History of earthquakes along the creeping section of the San Andreas fault, California, USA

Creeping faults are difficult to assess for seismic hazard because they may participate in rupture even though they likely cannot nucleate large earthquakes. The creeping central section of the San Andreas fault in California (USA) has not participated in a historical large earthquake; however, earthquake ruptures nucleating in the locked northern and southern sections may propagate through the creeping section. We used biomarker thermal maturity and K/Ar dating on samples from the San Andreas Fault Observatory at Depth to look for evidence of earthquakes. Biomarkers show evidence of many earthquakes with displacements >1.5 m in and near a 3.5-m-wide patch of the fault. We show that K/Ar ages decrease with thermal maturity, and partial resetting occurs during coseismic heating. Therefore, measured ages provide a maximum constraint on earthquake age, and the youngest earthquakes here are younger than 3 Ma. Our results demonstrate that creeping faults may host large earthquakes over longer time scales.


biomarker. A) Methylphenanthrene structural isomes and the methylphenanthrene index (MPI-4). B) C29 Steranes α and β isomers and the sterane index (SI). ααα -Ster
and CDZ (bottom). CPI decreases with increasing temperature. CPI is low in SAFOD samples, mostly hovering around 1, indicating they have reached maximum maturity and therefore show no thermal maturity anomaly.  SDZ (top) and CDZ (bottom). ADI decreases with increasing temperature. Samples are at or approaching the maximum value (~1.5) for ADI\ and therefore show no thermal maturity anomaly.    Schematic demonstrating the three different pathways that can lead to a measured K/Ar age. 1) Temperature is high enough that complete resetting occurs, this results in a zero age immediately after heating. The measured K/Ar age in this case reflects the time since the earthquake. Scenarios 2) and 3) reflect partial resetting resulting in a non-zero age immediately after the earthquake. In these cases, the measured age is older than the earthquake. Fig S12. average friction during sliding plotted against displacement for a range of normal stresses with hydrostatic pore pressure. At larger normal stress and displacement, the thermal breakdown distance is small relative to displacement and the average friction is low.

Average friction for SAFOD normal stress conditions (49 MPa) is shown in red.
Table S1 -Parameters used to model SAFOD earthquake displacements and apparent ages resulting from thermal resetting. The range of friction values used is consistent with steadystate friction values measured from Di Toro et al. (2011) and with calculations of average friction for sliding at SAFOD (see supplementary methods and Fig. S14). Slip layer thicknesses represent the distribution of localized layers throughout the BFR.

Biomarker thermal maturity analysis
Samples were either subsampled if localized structures were present or processed whole. In the preliminary round of SAFOD sampling we separated and measured the biomarker maturity of the center and outside of a sample but found no difference in maturity between those aliquots. Samples were rinsed with dicholoromethane to remove any contamination and disaggregated using a mortar and pestle. Samples were extracted with a Dionex Accelerated Solvent Extractor (ASE-350) with 9:1 DCM:methanol at 1500 psi and a temperature of 100 °C for 3x5 minute static cycles to isolate the total lipid extract (TLE). A recovery standard consisting of 5α-androstane, 1-1' binapthyl, and stearyl stearate, was added to each TLE and the TLE was evaporated with nitrogen and transferred to 4 mL vials. The TLE was brought up in 0.5 ml of hexane and separated into aliphatic, aromatic/ketone, and polar fractions using 0.5 g silica gel (stored at 75 °C) in 5-inch Pasteur pipettes. The sample was loaded onto the columns in hexane, and the aliphatic fraction (F1) eluted with 4 ml of hexane, the aromatic/ketone fraction (F2) with 4 ml of dichloromethane, and the polar (F3) with 4 ml of methanol. The aliphatic and aromatic/ketone fractions were brought up in 0.25 mL of hexane and transferred to 2 mL high-recovery vials for analysis on an Agilent 7890A gas chromatograph with a 5975C mass selective detector (GC-MSD) equipped with a multi-mode inlet (MMI, deactivated single-taper liner with wool packing) and DB-5ms column (30 m length, 250 µm i.d., 0.25 µm phase thickness) at 1.0 ml/min helium flow. Samples were diluted in 100 to 500 µl hexane, depending upon their concentration, with an injection volume of 1 µl. The aromatic fraction containing phenanthrenes and methylphenanthrenes was analyzed in hybrid selected ion monitoring (SIM)/full scan mode (SIM/scan) with external calibration as described in Sheppard et al. (2015). The aliphatic fraction containing n-alkanes, steranes, and hopanes was analyzed in full scan mode. The sample in hexane was injected splitless into the MMI and the MMI temperature held at 60 °C for 0.1 minutes and then ramped to 320 °C at 15 °C/s and held for the remaining acquisition time. The oven temperature was held at 60 °C for 1.5 minutes, ramped to 150 °C at 15 °C/min and then to 320 °C at 4 °C/min where it was held for 10 minutes. The MSD ion source was held at 300 °C with an electron energy of 70 eV and a quadrupole temperature of 150 °C. The MSD was operated in full scan mode, scanning from 50 -550 dalton with a cycle time of ~3 scans/s. Peaks were integrated with the Agilent Chemstation software, using extracted ion peak areas for n-alkanes (m/z 57), C29 steranes (m/z 217), C31 hopanes (m/z 205) and the recovery standard (5a-androstane, m/z 245). Concurrent analyses of a standard mixture of C8 to C40 n-alkanes plus 5a-androstane was used to calibrate the relative response ratio of each n-alkanes to the recovery standard daily. Individual ion peak areas were used to calculate sterane and hopane ratios without any further treatment.

K/Ar measurements
Argon measurements were made on samples after biomarker measurement. Bulk and < 2 µm grain size fractions were measured to assess whether measurements demonstrated any grain size dependence. The < 2 µm fraction was isolated using gravitational settling techniques. Argon measurements were made using a VG 5400 mass spectrometer with a CO2 laser extraction system, and potassium concentrations measured using inductively coupled plasma optical emission spectroscopy (ICP-OES). Replicates were measured for all samples. Ages were similar in age between each grain size fraction and thin section observations demonstrated no difference in grain size between unsettled and settled sediment fractions (this discrepancy may be due to clumping in the bulk fraction). As a result, we group the grain size fractions together and report the mean for each sample set.
To measure potassium, an open beaker total digest was performed using HNO3/HF/HClO4 in order to achieve a complete digestion of the sample material. Due to the potential to form insoluble potassium perchlorate, HClO4 was used sparingly, and the samples were evaporated to dryness several times in the presence of nitric. Samples were taken up in ~3% nitric acid and brought to a final dilution of 3,000 -10,000x. Replicate samples and a USGS certified reference material (SCo-1 Cody Shale) was prepared with each sample batch to evaluate reproducibility and precision. Samples were measured by Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES).

Laser heating experiments
Aliquots of a single background SAFOD sample were weighed out and wrapped in tantalum foil. The sample packets were folded over a type K thermocouple and placed in a diffusion cell for analysis. A schematic of this set up can be seen in Fig S14. Samples were heated to temperatures of 500 -820 °C for 10s within diffusion cells (Farley et al. 1999) using a diode laser. Temperature was controlled by manipulating the power of the laser while recording the temperature output from the thermocouple. The amount of argon released during heating was measured, and the sample was then heated again to 900 °C for 3 minutes to completely degas it and the total argon measured. Suspending the samples on thin thermocouple wires in individual diffusion cells allows us to heat and cool the samples quickly enough to simulate earthquake conditions. We use the linear relationship between fraction degassed and temperature from these experiments to model the apparent age resulting from each possible SAFOD heating event for each sample.

Thermal modeling
To constrain the temperature rise associated with a given high MPI4, heat generation and diffusion equations (Fulton & Harris, 2012;Lachenbruch 1986) for a fault are coupled with the reaction kinetics for MPI4 (Sheppard et al. 2015). The adiabatic temperature rise that occurs depends on properties of the fault zone are as follows 2) where τ is shear stress, ρ is density, c is the heat capacity, a is the fault half width, v is slip velocity, is thermal diffusivity, x is distance from the slipping layer, and t is time. Temperature profiles are used to simulate biomarker reaction for different displacements, frictions, and slip layer thicknesses. MPI4 resulting from these scenarios are calculated using experimentally determined reaction kinetics (Sheppard et al. 2015) and the Easy%R_0 method (Sweeney & Burnham, 1990). This allows identification of MPI4 profiles that best fit core measurements and the extraction of possible coseismic temperatures. Temperature rise and fault properties that fit our measurements are then, along with the kinetics of argon degassing used to model argon concentration and calculate the apparent ages expected for these conditions.

Average friction calculation
Under the normal stress conditions at SAFOD, friction during sliding evolves from a peak value to steady state over a thermal weakening distance. Because the peak friction has a larger effect on the average friction for small earthquakes compared to large, we calculate the range of average friction for displacements used in our thermal model as follows.
Calculation of average friction and Fig. S14 were done using the relationship for thermal breakdown distance ( ℎ ) and normal stress: where and are experimental constants and is normal stress (Di Toro et al. 2011). From this, friction was calculated using the following equation for stress ( ) established by fitting a shear stress curve to experimental data (Seyler et al. 2020): where is the steady state shear stress, is the peak shear stress, and is the slip accumulated after ℎ . Values used in this calculation for SAFOD are shown in the Table S1.

Earthquake magnitude scaling
We use the following scaling relationship developed by Ellsworth (2003) from a database of strike-slip earthquakes to estimate magnitude of these earthquakes identified at SAFOD.
= 4.2 + 10 ( ) where A is rupture area. We assume for earthquakes that do not rupture the entire seismogenic zone that A is equal to L 2 , where is rupture length and the ratio of displacement to rupture length is 0.0001 (Scholz 2002).