Abstract

We use geophysical data to examine the structural framework of the northern San Francisco Bay region, an area that hosts the northward continuation of the East Bay fault system. Although this fault system has accommodated ∼175 km of right-lateral offset since 12 Ma, how this offset is partitioned north of the bay is controversial and important for understanding where and how strain is accommodated along this stretch of the broader San Andreas transform margin. Using gravity and magnetic data, we map these faults, many of which influenced basin formation and volcanism. Continuity of magnetic anomalies in certain areas, such as Napa and Sonoma Valleys, the region north of Napa Valley, and the region south of the Santa Rosa Plain, preclude significant (>10 km) offset. Much of the slip is partitioned around Sonoma and Napa Valleys and onto the Carneros, Rodgers Creek, and Green Valley faults. The absence of correlative magnetic anomalies across the Hayward–Rodgers Creek–Maacama fault system suggests that this system reactivated older basement structures, which appear to influence seismicity patterns in the region.

INTRODUCTION

The San Andreas fault consists of a single main active strand between the San Francisco Bay area and the Big Bend in southern California. In the San Francisco Bay area, however, the fault branches into several strands with significant offset (Fig. 1). The San Andreas fault zone (including the Pilarcitos fault; Fig. 1) that transects the San Francisco Peninsula has ∼130 km of offset (Jachens et al., 1998), whereas the East Bay fault system has a well-documented displacement of 160–175 km (McLaughlin et al., 1996; Jachens et al., 1998; Graymer et al., 2002). Although the East Bay fault system has been extensively studied in the eastern and southern bay area, how displacement is parsed onto faults north of San Pablo Bay is controversial. For example, offset estimates for the Rodgers Creek fault, the present-day northward continuation of the Hayward fault, range from as low as 5–10 km (Fox, 1983, p. 16; Randolph-Loar, 2002; Allen, 2003), to 28 km (Sarna-Wojcicki, 1992; McLaughlin et al., 2008a), to as high as 65 km (Graymer et al., 2002). In Graymer et al. (2002), correlative, but now offset, geologic units and other geologic relations were used to constrain the amount and timing of offset on the East Bay fault system, but did not, however, account for most of the fault offset in the area north of the bay. Geophysical data, in particular magnetic data, have been used to estimate cumulative displacements along the San Andreas fault (Griscom and Jachens, 1990). Here we use geophysical data to examine basement offsets and to illuminate fault locations, particularly those concealed beneath Neogene sedimentary and volcanic deposits in the northern San Francisco Bay region (Fig. 2).

We also examine mechanisms of basin formation for the region. Prior to the development of discrete right-lateral faults within the San Andreas fault system, the San Andreas transform system was marked by diffuse transtension along the continental margin (Atwater, 1970; McCrory et al., 1995) that led to development of Miocene sedimentary basins along the coast. Northward passage of the Mendocino Triple Junction (inset in Fig. 1) and opening of a slab window have led to rapid subsidence synchronous with volcanism documented in basins west of the San Andreas fault along coastal California (Wilson et al., 2005); could such a mechanism form some of the basins in the North Bay? Basins may also have formed in response to folding and faulting associated with transpressional tectonics resulting from a change in plate motion ca. 8 Ma (Atwater and Stock, 1998). Calderas that produced some of the widespread tuffs within the Sonoma Volcanics (see Sweetkind et al., 2005) may also have contributed to basin formation as well as right steps or bends in the right-lateral faults producing pull-apart basins. To gain insight into the mechanisms of basin formation, we interpret gravity data in terms of basin geometry. For example, these data reveal a complex basin configuration concealed beneath Napa and Sonoma Valleys formed by folding, strike-slip faulting, and volcanic processes.

This paper builds upon several recent papers that examine the geophysical framework of the area, in particular the geometry and continuity of the West Napa fault (Langenheim et al., 2006a), the basin configuration beneath Sonoma Valley for groundwater flow modeling (Langenheim, 2006), and the basin geometry beneath the Santa Rosa Plain that influenced where strong shaking was focused in the 1906 San Francisco and 1969 Santa Rosa earthquakes (McPhee et al., 2007). These previous studies showed successful application of potential-field data for seismic hazard and groundwater resource assessments, allowing us to address the larger question of how slip is partitioned into this complex section of the transform margin and influenced by Cenozoic volcanism related to the passage of the Mendocino Triple Junction. The basin configuration presented here also forms the base of three-dimensional geologic modeling in the Santa Rosa Plain (Sweetkind et al., 2010). In particular, we find that the large gravity low beneath San Pablo Bay results from multiple events that resulted in thick sediments collected in a basin formed within a right step in the East Bay fault system, active between ca. 8.5 and ca. 6 Ma. This basin was subsequently overprinted by an additional basin associated with the active right step between the Hayward and Rodgers Creek faults. We find that superposition of basins results from superposed faulting due to fault reorganization within the North Bay system. This superposition of structure is guided by preexisting basement structures, suggested by the absence of correlatable magnetic anomalies across the Hayward–Rodgers Creek–Maacama fault system.

GEOPHYSICAL DATA AND METHODS

In this study we concentrate on gravity and aeromagnetic data sets because these data cover much of the study area in a uniform fashion, whereas other geophysical data, such as crustal seismic sections, are areally limited, proprietary, or nonexistent. Gravity and aeromagnetic data reflect density and magnetization contrasts within the upper and middle crust from which we can define the three-dimensional geometry of Cenozoic basins and structures.

Gravity

The regional gravity data (Langenheim et al., 2006b), augmented by 329 stations collected in 2006–2007, were gridded at 300 m to produce an isostatic residual gravity map of the study region (Figs. 3 and 4). This map reflects density variations in the upper and middle crust (Simpson et al., 1986). Density measurements indicate that one of the most significant density contrasts in the upper crust of this area is that between dense Mesozoic basement rocks and lower density Cenozoic, especially Neogene, sedimentary and volcanic rocks (Langenheim, 2006). This is corroborated in that positive gravity values generally coincide with outcrops of Franciscan Complex or Coast Range ophiolite (Fig. 4), whereas the most prominent gravity low over San Pablo Bay marks thick sedimentary and volcanic fill penetrated by several deep (>2 km) drill holes (Wright and Smith, 1992).

We used the method in Jachens and Moring (1990) to separate the isostatic gravity field into that component produced by variations in basement density (basement gravity field) and that caused by thick sedimentary and volcanic deposits, which can then be inverted to estimate basin thickness (Fig. 5). The inversion method allows the density of basement to vary horizontally as needed, whereas the density of basin-filling deposits is specified by a predetermined density-depth relationship (Table 1) based on density logs (Brocher et al., 1997) from the Chevron Bethlehem No. 1 and Clarement Energy John Rice No. 1 wells (drill holes 1 and 5 in Fig. 2). Phelps et al. (1999) showed that lateral variations in basin-fill density, unless abrupt, do not change the overall modeled shape of the basin and that the method can be less effective in estimating the magnitude of basin thickness, especially in basins containing thick basalt flows or in areas of poor well control.

Previous basin inversions for parts of the North Bay region (Langenheim, 2006; Langenheim et al., 2006a; McPhee et al., 2007) defined basement as Mesozoic in age, including Mesozoic Great Valley Sequence rocks, which are of intermediate density relative to Franciscan Complex and Coast Range ophiolite rocks and Cenozoic basin fill. For this study we limit basement to Franciscan Complex or Coast Range ophiolite. This change in basement definition avoids unrealistically high (>10 km) thickness estimates of Cenozoic fill beneath San Pablo Bay that do not match the scattered well data (Wright and Smith, 1992). The effect of this change in density-depth function and basement definition did not change the overall shape of the basins, but did alter the magnitude of the basin thickness.

To help delineate trends and gradients in the gravity field, we used a computer algorithm to locate the maximum horizontal gravity gradient (Cordell and Grauch, 1983; Blakely and Simpson, 1986). Locations of gradient maxima (aligned blue circles in Fig. 4) occur approximately over vertical or near-vertical contacts that separate rocks of contrasting densities. For moderate to steep dips (45° to vertical), the horizontal displacement of a gradient maximum from the top edge of an offset horizontal layer is always less than or equal to the depth to the top of the source (Grauch and Cordell, 1987).

Magnetics

Aeromagnetic anomalies in the area generally reflect Mesozoic basement rock types (ophiolitic rocks including serpentinite, gabbro, and basalt) and magnetic rock types within the various Tertiary volcanic rocks either exposed or in the subsurface. Sedimentary rocks generally are not magnetic enough to produce measurable aeromagnetic anomalies; however, in our study area, sandstones of the Neroly Formation (Hillhouse and Jachens, 2005) and beds in the Late Cretaceous part of the Great Valley Sequence produce anomalies with amplitudes as high as 200 nT (ne and k in Figs. 6 and 7). Aeromagnetic data consist of three detailed surveys flown at 305 m above average terrain along flight lines spaced 536 m apart (U.S. Geological Survey, 1992, 1996, 1997). The aeromagnetic data were adjusted to a common datum and merged by smooth interpolation across survey boundaries.

Magnetic (and gravity) anomalies are produced by a variety of sources of variable size and depth. Superposition of anomalies from multiple sources can result in interpretational ambiguities. For example, both Mesozoic serpentinites and some Tertiary volcanic rock types are capable of producing magnetic anomalies, but may be characterized by anomalies of differing wavelengths (or characteristic length). Shallow sources typically cause short-wavelength anomalies, whereas deep sources produce long-wavelength anomalies. Generally, the Tertiary volcanic rocks, comparatively thinner and shallower than Mesozoic serpentinite or ophiolite, should produce shorter wavelength, highly variable anomalies that produce a rough texture when viewed in shaded-relief form (Fig. 6).

To emphasize both short-wavelength anomalies caused by shallow sources (e.g., Tertiary volcanic rock) and long-wavelength anomalies (e.g., serpentinite or ophiolite), a match filter was applied to the aeromagnetic data (Phillips, 2001). Match filtering separates the data into different wavelength components by modeling the observed anomalies as a sum of anomalies from distinct equivalent source layers at increasing depths (see Phillips, 2001). Figures 8–10 show the resulting separated fields produced by the dipole equivalent-source layers at 8.089, 1.546, and 0.4141 km depths. For each of these filtered fields to help delineate trends and gradients, we also calculated magnetization boundaries (as described in Blakely and Simpson, 1986), but using aeromagnetic data that were reduced to the pole to shift the anomalies over the sources and transformed to magnetic potential anomalies. Magnetization boundaries shown in Figure 7 (reddish-brown crosses) were calculated using the data filtered to emphasize shallow sources (Fig. 10).

RESULTS

Cenozoic Faults—Extent, Location, and Offset

Green Valley Fault

The Green Valley fault is a Holocene active fault mapped from Suisun Bay north to the lat ∼38°22.5′ (Fig. 2; Graymer et al., 2006a). Its continuation northward is conjectural, but may connect to Quaternary faults mapped west of Lake Berryessa (Hunting Creek fault; D.G. Herd cited inBryant, 1982) or step westward to connect with the Maacama fault (Baldwin et al., 1998; Baldwin and Unruh, 2004). Correlation of volcanic rocks between the fault and Suisun Bay with those east of the city of Napa leads to an estimated cumulative right-lateral offset of ∼20 km since 3.5–5.5 Ma (Graymer et al., 2002; Table 2).

The fault is well expressed geophysically where it forms the western edge of the Suisun Bay basin southeast of Napa Valley, and coincides with east-facing gravity and magnetic gradients (GVF in Figs. 3 and 6). North of Green Valley (Gv in Fig. 3), however, the gravity gradient changes to a west-facing direction, reflecting the contact between Sonoma Volcanics on the west against Great Valley Sequence rocks to the east. Strands of the Green Valley fault are mapped cutting through volcanic rocks with as much as several kilometers of dextral offset (Bezore et al., 2004). The west-facing gravity gradient most likely marks an edge of one of several overlapping calderas that were the source area for several large ash-flow tuff and Plinian eruptions, the coarsest facies and greatest thicknesses of which are found in southeastern Napa Valley (Sweetkind et al., 2005).

North of this inferred caldera complex, one possible continuation of the Green Valley fault is an active Quaternary fault that is mapped to curve westward (Graymer et al., 2006a). The fault coincides with a strong west-northwest–striking gravity gradient east of Lake Hennessey (Fig. 4). The position of the gravity gradient relative to the fault juxtaposing Franciscan Complex against Sonoma Volcanics suggests a steep northeast-southwest–dipping fault. Baldwin and Unruh (2004) proposed that the Green Valley fault steps westward in a restraining stepover to the Maacama fault, folding the Sonoma Volcanics with locally several kilometers of structural relief, which is not readily apparent in the gravity field (Fig. 3) or basin inversion results (Fig. 5). The steep gravity gradient bifurcates immediately west of Lake Hennesey, with the stronger gradient marking the southwestern edge of Franciscan Complex and Great Valley Sequence rocks stepping ∼10 km north, where it again regains a northwest strike at the contact between the volcanic rocks and the Great Valley Sequence.

The other possible northward continuation of the Green Valley fault is along the eastern margin of the large prominent gravity high north of Lake Hennessey, and is mapped along the west side of Lake Berryessa. A nearly 10-km-long gap and a possible eastward step of ∼5 km separate the mapped Quaternary active fault traces. In the area between the mapped fault traces, northwest-striking magnetic anomalies (Figs. 6, 9, and 10) may reflect a structural connection. This branch of the Green Valley fault appears to roughly follow the axis of a deep magnetic source (Fig. 8), whereas seismicity follows the western edge of the deep magnetic anomaly.

Soda Creek Fault

The north-striking Soda Creek fault is mapped along the southeastern margin of Napa Valley (Fig. 2) and has >210 m of vertical displacement (Weaver, 1949, p. 140). The fault also forms a partial groundwater barrier and coincides with elevated levels of arsenic (Farrar and Metzger, 2003).

The fault is expressed as a weak gravity gradient (SCF in Fig. 3). As the fault projects northward into outcrops of Sonoma Volcanics, it appears to step left (or west) ∼3 km to again form the eastern valley margin as the valley margin progressively curves to the west. The left step in the Soda Creek fault coincides with an elevated ridge, and an apparent right-lateral offset (<5 km) of the southern margin of the Yountville gravity high leads us to suggest that the Soda Creek fault has both vertical and horizontal offset. At and west of Lake Hennessey, the fault is located at the base of a prominent gravity gradient that marks the southern edge of Franciscan Complex and Great Valley Sequence rocks, suggesting that this boundary dips to the northeast. West of Lake Hennessey, the gradient bifurcates, with a weaker gravity gradient continuing northwest of the mapped tip of the Soda Creek fault. The gradient projects along the eastern margin of the valley floor, diverging from the valley margin westward, parallel to the gravity gradient of the West Napa fault, west to within 2 km of the Maacama fault. This gradient suggests a possible structural continuity across the valley, which would influence the location of upwelling thermal fluids in the Calistoga area.

West Napa Fault

The West Napa fault is mapped at the surface from north of Vallejo to the town of St. Helena (Fox, 1983; Clahan et al., 2004, 2005; Graymer et al., 2007; Fig. 2). The fault coincides with nearly continuous gravity (Figs. 3 and 4) and aeromagnetic gradients (Figs. 6 and 7) and was the likely source of the 2000 ML 5.2 Yountville earthquake (Langenheim et al., 2006a). South of the town of St. Helena, the gravity gradient associated with the fault results from the juxtaposition of Great Valley Sequence and Franciscan Complex rocks to the southwest with less dense Cenozoic Sonoma Volcanics and alluvium to the northeast (Fig. 4). North of the town of St. Helena, the gravity gradient changes to a slightly more westward strike and continues, apparently reflecting the West Napa fault, concealed beneath Sonoma Volcanics dated as 2.6–3.7 Ma by K-Ar and fission-track methods (Fox et al., 1985) and as 2.8–3.4 Ma using Ar/Ar and tephrochronologic techniques (McLaughlin et al., 2004, 2008b). Magnetic gradients of shallow and medium depth sources (Figs. 9 and 10) also coincide with much of the pronounced westward-curving gravity gradient.

Gravity data and, to a lesser extent aeromagnetic data, suggest that the West Napa fault extends to the Maacama fault; whether the West Napa fault merges or is truncated by the Maacama fault is not apparent based on the geophysical data. Carefully mapped field relations (McLaughlin et al., 2004) indicate that the Petrified Forest fault (Fig. 2), one of the major compressional, west-northwest–striking faults that involves 3.2–3.4 Ma Sonoma Volcanics, is bluntly truncated by or abuts the Maacama fault zone. However, no obvious offset equivalents of the magnetic or gravity gradients that mark the West Napa fault northeast of the Maacama fault zone are present on the southwest side of the Maacama fault. Such equivalent features would be expected, given 17–21 km of right-lateral offset on the Maacama fault since 3.2–3.3 Ma (McLaughlin et al., 2005a). Thus, our interpretation based on these geophysical gradients is that the fault merges with the Maacama fault. Alternatively, the continuation of the West Napa fault north of St. Helena may maintain more of a northwest trend and connect to the Geyser Peak fault zone (Fig. 2). This path for the West Napa fault is weakly supported by a gradient in the magnetic field, filtered to enhance deep sources (Fig. 8), that extends the mapped trace of the Geyser Peak fault southeast ∼3–4 km toward the West Napa fault. A similar gradient in the magnetic field filtered for medium depth sources (Fig. 9) extends the mapped trace of the Geyser Peak fault zone even farther to the southeast, but a straighter trace of the West Napa fault cuts across the prominent gravity gradients of Napa Valley, with no obvious offset equivalent on the east side of the fault.

The amount and sense of displacement on the West Napa fault are not well known, but gravity data indicate that the fault must have a long-term dip-slip component, east side down, to form Napa Valley and its underlying basin topography. The location of the mapped fault relative to the gravity gradient and a joint gravity and magnetic model (Langenheim et al., 2006a) indicate that the fault dips slightly to the southwest and thus has a reverse component of displacement. Right-lateral displacement is also likely given the northwest strike of the fault and the focal mechanism of the 2000 Yountville earthquake. Correlation of Late Cretaceous and Paleogene strata west of Napa with similar strata in the Cordelia area (Fig. 2) may suggest a long-term right-lateral offset of 5 km, but detailed comparisons required for positive correlation have not been made. In Langenheim et al. (2006a), a pair of magnetic anomalies was speculatively correlated across the fault, providing an estimate of 10 km of right-lateral displacement. This estimate is speculative because of the oblique trend of the correlated anomalies relative to that of the fault. Another estimate suggests as much as 40 km of right-lateral offset by correlating the deep magnetic and dense source southwest of St. Helena with the deep magnetic and dense body near Vallejo. Although the northern deep anomaly appears to cross the West Napa fault without apparent offset, the shallow wavelength, medium wavelength, and unfiltered magnetic data indicate that the apparent continuity of the deep magnetic anomaly across the West Napa fault separates two distinct magnetic sources. The advantage of this correlation is the nearly perpendicular trend of the northern anomaly to that of the West Napa fault.

Carneros Fault

The Carneros fault juxtaposes Tertiary marine strata, including the Neroly Formation and overlying Sonoma Volcanics on the west, with Great Valley Sequence rocks on the east in Carneros Valley at the southern end of the Mayacmas Mountains (Fig. 2). It is a vertical to near-vertical fault (Weaver, 1949) that cuts older Sonoma Volcanics but appears to be overlapped by undated, younger Sonoma Volcanics north of Carneros Valley.

The mapped strand of the Carneros fault east of Sonoma coincides with a strong gravity gradient (Fig. 4), suggesting that the vertical dip observed at the surface continues at depth. Using this prominent gravity gradient, we extend the Carneros fault northwest of its mapped extent to upper Sonoma Valley and the southeast tip of the Maacama fault. The gravity gradient links the mapped extent of the Carneros fault to faults that juxtapose San Pablo Group sediments and overlying Sonoma Volcanics against Great Valley Sequence ∼7 km due west of Yountville. Thus, the Carneros fault forms a major fault that bounds the Mayacmas Mountains on the southwest. However, west of 122°30′W the Carneros fault is overthrust and obscured by reverse or thrust faults associated with the Mount St. John thrust (Graymer et al., 2007).

We can also extend the Carneros fault using the prominent gravity gradient south of its mapped extent to the north shore of San Pablo Bay and mouth of the Napa River. Fox (1983) suggested that the Carneros fault was a likely northward continuation of the Franklin-Sunol-Calaveras fault. Another possibility, which we discuss in more detail in the San Pablo Bay basin section, is that the Carneros fault is a continuation of the Pinole fault.

Offset of the Carneros fault includes both vertical and horizontal components. Based on stratigraphic relations, Weaver (1949) proposed at least 460 m of west-side-down offset on the fault. The facing direction of the gravity gradient supports this sense of offset, and the amplitude of the gradient suggests an even greater amount of offset. Fox (1983) postulated ∼35 km of right-lateral displacement by correlating Tertiary strata of the San Pablo Group south of Suisun Bay east of the fault with a similar sequence near Mount Veeder (Fig. 2). As pointed out by Fox (1983) and elaborated on in McLaughlin et al. (1996), the original configuration of the Tertiary depositional basin is not well known and thus estimates of offset must be viewed with caution. Correlating magnetic anomalies across the Carneros fault sourced by Neroly Sandstone (anomaly ne in Fig. 6) argues for a minimum of 15–20 km of offset.

The Carneros fault must mostly predate undated Sonoma Volcanics that overlap the Carneros fault northeast of the town of Sonoma. To the north (Wagner et al., 2004), the Carneros fault cuts the rhyolite of Bismarck Knob (dated as 5.7–6.1 Ma; R. Fleck, 2008, personal commun.) and is truncated by a younger fault that cuts the Lawlor Tuff (4.83 Ma). Thus most of the movement on this fault predates 4.83 Ma.

Sonoma Valley Faults

Faults mapped in Sonoma Valley (Wagner et al., 2003) are limited to a north-striking set of faults along the west side of the valley that extend as far north as Glen Ellen (West Sonoma Valley fault zone; WSVF in Fig. 2). These faults were originally proposed based on gravity data (Campion et al., 1984). These faults also are expressed in the magnetic data, truncating northwest-striking magnetic anomalies west of the fault zone (Figs. 6 and 7). Near Glen Ellen, northwest-striking faults appear to splay off of the north end of the fault zone and can be extended several kilometers to the northwest using gravity and magnetic gradients. At the south end, the fault zone appears to splay off of the Bennett Valley fault zone. Another fault on the east side of the valley was proposed based on an ∼1 mGal step (Campion et al., 1984; red dashed line in Fig. 3). This fault also is locally expressed at the surface (Wagner et al., 2004). We discuss how these faults relate to basins beneath Sonoma Valley in a later section.

Allen (2003) proposed a concealed fault beneath Sonoma Valley that he interpreted as the main northward continuation of the Hayward fault. He attributed as much as 67 km of right-lateral displacement to this fault since 5 Ma. A north-striking magnetic anomaly, however, extends across Sonoma Valley (dc in Figs. 6 and 7). Possible sources for this anomaly are Mesozoic serpentinite or ultramafic rocks or Sonoma Volcanics. The unusual orientation of the anomaly (N20°E), nearly parallel to the N14°E average azimuth measured for dikes in the Sonoma Volcanics by Fox (1983), leads us to interpret the anomaly as caused by a dike complex within the Sonoma Volcanics. The uninterrupted character of the anomaly argues against significant strike-slip offset through this part of Sonoma Valley since the intrusion of the dike complex.

Hayward–Rodgers Creek–Healdsburg–Bennett Valley–Maacama System

This fault system consists of several right-lateral, right-stepping, overlapping faults and is considered one of the main branches of the San Andreas transform system in northern California. From south to north, the system consists of the Hayward, Rodgers Creek, Bennett Valley, Healdsburg, and Maacama faults (Fig. 2). The Hayward and Rodgers Creek faults are not aligned but are offset by a right step of several kilometers. The Bennett Valley fault steps right of the Rodgers Creek fault ∼10 km north of San Pablo Bay and parallels the Rodgers Creek fault to the latitude of the city of Santa Rosa, where it changes to a northward strike and projects toward the Maacama fault. Also at the city of Santa Rosa, the Rodgers Creek fault steps or bends ∼1 km right onto the Healdsburg fault. Slip from the Rodgers Creek, Healdsburg, and Bennett Valley faults steps right onto the Maacama fault in a complex zone as wide as 10 km.

Northward continuation of the Hayward fault. The Hayward fault, last mapped at Point Pinole, is associated with a prominent gravity gradient that extends across San Pablo Bay and steps right 3–6 km just south of the northern San Pablo Bay shoreline. The maximum horizontal gradient (aligned blue circles in Fig. 4) at the southern bay shoreline does not exactly coincide with the mapped active trace of the Hayward fault, but is offset consistently ∼1 km to the northeast. The fault instead coincides with the top of the steep, east-facing gradient. The gravity gradient, however, coincides with a significant lateral contrast in seismic velocity and reflection character, interpreted to be an abandoned trace of the Hayward fault (Parsons et al., 2003). The active mapped trace of the Hayward fault coincides with the southwest edge of a magnetic anomaly from Point Pinole across San Pablo Bay to at least Sears Point (anomaly n in Fig. 6; Jachens et al., 2002). Tertiary volcanic rocks folded up against the fault (Wright and Smith, 1992) are the main source of this anomaly, although underlying magnetic Tertiary sedimentary rocks of the San Pablo and Contra Costa groups and magnetic ophiolite may also contribute to this anomaly (Jachens et al., 2002). Another magnetic anomaly is present on the southwest side of the fault along the northern stretch of the fault within the bay (anomaly m; Fig. 6). The wavelength of this anomaly suggests that the source is considerably shallower than the source of the anomaly on the northeast side of the Hayward fault, consistent with the facing direction of the gravity gradient. The source of the magnetic (and gravity) high is likely gabbro and related ophiolitic rocks because of nearby outcrops of lower Great Valley Sequence strata at Black Point (Fig. 2). This magnetic high was correlated with the magnetic and gravity high near San Leandro, arguing for a minimum total offset on the Hayward fault of 38 km (Jachens et al., 2002). In Graymer et al. (1995), it was shown that the source of the magnetic and gravity high, the San Leandro gabbro, is entirely within the long-term Hayward fault zone, such that the calculated offset represents only offset on the faults of the western part of the long-term fault zone. This closely approximates the ∼40 km offset along the western margin faults of the correlated Burdell Mountain Volcanics and the Northbrae rhyolite (Graymer, 2006), located at the north end of the Berkeley Hills volcanics (BHV in Fig. 1).

In Jachens et al. (2002), it was argued that the continuity of anomaly n precluded any direct connection between the upper parts of the Hayward and Rodgers Creek faults north of the south edge of this body; they allowed for several kilometers of offset within the right step south of the body, which coincides with a similar-trending gravity gradient. It was also argued (Jachens et al., 2002) that the apparent continuity of another anomaly (o in Fig. 6) parallel and 5–6 km east of anomaly n precluded any significant strike-slip offset on a southeastward extension of the Rodgers Creek fault. The continuity of anomaly o in filtered versions of the magnetic field (Figs. 9 and 10) is not obvious, but the anomaly changes to a more southward trend at or near the northward continuation of the Pinole fault. The lack of a clear connection between the Rodgers Creek and Hayward faults, however, does not preclude a throughgoing, segmented rupture via a normal fault link (Parsons et al., 2003), but suggests some kind of decoupling at depth to reconcile the magnetic evidence for limited offset versus geologic evidence for tens of kilometers of offset on the Hayward (as much as 100 km since ca. 12 Ma; Graymer et al., 2002) and Rodgers Creek (∼28 km since ca. 7 Ma; McLaughlin et al., 2005b) faults. Alternatively, the continuity of magnetic anomaly n may be fortuitous; a very slight inflection in the magnetic gradient points toward juxtaposition of two magnetic bodies, rather than a single continuous magnetic body.

Rodgers Creek–Bennett Valley–Healdsburg faults. North of San Pablo Bay, the prominent gravity gradient associated with the northward continuation of the Hayward fault steps right and weakens, and parts of the gradient continue along the Tolay, Rodgers Creek, and Bennett Valley faults (Fig. 4). Although the gravity gradient along the Rodgers Creek fault dissipates north of lat 38°15′N in the Sonoma Mountain area, the fault is well expressed in the magnetic field, bounding the northeast edge of a broad magnetic anomaly for nearly 20 km (Fig. 6). The source of the broad anomaly is likely serpentinized ophiolitic rocks as well as the Tolay Volcanics (as defined by Wagner et al., 2005, which also includes the Donnell Ranch Volcanics of Youngman, 1989). A prominent magnetic gradient also coincides with the north end of the Bennett Valley fault zone east of Santa Rosa, where it changes from a northwest to a northward strike and coincides with Holocene fault scarps (Figs. 6 and 7).

West of Glen Ellen, a gravity high straddles the Rodgers Creek and Bennett Valley fault zones, but then narrows to the northwest and reaches its maximum value beneath Bennett Valley (BVH in Fig. 3). In Bennett Valley, the eastern margin of the gravity high (pink shaded region in Fig. 11) roughly coincides with the Bennett Valley fault, whereas the western margin of the high coincides with the Rodgers Creek fault. The source of this gravity high is not obvious from surface outcrops, but is most likely dense Mesozoic rocks associated with a small outcrop of serpentinite in the Rodgers Creek fault zone (McLaughlin et al., 2008b). An oil-test well (Williams Jacobs No. 1; Jacobs in Fig. 11) encountered basalt at a depth of 305 m (Youngs et al., 1985). Despite a minimum thickness of 390 m of basalt encountered in this well, the gravity value at the well is still 10 mGal lower than the apex of the Bennett Valley gravity high. In Langenheim et al. (2008), dense basaltic andesite of the Sonoma Volcanics were ruled out as a source for this gravity high because, although these rocks are extensively exposed throughout the area, they do not always coincide with gravity highs and an unreasonably thick section (1–2 km) of basaltic andesite would be required to produce the amplitude of the gravity high.

A prominent magnetization and density boundary coincides with the Rodgers Creek fault south of where it is concealed by alluvium in downtown Santa Rosa and bends or steps ∼1 km to the right to coincide with the Healdsburg fault (Figs. 12 and 13). Northwest-striking magnetic anomalies of the Taylor Mountain area (Figs. 11 and 13) are truncated on the east by the Rodgers Creek fault; these anomalies highlight the structural grain of the transpressional Taylor Mountain and Cooks Peak fault zones and become attenuated beneath alluvial cover to the west, suggesting that these east-vergent thrust faults lose displacement to the north or plunge to the northwest. At lat 38°30′N, the gravity gradient deviates 2 km to the northeast of the Holocene strand of the Healdsburg fault (HF in Fig. 12). The position of the fault relative to the gravity gradient may signify a significant northeast dip and reverse slip on the Healdsburg segment or an earlier strand of the fault.

Maacama fault. The northernmost fault of this system in our study area is the Maacama fault. A cumulative right-lateral offset of 20–22 km of a distinctive terrane of Great Valley ophiolite, posited in McLaughlin et al. (2008a), is roughly the same as that documented for offset volcanic rocks dated as ca. 3 Ma. The fault forms the northeast margin of a 25-km-long band of gravity lows (Fig. 3). These lows cross both topographic highs and lows composed mainly of gravels and some units of the Sonoma Volcanics (Fig. 4). The gravity data provide evidence for long-term, southwest-side-down offset on the Maacama fault. The Maacama fault, along with the Healdsburg and Rodgers Creek faults, also marks the eastern termination of deep magnetic sources within the basement (Fig. 8).

Trenton Thrust Fault

The Trenton thrust fault (Fig. 11), exposed in a road cut ∼10 km north of Sebastopol, strikes ∼N70°–75°W and dips shallowly to the northeast (Fox, 1983). The fault places Franciscan Complex rocks over Miocene Wilson Grove Formation. The southeastward projection of the fault, covered by alluvium, is based on subdued scarps. The projection of the Trenton fault was linked to a system of thrust faults in the Taylor Mountain area southeast of Santa Rosa (McLaughlin et al., 2008b).

The Trenton thrust fault is aligned with the southern edge of a magnetic anomaly just east of basement outcrops. The magnetic anomaly broadens southeastward across the Santa Rosa Plain (Fig. 6). The southeastward projection of the Trenton thrust fault across the plain follows a weak magnetization boundary that is aligned with the southwest edge of a more intense, oblong-shaped anomaly at the eastern end of the Trenton Ridge (A′ in Fig. 13). The breadth of the magnetic high, however, suggests that magnetic rocks are not confined northeast of the projection of the Trenton thrust fault, but may extend southwestward at depth, consistent with a nearly coincident gravity high. The projected Trenton thrust fault bisects the gravity high (Fig. 12), indicating that the development of the subsurface basement ridge is not solely related to movement on the thrust fault (McPhee et al., 2007). The gravity low nestled between the Rodgers Creek fault and the basement ridge also suggests that the proposed connection between the Trenton thrust fault and thrust faults mapped to the southeast in the Taylor Mountain area would have significant relief.

Tolay and Other Northwest-Striking Faults South of the Santa Rosa Plain

The Tolay fault, the northernmost of a family of northwest-striking faults west of the Rodgers Creek fault and south of the Santa Rosa Plain (Fig. 2), has been attributed significant right-lateral offset, as much as 110 km (McLaughlin et al., 1996), to account for juxtaposition of Sonoma, Tolay (as defined by Wagner et al., 2005), and Burdell Mountain Volcanics. This fault and others, such as the Petaluma Valley, Bloomfield, and Burdell Mountain faults, have been considered extensions of the Hayward fault (proto-Hayward fault). Although Fox (1983) and Sarna-Wojcicki (1992) attributed 45 ± 10 km of right-lateral offset since 8 Ma to the Tolay fault, in Wagner et al. (2002a, 2002b, 2005) it was shown that the Tolay fault is a zone of disparate, generally southwest-dipping thrust and reverse faults with no significant strike-slip offset since the deposition of the Roblar Tuff (6.26 Ma).

The Tolay fault coincides with a north-facing gravity gradient that is consistent with significant vertical offset, with southwest-side-up movement (Fig. 4). The fault also coincides with a significant south-facing magnetic gradient in medium and deep wavelength fields (Figs. 8 and 9). The gravity and magnetic gradients diverge from the fault near the western terminus of the mapped fault zone. The southern edge of the broad magnetic high (just south of the Murphy well in Fig. 9) steps to the northeast ∼3 km at about the same place the gravity gradient changes to a more westward trend. We speculate that the right step in the southern boundary of the magnetic high restores against the magnetic rocks along the southern margin of the Cotati basin (black arrows in Fig. 9), suggesting perhaps ∼8–10 km of right-lateral offset.

Alternatively, 35 km of right-lateral offset on the Petaluma Valley fault (gray line in Fig. 2) since 6 Ma was proposed in Graymer et al. (2002). The northern part of the fault coincides with a gravity gradient marking the southern margin of the Cotati low (Co in Fig. 3), but obliquely crosses the gravity gradient associated with the Tolay fault. The location of the southern extent of the Petaluma Valley fault is inferred to pass south of the Tolay fault, based on the distribution of Roblar Tuff. Its location is plausible given that it does not cut across continuous magnetic anomalies. However, its offset may be closer to 10–20 km to align better with gravity and medium wavelength magnetic gradients, the Roblar Tuff as mapped in Wagner et al. (2002a) and Clahan et al. (2003), and fold and thrust belts at Meacham Hill and along the Tolay fault.

The Bloomfield fault (Fig. 2), considered to be an extension of the Tolay fault (McLaughlin et al., 1996), places Franciscan Complex rocks on the northeast against Miocene Wilson Grove Formation to the southwest (Bezore et al., 2003). The facing direction of the gravity gradient associated with this fault is consistent with northeast-side-up displacement. This fault, along with other N60°W striking faults in the area, is parallel to a weak to moderate grain in the filtered magnetic field (Fig. 10) that reflects structural grain in the Franciscan Complex. This observation is consistent with that of Fox (1983), who noted that these faults probably originated as shear zones in the Franciscan Complex, but that some were later reactivated during or after deposition of the Wilson Grove Formation.

The southernmost mapped fault in this family is the Burdell Mountain fault, which bounds the northeast side of Burdell Mountain and juxtaposes Great Valley Sequence rocks and Burdell Mountain Volcanics on the southwest against Franciscan Complex rocks to the northeast (Fig. 2). Ford (2007) proposed 10 km of right-lateral offset of the Novato Conglomerate. The fault is poorly expressed in the gravity field, but a speculative correlation of shallow and medium depth magnetic anomalies suggests between 5 and 10 km of right-lateral offset (blue arrows in Figs. 9 and 10).

Normal Faults on West Side of Santa Rosa Plain

A poorly defined zone of north-striking normal faults marks the west side of the Santa Rosa Plain (Fig. 11; McLaughlin et al., 2005a). Only two localities show evidence of Quaternary normal faulting (McLaughlin et al., 2005a, p. 62; n in Figs. 11 and 12), but continuity of faulting is supported by the abundant springs and ponded water along the western curvilinear margin of the Santa Rosa Plain.

The lack of strong geophysical expression of these faults suggests that the offset, both horizontal and vertical, is not large (<500 m). Locally, the mapped faults coincide with gravity gradients marking the edge of concealed basins beneath the Santa Rosa Plain (that is, the east-side-down normal fault north of the Trenton thrust fault), but in general, the strongest gravity gradients are 2–4 km east of the mapped faults (Fig. 12). Thus, the faults with the greatest displacement along the west side of the plain are concealed beneath alluvial deposits.

Cenozoic Basins and Volcanic Centers

Napa Valley

The basin configuration beneath Napa Valley indicates three main subbasins just east of the West Napa fault partitioned by a shallow ridge of dense rock in the area of Yountville (Y in Fig. 5). East of Yountville, Great Valley Sequence rocks and the Stags Leap stock exposed in low hills that protrude above the valley floor support the gravity inversion results for shallow basement. Although Howell and Swinchatt (2003) interpreted these low hills as megalandslide deposits, the high gravity values suggest that the stock and Great Valley Sequence rocks are in place or, if indeed landslide deposits, were slid into a depression with very thin basin fill. The gravity high also coincides with the extent of the tilted remnants of a 3.9 Ma basaltic andesitic stratovolcano (Sweetkind et al., 2005; J. Rytuba, 2008, personal commun.). The stratovolcano remnant coincides with a complex magnetic anomaly pattern that is muted in the center of the valley (red circle labeled Y in Fig. 10).

The thickest part of the basin fill is located in the northern subbasin, south of Calistoga (C in Fig. 5) over exposures of Sonoma Volcanics. The magnitude of the thickness may be overestimated in this area (4 km) and other highlands composed of Sonoma Volcanics (such as those north of Calistoga and east of Napa) because of underestimating the density contrast in this area. In the southern part of the valley, gravity data reflect the presence of a generally circular accumulation of low-density volcanic rocks, with a slightly higher density feature centered within it possibly reflecting a resurgent dome associated with the youngest caldera. This is mirrored in the magnetic data, indicating a roughly circular pattern (red circle labeled N in Fig. 10) with a central magnetic high. Nonetheless, the thickest basin fill is adjacent to the West Napa fault, both north and south of the basement ridge near Yountville.

The magnetic pattern of the Napa Valley reflects the imprint of volcanism and, in the northeastern part of the valley, the effects of hydrothermal alteration (Sweetkind et al., 2005). Low magnetic values near Yountville (Y in Fig. 6) in the apex of the gravity high reflect nonmagnetic Great Valley Sequence rocks and Stags Leap stock, but may also reflect associated hydrothermal alteration in this area. A broad magnetic high to the northwest most likely reflects ophiolite and the absence of extensive hydrothermal alteration.

San Pablo Bay

The basin configuration (Fig. 5) shows that the deep rhombochasm beneath San Pablo Bay has thicker pockets of basin fill elongated along the Hayward and Carneros faults separated by an intrabasinal high. This configuration mimics lows in gravity data filtered to enhance density sources in the shallowest part of the crust (Parsons et al., 2003). Parsons et al. (2003) interpreted the residual gravity low along the Hayward fault to indicate a pull-apart basin developed between the Rodgers Creek and Hayward faults. Thus, we would argue that the subbasin bounded on the northeast by the Carneros fault is also a pull-apart basin transferring slip between the Carneros and the Pinole faults (Fig. 5). The age of this proposed pull-apart basin is not well known; however, a drill hole (General Crude Cullinan #1; Wright and Smith, 1992; 2 in Fig. 5) penetrated a 1.7 km section of continental or shallow-marine sediments overlying the Late Miocene Neroly Sandstone. The post-Neroly age of these sediments is consistent with the hypothesis that they were deposited in a sedimentary basin formed after 8.5 Ma and consistent with mapped relationships along the Carneros fault indicating that the Carneros fault is older than 4.83 Ma.

Sonoma Valley

Beneath Sonoma Valley, a shallow basement ridge near the town of Glen Ellen (Gl in Fig. 5) separates two main basins that are 2–3 km deep. The northern basin, centered near the town of Oakmont (Oak in Fig. 5), is ∼2–3 km wide and 6–8 km long. Its margins are roughly parallel, have a northwest trend, are aligned with mapped faults at its southwest and northeast ends, and coincide with magnetization boundaries. Its geometry and aspect ratio suggest a pull-apart origin.

The age of the Oakmont basin is not directly known, but a possible constraint comes from an inferred eruptive center at the southeast margin in the Sugarloaf Ridge–Nunns Canyon area, (Su in Figs. 3 and 4). This area is marked by a gravity high around a central moderate gravity low, and forms an arcuate topographic high comprised of rhyolite, tuff breccia, and basaltic andesite ring dikes (Delattre et al., 2007). The volcanic section, which also includes near-vent breccia deposits containing volcanic bombs, coincides with strong negative magnetic anomalies (Su in Fig. 6). The negative anomalies are consistent with paleomagnetic data on rhyolite dated by K-Ar methods as 5.3 Ma near the base of the volcanic section, indicating reversed directions (Mankinen, 1972). The geologic and geophysical data indicate the presence of an eruptive center at the southeast margin of the Oakmont basin. Mapping relations suggest that some of the mafic flows from this volcanic center predate opening of the Oakmont sedimentary basin (Delattre et al., 2007); these flows are interbedded with tuff that contains tephra correlated with Carriger tuff (4.81 Ma) and Mark West Springs tuff (between 4.83 and 5.0 Ma).

The southern basin is more areally extensive and complex, with subbasins extending northwest from San Pablo Bay into the eastern and central parts of Sonoma Valley (Fig. 5). The east-side fault forms part of the eastern margins of these subbasins, and based on the inversion, the fault (or faults) extends another 7–8 km to the southeast. The southwest margin of the southern basin beneath Sonoma Valley and San Pablo Bay appears to be stepped, with steps coincident with the Tolay, Rodgers Creek, Bennett Valley, and West Sonoma Valley faults. The West Sonoma Valley fault zone forms the western edge of a 2–3-km-wide subbasin (West Sonoma basin, immediately west of S in Fig. 5). This fault has evidence for both normal and strike-slip displacements. The eastern margin of the West Sonoma basin coincides with a 2-km-wide magnetic anomaly (dc in Fig. 6) that we attribute to a dike complex in the Sonoma Volcanics.

The age of the western Sonoma Valley basin is unknown, but may be related to volcanic centers inferred by magnetic anomaly patterns at the northern and southern ends of the north-striking magnetic anomaly (red circles labeled SoH and vc in Fig. 10). The magnetic pattern at the southern end is more subdued and broader than that at the northern end near Mount Veeder, but both areas coincide with groundwater temperatures exceeding 30 °C (Campion et al., 1984; Farrar et al., 2006). In addition, the northern magnetic pattern coincides with a region where the basin inversion predicts a 2–4-km-thick section of low-density materials in the highlands north of the town of Sonoma (SoH in Fig. 5). If the volcanic section is composed of substantial amounts of tuff or other low-density rocks, the inversion may locally overestimate the thickness of fill in this area. In this area, Wagner et al. (2004) mapped a complex and heterogeneous package of volcanics, the textures and areal extents of which suggest proximity to or inclusion of one or more eruptive centers. These rocks are older than 4.83 Ma (Wagner et al., 2004) and perhaps constrain the age of the opening of the West Sonoma basin if the volcanics are related to the inferred dike complex beneath Sonoma Valley. If the dike complex intruded into the basin margin fault, the basin is older than 4.83 Ma. Alternatively, if the dikes provided a structural break that the basin-margin fault or faults later occupied, then the opening of the basin could be younger than 4.83 Ma.

Bennett-Rincon Valley

Bennett and Rincon Valleys form a topographic depression between the Rodgers Creek and Maacama faults (Fig. 11). The 6-km-wide right step between the two faults produces a pull-apart basin that is moderately well expressed topographically but poorly expressed in the gravity field, consistent with geologic evidence that the basin formed only during the past 1 m.y. (McLaughlin et al., 2005a, 2008b). The low-density fill of the Rincon-Bennett Valley pull-apart basin is thin, with older volcanics exposed in the central part of the depression, and its gravity signature is in part masked by the Bennett Valley gravity high. Only in Rincon Valley do the gravity data image thicker Cenozoic fill. This local accumulation appears to be related to folding (Langenheim et al., 2008).

Santa Rosa Plain

The relatively flat valley floor of the Santa Rosa Plain conceals the deepest sedimentary basins west of the Rodgers Creek fault. Gravity data indicate that a shallow concealed basement ridge (Trenton Ridge; TR in Figs. 3 and 4) at the latitude of the city of Santa Rosa separates the Windsor basin (W in Figs. 3 and 4) to the north from the larger and more complex Cotati basin (Co in Figs. 3 and 4) to the south (McPhee et al., 2007). Seismic-reflection profiling confirms the general shape and depth extent of the Trenton Ridge derived from gravity data (Williams et al., 2008). The concealed ridge does not extend completely across the basin; damage from the 1906 M7.9 San Francisco earthquake and the 1969 M5.6 and M5.7 Santa Rosa earthquakes was concentrated in Santa Rosa (SR in Fig. 5) between the eastern end of the Trenton Ridge and shallow basement marking the eastern margin of the Cotati basin along the Rodgers Creek fault zone. Ground-motion simulations of the 1906 and 1969 earthquakes using the gravity-defined basement surface (McPhee et al., 2007) show enhanced ground motions in the northeast corner of the Cotati basin, suggesting that basin edge effects were important in producing shaking damage (McPhee et al., 2007).

Analysis of seismic-reflection data points to a steeply dipping fault that bounds the southwest edge of the Trenton Ridge and displaces reflections attributed to basement and Pliocene–Quaternary sedimentary rocks down to the southwest (Williams et al., 2008), rather than a gently dipping fault (for example, an extension of the mapped Trenton thrust fault). If so, the structures that produced the ridge may also include oblique-slip faults given the current stress field (maximum horizontal stress oriented N14°E; Provost and Houston, 2003, their site NCA29). Maximum horizontal stress directions from average dike azimuths (Fox, 1983) and slip vector data from faults cutting 4 Ma volcanics (McLaughlin et al., 2002) range from N10°W to N20°E, which would produce right-lateral to oblique thrust movement on a west-northwest–striking fault. Magnetic data filtered to enhance shallow sources suggest an apparent right-lateral offset along the mapped Trenton thrust (arrows in Fig. 13). We speculate for even more right-lateral offset, as much as 8–10 km, on associated faults, to restore the magnetic and gravity high at the eastern end of the Trenton Ridge (A′ in Figs. 12 and 13) back to the gravity and magnetic highs at the western end of the ridge (A in Figs. 12 and 13), thus opening the Windsor basin to the north. Increasing downwarp of the reflections across the Trenton Ridge suggests that the ridge formed over time, coincident with the development of the Cotati and Windsor basins (Williams et al., 2008).

The Windsor basin is a roughly triangular basin bounded on the northeast by the Healdsburg fault, on the south by the Trenton Ridge, and on the west by a poorly defined zone of normal faults (Figs. 11 and 12). The Holocene strand of the Healdsburg fault is parallel to but 2 km southwest of the geophysically determined northeast basin margin and does not offset the basin margins. The discrepancy in location between the Holocene strand of the Healdsburg fault and the basin margin may reflect an earlier geometry of the Healdsburg fault, the Holocene strand reflecting a young adjustment of the fault within an older right step. Gravity data filtered to enhance subtle features show an 8-km-long, 2-km-wide gravity low (Fig. 12) that we speculate represents an older pull-apart subbasin superposed on the northeastern margin of the larger Windsor gravity low. Closing this subbasin by restoring 8 km of right slip brings magnetic highs into alignment (m in Figs. 13 and 14). The Holocene fault trace cutting across the subbasin and seismicity indicating a steep northeast dip and right-lateral slip on the fault (Waldhauser and Ellsworth, 2000; Wong and Bott, 1995) are both consistent with the fault having successfully propagated across the right step. Sandbox models of strike-slip basins show that faulting in extensional stepovers evolves with time such that younger fault strands often bisect the strike-slip basin as the fault zone geometry changes (Dooley and McClay, 1997). The older configuration of the Healdsburg fault probably dates to before ca. 3 Ma, because paleocurrent directions in the Glen Ellen Formation (McLaughlin et al., 2005a) cross the subbasin without deflection. The offset estimate is similar to that from matching obsidian clasts in gravels southwest of the Healdsburg fault with obsidian dated as 4.5 Ma northeast of the Bennett Valley fault (6–8 km; McLaughlin et al., 2005a).

The southern and western margins of the Windsor basin appear to step downward into the basin. The strongest gravity gradients on the west side of the basin are 2–3 km east of the westernmost north-striking faults (as mapped by McLaughlin et al., 2005a). The western basin margin appears to step east at its northern end (Figs. 11 and 12) near the town of Windsor. The southern margin appears to slope gently from the basement high of the Trenton Ridge into the basin, but a gravity gradient 2 km north of the main Trenton Ridge gradient suggests that the gentle slope has at least one step that may be fault related.

The Cotati basin is more complex than the Windsor basin, with two gravity-defined subbasins divided by an east-trending basement ridge that is within 1–1.2 km of the ground surface (Warrington high in Fig. 11; WH in Figs. 3, 4, and 5). The northern subbasin, named the Bellevue low in McLaughlin et al. (2008b), is bounded on the east by the Rodgers Creek fault and by the uplifted Cotati basin margin exposed west of the Rodgers Creek fault in the Taylor Mountain area (TM in Fig. 5). In this area, rhyodacitic breccia interbedded with the lower part of the Petaluma Formation was interpreted in McLaughlin et al. (2008b) to be formed along a fault scarp, indicating an early episode of transtensional faulting along the basin margin ca. 6–7 Ma.

The Warrington high (WH in Figs. 12 and 13) forms the southern margin of the Bellevue subbasin (Be in Figs. 5 and 12). The orientation of the high within past stress field orientations (Fox, 1983; McLaughlin et al., 2002) suggests it is likely thrust related, either an uplifted block along a concealed thrust fault or an anticlinal fold, or both. It may also have a right-lateral component of slip where its strike changes to a more northward direction as the high approaches the Rodgers Creek fault to the east (Fig. 11). The western margin of the Bellevue subbasin is 3–4 km east of the north-striking normal fault zone marking the western edge of the Santa Rosa Plain, suggesting an earlier history concealed beneath the Santa Rosa Plain.

The subbasin south of the Warrington high reaches its deepest extent beneath the city of Rohnert Park. This subbasin contains the only deep wells that penetrate basement, such as the Stephens-Rohnert No. 1 (Stephens in Fig. 11), which encountered Franciscan Complex rocks at a depth of 1.68 km. The subbasin's northeastern margin as defined geophysically does not coincide with the Rodgers Creek fault, but is the southwest edge of the Warrington high and is marked by the northeast edge of a 250 nT magnetic high (Fig. 6). The source of the magnetic high is probably composite, the higher frequency part of the anomaly being caused by volcanic rocks (127 m of volcanic rock were penetrated at the bottom of drill hole RN-14 in Fig. 11; Herbst, 1979) and the broader part of the anomaly (Fig. 8) that extends ∼35 km to the southeast being caused by serpentinized ophiolitic rocks (and possibly Tolay Volcanics as defined in Wagner et al., 2005). We speculate that this belt of magnetic highs may be related to the broader Trenton Ridge magnetic high, compatible with the idea of a transtensional origin for the Cotati subbasins and an additional 10 km of right-lateral offset. The southwest edge of this magnetic belt extends south of the gravity-defined basin margin, discussed in more detail in the following.

The southern margin of the Cotati basin defined by gravity gradients (dotted white lines in Fig. 5) is as much as 3–4 km north of the nearest mapped Quaternary faults of Graymer et al. (2006a). Part of the margin, however, coincides with the Petaluma Valley fault of Graymer et al. (2002), although our basin margin is less linear than that shown in Graymer et al. (2002), reflecting detail from recent gravity measurements (Langenheim et al., 2006b). The basin margin is immediately north of a folded section consisting of the Petaluma Formation (Clahan et al., 2003; Bezore et al., 2003). The folded section is overlain by nearly flat-lying Wilson Grove Formation containing the Roblar Tuff (Bezore et al., 2003), dated as 6.26 Ma. Using the gravity gradient as a guide, we suggest that the basin margin (and presumably the folded section) continues concealed beneath the Wilson Grove Formation west-northwest of 122°45′W for 7–8 km (Fig. 4).

The southern basin margin strikes nearly east-west before curving to the southeast into the Petaluma basin. Where the basin margin curves to the southeast, the gravity gradient appears to step right just south of the faulted anticline at Meacham Hill (MH in Fig. 5). Tolay Volcanics (as defined by Wagner et al., 2005) exposed in the core of the anticline coincide with high-frequency magnetic anomalies (Fig. 10); this anomaly pattern can be traced to the north, where it appears to be truncated by the basin margin as defined by gravity. The gravity gradient passes uninterrupted across the topographic divide between the Santa Rosa Plain and Petaluma Valley.

DISCUSSION

Speculations on Basin Formation

Napa and Sonoma Valleys conceal complex basin configurations that reflect the influence of the strike-slip faulting, folding, and volcanism, whereas basins beneath the Santa Rosa Plain likely formed by transtension, now overprinted by transpression. On the western side of Napa Valley, long, linear basins point to a combination of folding and reverse faulting associated with the West Napa fault. Napa Valley is a synform, with measured attitudes in the Sonoma Volcanics on either side of the valley dipping toward the valley (Graymer et al., 2007). Superposed on these long, narrow basins is the central high near Yountville and the circular depression east of the city of Napa that reflect the influence of volcanism. The Yountville high reflects not only the remnants of a tilted stratovolcano, but also relative uplift because of exposures of the Stags Leap stock and Great Valley Sequence in this area. The Yountville high coincides with a change to more westward trends in the Soda Creek and Green Valley faults that would produce contraction and uplift in this region. In southern Napa Valley a circular depression with a central high is inferred to be the source area for several ash flows and Plinian eruptions. The greater depth of the narrow basin between the West Napa fault and the central high of this depression probably reflects superposition of the inferred caldera and downdropping along the West Napa fault. The depth of the basin may also reflect part of the right-lateral slip along the West Napa fault that stepped right onto the Soda Creek fault.

In the northern part of the block northeast of Calistoga, thicker basin fill reflects lower density volcanic rocks associated with the Mount St. Helena and Wildlake eruptive centers (J. Rytuba, 2008, personal commun.). We speculate that the apparent right step in the western branch of the Green Valley fault may have facilitated these eruptive centers, as has been suggested for other calderas (Self et al., 1986; Moore and Kokelaar, 1998). To the south, the Green Valley fault may have been influenced by the location of the southeastern caldera complex; volcanic rocks clearly postdate faulting because they are offset as much as 20 km by the Green Valley fault (Graymer et al., 2002; Table 2).

Sonoma Valley, like Napa Valley to the east, is a synform. The basin inversion, however, suggests that two strike-slip basins are superposed on this synform; the bounding faults at the corners of the basins led to eruptive centers. If basin lengths are proxies for cumulative offsets, then the Oakmont basin argues for 6–8 km of offset. The West Sonoma basin argues for ∼5–6 km of right-lateral offset. The similar amount of offset on these two basins suggests that they are related, perhaps the basement high near Glen Ellen reflecting a left bend from the West Sonoma basin fault system onto the Oakmont basin fault system. Slip from this system of faults would likely be transferred to the Maacama fault to the north. The timing of basin opening is poorly constrained and indirect, but could be younger than 4.83 Ma.

The geometry of basins beneath the Santa Rosa Plain suggests a transtensional origin that has superposed oblique-slip thrust faulting along its southern and eastern margins (McLaughlin et al., 2008b). Oblique thrust faulting probably also led to the formation of the concealed Trenton Ridge and Warrington high, although the timing and magnitude of thrusting likely vary across the Santa Rosa Plain. We speculate that the concealed ridges may have been formed by right-lateral faults, the current orientations of which are favorable for thrust or oblique-slip faulting. This implies that stress orientations may have rotated clockwise with time.

The Petaluma basin, bounded on the southwest by the Tolay fault, may also be an old transtensional basin that was subsequently deformed by transpressional deformation manifested by folding (e.g., the faulted Adobe anticline drilled into in the Petaluma oil field; Wright, 1992) and thrusting. The Petaluma basin may have been once continuous with the Cotati basin, but folding and thrusting in the Meacham Hill area may have subsequently deformed and obscured the basin margins.

Partitioning of Right-Lateral Offset North of San Pablo Bay

We discuss here how 175 km of right-lateral offset on the East Bay fault system is partitioned north of San Pablo Bay, based on geophysical and geologic interpretations. Evidence for the large-scale offset on the East Bay fault system includes correlation of the Burdell Mountain Volcanics near Novato with the Quien Sabe Volcanics south of Hollister (McLaughlin et al., 1996; Graymer et al., 2002; Fig. 1). Geophysical data shed light on cumulative offset, mostly by limiting where significant offset can be accommodated in the valley areas, north of Napa Valley, and south of the Santa Rosa Plain.

Offset through the Valleys

Several studies have proposed significant offset through alluvial valleys north of San Francisco Bay to carry some of the 175 km of right-lateral offset on the East Bay fault system. Our estimates of offset to produce the basins beneath the Santa Rosa Plain are no more than 30 km, <10 km in Sonoma Valley, and 10 km in Napa Valley. The lengths of the pull-apart basins beneath San Pablo Bay give estimates of cumulative offset along the Rodgers Creek–Hayward and Carneros-Pinole faults. The basin between the Rodgers Creek and Hayward faults suggests an offset of 10–15 km, whereas the basin between the Carneros and Pinole faults suggests ∼20 km, significantly less than geologic estimates of 28 and 35 km (Table 2). This discrepancy argues for the evolution of a single strand to a stepped-fault geometry. The structure beneath San Pablo Bay still awaits full resolution.

Offset South of Santa Rosa Plain

Although as much as 80–110 km of offset has been attributed to west-northwest–striking faults south of the Santa Rosa Plain, we argue that offset is likely to be <20 km for two reasons. One, such a fault geometry produces a large left or restraining bend (change in strike of >30°) that is only moderately topographically apparent south of that fault segment. Thermochronologic data (McLaughlin et al., 1996) suggest that any significant unroofing and uplift expected from such a restraining bend took place in this area before initiation of the East Bay fault system ca. 11–12 Ma, based on the age of subaerially erupted Burdell Mountain Volcanics and the age of shallow (<1.2 km) epithermal veins (McLaughlin et al., 1996). A smaller restraining bend (change in strike of <15°) in the San Andreas fault in the northern Santa Cruz Mountains produces larger topographic relief (Fig. 1) and ∼3 km of unroofing during the past 4.6 m.y. (Burgmann et al., 1994). Two nearly identical offset estimates on the San Andreas fault north and south of this region (∼300 km; Jachens et al., 1998) argue against significant offset on the west-northwest–striking faults. One would predict San Andreas fault offset to increase abruptly north of the intersection with these faults, if these faults extend all the way west to the San Andreas fault. The predominantly parallel northwest-trending magnetic grain coincident with these faults argues that these faults continue to the San Andreas fault.

Offset Partitioning onto the Maacama Fault

Both the Carneros (35 km offset) and West Napa faults (10 km or more offset), based on their geophysical expression, curve toward the southern end of the Maacama fault, suggesting that their slip is transferred onto the Maacama fault. The offsets on the Rodgers Creek (28 km; McLaughlin et al., 2005b) and Bennett Valley (unconstrained, but most likely <5 km) faults, as well as offset from the pull-apart basins beneath Sonoma Valley (5–8 km), also feed onto the Maacama fault. Yet the amount of cumulative offset on the Maacama fault north of these junctions is posited to be only 20–22 km, based on correlating ophiolite terranes near Hopland west of the fault (Feliz Creek section; FC in Fig. 1) with the Geyser Peak ophiolite east of the fault (GP in Fig. 1; McLaughlin et al., 2008a). The Coast Range ophiolite consists of two distinctive terranes, the Elder Creek and Del Puerto. The only known occurrence of the Elder Creek terrane west of the Maacama fault is the Feliz Creek section, whereas the Elder Creek terrane is interleaved with Franciscan Complex rocks east of the fault and northwest of Lake Berryessa. Either the posited offset on the Maacama fault is in error, or modification of the Carneros and West Napa fault traces is suggested.

Modification of the original pathways for the Carneros and West Napa faults is possible. The northward continuations of these faults begin to curve near their junction with the Mount St. John thrust, suggesting that these faults may be linked. If the West Napa and Carneros faults were originally straight, offset from the West Napa and Carneros faults need not continue onto the Maacama fault. However, the continuity of the northwest-striking magnetic anomaly ∼10 km north of Mount St. Helena (Fig. 8) suggests that significant strike-slip offset does not continue north of Mount St. Helena. Bending of the faults would lead to contraction between the West Napa and Carneros faults and is supported by a series of folds and thrust faults (such as the Mount St. John thrust; Fig. 2) that involve both the Sonoma Volcanics and the basement rocks west of the fault. The contractional deformation in this area may reflect a complex transfer of right-lateral slip from the West Napa to the Maacama fault, if not at the surface, possibly at depth, as suggested by relocated seismicity of Waldhauser and Ellsworth (2000; Fig. 8).

The timing of the contraction and accompanying uplift of the southern Mayacmas block must be in part younger than early Pliocene because obsidian from Napa Glass Mountain dated as ca. 2.8 Ma on the east side of Napa Valley is found in gravels in the Glen Ellen Formation west of this block (McLaughlin and Sarna-Wojcicki, 2003, McLaughlin et al., 2004, 2005a). This correlation and the sedimentary characteristics of the Glen Ellen Formation suggest that the gravels were deposited as westward-prograding alluvial fans prior to development of the present topography of the southern Mayacmas block (Graymer et al., 2007). We speculate that some of the elevated topography of the Mayacmas Mountains may have been a response to reverse movement on the West Napa fault and contractional structures, such as the Mount St. John thrust, associated with the gradually more pronounced left bend in the West Napa fault north of the town of Napa since ca. 3 Ma. Furthermore, slip from the West Napa onto the Mount St. John thrust fault since ca. 3 Ma may explain the absence of geomorphic expression of the West Napa fault north of St. Helena. A detailed paleomagnetic study is needed to test this hypothesis since vertical-axis rotation would be predicted in the area southwest of the curvilinear West Napa fault.

Influence of Preexisting Structure

We can account for 175 km of right-lateral offset to restore the Burdell Mountain Volcanics against the Quien Sabe Volcanics, if we assume the maximum offset estimate for each of the 8 faults listed in Table 2. However, no matter the pathways for the 175 km of offset north of San Francisco Bay, restoring the two volcanic fields juxtaposes very different magnetic patterns across the East Bay fault system (Fig. 15). Although we only show magnetic anomalies filtered for deep sources, this mismatch in magnetic pattern also holds true for intermediate and short wavelength anomalies, and indicates that immediately north of the restored volcanic fields is a fundamental difference in basement terranes across the East Bay fault system that is mirrored in the mapped geology. This suggests that the older basement structure guided the location of the Hayward–Rodgers Creek–Maacama fault system. Double-difference relocated seismicity suggests that Mesozoic basement structure as reflected by deep magnetic sources in the North Bay (Fig. 8) continue to influence where slip can be accommodated.

The influence of preexisting structure is also evident at a smaller scale within the Rodgers Creek–Maacama fault system. The presence of the Bennett Valley gravity high may restrict where one places the connection between the Rodgers Creek and Bennett Valley–Maacama faults through Bennett Valley. The apparently continuous dense body associated with the gravity high seems to preclude any fault with significant offset connecting the two faults between Santa Rosa and Sonoma Mountain. The dense body may also have influenced the change in trend of the Bennett Valley fault at its southern end, as evidenced by the uninterrupted and linear northeast edge of the gravity high. Seismicity (Waldhauser and Ellsworth, 2000; Fig. 11) also appears to be concentrated along the margins or outside of the gravity high. Perhaps the source of the gravity high, inferred here to be mafic rock, is more rigid than the surrounding basement rocks and concentrates stress along its margins, similar to the mechanism for concentrating stress for intraplate earthquakes proposed by Campbell (1978).

CONCLUSIONS

Restoring the Burdell Mountain Volcanics against the Quien Sabe Volcanics juxtaposes very different magnetic anomaly patterns, a juxtaposition that suggests that the East Bay fault system and its extension north of San Francisco Bay took advantage of a preexisting basement structure. This older basement structure also appears to influence seismicity patterns in the North Bay region. Complex basin configurations beneath the alluvial valleys point to a superposition of faulting, folding, and volcanism that reflects the interplay of processes along the transform margin, the northward migration of the Mendocino Triple Junction, and influence of older basement structure.

We thank the National Cooperative Geologic Mapping Program for financial support. We are grateful to Ramon Arrowsmith, Harvey Kelsey, Ed Mankinen, and George Saucedo, whose comments helped focus and improve the paper. We also would like to thank the many landowners in Napa, Sonoma, and Solano counties who granted us access.