JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, B03S08, doi: /2006jb004320, 2007

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112,, doi: /2006jb004320, 2007 Fault-related folds above the source fault of the 2004 mid-niigata Prefecture earthquake, in a fold-and-thrust belt caused by basin inversion along the eastern margin of the Japan Sea Yukinobu Okamura, 1 Tatsuya Ishiyama, 1 and Yukio Yanagisawa 2 Received 1 February 2006; revised 10 October 2006; accepted 16 January 2007; published 28 March [1] The 2004 mid-niigata Prefecture earthquake occurred in a fold-and-thrust belt that has been growing since late Pliocene time in a Miocene rift basin along the eastern margin of the Japan Sea. We constructed the trajectory of the subsurface faults responsible for the growth of the folds from the geologic structure and stratigraphy in the source region, assuming that the folds have been growing as fault-related folds because of inclined antithetic shear with a dip of 85 in the hanging wall above a single reverse fault. The fault trajectory constructed from the fold geometries nearly coincides with the geometries of the source fault of the main shock of the 2004 earthquake revealed by aftershocks, which supports that the rupture was along a geologic fault that has ruptured repeatedly during the last a few million years. A three-dimensional fault model based on 12 fault trajectories constructed along parallel sections revealed that the main shock occurred on a convex bend in the fault surface and that the southern termination of the aftershock distribution nearly coincides with a concave bend in the fault. The close relation between the source fault and the geologic structure shows that it is possible to construct source fault geometry assuming inclined shear as deformation mechanism of a hanging wall and to infer the rupture areas from geologic data. Citation: Okamura, Y., T. Ishiyama, and Y. Yanagisawa (2007), Fault-related folds above the source fault of the 2004 mid-niigata Prefecture earthquake, in a fold-and-thrust belt caused by basin inversion along the eastern margin of the Japan Sea, J. Geophys. Res., 112,, doi: /2006jb Introduction [2] The concept of fault-related folding suggests that it is possible to infer the geometry of a subsurface dip-slip fault from the geometry of surface folds [e.g., Suppe, 1983; Gibbs, 1983]. This methodology has previously been applied to the evaluation of active faults and earthquake potential in sedimentary basins [Shaw and Suppe, 1994, 1996; Shaw et al., 2002; Ishiyama et al., 2004]. By comparing faults constructed on the basis of the theory of fault-related folding with source faults of earthquakes, we can examine the reliability of this fault modeling method. Davis and Namson [1994] and Carena and Suppe [2002] showed that the source fault of the 1994 Northridge earthquake is consistent with a fault inferred from fold geometry in the source area; however, in this case it is difficult to obtain a precise geologic structure related to the source fault because the structure is masked by overlying complex deformation [Carena and Suppe, 2002]. [3] The 2004 mid-niigata Prefecture earthquake occurred in a fold-thrust belt that has been growing during the 1 Active Fault Research Center, National Institute of Advanced Industrial Science and Technology, Tsukuba, Japan. 2 Institute of Geology and Geoinformation, National Institute of Advanced Industrial Science and Technology, Tsukuba, Japan. Copyright 2007 by the American Geophysical Union /07/2006JB004320$09.00 Quaternary, and aftershocks related to this event were observed by a dense seismic network deployed after 2 to 5 days of the main shock [e.g., Kato et al., 2005a, 2006; Okada et al., 2005]. These data show the precise geometry of the source fault and thus provide an opportunity to examine whether the method of fault-related folding can be used to infer successfully the subsurface source fault. Okamura and Yanagisawa [2005] and Sato and Kato [2005] have shown that there is close relation between fold geometry and the source fault based on construction of a balanced cross section. In this study we constructed threedimensional subsurface fault models from the detailed geologic structures in the source area assuming inclines shear as deformation mechanism of the hanging wall, and examined the relationship between the fault models and the source fault of the 2004 earthquake. 2. Seismotectonic Setting [4] The 2004 mid-niigata Prefecture earthquake occurred at the southern end of the eastern margin of the Japan Sea, where the convergent plate boundary between the Amurian and Okhotsk plates is thought to be located (Figure 1). Wei and Seno [1998] estimated the convergence rate of these plates to be about 1.5 cm/yr in this region. The margin was initially formed as a rifted passive margin, mainly during the early Miocene when the Japan Sea opened, and it has 1of14

2 Figure 1. Index and tectonic maps. (right) Plate tectonic framework around the eastern margin of the Japan Sea. OK, Okhotsk plate; AM, Amurian plate; PA, Pacific plate; PH, Philippine Sea plate. (left) Tectonic map showing the contraction deformation zones (dotted areas) concentrated in several fold belts (modified from Okamura [2002]) and fault models of recent major earthquakes (rectangles), based on work by Okamura et al. [2005], Tanioka et al. [1995], Satake [1985], and Abe [1975]. Magnitude of the earthquakes is shown by M JMA. accommodated E-W shortening since the late Pliocene [Sato, 1994; Okamura et al., 1995]. Many rift basins formed in the early Miocene have been inverted as a result of reactivation of normal faults as reverse faults during the last 2 to 3 Myr [Okamura et al., 1995]. Okamura [2002] pointed out that contraction is localized mainly in the rifts, whereas the horsts between the rifts, composed of pre-neogene rocks are scarcely deformed (Figure 1). The Okushiri ridge is one of the largest currently contracting zones in the northern part of the margin, and its southern extension is inferred to be continuous with the Awashima uplift (Figure 1) and the active fault zone along the Shinano River (Figure 2). [5] Four major earthquakes larger than M = 7.5 occurred in the twentieth century in the offshore area along the Okushiri ridge and Awashima uplift (Figure 1) [Ohtake, 1995]. Along the Shinano River, two destructive earthquakes occurred in the nineteenth century (Figure 2) [Usami, 2003]. Recent GPS measurements have shown that contraction has been focused along the Niigata-Kobe tectonic zone [Sagiya et al., 2000] and along the Shinano River in the northern part of this tectonic zone. [6] These studies indicate that active contraction has occurred for the last few million years along the Okushiri ridge, Awashima uplift, and Shinano River, and these zones are still active and have high potential for future earth- 2of14

3 Figure 2. Geologic map of the Niigata sedimentary basin and adjacent areas. ISTL is the Itoigawa- Shizuoka Tectonic Line, and SKTL is the Shibata-Koide Tectonic Line. The star indicates the epicenter of the 2004 earthquake [Kato et al., 2005a], and crosses show the inferred epicenters of historical earthquakes [Usami, 2003]. quakes. Ohtake [1995, 2002] pointed out that there are several seismic gaps along this active zone where the potential for future earthquakes may be even higher. The 2004 mid-niigata Prefecture earthquake occurred in one of these gaps. 3. Geology of the Source Area 3.1. Niigata Sedimentary Basin [7] The Niigata sedimentary basin, one of the largest rift basins in the eastern margin of Japan Sea, formed mainly in early Miocene time. The basin is about 50 km wide and 100 km long, trends NNE SSW, and is bounded by the Shibata-Koide Tectonic Line (SKTL) on the east and by the Itoigawa-Shizuoka Tectonic Line (ISTL) on the west (Figure 2). The sediments in the basin are up to 6000 m thick and have been folded during the last 2 to 3 Myr [e.g., Niigata Prefecture Government, 2000]. Suzuki et al. [1974] classified anticlines in the basin into 3 orders and showed that the first-order anticlines, which have axes longer than 30 km, tend to be separated by about 10 km. Okamura [2003] noted that the first-order anticlines in the northwestern part of the basin are fault related and lie above thrust faults that cut through the upper crust. He further suggested that other folds of similar scale in the basin are also fault related. [8] The 2004 mid-niigata Prefecture earthquake occurred in the eastern margin of the basin along one of the firstorder anticlinoriums. The anticlinorium deforms Miocene to early Pleistocene sedimentary sequences and volcanic rocks 3of14

4 Figure 3. Detailed geologic map of the Uonuma hills (simplified from a digitized geologic map of the Chuetsu area [Takeuchi et al., 2004]). [Yanagisawa et al., 1985, 1986; Kobayashi et al., 1991]. The anticlinorium forms topographic highs that are divided into Higashiyama hills to the north and Uonuma hills to the south by Uono river valley, although the two hills are structurally continuous. In this paper, we refer to the entire complex as the Uonuma hills Structure of the Uonuma Hills [9] The Uonuma hills can be subdivided into northern, middle, and southern segments as defined by structural style, as pointed out by Sato and Kato [2005]. North of N, the hills are composed of two anticlines (Figure 3). On the east is the Nigoro anticline, which trends NNE SSW, parallel to the general trend of the hills, and has a gentle symmetrical profile. The western anticline is a west vergent asymmetric anticline trending N S. The western anticline has been called the Higashiyama anticline in previous papers and geologic maps [e.g., Kobayashi et al., 1991], but we use the name North Higashiyama anticline in this paper to distinguish it from the Higashiyama anticline in the middle segment of the Uonuma hills. The western limb of the North Higashiyama anticline is steeply inclined and partly overturned at its base, and these deformation of Pleistocene sediments were attributed to an east dipping high-angle reverse fault system, the Yukyuza fault [Tsutsumi et al. 2001; Sato and Kato, 2005]. The North Higashiyama and Nigoro anticlines merge into the Higashiyama anticline in the middle segment. [10] In the middle segment of the hills, Yanagisawa et al. [1986] defined the Higashiyama anticlinorium and the 4of14

5 Figure 4. Geologic structure in the Ojiya district based on the geologic map by Yanagisawa et al. [1986]. The fold axes in the Higashiyama anticlinorium are discontinuous and often overlapping each other. In contrast, there is no structural offset in the backlimb of the anticlinorium except for a bend of strike from NNE in the north of N S in the south. The forelimb of the anticlinorium is bounded by the Suwatoge flexure, which is continuous through the anticlinorium. Horinouchi synclinorium (Figure 4). The anticlinorium is composed of two major asymmetric anticlines, the Higashiyama and Tamugiyama anticlines, and other smaller anticlines and synclines. The fold axes of them are discontinuous in the anticlinolium, however, the western and eastern limbs of the anticlinorium have no structural offset along the anticlinorium (Figure 4). The Horinouchi synclinorium is a large, gentle basin like structure including small, gentle folds. [11] The southern segment of the Uonuma hills, south of N, is characterized by a simple east vergent asymmetric anticline with a wide monoclinal western limb [Yanagisawa et al., 1985]. The width of the hills decreases to the south from 14 km to 10 km (Figure 3). [12] Two active faults and fold scarps were identified before the 2004 earthquake along the eastern margin of the middle segment (Figure 3). The Obiro fault, about 6 km long, was defined by fold scarps that deforms late Pleistocene terrace along the synclinal axis by Watanabe et al. [2001], but no fault or a flexure was observed on a highresolution seismic profile obtained by Kato et al. [2005b]. The Muikamachi fault bounds the eastern margin of the middle to southern segments of Uonuma Hills [Kim, 2004], and its uplift rate was estimated to be 2 m/ky in maximum Stratigraphy of the Uonuma Hills [13] Sediments deformed by the anticlines range in age from middle Miocene to Pleistocene [Yanagisawa et al., 1985, 1986; Kobayashi et al., 1991]. Miocene sediments are composed mainly of muddy sediments and turbidites deposited in deep marine environments. The Pliocene sediments comprise a coarsening upward sequence that thickens to the west, indicating that they were deposited on the westward deepening slope along the eastern margin of the rift basin. The Uonuma Formation consists of latest Pliocene to middle Pleistocene fluvial to shallow marine sediments. Facies changes in these strata make it difficult to correlate them across the Uonuma hills. However, the many volcanic tephras contained in the Pliocene to Pleistocene sediments provide useful key horizons for the correlation of synchronous stratigraphic surfaces in this area [Kurokawa, 1999]. The youngest strata deformed in the Uonuma hills include middle to late Pleistocene terraces [Kim, 2004], preserved in river valleys and fans. 4. The 2004 Mid-Niigata Earthquake [14] The 2004 mid-niigata Prefecture earthquake (M w = 6.6) occurred on 23 October 2004 in the southwestern part of the Higashiyama anticline [e.g., Aoki et al., 2005] and caused severe damage because of strong ground shaking [Honda et al., 2005] and numerous landslides [Yoshimi et al., 2005]. A dense seismic network was deployed in and around the source area of the earthquake from 24 October and provided precise locations of aftershocks distributed mainly in the middle segment of the Uonuma hills 5of14

6 Figure 5. Main shocks and aftershocks of the 2004 mid-niigata Prefecture earthquake [Kato et al., 2005a]. Lines indicate the locations of fault modeling. (Figure 5), several rupture surfaces related with large aftershocks, and detailed velocity structure (Figure 6) [Kato et al., 2005a, 2006; Okada et al., 2005; Sakai et al., 2005]. The main shock had a reverse-fault-type mechanism, and the aftershock distribution indicated that slip on the main shock occurred along a west dipping high-angle reverse fault (Figures 5 and 6). The aftershocks suggested a flatter, low-angle thrust at shallower level of less than about 5 km (Figure 6) [Kato et al., 2005a, 2006]. Tomographic analysis of aftershock data shows that the P wave velocity of the hanging wall is slower than that of the footwall [Kato et al., 2005a, 2006; Korenaga et al., 2005; Okada et al., 2005]. Analyses of coseismic crustal deformation [Ozawa et al., 2005] and strong ground motion records [Hikima and Koketsu, 2005; Honda et al., 2005] support the high-angle fault model. The aftershocks extended for a distance along strike of 35 km, from the northern end of the Higashiyama anticline to the southern termination of the Tamugiyama anticline (Figures 3 and 5). [15] The largest aftershock (M w = 6.3) occurred about 36 min later at the deepest end of a reverse fault parallel to and about 5 km to the east of the source fault of the main 6of14

7 Figure 6. Fault models and aftershocks on cross sections. Fault models were constructed along the cross sections from geologic structure along which Kato et al. [2005a] depicted the precise aftershock distribution and velocity structure. The locations of the cross sections are shown in Figure 5. shock (Figure 6). Other aftershock occurred on 27 October along an ESE dipping thrust fault in the footwall, which is interpreted as a fault conjugate to those of the main shock and largest aftershock (Figure 6). There were many aftershocks in the hanging wall of the source fault of the main shock, suggesting complicated deformation there [Aoki et al., 2005]. 5. Construction of Fault Models 5.1. Single-Thrust Model [16] The source area of the 2004 earthquake consists of many folds, and the main shock was followed by many aftershocks, suggesting that several faults may have caused the growth of the folds. We, however, assumed that a single reverse fault along the source fault of main shock is responsible to growth of the folds in the Uonuma hills based on following reasons. [17] Surface ruptures were found along the eastern margin of the Higashiyama Hills [Suzuki et al., 2004; Maruyama et al., 2005]. The displacement at the ground surface was less than 30 cm but the trenching survey revealed that the surface ruptures occurred repeatedly along the active fault that has slipped [Maruyama et al., 2007]. A high-resolution seismic profile presented by Kato et al. [2005b] showed that the surface rupture continues to a low-angle thrust fault down to several hundred meters in depth. The fact that the surface ruptures continues to the thrust fault indicates that coseismic slip reached the eastern margin of the hills from one of deep seismogenic faults along the main shock or one of major aftershocks. [18] Around the location of the surface rupture, two large aftershocks occurred which accompanied linear alignments of smaller aftershocks showing the source faults of the large aftershocks. The one is the largest aftershock that has a source fault parallel to the main shock and the other is an aftershock that has an east dipping source fault occurred on 27th October (Figure 6). [19] Kato et al. [2005a, 2006] have determined the precise location of aftershocks and detailed velocity structure (Figure 6) and showed that the two large aftershocks occurred in a high-velocity zone deeper than 8 km. The velocity above the upper limit of the aftershocks is higher than 5 km/s, suggesting that the shallower part has enough rigidity to generate earthquakes. These observations indicate that the slip of the two large aftershocks have not reached the surface rupture. [20] In contrast, the aftershocks along the main shock continue to the depth 2 3 km and show a bend from high to low angle around 5 km deep (Figure 6). The aftershocks above the bend probably occurred in sediments or boundary between the sediments and basement that has velocity less than 4 km/s. Kato et al. [2005b] and Sato and Kato [2005] showed that the low-angle thrust shown by the shallow seismic profile at the eastern margin of the hill can be consistently connected with the source fault of the main 7of14

8 Figure 7. Cartoon illustrating geometric model for determining fault trajectory by inclined shear of hanging wall. (a) Three key geometric elements, (b) passes of points (arrows) on the initial key horizon during the deformation of hanging wall. Fault trajectory can be constructed by connecting the arrows from top to down. (c) Fault trajectory constructed assuming lower angle (65 ) of shear planes. Note that inclination of the fault trajectory becomes steeper as the decrease of the shear angle of antithetic shear plane. shock. The results of the detailed study of aftershocks and velocity structure strongly suggest that the surface rupture were caused by the slip along the source fault of the main shock. [21] The detailed velocity structure deduced by tomographic analyses [Kato et al., 2005a, 2006] showed that the source fault of the main shock nearly coincides with the boundary between the low-velocity hanging wall and the high-velocity footwall (Figure 6). This indicates that the reverse fault on the main shock was a major normal fault bounding a rift basin. In contrast, the two aftershocks on 23 and 27 October occurred in the high-velocity footwall suggesting that their source faults were not major normal faults. Because basin inversion and fault reactivation are dominant tectonic processes in the eastern margin of Japan Sea [e.g., Sato, 1994; Okamura et al., 1995], it is reasonable to assume that the growth of the folds is mainly attributed to slip of the major reverse fault on the main shock rather than the faults related aftershocks in the high-velocity basement. [22] Geologic structure supports our single reverse fault model. The eastern and western limbs of the Higashiyama anticlinorium have no structural discontinuity along the margin, whereas the fold axes are discontinuous in the anticlinorium (Figure 4). The back limbs of the Higashiyama and Tamugiyama anticlines, which are located above the source fault, have continuous dips and strikes, indicating they are underlain by a common reverse fault. In contrast, the forelimbs of the two anticlines are discontinuous. The Suwatoge flexure, the forelimb of the Tamugiyama anticline, continues north to the forelimb of the Komatsukura anticline, one of the smaller anticlines to the east of the Higashiyama anticline. This strongly suggests that the smaller anticlines to the east of the Higashiyama anticline are also related to the source fault. These arrangements of folds and flexures can be simply explained by a single reverse fault model with lateral ramps. Gentle folds in the Horinouchi synclinorium is also able to be explained by a low-angle thrust. [23] Our single reverse fault model may not be a unique one, but reasonably explains geologic structure, aftershock distribution and the surface rupture Inclined Shear [24] The inclined shear is a simple deformation mechanism of a hanging wall above a dip-slip fault, which was presented as a vertical shear model or a constant heave model [Verrall, 1981; Gibbs, 1983] and developed to inclined shear model by White et al. [1986]. Clay and sand box experiments showed that inclined shear can approximate the deformation of hanging wall [Dula, 1991; Yamada and McClay, 2003a, 2003b]. In this model, we assume that a hanging wall is composed of many slices bounded by parallel shear plains and is deformed only by slips along the shear plains retaining the thickness of the slices (Figure 7). If we have the geometries of a key bed before and after deformation and a leading edge of a fault (Figure 7a), passes of the point (a, b,... i) on a key bed during deformation can be determined as one of arrows shown in Figure 7b. The arrows are parallel to intervals of the fault segments bounded by shear planes, then we can construct a fault trajectory from top to downward by connecting the arrows (a-a 0,b-b 0,... i-i 0 ) as shown in Figure 7b. [25] The parallel shear assumed in this model represents bulk deformation approximating real deformation [Dula, 1991, White, 1992]. The real deformation in a hanging wall is inferred to be complicated, and may have been caused by antithetic and synthetic shears [Dula, 1991]. The fold geometry constructed by incline shear modeling varies by the change of the angle of shear plain (Figure 7c), and Yamada and McClay [2003b] showed that the angle of shear plain varies by change of three dimensional fault geometry. [26] In this modeling method, geometry of a certain interval of a fault trajectory is determined by amount of a slip and geometry of a key bed before and after deformation above the fault interval between the shear planes bounding the interval (Figure 7b). The amount of slip is a key parameter to construct the fault trajectory, but it is generally 8of14

9 Figure 8. Cartoon showing the fault construction method. (a) Three key geometries for fault modeling. The uplifted area is the product of the amount of slip and the depth of detachment. (b) (top) Geologic map split from the digitized geological map along the section line and (bottom) dips of beds shown on the map projected onto the topographic profile. (c) Parallel lines drawn using the dips of beds as reference lines. The geometry of the folded key bed was built taking into account the parallel lines and the location of the key bed. Ko and Ok I are tephra key beds of nearly the same age. very difficult to estimate precise slip amount from surface geologic structure. To solve this difficulty, we assumed the depth of the fault to be 13 km from the cutoff depth of the aftershocks. If we can determine the depth of the fault end (D), we can calculate slip amount S by S = A/D, where A is an uplift area that can be measured on sections (Figure 7b). [27] If we can determine the fault depth from the depths of the aftershocks, we can fix the position of the lower end of the fault. This means that the geometry of deeper part of the fault trajectory in a seismogenic zone is mainly determined by the geometry of a backlimb. The Higashiyama and Tamugiyama anticlines have nearly constant dip in their back limb from their western margin to the anticlinal axes, indicating that the geometry of the key bed is reliable. We can construct the reliable fault trajectory in the seismogenic zone, if we can determine the fault depth and the geometry of the backlimb precisely Construction of Fault Trajectory [28] We constructed 12 serial cross sections using the method described above. The section lines were established parallel to the cross sections along which Kato et al. [2005a] depicted the precise aftershock distribution and P wave velocity, and four of them coincide with those cross sections (Figure 5). [29] For the modeling, we needed to determine three geometric elements: the geometry of folded key beds, the original geometry of the key bed before folding, and the geometry of the fault interval connecting the tips of the folded and original key beds (Figure 8a). We used commercial software for balanced cross section modeling to construct fold geometry and fault models. Geologic data were loaded from a digitized geological map [Takeuchi et al., 2004] based on quadrangle geological maps at a scale of 1:50,000 for Nagaoka [Kobayashi et al., 1991], Ojiya 9of14

10 Table 1. Parameters of Fault Modeling a Line Key Bed Age, Ma Area, km 2 Contraction, km Y = 10 NA Y = 5 NA Y = 2.5 Ko-OkI Y = 0 Ko Y= 2.5 Ko Y= 5 T Y= 7.5 Kk-Tz Y= 10 Tg Y= 12.5 N/A Y= 15 N/A Y= 17.5 N/A Y= 20 N/A a Key beds and age indicate the name of tephra and its age used for modeling of each section. N/A indicates that no tephra exists on the section and base of the Uonuma formation was used as a key bed. Area was measured between original and fold geometries of a key bed, and horizontal contraction was calculated assuming the detachment depth to be 13 km. [Yanagisawa et al., 1986], Tokamachi [Yanagisawa et al., 1985] and part of Suhara [Takahashi et al., 2004]), and part of the Sumondake area [Ijima, 1974; Kageyama and Kaneko, 1992]. [30] Topographic profiles were constructed from a 50 m digital elevation model (DEM) presented by Geographical Survey Institute of Japan. The dip and strike of beds within about 800 m from a section line were projected onto each of the topographic profiles (Figure 8b), and references lines were drawn parallel to the dip of beds projected on the section (Figure 8c). Because the thicknesses of the sedimentary units were not constant through the section, the parallel lines do not indicate synchronous surfaces. We corrected for the differences in sediment thickness by using the contained tephras. The geometry of the folded key bed was drawn taking into account the reference lines and the location of the key tephra exposure (Figure 8c). We selected the youngest tephra exposed on both forelimb and backlimb of the anticline along each of the transects as the key bed. Because there was no tephra on the southern four sections, we used the base of the Uonuma formation as a key bed (Table 1). The Nokogiriyama fault along the axis of the Higashiyama anticlines was ignored in the construction of the fold geometry because the stratigraphic offset caused by the fault is less than a few hundreds of meters, which is of little significance on the scale of the Higashiyama anticline. Original geometries of key beds were defined by a linear or simple concave-upward profile increasing in depth to the west (Figure 8c). The paleoenvironment indicated by the sedimentary facies of Pliocene and Pleistocene units supports that the western part of the hills was in a deeper environment than the eastern part. A dip of inclined shear was determined to be 85 ESE by comparison of a fault trajectory and the aftershock alignment along the section on the main shock (Y = 0 km). 6. Results and Discussion 6.1. Comparison Between Fault Models and Aftershocks on Cross Sections [31] Fault trajectories were compared with the distribution of the aftershocks along the four cross sections presented by Kato et al. [2005a] (Figure 6). The fault trajectories are composed of a high-angle segment deeper than 5 km and a low-angle segment shallower than 5 km, which is consistent with the aftershock distribution along the source fault of the main shock. Two of the sections (Y = 0 and 10 km) show good agreement between the estimated fault trajectories and the aftershock distribution, and another section (Y = 5 km) is consistent in part with the aftershock distribution. On the southernmost section (Y = 5 km), there are two cluster of aftershock distribution that does not show a west dipping plane; thus it is difficult to discuss the relationship between the fault model and the source fault. [32] The result indicates that the geometry of the source fault of the 2004 earthquake can be constructed from fold geometry assuming inclined antithetic shear of hanging wall at 85, although more detailed comparison between the fault model and aftershocks are necessary. The high-angle shear of hanging wall presumably indicates that antithetic and synthetic shears exist in the hanging wall. [33] It is not difficult to apply this method to other active fold-and-thrust zones and to construct geometry of seismogenic faults, if detailed geologic structure of ground surface and depth of the seismogenic zone in the upper crust can be determined. The constructed fault is expected to be a source fault of future earthquakes, implying that the model would provide important information to reduce damage by future earthquakes Lateral Change in Fault Geometry and Its Relationship to the 2004 Earthquake Rupture [34] We constructed eight additional fault models by the same method along sections parallel to the four cross sections discussed above (Figure 9), mainly in the southern part of the Uonuma hills (Figure 5). The fold geometry, consisting of an asymmetric anticlinorium in the source area, changes to a simple asymmetric anticline in the southern part of Uonuma hills, and the fault trajectories also become simpler as the geometry of the folds changes (Figure 9). The depth contours of the subsurface fault plane based on the 12 fault models show that the strike of the fault north of the main shock is NNE SSW; south of the main shock it bends to strike N S, and then bends again near the southern margin of the aftershock area to strike NNE SSW (Figure 10). Kato et al. [2006] showed that the NNE SSW trend of the aftershock alignment change to roughly N S from north to south around the main shock, which appears to be consistent with our fault model based on geologic structure. If our model is correct, the earthquake rupture started at the northern convex bend of the fault surface and stopped at the concave bend at its southern end. [35] We could not construct the fault around the northern margin of the source area because the northern margin may be underlain by two oppositely dipping faults. The Higashiyama anticline in the source area is continuous with the Nigoro anticline, and the aftershock distribution appears to follow the trend of the Nigoro anticline (Figures 3 and 5), suggesting that the west dipping source fault extends to the northern segment of Uonuma hills. In contrast, the Yukyuzan fault along the western margin of the North Higashiyama anticline is an east dipping reverse fault under the northern part of the source area. It is difficult to determine the relationship between the east dipping and 10 of 14

11 Figure 9. Fault models along 12 cross sections. The locations of the cross sections are shown in Figure of 14

12 Figure 10. Depth contours of the fault surface constructed by the fault modeling. The fault plane bends twice from north to south: from NNE SSW to N S and then back to NNE SSW. The main shock (star) is located on the convex northern bend, and the southern boundary of the aftershock (gray circles) area coincides with the southern concave bend. west dipping faults only from the geologic structure at the ground surface. The geometric change of the anticlines at the northern end of the source area suggests that a change in fault geometry determined the northern margin of the rupture. Sato and Kato [2005] proposed that the east vergent reverse fault in the middle segment abruptly changes into a west vergent fault around the southern margin of the northern segment; however, the gradual change in the geometry of the anticline from the source area to the Nigoro anticline indicates that the east verging fault continues to the north for some distance Slip Amount and Rate [36] On the basis of our fault models, the amounts of horizontal contraction along each of the constructed faults were estimated to be 0.4 to 1.7 km (Table 1). The contraction was largest at the Higashiyama anticline and decreased to the south. The key beds used in the models were tephras of different ages, ranging from 4.5 to 1.3 Ma, and the largest slip was deduced from the cross section based on the tephra of 4.5 Ma. This suggests that the growth of the fold started in the early Pliocene; however, the fact that structure of the Pliocene sedimentary sequences is almost uniform probably 12 of 14

13 indicates that the major growth of the folds occurred after the late Pliocene. If we assume that the growth of the anticlines began about 2 Myr ago, then the average horizontal contraction rate is calculated to be 0.2 to 0.85 m/kyr (Table 1), but it is difficult to estimate the precise time of onset of folding or the slip rate of the fault from our model. [37] Kim [2004] estimated the uplift rate of the Uonuma hills to be m/kyr in its middle segment during the last several tens of thousands of years, based on structural relief of folded middle Pleistocene fluvial terraces deposits in the middle segment of the Uonuma Hills. The uplift rate is estimated to be 1.7 times larger than contraction rate, if the fault inclination is 60. Thus the maximum contraction rate 0.85 m/kyr is consistent with the maximum uplift rate 1.3 m/kyr. This comparison supports that our fault model is valid. 7. Summary [38] The subsurface fault that are responsible for the growth of anticlines in the source region of the 2004 mid- Niigata Prefecture earthquake were constructed from fold geometries, assuming that the folds have been growing by the deformation of the hanging wall because of inclined shear above the single reverse fault. The depth of the detachment was determined to be 13 km, based on the aftershock cutoff depth. The fault trajectories agree well with the distribution of aftershocks along the fault of the main shock. The results indicate that the source fault of the earthquake is responsible to the growth of the geologic structure. The key data for the fault construction are the depth of the detachment and the fold geometry, especially of the back limb. If these data are available, it is possible to infer the geometry of source faults of future earthquakes in the active fold-and-thrust zone that is underlain by a simple thrust using inclined shear modeling. A three-dimensional fault geometry constructed the from fault trajectories along 12 section lines indicates that from north to south, the strike of the fault changes from NNE SSW to N S at the site of the main shock and from N S to NNE SSW near the southern termination of the aftershock area. The relationship suggests that the rupture started at the convex bend of the fault and stopped at the concave bend of the fault. The horizontal contraction rate of the fault was inferred to be 0.2 to 0.85 m/kyr, which is consistent with the uplift rates of the hills inferred from the elevation of fluvial terraces by Kim [2004]. [39] Acknowledgments. We are grateful to James Dolan, Karl Mueller, and Tom Pratt, reviewers of JGR, for their useful comments and suggestions, which helped greatly to improve our manuscript. We also thank Yuichi Sugiyama for his encouragement of our study and to Aitaro Koto, who kindly provided precise aftershock data. This work was supported by the Special Coordination Fund for Promotion of Science and Technology offered by the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT) under the title of Urgent Research for the 2004 Mid-Niigata Prefecture Earthquake. Some figures are created using Generic Mapping Tools (GMT). References Abe, K. (1975), Re-examination of the fault model for the Niigata earthquake of 1964, J. Phys. Earth, 23, Aoki, S., M. Nishi, N. Nakamura, T. Hashimoto, S. Yoshikawa, and H. M. Ito (2005), Multi-planar structures in the aftershock distribution of the mid Niigata Prefecture earthquake in 2004, Earth Planets Space, 57, Carena, S., and J. Suppe (2002), Three-dimensional imaging of active structure using earthquake aftershocks: The Northridge thrust, California, J. Struct. Geol., 24, Davis, T. L., and J. S. Namson (1994), A balanced cross-section of the 1994 Northridge earthquake, southern California, Nature, 372, Dula, F. W., Jr. (1991), Geometric models of listric normal faults and rollover folds, AAPG Bull., 75, Gibbs, A. D. (1983), Balanced cross-section construction from seismic sections in areas of extensional tectonics, J. Struct. Geol., 5, Hikima, K., and K. Koketsu (2005), Rupture processes of the 2004 Chuetsu (mid-niigata Prefecture) earthquake, Japan: A series of events in a complex fault system, Geophys. Res. Lett., 32, L18303, doi: / 2005GL Honda, R., S. Aoi, N. Morikawa, H. Sekiguchi, T. Kunugi, and H. Fujiwara (2005), Ground motion and rupture process of the 2004 mid Niigata Prefecture earthquake obtained from strong motion data of K-NET and KiK-net, Earth Planets Space, 57, Ijima, S. (1974), Nature of the middke Miocene unconformity on the midstream region of the R. Aburuma-gawa, Shinano-gawa river group (in Japanese with English abstract), Rep , pp , Geol. Surv. of Jpn., Tsukuba. Ishiyama, T., K. Mueller, M. Togo, A. Okada, and K. Takemura (2004), Geomorphology, kinematic history, and earthquake behavior of the active Kuwana wedge thrust anticline, central Japan, J. Geophys. Res., 109, B12408, doi: /2003jb Kageyama, K., and N. Kaneko (1992), Central part of Niigata Prefecture, Geol. Map Oil Gas Fields Jpn. 13, Geol. Surv. of Jpn., Tsukuba. Kato, A., E. Kurashimo, N. Hirata, S. Sakai, T. Iwasaki, and T. Kanazawa (2005a), Imaging the source region of the 2004 mid-niigata Prefecture earthquake and the evolution of a seismogenic thrust-related fold, Geophys. Res. Lett., 32, L07307, doi: /2005gl Kato, N., et al. (2005b), Geologic fault model based on the high-resolution seismic reflection profile and aftershock distribution associated with the 2004 mid-niigata Prefecture earthquake (M6.8), central Japan, Earth Planets Space, 57, Kato, A., S. Sakai, N. Hirata, E. Kurashimo, T. Iidaka, T. Iwasaki, and T. Kanazawa (2006), Imaging the seismic structure and stress field in the source region of the 2004 mid-niigata Prefecture earthquake: Structural zones of weakness and seismogenic stress concentration by ductile flow, J. Geophys. Res., 111, B08308, doi: /2005jb Kim, H. Y. (2004), Relationship between the upheaval process of the Uonuma Hills and the cumulative nature of the Muikamachi fault, central Japan (in Japanese with English abstract), Active Fault Res., 24, Kobayashi, I., M. Tateishi, M. Yoshioka, and M. Shimazu (1991), Geology of the Nagaoka District (in Japanese with English abstract), quadrangle series, scale 1:50,000, 132 pp., Geol. Surv. of Jpn., Tsukuba. Korenaga, M., S. Matsumoto, Y. Iio, T. Matsushima, K. Uehira, and T. Shibutani (2005), Three dimensional velocity structure around aftershock area of the 2004 mid Niigata Prefecture earthquake (M6.8) by the double-difference tomography, Earth Planets Space, 57, Kurokawa, K. (1999), Tephrostratigraphy of the Nanatani to Uonuma formation of 13 Ma to 1 Ma in the Niigata region, central Japan (in Japanese with English abstract), J. Jpn. Pet. Technol., 64, Maruyama, T., Y. Fusejima, T. Yoshioka, Y. Awata, and T. Matsu ura (2005), Characteristics of the surface rupture associated with the 2004 mid Niigata Prefecture earthquake, central Japan and their seismotectonic implications, Earth Planets Space, 57, Maruyama, T., K. Iemura, T. Azuma, T. Yoshioka, M. Sato, and R. Miyawaki (2007), Paleoseismological evidence for non-characteristic behavior of surface rupture associated with the 2004 mid-niigata Prefecture, central Japan, Tectonophysics, 429, Niigata Prefecture Government (2000), Geological map of the Niigata Prefecture (in Japanese), 200 pp., Niigata, Japan. Ohtake, M. (1995), A seismic gap in the eastern margin of the Sea of Japan as inferred from the time-space distribution of past seismicity, Island Arc, 4, Ohtake, M. (2002), Earthquake potential along the eastern margin of the Japan Sea (in Japanese), in Active Faults and Seismotectonics in the Eastern Margin of Japan Sea, edited by M. Otake et al., pp , Univ. of Tokyo Press, Tokyo. Okada, T., et al. (2005), Aftershock distribution and 3D seismic velocity structure in and around the focal area of the 2004 mid Niigata Prefecture earthquake obtained by applying double-difference tomography to dense temporary seismic network data, Earth Planets Space, 57, Okamura, Y. (2002), Deformation zone since Neogene along the eastern margin of Japan Sea (in Japanese), in Active Faults and Seismotectonics in the Eastern Margin of Japan Sea, edited by M. Otake et al., pp , Univ. of Tokyo Press, Tokyo. 13 of 14

14 Okamura, Y. (2003), Fault-related folds and an imbricate thrust system on the northwestern margin of the northern Fossa Magna region, central Japan, Island Arc, 12, Okamura, Y., and Y. Yanagisawa (2005), Relationship between the geologic structure and Chuetsu earthquake (in Japanese), Chishitsu News, 607, Okamura, Y., M. Watanabe, R. Morijiri, and M. Satoh (1995), Rifting and basin inversion in the eastern margin of the Japan Sea, Island Arc, 4, Okamura, Y., K. Satake, K. Ikehara, A. Takeuchi, and K. Arai (2005), Paleoseismology of deep-sea faults based on marine surveys of northern Okushiri ridge in the Japan Sea, J. Geophys. Res., 110, B09105, doi: /2004jb Ozawa, T., S. Nishimura, Y. Wada, and H. Ohkura (2005), Coseismic deformation of the mid Niigata Prefecture earthquake in 2004 detected by RADARSAT/InSAR, Earth Planets Space, 57, Sagiya, T., S. Miyazaki, and T. Tada (2000), Continuous GPS array and present-day crustal deformation of Japan, Pure Appl. Geophys., 157, Sakai, S., N. Hirata, A. Kato, E. Kurashimo, T. Iwasaki, and T. Kanazawa (2005), Multi-fault system of the 2004 mid-niigata Prefecture earthquake and its aftershocks, Earth Planets Space, 57, Satake, K. (1985), The mechanism of the 1983 Japan Sea earthquake as inferred from long-period surface waves and tsunamis, Phys. Earth Planet. Inter., 37, Sato, H. (1994), The relationship between late Cenozoic tectonic events and stress field and basin development in northeast Japan, J. Geophys. Res., 99, 22,261 22,274. Sato, H., and N. Kato (2005), Relationship between geologic structure and the source fault of the 2004 mid-niigata Prefecture earthquake, central Japan, Earth Planets Space, 57, Shaw, J. H., and J. Suppe (1994), Active faulting and growth folding in the eastern Santa Barbara Channel, California, Geol. Soc. Am. Bull., 106, Shaw, J. H., and J. Suppe (1996), Earthquake hazards of active blind-thrusts under the central Los Angeles basin, California, J. Geophys. Res., 101, Shaw, J. H., A. Plesch, J. F. Dolan, T. L. Pratt, and P. Fiore (2002), Puente Hills blind-thrust system, Los Angeles, California, Bull. Seismol. Soc. Am., 92, Suppe, J. (1983), Geometry and kinematics of fault-bend folding, Am. J. Sci., 283, Suzuki, Y., T. Mitsunashi, K. Kageyama, T. Shimada, M. Miyashita, and K. Kodama (1974), On the mechanism of folding in the Niigata tertiary basin (in Japanese with English abstract), Rep. Geol. Surv. Jpn., 250-2, Suzuki, Y., M. Watanabe, and D. Hirouchi (2004), Surface earthquake fault of the mid-niigata Prefecture earthquake in 2004 (in Japanese with English abstract), J. Geogr., 113, Takahashi, H., T. Toyoshima, T. Shimura, H. Hara, K. Takeuchi, A. Sakai, and S. Nakano (2004), Geology of Suhara district (in Japanese with English abstract), Quadrangle series, scale 1:50,000, 80 pp., Geol. Surv. of Jpn., Tsukuba. Takeuchi, K., Y. Yanagisawa, J. Miyazaki, and M. Ozaki (2004), Digital geological map of the Chuetsu-Uonuma area at scale 1 to 50,000 (in Japanese), Geol. Surv. Jpn. Open File Rep., 412. Tanioka, Y., K. Satake, and L. Ruff (1995), Total analysis of the 1993 Hokkaido Nansei-oki earthquake using seismic wave, tsunami, and geodetic data, Geophys. Res. Lett., 22, Tsutsumi, H., M. Yogo, M. Watanabe, Y. Kim, and N. Sato (2001), Active fault map in urban area: Nagaoka (in Japanese), Tech. Rep. D1 388, Geogr. Surv. Inst., Tsukuba. Usami, T. (2003), Materials for Comprehensive List of Destructive Earthquakes in Japan (in Japanese), 605 pp., Tokyo of Univ. Press, Tokyo. Verrall, P. (1981), Structural interpretation with application to North Sea problems, Rep. 3, 156 pp., Jt. Assoc. Pet. Explor. Courses, London. Watanabe, M., H. Tsutsumi, Y. Suzuki, Y. Kim, and N. Sato (2001), Active fault map in urban area: Ojiya (in Japanese), Tech. Rep. D.1, 388, Geogr. Surv. Inst., Tsukuba. Wei, D., and T. Seno (1998), Determination of the Amurian plate motion, in Mantle Dynamics and Plate Interactions in East Asia, Geodyn. Ser., vol. 27, pp , AGU, Washington, D. C. White, N. J. (1992), A model for automatically determining normal fault geometry, J. Geophys. Res., 97, White, N. J., J. A. Jackson, and McKenzie (1986), The relationship between the geometry of normal fault and that of the sedimentary layers in their hanging walls, J. Struct Geol., 8, Yamada, Y., and K. McClay (2003a), Application of geometric models to inverted listric fault systems in sandbox experiments. paper 1: 2D hanging wall deformation and section restoration, J. Struct Geol., 25, Yamada, Y., and K. McClay (2003b), Application of geometric models to inverted listric fault systems in sandbox experiments. paper 2: Insights for possible along strike migration of material during 3D hanging wall deformation, J. Struct Geol., 25, Yanagisawa, Y., Y. Chihara, Y. Suzuki, T. Uemura, K. Kodama, and T. Kato (1985), Geology of Tokamachi district (in Japanese with English abstract), quadrangle series, scale 1:50,000, 104 pp., Geol. Surv. of Jpn., Tsukuba. Yanagisawa, Y., K. Kobayashi, K. Takeuchi, M. Tateishi, K. Chihara, and K. Kato (1986), Geology of Ojiya district (in Japanese with English abstract), quadrangle series, scale 1:50,000, 177 pp., Geol. Surv. of Jpn., Tsukuba. Yoshimi, M., et al. (2005), Landform and local site effects on the colonies worst-hit by the 2004 mid-niigata Prefecture, earthquake, Japan, Earth Planet Space, 57, T. Ishiyama and Y. Okamura, Active Fault Research Center, C7, AIST, 1-1-1, Higashi, Tsukuba, Ibaraki , Japan. (okamura-y@aist.go.jp) Y. Yanagisawa, Institute of Geology and Geoinformation, C7, AIST, 1-1-1, Higashi, Tsukuba, Ibaraki , Japan. 14 of 14