Detailed analysis of continuously cored boreholes and cone penetrometer tests (CPTs), high-resolution seismic-reflection data, and luminescence and 14C dates from Holocene strata folded above the tip of the Ventura blind thrust fault constrain the ages and displacements of the two (or more) most recent earthquakes. These two earthquakes, which are identified by a prominent surface fold scarp and a stratigraphic sequence that thickens across an older buried fold scarp, occurred before the 235-yr-long historic era and after 805 ± 75 yr ago (most recent folding event[s]) and between 4065 and 4665 yr ago (previous folding event[s]). Minimum uplift in these two scarp-forming events was ∼6 m for the most recent earthquake(s) and ∼5.2 m for the previous event(s). Large uplifts such as these typically occur in large-magnitude earthquakes in the range of Mw 7.5–8.0. Any such events along the Ventura fault would likely involve rupture of other Transverse Ranges faults to the east and west and/or rupture downward onto the deep, low-angle décollements that underlie these faults. The proximity of this large reverse-fault system to major population centers, including the greater Los Angeles region, and the potential for tsunami generation during ruptures extending offshore along the western parts of the system highlight the importance of understanding the complex behavior of these faults for probabilistic seismic hazard assessment.
Emerging Recognition of Thrust Fault Hazards
The recognition of the hazards posed by thrust fault earthquakes to urban centers around the world has been highlighted by several recent events (e.g., 1994 Mw 6.7 Northridge, 1999 Mw 7.6 Chi-Chi, 2005 Mw 7.5 Kashmir, 2008 Mw 7.9 Wenchuan). These earthquakes demonstrate the need to better understand the behavior of these faults and their associated folds, particularly when these faults are “blind,” that is, where the faults do not reach (or “see”) the surface. The 2008 Mw 7.9 Wenchuan earthquake further illustrated that thrust fault ruptures may link together adjacent faults to generate extremely large-magnitude earthquakes. In southern California, the prospect for large, multiple-segment thrust fault ruptures remains poorly understood. However, the numerous large reverse- and oblique-slip faults in the Transverse Ranges suggest that such large earthquakes are possible, and their occurrence would represent a serious hazard to property and life in the densely populated southern California region.
Most seismic hazard assessments and models of reverse fault earthquakes in southern California involve the rupture of individual faults in the Transverse Ranges in moderately large magnitude (≥Mw 7) events (e.g., Sierra Madre fault, San Cayetano fault; WGCEP, 1995; Field et al., 2009), although more recent models allow cascade failures encompassing multiple faults (e.g., Field et al., 2013). While the seismic threat posed by these individual faults is significant, as was demonstrated by the 1994 Mw 6.7 Northridge earthquake (the costliest natural disaster in U.S. history prior to Hurricane Katrina [Scientists of the U.S. Geological Survey and the Southern California Earthquake Center, 1994]), a larger threat presents itself if several of these faults rupture together. The reverse- and oblique-slip faults of the Transverse Ranges form an interconnected, >200-km-long network of faults that could potentially rupture together to cause large-magnitude events. Although the potential for these faults to link and rupture together has been recognized (e.g., Dolan et al., 1995; Hubbard et al., 2014), relatively little is known about the ages, repeat times, and magnitudes of paleoearthquakes generated by faults within the Transverse Ranges.
In this paper, we apply a multidisciplinary approach utilizing continuously cored borehole and cone penetrometer test (CPT) data, in conjunction with high-resolution and deeper-penetration petroleum industry seismic-reflection data, to document the structural evolution of young folds generated by earthquakes on the Ventura fault, a major reverse fault in the western Transverse Ranges. Together, these new data allow us to assess the geometry of buried fold scarps and identify periods of stratigraphic growth that record discrete uplift events along the Ventura fault. We use these data to determine the timing and displacements of Ventura fault paleoearthquakes and discuss these results in light of their implications for assessing the prospects for multisegment ruptures in the western Transverse Ranges, and more generally for seismic hazard in southern California.
The western Transverse Ranges are dominated by several major east-west–striking faults and folds that extend “transverse” to the general northwesterly structural grain of coastal California. These structures have developed in response to the north-south compressional forces that have characterized the region since early Pliocene time (e.g., Luyendyk et al., 1985). The deformation of Pleistocene and younger deposits along a similar structural grain, together with current geodetic data, illustrates that this style of deformation is ongoing within this region.
Located ∼75 km northwest of Los Angeles, the Ventura Basin is a narrow, ∼50-km-long trough bounded on both the north and south by a complex network of east-west–striking reverse and oblique–left-lateral reverse faults (Fig. 1). These structures include the Oak Ridge fault to the south, and the faults responsible for uplift of the Ventura Avenue anticline and mountains of the western Transverse Ranges to the north, including the San Cayetano and Ventura–Pitas Point fault systems (Fig. 2). The Ventura Basin is ∼4 km across at its widest near the City of Ventura and narrows toward its eastern end, where the northern basin-bounding San Cayetano fault overrides the south-dipping Oak Ridge fault (Huftile and Yeats, 1995).
The Ventura Basin, which is >10 km thick at its deepest point, is thought to have formed during the mid-Miocene clockwise rotation of the western Transverse Ranges block to its current east-west trend, and it has experienced left-lateral oblique, north-south shortening since the Pliocene (e.g., Hornafius et al., 1986; Jackson and Molnar, 1990; Luyendyk, 1991). With a maximum thickness of 15–17 km, the Ventura Basin is one of the deepest Pliocene–Pleistocene sedimentary basins in the world (up to 12 km; Yeats, 1977, 1983; Huftile and Yeats, 1995). Rapid sediment deposition and tectonic loading rates produce anomalously low heat-flow values (∼30% lower than are typical for southern California), which account for the anomalously deep background seismicity observed within the area (Bryant and Jones, 1992).
Geodetically determined north-south convergence across the Ventura Basin is as rapid as 7–10 mm/yr (Donnellan et al., 1993a, 1993b; Hager et al., 1999; Marshall et al., 2008, 2013). The Ventura Avenue anticline, located on the north side of the basin (Fig. 1), is rising at a rapid rate of ∼5 mm/yr (Rockwell et al., 1988). This structure, which is underlain by the Ventura fault, is thought to accommodate much of the north-south shortening across the western basin (Rockwell et al., 1988; Stein and Yeats, 1989; Hubbard et al., 2014).
The blind Ventura fault is an ∼12-km-long, east-west–striking, north-dipping reverse fault that is expressed at the surface by a monoclinal south-facing fold scarp extending across the City of Ventura (Figs. 3 and 4; Ogle and Hacker, 1969; CDMG, 1976, 1978; Sarna-Wojcicki et al., 1976; Yeats, 1982; Perry and Bryant, 2002). Historically, several different models have been proposed for the subsurface geometry and downdip extent of the Ventura fault at depth (e.g., Yeats, 1982; Huftile and Yeats, 1995; Sarna-Wojcicki and Yerkes, 1982). These differing interpretations have important implications for understanding the seismogenic potential of the fault. One model suggests that the fault extends to seismogenic depth beneath the Ventura Avenue anticline (Sarna-Wojcicki et al., 1976; Sarna-Wojcicki and Yerkes, 1982), while an alternative interpretation considers the fault to be rooted at ∼300 m depth in the syncline at the southern end of the Ventura Avenue anticline (Yeats, 1982; Huftile and Yeats, 1995). Hubbard et al. (2014) presented several lines of evidence to support the former interpretation. These include: (1) seismic-reflection images with interpreted fault cutoffs at more than 2 km depth; (2) planar alignment of interpreted fault penetrations in wells extending down to 5 km below the surface scarp; and (3) a downdip increase in fault displacement, whereas a shallowly rooted fault would exhibit downward-decreasing fault displacement. The deep, seismogenic ramp interpretation is also consistent with the geometry of the Pitas Point fault, which is imaged by three-dimensional (3-D) seismic data and lies offshore along strike to the west of the Ventura fault trace. Several studies have mapped the Pitas Point fault as the offshore extension of the Ventura fault (Sarna-Wojcicki et al., 1976; Yeats, 1982; Yerkes and Lee, 1987; Yerkes et al., 1987; Dahlen, 1989; Kamerling and Nicholson, 1995; Kamerling and Sorlien, 1999). Hubbard et al. (2014) presented a breakthrough fault-propagation fold model for the development of the Ventura Avenue anticline and underlying Ventura fault that is consistent with the observed acceleration of Quaternary terrace uplift rates observed by Rockwell et al. (1988) on the flanks of the fold. Our study also provides insights that can help to distinguish between these alternative interpretations. Specifically, as described herein, our evidence for large-displacement paleoearthquakes would be difficult to reconcile with a fault rooted at only 300 m depth.
A marked decrease in uplift rates of the anticline at ca. 30 ka (Rockwell et al., 1988) is consistent with breakthrough of the Ventura fault to the near surface at that time (Hubbard et al., 2014). This breakthrough shifted the tip line of the Ventura fault to the south, corresponding to the prominent monoclinal fold scarp that extends east-west through the City of Ventura. Near our study area, the fault remains buried by a thin sedimentary cover (∼200 m thick), and it is therefore technically blind. Further west along the scarp, however, the fault is less deeply buried and has been observed directly in trenches and large-diameter bucket-auger holes examined downhole by geologists (Sarna-Wojcicki et al., 1976; Prentice and Powell, 1991). The Ventura fault trace, marked by the surface scarp extending through the City of Ventura, has been zoned as an active fault under the State of California Alquist-Priolo Special Studies Zone Act (CDMG, 1976, 1978).
Regionally, the Ventura fault acts as a transfer structure to accommodate significant north-south shortening as slip is transferred between the San Cayetano fault to the east and Pitas Point and related faults to the west. Hubbard et al. (2014) suggested that these faults all merge below 7.5 km depth onto a regional horizontal detachment (Sisar décollement) to form a nearly continuous fault surface, despite having separate surface traces (Fig. 2). Previous slip rates on the Ventura fault have been calculated at 0.2–2.4 mm/yr (Petersen and Wesnousky, 1994; Perry and Bryant, 2002) and 2.3 ± 0.7 mm/yr (Marshall et al., 2013) on the basis of 3-D mechanical and kinematic models of active faulting. Hubbard et al. (2014), in contrast, determined a considerably faster fault slip rate of ∼4.4–6.9 mm/yr for the past 30 ± 10 k.y. based on terrace uplift rates and an updated interpretation of the fault kinematics.
Day Road Study Area
The City of Ventura extends east-west along the base of a steep, south-facing mountain front at the western end of the Ventura Basin. This mountain front is coincident with the forelimb of the Ventura Avenue anticline. The city itself is built mainly on low-relief, Upper Pleistocene to Holocene alluvial fans deposited from over half a dozen rivers and creeks draining southward from the Ventura Avenue anticline. Most of these alluvial fans exhibit drainages that have been incised by ∼1 to 6 m into remnant fan surfaces, with the exception of the source drainage for the alluvial fan emanating from the north end of Day Road (Fig. 3; Supplemental Figure S11). We refer to this latter fan as the Arroyo Verde fan after the city park located within the source drainage.
The topographic expression of the Ventura fault through the city is marked by a prominent south-facing scarp (Fig. 4) that extends roughly eastward for ∼12 km from the eastern edge of the active channel of the Ventura River to where the mountain front takes an ∼2 km step to the north at the eastern end of the City of Ventura (Fig. 1). At the eastern end of the scarp, it appears that slip is transferred northward in left-stepping, en echelon fashion onto the southern San Cayetano fault.
Based on high-resolution seismic-reflection profiles collected at three sites along the Ventura fault scarp (McAuliffe, 2014; Supplemental Figure S1 [see footnote 1]), we chose Day Road as the preferred site for our borehole and CPT study due to its location on an active alluvial fan with the potential for continuous deposition. Elsewhere along the fault, south-flowing drainages have incised into the alluvial-fan surfaces, isolating them from active deposition (Fig. 3). The absence of incision into the Arroyo Verde fan surface suggested to us that sediment accumulation is ongoing and that this fan might preserve a more-continuous sedimentary record of near-surface deformation associated with slip on the Ventura fault.
High-Resolution Seismic Reflection Data
We collected a 2.24-km-long, high-resolution seismic-reflection profile along Day Road as part of a broader effort to characterize the deformation of strata above the tip line of the Ventura fault (reflection profiles and details of acquisition and processing can be found in the Supplemental Data File2). The data quality at Day Road was poor, however, because of high traffic noise and signal attenuation within the unsaturated alluvial-fan strata (Supplemental Figure S23). In contrast, our high-resolution profile collected along Brookshire Avenue, 1.4 km east of our Day Road site (Fig. 3), yielded a better-quality image of the structure beneath the scarp because of the linearity of this transect and the low traffic noise on this quiet street (Fig. 5). The Brookshire Avenue profile extends northward along Brookshire Avenue for 1.06 km from its intersection with Woodland Street to the north end of Brookshire Avenue, where it intersects Kearny Street (Supplemental Figure S34).
At Brookshire Avenue, our high-resolution seismic profiles reveal a panel of south-dipping beds (between lines A1 and S1 on Fig. 5) flanked by subhorizontal strata. This profile provides a clear image to a depth of ∼500 m. A well-defined, north-dipping active synclinal axial surface can be traced from near the tip line of the fault at a depth of ∼200 m below sea level to the surface (line S1 in Fig. 5). The south-dipping strata between the synclinal and anticlinal axial surfaces extend to the surface at the prominent south-facing fold scarp, which at this location occurs ∼500 m south of the topographic range front. This scarp defines the surface expression of deformation associated with the most recent folding events on the underlying thrust ramp. Our structural interpretation of this profile differs slightly from that of Hubbard et al. (2014) in that we take advantage of the kinematic constraints provided by our borehole analysis of the youngest folding geometry to modify the dip of the axial surface extending upward from near the tip line of the fault at ∼200 m depth. This interpretation fits the seismic-reflection data equally well, and it provides a more compelling and kinematically consistent interpretation of the relationship between the young folding and the tip line of the fault. Using the structure visible on the Brookshire profile as a guide to the overall structure, a similar structure can be interpreted on the poorer quality Day Road profile (Supplemental Figure S2 [see footnote 3]).
To determine the geometry of recent folding of Arroyo Verde fan strata above the Ventura fault tip line at Day Road, we acquired six continuously cored hollow-stem auger boreholes, 8 cm in diameter and 15 to 21 m deep, along the central section of the Day Road transect across the prominent fold scarp (Figs. 4 and 6). The cores facilitated detailed observation of the subsurface structure and stratigraphy through correlations of the upper 20 m of alluvial strata. In addition to allowing sampling for radiocarbon and luminescence dating, the continuous sampling method allowed us to observe basic sediment characteristics, including grain size, sediment color, and degree of soil development. These sediment characteristics were used to identify and correlate the subsurface stratigraphy between the six boreholes.
We also conducted 13 CPTs, which provided detailed measurements of grain-size variations and other sediment characteristics with depth (see Supplemental Data File [see footnote 2]). The much denser spacing of the CPTs provided valuable data that allowed robust correlations of strata between boreholes. In addition, we excavated two sampling pits, 1.8 m deep, 1 m × 1 m, at the top and base of the surface fold scarp to determine whether any erosion or deposition has taken place subsequent to the most recent event. At each pit, we collected sediment samples for luminescence dating, and we logged the upper 1.8 m of sediment (Supplemental Figures S45 and S56).
The borehole-CPT transect at Day Road extends a total of 368 m from the northernmost borehole at 34.281901°N, 119.227480°E, which is located ∼210 m north of the top of the fold scarp, to the southernmost borehole at 34.278844°N, 119.227021°E, ∼150 m south of the base of the fold scarp (Figs. 3 and 4; Supplemental Figure S67). The fold scarp at the Day Road site lies at the north side of the intersection of Day and Loma Vista Roads, with Loma Vista Road extending approximately along the base of the scarp.
The stratigraphy along Day Road consists of alternating silt and fine- to coarse-grained sand beds interbedded with several gravel layers. The results from our borehole and CPT analyses can be generalized to show that nine distinctive stratigraphic packages can be traced along the entire length of the transect (Figs. 6 and 7; see Supplemental Figure S78 for raw CPT data). The uppermost 4 m of section consist predominantly of fine-grained sands and silts (units A and B). These are underlain by a sequence of silty sands to coarse-grained gravel units (units C, C′, and D), which in turn overlie unit E. Unit E is a distinctive, fine-grained, predominantly silt interval that was deposited on top of a sand to gravel unit (unit F), which overlies fine-grained basal silt unit G.
The sedimentary section is thicker on the downthrown side of the scarp, and there the package comprising unit C includes three distinctive gravel subunits referred to as C′1, C′2, and C′3. These three gravel subunits appear to fan downslope, and C′1 and C′2 may represent onlapping of material onto a paleo-event scarp, as discussed later herein. Our stratigraphic correlations were aided by the presence of distinctive gypsum nodules in the upper 0.5 m of unit C in boreholes DY-2B and DY-3, and by distinctive 0.5 to 3 cm detrital chips of what appear to be fire-baked clay from ancient wildfires in the mountains north of Ventura that were found between 6 and 10 m depth in boreholes DY-2, DY-2B, DY-2C, and DY-3 (red circles in Fig. 6).
Age control for the Day Road transect is provided by 26 infrared stimulated luminescence (IRSL) samples and eight radiocarbon ages from small detrital charcoal fragments collected from the six boreholes and the two sampling pits (Table 1). In addition to reporting the calibrated IRSL and radiocarbon ages, we also report posterior age estimates for each sample based on an OxCal stratigraphic ordering model (Table 1). Our preferred OxCal ordering model (Fig. 8), which is used to constrain both the IRSL and 14C ages, includes only 20 of the 26 total IRSL samples and only one of the eight radiocarbon samples; as discussed in the following, most of the radiocarbon samples were reworked, and the 20 IRSL ages used yielded an internally consistent age model. Additional details concerning sample collection and dating can be found in the Supplemental Data File (see footnote 2).
IRSL Dating of Sediment
The recently developed post-IR IRSL225 dating approach for potassium feldspar (Buylaert et al., 2009, 2012; Thiel et al., 2011), modified for single-grain application (Brown et al., 2014; Rhodes, 2015), was used to date our luminescence samples. Samples were collected using 15 cm brass rings that were placed inside the core barrels at different depths. Once filled with sediment and brought to the surface, these samples were wrapped with foil and placed in light-proof bags to prevent light exposure. The luminescence ages reveal that the borehole transect spans almost the entire Holocene, with the youngest samples collected from the middle of unit A in the sample pits at a depth of ∼1.1 m having ages of 720 ± 90 and 710 ± 150 yr before A.D. 2015, and the oldest sample from a depth of 18.21 m in borehole DY-1 yielding an unmodeled age of 11,720 ± 770 yr before A.D. 2015. Detailed explanation of the IRSL dating protocol can be found in the Supplemental Data File (see footnote 2).
Only eight of the 28 radiocarbon samples that were sent to the Keck Carbon Cycle Accelerator Mass Spectrometry (AMS) facility at the University of California, Irvine, yielded recordable ages (Table 1). The remaining samples were not measured because either the samples were too small, and/or no organic material was left after the standard acid-base-acid pretreatment. Several of the samples that were measured and that did provide ages had extremely large uncertainties due to the small sample size (e.g., DY-C12 and DY-2C:CL-1). Furthermore, many of the radiocarbon samples appear to have been reworked, because they show ages that are much older than other radiocarbon ages and luminescence dates from shallower strata. For example, the 44,193–48,591 calendric year (cal. yr) before A.D. 2015 age of sample DY-C14, the >54,767 cal. yr before A.D. 2015 age of sample DY-2C:CL-01, and the >52,857 cal. yr before A.D. 2015 age of sample DY-C12 are all tens of thousands of years older than the mid- to late-Holocene strata within which they were found. In addition, the 8116–8474 cal. yr before A.D. 2015 age of charcoal sample DY-C1 is >1000 yr older than underlying charcoal sample DY-C7, which was collected from the same core and yielded an age of 6964–7223 cal. yr before A.D. 2015. Finally, the 1400–1480 cal. yr before A.D. 2015 radiocarbon age for sample DR-14:CL-01 from the northern pit is several hundred years older than the internally consistent ca. 1000-yr-old IRSL ages from samples of the underlying silt bed (pit samples DR13-04 and DR14-04). All the radiocarbon results in Table 1 have been corrected for isotopic fractionation according to the conventions of Stuiver and Polach (1977), with δ14C values measured on prepared graphite using the AMS.
The IRSL and 14C age data reveal that the uppermost part of the Arroyo Verde alluvial fan is Holocene in age and that the fan has been actively receiving sediment for at least the past ∼12,000 yr. In coastal environments along the Ventura Basin, soil development occurs at much faster rates relative to soils of equivalent age in California’s inland areas (Rockwell et al., 1985). Favorable soil development conditions are promoted by the presence of sodium ions (a clay deflocculant) caused by sea fog in these coastal areas (Rockwell, 1983). The luminescence and radiocarbon ages, together with the absence of any well-developed soils within the upper 20 m, suggest relatively continuous sediment accumulation at our Day Road study site, without any substantial hiatuses (Figs. 8 and 9). It is worth noting, however, that the age data do not rule out the possibility of relatively brief (a few centuries) hiatuses, and we cannot rule out potential minor stripping of paleosols in this alluvial environment.
The four IRSL ages from the two sample pits provide evidence for lateral continuity of the shallowest strata. In each pit, we dated two samples, one from ∼1.1 m depth in a silt bed, and a second from ∼1.6–1.8 m depth in a different silt layer beneath a very weakly developed topsoil. Despite the 275 m down-fan distance between the sample pits, the resulting pairs of ages are remarkably consistent, with the two shallow samples yielding ages of 710 ± 150 and 720 ± 90 yr before A.D. 2015 (DR1-02 and DR13-02, respectively), and the lower samples in each pit yielding ages of 930 ± 80 yr before A.D. 2015 (DR14-04) and 960 ± 110 yr before A.D. 2015 (DR13-04). This internal consistency, and the similar stratigraphy of the two pits, strongly suggests that these are the same strata (unit A), encountered at the same depths, both above and below the scarp. The deepest identifiable sedimentary unit that can be correlated along the entire transect is unit I, which is dated at ca. 9 ka.
Most Recent Event
Several observations from the two shallowest units in our cross section indicate folding of sediments during the most recent event (MRE) on the Ventura fault at Day Road. Specifically, the stratigraphy of the uppermost 4 m (units A and B in Fig. 6) tracks the ground surface across the fold scarp without a significant change in thickness, indicating that: (1) these strata were deposited during a period of structural quiescence at the gently south-dipping slope of the Arroyo Verde alluvial fan and were subsequently folded; (2) the fold scarp has not yet been buried by young alluvial strata following the most recent event(s); and (3) the ∼6.0–6.5 m height of the fold scarp (measured vertically from the northward and southward projections of the average far-field ground surface slope) records the amount of uplift during the most recent large-magnitude earthquake (or earthquakes) on the Ventura fault. Moreover, the similar ages of the pairs of IRSL ages from the uppermost 1.1–1.8 m collected from the sample pits above and below the scarp suggest minimal post–most recent event erosion of the hanging wall and negligible post–most recent event deposition on the footwall since the scarp formed. The absence of thickening in these latest Holocene strata supports our interpretation that deposition of these beds predates the scarp, and thus that the height of the current topographic scarp records the approximate amount of uplift during the most recent event(s) on the Ventura fault.
The similar IRSL ages from 1.1 m depth in the pits coupled with apparent lateral continuity of the thickness of these young strata across the entire fold scarp and the absence of any evidence for onlap of the base of the scarp indicate that the most recent folding event(s) occurred after deposition of the strata from 1.1 m depth, the youngest of which has an OxCal stratigraphic ordering model age of 805 ± 90 yr before A.D. 2015. The minimum-possible age of this uplift is not directly constrained by our age data, but it must have occurred before the historic era, which began in Ventura with the establishment of the Spanish mission there in early 1782. An event of this size would certainly have been noted and commented upon by the mission authorities, as the scarp that grew by ∼6 m in the most recent event extends <25 m north of the main mission buildings. Moreover, the 6 m scarp height indicates that the earthquake (or earthquakes) responsible for this uplift must have been of large magnitude (Mw 7.5–8.0). The only possible historical event that could have been this large in this region was the 21 December 1812 earthquake, although the historical data suggest that that event was likely of smaller magnitude (Toppozada et al., 1981). Thus, the fold scarp at Day Road developed after the age of the 1.1-m-deep strata in the upper part of unit A at 805 ± 75 yr before A.D. 2015, and before the historic era began 230+ yr ago. An additional consideration is that the deposition of ∼1 m of pre-most recent event sediment above the shallowest ∼800-yr-old IRSL samples suggests that the most recent event may actually have occurred relatively late in the allowable time interval. We cannot determine whether this large uplift occurred in a single earthquake, or more than one event, because of the absence of growth strata across the scarp. If the scarp records uplift in more than one earthquake, then all of those events, which themselves must have been large-magnitude events, each generating several meters of slip, must have occurred during this relatively brief time interval during the latter part of the allowable time range between ∼240 and 800 yr ago.
Evidence for an older event (or events) at Day Road comes from a second episode of uplift and folding that is recorded by stratigraphic growth of unit C, which thickens by ∼4.5 m southward across the fold scarp (Fig. 6). We interpret this sedimentary growth as evidence for deposition against a now-buried paleo–fold scarp that developed during the previous folding event(s). The event horizon for this period of fold growth is located in the lower part of unit C at the base of the growth interval. Specifically, the event horizon is above unit C′3, the deepest of the three distinctive gravel layers within the unit C-C′ growth interval. Unit C′3 is folded parallel to underlying strata at the scarp, indicating that all strata below C′3 have experienced the same folding history. The unit C′3 gravel bed has been truncated on the upthrown side by erosion of the hanging wall (Fig. 6). Thickening of unit C by ∼4.5 m on the downthrown side of the scarp indicates that at least this much uplift occurred during fold growth. This is a minimum estimate because it does not take into account any erosion of unit C that may have occurred on the upthrown side of the fold scarp. If we assume that minimal deposition has occurred on the downthrown side of the scarp following the most recent event(s), in keeping with the similar IRSL ages from the two sample pits, then the top of unit D can be used to document the combined total uplift in the two folding events. Measured at the top of unit D, this total uplift is ∼11.2 m. Allocating ∼6 m of this total to uplift during the youngest folding event(s) that produced the surface scarp provides a best estimate of ∼5.2 m of uplift in the previous folding event(s), marked by the unit C-C′ growth section. IRSL samples Dy-OSL-2/1 and Dy-OSL-2B/3 in the OxCal stratigraphic ordering model constrain the age of this event(s) to between 4590 ± 230 and 4165 ± 250 yr before A.D. 2015, with a stratigraphic ordering modeled boundary age range for this uplift event of 4365 ± 300 yr before A.D. 2015 (Fig. 8).
Below the unit C-C′ growth section, the underlying ∼7-m-thick section consisting of units D, E, F, and G does not change thickness across the fold, indicating that those lower- to mid-Holocene sediments were deposited during a period of structural quiescence.
Possible Additional Event?
Stratigraphic correlations show a distinct change in bed dip between gravel units C′2 and C′3 within the growth stratigraphic section, with C′2 dipping more gently than the underlying C′3 (7.62 m to 9.75 m depth in borehole DY-3). It is possible that this change in bed dip may record progressive limb rotation (Novoa et al., 2000) during an additional folding event, with two event horizons located just above unit C′3. This interpretation requires two closely spaced events, with an earlier event occurring just after the deposition of C′3 (i.e., event described in previous section), and an additional event between the deposition of C′3 and C′2, causing the beds beneath C′2 to have distinctly steeper dips than those above. If this fanning of bedding dips records a separate folding event during a different earthquake, then that second event must have also occurred during the same relatively brief interval as the event described previously, at 4365 ± 300 yr before A.D. 2015, based on the stratigraphic ordering modeled boundary age range for this event constrained by over- and underlying sample ages of 4165 ± 250 yr before A.D. 2015 (sample Dy-OSL-2B/3) and 4590 ± 230 yr before A.D. 2015 (sample Dy-OSL-2/1), respectively.
Alternatively, this change in bed dip may simply record fanning of material off the paleo–fold scarp as the fan built back up to its original unfaulted geometry. In this case, the geometry of the unit C growth section could all be due to a single earthquake on the Ventura fault (i.e., previous event), with an event horizon located at the top of unit C′3. We reiterate that aside from this slight fanning of bedding dips in the unit C growth strata, all of our stratigraphic observations can be explained by two folding events, one that produced the surface scarp and an older event that generated the buried fold scarp onlapped by the unit C growth section (Figs. 10 and 11).
Uplift Measurements and Fault Displacement Estimates
To more accurately determine the uplift record of the Day Road section, we incrementally restored (i.e., “unfolded”) the strata using the inclined shear-restoration method of Novoa et al. (2000) (Fig. 10). In this procedure, all points along each depositional contact are restored parallel to the dip of the active axial surface, such that the youngest deformed units are restored to the undeformed, depositional geometry, which in this case is presumed to be the regional slope of the alluvial-fan surface. Then, older strata are incrementally unfolded to constrain deformation in earlier events. The total minimum scarp height for each paleofolding event is determined from these restorations.
In the case of the most recent folding event(s), restoration of the folded unit A and B stratigraphic section to the far-field alluvial-fan slope results in a planar depositional geometry for both of these units, indicating that they were deposited prior to the most recent folding event(s) in the absence of a topographic scarp. Very little, if any, sediment accumulation has occurred since the deposition of this shallowest unit, as shown by both the similar ages of samples collected at ∼1.1 m depth from our sample pits upslope and downslope from the scarp, as well as by the overall geometry of the strata relative to the surface scarp (Fig. 6; Table 1). The shear restoration shown in Figure 10 yields 6 m of surface uplift during the most recent folding event(s).
Measuring uplift in the previous event(s) is slightly more complicated due to the apparent truncation of unit C′3 on the upthrown side of the scarp. Based on sedimentary growth occurring from the top of unit C′3 to near the top of unit C, we measure the minimum uplift in the previous event(s) as ∼4.5 m. As noted already, this minimum estimate does not take into account any erosion of the upthrown strata. As described in previous sections, subtracting the ∼6 m of uplift from the ∼11.2 m of total uplift that occurred in events 1 and 2 combined yields a best estimate of uplift in the penultimate uplift event(s) of ∼5.2 m, indicating that ∼0.7 m of erosion must have occurred on the upthrown side of the scarp following the previous event(s). This erosion explains the northward truncation of gravel unit C′1 at the scarp (Figs. 10 and 11).
To convert the scarp heights to reverse displacements on the underlying Ventura fault, we divide the scarp height by the sine of the 50° ± 5° dip of the fault from Hubbard et al. (2014). We also make the conservative assumption that coseismic slip is constant on the fault ramp, rather than increasing with depth, as is observed for the total slip that has accumulated over geologic time scales (Hubbard et al., 2014). This will render any displacement measurements we make minima. Assuming that the most recent uplift event between ∼235 and 800 yr ago is a single earthquake, 6 m of uplift yields a fault displacement of 7.3–8.5 m (the range in displacement is due to the uncertainty in fault dip angle). For the previous event(s), the minimum 4.5 m of uplift yields 5.5–6.4 m of thrust slip, and our best estimate of 5.2 m of uplift yields 6.4–7.4 m of thrust slip (Table 2).
Paleoearthquake Magnitude Estimates
We can estimate paleoearthquake magnitudes for the two most recent events on the eastern part of the Ventura fault at Day Road by using published empirical equations based on global regressions that relate earthquake magnitude, fault area, and average displacement (Wells and Coppersmith, 1994; Biasi and Weldon, 2006). Although the calculated displacements at the Day Road site are only single measurements along the 60-km-long fault Ventura–Pitas Point fault system, which likely exhibits some degree of lateral variability in displacements, even larger uplift values for the four most recent earthquakes at Pitas Point (Rockwell, 2011) suggest that the slip values of >5.5–8.5 m derived from our uplift measurements are suitable for use as the average slip during a systemwide rupture of the Ventura–Pitas Point fault. Our Day Road values are within the “most likely” range suggested by Hubbard et al. (2014) of 6.2–9.9 m based on uplifted coastal terraces.
Results from two different empirical equations relating earthquake magnitude and average displacement are shown in Table 2. We calculated a paleoearthquake magnitude for both Day Road events using our preferred interpretation of the evidence that these two uplift and folding events each records a single earthquake, rather than multiple events. Assuming that the most recent rupture slipped with an average displacement of 7.3–8.5 m during the most recent event, the Wells and Coppersmith (1994) regression yields a paleoearthquake magnitude range of Mw 7.64–7.69. Applying this same regression to the earlier uplift event yields a minimum paleoearthquake magnitude of Mw 7.54–7.59 and a best estimate magnitude of Mw 7.59–7.64. In contrast, the empirical equations of Biasi and Weldon (2006) yield much larger estimated paleoearthquake magnitudes of Mw 7.91–7.98 for the most recent event and Mw 7.84–7.91 for the previous event. These paleo-event magnitudes are similar to those estimated by Hubbard et al. (2014) using slip values based on the uplifted marine terraces measured at Pitas Point, ∼15 km west of the Day Road site.
It is, of course, possible that smaller-magnitude events could have occurred on the Ventura fault that did not leave a detectable uplift record in the Day Road section, but these would have had to have been few in number and of very small displacement, as the shear “unfolding” restorations shown in Figure 10 yield sedimentologically reasonable restored geometries for the folded Day Road strata with only two uplift events.
An additional consideration in determining paleomagnitude estimates for past ruptures on the Ventura fault is the structural position of the Day Road transect along the thrust system. The eastern end of the Ventura fault, ∼3 km east of the Day Road site, forms a “soft,” en echelon segment boundary with the southern San Cayetano fault to the east. Hubbard et al. (2014) provided new data indicating that slip on the Ventura thrust dies out eastward ∼6 km east of our study site, corresponding with the eastward disappearance of the fault scarp. This eastward-diminishing slip corresponds with a major en echelon left step in the thrust system and the eastward-increasing prominence of the south-facing fold scarp along the northern edge of the Ventura Basin that marks the surface manifestation of the southern San Cayetano fault. Studies of similar “soft” segment boundaries between the Puente Hills and Compton blind thrust faults beneath the Los Angeles Basin by Shaw et al. (2002) and Lehle (2007) illustrate lateral decreases in fault slip as these faults approach segment boundaries and the transfer of slip onto en echelon fault segments. The overlapping, en echelon “soft” segment boundary in the Ventura–southern San Cayetano fault thrust system just to the east of our Day Road site is structurally similar to those observed along the Puente Hills and Compton blind thrust faults, and we might expect total displacement along the Ventura fault to be decreasing eastward at our site as it begins to transfer eastward to the southern San Cayetano fault. Thus, the reverse displacements that we calculate from paleo-uplift events at Day Road likely underestimate the average displacement of any multisegment ruptures involving both the Ventura and southern San Cayetano faults.
Implications for Seismic Hazard in Southern California
From a seismic hazard assessment standpoint, one of the most important issues concerning the faults of the western Transverse Ranges is the size of future earthquakes that they might produce. As described herein, the large vertical uplift events that occurred during the past two earthquakes observed at Day Road indicate very large thrust displacements on the order of >5.5–8.5 m, despite the fact that this study site is only a few kilometers from the eastern end of the Ventura fault. The seismogenic potential of the Ventura fault has been debated for some time (Sarna-Wojcicki et al., 1976; Sarna-Wojcicki and Yerkes, 1982; Yeats, 1982; Huftile and Yeats, 1995), with the persistent disagreement on this matter stemming from uncertainty of the fault geometry at depth. The new 3-D model of Hubbard et al. (2014) not only affirms the validity of the deep, seismogenic model for the Ventura fault, but it also plausibly hypothesizes connectivity of the Ventura–Pitas Point, San Cayetano, and Red Mountain faults, which might allow for large-magnitude, multisegment ruptures in the western Transverse Ranges. Specifically, the Ventura fault forms the middle section of a >200-km-long, east-west belt of large, discrete, yet interconnected reverse- and oblique-slip faults that extends across the western and central Transverse Ranges.
Although each individual fault in the Transverse Ranges fault system represents a major seismic source in its own right, a systemwide, multifault rupture involving the Ventura fault together with other major faults of the western Transverse Ranges could cause catastrophic damage to the densely urbanized areas of the Ventura and Los Angeles Basins. One of the largest of these potential multifault earthquakes involves rupture of the rapidly slipping eastern San Cayetano fault westward via the blind southern San Cayetano fault, onto the blind Ventura thrust fault together with correlative faults to the west (e.g., Pitas Point fault; Fig. 1). Such a 75- to 100-km-long multisegment rupture could potentially encompass a fault-plane area of as much as several thousand square kilometers—similar to the rupture area of the great 1857 Mw 7.8 Fort Tejon and 1906 Mw 7.9 San Francisco earthquakes on the San Andreas fault, and comparable to other large-magnitude reverse fault earthquakes such as the 2008 Mw 7.9 Wenchuan event. To the east, potential subsurface connectivity of the San Cayetano fault with the Santa Susana and Sierra Madre faults may provide a mechanism to extend the ruptures further eastward, but this subsurface structure remains poorly understood. Recent modeling of geodetic data suggests that longer faults or a series of connected fault surfaces do a better job of fitting current global positioning system (GPS) rates (Marshall et al., 2013). Such studies seem to support our results of large-magnitude events with likely long rupture lengths.
Although there is abundant microseismicity in the Ventura area (Bryant and Jones, 1992; Hauksson, 2011), there are few historical and paleoseismic data available to test the validity of the various rupture scenarios. We summarize the few available data, including those from this study, in Figure 12. Although the limited slip-per-event data available from western Transverse Ranges faults suggest that large-magnitude earthquakes have occurred in the past (e.g., Dolan and Rockwell, 2001; Rockwell, 2011; Hubbard et al., 2014; this study), no large Mw (M > 7) earthquakes have occurred on any of the faults surrounding the Ventura Basin for at least 200 yr, suggesting the possibility that recurrence intervals for these faults are relatively long and that they may therefore rupture in larger, multisegment events. The most recent potentially large-magnitude earthquake in the Ventura region occurred on 21 December 1812. Toppozada et al. (1981) originally suggested that this Mw ∼7 earthquake was generated by rupture of an offshore fault beneath the Santa Barbara Channel. Toppozada et al. (2002), however, subsequently speculated that this earthquake may have occurred on the western Big Bend section of the San Andreas fault, effectively extending the 8 December 1812 San Andreas fault Mojave segment rupture to the northwest. There is no direct evidence, however, that the second 1812 earthquake occurred on the San Andreas fault, and felt intensity reports are consistent with a western Transverse Ranges source. Dolan and Rockwell (2001) documented a large-displacement (>∼5 m) thrust event on the eastern San Cayetano fault sometime after A.D. 1660. If this event was not the 21 December 1812 earthquake, then it occurred between A.D. 1660 and the beginning of the historic period, which likely began in the 1780s for an earthquake of this size (Dolan and Rockwell, 2001). It is worth noting that this eastern San Cayetano fault earthquake occurred during the latter part of the allowable ca. A.D. 1200–1780 age range of the most recent earthquake(s) that formed the Day Road fold scarp, and these two sites may record the same earthquake (Fig. 12). The large displacements in the most recent earthquake(s) at Day Road and at the Dolan and Rockwell (2001) trench site along the eastern San Cayetano fault suggest that this is a plausible scenario. Moreover, we reiterate that the Day Road site is only ∼3 km from the eastern end of the well-defined Ventura fault fold scarp, and that slip in this area is gradually transferred eastward from the Ventura fault onto the southern San Cayetano fault across a “soft” segment boundary. Thus, the large minimum displacement that occurred during the most recent earthquake at Day Road (7.3–8.5 m) close to the eastern end of the Ventura fault strongly suggests that the most recent earthquake rupture continued eastward onto the southern San Cayetano fault.
However, could this rupture have also continued westward along the Ventura–Pitas Point fault system? At Pitas Point, which lies 15 km to the west-northwest of the Day Road site along the structural crest of the Ventura Avenue anticline, Rockwell (2011) and Hubbard et al. (2014) pointed out four paleoshorefaces along the Ventura coastline that they argued were uplifted 5–10 m in each of the four most recent events. Uplift of the youngest two shorefaces at Pitas Point occurred during earthquakes at ∼800–1000 yr ago for the most recent earthquake(s) and ∼1900 yr ago for the previous event (Fig. 12). The ages of the two earlier events recorded at Pitas Point are being reconsidered in light of new data (T. Rockwell, 2015, personal commun.) and are not discussed in this paper. Individual uplift events of as large as 9 m observed at Pitas Point would require large (Mw 7.7–8.1) earthquakes (Hubbard et al., 2014), similar in size to those inferred to have occurred at Day Road and likely involving rupture of multiple faults within the central-western Transverse Ranges (e.g., Ventura, Pitas Point, southern San Cayetano, eastern San Cayetano, and western Santa Barbara Channel faults; Hubbard et al., 2014), and/or the deep décollements that underlie many of these faults.
Interestingly, the two most recent events observed at Pitas Point (Rockwell, 2011; Hubbard et al., 2014) do not seem to correlate with either of the uplift events recorded at Day Road. Specifically, the earliest part of the ca. A.D. 1200–1780 age range of most recent earthquake(s) that generated the Day Road scarp barely overlaps with the last part of the ∼800- to 1000-yr-old age range of the most recent earthquake at Pitas Point (Rockwell, 2011; Hubbard et al., 2014), suggesting the possibility that the most recent earthquakes at both sites may the same event. As noted previously herein, however, the Day Road most recent earthquake must have occurred after deposition of the ∼1 m of strata that overlie the stratum from which the ∼800-yr-old IRSL samples were collected at Day Road. While the duration of this period of deposition is not directly constrained, it likely spanned centuries, based on the sediment accumulation rates at the site from deeper intervals of similar sedimentary character. Thus, while the most recent earthquakes at Day Road and Pitas Point could be the same event, it appears more likely that the Day Road most recent earthquake occurred somewhat later than the ca. 800–1000 yr B.P. most recent earthquake at Pitas Point. Similarly, the post–A.D. 1660 eastern San Cayetano fault event is not observed at Pitas Point, demonstrating that, if the most recent earthquake at Day Road is the same event as the most recent earthquake on the eastern San Cayetano fault, this eastern San Cayetano–plus–Ventura fault rupture did not extend as far west as Pitas Point. If the rupture did not extend a significant distance westward from the Day Road site, it must have either: (1) involved rupture of additional faults located to the east of the San Cayetano fault (e.g., Santa Susana and Sierra Madre faults; it is noteworthy that the western part of the Sierra Madre fault experienced a young, barely prehistoric surface rupture sometime within the past ∼500 yr [Fumal et al., 1995; T. Fumal, 1996, personal commun.], suggesting that this scenario is not necessarily implausible, despite the structural complexities of the fault system between the eastern San Cayetano and Sierra Madre faults); and/or (2) ruptured much deeper along the downdip extents of the Ventura, southern San Cayetano, and eastern San Cayetano faults, possibly including rupture of the relatively gently dipping décollements modeled by Hubbard et al. (2014) along the deep parts of these faults. It is also noteworthy that if the Ventura fault most recent event observed at Day Road is not the same earthquake as the most recent event at Pitas Point (as seems likely in light of the age data), then these events must have occurred within a few hundred years of one another. Such a scenario could reflect relatively brief, laterally propagating sequences of very large-magnitude events along the Transverse Ranges fault system, perhaps in response to progressive lateral changes in ΔCoulomb failure function stresses related to earlier events in the sequence, as has been suggested for laterally propagating sequences of thrust fault earthquakes elsewhere (e.g., Dolan and Bowman, 2004).
The penultimate event observed at Pitas Point (ca. 1.9 ka; Rockwell, 2011) does not appear to have produced any detectable paleoearthquake signal on the Ventura fault at the Day Road site, as this date falls within the middle of the ∼800- to 3000-yr-old stratigraphic section, which does not change thickness across the fold, thus indicating that it was deposited during a period of structural quiescence. As with the most recent earthquakes at both sites, this observation suggests that at least sometimes the Ventura–Pitas Point system does not rupture in its entirety. One possibility may be that the fault may rupture partial segments in smaller-magnitude events at times between the multisegment ruptures. Arguing against this possibility, however, is the similarity in uplift height (and presumably magnitude [i.e., 7.7–8.1]; Hubbard et al., 2014) during each of the past four uplift events at Pitas Point, which suggests that each event there records a similar-sized rupture, albeit not necessarily generated by rupture of the same parts of the same faults in different events. Thus, it seems unusual for the most recent event and second event back at Pitas Point to not have ruptured as far east as Day Road. One possible alternative scenario may involve the rupture of both the Pitas Point fault and the southern San Cayetano fault, with slip transferred eastward along the deep, blind Ventura thrust ramp and southern San Cayetano fault, bypassing the shallower part of the easternmost Ventura fault encompassing the Day Road study site.
As noted earlier, the unit C-C′ previous event(s) growth interval at Day Road is bracketed by two IRSL ages of 4165 ± 250 and 4590 ± 230 yr before A.D. 2015, yielding a boundary age from the OxCal stratigraphic ordering model for the penultimate uplift event(s) at Day Road of 4365 ± 300 yr (Fig. 8). The age of this event is older than currently available paleoseismic data from the eastern San Cayetano fault and from the Pitas Point site, and any space-time correlation (or lack thereof) of this event with paleoearthquakes on faults to the east and west awaits further age constraints from those faults.
Collectively, the data from the Day Road site along the Ventura fault (this study), the Pitas Point site along the Pitas Point fault (Rockwell, 2011; Hubbard et al., 2014), and the eastern San Cayetano fault trench site of Dolan and Rockwell (2001) point to complex space-time patterns of large-magnitude earthquake occurrence along the major reverse faults of the western Transverse Ranges. Whereas the most recent events at the eastern San Cayetano and Day Road sites could be the same post–A.D. 1660 event, the age range of the most recent event at the eastern San Cayetano fault site does not overlap with the age range of the most recent event at Pitas Point, precluding the possibility that this was a systemwide event encompassing simultaneous rupture of the Ventura–Pitas Point, southern San Cayetano, and eastern San Cayetano faults. Moreover, the ca. 1.9 ka second event back at Pitas Point is not observed at the Day Road site, only 15 km to the east, along what is thought to be a continuous fault system at depth (Hubbard et al., 2014). Additional paleoearthquake age constraints at more sites along this fault system will be needed to resolve more fully the space-time patterns of large-magnitude earthquakes generated by these faults.
Although the space-time pattern of earthquakes in the western Transverse Ranges remains poorly resolved, the available paleoearthquake data from the Day Road (this study), Pitas Point (Rockwell, 2011), and eastern San Cayetano fault (Dolan and Rockwell, 2001) sites do provide significant insights into the behavior of this system. Most basically, these observations indicate that these faults appear to rupture relatively frequently in large-magnitude (Mw 7.5–8.0) earthquakes. Moreover, these data facilitate comparison with recent models of expected seismicity in the region (e.g., UCERF3 [Uniform California Earthquake Rupture Forecast]; Field et al., 2013). For example, the combination of the range of magnitude estimates and the recurrence of two such events (or brief clusters of events) at Day Road with the four-event record provided by the Pitas Point uplift data (Rockwell, 2011; Hubbard et al., 2014) suggests that such events recur every 1000 yr to several thousand years. For comparison, the recent UCERF3 model suggests that the Ventura–Pitas Point fault system will participate in Mw ≥7.5 earthquakes with a mean recurrence of 2500 yr (Field et al., 2013). At this magnitude threshold, the UCERF3 model appears to reflect the recurrence characteristics of this fault system reasonably well. However, as described earlier herein, Mw 7.5 lies at the extreme low end of the magnitude range suggested by the large displacements inferred from the Day Road uplift data, and it is smaller than the lowest Mw estimate for earthquakes on the Pitas Point part of the system (Mw 7.7–8.1; Hubbard et al., 2014), assuming our preferred interpretation that each fold scarp records uplift during an individual earthquake. If, instead, each of the uplift events observed at Day Road and Pitas Point records clusters or couplets of multiple events, then recurrence intervals will be shortened commensurately. The UCERF3 model does more poorly when larger magnitudes are considered. For example, the model predicts that the Ventura fault will participate in Mw ≥7.75 earthquakes (the middle of our inferred magnitude range for paleoearthquakes at the Day Road site) with a mean recurrence time of ∼10,000 yr, and Mw ≥8.0 events (the high end of our magnitudes estimates) with an average recurrence rate of once every million years. This seeming disconnect between the predicted and observed recurrence rates of large-magnitude earthquakes needs to be addressed in future such models. Part of the discrepancy likely lies in the use of a relatively slow best-estimate slip rate of 1.6 mm/yr in the UCERF3 model (with upper and lower bounds of 0.5 and 10 mm/yr, respectively), relative to the preferred slip rate of ∼4.4–6.9 mm/yr in Hubbard et al. (2014). However, this discrepancy may also reflect higher-than-expected levels of potential rupture connectivity between the major reverse faults of the central and western Transverse Ranges.
Results from newly acquired high-resolution seismic-reflection data, borehole cores, cone penetrometer tests, and luminescence and radiocarbon geochronology at the Day Road study site in eastern Ventura reveal evidence for two uplift and folding events that we interpret as evidence for the occurrence of large-magnitude earthquakes on the underlying Ventura fault. The most recent folding event(s), which occurred after deposition of pre-event strata dated at ca. A.D. 1200 and before the beginning of the historic record (A.D. 1782), generated the 6-m-high fold scarp observed at our study site. This prominent surface scarp is underlain by a 4-m-thick, post–3 ka sequence of alluvial-fan strata that does not change thickness across the fold scarp, recording structural quiescence during this period and indicating that: (1) these sediments were folded in the most recent folding event(s), and (2) the surface scarp records uplift during that event(s). The previous event (or events) at this site is recorded by a southward-thickening interval of sedimentary growth strata that onlaps a now-buried, >4.5-m-tall fold scarp that formed between 4065 and 4665 yr ago. This growth interval is underlain by an ∼5-m-thick section spanning 5–9 ka that does not change thickness across the fold, indicating that this was a period of structural quiescence. Although it is also possible that each of these scarp-forming events records uplift in multiple earthquakes, we think this possibility is unlikely, especially in the case of the most recent folding event, as this would require the occurrence of multiple large-magnitude (M >7) earthquakes during the brief allowable interval between ∼800 yr ago and the beginning of the historic period ∼235 yr ago. Moreover, evidence for similarly discrete 5–10 m uplift events from farther west along the Ventura–Pitas Point fault system in the form of uplifted marine terrace inner edges (Rockwell, 2011; Hubbard et al., 2014) also argues for these as single-earthquake folding events. If, alternatively, each of these uplift events records multiple earthquakes, then these earthquakes must have occurred in brief temporal clusters.
The very large reverse displacements required to generate the 5–6 m uplifts observed at the Day Road site typically occur in large-magnitude earthquakes (e.g., Wells and Coppersmith, 1994). This suggests that the Day Road events were of large magnitude, likely in excess of Mw 7.5, and potentially approaching Mw 8.0, if each folding event represents a single earthquake. The potential occurrence of such large-magnitude earthquakes on the Ventura fault would likely involve simultaneous rupture of multiple faults along strike to the east and west and/or ruptures extending downward onto the deep, gently dipping décollements that are thought to underlie several major faults in the region (see Hubbard et al., 2014). Comparison of our paleoearthquake ages and displacements with similar data generated by Rockwell (2011) and Hubbard et al. (2014) from uplifted paleoshorelines at Pitas Point, 15 km to the west along the structural crest of the Ventura Avenue anticline, indicates that: (1) although the earliest part of the post–ca. A.D. 1200–1780 most recent event at Day Road slightly overlaps with the age of the most recent event at Pitas Point (ca. A.D. 1000–1200), the most recent event at Day Road probably records a separate, younger event that occurred relatively late during the allowable ca. A.D. 1200–1780 age range. In contrast, the allowable most recent event age range at Day Road does overlap in age with the most recent event on the eastern San Cayetano fault (A.D. 1660–1812; Dolan and Rockwell, 2001), suggesting that these could be the same earthquake; and (2) the ca. 1.9 ka penultimate event at Pitas Point did not extend through the Day Road site on the eastern Ventura fault, indicating that these sites do not always rupture together, despite being large-displacement earthquakes on the same fault system only 15 km apart.
These observations point to complex recurrence patterns for large-magnitude earthquakes involving the Ventura–Pitas Point fault and adjacent faults during mid- to late Holocene time. The potential recurrence of such large-magnitude events has critically important implications for seismic hazard assessment in southern California. Specifically, the occurrence of large thrust fault earthquakes adjacent to the deep (>10 km) Ventura Basin would cause significant amplification of seismic waves, leading to damaging ground motions over much of the region, perhaps extending into the Los Angeles metropolitan area, the San Fernando Basin, and the San Gabriel Valley. Moreover, large-displacement ruptures of the Ventura fault along its offshore western continuation, the Pitas Point fault, could potentially generate significant tsunamis near the coast, with limited potential warning times. It is worth noting, however, that the relatively shallow water depths at the fault-seafloor interface will reduce the overall volume of the water mass involved in any such tsunamis.
The recurrence intervals for the large-displacement Ventura fault earthquakes documented at our Day Road site are significantly longer than those for the recurrence of “Big Ones” on the San Andreas fault system, with interevent times measurable in thousands of years, rather than hundreds. Nevertheless, the potential magnitude of multifault and/or very deep western Transverse Ranges earthquakes involving the Ventura fault may approach or even exceed those of San Andreas earthquakes, indicating that it is crucial that the prospects for the recurrence of large-magnitude, multifault earthquakes on the Ventura and mechanically interconnected faults in the western and central Transverse Ranges be properly considered in future regional seismic hazard assessments.
We greatly appreciate the help of Amir Allam, Jessica Grenader, Chris Milliner, Jason Williams, and Robert Zinke, who helped us with the borehole and cone penetrometer test excavations, Nathan Brown and Chris McGuire, who helped with the luminescence sampling at Day Road, and Chris Cothrun, Ben Haravitch, Zurriya Hasnan, Alison Koop, Samuel Rosenbaum, and Rachel Zucker for their assistance in the field acquiring seismic data. We thank Kate Scharer, Sally McGill, and an anonymous reviewer for their constructive reviews, and Ned Field for helpful discussions regarding the UCERF3 seismic hazard model. Finally, we would like to especially thank Chandra Shaker, Roger Markel, and the City of Ventura for facilitating our field work and for help with logistics and permitting. Thomas Pratt was supported for this work by the U.S. Geological Survey (USGS) National Earthquake Hazards Reduction Program (NEHRP). Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. government. This research was supported by the Southern California Earthquake Center (SCEC). SCEC is funded by the National Science Foundation cooperative agreement EAR-1033462 and USGS cooperative agreement G12AC20038. This is SCEC contribution 2076.