The Tokyo metropolitan area, underlain by Neogene and Quaternary sediments more than 5 km thick, is currently deformed by blind thrusts that could generate hazardous earthquakes. However, their little geomorphic expression and dense urbanization make understanding of folds produced above them and recent deformation highly elusive. Here we show subsurface geometries of several active blind thrusts beneath this highly urbanized area, based on tectonic landforms, high-resolution seismic reflection data, and Quaternary stratigraphy. Deep seismic reflection profiles corroborate the notion that steeply dipping blind thrusts are reactivated normal faults originally formed by middle Miocene extensional tectonics. Despite very slow (less than 0.1 mm/yr) late Quaternary slip rates, our work suggests the presence of previously unrecognized faults that pose seismic hazards to Tokyo and outlying communities, highlighting the need for additional information to define recent slip rates, magnitude, and recurrence of past earthquakes on them.
 Metropolitan Tokyo is one of the largest and most densely urbanized cities in the world, with a population of more than 35 million people. Seismic hazards that pose a risk to Tokyo include faults that lie in the Kanto Basin, as well as subduction zones between the Eurasian, Pacific, and Philippine Sea plates [Wu et al., 2007]. The region has repeatedly suffered earthquakes in historic time, including the devastating 1923 Kanto earthquake (M 7.9) [Sato et al., 2005]. Although seismic hazards are dominated by great earthquakes on the subduction megathrusts, complex intraplate strain is also accommodated by active faults formed in response to subduction processes. These faults are associated with relatively smaller structures, yet their proximity or location near Tokyo makes them disproportionally more hazardous. In addition, active structures in the Kanto Basin are typically blind with little expression at the ground surface, making understanding of recent slip histories across them elusive.
 Active thrust faults and folds are mainly identified at the ground surface near the southwestern Kanto Basin by minor deformation of late Pleistocene marine and fluvial terraces [Kaizuka, 1987; Sugiyama et al., 1997] (Figure 1), with much greater displacement across them being hidden at depth. As blind structures however, the subsurface geometry of the thrusts within the region is poorly understood, and this is the main focus of the work presented in this paper. Earthquake scenarios associated with intraplate earthquakes in this region are poorly constrained because these depend in part on knowledge of fault length as well as magnitude-frequency relationships from paleoseismic studies. Deformation in the shallow subsurface is typically defined by folding as opposed to surface fault ruptures, consistent with fault-related fold theory where strain is accommodated by flexural slip in sediments above the upwardly propagating fault tips [Niño et al., 1998; Roering et al., 1997]. A challenge posed by characterizing surface folding is the densely urbanized metropolitan area itself, where already subtle tectonic landforms become more difficult to identify and measure strains across. In spite of this, significant Coulomb stress changes in this area suggested by increased seismicity after the Mw 9.0 in the 2011 Tohoku-oki earthquake [Ishibe et al., 2011] may bring even moderate (Mw ~7.0) but devastating earthquakes generated by blind active faults closer to failure, which would cause damage approaching to that of a much larger earthquake on the subduction interface. The economic impact of such a seismic event would be profound and include great loss of infrastructure in parts of the Tokyo metropolitan area, and surrounding regions, with economic effects extending outward on a global scale.
 In this study, we present new structural models and interpretations based on new seismic reflection data that shed light on the subsurface geometries and tectonic origins of active structures in the Kanto Basin. Collected over the past decade, these deep seismic reflection profiles have successfully overcome the difficulties described above and defined a number of active blind structures and identified previously unrecognized faults (Figure 1). The seismic reflection data consists of profiles that imaged the entire stratigraphic section deposited in the basin (Figures 2 and S2). Additional high-resolution shallow seismic reflection profile image late Quaternary strata to depths of about 1 km, revealing remarkably clear compressive growth strata (Figure 3) deposited across the forelimb of the Ayasegawa fault. Shallow borehole data (Figure 4) are also used to characterize displacement across the fold formed above the fault to define more recent rates of displacement across it.
2 Data Set and Methodology
 Data used in this study include over 1300 km of deep seismic reflection profiles, acquired over the last decade to resolve the geometry and history of displacement on deeply buried active faults and folds [Sato et al., 2010]. Seismic sources used to acquire the profiles were provided by Vibroseis trucks, and average common depth point (CDP) intervals are 25–50 m. The quality of the data is significantly affected by urban noise, but in general, the profiles image structures to a depth of about 5–6 km (see Table S1 for acquisition and processing informations) [Sato et al., 2006].
 The Pliocene to Pleistocene Kanto Basin is more than 100 km wide and contains sedimentary fill deposited in a variety of sedimentary environments. In the western portion of the basin, sediments are mainly composed of alternating shallow marine and fluvial sequences, whereas in the east, sediments were deposited as deep-sea turbidites in the fore arc of the subduction zone between the Pacific and Eurasian plates. In spite of the large difference in depositional environments across the basin, stratigraphic correlation is well constrained by the presence of numerous volcanic tephra (Figure S1). These tephra are widely identified and correlated throughout this region and have been partly dated with fission track dating [Geological Society of Japan (Eds.), 2008]. Further dating and correlation of Neogene strata in the Kanto Basin is provided from magnetic reversals in cores and outcrops and calcareous nannofossil biostratigraphy, which are correlated with marine eustatic records [e.g., Geological Society of Japan (Eds.), 2008].
 We tied reflectors in the seismic reflection profiles with well-established Neogene stratigraphic units exposed in the Boso and Miura Peninsulas and in the Tama Hills (Figure S1). The reflectors were also cross-correlated and loop-tied through the grid of seismic profiles used in our analysis. Strata defined by cores recovered from boreholes were also correlated with Neogene stratigraphy.
3 Blind Fault and Fold Architecture in the Kanto Basin
 Interpretation of Line A suggests that the section images an array of extensional half grabens filled with lower to middle Miocene sediments [Takahashi et al., 2006] (Figures 1 and S2). An enlargement of the eastern portion of the section shown in Figure 2a illustrates a west dipping normal fault and half graben. In this area, Miocene to late Pliocene, synextensional and postextensional sediments are gently folded upward and define a west dipping backlimb and more steeply east dipping forelimb, typical of an asymmetric fault-related fold. In addition, the lower to middle Pleistocene strata become thinner over the crest of the anticline, suggesting reactivation of the west dipping normal fault as a reverse fault and structural growth of the west dipping normal fault as a thrust, consistent with compressive growth strata deposited above an actively growing fold [Hardy and Ford, 1997]. Uplifted late Pleistocene fluvial terraces preserved on the crest of the fold were first recognized by Sugiyama et al. . This fold is interpreted as recording recent movement on the underlying reactivated, blind thrust. Marine sediments that correlate with the last interglacial period have been recovered from boreholes that penetrate the west dipping backlimb, and these strata are also folded and dip steeply [Nakazawa and Tanabe, 2011], suggesting recent structural growth of the compressive anticline.
 Similar reactivated structures can be seen in the center of seismic section A shown in Figure 2c. These structures include two west dipping normal faults that contain a thick sequence of Miocene to late Pliocene strata in their hanging walls. Folding of the uppermost part of the synextensional stratigraphic sequence is evident and is consistent with inversion of slip on the normal faults as thrust faults in the Pleistocene. The eastern half of the pair of reactivated normal faults seen (Figure 2c) is overlain by incised fluvial terraces that are separated from flat-lying deposits in the alluvial plain across the synclinal axial surface formed above the thrust.
 A fourth structure, named the Ayasegawa fault, is also associated with an overlying, asymmetric fold that deforms Miocene to Pleistocene strata (Figures 2b and 3). The fold is readily recognized at the ground surface by anticlinal folding of late Pleistocene terraces [Kaizuka, 1987; Ishiyama et al., 2005]. An east facing fold scarp at the surface coincides with the base of the forelimb defined by the folded late Pleistocene marine and fluvial terraces, suggesting recent displacement across it. We interpret this active blind thrust to be a reactivated antithetic normal fault rather than a master fault, given that units on its hanging wall are apparently thinner than on its footwall (see Figures S2b and S2c).
4 Discussion and Summary
 We identified an array of normal faults reactivated as blind thrusts beneath the northwestern portion of the Kanto Basin. Imaging of late Quaternary growth strata suggests that structural inversion began after the end of Pliocene time and that shortening across the thrusts was accommodated by folding in overlying postextensional Pleistocene deposits. The geometry of folded growth strata deposited above the rising folds is consistent with fault propagation folding, where younger strata dip progressively less and hence accommodate lesser amounts of shear across the underlying thrust [Allmendinger, 1998; Hardy and Ford, 1997].
 Growth strata deformed across the folds range in age from middle to late Pleistocene, suggesting that they have been active for this period. Moreover, the synclinal axial surface at the base of the forelimb of at least one fold (i.e., the Ayasegawa fault) coincides with displacement of a surface fold scarp that deforms shallow marine and fluvial terrace deposits, as defined by high-resolution P wave shallow seismic reflection data with 2.5 m CDP intervals (Figure 3; see data acquisition and processing parameters in Table S2) and shallow boreholes drilled along the seismic line [Ishiyama et al., 2005; Yamaguchi et al., 2009] (Figure 4). Shallow boreholes acquired across the fold scarp define about 8 m of structural relief on the top of the shallow marine sediments, correlated with marine oxygen isotope stage 5e (circa 125 ka) based on tephrostratigraphy. This yields a vertical slip rate of about 0.07 mm/yr on the underlying blind thrust. The upper surface of the shallow marine sediments on the hanging wall is separated from overlying fluvial deposits with unconformity. We thus suggest that the actual vertical slip rate across the thrust-related fold may be slightly larger than 0.07 mm/yr.
 Other thrusts discussed in this study appear to slip at lower rates, based on little or no evidence for surface deformation on them and relatively less displacement shown in older strata in the seismic reflection data (Figures 2a, 2b, 2c, and S2). Although more detailed information is required to determine whether these structures deform late Pleistocene strata, it seems prudent to consider them as a source of large earthquakes that might occur more infrequently than those on the Ayasegawa fault.
 In summary, our work suggests the presence of previously unrecognized faults in the Kanto Basin that pose seismic hazards to Tokyo and outlying communities. Further work aimed at additional shallow imaging of these faults, shallow borehole transects, and geomorphic analyses of active folds based on lidar data may shed more light on the late Quaternary rates of slip across them and the magnitude and recurrence of recent earthquakes. These are likely to be the only indicators of the history of recent earthquakes on these elusive blind thrust faults and their otherwise inaccessible seismic hazards, similar to other active structures in such highly urbanized areas as Los Angeles [Dolan et al., 2003; Shaw et al., 2002; Pratt et al., 2002]. Further construction of structural fault models in this region based on subsurface fault geometry, similar to the Community Fault Model in Southern California [Plesch et al., 2007], will also be significant in that they would be useful not only for predicting strong ground motions and assessing probabilistic seismic hazards but also calculating Coulomb stress transfer associated with coseismic slip, afterslip, and viscous relaxation of the 2011 Tohoku-oki event to active faults in this region, as exemplified by studies along other plate boundaries [Freed et al., 2007; Ali and Freed, 2010].
 This work has been financially supported by the Special Project for Earthquake Disaster Mitigation in Urban Areas of Japan's Ministry of Education, Culture, Sports, Science, and Technology. Our thanks also go to Tanio Ito and David Okaya for helpful discussions and inspiration that contribute a lot to this article.
 The Editor thanks Karl Mueller and an anonymous reviewer for their assistance in evaluating this paper.