stress drop, as each migrating earthquake is not located at the edge of the rupture zone of the previous event (Supplementary Fig. S8).

Diffusional earthquake migration has been observed in fluid-induced earthquake experiments36, during tectonic swarm activity12, and in active volcanic areas43. Upwelling of deep fluids into the San-in shear zone has also been implied from seismic tomography and geo-chemical analysis44.

From the center of the ML2.2 fault to the north, three sets of sub-clusters can be recognized. Each cluster is ~30 m long and rotates slightly counterclockwise from the main trend (gray arrows in Fig. 4a). The sub-clusters are almost parallel to each other and have a spacing of ~15 m. These stepped, subparallel faults are similar to en echelon cracks, which can be observed at the surface following a strike-slip earthquake rupture45. The orientation of the sub-cluster relative to the main fault suggests that it could be Riedel shears formed in association with left-lateral strike-slip faulting. Riedel shear structures often form at low levels of cumulative displacement. With a progressive increase in cumulative displacement, Riedel structures will organize into a dense elongate network, eventually coalescing to form a smooth mature fault zone46,47.

The conjugate faults, potential blind fault, heterogeneous velocity structure, and Riedel shears indicate that the 2000 Tottori earthquake occurred as a part of a fault growth process in a young fault system. It appears that the fault system will develop into a mature, large-scale tectonic boundary over long time. The spatio-temporal pattern of seismic activity and the fault-slip distribution appear to be controlled by complexities within the young fault system. The smallest scale of crustal deformation that is resolved by our current seismic network (~10 m) is approaching that of a geological survey. Future hyperdense seismic networks could unravel deformation across multiple length scales, with meter or sub-meter spatial resolution. This could lead to a greater understanding of ongoing strain localization in young fault systems possessing unique complexity.

Methods

Hyperdense seismic observation

We deployed 1000 temporary seismic stations in and around the source region of the 2000 Tottori earthquake from April 2017 to April 2018. The average spatial interval of the hyperdense seismic network was ~1 km, covering the entire aftershock area (Fig. 1). Each seismic station was equipped with a short-period vertical-component sensor with a natural frequency of 2.0 or 4.5 Hz. The waveform data were recorded continuously at a sampling rate of 100 Hz and was stored on an SD card in each data logger. The internal clock accuracy was maintained to within 1 ms by calibrating at 6-h intervals, using the GPS receiver on each data logger. Approximately 800 stations transferred continuous data every 6 h via mobile internet, using a SIM card installed in each data logger. The data loggers were operated by dry batteries for over 1 year. This low-power-consumption data logger has been newly developed in cooperation with Kinkei System Corporation for the purpose of hyperdense seismic observation.

Seismic tomography analysis

To detect micro-earthquakes, we applied an automatic processing workflow (composed of phase picking and phase association) to the continuous waveform data retrieved from the hyperdense seismic network. All of the sensors were vertical components, so we focused exclusively on P-wave arrival time data. The initial hypocenter was calculated using P-wave travel times, assuming the one-dimensional velocity structure routinely used by JMA.

To relocate the detected earthquakes and estimate velocity structure, we applied Double-difference travel-time tomography code48 to the arrival time data. The earthquakes with a sufficient number of arrival times (i.e., the number of P-arrivals must be greater than 23; the median value of P-wave arrivals for each event is 209) and a low average residual of P-wave travel times were then selected for the tomographic step. The total number of selected events was 4033 (Fig. 1b), which is about nine times the number of events recorded in the JMA catalog for the same period (455 events) (Supplementary Fig. S1). This is a significant improvement in the completeness of magnitude down to ML−0.6 (Fig. 1c). The number of absolute arrival times of P-waves used for tomography analysis was 1.13 million, and the number of differential travel-time data for the catalog-picked P-waves exceeded 3.64 million. The JMA velocity model was used as the starting model for tomographic analysis (Supplementary Table S1). To obtain more precise differential P travel-time data, we applied the waveform cross-correlation technique for earthquake pairs with a spatial separation of less than 2 km. We applied a two-way 4–10 Hz Butterworth filter to waveform data and used a 1.2 s window length beginning 0.6 s before the automatically picked arrival time. We then computed the differential travel-time for each earthquake pair to subsample precision using time domain cross-correlation. After selecting the differential P travel-time with cross-correlation exceeding 0.85, we obtained a dataset of more accurate differential travel-time data that contained 8.47 million P-wave observations.

The grid nodes for the tomography were located at −300.0, −100.0, −20.0, −18.0, −16.0, −14.0, −12.0, −11.0, −10.0, −9.0, −8.0, −7.0, −6.0, −5.0, −4.0, −3.5, −3.0, −2.5, −2.0, −1.5, −1.0, −0.5, 0.0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 14.0, 16.0, 18.0, 20.0, 100.0, and 300.0 on the X-axis (E35N); −300.0, −100.0, −20.0, −18.0, −16.0, −14.0, −12.0, −11.5, −11.0, −10.5, −10.0, −9.5, −9.0, −8.5, −8.0, −7.5, −7.0, −6.5, −6.0, −5.5, −5.0, −4.5, −4.0, −3.5, −3.0, −2.5, −2.0, −1.5, −1.0, −0.5, 0.0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5, 7.0, 7.5, 8.0, 8.5, 9.0, 9.5, 10.0, 10.5, 11.0, 11.5, 12.0, 14.0, 16.0, 18.0, 20.0, 100.0, and 300.0 on the Y-axis (N35W); and −150.0, −20.0, 0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 12.0, 15.0, 20.0, 80.0, and 300.0 on the Z (depth) axis (Fig. 1d). Near the mainshock failure area, the grid spacing was 0.5 km (Fig. 3j), which is much finer than that used in previous studies19.

We also used a velocity model estimated by ref. 19 as the starting model for the tomographic analysis. However, the final velocity model does not show any significant change from the model that was inverted using the JMA velocity model as the initial model (Fig. 3). That is because the present study has used more than 20 times seismic stations and huge number of arrival time data, compared with ref. 19.

Spatial resolution of the P-wave velocity model

We tested different spatial smoothing weights of 1, 2, 3, 5, 7, 10, 15, 20, 30 45, and 60 (see ref. 48 for details) to verify the robustness of the velocity model. Despite changing the spatial weights, the main features of the tomograms remained similar in areas with good ray coverage (Supplementary Figs. S5 and S6). Considering the trade-off between roughness and stabilization of the model (Supplementary Fig. S9), we chose the velocity model with a weighting of 10 for our preferred final model (Fig. 3). The root mean square of the absolute travel-time residual was reduced from 0.057 s to 0.022 s after 28 iterations.

To evaluate the spatial resolution of the final velocity model, we conducted a checkerboard resolution test. We computed synthetic travel times for a three-dimensional velocity model using perturbations of ±3% amplitude of the initial velocity model at each inversion grid, adding random noises uniformly distributed from −0.025 s to 0.025 s for P-waves. We then inverted the synthetic data using the same distribution of hypocenters and stations as the real data, starting from the initial velocity structure, and verified how well the test model is recovered (Supplementary Figs. S2 and S4). We demonstrated that P-wave velocity anomalies are well recovered in the areas where there are sufficient seismic events (Supplementary Figs. S2 and S4) and the derivative weight sum (DWS)49 values are sufficiently large. Note that the entire aftershock area is well resolved, with horizontal and vertical spatial grid intervals of 0.5 km and 1.0 km, respectively, owing to the high density of seismic stations.

In order to assess the reliability of the model resolutions for the velocity anomalies described in the present study (Fig. 3), we also conducted a sensitivity (spike) test using a validation model shown in Supplementary Fig. S10a. Low-velocity anomalies were assigned to grid nodes at the northwestern side of the mainshock epicenter (−4% at depths from 2 to 6 km, and −2% at depths from 7 to 10 km). Furthermore, a thin low-velocity band of −3% are given to grid nodes at depths from 5 to 8 km, and a high-velocity zone with +4% are assigned to grid nodes along the fault-strike nearby the mainshock epicenter (at depths from 2 to 10 km) (Supplementary Fig. S10a). The calculation of the synthetic travel-time and the inversion method were the same as those employed for the checkerboard model resolution test. The results shown in Supplementary Fig. S10b indicate that the assumed velocity anomalies of P-wave were well recovered, although the recovery of the greater depths was weaker than that of the shallow depths. Based on these results, we consider that the presence of the velocity anomalies discussed in the present study is reliable.

Relocation of a seismic cluster associated with M L2.2

We relocated a tiny seismic cluster, boosted by an event with ML2.2. We used differential arrival times from the nearby seismic stations, with an epicentral distance of 8 km from the cluster centroid (189 stations among 1000). This seismic cluster contains 66 events that occurred from 24 December 2017 to 30 January 2018 (Fig. 4). We applied a double-difference relocation algorithm50 to differential travel times constructed by automatic picking (46 thousand P-wave data) and waveform cross-correlation technique (84 thousand P-waves with cross-correlation coefficient ≥ 0.85). We used the one-dimensional velocity structure routinely employed in the JMA location procedures. To estimate the relative location uncertainties (95% confidence limit) in the horizontal and vertical directions, we conducted a jackknife test. After generating 300 sets of differential travel times with 80% of randomly selected catalog and cross-correlation data, we applied the same relocation algorithm to all of the resampled datasets. After removing events with relatively large location uncertainties, we finally obtained the highly accurate hypocenters of 56 events. The relative location errors (95% confidence limit) are estimated to be 3 m and 12 m in longitudinal and latitudinal directions, respectively, and 19 m vertically.

Focal mechanisms

We estimated focal mechanisms of several earthquakes with ML greater than 1.3 included in the tiny relocated seismic cluster, boosted by the ML2.2 event (Fig. 4a), applying the HASH code51 to the first-motion polarities picked by automatic processing. The takeoff angles were computed using the one-dimensional velocity structure routinely used in the JMA location procedure. The focal mechanisms have the highest grade (quality A) and are well constrained based on large number of polarity data (>640). Ref. 52 has discussed a non-double couple component of several focal mechanisms using waveform data retrieved by the hyperdense seismic network consisting of ~1000 seismic stations.

Data availability

All seismic data needed to evaluate the conclusions in the paper are available in the Zenodo repository, https://zenodo.org/deposit/4059423. Additional data related to this paper are available in the paper and/or the Supplementary Materials and can be requested from the authors.

Code availability

The code of the “HASH” used to determine focal mechanism can be found at https://www.usgs.gov/software/hash-12. The code of the “hypoDD” used to relocate hypocenters can be found at https://www.usgs.gov/software/hypodd. The code of the “tomoDD” used to conduct seismic tomography with simultaneous relocation of hypocenters is available from H.Z. (email: zhang11@ustc.edu.cn) on request. We plotted figures using GMT5 (http://gmt.soest.hawaii.edu/projects/gmt). The other custom codes used in this study are available from A.K. on request.

References

  1. 1.

    Scholz, C. H. The Mechanics of Earthquakes and Faulting. https://doi.org/10.1017/9781316681473 (2019).

  2. 2.

    Wesnousky, S. G. Seismological and structural evolution of strike-slip faults. Nature 335, 340–343 (1988).

    Article  Google Scholar 

  3. 3.

    King, G. & Nábělek, J. Role of fault bends in the initiation and termination of earthquake rupture. Science 228, 984–987 (1985).

    CAS  Article  Google Scholar 

  4. 4.

    Wesnousky, S. G. Predicting the endpoints of earthquake ruptures. Nature 444, 358–360 (2006).

    CAS  Article  Google Scholar 

  5. 5.

    Toda, S. Toggling of seismicity by the 1997 Kagoshima earthquake couplet: a demonstration of time-dependent stress transfer. J. Geophys. Res. 108, 2567 (2003).

    Article  Google Scholar 

  6. 6.

    Tanaka, M., Asano, K., Iwata, T. & Kubo, H. Source rupture process of the 2011 Fukushima-ken Hamadori earthquake: How did the two subparallel faults rupture? Earth Planets Space 66, 101 (2014).

    Article  Google Scholar 

  7. 7.

    Hamling, I. J. et al. Complex multifault rupture during the 2016 Mw 7.8 Kaikōura earthquake, New Zealand. Science 356, eaam7194 (2017).

    Article  CAS  Google Scholar 

  8. 8.

    Ross, Z. E. et al. 2019 Ridgecrest earthquake sequence. Science 366, 346–351 (2019).

    CAS  Article  Google Scholar 

  9. 9.

    Miller, S. A. et al. Aftershocks driven by a high-pressure CO2 source at depth. Nature 427, 724–727 (2004).

    CAS  Article  Google Scholar 

  10. 10.

    Kato, A. et al. Anomalous depth dependency of the stress field in the 2007 Noto Hanto, Japan, earthquake: Potential involvement of a deep fluid reservoir. Geophys. Res. Lett. 38, L06306 (2011).

    Article  Google Scholar 

  11. 11.

    Kato, A. et al. Imaging the source regions of normal faulting sequences induced by the 2011 M9.0 Tohoku-Oki earthquake. Geophys. Res. Lett. 40, 273–278 (2013).

    Article  Google Scholar 

  12. 12.

    Yoshida, K., Saito, T., Urata, Y., Asano, Y. & Hasegawa, A. Temporal changes in stress drop, frictional strength, and earthquake size distribution in the 2011 Yamagata-Fukushima, NE Japan, earthquake swarm, caused by fluid migration. J. Geophys. Res. Solid Earth 122, 10,379–10,397 (2017).

    Article  Google Scholar 

  13. 13.

    Ogata, Y., Katsura, K. & Tanemura, M. Modelling heterogeneous space-time occurrences of earthquakes and its residual analysis. J. R. Stat. Soc. Ser. C Appl. Stat. 52, 499–509 (2003).

    Article  Google Scholar 

  14. 14.

    Kawanishi, R., Iio, Y., Yuyutake, Y., Shibutani, T. & Katao, H. Local stress concentration in the seismic belt along the Japan Sea coast inferred from precise focal mechanisms: Implications for the stress accumulation process on intraplate earthquake faults. J. Geophys. Res. Solid Earth 114, B01309 (2009).

    Article  Google Scholar 

  15. 15.

    Nishimura, T. & Takada, Y. San-in shear zone in southwest Japan, revealed by GNSS observations. Earth Planets Space 69, 85 (2017).

    Article  Google Scholar 

  16. 16.

    Okada, A. Characteristics of the active faults in the San’in district, western Japan. Act. Fault Res. 2002, 17–32 (2002).

    Google Scholar 

  17. 17.

    Hansen, S. M. & Schmandt, B. Automated detection and location of microseismicity at Mount St. Helens with a large-N geophone array. Geophys. Res. Lett. 42, 7390–7397 (2015).

    Article  Google Scholar 

  18. 18.

    Inbal, A., Ampuero, J. P. & Clayton, R. W. Localized seismic deformation in the upper mantle revealed by dense seismic arrays. Science 354, 88–92 (2016).

    CAS  Article  Google Scholar 

  19. 19.

    Shibutani, T. & Katao, H. High resolution 3-D velocity structure in the source region of the 2000 Western Tottori Earthquake in southwestern Honshu, Japan using very dense aftershock observations. Earth Planets Space 57, 825–838 (2005).

    Article  Google Scholar 

  20. 20.

    Kato, A., Miyatake, T. & Hirata, N. Asperity and barriers of the 2004 Mid-Niigata prefecture earthquake revealed by highly dense seismic observations. Bull. Seismol. Soc. Am. 100, 298–306 (2010).

    Article  Google Scholar 

  21. 21.

    Kato, A. & Igarashi, T. Regional extent of the large coseismic slip zone of the 2011 Mw 9.0 Tohoku-Oki earthquake delineated by on-fault aftershocks. Geophys. Res. Lett. 39, L1530 (2012).

    Article  Google Scholar 

  22. 22.

    Yukutake, Y. & Iio, Y. Why do aftershocks occur? Relationship between mainshock rupture and aftershock sequence based on highly resolved hypocenter and focal mechanism distributions. Earth Planets Space 69, 68 (2017).

    Article  Google Scholar 

  23. 23.

    Di Carli, S., Voisin, C., Cotton, F. & Semmane, F. The 2000 western Tottori (Japan) earthquake: Triggering of the largest aftershock and constraints on the slip-weakening distance. J. Geophys. Res. Solid Earth 113, B05307 (2008).

    Article  Google Scholar 

  24. 24.

    Meng, L. et al. Earthquake in a maze: compressional rupture branching during the 2012 Mw8.6 Sumatra earthquake. Science 337, 724–726 (2012).

    CAS  Article  Google Scholar 

  25. 25.

    Kato, A. & Ueda, T. Source fault model of the 2018 M w 5.6 northern Osaka earthquake, Japan, inferred from the aftershock sequence. Earth Planets Space 71, 11 (2019).

    Article  Google Scholar 

  26. 26.

    Peyrat, S. & Olsen, K. B. Nonlinear dynamic rupture inversion of the 2000 Western Tottori, Japan, earthquake. Geophys. Res. Lett. 31, L05604 (2004).

    Article  Google Scholar 

  27. 27.

    Semmane, F., Cotton, F. & Campillo, M. The 2000 Tottori earthquake: A shallow earthquake with no surface rupture and slip properties controlled by depth. J. Geophys. Res. Solid Earth 110, B03306 (2005).

    Article  Google Scholar 

  28. 28.

    Thurber, C. et al. Three-dimensional compressional wavespeed model, earthquake relocations, and focal mechanisms for the Parkfield, California, region. Bull. Seismol. Soc. Am. 96, 38–49 (2006).

    Article  Google Scholar 

  29. 29.

    Liu, X. & Zhao, D. Upper and lower plate controls on the great 2011 Tohoku-oki earthquake. Sci. Adv. 4, eaat4396 (2018).

    Article  Google Scholar 

  30. 30.

    Rempe, M. et al. The relationship between microfracture damage and the physical properties of fault-related rocks: the Gole Larghe Fault Zone, Italian Southern Alps. J. Geophys. Res. Solid Earth 123, 7661–7687 (2018).

    Article  Google Scholar 

  31. 31.

    Li, Y.-G., Vidale, J. E., Aki, K. & Xu, F. Depth-dependent structure of the Landers fault zone from trapped waves generated by aftershocks. J. Geophys. Res. Solid Earth 105, 6237–6254 (2000).

    Article  Google Scholar 

  32. 32.

    Li, Y.-G., Vidale, J. E., Oglesby, D. D., Day, S. M. & Cochran, E. Multiple-fault rupture of the M 7.1 Hector Mine, California, earthquake from fault zone trapped waves. J. Geophys. Res. Solid Earth 108, 2165 (2003).

    Google Scholar 

  33. 33.

    Cochran, E. S. et al. Seismic and geodetic evidence for extensive, long-lived fault damage zones. Geology 37, 315–318 (2009).

    Article  Google Scholar 

  34. 34.

    Yang, H., Zhu, L. & Cochran, E. S. Seismic structures of the Calico fault zone inferred from local earthquake travel time modelling. Geophys. J. Int. 186, 760–770 (2011).

    Article  Google Scholar 

  35. 35.

    Fukuyama, E., Ellsworth, W. L., Waldhauser, F. & Kubo, A. Detailed fault structure of the 2000 Western Tottori, Japan, earthquake sequence. Bull. Seismol. Soc. Am. 93, 1468–1478 (2003).

    Article  Google Scholar 

  36. 36.

    Shapiro, S. A., Rothert, E., Rath, V. & Rindschwentner, J. Characterization of fluid transport properties of reservoirs using induced microseismicity. Geophysics 67, 212–220 (2002).

    Article  Google Scholar 

  37. 37.

    Peng, Z. & Zhao, P. Migration of early aftershocks following the 2004 Parkfield earthquake. Nat. Geosci. 2, 877–881 (2009).

    CAS  Article  Google Scholar 

  38. 38.

    Roland, E. & McGuire, J. J. Earthquake swarms on transform faults. Geophys. J. Int. 178, 1677–1690 (2009).

    Article  Google Scholar 

  39. 39.

    Kato, A. et al. Propagation of slow slip leading up to the 2011 Mw9.0 Tohoku-Oki earthquake. Science 335, 705–708 (2012).

    CAS  Article  Google Scholar 

  40. 40.

    Obara, K. & Kato, A. Connecting slowearthquakes to huge earthquakes. Science 353, 253–257 (2016).

    CAS  Article  Google Scholar 

  41. 41.

    Sibson, R. H. Rupturing in overpressured crust during compressional inversion-the case from NE Honshu, Japan. Tectonophysics 473, 404–416 (2009).

    CAS  Article  Google Scholar 

  42. 42.

    Ellsworth, W. L. & Bulut, F. Nucleation of the 1999 Izmit earthquake by a triggered cascade of foreshocks. Nat. Geosci. 11, 531–535 (2018).

    CAS  Article  Google Scholar 

  43. 43.

    Yukutake, Y. et al. Fluid-induced swarm earthquake sequence revealed by precisely determined hypocenters and focal mechanisms in the 2009 activity at Hakone volcano, Japan. J. Geophys. Res. Solid Earth 116, B04308 (2011).

    Article  Google Scholar 

  44. 44.

    Nakajima, J. & Hasegawa, A. Tomographic evidence for the mantle upwelling beneath southwestern Japan and its implications for arc magmatism. Earth Planet. Sci. Lett. 254, 90–105 (2007).

    CAS  Article  Google Scholar 

  45. 45.

    Quigley, M. et al. Surface rupture during the 2010 Mw 7.1 darfield(canterbury) earthquake: Implications for fault rupture dynamics and seismic-hazard analysis. Geology 40, 55–58 (2012).

    Article  Google Scholar 

  46. 46.

    Tchalenko, J. S. Similarities between shear zones of different magnitudes. Geol. Soc. Am. Bull. 81, 1625–1640 (1970).

    Article  Google Scholar 

  47. 47.

    Katz, Y., Weinberger, R. & Aydin, A. Geometry and kinematic evolution of Riedel shear structures, Capitol Reef National Park, Utah. J. Struct. Geol. 26, 491–501 (2004).

    Article  Google Scholar 

  48. 48.

    Zhang, H. & Thurber, C. H. Double-difference tomography: The method and its application to the Hayward Fault, California. Bull. Seismol. Soc. Am. 93, 1875–1889 (2003).

    Article  Google Scholar 

  49. 49.

    Thurber, Clifford, Eberhart-phillips, D. Local earthquake tomography with exible gridding. Comput. Geosci. 25, 809–818 (1999).

    Article  Google Scholar 

  50. 50.

    Waldhauser, F. & Ellsworth, W. L. A Double-difference Earthquake location algorithm: Method and application to the Northern Hayward Fault, California. Bull. Seismol. Soc. Am. 90, 1353–1368 (2000).

    Article  Google Scholar 

  51. 51.

    Hardebeck, J. L. & Shearer, P. M. A new method for determining first-motion focal mechanisms. Bull. Seismol. Soc. Am. 92, 2264–2276 (2002).

    Article  Google Scholar 

  52. 52.

    Hayashida, Y., Matsumoto, S., Iio, Y., Sakai, S. & Kato, A. Non-double-couple microearthquakes in the focal area of the 2000 Western Tottori Earthquake (M 7.3) via hyperdense seismic observations. Geophys. Res. Lett. 47, e2019GL084841 (2020).

    Article  Google Scholar 

Download references

Acknowledgements

We deeply appreciate all members including researchers, technicians, and students who participated in and contributed to the deployment of the hyperdense seismic network. We are grateful to the residents and local governments of the Tottori, Shimane, and Okayama Prefectures for their strong support and understanding of the hyperdense seismic observations. We used the JMA earthquake catalog and waveform data from seismic stations operated by the National Research Institute for Earth Science and Disaster Prevention. This work was supported by the Grant-in-Aid for Scientific Research on Innovative Areas (KAKENHI No. 26109004) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan. This study was also partially supported by JST CREST Grant Number JPMJCR1763.

Author information

Affiliations

Authors

Contributions

A.K. carried out the data processing and seismic tomography analysis including relocation of micro-earthquakes and drafted the manuscript. A.K., S.S., S.M. and Y.I. managed and conducted the hyperdense seismic observation. S.S. conducted the data pre-processing. SS, SM and Y.I. helped draft the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Aitaro Kato.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Primary handling editor: Joe Aslin

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Kato, A., Sakai, S., Matsumoto, S. et al. Conjugate faulting and structural complexity on the young fault system associated with the 2000 Tottori earthquake. Commun Earth Environ 2, 13 (2021). https://doi.org/10.1038/s43247-020-00086-3

Download citation

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing