Stable, rapid rate of slip since inception of the San Jacinto fault, California



[1] In California, where the San Jacinto fault (SJF) and San Andreas fault (SAF) accommodate the majority of the dextral shear deformation between the Pacific and North American plates, initiation of the SJF led to an apparent decline in the slip rate of the SAF. Previous studies suggest that since then, slip rate has covaried between these faults (possibly due to changes in fault strength, variation in topographic loading along a fault, or the development of new faults) and that presently the SJF is the dominant plate boundary structure. However, we dated displaced sedimentary deposits and landforms over three distinct time intervals since ~700 ka, and our results imply a constant slip rate of 12.1+3.4/−2.6 mm/yr. This rate is similar to the fault's lifetime rate and from rates derived from geodesy, suggesting that since the SJF initiated, its slip rate has remained relatively stable and does not exceed that of the SAF.

1 Introduction

[2] Quaternary fault slip rates provide important information on the recurrence of past earthquakes, as displacements along active faults mostly accrue during large surface rupturing events. These rates are generally considered steady when averaged over many earthquake cycles [e.g., Reid, 1910; Schwartz and Coppersmith, 1984; Mouslopoulou et al., 2009], however, within regions of the San Andreas fault system, paleoseismic evidence for earthquake clustering suggests rate variability over 1 kyr timescales [e.g., Rockwell et al., 2000; Peltzer et al., 2001; Dolan et al., 2007]. This, in turn, implies time-dependent changes in fault rheology, geometry, segmentation, and the transfer of stress between faults, with consequences for the stress distribution within the crust and fault localization below the seismogenic zone [Friedrich et al., 2003; Fay and Humphreys, 2006]. Discrepancies between slip rates on geologic (Myr), late Quaternary (kyr), and geodetic (decadal) time scales indicate that transient fault loading may play a role in this clustering behavior [e.g., Kenner and Simons, 2004; Oskin et al., 2008]. Because far-field plate motions are steady [DeMets and Dixon, 1999], increased loading rate in one area within a plate boundary fault system must be balanced by diminished rates elsewhere. A dramatic example of this may be the hypothesized codependence of slip between the San Andreas fault (SAF) and San Jacinto fault (SJF), which share the majority of Pacific-North America (PA-NA) plate motion [Sharp, 1981; Bennett et al., 2004] (Figure 1A). However, the uncertainty of published slip rates [Rockwell et al., 1990; Janecke et al., 2011] and along strike-slip rate gradients due to transfer of deformation to other faults [Blisniuk et al., 2010; Spinler et al., 2010] casts doubt on this hypothesis.

Figure 1.

Maps of study area and individual sites. (a) Overview map of the study area, showing the location of the San Andreas, San Jacinto, and Elsinore fault zones. References: (1) van der Woerd et al. [2006], Behr et al. [2010], (2) Blisniuk et al. [2010], and (3) Janecke et al. [2011]. (b) Geologic map of Thomas Mountain and the present-day location of exposed metamorphic and plutonic alluvium along the fault. Black arrows indicate the offset of the buried metamorphic alluvial unit. (c) Lidar image showing the present-day configuration of channels 1 and 2 offset from their upstream sources channels A and B. The black frame indicates the location of Figure 1e. The red dots denote the average GPS location for groups of cobbles or amalgamated samples collected. Inset ages are determined in this study. (d) Reconstruction showing 605 m offset of channels 1 and 2 from channels A and B. The pie charts show the results of clast counts. (e) Location of stream offset and trenches T7, T17, and T14 by Merifield et al. [1991]. The black arrows point to the trace of the fault. See text for description of offset interpreted by Merifield et al. [1991].

[3] It is well established that the SAF and SJF rapidly evolved over the early Quaternary [Morton and Matti, 1993]. The SJF with a total bedrock offset of 22–29 km [Sharp, 1981; Janecke et al., 2011] initiated at ~1.5 Ma [Morton and Matti, 1993] shunting slip around the evolving San Gorgonio Pass stepover on the SAF. Within its considerable uncertainties, the 20.1+6/−10 mm/yr [Janecke et al., 2011] lifetime slip rate of the SJF equals or exceeds the ~7–22 mm/yr late Quaternary slip rate of the longer-lived SAF to the east [van der Woerd et al., 2006; Behr et al., 2010; McGill et al., 2012].

2 Methods

[4] Building on earlier studies by Sharp [1981], Rockwell et al. [1990], and Merifield et al. [1991], we revised the slip-rate estimate for the SJF at Anza using 0.5 m resolution B4 Lidar [Bevis et al., 2005] topography data and cosmogenic nuclide geochronology. In the field, lithologic characteristics of alluvial deposits were quantified to reinforce links between offset deposits described by Sharp [1981] and their respective source areas (Figure 1b). The ages of these deposits are updated based on 10Be surface exposure dating of 53 samples: 34 quartz-bearing cobbles collected from alluvial fan and abandoned-channel surfaces, 13 scattered samples from the shallow subsurface (10–85 cm depth), and a six sample depth profile (Figure 1c, Text S1, and Tables S1 and S2). We carried out one 26Al/10Be analysis of a burial sample (sample SJF64) to provide a minimum age for an older, underlying deposit matched across the SJF by Sharp [1981] (Text S1 and Table S1). We note that postdepositional production of 26Al and 10Be before a sample becomes sufficiently shielded from surface exposure will always introduce a bias toward ages that are younger than the true burial event. Accordingly, a burial age provides a robust minimum age estimate for underlying deposits. Prior burial during sediment transport, which can yield apparently older ages [Hu et al., 2011], is assumed to be minimal due to the relatively small catchment areas of the sediment sources. All samples were processed following the procedures described in Blisniuk et al. [2012] or Mériaux et al. [2012].

[5] The final age estimate for each offset landform is reported at the 95% confidence level. Fault slip rates are calculated following a boxcar approach of age and displacement [Zechar and Frankel, 2009] (Table S3) with uncertainties in age and displacement reported at 2σ. The final slip rate is calculated by taking the weighted mean of both the positive and negative uncertainty of all slip rates.

3 Slip-Rate Estimates

[6] To refine the SJF slip rate and test for variations over time, we determined rates from well-defined offsets of Quaternary landforms whose ages are constrained by 10Be surface exposure or 26Al/10Be burial dating over three time intervals: since the mid-Quaternary (~700 ka), late Pleistocene (~45 ka), and mid-Holocene (5 ka). To isolate temporal from spatial variability, we analyzed closely spaced sites near Anza, California, where the SJF is localized onto a single fault strand (Figure 1).

[7] In the study area, the SJF (Clark fault strand) progressively offsets alluvial deposits as old as mid-Quaternary along the southwestern range front of Thomas Mountain [Sharp, 1981], where four types of crystalline rocks are exposed adjacent to the SJF (Figure 1b). Importantly, the relative abundance of metamorphic and tonalitic rocks characterizes distinctive, nearly monolithologic alluvial fans that gradually accumulated in the piedmont southeast of the fault. Of particular interest is a distinctive, poorly sorted sandy cobble conglomerate containing >90% metamorphic clasts that overlies the Bishop Ash (Figure S2). This deposit is displaced along the fault from its source area by 5.7 to 8.6 km [Sharp, 1981] (Figure 1b), and the combination of this offset with the maximum age provided by the underlying Bishop Ash, 758 ± 1.8 ka [Sarna-Wojcicki et al., 2000], yields a minimum slip rate of 8–11 mm/yr.

[8] To improve this rate, we provide a minimum age for this offset by applying 26Al/10Be burial dating to a ~50 m thick capping alluvial unit, dominated by >80% tonalitic cobble clasts in a coarse sand matrix, that disconformably overlies the distinctive offset unit. The dated sample was collected 10–20 cm above the contact with the underlying metamorphic-clast dominated alluvium, along a recent road-cut (Figure S2). The burial date yielded a minimum deposition age of 698 ± 94 ka (Table S1), which also provides a minimum age for the underlying alluvium. Combining this minimum age with the maximum age provided by the Bishop Ash and the offset from Sharp [1981] yields a slip rate of 10.5+2.9/−2.5 mm/yr since 600–760 ka (Table S3).

[9] To determine late Quaternary slip rates, we applied quantitative age estimates to offset alluvial fans and reevaluated the offset interpretation of these fans and their incised channels (Figure 1c). A larger offset was reconstructed using two beheaded channels incised into an alluvial fan complex. A smaller offset was reconstructed using the most recently incised and abandoned stream channel.

[10] Based on ages inferred from soil development, alluvial deposits at Anza were interpreted to become progressively younger southeastward as they were displaced along the fault [Rockwell et al., 1990]. From this, slip rates of 13+10/−6 (since circa 48 ka) and 12+9/−5 mm/yr (since circa 17 ka) were established [Rockwell et al., 1990]. Here we reevaluate this interpretation with high-resolution topography data to better constrain the offsets, and new age determinations that show the uppermost fan units were not deposited sequentially [Rockwell et al., 1990] but rather during a single aggradation event.

[11] Figure 1d illustrates our revised interpretation of the offset as 580 to 630 m. This reconstruction realigns the two northwest walls of channels 1 and A and the southeast walls of channel 1 to the active part of channel 1 (Figure S1). The orientation of the channels orthogonal to the fault, coupled with the similarity in channel width on both sides of the fault, supports the assumption that the channels were originally incised directly across the fault trace after deposition of the uppermost part of the alluvial fan complex. This reconstruction also realigns the walls of channel B with channel 2 (Figures 1d and S1). To test whether channel A is the dominant source of the alluvial fan, we compared clast counts from the recent alluvium in channel A and the alluvial surface at location b. The results support the correlation of this alluvial deposit, dominated by 85% hornblende-biotite tonalitic clasts with 13% foliated biotite adamellitic clasts, to deposits in channel A, composed of 71% tonalitic and 17% adamellitic clasts (Figure 1d).

[12] Using 10Be surface exposure dating, we determined the abandonment age of fan surfaces adjacent to channel 1 and channel 2 to provide a maximum date for the channel incisions. The surface dates, uncorrected for inheritance or erosion, yield weighted mean ages of 46.6 ± 27 ka (n = 3), 47.0 ± 11 ka (n = 8), and 44.5 ± 5 ka (n = 16) for surfaces at locations a, b, and c, respectively (Figure 1c; Table S1). Overall, these ages are all identical within errors and average 45.2 ± 4.4 ka (±2σ, Figure 2a). We also collected subsurface cobbles from alluvial deposits a and b, respectively. The depth-corrected dates for subsurface samples range from 17 to 71 ka and yield an average of 43.5 ± 17.2 ka (n = 13), which is also consistent with the surface ages of the alluvial complex.

Figure 2.

Summary of ages. (a) Ages from surfaces a, b, and c. (b) Ages from the bottom of channel 2. Black bars are dates of individual surface cobbles. Light gray bars are ages outside the 95% confidence interval of the remaining ages. (c) Summary of age data results from simulating 10Be concentration versus depth at surface d, showing the 2σ solution spaces (grey shaded region) and lowest chi-square solution (black dashed line).

[13] For fan surface d, we obtained a 10Be depth profile model age of 45.0+10/−11 ka (Figure 2b; Table S2). The base of the depth profile indicates a minimal inheritance age, for gravel-sized clasts, of 0.13+2.4/−0.01 kyr. The resulting date is identical, within error, to the means of dates from surfaces a, b, and c, 45.2 ± 4.4 ka, suggesting that this fan complex formed during a single emplacement event, contrary to the previous interpretation by Rockwell et al. [1990]. To calculate the slip rate, we combine the offset of channel A (the dominant fan source) with the depth profile model age, which we consider the most reliable age estimate as it accounts for both inheritance and surface lowering. This yields a slip rate estimate of 13.4+3.8/−2.5 mm/yr since ~45 ka.

[14] A Holocene offset of 51–61 m is determined from the most recently incised and abandoned stream channel located within the wide, sediment-filled floor of channel 2 (Figure 1e). Based on a 4.3+0.5/−0.3 ka age of charcoal from within deposits at the base of the abandoned-channel fill and realigning the closest abandoned-channel risers north and south of the fault, a slip rate of 10–15 mm/yr was estimated by Merifield et al. [1991]. Their age constraint is corroborated by the ages of 7 10Be samples collected in channel 2 which range from 2.1 to 8.4 ka, with four dates ranging from 5.0 to 6.0 ka (Figure 2c; Table S1). Combining the Gaussian probability distribution functions of these two age constraints into a single distribution yields an age of 4.5 ± 0.6 ka (±2σ) for the channel abandonment, yielding a slip rate of 12.4+2.5/−2 mm/yr for the Holocene.

4 Discussion/Conclusion

[15] Combining all the new offset and age constraints yields a best-fit constant slip rate of 12.1+3.4/−2.6 mm/yr for the SJF, with interval rates between 10 and 15 mm/yr that overlap within error (Figure 3). The consistency of these rates over 3 orders of magnitude of time does not support the hypothesis that the SJF rate has significantly varied over the latter half of the Quaternary [Sharp, 1981; Bennett et al., 2004]. This SJF slip rate agrees well with estimates from paleoseismology and GPS studies, but is inconsistent with faster interferometric synthetic aperture radar (InSAR)-derived rates and some geologic estimates. Paleoseismic studies at Hog Lake, located ~5 km north of our site, suggest a slip rate of 14+4/−4 mm/yr, based on an average earthquake return period of ~230 years over the past 4000 years, and coseismic slip per event of 3–4 m [e.g., Salisbury et al., 2012]. Elastic and viscoelastic models of GPS data also suggest rates of 9–15 mm/yr [Meade and Hager, 2005; Fay and Humphreys, 2005; Spinler et al. 2010; Chuang and Johnson, 2011]. InSAR-derived rates of ~20 mm/yr [Fialko, 2006; Lundgren et al., 2009] are substantially higher, but may overestimate the actual slip rate due to the assumption that all deformation west of the SJF is accommodated by the fault. A geologic slip rate estimate of ~20 mm/yr for the northern SJF [Kendrick et al., 2002] may be problematic because it is not directly determined from dated, laterally offset features, but based on a model of fault-related uplift.

Figure 3.

Summary of slip rates for the SJF at Anza and the SAF. (a) Plot showing mid-Quaternary to Holocene slip rates for the SJF determined in this study. The consistency of slip rates across three time intervals implies time-constant rates, within error. (b) Plot summarizing published slip rates for SJF and SAF. The gray shaded region shows the average slip rate determined in this study. References for plotted GPS are derived in text.

[16] The new slip rate estimates reported here are within the uncertainty of the lifetime slip rate of the SJF [Janecke et al., 2011] (Figure 3b), but are very close to the lower bound. A modest decrease of the SJF slip rate in the mid-Quaternary is permitted, but not required. The 10–15 mm/yr slip rate we determine for the SJF overlaps with the lower end of the 14–17 mm/yr preferred slip rate for the SAF [van der Woerd et al., 2006; Behr et al., 2010], suggesting the SJF is subordinate to the SAF or deformation across the PA-NA boundary at this latitude is equally shared by the two structures. These results indicate that the SJF is likely not the dominant structure across the plate boundary at this latitude.

[17] The codependent slip rate hypothesis, if correct, would imply a mechanism by which fault zone strength covaried with fault slip [Bennett et al., 2004] or earthquake stress changes on the SJF reduced the Coulomb stress on the SAF and vice-versa [Li and Liu, 2007]. However, our results show that the SJF slip rate is instead more or less constant (to within 5 mm/yr) since the mid-Quaternary, and perhaps since early inception of faulting. Thus, the covariant slip-rate hypothesis of Bennett et al., [2004] is not supported by the available data—in good agreement with the results of a GPS study by Spinler et al. [2010], which showed that the apparent slip-rate variation in time on the SAF was instead the result of spatial slip rate gradients. Stable, rapid rate of slip of the SJF implies that the fault became relatively mature and strain-weakened early in its evolution [e.g., Wesnousky, 1988; Ben-Zion and Sammis, 2009]. Accordingly, the geometrical configuration of the PA-NA plate boundary fault system may be the primary control on the partitioning of slip rate between the SAF and SJF [cf. Bourne et al., 1998]. Either (1) the effects of growing topography of the Transverse Ranges [Fay and Humphreys, 2006] and evolution of the Eastern California shear zone [Liu et al., 2010], are unimportant relative to the strength and the transfer of stress between the SJF and SAF, or (2) a balance is achieved between gradual weakening of the SJF over time (enhancing its favorability for slip) and inhibition of its slip rate via growth of topography and evolution of other parts of the southern California fault system.

[18] In conclusion, our new geologic slip rate estimates for the SJF, combined with the revised slip rate at the Biskra Palms site on the SAF [Behr et al., 2010], do not support the model of strong temporal slip rate variation and codependency between these faults [Bennett et al., 2004]. Though initiation of the SJF and a reorganization of plate boundary deformation in the early Quaternary likely reduced the rate of slip on the SAF [Morton and Matti, 1993], the division of slip rate between the SAF and SJF appears to have been constant since at least the mid-Quaternary, and likely since initiation of the SJF in the early Quaternary. This suggests that evolving fault-network geometry may exert the primary control on the distribution of fault slip rates across a plate boundary.


[19] This research was supported by the Southern California Earthquake Center, which is funded by the National Science Foundation Cooperative Agreement EAR-0529922 and by the U.S. Geological Survey Cooperative Agreement 07HQAG0008. K. Blisniuk thanks G. Balco for his help in calculating the burial age. We thank R. Bennett and R. Arrowsmith for insightful and constructive reviews of the manuscript, and S. McGill and E. Kirby for reviewing an earlier version.

[20] The Editor thanks Richard Bennett and Ramon Arrowsmith for their assistance in evaluating this paper.