The Pondosa fault zone: A distributed dextral-normal-oblique fault system in northeastern California, USA Open Access

The Cenozoic tectonic history of western North America is complicated, evolving from a convergent margin to a transform margin, with inboard extension in the Basin and Range (Atwater, 1970; Dickinson and Snyder, 1979). Northeastern California is located at an intersection of three primary tectonic regimes: Walker Lane dextral transtension, Basin and Range extension, and Cascadia convergence. The Walker Lane in northeastern California (Stewart and Ernst, 1988; Wesnousky, 2005) is a zone of distributed, northwest-directed transtensional faults located along the eastern edge of the Sierra Nevada that accommodates up to 10 mm/yr of relative North America–Sierra Nevada motion across a region ~135 km wide (Fig. 1; Kreemer et al., 2009). Previous studies have focused on characterizing the geologic, seismic, geodetic, and geophysical nature of deformation in northeastern California (Wells and Simpson, 2001; Unruh and Humphrey, 2017; Langenheim et al., 2016), yet questions remain regarding the regional style of deformation and the kinematics of late Pleistocene to Holocene active faults because of the prevalence of low-slip-rate, incompletely characterized, distributed fault systems (e.g., Pezzopane and Weldon, 1993).

Distributed low-slip-rate faults in high-relief terrain occur in parts of the Intermountain West and Pacific Northwest and can be challenging to identify (e.g., Sherrod et al., 2004; Gold et al., 2014; Bacon and Robinson, 2019; Horst et al., 2021). Whereas geodetic data can broadly identify zones or regions of deformation, these techniques generally are unable to detect and resolve individual fault strands in distributed fault systems. As light detection and ranging (LiDAR) data coverage expands, it allows us to fill in gaps in mapping and databases to identify and characterize faults in high-relief, vegetated terrain (e.g., Sherrod et al., 2004; Cunningham et al., 2006; Lin et al., 2013). Northeastern California has numerous faults visible on digital elevation models (DEMs) in low- and high-relief terrain not included in the U.S. Geological Survey (USGS) Quaternary Fault and Fold Database (USGS and CGS, 2020) or in the California Fault Activity Map (Jennings and Bryant, 2010). These faults are also poorly characterized in the Uniform California Earthquake Rupture Forecast, version 3 (UCERF3), fault model (Field et al., 2014), and the National Seismic Hazard Model (NSHM; Petersen et al., 2014, 2020). Mapping and understanding the extent, rate, and style of deformation on these faults are important due to their proximity to critical facilities and hydropower infrastructure in addition to interpreting the regional tectonic processes.

In this study, we identified, mapped, and characterized active faults near the Pit River in northeastern California to better understand the character, style, and rates of active and recent deformation. We documented the geologic expression of late Pleistocene faulting in the Mushroom Rock region (defined as the area shown in Fig. 2B) in northeastern California located between Mount Shasta and Lassen Peak. We focused on the region north of the Pit River, where the strike of newly recognized Quaternary faults rotates distinctly counterclockwise to a west-northwest orientation from the generally north-northwest strike of faults in the Hat Creek graben south of the Pit River (i.e., northern end of Rocky Ledge and Hat Creek fault zones; Fig. 2B). We integrated new fault mapping from 1-m-resolution bare-earth LiDAR and 10-m-resolution National Elevation Data set (NED) DEMs, structure-from-motion (SfM) models of key outcrops, field-based bedrock mapping, geochemistry and ⁴⁰Ar/³⁹Ar dating of volcanic rocks, (U-Th)/He apatite thermochronology of bedrock terranes, and field observations with modern geodesy and earthquake focal mechanisms to document the style and timing of faulting. Our data revealed evidence of late Pleistocene dextral-normal-oblique faulting within the Mushroom Rock region, called the Pondosa fault zone, which may represent a dextral-oblique accommodation zone between north-south–oriented Basin and Range extensional fault systems and/or part of the Sierra Nevada–Oregon Coast block microplate boundary.

The study area encompasses the intersection of the northernmost Sierra Nevada microplate, the southernmost Oregon Coast block microplate, Basin and Range extension, and Walker Lane dextral transtension. The Mushroom Rock study area is located at the boundary between the Modoc Plateau and the Klamath Mountains and spans the southernmost Cascade Ranges (Fig. 2); Mushroom Rock is a peak in the center of the study area with an elevation of 1900 m (Fig. 1). Mushroom Rock ridge is an east-west–trending ridgeline (Fig. 2B). The Mushroom Rock region is bounded by the McCloud and Pit Rivers to the north and south, respectively, and by Grizzly Peak and Highway 89 to the west and east, respectively (Fig. 3; Mushroom Rock region spans the area of Fig. 4). Near the southern end of the study area, the Pit River flows east to west. Kosk Creek, a tributary to the Pit River, drains the central part of the study area (Fig. 3).

The study area is characterized by Paleozoic through Jurassic metasediments and Tertiary clastic sedimentary rocks that are overlain by Miocene through Quaternary volcanic deposits. The Mushroom Rock region straddles the eastern edge of the Eastern Klamath terrane, which is composed of fragments of island-arc and ophiolitic lithosphere that accreted in the early Paleozoic through Middle Jurassic (Fig. 3; Irwin, 1985). The Klamath terranes in northern California and southern Oregon are broadly described as westward-younging series of thrust sheets bounded by gently east-dipping thrust faults and composed of Paleozoic through Late Jurassic accreted oceanic terranes (Irwin, 1972, 1985, 1994). Numerous postaccretionary plutons intrude the various belts and provide age control for their accretion during the late Paleozoic and Mesozoic westward growth of North America (Irwin, 1985; Snoke and Barnes, 2006).

The Eocene to Late Cretaceous Montgomery Creek Formation (Gay and Aune, 1958; Sanborn, 1960; Higinbotham, 1986; Jennings et al., 2010) is exposed in the headwaters of Kosk Creek and along the Pit River. This sandstone and conglomerate forms the lowest stratigraphic unit exposed in the headwaters of Kosk Creek (Fig. 3). The Montgomery Creek Formation is interpreted as a braided river deposit fed by erosion of the early Klamath mountains (Higinbotham, 1986).

Much of the present-day landscape is capped by andesite and basalt related to Cascadia volcanism and Basin and Range extensional deformation (Fig. 3). Modern Cascade volcanic arc centers, such as Mount Shasta, Lassen, and Medicine Lake, formed within the past ~850 k.y. (Clynne and Muffler, 2010, 2017; Muffler and Clynne, 2015; Christiansen et al., 2017) adjacent to regional east-west backarc spreading or Basin and Range extension (Fig. 2A; Guffanti et al., 1990; Blakely et al., 1997). Although the volcanic rocks comprising the Mushroom Rock ridge (Fig. 2B) were undated prior to this study, tholeiitic basalts that fill the erosional topography surrounding the ridge range in age from ca. 1.8 Ma to ca. 262 ka (Clynne and Muffler, 2017; Downs et al., 2020). The basalt of Pole Creek, dated at 991 ± 12 ka (Clynne and Muffler, 2017), is the most spatially extensive unit and serves as a key marker flow recording Pleistocene fault offset. Additional bedrock units and stratigraphy mapped in this study, including the Miocene Curtis Lake and Red Mountain units and the Miocene–Pliocene Bartle Gap unit, are described in the Results section and in the Supplemental Material.¹

A distinct and pervasive system of Quaternary faults crosses the study region (Jennings and Bryant, 2010). Few of the faults have been studied in detail, however, and most are unnamed (Jennings and Bryant, 2010). Numerous faults south of the Pit River have been the subject of site-specific studies for seismic hazard analyses (e.g., Kozaci et al., 2014; Lahontan GeoScience, Inc., 2012; Lettis Consultants International, 2013; O’Brien, 2013; Pacific Gas and Electric Company, 2018); however, some of these data and reports are not publicly available.

The 47-km-long, west-dipping Hat Creek fault zone forms the eastern edge of the Hat Creek graben (Fig. 2B). The Hat Creek fault zone has a primarily normal sense of slip, although there is evidence for dextral-oblique offset (Gray et al., 2017; Blakeslee and Kattenhorn, 2013). The fault forms a west-facing escarpment, up to ~490 m high, in middle Pleistocene volcanic rocks (940 ± 24 ka; Clynne and Muffler, 2010) and displaces units as young as the late Pleistocene Hat Creek Basalt (24 ± 6 ka; Turrin et al., 2007) and ca. 15 ka periglacial deposits (Muffler et al., 1994). Estimates of Quaternary dip-slip rates range from 1.2–1.9 mm/yr (Turrin et al., 2007) to 2.2–3.6 mm/yr (Blakeslee and Kattenhorn, 2013). Recent constraints from geodetic data indicate that the modern horizontal slip rate is <1 mm/yr (Thatcher et al., 2014).

The ~17-km-long Rocky Ledge fault zone forms a portion of the western boundary of the Hat Creek graben (Fig. 2B; Wills, 1990; Sawyer and Bryant, 1995; Martin, 2020). The Rocky Ledge fault zone also has a primarily normal sense of slip along an east-dipping primary fault, with some evidence for dextral offset (Martin, 2020). The fault scarp is ~50 m high and offsets the Rocky Ledge flow, dated to 200 ± 8 ka (Muffler et al., 2012, 2017; Muffler and Clynne, 2015).

The Mayfield fault zone is ~32 km in length, strikes 165°, and dips to the west at a high angle (Fig. 2B; Sawyer et al., 1995; Page, 1995). The fault zone has a primarily normal sense of slip and is considered to have been active in the Holocene because it offsets the Giant Crater flow, dated at 12,430 ¹⁴C yr B.P. (Donnelly-Nolan et al., 1990), by as much as 10 m vertically, yielding an estimated maximum slip rate of ~1 mm/yr.

Sparse and scattered seismicity in the region is not easily attributed to mapped faults or folds (Fig. 2B). A regional analysis of relocated microseismicity and focal mechanisms revealed that the Mushroom Rock region comprises primarily transcurrent and transtensional domains (Unruh and Humphrey, 2017). Localized transpression occurs well southwest of the study area in a region of recognized contractional Quaternary faulting (Inks Creek fold belt, Red Bluff and Battle Creek faults; Fig. 2A; Helley et al., 1981; Angster et al., 2021). Kinematic analysis by Unruh and Humphrey (2017) showed that focal mechanisms near Mushroom Rock ridge, derived from earthquakes ranging from magnitude (M) 2.5 to M 4.3 at depths of 3.1–21.7 km, display primarily dextral to dextral-normal sense of slip on steeply dipping faults that strike southeast-northwest or east-west (or sinistral-oblique slip on southwest-northeast–striking faults; Fig. 2B).

Historical small and moderate (M 4–5.9) earthquakes are rare in the study area (Fig. 2A). The 1978 M 4.6 Stephens Pass earthquake sequence near Medicine Lake, ~30 km north of the Mushroom Rock ridge, is the only recorded historical surface-rupturing event in the region (Fig. 2A). This earthquake occurred on the east-dipping Stephens Pass section of the Cedar Mountain fault, creating a 2-km-long surface rupture characterized by a 1-m-deep and 4.5-m-wide zone of normal faulting (Bennett et al., 1979). A series of earth-quake sequences occurred near Lassen Peak, south of the study area, in 1936, 1945–1947, and 1950; the last two of these culminated in M 5.0 and M 5.5 earth-quakes, respectively (Norris et al., 1997). The similarity of the magnitudes and time histories of these sequences to others in the Basin and Range indicates tectonic, rather than magmatic, earthquake processes (Norris et al., 1997). No historical large (>M 6.0) earthquakes have been recorded in the study area, although the Hat Creek fault zone has an estimated earthquake potential of moment magnitude (M_w) 6.7 (Blakeslee and Kattenhorn, 2013).

Patterns of regional strain interpreted from space geodesy indicate that northeastern California is actively deforming. Global navigation satellite system (GNSS) and regional-scale strain mapping show that the Mushroom Rock region is a zone of crustal shear and transtension (Kreemer et al., 2014; Zeng and Shen, 2017). North of the Pit River, regional geodetic studies have shown that deformation relative to stable North America and corrected for Cascadia subduction zone coupling is dominated by a northwest velocity component (Poland et al., 2006; McCaffrey et al., 2013), reflecting clockwise rotation of the Oregon Coast block microplate (Fig. 5). South of the Pit River, velocities reflect translation and counterclockwise rotation of the Sierra Nevada microplate relative to stable North America (Argus and Gordon, 1991; Dixon et al., 2000; Williams et al., 2006).

Published bedrock mapping in the region between the Medicine Lake and Lassen Peak areas (Fig. 2) is limited to Quaternary volcanics (basalts and calc-alkaline cones; e.g., Champion et al., 2017; Muffler et al., 2017), local work on the Klamath terranes and Montgomery Creek Formation to the south of Mushroom Rock ridge (e.g., Sanborn, 1960), and detailed maps of Lassen Volcanic National Park (Clynne and Muffler, 2010) and Medicine Lake volcano area (Donnelly-Nolan, 2010) (Fig. 3; see Fig. S1 for outlines of previous mapping efforts, ^{footnote 1}). Geologic mapping is not entirely consistent in the study area along Mushroom Rock ridge (Gardner, 1964; Gay and Aune, 1958; Jennings et al., 2010). Resolving these inconsistencies (summarized in the Results and described in detail in the Supplemental Material) in the bedrock mapping was an essential first step in our investigation to understand the Miocene–Pleistocene volcano-tectonic history of the area.

We mapped the bedrock (i.e., older than late Pleistocene metamorphic and volcanic units) to place maximum age constraints on the timing of faulting. We synthesized previous regional and local mapping (Gardner, 1964; Jennings et al., 2010; Champion et al., 2017; Christiansen et al., 2017; Muffler et al., 2017) and added details along the Mushroom Rock ridge (Fig. 2; area mapped in this study shown in Fig. 3; Table S1, ^{footnote 1}).

In addition, we collected samples for ⁴⁰Ar/³⁹Ar dating, geochemistry, and petrographic description to help distinguish between different volcanic units along the Mushroom Rock ridge (Table 1; Tables S2 and S3; Figs. 3 and 4). Geochemical samples were analyzed at the GeoAnalytical Laboratory at Washington State University. The ⁴⁰Ar/³⁹Ar samples were analyzed at the U.S. Geological Survey Menlo Park facilities. Radiometric ages cited throughout the paper have been recalculated to the same standards as results reported herein (GA1550 biotite = 98.79 Ma; Renne et al., 1998; Fleck et al., 2019). See the Supplemental Material for more details on the ⁴⁰Ar/³⁹Ar dating and geochemical sample analysis (Table S4).

To better understand faults in the region, we mapped scarps and topographic lineaments (hereafter “lineaments”) from topographic and imagery data in the Mushroom Rock study area (Figs. 2 and 3). In addition to the 10-m-resolution NED DEM, we used two <1-m-resolution bare-earth DEMs derived from airborne LiDAR data spanning eastern Mushroom Rock ridge (Figs. 2 and 3; see Fig. S1 for LiDAR extents, ^{footnote 1}) and two publicly available 1-m-resolution bare-earth LiDAR DEMs covering Mount Shasta (DOGAMI, 2010) and spanning the region south of Medicine Lake in the Pondosa/Hambone regions (NoCal Wildfire GMEG 2018 LiDAR; U.S. Geological Survey, 2019). We mapped scarps and lineaments in the topographic data as abrupt topographic steps that commonly crossed elevation contours and sometimes created sediment ponding in fluvial systems. We mapped lineaments of uncertain origin as topographic disruptions in the LiDAR data or tonal contrasts in aerial and satellite imagery. We field-checked lineaments to determine if they were likely tectonic scarps, and if so, we collected additional relevant geomorphic and stratigraphic data. In volcanic topography, eroded contacts, flow margins, sills, and dikes may also appear as scarps and lineaments in the landscape; thus, field investigation was necessary to determine whether lineaments corresponded to volcanic or possible tectonic features.

We classified each scarp and lineament based on our level of confidence that the features were tectonic in origin. We mapped scarps that clearly offset correlative surfaces in the field or on LiDAR data, and that crosscut elevation and offset several surfaces of various ages and could not be attributed to a fluvial, volcanic, or mass movement origin, as high-confidence faults. We mapped scarps that did not clearly offset a correlative surface and where fluvial, volcanic, or mass movement origin could not be entirely ruled out as medium-confidence (or inferred) faults. We mapped lineaments identified on LiDAR data that could not be field verified due to land access restrictions, field conditions, or vegetation masking as low-confidence (or queried) faults.

To facilitate bedrock mapping in inaccessible areas south of Mushroom Rock ridge in the Kosk Creek headwaters and near Red and Bald Mountains (Fig. 3), we created six SfM models of the volcanic units using photographs taken from a helicopter at three key outcrops: Curtis Lake North, Nelson Slides, and Red Mountain (Fig. 4). Models were created using imagery collected with a Nikon D750 digital single-lens reflex (DSLR) camera and a GoPro HERO7 camera (Table S5, ^{footnote 1}). The models were created using Agisoft Metashape Professional v1.5 software and exported as LAS (laser) red-green-blue (RGB) point clouds. To georeference the point clouds, we coregistered the SfM model point clouds to the Mushroom Rock LiDAR data set point cloud using an iterative closest point algorithm as implemented in CloudCompare v2.10.2. We estimate uncertainties of 5° in the measurements derived from the SfM models. The models are available as a USGS ScienceBase Data Release² (Data Sets S1, S2, and S3) (Thompson Jobe et al., 2022).

To characterize exhumation and uplift and constrain crustal cooling related to tectonics in the eastern Klamath Mountains and Mushroom Rock region, we collected two low-temperature thermochronology (apatite [U-Th]/He or AHe) samples from different elevations to represent different structural depths (Table 2; Fig. 3) from Triassic and Jurassic sedimentary rocks exposed in the Kosk Creek drainage (Gay and Aune, 1958; Sanborn, 1960; Gardner, 1964). We then determined probable thermal histories using the QTQt inverse modeling software (Gallagher, 2012). In addition to the AHe ages from the samples, we used deposition and burial during the Jurassic (ca. 200–150 Ma) as model constraints, because the sampled units are Triassic (Gay and Aune, 1958; Jennings et al., 2010), and we explored scenarios of heating from the overlying volcanic units. Additional details on the constraints and QTQt software are provided in the Supplemental Material (see ^{footnote 1}).

Bedrock descriptions and ages are described below; additional details, field photographs, and geochemical data on the bedrock units are available in the Supplemental Material.

The Curtis Lake unit (Tvcl) is a >300-m-thick volcaniclastic unit with inter-bedded andesite flows, tuffs, and well-stratified volcanic breccias that extend from north to south for over 20 km in the Kosk and Nelson Creek headwaters (Fig. 4). The beds generally dip ~20°–25° east-southeast and unconformably overlie the Montgomery Creek Formation.

The Red Mountain unit (Tvrm) is a >100-m-thick sequence of andesite flows that generally dip 10°–30° southeast where exposed on the north side of Red Mountain; this unit lies unconformably above the Curtis Lake unit. A ⁴⁰Ar/³⁹Ar age from a sample (VR-f; Fig. 4) collected near the top of Red Mountain is 9.6 ± 0.3 Ma (Table 3; Table S4, ^{footnote 1}).

The Bartle Gap unit (Tvbg) spans the study area north of the Pit River, west of Highway 89, and north to the McCloud River, except for the area excavated by Kosk Creek (Fig. 4), and it comprises the higher topography that is dissected by Quaternary-active faults. The base of the Bartle Gap unit has variable volcanic lithology, including andesite and basaltic andesite flows, small scoria cones, tuffs, and volcaniclastic and pyroclastic beds (this study; Gardner, 1964). In general, andesite flows, tuff, and volcaniclastic beds are interbedded with a total thickness of >300 m. Individual beds have variable orientations, although on the northern side of Mushroom Rock ridge, the basaltic andesite flows generally dip 10°–20° north, northeast, and east. The ⁴⁰Ar/³⁹Ar ages from three samples ranged between 6.2 and 4.9 Ma (Table 3; Table S4, see ^{footnote 1}) and were collected from volcanic flows near the stratigraphic top of the Bartle Gap unit (VR-a, VR-b, VR-d; Fig. 4). Another sample from a dike (VR-g; Fig. 4), dated to 6.3 ± 0.2 Ma, is similar to the age of a sample (VR-a; Fig. 4) from the overlying volcanic flow of 6.2 ± 0.2 Ma (Fig. S2).

Soldier Mountain, located east of Mushroom Rock ridge, is calc-alkaline andesite (labeled Qev near Soldier Mountain on Fig. 3). A ⁴⁰Ar/³⁹Ar age from a sample (VR-e; Fig. 3) collected near the top of Soldier Mountain is 2.04 ± 0.01 Ma (Table 3; Table S4).

In this section, we outline the observations from the three techniques we used to identify and characterize the faults. For each technique, we first describe the observations, followed by a section summarizing the interpretation of the observations.

We mapped >50 scarps in the Mushroom Rock region (Figs. 6 and 7; Data Set S4, ^{footnote 1}), most of which we interpret to be tectonic in origin. Fault traces form two distinct sets that strike northwest-southeast and north-south (Fig. 7). Our mapping shows that the north-south–striking faults generally lie north and south of Mushroom Rock ridge, whereas faults within the eastern Mushroom Rock ridge strike northwest-southeast (Fig. 7C). Fault scarps are typically spaced ~300–1200 m apart and are commonly subparallel but also display both left- and right-stepping patterns, with <200 m between adjacent fault scarps or features. Individual mapped fault traces are typically 1–4 km in length, with a maximum length of ~10 km, and they generally strike northwest-southeast or north-south. Fault zones (including left- and right-stepping surface traces) reach a maximum length of 25 km. Faults that strike northwest-southeast are generally subparallel, with some faults exhibiting left-stepping patterns, whereas faults that strike north-south are also generally subparallel, but they display more right-stepping patterns.

We focused on the set of northwest-southeast–striking faults to the north of the Pit River that interrupts the dominantly north-south tectonic fabric across the region (yellow box on Fig. 7C). We term this collection of newly mapped faults the Pondosa fault zone. The Pondosa fault zone is an ~30-km-wide, ~20-km-long zone of distributed and discontinuous faulting (Fig. 6 and 7). Two populations of faults are recognized in the Pondosa fault zone: (1) subparallel en echelon faults that displace the 0.991 Ma basalt of Pole Creek (Champion et al., 2017), 0.3055 Ma basalt of Hammond Crossing (Downs et al., 2020), and ca. 12.4 ka Giant Crater flow (labeled 1 on Fig. 7B; age from Donnelly-Nolan, 2010); and (2) faults that extend into the high, deeply dissected terrain underlain by the Miocene–Pliocene (6.3–4.9 Ma) Bartle Gap unit south of Highway 89 above ~1300 m elevation (labeled 2 on Fig. 7B). The faulted portion of the Bartle Gap unit has an approximately northwest-southeast preferred orientation of captured and deflected drainages, asymmetric canyons, and perched basins with apparently young alluvial fill, which indicate fault-controlled topography. Faults in the higher-elevation terrain are expressed as discontinuous, block-bounding scarps or oversteepened slopes, some of which can be traced across late Quaternary, inset alluvial deposits (described below in Observations from Field Mapping section).

Scarps and lineaments within the high topography in the Pondosa fault zone, just east of Kosk Creek, have medium or low confidence ratings (Fig. 6B). These scarps and lineaments are influenced by the high rate of slope transport processes, variable dips and lithologies within the Bartle Gap unit, and the possibility that scarps have formed due to strength contrasts in layered volcanics rather than faulting. Scarps and lineaments in the Kosk Creek headwaters offset the Bartle Gap and Curtis Lake units and are clearly tectonic in origin. Within the relatively flat-lying basalt of Pole Creek, we generally mapped scarps and lineaments with a high degree of confidence of tectonic origin because scarps are continuous for several kilometers based on LiDAR and field mapping and offset correlative surfaces.

We mapped units on the SfM models based on bedding orientation and thickness, apparent layer resistance to weathering, clast size and orientation, and color. The thicknesses and orientations of the Curtis Lake volcaniclastic unit (Tvcl) and, to a lesser extent, the Bartle Gap unit (Tvbg) change along the outcrop and are offset by faults. We describe the observations from the Curtis Lake North and Red Mountain sites in detail here, and the observations from the Nelson Slides site can be found in the Supplemental Material (see ^{footnote 1}).

Along the Curtis Lake North outcrop exposure (Area A polygon on Fig. 4; see also Figs. 8 and 9), we defined three stratigraphic packages within unit Tvcl (Fig. 9B) and measured apparent thicknesses and orientations where possible. A chaotic altered package (I) is present at the base of the section. Bedding within this package is difficult to discern, and outcrop exposures are heavily altered. Above this package, there is a well-bedded volcaniclastic unit (II) (Fig. 9D). The apparent unit thickness varies from ~50 m in the north to ~25 m in the south (Figs. 8A and 9F). Bedding orientations vary in strike but are relatively consistent in dip, except for the northernmost exposures, where beds strike 043° and dip 46° southeast. In the central area, beds strike 052° and dip 27° southeast, and at the southern end of the exposure, beds strike 077° and dip 23° south-southeast (Fig. 8A). The next stratigraphic package (III) is defined by a thick basal unit that we interpret as a tuff, based on the rounded and smooth outcrop exposure and lighter color compared to the other volcanic deposits in the region (Fig. 9D). The basal tuff has an apparent thickness of ~10–15 m and can be traced capping volcaniclastic unit II along the length of the outcrop. Above the basal tuff, interbedded lighter- and darker-colored intervals are visible; we interpret the lighter-colored units as tuff and the darker units as lava flows. In total, this package (III) is 150+ m thick, with the uppermost part of the package defining the ridgeline (Fig. 9D). At the top of the outcrop, in only a few locations, the Bartle Gap unit (package IV) dips gently northeast at 6° (Fig. 9F).

We observed four discontinuities within the stratigraphic units that define four lateral domains in the Curtis Lake North stratigraphic packages (Fig. 9B). We interpret these discontinuities as faults because we observed lateral truncations, offset of units, and thickness changes. For convenience, we numbered the domains from 1 in the north to 4 in the south (Fig. 9B). We used stratigraphic package II as a marker bed to measure thickness changes and offsets because it is prominent with visible individual beds.

The first fault (A) is defined by a light-toned vegetation lineament that crosscuts packages I, II, and III (Fig. 9). The fault strikes 135° and dips 60° south-west (Fig. 8A). We could not confidently correlate offset markers to estimate displacement, and a prominent vertical component is not recorded along the surface projection of the fault. We interpret the lighter-toned bedrock (and possibly soil) along the fault to be alteration or fault zone damage.

The second fault (B) strikes ~300° and dips steeply (60°–80°) to the northeast (Fig. 9). Fault B aligns with a prominent steep drainage that separates domains 1 and 2. Stratigraphic packages II and III display apparent offset across the steep drainage, but mapping of the stratigraphic packages within the fault zone is difficult. Stratigraphic package II changes apparent thickness from ~50 m in domain 1 to ~30 m in domain 2 (Figs. 8A and 8B) and appears to be vertically offset by ~100 m across fault B. In addition, the orientation of the bedding changes across fault B (043/46 in domain 1 to 052/27 in domain 2; Figs. 8A and 8B). Fault B is aligned with a prominent fault escarpment ~100 m high east of the Kosk Creek headwaters (areas shown in Figs. 8A, 9A, 9B, and 9F).

The third fault (C) separates domains 2 and 3 and strikes ~340°, creating a prominent vegetated drainage (Figs. 9C and 9D). Package II changes apparent thickness from ~30 m in domain 2 to ~25 m in domain 3 across fault C (Figs. 8A and 8B). Although correlation of individual beds within package II across the vegetated drainage formed along fault C is challenging, we estimate the top of package II has apparent vertical separation of ~35 m with a down-to-the-northeast sense of slip and apparent horizontal separation of ~10 m across the fault.

The fourth fault (D) is located at the southernmost end of the exposure and separates domains 3 and 4. Packages II and III are easily correlated across fault D due to prominent thin lighter-colored beds interbedded with volcaniclastic beds and an apparent vertical separation of ~50 m down to the west (Fig. 9D). A subtle vegetation lineament and juxtaposed rocks/soils of different colors (light brown against tan) project down the slope, at a strike of ~350°–355° and a steep dip (73°) to the northeast. This lineament can be followed across the valley until fault D intersects with fault A at the northern end of the exposure.

The Red Mountain outcrop model, located north-northwest of Red Mountain (Area B polygon on Fig. 4; see Figs. 10A and 10B), reveals two stratigraphic packages: (1) rm-I, the volcaniclastic unit at the base, and (2) rm-II, a thick series of volcanic flows at the top. The flows at the top of the section have an apparent thickness of ~1–10 m, whereas the thicker lower flows exhibit columnar jointing. The flows strike ~040° and dip 8° southeast. The stratigraphy is cut by a fault striking ~315° and steeply dipping northeast that projects into a vegetated steep drainage. Volcanic flow packages are present on either side of this fault (rm-II), but they have different character and do not correlate. On the southwest side of the fault, the flows are thicker, and several flows have columnar jointing. On the northeast side of the fault, flows are generally thinner, no columnar jointing is apparent, and the rock is lighter and pinker in color. A lineament to the southeast, ~4 km long, mapped on the LiDAR data (Figs. 10C and 10D), also strikes ~315° and projects into the fault mapped on the SfM model, but it cannot be traced across the intervening canyon. The lineament has a maximum vertical separation of ~30 m.

We found clear evidence of distinctive map units and faults across the three high-resolution outcrop models. A continuous, correlative set of stratigraphic packages exists within the Curtis Lake unit (Tvcl) across the Curtis Lake North, Red Mountain, and Nelson Slides models. The Curtis Lake unit (Tvcl) is in turn overlain by the Bartle Gap unit (Tvbg) in the Curtis Lake North model (Fig. 9) and the Red Mountain unit (Tvrm) in the Red Mountain (Fig. 10) and Nelson Slides models (Fig. S5). At least two of these outcrop models (Curtis Lake North and Red Mountain) record evidence of oblique slip through thickness changes and vertical separations across the faults.

The orientation of the Curtis Lake unit is consistent at all sites and does not vary considerably north or south of Kosk Creek. Generally, the basal volcaniclastic flows and overlying interbedded flows and tuffs of the Curtis Lake unit strike north-northeast and dip 8°–40° southeast. At least four faults observed on the outcrop models at Curtis Lake North and Red Mountain align with adjacent fault surface traces mapped on the LiDAR data.

At the Curtis Lake North outcrop, the stratigraphic packages are vertically offset ~30–100 m each by at least four faults that generally trend north/northwest-south/ southeast, consistent with the newly mapped fault network. Additionally, fault B in the outcrop model is aligned with a fault mapped in the LiDAR data, which forms a prominent, ~100-m-high east-west lineament in the landscape (Figs. 9E and 9F) with the same relative vertical offset (down-to-the-north) and magnitude. Because of the variability in thickness and orientation of the stratigraphic packages on either side of the fault zone, and because the zone between faults A and B is relatively wide and does not display clear stratigraphy, we infer that fault B has a lateral component, in addition to the apparent normal offset observed in outcrop and on the LiDAR DEM (the 100-m-high linear ridge; Fig. 9). The lack of discrete lateral offsets along the LiDAR lineament may be explained by the lack of good strain markers or piercing points, such as channels or gullies, to record offset. In addition, evidence of landsliding at the western end of the escarpment is clear; landslides and colluvial processes in the steep terrain may mask any lateral offsets.

At Red Mountain, the rm-II stratigraphic package, which we interpret to be the ca. 9.6 Ma Red Mountain volcanic unit, exhibits different apparent thicknesses and textural characteristics (i.e., the southwest side has columnar jointing) on either side of the fault. A mapped surface fault to the southeast projects northwest into the fault visible in the exposure (Figs. 10C and 10D). We interpret highly oblique to pure strike-slip motion on this steeply dipping fault due to the little apparent vertical offset on the stratigraphic packages and in the topography, near-vertical dip, and the mismatch of stratigraphic packages across the fault. The lack of lateral offset observed along the lineament may result from the lack of features, such as channels, that could record lateral motion, in addition to the higher-relief topography with ongoing colluvial processes that may mask or erode subtle lateral offsets. Alternatively, flows can have highly variable thicknesses and textures; the mismatch we observe in the stratigraphic packages across the fault could also result from a vertical sense of offset on a highly variable volcanic flow.

Approximately 20 scarps of the Pondosa fault zone were evaluated in the field (Fig. 6; Table S6, see ^{footnote 1}). Although we interpreted most of the scarps to be tectonic in origin (Fig. 11), several may be nontectonic flow margins and fluvial scarps. We depict these as probable nontectonic scarps (queried or low-confidence faults; Fig. 6) to contrast them with scarps interpreted as tectonic and to highlight features in the landscape that might be misinterpreted as having a tectonic origin.

In general, fault scarps extend across drainages and displace young alluvial and colluvial deposits (Fig. 11) by as much as 5 m vertically. At some locations, the mapped faults coincide with distinct soil color changes, closed basins, sag ponds, and springs near the base of hillslopes (Fig. 6).

On the northern side of Mushroom Rock ridge, several northwest-striking scarps and tectonic lineaments extend northward and offset the alluvium and middle Pleistocene basalt of Deadhorse Canyon and overlying alluvium (1.483 ± 25 Ma; Christiansen et al., 2017). Where the fault offsets alluvium (Figs. 11A and 11B), vertical displacements along the densely vegetated scarp decrease from 5 m at the southern end (Fig. 11D), where the scarp first exits the high topography, to 1 m at the northwestern end (Fig. 11C). At this site, the scarp intersects the flow direction of an active stream channel at a high angle, indicating the scarp is not a fluvial terrace, but preservation may be affected by fluvial processes. The scarps to the northeast juxtapose the Miocene–Pliocene Bartle Gap unit against middle Pleistocene basalt of Deadhorse Canyon, with an apparent minimum vertical separation of ~10 m (Fig. 11D). To the northeast and along strike to the northwest, down-to-the-northeast scarps offset the basalt of Deadhorse Canyon and pond Quaternary alluvium upstream of the scarp. We were unable to document any offset of the Quaternary alluvium where these lineaments intersect the channels (Fig. 11A).

A fault strand offsets a small alluvial fan with <2 m of vertical separation on the northeastern side of the Pondosa fault zone (Figs. 11G and 11H). Although these deposits are undated, we interpret them to be late Pleistocene or Holocene based on their fresh and youthful morphology and landscape position. To the southeast along strike of this fault, the offset alluvial fan coincides with a steep, linear ridge, possibly fault controlled, which offsets early Pleistocene basalt of Pole Creek with as much as ~30 m of vertical separation (Fig. 11H).

Mapped fault scarps show predominantly vertical separation, with limited evidence for lateral motion. The Curtis Lake North and Red Mountain outcrops provide the strongest evidence for lateral fault motion. We speculate that the lack of discrete lateral offsets on many of the faults may arise from a lack of preservation due to a lack of reliable strain markers or piercing points or a high vertical-to-lateral slip ratio.

The oldest faulted volcanic units place maximum age constraints on the timing of faulting along the Pondosa fault zone (Fig. 7B). The youngest faulted deposits place constraints on the recency of fault activity. In the higher-relief, more-dissected portion of the fault zone, faults offset the Miocene–Pliocene Bartle Gap unit, indicating that faulting postdates 4.9 Ma. The fresh geomorphic expression and sharp topographic profiles of these scarps indicate that faulting is likely much younger and may be as young as late Pleistocene, based on fault scarps formed in alluvium along strike. Faulting clearly postdates the basalt of Pole Creek, basalt of Hammond Crossing, and the Giant Crater flow in the geomorphically best-expressed part of the fault zone. We did not observe any faulting of inset alluvium or terraces in this area.

The Pondosa fault zone offsets the ca. 12.4 ka Giant Crater flow (approximately encompassed within unit Qrv, with flow outline shown on Fig. 3; Donnelly-Nolan, 2010) vertically by <8 m over its western extent (Figs. 12C and 12D). At least part, if not most, of this offset appears to be due to a remnant older flow preserved in the footwall of the fault in at least one location; thus, the 8 m apparent vertical offset is likely an overestimate of the actual fault offset. The Giant Crater flow is vertically offset by larger amounts on its eastern and southeastern margins, where the Holocene-active Mayfield and McArthur faults, respectively, displace the flow (Page, 1995). This observation indicates that most of the fault network in the eastern part of the study area has experienced at least one surface-rupturing event since ca. 12.4 ka.

We report AHe measurements from five grains from each of the two samples from the Klamath terrane basement in Table 4. Grains from the lower sample, T-a, have ages between 18.0 and 9.7 Ma, correcting for effects of ⁴He ejection (note that plots of model results in Fig. S7 present uncorrected He ages, see ^{footnote 1}). Ages from the upper sample, T-b, span 98.3–14.4 Ma; however, the oldest grain is much older than the younger four grains from that sample, for which ages range from 41.3 to 14.4 Ma.

Thermochronology samples from elevated topography near Grizzly Peak (Fig. 3) yielded relatively consistent (Table 4) Miocene ages. When modeled, the AHe data from the Klamath terrane basement below the Mushroom Rock ridge are consistent with a simple history of burial reheating following sedimentation, followed by a protracted period of relatively slow cooling (e.g., 2–3 °C/m.y.) for as much as tens of millions of years, although the data and models used here do not resolve the start of this cooling. While more complicated thermal histories can reproduce the data, the least complicated models indicate minimal (~0.02 km/m.y.) tectonic exhumation during the Miocene. See the Supplemental Material for further details on the results and models.

The distributed, northwest-southeast–striking, dextral-normal-oblique, late Pleistocene–active Pondosa fault zone disrupts the regional pattern of north-south–striking normal faults north and south of the Pit River (Fig. 7). Several faults in the Pondosa fault zone offset late Pleistocene geomorphic surfaces and deposits (Fig. 11).

Despite evidence for active faulting in the region, the fault kinematics remain ambiguous. Focal mechanisms from small earthquakes indicate oblique-dextral slip (Unruh and Humphrey, 2017) and align with newly mapped faults. Vertical offsets and thickness changes in volcanic stratigraphy in the headwaters of Kosk Creek indicate oblique slip that aligns with topographic scarps. Regionally, prominent north-south–striking faults, such as the Hat Creek and Rocky Ledge faults, record primarily dip-slip normal motion (Kattenhorn et al., 2016; Martin, 2020) with limited recently recognized dextral movement on more northwest-southeast–striking fault strands (Gray et al., 2017; Martin, 2020). On the other hand, northwest-southeast–striking faults in the Walker Lane, such as the Likely and Honey Lake faults (Fig. 2; Wills and Borchardt, 1993), generally accommodated significant dextral movement during the late Pleistocene and Holocene between regions of north-south–striking normal faulting. Moreover, the en echelon fault pattern is indicative of an oblique-slip environment (e.g., Crider, 2001; Wilcox et al., 1973), similar to fault patterns observed in other oblique-slip environments (e.g., Summer Lake region, Oregon; Pezzopane and Weldon, 1993).

We suspect the apparent lack of obvious dextral slip on individual strands of the Pondosa fault zone, despite the clear oblique slip observed in the Kosk Creek exposures, reflects a lack of preservation of the lateral component due to a higher vertical component on most individual fault strands or a lack of geomorphic strain markers or piercing points. Seismic reflection data (Kozaci et al., 2014) and outcrop observations (O’Brien, 2013) also provide evidence for oblique slip on similarly oriented faults south of the Pit River. Regardless, total right-lateral offset across the fault system may be limited based on observations that a regional aeromagnetic lineament, the Eastern Klamath boundary, does not appear to be offset more than 1–2 km (Langenheim et al., 2016).

We estimated minimum vertical separation rates for the Pondosa fault zone using the dated volcanic flows and deposits and scarp heights measured from topographic profiles across faults. These rates are subject to limitations; however, the calculations are useful for estimating deformation rates across the Pondosa fault zone. We focused on two representative transects across the Pondosa fault zone that extend from higher topography in the Bartle Gap unit northeastward to the late Pleistocene volcanic units (Fig. 12). Along these transects, we estimated early Quaternary vertical separation rates across individual faults and calculated net rates across subsections of the Pondosa fault zone. We measured scarp heights to approximate minimum vertical separations across faults in the Quaternary low-potassium olivine tholeiite (LKOT) basalts. We estimated the net vertical separation rate since the Pliocene on the Bartle Gap unit using the overall elevation difference across all fault traces. However, these rates should be treated with caution for several reasons: (1) Scarp heights may underestimate the true vertical separation or slip across the fault due to recent alluvial and colluvial deposition in the hanging wall against the base of the scarp; all measured offsets are then considered minimums because of the unknown amount of Quaternary alluvium. (2) The rates do not account for the lateral component of the total fault slip; the apparent vertical separation rate represents a minimum of the total slip rate. (3) Because the fault dips are poorly constrained, we cannot estimate a vertical slip rate or extension rate as has been done in previous studies of distributed extensional fault systems (e.g., Taupō Volcanic Field, New Zealand; Villamor and Berryman, 2001). (4) Within the Bartle Gap unit, the lack of identifiable correlative geomorphic surfaces or strain markers challenges scarp height or apparent minimum vertical separation measurements across individual fault traces. Using the total elevation difference between the highest and lowest mapped Bartle Gap units, we estimated a minimum long-term net vertical separation rate, but this rate also fails to account for any constructional volcanic topography that occurred during deposition. Whereas the basalt of Pole Creek may have formed a generally planar surface during deposition, the interbedded calc-alkaline andesitic basalt, tuffs, and pyroclastic flows of the Bartle Gap unit are likely to have formed irregular and variable topography during deposition. Despite these limitations, this exercise offers useful insight into the approximate deformation rates in the region.

Minimum vertical separation rates on individual fault traces and across the Pondosa fault system are low (<0.13 mm/yr) on Quaternary volcanic units (Fig. 12). In the basalt of Pole Creek (0.99 Ma), approximate vertical separations are <30 m to ~75 m on individual strands, resulting in minimum vertical separation rates of 0.03–0.08 mm/yr. However, the net vertical separation across the entire graben, which accounts for possible antithetic faulting, is ~100–130 m, resulting in a net vertical separation rate of ~0.1–0.13 mm/yr. On younger Pleisto cene volcanic units (basalt of Roseburg Timber, preliminary age of 0.795 Ma [Clynne and Muffler, 2017]; basalt of Hammond Crossing, 0.3055 Ma [Downs et al., 2020]), minimum vertical separation on individual strands varies from ~7 m to ~60 m, and estimated minimum vertical separation rates are <0.1 mm/yr. On the youngest mapped unit, late Pleistocene and Holocene volcanics including the Giant Crater flow (12,430 cal yr B.P.; Donnelly-Nolan et al., 1990), vertical separations are <8 m, corresponding to a vertical separation rate of 0.6 mm/yr. However, based on mapping and a smoother surface morphology, the footwall may have remnants of an older flow as mapped by Donnelly-Nolan (2010), and thus 8 m of vertical separation since ca. 12.4 ka may overestimate the fault movement and vertical separation rate (described in Interpretation of Faulting section; Fig. 12D). These rate calculations are further limited because they represent an average over limited (possibly only one surface-rupturing earthquake) earthquake cycles.

Apparent vertical offset of the Bartle Gap unit provides constraints for deformation rates over longer time scales. Using the highest and lowest elevations of the Bartle Gap unit, and assuming that the Bartle Gap unit underlies the basalt of Pole Creek, and that the basalt is <30 m thick, we estimated minimum vertical separations of ~280–330 m (Fig. 12). Using the youngest age of the Bartle Gap unit (4.9 Ma) and a vertical separation of ~330 m, the net vertical separation rate is ~0.07 mm/yr.

The vertical separation rates of ~0.07, ~0.1–0.13, and ~0.1 mm/yr on Pliocene through middle Pleistocene units imply that vertical slip across the Pondosa fault zone has been slow and uniform but likely distributed across many fault traces. The larger vertical separations on older units and smaller vertical separations on progressively younger units indicate that the Pondosa fault system has been active through the Quaternary, although likely at low (<0.13 mm/ yr) vertical rates.

Here, we place our observations of faulting along the Pondosa fault zone in a regional geodetic and geologic context. Several models exist to explain the regional style of deformation in northeastern California in general and the Mushroom Rock region in particular. These models include (1) the Basin and Range accommodation zone, with the Mushroom Rock region serving as a left step in the overall east-west–directed westward propagation of Basin and Range extension; (2) the microplate boundary model; and (3) the regional contraction model. Our observations of deformation along the Pondosa fault zone do not favor the regional contractional model but cannot discriminate between the microplate or accommodation zone models, although these latter two models are not incompatible with each other.

We summarize our relevant observations in the context of tectonic models here. Evidence for primarily dextral transtension is recorded in the surface faulting immediately north of the Pit River and is consistent with seismicity and focal mechanisms (Unruh and Humphrey, 2017), seismic reflection data (Kozaci et al., 2014), and outcrop data (O’Brien, 2013) from the region. Thermochronology and local structure are consistent with regional slow exhumation during the late Cenozoic (or rapid exhumation in the Miocene; see Supplemental Material, ^{footnote 1}), followed by normal-oblique dismemberment of Miocene–Pliocene volcanic topography.

Slip rates on individual fault strands appear to be low (<0.2 mm/yr; Fig. 12). The youngest faulted deposits along the Pondosa fault zone are assumed to be late Pleistocene in age but could be Holocene. However, faulting is highly distributed, and faults appear to rupture in complex patterns, as no simple, through-going systems are observed. These fault patterns and rates indicate an incipient, distributed fault zone (Tchalenko, 1970; Segall and Pollard, 1983).

In Figure 5, we depict strain accumulation in a local reference frame to highlight a contrast in relative GNSS horizontal station velocities north and south of the Mushroom Rock region and the Pit River. We plotted GNSS velocities relative to a fixed location (Hat Creek Radio observatory [HCRO]) near the approximate northeast edge of the Sierra Nevada microplate (Fig. 5) but note that selecting other stations near the Pit River provides similar results. We plotted no-net-rotation ITRF2014 (Altamimi et al., 2016) velocities with respect to station HCRO. We did not apply corrections for block rotations or subduction-related interseismic strain accumulation, because our intent was to provide an instantaneous snapshot of crustal strain with respect to a point near the Sierra Nevada–Oregon Coast block microplate boundary.

Using an HCRO-fixed reference frame, the velocity fields north and south of the Pit River appear to have different motions (Fig. 5). The velocity field south of the Pit River reflects northwestward-directed dextral shear extending inland at least 300 km from the coast. This portion of the velocity field appears to represent distributed northwest-directed dextral shear east of the Pacific plate with a small component of Sierra Nevada microplate rotation. North of the Pit River, horizontal velocities transition sharply to east-northeastward–directed motion relative to station HCRO. In general, the velocity field north of the Pit River integrates subduction-related, eastward-directed interseismic elastic upper-plate shortening with a much smaller component of Oregon Coast block microplate rotation. Mount Shasta moves coherently eastward relative to station HCRO, in contrast to more scattered, presumably magmatically dominated velocities at Lassen Peak (Poland et al., 2017). The general conclusion we draw from the GNSS data is that the Mushroom Rock region and the Pondosa fault zone are located at a transition in crustal deformation styles roughly centered on the Pit River, which represents the southern boundary of subduction-related elastic strain (e.g., Muffler et al., 2021).

A pervasive system of approximately north-south–oriented faults is clearly expressed on the landscape throughout northeastern California. The west-northwest–trending Pondosa fault zone deviates substantially from the north-south strike of faults in the region (Fig. 7). The orientation of these faults and the offset volcanic stratigraphy at the headwaters of Kosk Creek (Fig. 9) are consistent with normal-oblique motion. Thus, the Pondosa fault system may represent an accommodation zone between normal fault domains that has an oblique-dextral-normal sense of motion (Fig. 13).

Although the Walker Lane at this latitude is also interpreted to be an incipient fault zone that likely initiated ca. 3.5 Ma (Faulds and Henry, 2008), the different orientations (west-northwest for the Pondosa fault zone versus northwest for the Walker Lane, i.e., the Likely and Honey Lake faults; Fig. 2A) are consistent with the interpretation that the rate and kinematics of the Pondosa fault zone are distinct from the Walker Lane. These observations do not preclude interpretations of incipient Walker Lane deformation propagating within the Cascadia backarc region, supported by limited observations of dextral shear within the Klamath graben (Waldien et al., 2019), which has migrated northward over the past ~8 m.y. with the Mendocino triple junction (Faulds and Henry, 2008). However, if dextral transtension from the Walker Lane played an important role at this latitude, the orientation of these faults indicates that the Pondosa fault zone would be a compressional stepover, as described in model 3 for regional contraction below.

The microplate boundary model hinges on differential motion between the Sierra Nevada and Oregon Coast block microplates, rather than a transfer of Walker Lane shear, and predicts that active faulting north of the Pit River results from localized motion near the microplate boundary. This second model is based primarily on the observation that macroscopic dextral shear, inferred from seismicity and geodesy (McCaffrey et al., 2013; Unruh and Humphrey, 2017), trends west-northwest across an ~80-km-wide zone corresponding to the microplate boundary at ~40.5°N. This shear is parallel to Sierra Nevada–Oregon Coast block motion but counterclockwise to Sierra Nevada–North America–parallel motion in the northern Walker Lane. In this framework, Sierra Nevada–Oregon Coast block relative motion is accommodated by distributed shear and localized transpression, distinct from the Walker Lane dextral shear transitioning to oblique extension in the Cascade backarc, where the Walker Lane and Basin and Range open in a fan-like manner by rotation about a pole or poles near the mutual boundary of Oregon, Washington, and Idaho (Blakely et al., 1997; Faulds and Henry, 2008; Oldow and Cashman, 2009; McCaffrey et al., 2013; Unruh and Humphrey, 2017). The interpretation of distinct expressions of Sierra Nevada–Oregon Coast block–related versus Walker Lane–related shear is consistent with the overall structural fabric visible in the landscape, particularly the continuous network of north-striking (primarily normal) and northwest-striking (inferred oblique-dextral) Holocene-active faults extending well north of the Mendocino triple junction throughout northwestern Nevada, northeastern California, and southern Oregon (Pezzopane and Weldon, 1993). Moreover, derivation of an Euler pole for relative Sierra Nevada–Oregon Coast block rigid motion predicts relative west-northwest–directed dextral shear between the microplates (McCaffrey et al., 2013), which is supported by the observations in this study. The orientation of these faults may be influenced by preexisting basement structure inferred from gravity lows and Cascade volcanism (Blakely et al., 1997), where zones of primarily normal faulting, such as the Hat Creek (Muffler et al., 1994; Blakeslee and Kattenhorn, 2013; Kattenhorn et al., 2016) and McArthur (Page, 1995) fault zones and the Klamath graben (Waldien et al., 2019), are separated by zones of oblique dextral shear (Blakely et al., 1997). Recent work on the Klamath graben posited that Walker Lane–related shear extends into southern Oregon, where it dissipates into the southern Cascade arc (Waldien et al., 2019). Studies on the oblique Hat Creek fault zone further indicate that substantial dextral motion has occurred since the middle Pleistocene (Gray et al., 2017).

The regional contraction model, also called the compressional stepover model, indicates that dextral shear in the northern Walker Lane belt is transferred west across the northern Sierra Nevada microplate toward the southern Cascadia subduction zone (Unruh et al., 2003). This transfer forms a broad left step in a right-lateral system, resulting in a series of primarily reverse faults. Other reverse faults occur in the region, such as the Inks Creek fold belt, Red Bluff fault, and Battle Creek fault (Fig. 2) in the northern Sacramento Valley (Unruh et al., 2003; Sawyer, 2015; Angster et al., 2021), but these faults are kinematically unrelated to a stepover in the Walker Lane. The left-step model relates the northward propagation of the Walker Lane to the northward migration of the Mendocino triple junction and predicts that faulting within the left step, which roughly coincides with the Sierra Nevada–Oregon Coast block microplate boundary, is expected to be mainly contractional, whereas faulting in northeastern California north of the Mendocino triple junction is expected to be dominated primarily by extension and normal faulting with limited dextral shear. Recent studies have documented Pliocene and Quaternary activity on the Inks Creek fold belt and Battle Creek fault zone (~60 km to the southwest; Fig. 2; Helley et al., 1981; Angster et al., 2021), but evidence for the Quaternary structures transferring dextral shear remains elusive.

The arcuate map pattern of older Klamath terrane and Eocene to Late Cretaceous sedimentary rocks ringed by younger Pliocene–Pleistocene volcanic deposits, apparent opposing dips on the “flanks” of this map pattern, and the high, linear topography of the Mushroom Rock ridge create an apparent anticline (Fig. 4), thus raising the question of whether the ridge was formed by a blind thrust fault (Sawyer, 2010; Langenheim et al., 2016). The idea that Mushroom Rock ridge might be the surface expression of a buried thrust fault draws from an analogy to reverse faults mapped at the north end of the Sacramento Valley (the Battle Creek and Red Bluff faults and Inks Creek fold system; Fig. 2; Helley et al., 1981; Angster et al., 2021) and stems from the model of a Northern California shear zone extending westward from the northern Walker Lane to the Mendocino triple junction, forming a large-scale left stepover across much of northern California (Unruh et al., 2003; Sawyer, 2015).

Data presented in this study do not favor the hypothesis that the Mushroom Rock ridge is an anticline cored by a blind reverse fault. First, structural data, in particular, the dips of volcanic units, are not consistent with an anticline. The 6.3–4.9 Ma Bartle Gap unit on the north has variable dips, whereas the 9.6 Ma Red Mountain unit on the south has a consistent eastward dip (Fig. 4). The underlying, older Curtis Lake unit maintains a nearly uniform strike and dip across the proposed fold (Figs. 3 and 4); thus, bedrock geometries are inconsistent with opposing dips of fold limbs. Second, the Pondosa fault zone on eastern Mushroom Rock ridge, combined with focal mechanisms of earthquakes (Unruh and Humphrey, 2017), shows primarily north-northwest strikes with a dextral transtensional or extensional sense of motion (Figs. 2B and 7). These faults were active in the Pleistocene, and motion on them is incompatible with the contraction necessary to create the present-day topographic and geomorphic expression of Mushroom Rock ridge. Third, low-temperature thermochronology data do not support rapid uplift-driven exhumation of Mushroom Rock ridge in the Quaternary and instead can be explained by slow cooling during the late Cenozoic (or rapid exhumation in the Miocene; see Supplemental Material and Fig. S7, ^{footnote 1}; Batt et al., 2010; Michalak et al., 2018). Finally, basaltic and andesitic dikes and sills crosscut both the Montgomery Creek Formation and the Curtis Lake unit, indicating Miocene–Pliocene volcanic activity in the Kosk Creek drainage and Mushroom Rock ridge area (Fig. 9; Fig. S2). Volcanic activity during the Miocene through the Pliocene provides evidence that the topography and bedrock forming Mushroom Rock ridge resulted from successive emplacement of volcanic deposits, in addition to providing markers that are not deformed into an antiform. Mushroom Rock ridge can be more easily explained by constructive volcanism in the Miocene–Pliocene that has since been eroded and faulted by normal and dextral faults in the late Pleistocene and Holocene.

Austin (2013) investigated the stress type and evolution through time in the Mushroom Rock region using an array of minor faults younger than 1 Ma exposed in a diatomite quarry and four natural and anthropogenic exposures (Figs. 3 and 4). These faults exhibit predominantly normal and strike-slip kinematics and were interpreted to record a stress state change from normal to strike slip to reverse (Austin, 2013), but observations are limited.

Although historical seismicity in the region is infrequent, the geomorphic expression of the dense network of faults in the Pondosa fault zone and beyond indicates an active fault network that may be capable of generating moderate to large earthquakes. Currently, seismic hazard models in the region, such as the USGS NSHM (Petersen et al., 2014, 2020) or UCERF3 (Field et al., 2014), do not include the Pondosa fault zone as a source fault. Incorporating this fault zone into seismic hazard models (i.e., Hatem et al., 2022), however, raises questions about fault linkages and segmentation during surface rupture. Without knowledge of fault slip during larger earthquakes and whether individual strands link or remain segmented at seismogenic depths and during surface-rupturing events, it remains difficult to accurately define a single source fault for the fault zone. Instead, we propose using a proxy fault (e.g., Hatem et al., 2022) for geodetic deformation models and an areal seismic source (fault zone polygon) to model shaking associated with distributed faulting, with fault slip distributed on mapped faults or fault-related features throughout the polygon (e.g., Thompson Jobe et al., 2020).

We mapped a dextral-normal-oblique fault network in northeastern California by integrating interpretations of high-resolution topographic data, such as bare-earth LiDAR and SfM models, with traditional geologic techniques like bedrock mapping and ⁴⁰Ar/³⁹Ar geochronology. Although some of these faults have been previously recognized and mapped, our mapping shows that the newly recognized Pondosa fault zone extends north-northwest into high-relief volcanic topography, with normal-oblique offset apparent in fault exposures at the headwaters of Kosk Creek. New and existing ⁴⁰Ar/³⁹Ar and radiocarbon ages date the offset volcanic units from 12.4 ka to 9.6 Ma. The scarp morphology and geomorphic expression of the Pondosa fault zone, in addition to the ages of offset volcanic units, indicate that the fault zone was active in the late Pleistocene (<12.4 ka). Because many of these faults were not previously included in seismic hazard assessments, we propose using an areal seismic source or fault zone polygon to characterize possible shaking hazard represented by distributed faulting in the Pondosa fault zone. We interpret that the regional style of faulting can be explained by either a leftstep, dextral-oblique accommodation zone within overall east-west Basin and Range extension and/or as a local manifestation of the Sierra Nevada–Oregon Coast block microplate boundary.

This work was partially supported by Pacific Gas and Electric Company via a cooperative agreement. Pacific Gas and Electric additionally provided the Mushroom Rock and Pit3 light detection and ranging (LiDAR) data. This work was also supported by the Earthquake Hazards Program and Cooperative Summer Field Training Program of the U.S. Geological Survey. Special thanks go to Campbell Global, Sierra Pacific Industries, and Hearst for permission to access private lands and collect samples. Data presented in this manuscript are available in Thompson Jobe et al. (2022). LiDAR data are available from the National Map (https://viewer.nationalmap.gov/basic/). The USGS earthquake catalog can be accessed at https://earthquake.usgs.gov/earthquakes/search/ (last accessed December 2020). The USGS Quaternary Fault and Fold Database can be accessed at https://www.usgs.gov/programs/earthquake-hazards/faults (last accessed December 2020). This work benefited from discussions with William Page, Chris Madugo, Sean Gallen, Gordon Seitz, Vicki Langenheim, Tim Dawson, Leah Morgan, Ralph Klinger, and Glenda Besana-Ostman, and laboratory support from Jae Erickson. Katie Herr and Casey Smith assisted with oblique air photographs. L.J. Patrick Muffler and Michael Clynne assisted with bedrock mapping and volcanic rock descriptions, in addition to discussions about the regional tectonic and volcanic history. Thermochronology sample analysis was performed at the Thermochronology Research and Instrumentation Laboratory (TRaIL) at the University of Colorado–Boulder, and we thank James Metcalf and Rebecca Flowers for their assistance. We thank Jeff Unruh, the late Doug Yule, and the Geosphere associate editor and science editor, in addition to USGS reviewer Steve Angster, for their constructive comments, which helped to focus the manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.