The San Bernardino segment is a critical segment for understanding the behavior of the San Andreas fault. Constraints on the ages of prehistoric earthquakes along this segment are needed to see whether or not past earthquakes have ruptured all the way from the Carrizo plain to the Salton Sea in a single, very large event. For example, paleoseismic evidence consistent with the occurrence of an earthquake around A.D. 1480 has been found at Bidart fan, Mill Potrero, Pallett Creek, Wrightwood, Pitman Canyon, and Indio (Grant and Sieh, 1994; Weldon and Seitz, 1994). If the paleoearthquake documented at all of these sites is indeed the same earthquake, then evidence for this earthquake should also be present along the central portion of the San Bernardino section of the fault. The most extensive paleoseismic work undertaken on the San Bernardino section of the fault to date is at Pitman Canyon, near the boundary between the San Bernardino section and the Mojave section (figure 1). Preliminary results from this site suggest 7-8 faulting events in the past 1150 years, with an average recurrence interval of about 150 years (Seitz and Weldon, 1994). This is close to the 200-year average recurrence interval obtained by dividing 4 meters of slip per event (inferred from a few channels offset this amount), divided by the slip rate of 24.5 mm/yr determined at Lost Lake (Weldon and Sieh, 1985; Weldon, 1987). My preliminary work on the central part of the San Bernardino section of the fault has been unable to confirm such a short recurrence interval, although my results are by no means conclusive at this point. The possibility that the recurrence interval along most of the San Bernardino section of the fault may be substantially longer than 150-200 years (either due to fewer events with larger displacement, or due to a slip rate of less than 25 mm/yr with more slip transferred to the San Jacinto fault or to other faults) must be tested. The fact that the San Bernardino section of the fault is the most heavily populated section of the fault in Southern California also makes this an important section to study.
In 1995 I excavated a trench across the San Andreas fault in the channel of City Creek (figure 1). The trench crossed the South Branch of the San Andreas fault zone, which is the dominant strand of the fault. The uppermost naturally deposited unit in the southern half of the trench (unit A) was unaffected by any of the fault strands visible in the trench. Pending radiocarbon dates from this unit may place an upper bound on the date of the most recent large earthquake at this site. Unfortunately, this unfaulted unit was eroded in the northern half of the trench by a younger channel which was then filled with bouldery gravel. No faults were visible within the bouldery channel fill, but it is possible that this young channel has destroyed evidence for faulting events that post-date unit A. Only one well-documented faulting event was visible within the trench. Its stratigraphic position is about 1/2 meter above a dark (organic-rich?) clayey sand layer (unit B), which has a calibrated radiocarbon age of A.D. 975-1235. Additional radiocarbon dates are pending that will bracket the age of this faulting event. Although the paleoseismic record at City Creek is probably not complete, 5 events would have to be missing at City Creek in order for the recurrence interval to be as short as it is at Pitman Canyon.
In 1996 I excavated 2 trenches across the San Andreas fault just southeast of Plunge Creek (figure 1). Mr. Stephen C. Suitt, a consulting geologist, had previously excavated 5 trenches at this site (figure 2). I excavated and logged one new trench (trench 6) and reopened trench 2, logging a new exposure about 1 m southeast of the southeastern wall of trench 2. This new exposure is referred to as trench 7 below and in figure 2. The site is located in an abandoned orchard between Plunge Creek on the northwest and Oak Creek on the south. The north-south topographic contours, perpendicular to Oak Creek, suggest that most of the site is underlain by alluvium deposited on the flood plain of Oak Creek (Qya2 on figure 2). A young alluvial fan at the mouth of a gully between Oak Creek and Plunge Creek (Qya on figure 2) buries the alluvium on the flood plain of Oak Creek. The topographic contours on figure 3 also suggest the presence of another young fan at the mouth of a gully at the northwestern corner of the property. I have received the results for a number of charcoal samples from Mr. Suitt's trenches 1-5 that I had submitted for radiocarbon dating. I have also collected over 150 charcoal samples from trenches 6 and 7, which will be used to further constrain the ages of faulting events at this site.
Figure 3 summarizes the salient relationships in each of the trenches at the Plunge Creek site, and shows the radiocarbon dates available so far. The earthquake horizons that I show in trenches 1 and 3-5 are inferred from Mr. Suitt's logs (Suitt, 1992). These earthquake horizons are only a first-cut interpretation and will need to be verified by detailed examination of additional trenches prior to publication. This is important because the goal of Mr. Suitt's study was primarily to locate and constrain the width of the fault zone for zoning purposes, not to identify prehistoric earthquake horizons.
From the data available so far, the ages of two earthquake horizons can be constrained. The younger one is visible in trench 7 (figure 4). A charcoal sample from trench 2 that most likely underlies this earthquake horizon has a calibrated radiocarbon date of A.D. 1315-1615. Another sample, which overlies the earthquake horizon has been dated at A.D. 1515-1950, and is not very useful in constraining the minimum age of the event. I have collected a number of charcoal samples from trench 7 which should allow me constrain the age of this event more tightly (see figure 4). An older earthquake horizon is clearly visible in Mr. Suitt's log of trench 3 (event A1 in figure 3, trench 3). Calibrated radiocarbon dates from below and above this horizon bracket it between A.D. 900-1215 and A.D. 1036-1405.
One hypothesis (hypothesis A) that could be tested at this site is that the two earthquake horizons mentioned above represent the only earthquakes that have ruptured through this site since A.D. 900-1215. In support of this hypothesis, Mr. Suitt's log of trench 3 only shows one earthquake horizon that is stratigraphically higher than the one mentioned above (event A2 in figure 3, trench 3). This is by no means conclusive, since the fault strands involved in this younger earthquake extend upward into a massive zone just below the ground surface, and they might represent more than one event. In Mr. Suitt's log of trench 3, however, the fault strands that extend up near the surface are not very well expressed, and do not give the appearance of having ruptured more than once. Further support for hypothesis A is found in trench 5. Only one faulting event is visible in trench 5, and it slightly postdates A.D. 975-1170. Thus, this faulting event is probably the same as the lower event in trench 3 (event A1). The upper event in trench 3 is not visible in trench 5, but this may be because trench 5 does not cross the entire fault zone. The fact that trench 5 does not cross the entire fault zone hinders interpretations. Nonetheless, if the fault strands that extend close to the ground surface in trench 3 had ruptured more than once, it is likely that at least one of these earthquakes would be visible in the young fan sediments in the upper part of trench 5. Figure 3 shows how the relationships in each of the trenches can be interpreted in a manner consistent with hypothesis A. In figure 3, earthquake horizons that are interpreted as the older of the two events are labelled event A1 and those that are interpreted as the younger event are labelled event A2. If hypothesis A is correct it would mean that the recurrence interval for this part of the fault is between 270 and 900 years.
To construct a second, endmember hypothesis (hypothesis B), I interpret the seven trenches in such a manner as to produce the greatest number of earthquakes that can be inferred from the data. Mr. Suitt's log of trench 1 shows fault strands rupturing up to three different elevations, but it is not clear whether these three elevations are three different stratigraphic levels, or whether fault strands rupturing to two different elevations might have formed in the same earthquake. For hypothesis B, I interpret 3 separate faulting events in trench 1, the oldest of which is the same as the oldest event in trench 3 (event B1, between A.D. 900-1215 and A.D. 1036-1405). The upper two possible faulting events in trench 1 are then events B2 and B3. I also interpret all three of these events as older than the event I have documented in trench 7 (which postdates A.D. 1315-1615). This makes the event in trench 7 event B4. For hypothesis B I also allow that the event exposed in trench 7 is not the most recent event. This is possible, since the unfaulted strata that overlie event B4 in trench 7 do not cross the entire fault zone. In fact, part of the main fault zone in trench 7 is occupied by massive fault-scarp colluvium. No discrete fault strands are visible within this colluvium, but faults could be present yet invisible due to the lack of bedding within the colluvium. Thus it is possible that a younger earthquake, event B5, has ruptured up through this colluvium. If hypothesis B is correct, there have been 5 large earthquakes since A.D. 900-1215, and the average recurrence interval would be between 150-225 years, comparable to previous inferences for the San Bernardino segment.
References cited:
Grant, L. B. and K. Sieh, 1994, Paleoseismic evidence of clustered earthquakes on the San Andreas fault in the Carrizo Plain, California, Jour. Geophys. Res., 99, pp. 6819-6841.
Seitz, G. and Weldon, R., II, 1994, The paleoseismology of the southern San Andreas fault at Pitman Canyon, San Bernardino, California, in McGill, S. F. and Ross, T. R., eds., Geological Investigations of an Active Margin, Geological Society of America, Cordilleran Section Guidebook: San Bernardino County Museum Association, pp. 152-156.
Suitt, S. C., 1992, Feasibility-level geological investigation, 45-acre East Highland parcel, San Bernardino County, California.
Weldon, R. J. II., 1987, San Andreas fault, Cajon Pass, southern California, in Hill, M. L., ed., Cordilleran Section of the Geol. Soc. Amer., Centennial Field Guide, vol. 1, pp. 193-198.
Weldon, R. J., II, and Sieh, K. E., 1985, Holocene rate of slip and tentative recurrence interval for large earthquakes on the San Andreas fault, Cajon Pass, southern California: Geological Society of America Bulletin, v. 96, no. 6, p. 793-812.
Publications from previous SCEC grants:
Sieh, K., ... S. McGill, ... , Near-field investigations of the Landers earthquake sequence, April to July 1992, Science, v. 260, pp. 171-176, 1993.
McGill, S. F. and C. M. Rubin, Surficial slip distribution on the central Emerson fault during the 28 June 1992 Landers earthquake, California, submitted to J. Geophys. Res. in June 1997(?). McGill, Sally F., Variability of surficial slip in the 1992 Landers earthquake: Implications for studies of prehistoric earthquakes, Proc. of Workshop on Paleoseismol., 18-22 Sept. 1994, U.S. Geol. Surv. Open-File Rpt. 94-56, pp. 118-120, 1994.
Rubin, C. M. and S. F. McGill, 1992, The June 28, 1992, Landers earthquake: slip distribution and variability along a portion of the Emerson fault, EOS, Transactions of the American Geophysical Union, v. 73, pp. 362-363. (INVITED)
McGill, Sally, 1996, Preliminary paleoseismic results from the San Andreas fault at City Creek, near San Bernardino, California, SCEC 1996 Annual Meeting.
