Evidence of Active Quaternary Deformation on the Great Valley Fault System near Winters, Northern California

The Great Valley fault system stretches nearly 500 km from north to south, and defines the tectonic boundary between the Coast Ranges and the Great Valley–Sierra Nevada microplate (Fig. 1; e.g., Wong et al., 1988). Since its initiation in the late Cretaceous (e.g., Wakabayashi, 2015), the Great Valley fault system accommodates significant, primarily dip slip, displacement (e.g., Suppe, 1979; Unruh and Moores, 1992). The modern Great Valley fault system is a seismically active structural system and is the locus of several significant historic earthquakes along its length, including the 1983 M 6.5 Coalinga earthquake in the San Joaquin Valley and the 1892 Vacaville–Winters earthquake sequence west of Sacramento. Quaternary activity on the Great Valley fault system is generally interpreted as dip slip and accommodating convergence, including at the latitude of the Vacaville–Winters earthquake sequence (e.g., O’Connell et al., 2001). Geodetic data allow for interpretation of several mm/yr of shortening across the Great Valley fault system (d’Alessio et al., 2005), consistent with geologic interpretations, though published geodetic models do not specifically include a structure accommodating active convergence (d’Alessio et al., 2005).

In detail, the locations, geometry, and connectivity of faults within the Great Valley fault system are not well constrained (e.g., USGS and California Geological Survey [CGS], 2022). Here, we report the results of detailed geomorphic mapping, and a high‐resolution seismic reflection and tomographic survey across a fault within the Trout Creek section of the Great Valley fault system, which we call the West Winters strand. We interpret that the West Winters strand is an active, surface‐deforming structure that accommodates modern convergence within the Great Valley fault system.

The eastern margin of the California Coast Ranges, where they meet with the western edge of the Great Valley, has long been recognized as a significant tectonic boundary (e.g., Suppe, 1979; Wong et al., 1988; Wong, 1990; Unruh et al., 1992, 1995; Wakabayashi and Unruh, 1995). This tectonic boundary is defined by a structural system that is herein referred to as the Great Valley fault system to avoid implied kinematics, while maintaining similarity to the “Great Valley thrust system” moniker used in hazards databases.

The Great Valley fault system was recognized in seismicity datasets by Wong et al. (1988), who suggest the existence of a major tectonic boundary defined by a zone of seismicity along the eastern margin of the Coast Ranges. At the surface, the Great Valley fault system emplaces Mesozoic strata of the informal Great Valley sequence in a series of east‐dipping thrust wedges atop Jurassic ophiolite of the Coast Ranges and the Franciscan Assemblage (e.g., Suppe, 1979; Unruh et al., 1992, 1995; Wakabayashi and Unruh, 1995). North of the Sacramento River, the Great Valley fault system consists of a series of right‐stepping en echelon sections, including from south to north the Gordon Valley, Trout Creek, and Mysterious Ridge sections (Fig. 1; Unruh and Sundermann, 2006).

Broadly defined, the Great Valley fault system represents an inherited tectonic boundary between material of Sierran arc affinity (east of the Great Valley fault system) from that of subduction forearc affinity, including Franciscan Complex and ophiolite of the Coast Ranges (to the west; Holbrook and Mooney, 1987; Wong et al., 1988). The Great Valley fault system likely initiated as a Paleogene forearc shortening structure (e.g., Wakabayashi and Unruh, 1995) before being incorporated into the modern dextral San Andreas plate boundary fault system following the passing of the northward‐migrating Mendocino Triple Junction at ∼10–8 Ma (Atwater and Stock, 1998).

The Great Valley fault system is widely assumed to be active throughout the Quaternary (e.g., O’Connell and Unruh, 2000; O’Connell et al., 2001; Prescott et al., 2001), but deformation rates across the Great Valley fault system are not well constrained. Geodetic data suggest that the Great Valley fault system in northern California may accommodate 2–5 mm/yr of plate‐boundary‐perpendicular convergence (e.g., Prescott et al., 2001); whether this apparent convergence is accommodated as dip‐slip displacement on structures is disputed (e.g., d’Alessio et al., 2005). Geologic data do record Quaternary plate‐boundary‐perpendicular shortening on the Great Valley fault system: approximately 800–1200 m of apparent reverse displacement of the Plio‐Pleistocene Tehama Formation (3.4–1.0 Ma; Steele, 1980), constrained by seismic reflection data, at the latitude of the Gordon Valley section implies an average dip‐slip rate of 0.5–2.0 mm/yr since deposition of that unit (O’Connell and Unruh, 2000).

Two alternative structural models have been proposed for the tectonic role the modern Great Valley fault system plays, both of which include an east‐directed thrust at depth beneath the eastern margin of the Coast Ranges (Fig. 1b,c). In one model, based primarily upon structural geologic data and seismic reflection images, the Great Valley fault system functions as a west‐directed roof thrust of a tectonic wedge, rooted into the east‐directed structure at depth by way of a shallow detachment that extends westward under the Sacramento Valley (Fig. 1b) (e.g., Unruh et al., 1995). This model has been proposed for the Great Valley fault system both north (Rumsey Hills, Unruh et al., 1995) and south (Coalinga, Guzofski et al., 2007) of the study site. In the second model, the Great Valley fault system accommodates forelimb flexural slip within a fault propagation fold above the tip of the east‐directed fault at depth. In this model, which is based in part upon ground‐motion modeling for the 1892 Vacaville–Winters earthquake sequence, west‐directed faults at the surface need not root into the east‐directed fault at depth (Fig. 1c; O’Connell et al., 2001). Both structural models predict east side‐up reverse displacement on the Great Valley fault system.

Active deformation in the eastern Coast Ranges and western Sacramento Valley manifests in part as a series of low foothills that incorporate deformed Paleogene‐ to Neogene‐age bedrock, including the English Hills located between the cities of Winters and Vacaville (Fig. 1). The uplift and eastward tilting of the English Hills is likely Quaternary in age based upon geomorphic relationships and interpretation of soil development on terrace surfaces (O’Connell et al., 2001). A large fan complex, deposited by Putah Creek, onlaps the Tehama formation, suggesting that any deformation of these surfaces has occurred since 1.0 Ma (O’Connell et al., 2001). Additional likely Quaternary deformation has been documented in a series of low‐relief (∼5 m), long‐wavelength (∼km) hills about 15 km west of the range front (Unruh and Moores, 1992). Bedrock mapping, supported by interpreted seismic reflection data, confirm that these folds also deform Tehama Formation, implying that folding has occurred since the early Pleistocene (Unruh and Moores, 1992). Structures within the Great Valley fault system are inferred to accommodate this deformation but significant uncertainties in location, kinematics, and deformation rates remain.

Significant seismicity (M > 3) on the Great Valley fault system is limited, particularly on its northern reaches along the western edge of the Sacramento Valley. A notable exception is the 1892 Winters–Vacaville earthquake sequence—a pair of M 6+ earthquakes separated by two days in April of 1892, interpreted to have occurred on the Gordon Valley section of the Great Valley fault system (Fig. 1; O’Connell et al., 2001). Large instrumentally recorded earthquakes along the Great Valley fault system include the 1983 M 6.5 Coalinga and 1985 M 6.1 Kettleman Hills earthquakes, ∼300 km to the south. The 1983 Coalinga earthquake occurred on a ramp beneath the tectonic wedge along the eastern front of the Coast Ranges (Guzofski et al., 2007)—a setting broadly structurally analogous to the 1892 Winters–Vacaville sequence (Fig. 1b,c; e.g., O’Connell et al., 2001), suggesting that other sections of the Great Valley fault system may also be capable of M 6+ earthquakes.

We combined products derived from an airborne lidar digital elevation model (U.S. Geological Survey, 2019; Fig. 2a), including shaded relief and slope maps, with an orthophoto mosaic and structure from motion (10 cm resolution) digital surface model from an unmanned aerial vehicle survey (Morelan et al., 2022) to produce a geomorphic map of the Bigelow Hills site (Figs. 2 and 3). Mapped geomorphic contacts and features were verified by field observations.

We use a residual relief model (Fig. 2b)—an elevation model detrended with respect to the local channel (e.g., Hiller and Smith, 2008), in this case the active channel of Putah Creek which drains eastward across the southern part of the study site—to investigate deformation signals in the topographic data. This product allows us to visualize low‐amplitude, long‐wavelength deformation in otherwise planar surfaces (e.g., the floodplain and fluvial terraces). Our workflow is detailed in the supplemental material available with this article.

We acquired active‐source (P wave and S wave) data along an ∼800 m long profile, centered on the Bigelow Hills (Fig. 3). We recorded data using SmartSolo UGU‐16HR 3C seismometers, spaced at 5 m intervals. Seismic sources were generated at ∼1 m offset from each seismometer using an ∼500 kg accelerated weight drop (P waves) and a 3.6 kg hammer and aluminum plate combination (S waves). We used the code of Hole (1992) to invert first‐arrival P‐ and S‐wave refractions in developing 2D tomographic P‐wave velocity (⁠$VP$⁠) and S wave velocity (⁠$VS$⁠) models (Fig. 4). We used the P‐wave data to generate an unmigrated reflection image (Fig. 5) along the profile. Data acquisition and modeling parameters, processing sequences, and additional model outputs are included in the supplemental material, and primary data are available via data release (Goldman et al., 2022).

Our geomorphic analysis documents a set of north–northwest‐trending scarps that delineate the western margin of the Bigelow Hills and continue northward across Quaternary terraces of Putah Creek to define the eastern margin of the Dry Creek drainage northwest of the city of Winters (Fig. 2). Though the westward‐facing scarps vary in height along strike, apparent displacement is consistently east side‐up and magnitude of apparent displacement is consistent with the relative age of displaced units. Other indicators of east side‐up displacement include a wind gap on the eastern edge of the Dry Creek drainage and apparent constriction in the eastward‐flowing Putah Creek drainage, consistent with expected incision resulting from an uphill‐facing scarp crossing an active stream (Fig. 2). The east side‐up geometry across the scarp is mirrored by bedrock uplift and eastward dip of Tehama Formation in the Bigelow Hills, east of the scarp (Fig. 3).

Despite clear geomorphic evidence of the scarp to the north, we do not find clear evidence of Quaternary deformation south of Putah Creek despite the apparent similarity in age of geomorphic surfaces. We suggest that this may result from regular avulsion or overtopping of the primary channel of Putah Creek, which may be biased to the south of the primary channel due to topography (including the Bigelow Hills) on the northern bank, possibly in combination with tapering displacement southward into the Trout Creek–Gordon Valley stepover zone of the Great Valley fault system.

From the seismic data acquired along the survey profile, we developed multiple models that provide indicators of faulting in several locations.

P‐wave velocities (⁠$VP$⁠) range from ∼600 to ∼2800 m/s along the profile, with the highest velocities at depth west of the Bigelow Hills (Fig. 4a). Prominent velocity changes between distance meters 300 and 350 underly the western edge of the Bigelow Hills, consistent with what we expect to see from fault‐related juxtaposition of fluvial terrace fill against the Plio‐Pleistocene Tehama Formation. East–west changes in $VP$ occur across a range of depths at this location in the model and, thus, do not appear to be a near‐surface topographic effect.

The groundwater table generally correlates with a $VP$ of about 1500 m/s in sediments and sedimentary rocks (Catchings, et al., 2009, 2014, 2019). Along our survey profile, the 1500 m/s $VP$ contour varies in depth but is generally within ∼20–35 m of the surface (Figs. 4 and 5). This depth is consistent with nearby groundwater monitoring wells with similar proximity to the active channel of Putah Creek, which record groundwater depths of 21–34 m during the date range of the survey (Yolo Subbasin Groundwater Agency [YSGA], 2022). We note an apparent vertical offset in the 1500 m/s contour between profile distance meters 300 and 350, which is consistent with our interpretation of a fault at that location (Fig. 4a). Measured groundwater depths in wells appear consistent with a slight vertical displacement across the hypothesized fault trace, though distribution of groundwater monitoring wells near the study site does not allow for a robust comparison (YSGA, 2022). Faults are typically groundwater barriers, and a discrete change in water table depth is often interpreted as a fault (Proctor, 1968; Wallace and Morris, 1986; Bredehoeft et al., 1992; Catchings et al. 2014).

S‐wave velocity (⁠$VS$⁠) ranges from about 300 m/s near the surface to about 1700 m/s in the upper 200 m near the Bigelow Hills, with the highest $VS$ values dominantly west of the Bigelow Hills (Fig. 4b). Much like the $VP$ tomographic model, an abrupt change in $VS$ near the western edge of the Bigelow Hills (distance meters 300–350) may indicate a change in lithology. We also observe a near‐surface, dipping zone of relatively low $VS$ (600–700 m/s) near distance meters 550–600 of the seismic profile. Zones of low $VS$ relative to surrounding areas have previously been interpreted as a result of shearing on fault planes (e.g., Catchings et al., 2014).

Our reflection stack (Fig. 5) shows that in the western 300 m of the seismic profile layered reflectors extend to about 50 m depth and are underlain by a strong reflector that represents a velocity contrast we interpret as the base of alluvial fill. For all but the easternmost ∼100 m, the strong reflector coincides with the 1500 m/s $VP$ contour used as a proxy for the top of groundwater (Fig. 5b). Near meter 325, the sequence of reflectors is steeply offset (down to the west) across an apparent fault or series of faults (Fig. 5b) and diverges from the 1500 m/s $VP$ contour. In addition, we observe a series of near‐vertically oriented diffractions (visible as sets of downward‐oriented “V”s); such diffractions are typically seen in unmigrated images in which layered strata are vertically offset by faulting. We observe at least two additional offsets (meters 450 and 620) of the apparent bedrock reflector, each of which is paired with a series of underlying diffractions. The apparent faults at approximately 325 and 620 meters appear to nearly extend to the surface. The apparent fault near meter 450 appears to offset only the deeper (>25 m) strata.

We interpret the subsurface images to show faulting at several locations along the survey profile (Figs. 4 and 5). Two faults extending to the near surface are particularly prominent in the seismic reflection image (Fig. 5), and correlate with topographic scarps and changes in $VP$⁠, $VS$⁠, and depth to groundwater. These faults result in apparent east side‐up vertical displacements of >25 m for a shallow‐depth (∼50 m) strong reflector that we interpret as the bedrock contact at the base of the fill terrace sequence deposited by modern Putah Creek. We observe two additional apparent blind faults along the profile. The relatively thin fill and shallow bedrock that we interpret in the seismic images to the west of the fault is consistent with geologic observations in the area surrounding the study site, which include >30 m high exposures of bedrock along Putah Creek at the east end of Lake Solano, ∼2 km west of the Bigelow Hills.

Fault dips are largely high angle (∼80°–85°) in the upper ∼100 m—an anomalously steep dip for dip‐slip structures. The steep fault dips we observe are also inconsistent with bedrock orientation data from the surrounding region, which generally record shallow‐to‐moderately east‐dipping bedding and west‐directed faults that are largely bedding‐parallel or shallowly cut upsection (e.g., Unruh et al., 1995). The Tehama Formation in the hanging wall of the West Winters strand at the study site is too poorly exposed for bedding‐orientation measurements; however, nearby exposures of the Tehama Formation south of Putah Creek dip 5°–10° east (Wiegers et al., 2007). We hypothesize that the steep fault dip at the study site may be a local aberration in an otherwise shallow‐to‐moderately dipping fault system that is also responsible for the very localized uplift of the Bigelow Hills; however, additional work could help to further refine understanding of structural geometries at this site.

The sharp lineament defining the western edge of the Bigelow Hills aligns with the western fault strand observed in seismic images. Based upon clear representation in both geomorphic and seismic datasets, we suggest that this is the primary active structure at this site. The eastern major strand aligns with a topographic saddle in the Bigelow Hills; we suggest that the saddle may represent surface expression of this structure. That this strand is not clearly expressed in Quaternary fluvial deposits along Putah Creek or Dry Creek suggests that any surface‐deforming activity on this strand is likely older than activity on the western strand.

The geomorphic evidence of active surface deformation at the Bigelow Hills aligns along strike with the scarp and water gap features along the Dry Creek drainage to the north, suggesting that these features all lie along the same structure. We collectively refer to this structure as the West Winters strand, and infer that it is the southern continuation of the Trout Creek section of the Great Valley fault system. Apparent displacement across these features is consistently east side‐up. A low‐amplitude scarp, particularly apparent in the residual relief model (Fig. 2b,c), continues between the water gap and the Bigelow Hills, though heavy agricultural activity in this area may have altered both the height and sharpness of this scarp.

The surface deformation we observe across the West Winters strand, at the southern end of the Trout Creek section of the Great Valley fault system, is the first quantitatively constrained evidence of latest Quaternary surface displacement on a fault in the northern Great Valley fault system. The identification of the surface‐deforming West Winters strand expands upon prior studies, which infer active dip‐slip displacement on structures within the Great Valley fault system based upon regional‐scale geomorphic and subsurface geophysical investigations (e.g., O’Connell et al., 2001). The location and east side‐up sense of displacement we document on the West Winters strand are consistent with west‐directed slip on an east‐dipping thrust fault—a feature of both the tectonic wedge model (e.g., Unruh et al., 1995) and the forelimb flexural slip model (e.g., O’Connell et al., 2001) proposed for the Great Valley fault system at this latitude. Our results do not clearly differentiate between the two models.

Although the West Winters strand strikes subparallel to dextral strike‐slip faults of the San Andreas plate boundary fault system, including the Bartlett Springs fault ∼20 km to the west of the West Winters strand, we do not observe any geomorphic evidence of dextral strike‐slip displacement across the structure. This result is consistent with prior studies interpreting the Great Valley fault system as a predominantly dip‐slip structural system, suggesting that plate boundary strain is partitioned into plate‐boundary‐normal and plate‐boundary‐parallel components, with the former being accommodated by the Great Valley fault system (e.g., Prescott et al., 2001).

The documented dip‐slip motion on this structure is significant for seismic hazard. Vertical displacement on Great Valley fault system structures, including the West Winters strand, could have significant consequences, given the region’s low relief and the presence of significant water conveyance infrastructure. Potentially impacted infrastructure includes the Putah South Canal, ∼1.5 km south of the Bigelow Hills, and Putah Dam, and the Lake Solano reservoir (750 acre feet) located ∼2 km west (Fig. 2). Monticello Dam, which impounds the Putah Creek drainage to create Lake Berryessa (1.6 million acre feet), is located 10 km west of the West Winters strand.

The source fault(s) for the 1892 Vacaville–Winters earthquake sequence is(are) very likely in the immediate vicinity of this study area and may well include the West Winters strand. Ground‐motion modeling and source characterization, based upon first‐hand observations combined with structural and seismotectonic datasets, suggest that the primary events in the 1892 sequence may have occurred along the Gordon Valley section of the Great Valley fault system approximately 5–10 km to the southwest of the Bigelow Hills site (O’Connell et al., 2001). However, those authors note that their modeling results may be biased by the lack of significant population centers in the area ∼15–20 km north of Winters, introducing an artificial northward reduction in observed shaking intensity. Even so, the region of modified Mercalli intensity >7 extends nearly as far north of Winters as to the south, suggesting possible directivity of shaking effects and/or potential rupture north of the Gordon Valley–Trout Creek stepover (Fig. 1; O’Connell et al., 2001). An eyewitness account during the second (M ∼6.2) earthquake in the sequence from a worker plowing the field immediately west of the Bigelow Hills gives a description of what may be coseismic surface deformation along this structure:

“The ground rose and fell like the sea in a storm …. Immediately after the shock, the eyewitness saw distinctly that the ground was broken up into reefs and furrows, which closed up even as he was watching them (Bennett, 1987).”

We have not found additional first‐person observations from along the West Winters strand, and it is difficult to draw any significant conclusions from this account. However, it raises the possibility that the West Winters strand accommodated surface‐deforming slip during the 21 April 1892 M ∼6.2 event. We note that the extensive agricultural development of the region (in particular plowing, grading, and planting of nut orchards), including on the Qt5 surface crossed by the West Winters strand, may have obscured or removed any hypothetical coseismic surface deformation along the fault trace in the more than a century since the 1892 event.

We present geomorphic, bedrock structural, and seismic reflection and tomography data that document Quaternary activity on a surface‐deforming fault, which we call the West Winters strand, within the Trout Creek section of the Great Valley fault system. Our observations record east side‐up, dip‐slip displacement across the fault, consistent with bedrock geometry and structural models along strike (e.g., Unruh et al., 1995). The lack of evidence for strike‐slip displacement on this structure suggests strain partitioning between the Great Valley fault system and plate boundary faults further west, including the Bartlett Springs fault.

Our observations suggest that the West Winters strand is a source of surface deforming and possibly surface rupturing earthquakes. Anecdotal reports and earthquake effects’ studies suggest that it is possible, though difficult to verify, that the West Winters strand played a role in the 1892 Vacaville–Winters earthquake sequence. Evidence of surface deformation on the West Winters strand raises questions about the role that this and other structures play in accommodating stress across the stepover to the Gordon Valley section ∼5 km southwest. The existence of a surface‐deforming fault accommodating vertical displacement within the Great Valley fault system has significant implications for scientific understanding of modern kinematics and for seismic hazard in the northern Sacramento Valley and the Sacramento–San Joaquin Delta region, and the Great Valley fault system and Coast Ranges as a whole. Future work at the Bigelow Hills, including excavations on deformed terrace surfaces to constrain timing and rate of deformation, may further elucidate the significance of individual earthquakes in the evolution of the topographic scarps at the site and provide constraints on the rate of active displacement across the West Winters strand of the Great Valley fault system.

Seismic imaging survey data are available via U.S. Geological Survey (USGS) Data Release by Goldman et al., 2022 (doi: 10.5066/P9MSPO83). Unmanned Aerial Vehicle (UAV) topographic survey are available via USGS Data Release by Morelan et al., 2022, hosted by OpenTopography (doi: 10.5069/G90G3HBK).

The authors acknowledge that there are no conflicts of interest recorded.

The authors thank C. Criley, J. Chan, R. Sickler, S. DeLong, A. Pickering, D. Samuel, and M. Esqueda for their invaluable help in conducting the shallow seismic survey. Thanks to M. Sharrock and M. Mariani for land access. Reviews by S. DeLong, R. Briggs, C. Amos, and M. Oskin greatly improved the article. This work was funded by the U.S. Geological Survey San Francisco Bay–Delta Priority Ecosystems Science program. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.