Gordon G. Seitz

Evidence for Large Earthquakes on the San Andreas Fault at the Wrightwood, California, Paleoseismic Site: a.d. 500 to Present

We present structural and stratigraphic evidence from a paleoseismic site near Wrightwood, California, for 14 large earthquakes that occurred on the southern San Andreas fault during the past 1500 years. In a network of 38 trenches and creek-bank exposures, we have exposed a composite section of interbedded debris flow deposits and thin peat layers more than 24 m thick; fluvial deposits occur along the northern margin of the site. The site is a 150-m-wide zone of deformation bounded on the surface by a main fault zone along the northwest margin and a secondary fault zone to the southwest. Evidence for most of the 14 earthquakes occurs along structures within both zones. We identify paleoearthquake horizons using in-filled fissures, scarps, multiple rupture terminations, and widespread folding and tilting of beds. Ages of stratigraphic units and earthquakes are constrained by historic data and 72 14C ages, mostly from samples of peat and some from plant fibers, wood, pine cones, and charcoal. Comparison of the long, well-resolved paleoseimic record at Wrightwood with records at other sites along the fault indicates that rupture lengths of past earthquakes were at least 100 km long. Paleoseismic records at sites in the Coachella Valley suggest that each of the past five large earthquakes recorded there ruptured the fault at least as far northwest as Wrightwood. Comparisons with event chronologies at Pallett Creek and sites to the northwest suggests that approximately the same part of the fault that ruptured in 1857 may also have failed in the early to mid-sixteenth century and several other times during the past 1200 years. Records at Pallett Creek and Pitman Canyon suggest that, in addition to the 14 earthquakes we document, one and possibly two other large earthquakes ruptured the part of the fault including Wrightwood since about a.d. 500. These observations and elapsed times that are significantly longer than mean recurrence intervals at Wrightwood and sites to the southeast suggest that at least the southermost 200 km of the San Andreas fault is near failure.

Paleoseismic Event Dating and the Conditional Probability of Large Earthquakes on the Southern San Andreas Fault, California

We introduce a quantitative approach to paleoearthquake dating and apply it to paleoseismic data from the Wrightwood and Pallett Creek sites on the southern San Andreas fault. We illustrate how stratigraphic ordering, sedimentolog- ical, and historical data can be used quantitatively in the process of estimating earth- quake ages. Calibrated radiocarbon age distributions are used directly from layer dating through recurrence intervals and recurrence probability estimation. The method does not eliminate subjective judgements in event dating, but it does provide a means of systematically and objectively approaching the dating process. Date dis- tributions for the most recent 14 events at Wrightwood are based on sample and contextual evidence in Fumal et al. (2002) and site context and slip history in Weldon et al. (2002). Pallett Creek event and dating descriptions are from published sources. For the five most recent events at Wrightwood, our results are consistent with pre- viously published estimates, with generally comparable or narrower uncertainties. For Pallett Creek, our earthquake date estimates generally overlap with previous results but typically have broader uncertainties. Some event date estimates are very sensitive to details of data interpretation. The historical earthquake in 1857 ruptured the ground at both sites but is not constrained by radiocarbon data. Radiocarbon ages, peat accumulation rates, and historical constraints at Pallett Creek for event X yield a date estimate in the earliest 1800s and preclude a date in the late 1600s. This event is almost certainly the historical 1812 earthquake, as previously concluded by Sieh et al. (1989). This earthquake also produced ground deformation at Wrightwood. All events at Pallett Creek, except for event T, about A.D. 1360, and possibly event I, about A.D. 960, have corresponding events at Wrightwood with some overlap in age ranges. Event T falls during a period of low sedimentation at Wrightwood when conditions were not favorable for recording earthquake evidence. Previously pro- posed correlations of Pallett Creek X with Wrightwood W3 in the 1690s and Pallett Creek event V with W5 around 1480 (Fumal et al., 1993) appear unlikely after our dating reevaluation. Apparent internal inconsistencies among event, layer, and dating relationships around events R and V identify them as candidates for further inves- tigation at the site. Conditional probabilities of earthquake recurrence were estimated using Poisson, lognormal, and empirical models. The presence of 12 or 13 events at Wrightwood during the same interval that 10 events are reported at Pallett Creek is reflected in mean recurrence intervals of 105 and 135 years, respectively. Average Poisson model 30-year conditional probabilities are about 20% at Pallett Creek and 25% at Wrightwood. The lognormal model conditional probabilities are somewhat higher, about 25% for Pallett Creek and 34% for Wrightwood. Lognormal variance rln estimates of 0.76 and 0.70, respectively, imply only weak time predictability. Conditional probabilities of 29% and 46%, respectively, were estimated for an em- pirical distribution derived from the data alone. Conditional probability uncertainties are dominated by the brevity of the event series; dating uncertainty contributes only secondarily. Wrightwood and Pallett Creek event chronologies both suggest varia- tions in recurrence interval with time, hinting that some form of recurrence rate modulation may be at work, but formal testing shows that neither series is more ordered than might be produced by a Poisson process.

A Record of Large Earthquakes on the Southern Hayward Fault for the Past 500 Years

The Hayward fault, a major branch of the right-lateral San Andreas fault system, traverses the densely populated eastern San Francisco Bay region, California. We conducted a paleoseismic investigation to better understand the Hayward fault9s past earthquake behavior. The site is near the south end of Tyson9s Lagoon, a sag pond formed in a right step of the fault in Fremont. Because the Hayward fault creeps at the surface, we identified paleoseismic events using features that we judge to be unique to ground ruptures or the result of strong ground motion, such as the presence of fault-scarp colluvial deposits and liquefaction. We correlate the most recent event evidence (E1) to the historical 1868 M 6.9 earthquake that caused liquefaction in the pond and recognize three additional paleoruptures since A.D. 1470 ± 110 yr. Event ages were estimated by chronological modeling, which incorporated historical and stratigraphic information and radiocarbon and pollen data. Modeled, mean age and 95-percentile ranges of the three earlier events are A.D. 1730 (1650-1790) yr (E2), A.D. 1630 (1530-1740) yr (E3), and A.D. 1470 (1360-1580) (E4). The ages of these paleoearthquakes yield a mean recurrence of 130 ± 40 yr. Although the mean recurrence is well determined for the period A.D. 1470-1868, individual intervals are less well determined: E1-E2, 140 +80/-70 yr; E2-E3, 100 +90/-100 yr; and E3-E4, 150 +130/-110 yr.

Serial ruptures of the San Andreas fault, Carrizo Plain, California, revealed by three-dimensional excavations

[1] It is poorly known if fault slip repeats regularly through many earthquake cycles. Well-documented measurements of successive slips rarely span more than three earthquake cycles. In this paper, we present evidence of six sequential offsets across the San Andreas fault at a site in the Carrizo Plain, using stream channels as piercing lines. We opened a latticework of trenches across the offset channels on both sides of the fault to expose their subsurface stratigraphy. We can correlate the channels across the fault on the basis of their elevations, shapes, stratigraphy, and ages. The three-dimensional excavations allow us to locate accurately the offset channel pairs and to determine the amounts of motion for each pair. We find that the dextral slips associated with the six events in the last millennium are, from oldest to youngest, !5.4 ± 0.6, 8.0 ± 0.5, 1.4 ± 0.5, 5.2 ± 0.6, 7.6 ± 0.4 and 7.9 ± 0.1 m. In this series, three and possibly four of the six offset values are between 7 and 8 m. The common occurrence of 7–8 m offsets suggests remarkably regular, but not strictly uniform, slip behavior. Age constraints for these events at our site, combined with previous paleoseismic investigations within a few kilometers, allow a construction of offset history and a preliminary evaluation of slip-and time-predictable models. The average slip rate over the span of the past five events (between A.D. 1210 and A.D. 1857.) has been 34 mm/yr, not resolvably different from the previously determined late Holocene slip rate and the modern geodetic strain accumulation rate. We find that the slip-predictable model is a better fit than the time-predictable model. In general, earthquake slip is positively correlated with the time interval preceding the event. Smaller offsets coincide with shorter prior intervals and larger offset with longer prior intervals.

Timing of Large Earthquakes since A.D. 800 on the Mission Creek Strand of the San Andreas Fault Zone at Thousand Palms Oasis, near Palm Springs, California

Paleoseismic investigations across the Mission Creek strand of the San Andreas fault at Thousand Palms Oasis indicate that four and probably five surface-rupturing earthquakes occurred during the past 1200 years. Calendar age estimates for these earthquakes are based on a chronological model that incorporates radiocarbon dates from 18 in situ burn layers and stratigraphic ordering constraints. These five earthquakes occurred in about A.D. 825 (770–890) (mean, 95% range), A.D. 982 (840–1150), A.D. 1231 (1170–1290), A.D. 1502 (1450–1555), and after a date in the range of A.D. 1520–1680. The most recent surface-rupturing earthquake at Thousand Palms is likely the same as the A.D. 1676 ± 35 event at Indio reported by Sieh and Williams (1990). Each of the past five earthquakes recorded on the San Andreas fault in the Coachella Valley strongly overlaps in time with an event at the Wrightwood paleoseismic site, about 120 km northwest of Thousand Palms Oasis. Correlation of events between these two sites suggests that at least the southernmost 200 km of the San Andreas fault zone may have ruptured in each earthquake. The average repeat time for surface-rupturing earthquakes on the San Andreas fault in the Coachella Valley is 215 ± 25 years, whereas the elapsed time since the most recent event is 326 ± 35 years. This suggests the southernmost San Andreas fault zone likely is very near failure. The Thousand Palms Oasis site is underlain by a series of six channels cut and filled since about A.D. 800 that cross the fault at high angles. A channel margin about 900 years old is offset right laterally 2.0 ± 0.5 m, indicating a slip rate of 4 ± 2 mm/yr. This slip rate is low relative to geodetic and other geologic slip rate estimates (26 ± 2 mm/yr and about 23–35 mm/yr, respectively) on the southernmost San Andreas fault zone, possibly because (1) the site is located in a small step-over in the fault trace and so the rate is not be representative of the Mission Creek fault, (2) slip is partitioned northward from the San Andreas fault and into the eastern California shear zone, and/or (3) slip is partitioned onto the Banning strand of the San Andreas fault zone.

A high-resolution seismic CHIRP investigation of active normal faulting across Lake Tahoe Basin, California-Nevada

We measured extension rates across Lake Tahoe Basin for the last 60 k.y. based on measured displacement of offset marker surfaces across three active faults beneath Lake Tahoe. Seismic chirp imaging with submeter accuracy, together with detailed multibeam and light detection and ranging (LIDAR)-derived bathymetry, was used to measure fault offset, thickness of earthquake-derived colluvial wedges, depth of wave-cut paleoterraces, and other geomorphic features. An analysis of these features provides refined estimates of extension rates and new information on Holocene faulting, and places Tahoe Basin deformation into the larger context of Walker Lane and Basin and Range tectonics. Measured offset marker surfaces include submerged wave-cut paleoterraces of Tioga age (19.2 ± 1.8 ka), McKinney Bay slide deposits (ca. 60 ka), and a winnowed boulder surface of Tahoe age (ca. 62 ka). Estimated vertical offset rates across submerged geomorphic surfaces are 0.43-0.81 mm/yr for the West Tahoe fault, 0.35-0.60 mm/yr for the Stateline-North Tahoe fault, and 0.12-0.30 mm/yr for the Incline Village fault. These offset rates indicate a combined east-west extension rate across Lake Tahoe Basin, assuming 60° fault dips, of 0.52-0.99 mm/yr. This estimate, when combined with the Genoa fault-slip rate, yields an extension rate consistent with the magnitude of the extension deficit across Carson Valley and Lake Tahoe Basin derived from global positioning system (GPS) velocities. The Stateline-North Tahoe, Incline Village, and West Tahoe faults all show evidence for individual Holocene earthquake events as recorded by either colluvial wedge deposits or offset fan-delta stratigraphy.

New Constraints on Deformation, Slip Rate, and Timing of the Most Recent Earthquake on the West Tahoe-Dollar Point Fault, Lake Tahoe Basin, California

High-resolution seismic compressed high intensity Radar pulse (CHIRP) data and piston cores acquired in Fallen Leaf Lake (FLL) and Lake Tahoe provide new paleoseismic constraints on the West Tahoe-Dollar Point fault (WTDPF), the western- most normal fault in the Lake Tahoe Basin, California. Paleoearthquake records along three sections of the WTDPF are investigated to determine the magnitude and recency of coseismic slip. CHIRP profiles image vertically offset and folded strata along the southern and central sections that record deformation associated with the most recent event (MRE) on the WTDPF. Three faults are imaged beneath FLL, and the maximum vertical offset observed across the primary trace of the WTDPF is ∼3:7 m. Coregis- tered piston cores in FLL recovered sediment and organic material above and below the MRE horizon. Radiocarbon dating of organic material constrained the age of the MRE to be between 3.6 and 4.9 k.y. B.P., with a preferred age of 4.1-4.5 k.y. B.P. In Lake Tahoe near Rubicon Point, approximately 2.0 m of vertical offset is observed across the WTDPF. Based on nearby core data, the timing of this offset occurred be- tween ∼3-10 k:y: B.P., which is consistent with the MRE age in FLL. Offset of Tioga- aged glacial deposits provides a long-term record of vertical deformation on the WTDPF since ∼13-14 k:y: B.P., yielding a slip rate of 0:4-0:8 mm=yr. In summary, the slip rate and earthquake potential along the WTDPF is comparable to the nearby Genoa fault, making it the most active and potentially hazardous fault in the Lake Tahoe Basin.

Geologic and Paleoseismic Study of the Lavic Lake Fault at Lavic Lake Playa, Mojave Desert, Southern California

Paleoseismic investigations of the Lavic Lake fault at Lavic Lake playa place constraints on the timing of a possible earlier earthquake along the 1999 Hector Mine rupture trace and reveal evidence of the timing of the penultimate earthquake on a strand of the Lavic Lake fault that did not rupture in 1999. Three of our four trenches, trenches A, B, and C, were excavated across the 1999 Hector Mine rupture; a fourth trench, D, was excavated across a vegetation lineament that had only minor slip at its southern end in 1999. Trenches A–C exposed strata that are broken only by the 1999 rupture; trench D exposed horizontal bedding that is locally warped and offset by faults. Stratigraphic evidence for the timing of an earlier earthquake along the 1999 rupture across Lavic Lake playa was not exposed. Thus, an earlier event, if there was one along that rupture trace, predates the lowest stratigraphic level exposed in our trenches. Radiocarbon dating of strata near the bottom of trenches constrains a possible earlier event to some time earlier than about 4950 B.C. Buried faults revealed in trench D are below a vegetation lineament at the ground surface. A depositional contact about 80 cm below the ground surface acts as the upward termination of fault breaks in trench D. Thus, this contact may be the event horizon for a surface-rupturing earthquake prior to 1999—the penultimate earthquake on the Lavic Lake fault. Radiocarbon ages of detrital charcoal samples from immediately below the event horizon indicate that the earthquake associated with the faulting occurred later than A.D. 260. An approximately 1300-year age difference between two samples at about the same stratigraphic level below the event horizon suggests the potential for a long residence time of detrital charcoal in the area. Coupled with a lack of bioturbation that could introduce young organic material into the stratigraphic section, the charcoal ages provide only a maximum bounding age; thus, the recognized event may be younger. There is abundant, subtle evidence for pre-1999 activity of the Lavic Lake fault in the playa area, even though the fault was not mapped near the playa prior to the Hector Mine earthquake. The most notable indicators for long-term presence of the fault are pronounced, persistent vegetation lineaments and uplifted basalt exposures. Primary and secondary slip occurred in 1999 on two southern vegetation lineaments, and minor slip locally formed on a northern lineament; trench exposures across the northern vegetation lineament revealed the post-A.D. 260 earthquake, and a geomorphic trough extends northward into alluvial fan deposits in line with this lineament. The presence of two basalt exposures in Lavic Lake playa indicates the presence of persistent compressional steps and uplift along the fault. Fault-line scarps are additional geomorphic markers of repeated slip events in basalt exposures.