P wave velocity structure in the Yucca Mountain, Nevada, region

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112,, doi: /2007jb005005, 2007 P wave velocity structure in the Yucca Mountain, Nevada, region Leiph Preston, 1 Ken Smith, 1 and David von Seggern 1 Received 20 February 2007; revised 18 July 2007; accepted 13 August 2007; published 8 November [1] We have performed a crustal tomographic inversion using over 250,000 P arrival times from local earthquake sources and surface explosions in the Yucca Mountain, Nevada, region. Within the shallowest 2 3 km, topographic features tend to dominate the structure with high velocities imaged under Bare Mountain, the Funeral Mountains, and higher terrain to the east of Yucca Mountain and low velocities imaged under Crater Flat, Jackass Flat, the Amargosa Desert, and the caldera complexes. Imaged shallow velocities also show correlation with several known gravity and aeromagnetic anomalies. Below the basins (2 3 km depth), velocities vary between 5.5 and 6.5 km/s and lose many of the correlations seen in the shallowest layers; however, a few major structures, such as the Bare Mountain block, can be traced to at least 10 km depth. Additionally, we image structures that may be associated with the Wahmonie intrusion and pre-tertiary structural trends. Yucca Mountain itself is underlain by a high-velocity upper crustal-scale structure similar to other structures in the region such as Bare Mountain and may represent a Basin and Range style back-tilted block, which may provide a structural explanation for Yucca Mountain s topographic expression. Additionally, the imaged, relatively low velocity basement under Crater Flat may provide a preferred conduit for magma intrusion into Crater Flat compared to Yucca Mountain, accounting for the lack of post-miocene volcanism observed at the mountain proper. We explore our tomographic results in the context of four major tectonic models that have been proposed for the Yucca Mountain region. Citation: Preston, L., K. Smith, and D. von Seggern (2007), P wave velocity structure in the Yucca Mountain, Nevada, region, J. Geophys. Res., 112,, doi: /2007jb Introduction [2] Yucca Mountain (YM) has evoked over 2 decades of extensive geological and geophysical research following its designation as a potential national high-level nuclear waste repository site. Despite advances in understanding many aspects of the mountain s geology, a consensus on a nearregional tectonic model remains elusive and several important questions still remain unanswered. Several tectonic models have been proposed (e.g., detachment mechanisms [Scott, 1990], a regional rift system [Carr, 1990], a half graben structure [Fridrich, 1999], and dextral shear on a local fault system [Schweickert and Lahren, 1997]), but mostly because of the lack of upper crustal information, little hard evidence is available to test these models. Also, these models lack a clear physical explanation for several important questions which include the following: Why is YM topographically elevated? Why is YM observed to have very little seismicity compared to surrounding regions? Why are no post-miocene volcanics observed within the Yucca Mountain block itself? These questions have important implications for evaluating the seismic and volcanic 1 Nevada Seismological Laboratory, University of Nevada, Reno, Nevada, USA. Copyright 2007 by the American Geophysical Union /07/2007JB005005$09.00 hazard of the proposed repository. We have developed for the YM region a three-dimensional (3-D) P wave velocity model, which images the critical upper 10 km of the crust in the Yucca Mountain region. Our results shed light on the answers to these important questions and provide important 3-D structural information necessary for the evaluation and development of tectonic models of the region. 2. Geology and Tectonic Models [3] YM (Figures 1 3) is situated within the southern Walker Lane Belt [Stewart, 1988], an active tectonic regime primarily defined by the NW striking, high-angle, strike-slip faults of the Eastern California Shear Zone (ECSZ) with subsidiary, low slip rate, normal and NE striking, leftlateral, strike-slip systems at the latitude of YM. The Furnace Creek Fault of the ECSZ, about 50 km west of YM with a slip rate of 4 5 mm/a and a recurrence interval of 5000 a [Klinger and Piety, 2000], is the most prominent late Quaternary regional fault system with seismic hazard implications for YM. In contrast, late Quaternary slip rates on YM area faults are about a factor of 400 less, and recurrence intervals of major earthquakes are on the order of 10 5 a reflecting the low strain rates of YM region faults [Scott, 1990]. YM is a structurally high Miocene tuff sequence between the Crater Flat Basin on the west and the Jackass Flat Basin to the east. The Crater Flat Basin, a 1of16

2 Figure 1. Map of the tomographic inversion region. Network stations are shown as red triangles. Important regional structures are labeled; YM, Yucca Mountain. Nevada Test Site is outlined in solid black. Focus region for most of the discussion is within dashed rectangle (Figure 2). fault-controlled basin bounded on the west by the Bare Mountain Fault, is composed of interbedded tuffs, post- Miocene basalt flows, and Quaternary alluvium, punctuated by several Plio-Pleistocene volcanic cones and basalt flows [O Leary et al., 2002; Vaniman et al., 1982]. These smallvolume basaltic sources represent waning post-miocene regional volcanism [Crowe et al., 1995]. North of YM the overlapping Timber Mountain, Oasis Valley, and Silent Canyon caldera complexes are the source regions of the voluminous Miocene tuff sequence found at YM. [4] Since all depths discussed in our tomography results will be relative to sea level, a short discussion of the topography of the area is warranted. Death Valley, about 50 km west of YM, has some elevations just below sea level. North of the Funeral Mountains (elevations 1500 m), elevations generally increase from south to north, with the southern Amargosa Desert having an elevation of 250 m increasing to 1250 m basin elevation for Yucca Flat and over 2000 m elevation for Pahute Mesa in the north. Local to Yucca Mountain, Crater Flat and Jackass Flat have elevations of 1000 m; Bare Mountain peaks at 2000 m; the Yucca Mountain crest averages about 1500 m, increasing from south to north where it approaches 2000 m; Timber Mountain peaks at 2250 m; Calico Hills and Shoshone Mountain reach approximately 2000 m; the Specter Range, Little Skull Mountain, Skull Mountain and higher terrain to the east of Jackass Flat reach to between 1500 and 2000 m. [5] Tectonic models for the YM-Crater Flat structures have invoked low-angle, normal-fault mechanisms [Scott, 1990], a local caldera within a regional rift system [Carr, 1990], a broader half graben structure [Fridrich, 1999], and dextral shear on an approximately N-S striking fault system through Crater Flat [Schweickert and Lahren, 1997]. As a prelude to presenting our tomographic work, we summarize each of these models, indicate primary observations that support each, and surmise what structures would be present to evaluate each with respect to the tomographic model. [6] Scott [1990] proposed the existence of a series of low-angle normal faults (the shallowest at 1 4 km beneath YM) based on observations from YM fault densities and geometries, on the local stress field, and on the presence of interpreted detachment structures in the vicinity of YM. This detachment system is viewed as accommodating relatively high Miocene extension. In conjunction with mid-miocene uplift of Bare Mountain, Scott [1990] suggests that the shallowest section of the detachment system under YM was interrupted, effectively isolating YM from regional tectonism. Consequently, local YM west dipping normal faults would be limited in depth extent by a subhorizontal shallow detachment system under the mountain. Since detachment faults would be parallel to expected isovelocity surfaces (subhorizontal), this is a difficult model to assess with tomography. Because of this limitation, we cannot refute the detachment model; however, we may be able to provide indirect evidence if we image subhorizontal offsets across vertical structures. 2of16

3 Figure 2. Base map of the Yucca Mountain focus region with shaded relief, simplified geology and labeled features: YM, Yucca Mountain; CF, Crater Flat; JF, Jackass Flat; BM, Bare Mountain; BMF, Bare Mountain Fault; BH, Bullfrog Hills; CH, Calico Hills; SP, Shoshone Peak; MV, Mid Valley; YF, Yucca Flat; CP, CP Hills; MMF, Mine Mountain Fault; LSM, Little Skull Mountain; SM, Skull Mountain; RV, Rock Valley; FF, Frenchman Flat; SR, Specter Range; SH, Striped Hills; GF, Gravity Fault; AV, Amargosa Valley; W, Wahmonie; FM, Funeral Mountains; TM, Timber Mountain; TMOVCC, Timber Mountain Oasis Valley Caldera Complex; PM, Pahute Mesa; SC, Silent Canyon Caldera. Thin black lines indicate schematic locations of the Bare Mountain Fault, Gravity Fault, and Mine Mountain Fault. These will be shown on velocity depth section maps for reference. Key indicates meaning of remaining symbols. The NPE shot (black circle) is just NE of SC. Blue E-W line indicates the position of cross section for Figures 7 and 9. Dashed rectangle demarks area of Figure 3. [7] Carr [1990] spatially associated Miocene deformation, caldera complexes, and Quaternary volcanic centers with what he interpreted to be the north-south oriented Kawich-Greenwater rift, bounded on the west by Bare Mountain and in eastern Amargosa Valley and in Jackass Flat by a gravity (Figure 4), aeromagnetic (Figure 5), and seismogenically defined structure. The rift is envisioned as a north-south striking zone of crustal and/or upper mantle weakness acting as a focus of regional Miocene extension. Within the rift, high-angle normal faults are simply gravitational break-away structures that most likely extend at least to the brittle-ductile transition, at roughly 15 km. Primary normal-fault break-away structures would dip east on the western margin of the rift and west along the eastern boundary. As a consequence of its location within the rift system and considering that YM s physiography is dominantly a remnant of Miocene deformation, it can be viewed as isolated, to some degree, from regional-scale post- Miocene tectonism and more directly affected by deformation processes within the rift itself. On the basis of this model, we would expect to see differences in structure between regions within the rift and those outside of it; structural elements within the rift would tend to be vertically extensive. Although emphasizing tectonic rather than volcanic processes, Wright [1989] also proposed a rift model that is generally spatially coincident with the Kawich- Greenwater rift and that he termed the Amargosa Desert Rift Zone. [8] Fridrich [1999] defined the Crater Flat Domain (CFD) as a structural and tectonic unit consisting of Crater 3of16

4 expression of this fault system into Crater Flat and northern Amargosa Desert, they interpret it to extend from at least Crater Flat SE into southern Amargosa Desert, where it connects to known fault systems, such as State Line and Stewart Valley fault systems. Recent geodetic observations do support local shear in the vicinity of YM, but are unable to isolate it to any particular localized fault zone [Hill and Blewitt, 2006; Wernicke et al., 2004]. According to this model, km of shear has occurred on this deep-seated fault zone; thus the tomography may be expected to image large horizontal offsets of deep structures or other signs of structural discontinuity near the fault zone. Figure 3. Close-up view of Yucca Mountain. See legend in Figure 2 for meaning of symbols. Red line, YM tunnel; FR, Fran Ridge; p#1, UE-25p#1 borehole. Flat, YM, and Jackass Flat as an asymmetric graben. The Bare Mountain fault is the primary down-to-the-east master fault bounding the half graben along Bare Mountain in western Crater Flat. This domain was proposed based on surrounding known or inferred faults, the area of total alluvial cover, and internal similarities in fault geometries and extensional history relative to surrounding areas. Fridrich characterizes the CFD as a hybrid structure produced by varying degrees of extension and dextral shear during its history. According to this model, the CFD may as a whole appear as a unit somewhat independent of surrounding domains. The predicted depth extent of the structures is not directly addressed by the model, but most likely major faults would be expected to be throughgoing while smaller faults would be antithetic to these major faults. In this model, YM is viewed as structurally linked to the CFD as a whole; thus YM may lack underlying structural expression independent of the CFD. [9] Schweickert and Lahren [1997] used gravity anomalies, fault geometries, alignments of volcanic centers and springs, and paleomagnetic evidence of tuff sheet rotation to infer the existence of a throughgoing right-lateral, strike-slip fault system. Although there is debate regarding the surface 3. Data, Method, and Modeling [10] Seismic tomography has the benefit of being able to characterize earth structure which is beyond direct measurement, and it has been used in many areas to aid in geologic and tectonic interpretation [e.g., van Wagoner et al., 2002]. The velocity structure in a region can be inferred through a tomographic inversion of P wave travel times from earthquakes and controlled sources. The resolution of the tomographic model is based on the ray coverage in a given area. Ray coverage varies significantly within the YM model region and is determined by the distribution of sources and receivers. Structure from a tomographic viewpoint is simply the distribution of seismic velocities implied by the observed traveltime data. [11] Because many materials can have similar seismic velocities and the velocity even of a single material may vary dramatically depending on its condition (weathered, fractured, unfractured, depth of emplacement, etc.), it is necessary to use other data, if available, to place constraints on material properties. Measured seismic velocities of common materials found in the Yucca Mountain region can be grouped into two categories. Cenozoic rocks, such as the tuffs that overlay Yucca Mountain, typically have P wave velocities <5 km/s, whereas pre-cenozoic rocks have velocities in excess of 5 km/s, with many rock types >5.5 km/s [Oliver and Ponce, 1995]. Seismic velocities in rocks also tend to increase as a function of depth due primarily to closure of fractures, so the velocities at depth will be higher than cited. Since Cenozoic rocks comprise the bulk of the basin fill material, we use the 5.5 km/s isovelocity surface as a proxy for depth to basement. Additionally, velocities will be lower in fractured, presumably weaker rocks compared with unfractured rocks of the same material. [12] The three-dimensional (3-D) P wave velocity model for the YM region was developed with a nonlinear inversion procedure [Preston et al., 2003], an iterative technique that simultaneously solves for optimal earthquake locations and 3-D P wave velocity structure from P wave arrival times. The solution is regularized by seeking a smooth velocity structure. Full 3-D ray-tracing and traveltime calculations are implemented through the Vidale-Hole algorithm [Vidale, 1990; Hole and Zelt, 1995]. The model region is defined to lie between 36.0 N and 37.5 N, W and W (Figure 1), at 6 km to 45 km depth relative to sea level with 2-km horizontal and 1-km vertical node spacing. Our study area, the region displayed in the depth sections, is a subset of the model region defined by 36.4 to 37.3 N, 4of16

5 Figure 4. USGS compilation of regional Bouguer gravity [Saltus and Jachens, 1995] draped on shaded relief USGS 80 m DEM with earthquake locations (no magnitudes shown, but ranging from 1.0 to 5.8) from the University of Nevada Reno catalog. Shown in dashed lines is interpreted extent of the Kawich-Greenwater gravity-defined rift proposed by Carr [1990] and reinterpreted here with respect to local seismicity and gravity gradients (roughly the 125 mgal gravity contour). The interpreted eastern boundary of the gravity trough is constrained by the continuity of the gravity gradient defined by Quaternary faults in eastern Amargosa Valley, the gravity fault in Jackass Flat, and the western physiographic extent of the Timber Mountain Oasis Valley Caldera complex to W (Figure 2). The starting model consists of the original earthquake catalog locations and a smoothed version of the 1-D velocity model currently used in routine event processing [Hoffman and Mooney, 1984] in the YM region. However, tests indicate practically no starting model dependence. The initial traveltime errors with respect to this model had a root-mean-square (RMS) error of 0.17 s. Following 3-D inversion, the RMS residual dropped to 0.06 s, amounting to a variance reduction of 88% Data [13] We have incorporated P wave arrival times from 19,125 earthquakes (250,000 phases), and 120 explosion sources (including 78 underground nuclear tests) recorded at 89 regional seismograph stations and 155 temporary stations in a traveltime tomographic inversion for the 3-D crustal velocity structure in the Yucca Mountain Nevada Test Site (NTS) region (Figures 1 3). These data include travel times from earthquakes recorded on the analog U.S. Geological Survey (USGS) network from 1981 to 1992, the same network operated by University of Nevada, Reno (UNR), from 1992 to 1995, and the combined analog and digital networks from 1995 to Travel times were collected from underground nuclear tests (S. Harmsen, personal communication, 1993) and from three major USGS deployments in [Hoffman and Mooney, 1984; Sutton, 1984, 1985] (T. M. Brocher, personal communication, 2005), and from the 1993 Non-Proliferation Experiment (NPE) in northeastern NTS [Smith et al., 2000]. These data were screened for outliers, and such data were removed. [14] Earthquake sources primarily provide constraints for the volume within the seismogenic zone and thus only weakly contribute to controlling structure in the upper 1 2 km of the crust, where most ray paths are nearly vertical to the regional network seismograph stations. Hypocentral parameters are treated as unknown and are resolved in the 5of16

6 Figure 5. Total intensity aeromagnetic map [Ponce and Blakely, 2001] of the YM region coded according to the color bar at the lower left. Region outlined by thick dashed line is the broad magnetic anomaly noted by Carr [1984]. The east-west dash-dotted line demarks the east-west magnetic gradient noted in the text. inversion, resulting in a trade-off between earthquake location and model velocities. Controlled sources, on the other hand, generally have known source locations and origin times, eliminating this trade-off. Also, controlled sources, such as the NPE and USGS experiments used in this study, often include portable instrument deployments that augment earthquake monitoring station coverage, thus enabling higher spatial resolution images from recordings at near source-receiver distances that sample the shallowest 1 2 km. The high-resolution images of Crater Flat, YM, and Jackass Flat in the upper 1 2 km are constrained by the controlled source data Resolution and Error Analysis [15] Model resolution, in general, is limited by several factors, most notably the model node spacing and the degree of ray coverage in a given area, which varies greatly from region to region. We have performed resolution tests to assess our ability to interpret the structures observed in the YM tomographic model. We perform this test by adding ±5% velocity perturbations to the final 3-D model in a regular checkerboard-like pattern. We test 6 6 km scale boxes in the horizontal dimension. We then calculate theoretical travel times through the perturbed model, add noise with the same distribution as the actual data, and use these times as observed times. The model is then inverted with these new times to see how we replicate the original input perturbation pattern. In general, horizontal resolution is best between 1 and 5 km depth and degrades rapidly with depth below 5 km. The 6 km block checkerboard tests (Figure 6a) indicate that the 0 km depth section is reasonably well resolved primarily east of the YM crest to west of Frenchman Flat (see Figures 1 3 for place names), south to the Specter Range and north to the expression of the resurgent dome of the Timber Mountain caldera, with some isolated, highly well resolved areas, notably over Bare Mountain and northern Yucca Flat. The 1 5 km depth sections show very well resolved areas from Bare Mountain to Frenchman Flat and the Specter Range to the Silent Canyon Caldera region under Pahute Mesa to the north. The areal extent of well-resolved regions in the model decreases significantly below 5 km depth. By 10 km depth, only the regions directly around the Little Skull Mountain aftershock sequence and under northern YM and Timber Mountain are 6of16

7 Figure 6a. Depth sections through the 6 6 km checkerboard test perturbed P wave model labeled according to depth relative to sea level over shaded relief. Fractional P wave velocity perturbations coded according to color bar below each panel. Grey regions are areas unsampled by rays. NTS outline shown in white, ESF tunnel in red, and faults of Figure 2 as black lines. Scale bar is in kilometers. well resolved. We also performed 12 km checkerboard tests (Figure 6b). Resolution is excellent at this scale for nearly the entire study region from 1 to 8 km depth. The 0-, 9-, and 10-km sections show degraded resolution SW of a line approximately running just west of Bare Mountain to the southern tip of YM to the SE portion of the Amargosa Desert. For the vertical resolution checkerboard tests, we used 2 km vertical by 6 km horizontal blocks at ±5% velocity perturbation (Figure 7). Similar to the horizontal resolution images, the best resolved depths are between 1 and 5 km depth, with degrading resolution, especially in the horizontal direction, below this depth. Directly under YM, vertical resolution is good to fair down to 9 km depth, whereas west of Bare Mountain, vertical resolution becomes poor. Resolution degrades shallower than 1 km depth because rays are primarily traveling vertically with few crossing rays, causing vertical smearing and good horizontal resolution only where stations or surface explosions are closely spaced. 7of16

8 Figure 6b. Same as Figure 6a, except the results of the 12 km block checkerboard tests are shown. [16] Because we are unable to directly access the error matrix due to the size of the inverse problem, we must use indirect methods to estimate the standard errors. The jackknife test is a statistical tool that we can use to derive these estimates. In this test, we randomly remove 10% of the travel times and invert this data subset. This is performed 10 times so that all of the data are eventually removed. The resulting 10 models are analyzed using the standard jackknife equations [Wonnacott and Wonnacott, 1985, p. 250] to yield standard error estimates at all nodes. On the basis of the jackknife tests, we find that, between 1 and 10 km depth, standard errors are generally between 0.1 and 0.2 km/s with no clear association with modeled structures except that higher velocity regions tend to show larger errors than lower velocity regions. At 1 km depth and shallower, standard error estimates also tend to be 0.15 km/s, but with the notable exception of the velocity hot spots (see Figure 8a) where shallow velocities exceed about 6 km/s. In these areas, standard error estimates are >0.5 km/s, which indicates that all of these areas could be 6 km/s, within the error. YM itself generally has standard errors between 0.1 and 0.2 km/s at all depth levels, with isolated nodes reaching 0.3 km/s error. 4. Model Results [17] Various slices through the tomographic model are presented in Figures Depth sections at 0 3 km depth 8of16

9 Figure 7. Cross section through the 6 km horizontal by 2 km vertical checkerboard test at the northern YM block at the repository site (blue line in Figure 2). Fractional P wave velocity perturbations are coded according to color bar. BM, Bare Mountain; BMF, Bare Mountain Fault; CF, Crater Flat; YM, Yucca Mountain; JF, Jackass Flat. Grey line is the topographic surface at this latitude. Dots are active and passive seismic sources within 0.05 latitude of this section. Vertical exaggeration is 3. (relative to sea level) are displayed in Figure 8a; depths 4, 6, 8, and 10 km are displayed in Figure 8b; a cross section at N, through the northern repository block of YM, is shown in Figure 9; the 5.5 km/s isovelocity surface, as a proxy for depth to bedrock, is depicted in Figure 10. We will first discuss large-scale findings and then proceed to smaller-scale features of interest starting with YM and progressing counterclockwise outward from YM General [18] In general, surficial topographic structures dominate the velocities to 2 km depth (all depths are relative to sea level). At 0-km depth, low velocities ( km/s) of the Amargosa Valley, Crater Flat and Jackass Flat basins are bounded by the high-velocity ( km/s) Bare Mountain, Funeral Mountains, and south central NTS regions. These modeled velocities match the expected velocities of basins and ranges with Cenozoic rocks comprising the bulk of basin material and pre-cenozoic rocks comprising the ranges. Although high-velocity material corresponds directly with Paleozoic exposures to the west, the high velocities to the east of southern Amargosa Valley and Jackass Flat are coincident with the exposed Paleozoic rocks of the Specter Range and extend north under the Tertiary tuff sequence into central NTS east of YM. In contrast, below 3 km depth, velocities are relatively homogeneous, mostly varying between 5.5 km/s and 6.5 km/s. Similar velocities have been derived with 1-D and 2-D models in the YM region [e.g., Hoffman and Mooney, 1984; Smith et al., 2000] as well as in the northern Basin and Range (B-R) province [Catchings, 1992]. These deeper velocities are within the range expected for pre-cenozoic rocks. However, due to the broad overlap of velocities of differing pre-cenozoic units (e.g., granite, limestone, dolomite, argillite), we cannot distinguish between these rock types based on velocities alone. The velocity variations ( km/s) could be due to differences in rock types; or, if the basement material primarily consists of a single rock type, then the velocity variations could be due to differences in fracture density which could imply differences in strength. [19] In the northern B-R province, broad structures with velocities at depth similar to those imaged in this study were observed from PASSCAL and COCORP 2-D lines by Catchings [1992], who interpreted the upper 5 km/s regions as composed of weaker, crushed or deformed basement material and the faster 6+ km/s regions as more competent, less deformed blocks. This argument was also supported by the fact that Tertiary volcanics, where present, were predominately found in areas with upper crustal velocities <6 km/s. He argued that a crushed, weaker material would more easily accommodate volcanic intrusion than areas where magma would need to break fresh rock [Catchings, 1992]. [20] Although slower velocities tend to occur within the rift structure proposed by Carr [1990] or Wright [1989] at depths shallower than 3 km, the deeper velocity structure in general does not show a clear correlation with the rift. In regions such as the Funeral Mountains, where detachment faulting has been documented geologically [Scott, 1990], we do not image any clear evidence for detachment faults. However, as mentioned before, given the nature of detachment faults we would not necessarily expect to unambiguously image them unless they happen to cut a vertically extensive structure. Strike-slip faulting, especially on the scale expected in the Schweickert and Lahren [1997] model, should be imaged within our tomography; but we see no clear evidence for this presumed structure in our images. [21] Imaged basement (below the basins) velocities exhibit close correspondence to known or inferred pre- Tertiary structural features. This is clearly seen on the 2-km depth section (Figure 8a) which is below most of the basins. The three dotted lines in Figure 8a follow structural trends 9of16

10 Figure 8. Depth sections through the P wave model labeled according to depth relative to sea level over shaded relief. P wave velocities coded according to color bar below each panel. Note that the velocities represented by the colors change with depth. Grey regions are areas unsampled by rays. Black dots are earthquake sources within 1 km of each section. Dotted lines in the 2 km depth section connect velocity highs (top and bottom lines) and lows (middle line) to indicate possible pre-tertiary trends in velocities. NTS outline shown in white, ESF tunnel in red, and faults of Figure 2 as black lines. Scale bar is in kilometers. of individual high- or low-velocity bands primarily east and NE of YM. These bands generally trend east-west at YM and turn northward in the regions east and north of YM. This may be a reflection of oroflexural bending [Albers, 1967; Cole and Cashman, 1999] of the pre-tertiary structures in central NTS. Below 4 km depth, the tomographic model loses any clear correlation with these pre-tertiary structures. However, at 8 10 km depth (Figure 8b), some NW and NE striking structures are evident. Although it is unclear what the significance of these structures may be, given their orientations they could be related to Walker Lane tectonics Yucca Mountain [22] At 0 km depth (1 2 km beneath the surface at YM) relatively high velocities ( km/s) are imaged 10 of 16

11 Figure 8. (continued) (Figure 8a) along an approximately 10-km-long north to NNE trending zone underlying the eastern topographic expression of YM (e.g., Fran Ridge, Busted Butte). At this depth, the model is most likely sensitive to the underlying Paleozoic bedrock. Borehole UE-25p#1, west of Fran Ridge, is the only borehole to have encountered Paleozoic basement rock within YM. The basement was intersected at a depth of 1.2 km (approximately sea level), although many boreholes exceed this depth, with three extending to 1.8 km depth below the surface [Scott, 1990]. Velocities imaged at YM are consistent with the borehole velocity logs when we consider that the tomography is vertically averaging the velocities. The borehole log at UE-25p#1 indicates velocities 4 km/s within the tuff sheets and velocities >6 km/s in the Paleozoic dolomite basement [Muller and Kibler, 1984]. Velocities decrease to the west and east of this highvelocity feature. This rise in basement under the eastern part of YM is also consistent with some gravity models [Ponce and Oliver, 1995]. At 1-km depth, the high-velocity basement structure is clearly evident extending from the YM crest eastward under Jackass Flat. In agreement with 2-D seismic reflection and refraction experiments through Crater Flat and across YM [Brocher et al., 1998], we image the basement (5.5 km/s contour) as gently west dipping under eastern Crater Flat in E-W cross section (Figure 9). East of and including the northern YM crest, velocities are predominately high from 1-km depth to at least 10-km depth. 11 of 16

12 Figure 9. Cross section through the northern YM block at the repository site (blue line in Figure 2). P wave velocities are coded according to color bar. White areas are unsampled by rays. Velocities are also contoured at 0.5 km/s intervals. Dots are sources within 6 km of this cross section. BM, Bare Mountain; BMF, Bare Mountain Fault; CF, Crater Flat; YM, Yucca Mountain; JF, Jackass Flat. Grey line is the topographic surface at this latitude. Vertical exaggeration is 3. [23] The YM block is a relatively high velocity (>6 km/s) structure, almost throughout the entire resolved portion of the model (1 10 km depth). The western extent of this structure (as defined by the 6 km/s contour, Figure 9) lies approximately under the YM crest and extends eastward under northern Jackass Flat and the Calico Hills. From 1 to 4 km depth it is imaged as a locally east-west trending highvelocity structure that merges with high velocities to its east (Figure 8). Its shallow eastern extent lies near Wahmonie where outcrops of granodiorite are present [Ponce, 1981]. This structure is spatially associated with inferred intrusives in the Calico Hills [Maldonado et al., 1979] and with aeromagnetic anomalies (Figure 5) that have been interpreted as the buried hydrothermally altered Eleana formation [Bath and Jahren, 1984] that Oliver et al. [1995] interpreted to extend under northern YM. Overall, the YM structure appears to be an eastward back-tilted structural block typical of Basin and Range deformation [Eaton, 1980; Stewart, 1978, 1971]. This structure is primarily confined to the northern half of YM (north of N). This underlying structural difference between the northern and southern halves of YM may account for why the southern half of the mountain has behaved tectonically differently from the north: The southern part of YM has experienced clockwise paleomagnetic rotation, whereas the north has not [Rosenbaum et al., 1991]. [24] Cross sections (Figure 9) do not suggest the existence of large-scale throughgoing detachment faults under YM [e.g., Scott, 1990], especially in the upper 5 km, where resolution is best. The western side of the YM structure is vertically extensive but there is no clear evidence of a subhorizontal offset of this structure in the 1 5 km depth range. Although we would be unable to image detachments with insufficient displacement (2 km) or velocity contrast, higher resolution seismic reflection surveys [e.g., Brocher et al., 1998] have found no evidence of their existence under YM. No YM faults demonstrate major structural influence in the tomographic images, except perhaps for the Solitario Canyon fault which may form the western boundary of the YM structure Jackass Flats and Regions East of YM [25] On the basis of the 5.5 km/s contour (Figure 10) east of YM, Jackass Flat is imaged as a shallow basin. It deepens southward eventually merging with the Amargosa Desert basin structure. The eastern boundary of southern Jackass Flat basin is better defined along a steep NE trending velocity gradient at 0-km depth (Figure 8a). This velocity gradient is coincident with the alignment of the Mine Mountain Fault Zone, mapped in NE Jackass Flat [Piety, 1996] as well as the general NE structural fabric of south central NTS [Carr, 1984]. It is also coincident with gravity and magnetic anomalies, and seismicity alignments east of YM. In particular, the Gravity Fault, defined based on both gravity [Winograd and Thorarson, 1975; Fridrich, 1999] and aeromagnetics [O Leary et al., 2002] represents the eastern boundary of Jackass Flat basin and the CFD of Fridrich [1999]. Physiographically, it coincides with the eastern boundary of the central section of the Kawich- Greenwater rift proposed by Carr [1984]. The structure is well imaged in the tomography and interpreted as a west dipping normal fault zone west of the Specter Range and Little Skull Mountain, kinematically consistent with N-S striking normal faults on the west side of Little Skull Mountain. On the basis of the 5.5 km/s isosurface, the Jackass Flat basin reaches its maximum depth against this structure at about 1 2 km. Note that, although there are some locations in the 5.5 km/s isosurface map which indicate locally deeper or shallower areas such as at the east side of Jackass Flat with a depth of 3 km, they represent 12 of 16

13 Figure 10. Depth relative to sea level to 5.5 km/s velocity isovelocity surface overlay shaded topography as proxy for depth to pre-cenozoic basement. NTS outline shown in white, ESF tunnel in red, and faults of Figure 2 as black lines. Scale bar is in kilometers. single nodes within the model and may not be trustworthy. In northeastern Jackass Flat, velocities are high even at 0-km depth as part of a NW trending gradient zone, indicating a very shallow part of the basin. [26] The model is characterized by higher velocities, especially at shallow depth (Figure 8a), east of Jackass Flat to about a longitude of 116W, but there are local complexities. The highest velocities east of Jackass Flat are primarily associated with the higher topography at Little Skull Mountain, Skull Mountain, Calico Hills, Shoshone Peak, and the CP Hills. Through 3 4 km depth, the predominant fabric in the velocity variations is NE to ENE (Figure 8a), consistent with the central NTS Paleozoic structural fabric [Cole and Cashman, 1999], orientation of Quaternary faults [Piety, 1996], and historical seismicity trends. The Rock Valley fault zone, a ENE striking zone of generally depressed topography, is imaged as lower velocities than the adjacent higher topography, but, shallower than 3 km, it shows higher velocities than those of the Crater Flat, Jackass Flat and Amargosa basins. [27] Deeper than 3 km (Figure 8b), the velocities directly to the southeast and east of YM become predominantly equal to or slower than the average model velocities at these depths. Below 4-km depth (Figures 8b and 9), a lowvelocity zone (<5.75 km/s) centered under NE Jackass Flat is imaged that persists to 10-km depth, with a minimum in velocity <5.5 km/s at 6 7 km depth. We have tested this structure using a variety of methods and it appears to be a robust feature of the model. Additionally, checkerboard tests indicate good resolution and jackknife tests estimate standard errors <0.3 km/s for this region. Velocities surrounding this low-velocity zone tend to be higher than the model average. Because of its location, this low-velocity zone may be associated with inferred intrusives under Calico Hills or with the Wahmonie intrusion Timber Mountain and Regions North of YM [28] The shallow structure (0 1 km depth) immediately to the north of YM is imaged as a northwest trending highvelocity zone (region >5 km/s in 0-km depth section, Figure 8a) which approximately follows the higher topography across the Calico Hills, Shoshone Peak, and under the resurgent dome of Timber Mountain. This velocity high follows a notable magnetic gradient identified by Carr [1984] that extends SE into Frenchman Flat (outline in Figure 5). Although this region is characterized by slightly higher velocities near the surface, gravity measurements do not indicate a similar rise into this region. This may be simply due to the fact that we are sampling slightly more buried material under the higher average terrane of Timber Mountain and surrounding regions. Velocities deeper than 1 km remain nearly constant in this area down to the base of the resolved portion of the model; the value is in the upper 5 km/s range, making it a slightly slower-thanaverage portion of the model at depth. North, into the Timber Mountain and Silent Canyon Caldera complexes, velocities are generally at or below the model average through 10-km depth. To the northeast, Yucca Flat is imaged as a lower than average velocity region down to 2 km depth and is generally surrounded by higher than average velocities associated with the higher terrain. From 3 to 6 km depth, Yucca Flat and the surrounding region are higher than average velocity Crater Flat, Bare Mountain, and Regions West [29] Crater Flat bounds the western side of YM. Following the 5.5 km/s contour surface (Figure 10) and consistent with 13 of 16

14 reflection studies [Mooney and Schapper, 1995; Brocher et al., 1998], the eastern flank of the basin is west dipping. The western boundary of Crater Flat, along the Bare Mountain Fault is clearly imaged in the shallow structure (0 1 km depth) as tightly spaced north-south trending velocity contours just east of Bare Mountain (Figure 8a). The tomography images the Bare Mountain Fault as a major structural feature in the area; this suggests that west dipping faults in the CFD such as the Solitario Canyon fault would be truncated at the Bare Mountain Fault. In east-west cross sections (Figure 9), the deepest part of the basin is adjacent to the Bare Mountain Fault where the 5.5 km/s isosurface is at about 3 km depth (Figure 10). The basin appears to be 1 2 km deeper in the south as compared to the north. The structure of Crater Flat in the tomographic model is consistent with the regional basin geometry derived from gravity data [Blakely et al., 1999]. A deeper southern section of the Crater Flat basin was also inferred by Fridrich [1999] based on evidence that the southern section has experienced more extension than elsewhere in the basin. Velocities beneath Crater Flat are less than 6 km/s until truncated by the east dipping Bare Mountain structure (Figure 9), described below. [30] Bare Mountain, a Paleozoic structural block, forms the footwall block of the Crater Flat basin. Velocities higher than 6 km/s characterize the Bare Mountain structure from 0 to 8 km depth; the Bare Mountain fault dips east and merges with the YM structure at 6 8 km depth. Crater Flat appears as a wedge of slower material (<6 km/s) between the Bare Mountain and YM structures. The continuity of the east dipping Bare Mountain structure down to at least 8 km depth casts doubt on the existence of a deep-seated fault zone such as suggested by Schweickert and Lahren [1997] running through Crater Flat. The northern part of Bare Mountain connects westward to the Bullfrog Hills, a series of east-west trending topography of lower relief than Bare Mountain. The near-surface velocities of these two features are almost identical Amargosa Desert and Regions South of YM [31] The Amargosa Desert comprises the entire area south of YM. It is characterized as a fairly shallow basin extending down to about 1 2 km depth in most areas (Figure 10). The western edge is not sharply resolved, but is defined by the high velocities comprising the Funeral Mountains core complex [Labotka and Albee, 1988]. The region to the east of the southern extension of the Gravity Fault into Amargosa Valley appears as a subbasin in the tomography (0 1 km depth, Figure 8a); this is consistent with gravity studies [e.g., Blakely et al., 1999]. South of the Specter Range, the high velocities that comprise the region to the north cease abruptly at this eastern lobe of the Amargosa Desert. The deep velocities under the Amargosa Desert do not indicate any clear offsets of structural blocks that would require a fault zone under this region Relocated Seismicity [32] There are several salient features in the seismicity relocated as part of this study that relate to the basement velocity model (Figure 11). The majority of seismicity in the NTS area occurs east of the Gravity and Mine Mountain faults at all depth levels even discounting the Little Skull Mountain sequence. Many of the earthquakes near previous nuclear testing areas such as in Pahute Mesa may be testinginduced, stress-triggered events [Hamilton et al., 1969]. Very little seismic activity occurs especially between the Gravity and Mine Mountain fault zones and the longitude of Bare Mountain, including YM itself. YM has been recognized for its low level of seismicity [Gomberg, 1991], and this has continued to be the case. The region of low seismicity rates in the vicinity of YM is spatially coincident with the CFD. Higher rates delineate the CFD boundaries, except on the west where the seismicity lies west of Bare Mountain instead of along the Bare Mountain fault. Just west of Bare Mountain, an approximately north-south lineation of small magnitude earthquakes extends from about 36.8 N to about N. North of Bare Mountain, this lineation is spatially associated with a steep down-tothe-east structure defined by the 5.5 km/s isosurface adjacent to the basins of the caldera complexes north of YM (Figures 10 and 11). At Bare Mountain the seismicity may indicate the division between the Bare Mountain structure and Bullfrog Hills structure to its west. On a broader scale, this linear trend of seismicity and a similar, but less well defined area of seismicity along the Gravity Fault and extending southward from the Little Skull Mountain aftershock sequence, may represent stress along the edges of a volcanic-tectonic rift as proposed by Carr [1990] or Wright [1989]. Seismicity rates, average shallow velocities (0 3 km) and Bouguer gravity inside this rift are lower than those observed to the west and east. 5. Conclusions [33] We have imaged the P wave velocity structure in the region surrounding Yucca Mountain to a depth of 10 km. The shallow velocity structure closely mimics the topography of the region, with high-velocity ranges separating variable depth low-velocity basins. Below the deepest basins (3 km), the velocity structure is fairly homogeneous with velocities ranging between 5.5 and 6.5 km/s. These results are similar to findings elsewhere, such as in the northern Basin and Range where Catchings [1992] interpreted the higher velocity regions with 6+ km/s velocities as being composed of stronger, less deformed blocks, whereas the slower sub-6 km/s regions consisted of weaker, more deformed materials. Velocities under YM are predominately faster than surrounding material (>6 km/s) from near the surface to at least 10 km depth. Regions east of the Gravity fault are generally faster at shallow depths and more seismically active than immediately west of this zone. Shallow high velocities and generally increased earthquake activity characterize the region west of the longitude of the Bare Mountain fault. Low velocities predominate in the caldera complexes to the north of YM. A prominent lowvelocity region is imaged under NE Jackass Flat from 5 to 10 km depth which may be related to mapped and inferred intrusives in the vicinity. [34] The regional-scale feature of a central north-south oriented low-velocity structure bounded by regions of higher velocity at less than 4 km depth support the rift models proposed by Wright [1989] and Carr [1990], although the lack of deep structural correlation with the proposed rift weakens the argument for the depth extent 14 of 16

15 Figure 11. Relocated seismicity (all depths) from this study on shaded topography. Earthquakes span the period from 1981 to 2003 and include events from magnitude 1.0 to 5.8. Outline of NTS shown in white, ESF tunnel in red, faults of Figure 2 shown as yellow. of the model. The local velocity structure in the YM area, illustrated in a cross section of the velocity structure between Bare Mountain and south central NTS (Figure 9), is consistent with the CFD model of Fridrich [1999]. However, the structural block underlying YM may divide the CFD into subdomains. The close spatial correlation between the region of low seismicity rates in the vicinity of YM [Gomberg, 1991] and the CFD, with higher rates at its margins, bolsters the argument that the CFD behaves as a tectonic domain. There is no evidence within our tomographic images that require detachment structures [Scott, 1990] under YM, especially at shallow depth, but due to the limitations of the tomography to image detachments, we cannot refute their existence. The imaged velocities also do not require a throughgoing fault zone running through Crater Flat and the Amargosa Desert [Schweickert and Lahren, 1997]. However, the strong evidence for local shear in the region may require at least a distributed shear zone, perhaps on the scale of the rift proposed by Carr [1990]. The high-velocity structure under (primarily northern and eastern) YM, including the repository, and Jackass Flat, which is imaged from shallow depths to the base of the well-resolved portion of the model, is most likely Paleozoic basement elevated relative to the Crater Flat basin. Given its similar velocities to Bare Mountain, it may represent a small, but typical Basin and Range-type structural block that is back tilted to the east. This may provide a structural explanation for the YM high topography. If the lower basement velocities under Crater Flat are due to weaker, more fractured basement rocks compared to YM, this may provide an explanation for the lack of Quaternary volcanism at YM proper [Coleman et al., 2004]. The Bare Mountain fault could provide a conduit for magma into the upper crust. The weaker basement rock under Crater Flat would be more conducive to eruption than the magma having to break fresh rock within the stronger YM block. [35] Acknowledgments. We would like to thank Steve Harmsen for providing the USGS SGBSN pick files for the underground nuclear tests We thank Tom Brocher for providing the data for the USGS controlled-source experiments and appreciate the time and effort he expended to put the traveltime data into an easily readable and usable electronic format from the original scanned images. The authors wish to acknowledge the continued support of the Yucca Mountain Project by the Department of Energy (DOE); this support has enabled utilization of the Southern Great Basin Digital Seismic Network (SGBDSN) for seismic monitoring and research. The funding for this research is provided by the DOE through a Cooperative Agreement (DE-FC28-04RW12232) with the Nevada System of Higher Education (NSHE), administered by the Harry Reid Center, Las Vegas, Nevada, and through a Lawrence Livermore National Laboratory subcontract of DOE grant DE-FC03-02SF References Albers, J. P. (1967), Belt of sigmoidal bending and right-lateral faulting in the western Great Basin, Geol. Soc. Am. Bull., 78(2), Bath, G. D., and C. E. Jahren (1984), Interpretations of magnetic anomalies at a potential repository site located in the Yucca Mountain area, Nevada Test Site, U.S. Geol. Surv. Open File, Blakely, R. J., R. C. Jachens, J. P. Calzia, and V. E. Langenheim (1999), Cenozoic basins of the Death Valley extended terrane as reflected in regional-scale gravity anomalies, in Cenozoic Basins of the Death Valley Region, edited by L. A Wright and B. W. Troxel, Spec. Pap. Geol. Soc. Am., 333, of 16