Interpretation of magnetic and new gravity data provides constraints on the geometry of the Hat Creek fault, the amount of right-lateral offset in the area between Mount Shasta and Lassen Peak (northern California, USA), and confirmation of the influence of preexisting structure on Quaternary faulting. Neogene volcanic rocks coincide with short-wavelength magnetic anomalies of both normal and reversed polarity, whereas a markedly smoother magnetic field occurs over the Klamath Mountains and Paleogene cover there. Although the magnetic field over the Neogene volcanic rocks is complex, the Hat Creek fault, which is one of the most prominent normal faults in the region and forms the eastern margin of the Hat Creek Valley, is marked by the eastern edge of a north-trending magnetic and gravity high 20–30 km long. Modeling of these anomalies indicates that the fault is a steeply dipping (∼75°–85°) structure. The spatial relationship of the fault as modeled by the potential-field data, the youngest strand of the fault, and relocated seismicity suggest that deformation continues to step westward across the valley, consistent with a component of right-lateral slip in an extensional environment.
Filtered aeromagnetic data highlight a concealed magnetic body of Mesozoic or older age north of Hat Creek Valley. The body’s northwest margin strikes northeast and is linear over a distance of ∼40 km. Within the resolution of the aeromagnetic data (1–2 km), we discern no right-lateral offset of this body. Furthermore, Quaternary faults change strike or appear to end, as if to avoid this concealed magnetic body and to pass along its southeast edge, suggesting that preexisting crustal structure influenced younger faulting, as previously proposed based on gravity data.
The Hat Creek fault is located at or near the junction of major tectonic provinces in northern California (USA): the Sierra Nevada–Great Valley block, southernmost Cascade arc, Basin and Range, Oregon coast block, and northern end of the Walker Lane (Fig. 1). The encroachment of the Walker Lane, a zone of right-lateral shear east of the Sierra Nevada–Great Valley block, into this region has been attributed to northward migration of the Mendocino triple junction 200–300 km to the west (Faulds et al., 2005), along with northwest translation of the Sierra Nevada–Great Valley block (Unruh et al., 2003). Many publications (e.g., Stewart, 1988; Muffler et al., 2008; Busby, 2013) do not show the Walker Lane extending north of Lassen Peak (Fig. 1). Yet others have argued that right-lateral shear continues north of Lassen Peak, perhaps as part of the enigmatic Northern California shear zone (Wesnousky, 2005, fig. 2 therein), which includes the predominantly normal Hat Creek and McArthur faults. The amount of right-lateral offset is unknown for the shear zone, although faults in the northern Walker Lane have as much as 30 km of displacement (Wesnousky, 2005; Faulds et al., 2005). Right-lateral shear may step to the west, producing uplift and folding through the Inks Creek fold belt of Harwood and Helley (1987) (ICF in Figs. 1 and 2) in the northern Great Valley (Unruh et al., 2003) and very speculatively to the Grizzly Peak anticline in the eastern Klamath Mountains (GPA in Fig. 2; Sawyer, 2013). Alternatively, the Walker Lane has been depicted as extending as far north as the Klamath graben (KG in Fig. 1; Oldow and Cashman, 2009), with extensional faults overprinted by possible young strike-slip movement (Waldien, 2012; Waldien and Meigs, 2013). Dextral shear in and north of the Hat Creek region might be expected, given oblique subduction of the Gorda plate beneath the North American plate (Fig. 1), and, although geodetic data indicate right-lateral shear north of Lassen Peak, it is clearly accommodated by clockwise crustal rotation north of the California-Oregon border (McCaffrey et al., 2007, 2013; Fig. 1 inset) rather than by discrete strike-slip faults. The boundary between the Oregon coast block, which has been rotating since the Miocene, and the Walker Lane is poorly defined, and the amount, if any, of right-lateral shear and how it may be accommodated in this region is not well known.
The most studied fault north of Lassen Peak is the Hat Creek fault (Muffler et al., 1994; Walker, 2008; Blakeslee and Kattenhorn, 2013; Kattenhorn et al., 2016), which is well expressed geomorphically and offsets units as young as the Hat Creek Basalt, dated as 24 ± 6 ka (Turrin et al., 2007; refined to 23.8 ± 1.4 ka by Rood et al., 2015) and ca. 15 ka periglacial deposits (Muffler et al., 1994). Although the Hat Creek fault is a significant normal fault that may reflect westward encroachment of Basin and Range extension (Muffler et al., 2008), its left-stepping geometry indicates a small right-lateral, strike-slip component (Wills, 1991; Blakeslee and Kattenhorn, 2013). Estimates of Quaternary normal slip rates range from 1.2 to 1.9 mm/yr (Turrin et al., 2007) to 2.2–3.6 mm/yr (Blakeslee and Kattenhorn, 2013), whereas the most recent estimate from modeling of global positioning system data is <1 mm/yr (Thatcher et al., 2014). The slip rates are based on an assumed fault dip of 60° because information on subsurface fault geometry is lacking. Where and in what manner this fault extends to the north is of interest, particularly for state, federal, and private infrastructure, such as dams along the Pit River.
The area has been relatively little studied, as compared to the Walker Lane. Most of the area has been mapped at only reconnaissance levels, except for a map of Lassen Volcanic National Park in the southern part of the study area (Clynne and Muffler, 2010), a map of the outline of a lava flow that extends from the Medicine Lake volcano area into the north-central part of the study area (Donnelly-Nolan, 2010), and detailed mapping (Muffler et al., 2012) that builds upon and extends earlier mapping of Hat Creek Valley (Muffler et al., 1994; Turrin et al., 2007). Little is known about pre-Pliocene geology in this area because of extensive Quaternary volcanic cover (Fig. 2). The subsurface geometry of the faults in this area is poorly known because seismicity is generally dispersed, even when hypocenters are relocated with double-difference methods (Waldhauser and Schaff, 2008). Previous work in the southern Cascade arc, based on regional gravity lows that transect the study area, proposed that the patterns of Quaternary faulting and volcanism are influenced by preexisting crustal structure (Blakely et al., 1997; Fig. 1). As part of a larger effort to characterize seismic hazard in the region north of Lassen Peak, we collected gravity data and reexamined the existing aeromagnetic data in order to refine the subsurface structure of this region. In doing so, we examine constraints on the amount of right-lateral slip, revisit the role of preexisting structure on Quaternary faulting, and investigate the geometry of the Hat Creek fault north of Lassen Peak.
DATA AND METHODS
We collected 1090 new gravity measurements (solid black circles, Fig. 3; Supplemental File1) and added them to the regional database (solid black triangles, Fig. 3), which consists of nearly 2000 measurements in the study area (Chapman et al., 1977; Robbins et al., 1976). Standard formulas for the free-air, Bouguer, curvature, and terrain corrections were used (Blakely, 1996; Telford et al., 1990). We used a standard reduction density of 2670 kg/m3 (typical of basement rocks, such as those exposed in the Klamath Mountains) shown in Figure 3; this assumes that crust above the geoid, including the terrain, has a density of 2670 kg/m3. However, much of the terrain, excluding the Klamath Mountains, is composed of Cenozoic volcanic rocks with lower average densities of ∼2500–2600 kg/m3 (LaFehr, 1965; Table 1). Thus we also show, in Figure 4, the gravity field reduced with a density of 2500 kg/m3 in order to attenuate anomalies that may be artifacts of reduction density (e.g., gravity lows that correlate with positive topographic features in Fig. 3). Figures 3 and 4 also include an isostatic correction that removes long-wavelength effects of deep crustal and/or upper mantle masses that isostatically support regional topography. The resulting isostatic residual gravity maps (Figs. 3 and 4) reflect lateral variations of density within the middle to upper crust. The accuracy of these data is estimated to be better than ±0.5 mGal.
Aeromagnetic data consist of a survey flown in 1980 along east-west flight lines spaced 1600 m apart at an altitude of 2740 m (Couch and Gemperle, 1982). These data were processed to account for the inclination (65°) and declination (17°) of the Earth’s magnetic field in this region and to shift the anomalies over their causative sources. This operation is referred to as reduction to the pole, and transforms the magnetic anomaly to that that would be measured at the magnetic north pole where the Earth’s magnetic field is directed vertically down. Although aeromagnetic profiles were flown closer to the ground (120 m) by the National Uranium Resource Evaluation Program (Hill et al., 2009), these lines were flown too far apart (4800 m) to map effectively the magnetic field between the lines.
We used gradients of gravity and magnetic fields to map the locations of geologic contacts (dashed gray and black lines, Figs. 3 and 4; magenta lines, Fig. 4; black dots, Figs. 5 and 6). The method (Cordell and Grauch, 1985; Blakely and Simpson, 1986) exploits the fact that gravity anomalies and magnetic anomalies converted to magnetic potential have steepest horizontal gradients directly over the contacts. A matched filter (Phillips, 2001) was applied to the aeromagnetic data to separate the data into different wavelength components that can then be related to depth. This is achieved by modeling the observed anomalies as a sum of anomalies from distinct equivalent source layers (fictional layers below the observation surface where the distribution of magnetization produces the observed magnetic field) at increasing depths. The matched-filter produced dipole-equivalent source layers at depths of 1.1 km (shallow), 2.6 km (intermediate depth), and 19.8 km (deep) (Fig. 6). Comparison of these fields with the geology in the following suggests that the upper two layers generally reflect anomalies caused by the Neogene volcanic rocks, whereas the deepest layer is sensitive to sources in the underlying basement. Note that the actual source depths for the layers may be shallower, given that broader wavelength anomalies can be fit by both deep and shallow sources. We also show gradients derived from these filtered versions of the magnetic data in Figure 6.
The dominant features in the gravity field are the pronounced lows over the Shasta and Lassen volcanic centers and Caribou volcanic field and the 30-km-wide band of gravity highs that extends east-northeastward from exposed pre-Cenozoic rocks of the Klamath Mountains (Figs. 3 and 4). Pronounced gravity highs also bracket the Lassen volcanic center and Caribou volcanic field on its eastern and western margins. Although they do not correspond with any exposed pre-Cenozoic basement in the study area, the gravity highs likely reflect concealed basement highs (Blakely and Jachens, 1990; Blakely et al., 1997), as basement is exposed ∼10 km to the south of the study area (Jennings et al., 2010). Although the gravity highs that bracket the Lassen volcanic center and Caribou volcanic field are of similar amplitude on either side of the gravity low, the basement rocks differ in their magnetic character (Fig. 6C) and may reflect different basement rock types. The alternating bands of gravity highs and lows extend to the east-northeast beyond the study area for ∼300 km (Blakely et al., 1997; green polygons in Fig. 1).
The gravity highs extending northeast from the Klamath Mountains are interrupted by gravity lows extending north-northwest from Fall River Mills and Big Valley (Figs. 3 and 4). These lows likely reflect low-density basin fill in young tectonic basins because the margins of these lows partly coincide with Quaternary faults. Some of the basin fill may be late Neogene in age, but the ages of the oldest lava flows exposed in the upthrown blocks of these faults (Page and Renne, 1994) suggest that these basins formed primarily during the Quaternary. The gravity low west of Big Bend coincides with the exposed Triassic sedimentary section of the Redding subterrane (Fig. 2) and thus cannot be caused by Cenozoic basin fill. The gravity low results from the sedimentary rocks being less dense than the mafic and ultramafic rocks of the Trinity subterrane exposed to the west and denser basement rocks concealed beneath Cenozoic volcanic cover to the east. These concealed basement rocks likely include metavolcanic rocks and hypabyssal diorite bodies, such as those exposed along the west bank of the Pit River near Big Bend (Renne and Scott, 1988).
The new gravity data refine the southern boundary of the gravity high between Big Bend and Fall River Mills. Here the southern boundary is stepped and marked by two gradients. The northern gradient is aligned along the Pit River for a distance of 20 km west of California Highway 89 (westernmost black dashed line in Fig. 3); the southern gradient is generally coincident with the edge of the low defined by Blakely et al. (1997; gray dashed line in Fig. 3). Part of the northern gradient in Figure 3 is caused by terrain north and south of the Pit River that is less dense than the reduction density (2670 kg/m3) based on the gradient’s alignment with the river and its steepness. However, an erroneous reduction density cannot explain the entire gradient because the northern gradient remains when the reduction density is reduced to a more representative density of 2500 kg/m3 (Fig. 4; Table 1), or even lower densities such as 2400 kg/m3. We conclude that part of the northern gradient is caused by a subsurface density contrast. The northern gradient is gentler east of California Highway 89 and is interrupted by gravity lows associated with thick basin fill in the Fall River Mills area and Big Valley.
Superposed on the gravity low that extends north in a subdued fashion from the Lassen volcanic center are higher gravity values that trend north and are west of the Hat Creek fault (HCF in Fig. 3) and within the eastern part of Hat Creek Valley (Fig. 4). The east edge of the gravity high coincides with the Hat Creek fault south of Cinder Butte. The gravity high weakens north of Cinder Butte. The presence of a gravity high over most of the valley contrasts with gravity lows that arise from low-density basin fill in the Fall River Mills area, Big Valley, and most basins of the Basin and Range province to the east. This feature is discussed more at length in the section on modeling of the Hat Creek fault.
The magnetic data reveal a complicated magnetic pattern over exposed Neogene volcanic rocks. The pattern consists of short-wavelength, high-amplitude magnetic highs and lows and is in contrast with smoother anomalies over the Klamath Mountains (Fig. 5). Although the magnetic pattern is complex over the young volcanic rocks, some of the anomalies can be clearly tied to geologic features exposed at the surface. Pronounced oblong magnetic highs are present over young lava cones or shield volcanoes at West Prospect Peak (WPP, Fig. 5; ca. 400–300 ka, Clynne and Muffler 2010) and Prospect Peak (PP, Fig. 5; 247 ± 56 ka, Clynne and Muffler, 2010), Burney Mountain (BM, Fig. 5; 40K/39Ar 243 ± 24 ka by Brent Dalrymple and 40Ar/39Ar 280 ± 6 ka by Marvin Lanphere, inMuffler et al., 2012), Freaner Peak (FP, Fig. 5), Magee Volcano (MV, Fig. 5; 210 ± 120 ka; Clynne and Muffler, 2010), and Ash Creek Butte (ACB, Fig. 5; 227 ka, Andrew Calvert, 2015, personal commun.), consistent with their formation during a normal polarity chron. Inverse modeling of the Ash Creek Butte magnetic high indicates that the bulk of the magnetization (declination of ∼23°, inclination of ∼33°) arises from at or below the topographic base of the butte, possibly as an intrusive plug (Christiansen et al., 1977). The hydrothermally altered core of Brokeoff Volcano (Clynne and Muffler, 2010) may be the source of a subdued magnetic low 5 km southwest of Lassen Peak. A pronounced magnetic low coincides with the dacite of Bogard Buttes (BB, Fig. 5; 2350 ± 70 ka and 2235 ± 49 ka, Clynne and Muffler, 2010), consistent with its having been erupted during a reversed polarity chron. The magnetic high over Burney Spring Mountain (BS, Fig. 5) suggests that it should have erupted during a normal polarity chron. An age of 2556 ± 142 ka (Andrew Calvert, inMuffler et al., 2012) makes it slightly younger than the Gauss normal chron (Cande and Kent, 1995), suggesting that the true age may be within the older range of the age uncertainty. Additional dates (Andrew Calvert, 2015, personal commun.) and paleomagnetic data (Duane Champion, 2015, personal commun.) support its age within the Gauss normal chron and are consistent with the magnetic high.
Circular to oblong magnetic lows over Saddle Mountain, Brush Mountain, and Chalk Mountain suggest reversed polarity for these volcanoes. Although the flows at the top of Brush Mountain have normal polarity, the magnetic low suggests that the bulk of the edifice may have formed during a reversed polarity chron, a conclusion supported by several paleomagnetic samples with reversed directions taken from the base of the mountain (Duane Champion, 2015, personal commun.).
Another set of magnetic anomalies can be attributed to juxtaposition of rock of differing magnetic properties by Quaternary faults. The faults bounding Hat Creek Valley provide the best example in the study area. The Hat Creek fault coincides with the east edge of magnetic high that extends north from the intersection of California Highways 89 and 44 (Fig. 5). The magnetic high broadens and appears to step slightly to the west, ∼5 km south of Cinder Butte. Just north-northeast of the source vents for the Hat Creek Basalt (star in Fig. 5) is a localized high superposed on the linear high bound by the fault. This high coincides with a shift in fault activity that Kattenhorn et al. (2016) hypothesize may be related to underlying magmatic influences. A magnetic low straddles the western margin of the valley, also marked locally by faults, from Highway 44 to the latitude of Burney over a distance of 25 km, and coincides with a belt of volcanic rocks that are considerably older than those to the east and west (Fig. 2).
The short-wavelength magnetic anomalies associated with young volcanic rocks contrast with the smoother pattern over the western third of the study area. Broader magnetic anomalies are present in the southwest part of the study area, despite outcrops of Neogene volcanic rocks. The smoother pattern in this area results from the survey altitude, which placed the magnetic sensor more than 2000 m above the ground surface. The smoother pattern over the Klamath Mountains, however, results from extensive outcrops of weakly magnetic Redding subterrane, which is composed of a Devonian to Jurassic sequence consisting of sandstone, shale, tuff, and limestone intercalated with island-arc volcanic rocks and intruded by plutons (Irwin, 1994). This sequence was affected by burial metamorphism up to greenschist facies (Renne and Scott, 1988). This subterrane is not a major magnetic source, although the filtered magnetic data that enhance shallow sources (Figs. 6A, 6B) highlight anomalies associated with the igneous parts of the subterrane. The Redding subterrane (Fig. 2) is faulted on the west against Ordovician gabbro and ultramafic rocks of the Trinity subterrane, which is considered to be the source of the prominent northeast-trending magnetic high in the northwest corner of the study area (Griscom, 1977; Blakely et al., 1985). The magnetic high increases in amplitude over outcrops of Trinity subterrane along the west edge of the map area (oph in Fig. 2), and the magnetic field filtered to enhance deeper sources (Fig. 6C) is consistent with the Trinity subterrane extending northeastward in the subsurface to beneath Mount Shasta (Blakely et al., 1985). Interpretation of a seismic refraction profile just north of the study area also supports the presence of an 8–10-km-thick slab of ultramafic rocks extending beneath Mount Shasta at 5 km depth (Fuis et al., 1987).
Southeast of the smooth magnetic pattern associated with the Redding subterrane is a prominent magnetic gradient that extends over a distance of ∼40 km northeast from Big Bend (northwest edge shown by white arrows in Figs. 5 and 6). Although many of the short-wavelength magnetic highs southeast of the magnetic boundary are clearly caused by Neogene volcanic rocks, these anomalies are superposed on a broad magnetic high (Fig. 6C). The northwest edge of the roughly rectangular high has a steeper gradient than the southeast edge, suggesting that the source dips to the southeast. Two sources have been proposed for the magnetic high: exposed Quaternary volcanic rocks (Glen et al., 2004) or a block of mafic or ultramafic rock buried at shallow depth (Blakely et al., 1985), probably 2–3 km deep, based on the width of the gradient on its northwest edge. We prefer the latter interpretation for four reasons. (1) The northwest edge of the magnetic high is present in the same place in all of the filtered magnetic maps, and the high extends across outcrops of weakly magnetic Montgomery Creek Formation, an Eocene nonmarine sandstone and conglomerate sequence as much as 800 m thick (Sanborn, 1960) near Big Bend. (2) Blakely et al. (1985) showed that uniformly magnetic volcanic terrain could not account for the broad magnetic high. (3) The inferred southeast dip of the body is consistent with southeastward younging and structure within this part of the Redding subterrane (Fig. 2). (4) Regional seismic tomography (Thurber et al., 2009) shows higher velocities at depths of 4–14 km in the area of the magnetic high. The age of the concealed body is unknown, but likely to be Mesozoic or even older, if, for example, it is a fragment of Ordovician Trinity subterrane.
Geometry of the Hat Creek Fault
We used the gravity and magnetic fields to model the geometry of the Hat Creek fault, the most prominent fault in the study area. We used a 2.5 dimensional simultaneous gravity and magnetic modeling program based on generalized inverse theory. The program requires an initial estimate of model parameters (depth, shape, magnetization, and density of suspected sources) and then varies selected parameters in an attempt to reduce the weighted root mean square error between the observed and calculated potential fields. The initial model estimate is based on mapped geologic relationships and physical property information. The amplitude of the anomaly is not the only attribute to match; matching its gradients and inflections are critical parameters that provide important information on the depth to the top of the source and its shape.
The interpretation of potential-field data yields nonunique solutions because an infinite number of geometrical models will have an associated field that closely matches the measured field. For example, increasing the density while decreasing the thickness of a proposed basalt body will generally not produce an appreciable change in the computed field. Although potential-field modeling is nonunique, the final models are based on numerous iterations and have been extensively tested to produce a model that is geologically reasonable with minimum structural complexities. The synthesis of the potential-field data and independent constraints described in the following leads to a family of similar models, regardless of the starting model, with characteristics that support our major conclusions on the subsurface geology.
To determine the geometry of the Hat Creek fault, a profile was modeled along section A–A′ (see Figs. 2–6). This profile was chosen because it coincides with detailed gravity measurements where the gravity anomaly associated with the fault is most pronounced and where the magnetic anomaly associated with the fault is more two-dimensional in character.
We first tested whether the gravity anomalies result from topographic effects. Rock of uniform density 2000 kg/m3 is needed to fit the amplitude of the gravity gradient across the Hat Creek fault (blue dashed line, center panel, Fig. 7). Although the sharpness of the gradient argues for a near-surface source, a density of 2000 kg/m3 is significantly lower than density measurements of the rocks in this area (Table 1), seems unlikely to account for all the observed gravity gradient across the Hat Creek fault, and does not fit the observed gravity anomalies away from the fault. The magnetic anomalies that span the valley also do not result from topographic effects, as indicated by the general lack of agreement between observed magnetic variations and those predicted from uniformly magnetic terrain along the profile (top panel in Fig. 7).
We modeled the profile using reasonable densities and magnetic properties to fit anomalies of interest, starting with a deep magnetic source (pink body, bottom panel, Fig. 7) to fit a broad magnetic high west of the valley (Fig. 6C); its exact geometry is not well constrained, but it may be related to the deep, oblong magnetic anomaly to the northwest (Fig. 6C). Its density is not constrained, but assumed to be higher than the surrounding crust and could result from partly serpentinized ultramafic rock.
To account for the long, narrow magnetic low along the western margin of the valley (Figs. 5, 6A, and 6B), we model a reversely magnetized, steeply dipping body (dark blue body, bottom panel in Fig. 7). Given the length and width of the magnetic low and the modeled depth extent of its source, it could reflect a concealed system of dikes. We have two possible simple models for the magnetic high and gravity gradient of the eastern margin of the valley. One model assumes that the gravity gradient results from juxtaposition of dense (2800 kg/m3) basin fill (brown body in Fig. 7), presumably basalt, against less dense (2590 kg/m3) volcanic rocks east of the fault. A good fit of the steep gravity gradient is achieved if the fault dips 75° and the basin fill is as much as 2 km thick. The fill must include basalt flows older than the Hat Creek Basalt, because the Hat Creek Basalt has a maximum thickness of ∼50–75 m (Walker, 2008; Kattenhorn et al., 2016). The dense basin fill, if also magnetic, can also account for the magnetic high in the eastern part of the valley, although the predicted gradient is somewhat steeper than observed.
The second model attempts to fit the gravity and magnetic data using a concealed, normally magnetized, steeply dipping body beneath the eastern part of the valley (red body, bottom panel in Fig. 7). The steeply dipping body model fits the shape of the magnetic high better (red line, top panel in Fig. 7), but the resulting gravity gradient (red line, middle panel in Fig. 7) is appreciably less steep than observed. A geologic interpretation of this steeply dipping body could be a dike swarm, presumably older than the Hat Creek Basalt and Cinder Butte (24 ± 6 ka and 38 ± 7 ka; Turrin et al., 2007), because no dikes are exposed in that part of the valley. The vents that erupted the Hat Creek Basalt are located only near the southern margin of the magnetic high (magenta star in Fig. 5). For both the basin-fill and steeply dipping model geometries, a 60° dipping interface for the gravity and magnetic highs bound by the Hat Creek fault produces gradients that are significantly less steep than observed. Regardless of the model, the potential-field data suggest that the Hat Creek fault dips steeply (75°–85°) in the upper 2–10 km.
We discuss the possible implications of the potential-field data for the style and evolution of Neogene deformation in the region north of Lassen Peak.
Hat Creek Fault
Although the Hat Creek fault is clearly a normal fault, the potential-field data suggest that it dips more steeply than a typical Basin and Range normal fault, which dips 40°–60° as documented by Stein and Barrientos (1985). Normal faults can dip more steeply, for example, in volcanic rocks. As mapped, the youngest scarps of the fault are subvertical (Muffler et al., 1994). The extent of these subvertical scarps into the subsurface, however, would be minimal, if they displace only the Hat Creek Basalt (∼50–75 m thick; Walker, 2008; Kattenhorn et al., 2016) along preexisting near-vertical cooling joints (Muffler et al., 1994; Blakeslee and Kattenhorn, 2013; Kattenhorn et al., 2016) that are assumed at depth to merge into a 60° dipping master normal fault. Near-vertical scarps produced by surface-breaking normal faults have been documented elsewhere, such as Hawaii, Iceland, and the East African Rift, but fissures associated with these scarps are considered to result from slip on more shallowly dipping normal faults at depth (Grant and Kattenhorn, 2004). However, relocated seismicity (Waldhauser and Schaff, 2008) projected onto our model profile suggests that deformation is accommodated along steeply west-dipping planes that extend through the seismogenic crust beneath the central part and west of the valley (Fig. 7). The eastern alignment of seismicity using a best-fit regression does not project up to the surface at mapped strands of the Hat Creek fault, but 1 km west of its westernmost strand at a dip of ∼72° (r2 = 0.85). The seismicity alignments are adjacent to our modeled steeply dipping body, a feature that could reflect dikes and accommodate at depth the extension manifested at the surface as fissures or brittle faulting. Such behavior has been documented in the Rahat Lunayyir in Saudi Arabia (Pallister et al., 2010), but not in Hat Creek Valley (Muffler et al., 1994, 2012). Furthermore, the amount of throw on the Hat Creek fault strands is probably too high to be attributed to a near-surface phenomenon above the underlying dikes. Although the relocated microearthquakes are precisely located with respect to each other (errors of 100–300 m), the absolute locations may be less accurate. The systematic 1–2 km westward offset of the seismicity alignments with respect to our modeled features and the surface traces of the Hat Creek fault may result from errors in the velocity model; no direct information on velocity variations at the scale of these features is available. Comparison of relocated earthquakes clustered near a quarry at Brush Mountain suggests that there is no systematic bias in the locations. In this case, the offset of the microseismicity locations is real and thus could be interpreted in terms of deformation stepping westward through time. No surface expression of such a fault has been found, which implies that the fault has not yet propagated to the ground surface. Westward-stepping faulting has been proposed for the Hat Creek fault, which consists of three subparallel scarps, the youngest scarp being the westernmost strand (Walker, 2008; Blakeslee and Kattenhorn, 2013; Kattenhorn et al., 2016). The distribution of microearthquakes beyond the plane of profile A–A′, although diffuse and primarily south of profile A–A′, suggests a more westward trend than that of the various strands of the Hat Creek fault.
The left-stepping geometry of the fault suggests some component of right-lateral slip in an extensional setting (Walker, 2008; Blakeslee and Kattenhorn, 2013); this is supported by focal mechanisms of microearthquakes in Hat Creek Valley (Lahontan Geoscience, Inc., 2012). Other faults in the study area also exhibit left-stepping geometry, such as the McArthur fault (Wills, 1991), the northern part of which forms the east margin of the Fall River Valley basin. The western margin of the Fall River Valley basin, as defined by steep gravity gradients (Fig. 4), is also left stepped, suggesting that these older, mostly concealed faults may also have accommodated some component of right-lateral slip.
A kinematic analog for the Hat Creek fault may be the Sierran frontal fault system between Bishop and Lake Tahoe. Although individual fault segments are normal faults, they form a left-stepping pattern (Unruh et al., 2003) with dominantly strike-slip focal mechanisms (Oldow, 2003). Although the Hat Creek fault has predominantly normal displacement, the steeply dipping structures implied by the potential-field modeling and seismicity suggest that it is active in a transtensional setting.
An 80° dip on the Hat Creek fault implies slightly slower slip rates as compared to a typical fault with 60° dip, for a given vertical separation (∼12%). However, if the basin fill model is correct, the amount of cumulative offset on the Hat Creek fault would be greater than the throw previously documented using topographic profiles across the fault scarps (at least 570 m; Kattenhorn et al., 2016); the throw measured from the topography does not account for offset concealed by young lava and thus represents a minimum estimate.
Northeast-Trending Magnetic Gradient
The boundary between the smooth magnetic pattern of the Klamath Mountains and the complex magnetic pattern to the southeast is very linear (white arrows in Fig. 5). The linearity of this boundary, which we name the Eastern Klamath boundary, persists in all filtered versions of the magnetic field (Fig. 6), suggesting that it is not simply a near-surface feature, but extends at least through the upper crust. The linearity of this magnetic boundary precludes any discrete right-lateral offsets of the causative body >1–2 km (Fig. 6A). Thus, no significant discrete right-lateral offset associated with the Walker Lane has propagated from the southeast across this boundary. Furthermore, given that this boundary is interpreted as arising from the Klamath basement, no significant right-lateral offset has propagated across this area since the Mesozoic.
Other interpretations, which we posit as less likely because of the absence of supporting data in the study area, could also explain the absence of discrete right-lateral offsets across the Eastern Klamath boundary, such as distributed right-lateral shear. The absence of discrete offsets of the linear magnetic boundary could result from slip on many closely spaced faults. South of the Pit River, the Hat Creek fault becomes one of many faults distributed between Burney and Pittville. Given this distributed nature of faulting, we consider an alternate interpretation of the Eastern Klamath boundary as having been rotated into its current orientation by distributed right-lateral shear. This shear could be related to the clockwise rotation of the Oregon coast block. Although there are no paleomagnetic data that pertain to Neogene and younger clockwise rotation in the study area, clockwise rotation has been documented in mid-Oligocene to early Miocene volcanic rocks to the north near the California border (14° ± 9°; Beck et al., 1986) and Miocene to mid-Pliocene volcanic rocks to the east near the Nevada border (11.9° ± 4.5°; Ritzinger et al., 2014). The amount of rotation in these rocks overlaps with the amount based on a few sites scattered north and west of our study area in Cretaceous sedimentary rocks (11.5° ± 15.8°; Mankinen and Irwin, 1982). However, paleomagnetic data from the volcanic rocks in this area are needed to document timing and evolution of any rotation near the southern boundary of the Oregon coast block.
The absence of discrete offsets of the magnetic boundary may indicate that right-lateral slip associated with the Walker Lane and other faults steps to the west in the area between Lassen Peak and Mount Shasta along a transpressional belt, as suggested by Sawyer (2013); his interpretation is based on the apparent change in strike of faults near the Pit River to a more western direction and the east-west prong of pre-Cenozoic basement near the crest of the mountains north of Big Bend that has been attributed to an enigmatic structure called the Grizzly Peak anticline (GPA in Fig. 2) (Lahontan Geoscience, Inc., 2012; Mushroom Rock anticline; Austin, 2013). Eocene Montgomery Creek Formation is exposed on the southern limb of the anticline and appears to wrap around the Redding subterrane rocks (Fig. 2). Some support for a compressional stress regime in this area comes from analysis of focal mechanisms that indicate transpression (and transtension) in this area (Lahontan Geoscience, Inc., 2012), and a very detailed study of faulting style in the Dicalite mine north of Burney Springs Mountain that indicates that the stress state there has evolved from predominantly normal to strike slip to reverse during the past 1 m.y. (Austin, 2013). A narrow, small-amplitude gravity high coincides with the proposed Grizzly Peak anticlinal structure and may extend east as far as California Highway 89 (Figs. 3 and 4), although the timing of basement uplift suggested by this anomaly is not constrained by the potential-field data. It is likely that the Grizzly Peak structure formed earlier than 1 Ma, given Neogene doming and uplift documented in other parts of the Klamath Mountains, such as the Condrey Mountain area 100–150 km to the northwest (Mortimer and Coleman, 1984), and thus is not deformation attributable to the Walker Lane. Recent geodetic modeling in the area (Thatcher et al., 2014) suggests a component of compression across the southern Klamath Mountains (Wayne Thatcher and Robert Simpson, 2014, written commun.), so it is possible that there is a Quaternary component to the development of the anticline. Detailed mapping of the volcanic rocks on either side of the proposed structure that provides amount, sense, and age of any tilting might address whether there is a measurable Quaternary component to the formation of the Grizzly Peak anticline.
Influence of Preexisting Structure
Preexisting structures appear to influence Quaternary faulting and volcanism in this region, as pointed out by Blakely et al. (1997) based on gravity anomalies. Quaternary and Neogene volcanoes and vents appear to be concentrated in the gravity lows associated with major volcanic centers of the southern Cascade Range (e.g., Lassen Peak, Mount Shasta, and Mount McLoughlin). These broad lows extend to the northeast 300–400 km (Fig. 1). East-west extension, as characterized by north-striking faults, is concentrated in the lows, and the lows are interpreted to be areas that are thermally weakened. In the intervening gravity highs, which coincide with areas that lack young volcanism and thus may possess greater elastic strength, faults strike northwest and are more optimally oriented for right-lateral slip. Here we expand on this motif by examining how preexisting structure, as manifested in the potential-field data, coincides with the character of the Hat Creek fault.
The Hat Creek fault north of Cinder Butte is one of many mapped faults distributed between Burney and Pittville. The gravity and magnetic character associated with the fault also changes in this area, with the weakening of the gravity high northward along the eastern margin of the valley (Fig. 3) and a change in strike of the magnetic high to the northeast that is parallel to the course of the Pit River (Fig. 5). The source of the lower gravity values is unknown but may be basin sediments that are concealed by the young volcanic cover. A possible component of that basin fill could be diatomite lake-bed deposits, which are exposed in an east-west belt along the Pit River and suggest that this area has been low-lying for ∼1–1.5 m.y. (Page and Renne, 1994). The low-lying area may have resulted from distributed normal faulting. This region is bracketed by two gravity gradients: the southern gradient roughly coincident with that identified in Blakely et al. (1997), and the northern gradient at the latitude of the Pit River identified by our new data where faults have more of a northwest strike (Fig. 2), as strain is deflected into the even stronger crust as indicated by the gravity highs.
The change in faulting style is roughly located along a weak, northeast-striking alignment of prominent, semicircular magnetic lows (Br, SM, a–c in Fig. 5) that may reflect volcanoes that erupted during reversed polarity epochs. The sources of the easternmost lows are not associated with volcanic edifices and are concealed beneath surficial deposits or volcanic rocks. We speculate that the distribution of these older volcanoes may also have been influenced by the preexisting crustal structure as delineated by the northeast-striking regional gravity high discussed by Blakely et al. (1997). Alternatively, the change in faulting style and distribution of volcanic rocks could result from a change in kinematic boundary conditions at this latitude.
The change in faulting character also coincides with the southeastern edge of the deep magnetic body that produces the prominent long-wavelength magnetic high in Figure 6C. If the magnetic body indicates more mafic rock, such as ophiolite, it may be more resistant to brittle deformation and thus influence the location and style of faulting. If this is the case, it is reasonable to examine the relationship of magnetic anomalies and faulting over a broader region than our study area. Such an examination (Fig. 8) shows that Quaternary faulting appears to avoid the most prominent magnetic highs. Most recently, based on its focal mechanism and aftershock distribution, the 2013 M5.7 Canyondam earthquake ruptured a 70° northeast-dipping fault (http://earthquake.usgs.gov/earthquakes/eqarchives/poster/2013/20130524CA.php, accessed 8/12/2014) that is aligned along the southwest edge of a prominent magnetic high. Strings of serpentinite and large bodies of metavolcanic rocks are exposed to the southeast of the epicenter (Jennings et al., 2010) and may be the source of the magnetic high.
Northeast-trending conductive features roughly coincident with the gravity lows are also imaged by magnetotelluric data (orange lines in Fig. 1; Bedrosian and Feucht, 2014) at middle to lower crustal depths (∼32 km). Given that the gravity data reflect structure within the upper and middle crust, the colocation of the northeast-trending zones in the lower crust suggests that these geophysical structures may be related to each other and thus penetrate the entire crust. The geophysical features are also parallel to, and envelop, a proposed northeast-trending Eocene to Miocene tear in the subducting slab (Colgan et al., 2011; dashed blue line in Fig. 1). The potential-field features also are on strike with the structure or structures responsible for the westward translation of the Klamath terranes from correlative terranes in the northern Sierra Nevada foothills. The translation of these terranes must have occurred during the Early Cretaceous, based on assemblages that overlap terranes in both the Klamath Mountains and Sierran foothills and on the absence of post–125 Ma plutons in the Klamath Mountains (Ernst, 2012). The evolution of these older features, however, is a discussion beyond the scope of this paper, but has led to an interesting and enigmatic superposition of preexisting structures that extend into the Hat Creek fault region and that appear to have influenced Quaternary deformation.
Modeling of gravity and magnetic highs that extend 20–25 km along the eastern margin of the Hat Creek Valley suggests that the Hat Creek fault is a 75°–85° dipping structure in the upper crust. The potential-field anomalies can be fit by dense, magnetic fill that is as much as 2 km thick, significantly more than the maximum thickness of the Hat Creek Basalt (∼50 m; Walker, 2008). Alternatively, the anomalies can be fit by a 75° dipping dense magnetic body that is 1–2 km east of relocated seismicity (Waldhauser and Schaff, 2008) that delineates a steep west-dipping structure. Given that the strands of the Hat Creek fault appear to step left through time, consistent with a component of right-lateral slip in an extensional environment, the location of seismicity relative to the youngest strand of the fault may suggest that deformation continues to step westward across the valley. Left steps along the west margin of the Fall River Mills basin are delineated by steep gravity gradients and suggest a style of deformation similar to that of the Hat Creek fault.
Magnetic data limit where right-lateral offsets associated with the Walker Lane can pass through the area north of the Pit River. The data reveal a 40-km-long linear gradient that separates the complex magnetic anomalies present over the young volcanic terrain and the smoother pattern over the Klamath Mountains. The source of the gradient is concealed and within the pre-Cenozoic basement. The absence of discrete offset within the resolution of the data (1–2 km) indicates that no significant right-lateral offset associated with the Walker Lane has propagated from the southeast across this boundary and, furthermore, no discrete offsets have been accommodated across this boundary since the Mesozoic.
Our interpretation of magnetic and new gravity data supports the hypothesis that Quaternary faulting is influenced by preexisting basement structure in the region between Mount Shasta and Lassen Peak (Blakely et al., 1997). New gravity data indicate that the southern margin of a gravity high in the eastern Klamath Mountains is stepped, with a newly delineated east-northeast–trending gradient extending parallel to the Pit River between Big Bend and California Highway 89. Across this gradient, Quaternary faults change to a more northwestward strike as strain is deflected into stronger crust, as suggested by higher gravity values. The change in faulting character also coincides with the southeast margin of a concealed magnetic high inferred to be a southeast-dipping fragment of ophiolite. Quaternary faulting becomes more distributed in nature and appears to avoid the body, presumably because its mafic composition is more resistant to deformation.
We acknowledge financial support by the Pacific Gas and Electric Company and the National Cooperative Geologic Mapping Program of the U.S. Geological Survey (USGS). Carson McPherson-Krutsky provided valuable field assistance, generously volunteering for field work after her internship with the USGS ended, and we thank Katherine (Kyeti) Morgan, a USGS intern, who helped with additional field work and physical property measurements. Reviews by Rick Blakely, Simon Kattenhorn, Cathy Busby, and an anonymous reviewer greatly helped improve the paper. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement of the U.S. Government.