University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

USGS Staff -- Published Research US Geological Survey

1999

Abrupt along-strike change in tectonic style: San Andreas fault zone, San Francisco Peninsula

Mary Lou Zoback U.S. Geological Survey, [email protected]

Robert C. Jachens U.S. Geological Survey

Jean A. Olson Timble Navigation Company

Follow this and additional works at: https://digitalcommons.unl.edu/usgsstaffpub

Part of the Earth Sciences Commons

Zoback, Mary Lou; Jachens, Robert C.; and Olson, Jean A., "Abrupt along-strike change in tectonic style: San Andreas fault zone, San Francisco Peninsula" (1999). USGS Staff -- Published Research. 467. https://digitalcommons.unl.edu/usgsstaffpub/467

This Article is brought to you for free and open access by the US Geological Survey at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in USGS Staff -- Published Research by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. B5, PAGES 10,719-10,742, MAY 10, 1999

Abrupt along-strike change in tectonic style: San Andreas fault zone, San Francisco Peninsula

Mary Lou Zoback, Robert C. Jachens,and Jean A. Olson• u.s. GeologicalSurvey, Menlo Park, California

Abstract. Seismicityand high-resolutionaeromagnetic data are usedto define an abrupt changefrom compressionalto extensionaltectonism within a 10- to 15-km-widezone along the San Andreas fault on the San FranciscoPeninsula and offshore from the Golden Gate. This 100-km-longsection of the San Andreas fault includesthe hypocenter of the Mw - 7.8 1906 San Franciscoearthquake as well as the highestlevel of persistent microseismicityalong that -470-km-long rupture. We define two distinctzones of deformationalong this stretchof the fault usingwell-constrained relocations of all post- 1969 earthquakesbased a joint one-dimensionalvelocity/hypocenter inversion and a redeterminationof focal mechanisms.The southernzone is characterizedby thrust- and reverse-faultingfocal mechanismswith NE trending P axesthat indicate "fault-normal" compressionin 7- to 10-km-widezones of deformation on both sidesof the San Andreas fault. A 1- to 2-km-wide vertical zone beneath the surface trace of the San Andreas is characterizedby its almostcomplete lack of seismicity.The compressionaldeformation is consistentwith the young,high topographyof the Santa Cruz Mountains/CoastRanges as the San Andreasfault makesa broad restrainingleft bend (-10 ø) throughthe southernmostpeninsula. A zone of seismicquiescence -15 km long separatesthis compressionalzone to the southfrom a zone of combinednormal-faulting and strike-slip- faultingfocal mechanisms(including a M L = 5.3 earthquakein 1957) on the northernmostpeninsula and offshoreon the Golden Gate platform. Both linear pseudo- gravitygradients, calculated from the aeromagneticdata, and seismicreflection data indicatethat the San Andreas fault makes an abrupt -3-km right step lessthan 5 km offshorein this northernzone. A similarright-stepping (dilatational) geometry is also observedfor the subparallelSan Gregorio fault offshore.Persistent seismicity and extensionaltectonism occur within the San Andreas right stepoverregion and at least 15 km along-strikeboth to the SE and NW. The 1906 San Franciscoearthquake may have nucleatedwithin the San Andreas right stepover,which may help explain the bilateral nature of rupture of this event. Our analysissuggests two seismichazards for the San FranciscoPeninsula in addition to the hazard associatedwith a M - 7 to 8 strike-slip earthquakealong the San Andreas fault: the potential for a M • 6 normal-faulting earthquakejust 5-8 km west of San Franciscoand a M -- 6 + thrust faulting event in the southernpeninsula.

1. Introduction This fault is part of the San Gregorio-Hosgri fault system,a mostlylate Tertiary, plate boundaryfault zone with a cumula- Approximately100 km southwestof the San FranciscoPen- tive offset of 100-150 km, that extends from Point Sal in insula the very linear central California segmentof the San southernCalifornia and mergeswith the San Andreas fault on Andreas fault bifurcates,with the westerlystrand heading to- the Golden Gate platform [Grahamand Dickinson,1978; Clark ward the peninsulaas the SanAndreas fault proper (Figure 1). et al., 1984]. Geologic data, including trenching of the San The Hayward-Calaverasfault systemsin the East Bay form an Grcgorio fault, indicate multiple Holocene surfacerupturing eastern, more northerly striking strand. Together these two eventsand a sliprate of -5 mm yr-1 [Simpsonet al., 1997; strandsaccount for roughly-32 mmyr -• right-lateralshear, WorkingGroup on Northern California EarthquakePotential, or approximately85% of the plate motion east of the Sierra 1996]. Nevada [WorkingGroup on Northern California Earthquake Four significanthistoric earthquakeshave occurredon the Potential,1996; Williamset al., 1994]. A third strike-slipfault San FranciscoPeninsula (Figure 1): (1)aM w • 6.8-7.5 event zone subparallelto the San Andreas fault, the San Gregorio in 1838 on the central peninsula,which on the basisof first- fault, crossesthe westernmostpart of the peninsulaand the hand accountsof the surface rupture appears to be a right- shelfoffshore from the Golden Gate (Golden Gate Platform). lateral strike-slipevent on the San Andreasfault [Louderback, 1947; Ellsworth, 1990; Tuttle and Sykes,1992; Toppozadaand 1Nowat TrimbleNavigation Company, Sunnyvale, California. Borchardt,1998; Bakun, 1999]; (2) the M w = 7.8 1906 San This paper is not subjectto U.S. copyright.Published in 1999 by the Francisco earthquake with an epicenter probably located American GeophysicalUnion. somewherenear where the San Andreas fault goes offshore, Paper number 1998JB900059. southof the Golden Gate [Bolt, 1969;Boore, 1981; Waldet al.,

10,719 10,720 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

122'45' 122'30' 122ø15 ' 122'

38' 38'

37'45' 37o45-

.:-

37'30' -•?111i,37'30' -fif•ß- *.::.

Menlo ::•...

ß :,:3 '•'

:., •;'•-'.•:::.::..

Jose

37'15'

37' ';;'6"•'•.• • 122'16' 122'

Figure 1. Shadedtopography map showingQuaternary faults in the greater San FranciscoBay area [Bor- tugnoet al., 1992]. The inset showsmajor strike-slipfault systemsin central California. Large circlesindicate approximateepicenters of large historic earthquakes.The location of the 1838 epicenter is unknown;it is plotted on the central peninsulanear the center of the assumedrupture.

1993; Thatcheret al., 1997]; (3) the M,• - 6.9 1989 Loma can plate boundary(including the East Bay fault systems)for Prieta earthquakeon the southernmostpeninsula; and (4) a the past 160 years [Bakun, 1999]. M • 6.5 event in 1865 probablylocated near the city of San Whereasgeologic, historic earthquake, and geodeticdata all Jose,which may have been a thrust or oblique thrust event indicate that the dominant style of deformation in the San similar to the 1989 Loma Prieta earthquakeand possiblynot FranciscoPeninsula region is right-lateral strike-slipfaulting, on the San Andreas fault [Tuttle and Sykes,1992; Jaum• and changesin geometry of the San Andreas fault systemin the Sykes,1996]. While there is considerablecontroversy about the vicinity of the peninsula coincide with marked along-strike magnitudeof the 1838 event (estimatesrange from 6.8 to 7.5 variations in its geomorphicand topographicexpression. In [Bakun,1999; Toppozada and Borchardt,1998]), at a minimum crossingthe southernpeninsula, the San Andreas fault under- the 1838 and 1906 earthquakessum to about two thirds of the goesa broad bend to the left (a 12- to 15-km westwardbend expectedmoment releasefor the entire Pacific-North Ameri- over ---90-kmlength of fault, resultingin a ---10ø-12ø compres- ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,721 sional"restraining" bend), analogousto the "Big Bend" in the the north; for more information on this event the reader is Transverse Ranges of southern California but on a much referred to Reasenberg[1997], Simpson [1994], Spudich [1996], smallerscale. As in the Big Bend regionin southernCalifornia, and Wellsand Vidale [1999]. there is a zone of uplift (25-30 km wide) straddlingthe San Andreasfault that is particularlypronounced on the SW side, 2. Offshore Fault Locations Inferred From forming the Santa Cruz Mountains. The 1989 Loma Prieta event occurredwithin this bend region and exhibited nearly Aeromagnetic Data equal componentsof right lateral strike-slipand reversefault Only limited informationhas been availableon the morphol- motion (rake = 140ø) [Oppenheimer,1990]. Both geodeticdata ogy of the San Andreas fault offshore on the Golden Gate and surface deformation features suggestthat this event did platform. The lack of a clear fault trace on the seafloor in not occur on the sameportion of San Andreas fault that rup- high-resolutionseismic profiling [Cooper,1973] as well as the tured in 1906 [Thatcheret al., 1997; Prenticeand Schwartz, lack of alignedoffset featureson detailed (1-m contour) ba- 1991; Segalland Lisowski, 1990]. However, this event drew thymetrymaps (T. E. Chase,U.S. GeologicalSurvey (USGS), attentionto the seismicpotential of the convergentcomponent written communication,1994) indicatesthat sedimentarypro- of plate motion in the Bay area. cesseson the shelf outpacetectonic processes in determining On the central and northern peninsulathe active strand of geomorphicexpression of the fault zone. In fact, the coastline the San Andreas fault is straight, with an average trend of where the San Andreas fault cuts offshore is one of the N36øW, only --•1ø oblique to the N34.5øW direction of relative straightestsegments of the California coastlinebecause of the Pacific-westernmostNorth American plate motion [Argusand combinationof easilyerodible Pliocene-Quaternarysediments Gordon, 1991]. On the northernmostpeninsula the zone of forming beach cliffs and swift, long-shorecurrents. uplift adjacentto the San Andreas fault becomesmuch nar- We used high-resolutionaeromagnetic data to help detail rower (total width of 5-10 km) and topographicallymore sub- the geometryof San Andreasfault offshorein the region of the dued than to the south. Although Quaternary thrust faults inferred right bend or step-over(Plate 1). These data contain subparallelto the San Andreas fault are mapped all along the prominent anomalieswhose edges are, in many cases,con- peninsula,the number and cumulativeoffset on these thrusts cealed fault traces.The aeromagneticdata are shownin Plate observedat the surfacegreatly decreasenorth of about 37o25' 1 and were collectedin 1994 and 1996 with fixed-wingaircraft [Aydinand Page, 1984]. A 2- to 3-km right step or bend in the along flight lines 500 m apart at a nominal height between San Andreas fault systemcan be inferred acrossthe Golden 250 m and 300 m above the surface of the Earth or ocean. For Gate platform by projectingonland tracesof the San Andreas details of the data collection and reduction techniques,see fault from the north and south.A right step in a right-lateral Jachensand Zoback [1999]. strike-slipsystem implies a componentof extension.It may not Offshore extensionsof the San Andreas and San Gregorio be a coincidence that the elevation of the San Andreas fault faults appear to bound, in map view, a triangular- or wedge- decaysnorthward on the peninsulatoward the Golden Gate shapedbody characterizedby a magnetic high (Plate 1). A platform, where it presentlylies below sea level. third fault, the Pilarcitos fault, which lies between the San The subjectof this paper is a seismotectonicand seismic Gregorio and San Andreas faults forms the southernboundary hazard analysisof the complex100-km-long stretch of the San of this block. The Pilarcitos fault locally forms the geologic Andreas fault extendingNW from the north end of the Loma boundary between late Mesozoic Franciscan forearc assem- Prieta aftershockzone to the Point ReyesPeninsula, just north blage to the northeastand primarily granitic Salinian rocks to of the Golden Gate platform. In addition to the along-strike the southwest,just as the San Andreas fault does in central variationsin fault geometry and geomorphicexpression dis- California. The Pilarcitos is believed to represent an aban- cussed above, this section of the San Andreas fault also in- doned strike-slip portion of the plate boundary [Brabb and cludesthe 1906 epicenteras well as some of the highestlevels Pampeyan,1983; Wagneret al., 1991;Page, 1990, 1992; Griscom of microseismicityanywhere along the 470-km-long 1906 rup- and Jachens,1989]. The magnetichigh in the wedge-shaped ture (includinga M L - 5.3 1957earthquake just southof San block (here called the Pilarcitosblock) indicatesthat mafic- Francisco(Figure 1)). Significantly,this stretch of the San rich Franciscanrocks within this block generally are more Andreas fault also representsthe only fault segmentin the magneticthan graniticrocks of the Salinianblock, southwestof United Stateswith a proven potential for a M w = 7.8 earth- the San Gregorio and Pilarcitosfaults, as well as the rocks of quake that directly transectsa major urban area. The objec- the predominately FranciscanCentral Belt northeast of the tives of our study are to (1) use accuratemicroearthquake San Andreasfault [Blakeet al., 1984].Because the three faults locationsand well-constrainedearthquake focal mechanisms bound the magneticPilarcitos block, their concealedpositions to determinethe relationshipbetween deformation on the San in the offshorewere determined by defining the edgesof the Andreas fault proper and that within in the crustal blocks magneticrocks that comprisethe block. Identifying the precise directly adjacentto the fault, (2) constrainthe geometryof locations of the fault traces from the contoured data shown in faultingoffshore using high-resolution aeromagnetic data, (3) Plate 1 is difficult becauseof the complexmagnetic anomalies analyze, in light of the along-strikegeometric and tectonic that result from the dipole nature of magneticsources. variations,possible seismic hazard and segmentationimplica- We inferred the offshore locations of the fault traces by tions for strike-slipfaulting along the San Andreas fault, and means of a numerical technique utilizing a linear filter, the (4) evaluatethe long-termseismic hazard potential of simul- pseudogravitytransform [Barahoy, 1957] that removes the taneouslyactive, subparallelnormal and thrust faults located magneticdipole effect and convertsa magneticanomaly to an directlyadjacent to the SAF and separatedalong strike by only equivalent gravity anomaly by assuminga constant ratio of 15-20 km. The 1989 Loma Prieta earthquakeand aftershock magnetizationto density.The maximum horizontal gradients zone at the southernmostend of the peninsulaare described of a gravity anomalyproduced by a shallowlyburied body lie only briefly in terms of seismicitypatterns observed directly to nearly over the sidesof the body, particularly if the sidesdip 10,722 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA steeply[Cordell and Grauch,1985; Blakely and Simpson,1986]. picksfrom a page size figure. Cooper[1973] also delineateda Therefore the maximum horizontal gradientsof the pseudo- distinct,somewhat sinuous 20-km-long, 2- to 3-km-wide "San gravityanomaly (transformed magnetic anomaly) can be used Andreas"graben sitting on top of a broader basinbetween the to define the locationsof the lateral boundariesof the mag- San Andreas and San Gregorio faults; the boundariesof this netic sourcebody. We modifiedthis techniqueslightly by ap- youngestgraben correlate well with our newly definedoffshore plying the maximumgradient analysisnot to the simplepseu- fault traces(Plate 2, inset). Further confirmationof the right dogravitytransformation, but rather to the differencebetween step in the San Andreas fault comesfrom reflectionsfrom a the transformedmagnetic data and those same data upward near-verticalfault identifiedby Hole et al. [1996] on a marine continuedto 200 m. This differencingis equivalentto usinga seismicprofile out the Golden Gate. They interpreted these high-passfilter; it focused the procedure on the shallowest reflectionsas comingfrom depthsof 3-7 km on a near-vertical parts (upper ---2 km) of the magneticbodies. The inferred San Andreas fault that coincidedat the surfacewith Cooper's shallowmagnetic edge boundaries (defined by localmaxima in [1973] eastwardtrace of the San Andreas. These reflections the horizontal pseudogravitygradient) are shown as small were not compatiblewith a San Andreaslocation correspond- "plus" symbolson Plates 1 and 2. Spatial resolutionof the ing to a simpleNW projectionof the onland fault trace shown locationof the inferred magneticboundaries is probablybetter in Figure 1. In addition, Cooper mappeda subtleleft bend in than 100 m based on our ability to accuratelylocate cultural the NW end of the "recenttrace" of the SanAndreas (Plate 2, features such as bridgesand large buildings. inset)in the regionSW of BolinasLagoon. The left benditself We have highlightedon Plate 2 those magneticboundaries is not reflectedin the magneticdata, but both endsof Cooper's that we interpret to represent the concealedtraces of the fault segmentnorth and south of the bend coincidewith linear Pilarcitos, San Gregorio, and San Andreas faults offshore on magneticboundaries (Plate 2, inset). the Golden Gate platform.We selectedthese specific magnetic The magneticdata also suggestthat the offshoreportion of boundariesfrom the total set of possiblechoices because (1) the San Gregoriofault consistsof two main, partly overlapping they define the lateral boundariesof the magneticbasement right-steppingstrands, with possiblyother parallel but lesswell rocksthat constitutethe fault-boundedPilarcitos block; (2) for definedstrands. The westernmoststrand is remarkablystraight the San Andreas and Pilarcitosfaults, they representoffshore for a distanceof >20 km, from a point ---7 km north of Half continuationsof the onshore magnetic boundariesthat coin- Moon Bay northward to a point nearly oppositethe Golden cidewith mappedfault tracesacross which magnetic basement Gate. Near the north end of thisstrand, the San Gregoriofault rocksare juxtaposedagainst nonmagnetic basement rocks; and stepsright about 1.5 km, to a slightlycurved, more eastern (3) for the San Gregoriofault, thesemagnetic boundaries not strand that continues north for an additional 15 km. This only occupythe proper structurallocation at the edge of the easternstrand bendswest near its north end, eventuallypar- Pilarcitosblock, but they are the only long,straight, continuous alleling the San Andreas fault, about 2.5 km to the southwest boundariesthat lie within the zone previouslydefined as the of BolinasLagoon, but neverjoining it. The two main strands San Gregorio fault zone. The lack of detailed correlationbe- of the San Gregorio fault overlapfor a distanceof at least 6 km tween our fault picks and the existingfault mapping offshore (and possiblyas muchas 12 km) at the right step.The Pilar- (blacklines on Plate2) is probablydue to the fact that previous citos fault can also be traced offshore for a distanceof 7-8 km, fault mapping was based on interpretation of relatively low bendingnorthward near its westernend to mergewith the San resolutionmagnetic data and very sparselydistributed marine Gregorio fault. seismicprofiling [McCulloch,1987]. The northernmostpart of the San Andreasfault on the San FranciscoPeninsula can be traced by its associatedmagnetic 3. Earthquake Relocation Procedure boundary out to sea on strike NNW for a distance of 6 km To improve the quality and precisionof earthquakeloca- (Plate 2), at whichpoint the magneticanomaly changes char- tions for the San FranciscoPeninsula region, we relocated all acter and the fault projectsinto the centerof the magnetichigh magnitude >1.0 earthquakesthat had previouslybeen rou- that characterizesthe Pilarcitos block (Plate 1). Similarly, tinely locatedby the USGS Northern California SeismicNet- north of the Golden Gate, the easternmapped strand of the work (NCSN; seeOppenheimer et al. [1993]for a descriptionof San Andreasfault at BolinasLagoon can be tracedoffshore by standardprocessing). The relocationprocedure was carried its associatedmagnetic boundary on strike in a straight line out in two stepsand utilized VELEST, a joint least squares SSW for a distanceof >20 km, to the shorelineof the penin- inversionprocedure that minimizestraveltime residuals [Ells- sulaat Lake Merced(Plate 1). The two strandsdo notjoin, but worth, 1977; Roecker, 1981]. First, a local best fitting one- rather define a right step of 3 km in the fault systemjust dimensional(l-D) velocitymodel was determined; then station offshore from Lake Merced. corrections were determined for stations both within and ad- Cooper [1973] used disruptionof the youngestsediments jacent to the studyarea. The area of relocatedevents is out- (upper ---5 m) imagedon shallow,high-resolution seismic re- lined by the inner small rectanglesin Figure 2. flection data to identify the "recent trace" of the San Andreas The startingvelocity model for the inversion(Peninsula fault offshore from the Golden Gate. He detected an eastward modelin Figure3) wasbased on a refractionvelocity model for offset in the San Andreas fault, which he mapped as an abrupt a NW trendingprofile alongthe peninsulagenerally coinciding bend in the projectedonshore portion of the SanAndreas. His with the SanAndreas fault zone [Catchingsand Kohler,1996]. abrupt right bend in the San Andreastrace occursin the same From this startingmodel a bestfitting 1-D velocitymodel was area that the magneticdata indicate an offset 3 km to the NE determinedusing only stationswithin the inner rectanglesin (see Plate 2). Cooper'seasternmost "recent trace" of the San Figure 4 and about 200 eventsincluding (1) 158 local earth- Andreas fault coincides within 500 m with our inferred offset quakeswith floating(unconstrained) hypocenters, (2) numer- segmentof the San Andreasfault, remarkableagreement since ous calibrationexplosions and quarry blastswith known loca- we do not have precise locations of Cooper's data but only tions,and (3) 15 moderateto regional(M > 5.0) earthquakes ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,723

122"40' 122'30' 122"20'

37"55' 37"55'

37ø50 37"50'

37'45' 37045 '

37'40' 37ø40 '

37'35' 37ø35 '

122'40' 122'30' 122'20'

-210 -180 -150 -120 -90 -60 -30 0 nT Plate 1. New high-resolutionaeromagnetic map of San FranciscoBay area. Quaternaryfaults from Bor- tugnoet al. [1992] are shownin red. Small white plusesindicate maximum horizontal gradient of the short wavelengthcomponent of the pseudo-gravityfield (see text for explanation). 10,724 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

122'40' 122'30' 122'20'

37'50'

'I,

37'40'

122'40' 122'30' 122'20' Plate 2. Maximum horizontal gradient locations(red pluses)from Plate 1 and inferred offshore fault structure(heavy yellow lines). Thin blacklines are the currentlymapped offshore Quaternary faulting from Bortugnoet al. [1992];the 1906rupture on the SanAndreas fault is in boldblack. The insetcompares new fault interpretationwith Cooper's[1973] "recent trace of the San Andreasfault" (heavyblack dashed line) and young"San Andreas" graben (thin greendashed line), both inferredfrom high-resolutionseismic profiling. ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,725

I I • Shot(fixed) Quarryblast(fixed) ß Regionalevent M _>5.0 (fixed)

Local event M >_2.5 ß (floating) 38 o_ /x Station

37 o_ 0 50 KM

/x /x

I I 123 o 122 o

Figure 2. Map of calibration explosions,quarry blasts, earthquakes,and stations used in the velocity inversion.Inner rectanglesmark northern (dashed)and southern(solid) studyareas in which earthquakes were relocated.The outer bold, dashedrectangle defines area additionalarea where stationcorrections were redetermined in order to better constrain focal mechanisms.

with fixed hypocenters.The regional earthquakeswere in- km s-• is identicalto the refraction-determinedMoho velocity cludedto better resolvethe refractorvelocities, especially Pn. in that survey. For the regional events we excluded arrivals within 20 km The secondstep in the inversionprocedure was to determine becausewe wanted the local eventsto provide the data for individualstation delays.These time adjustmentshelp correct determiningthe shallowervelocity structure. for variations in near-surface site conditions as well as for Figure 3 showsthe final velocitymodel (VELEST) in com- geologiccomplexity resulting in departuresfrom the assumed parisonwith severalvelocity models for the San FranciscoBay 1-D velocity model. Becauseof differencesin tectonismand region. Velocity layer boundariesin the inversionwere set at geomorphologythat couldbe reflectedin azimuthallyvariable 3-km intervals(except in the uppermost2 km). The resulting seismicvelocities, station delays were computedseparately for best fit velocitystructure is generallyslower than the starting eventsoccurring in the northern(dashed) and southern(solid) refraction-derivedPeninsula model of Catchingsand Kohler subareas(inner rectanglesin Figure 2). Station delayswere [1996]. However, it representsan average of Catchingsand also computedfor stationswithin the larger, outer rectangle Kohler's[1996] model and two other velocity models deter- shownin Figure 2 so that we could utilize first motion data mined in the peninsulaarea: one from a profile alongthe axis from these stationsto calculatefault plane solutions. of San FranciscoBay and the other from a profile runningout Determining station correctionsfor the southernarea was the Golden Gate, both from Holbrook et al. [1996]. We ran straightforwardbecause of the dense station spacing.Fixed testsinvolving 1-km variationsin the Moho depth utilizing the sourcesin the southernarea includequarry blasts with known regional(M -> 5.0 events)and found that an averageMoho locations as well as six chemical shots from Catchingsand depth of •--22-24 km best fit the regional earthquaketravel Kohler's [1996] active seismicexperiment. To obtain station times. The adopted averageMoho depth of •--23.5km in the delaysfor the northern subarea,we used three fixed sources: final velocity model is not well constrainedby these tests. Uhrhammer's[1981] relocationof a Mz. = 4.4 1979 San An- However,the Moho is probablydeeper than the 21-km depth dreasearthquake, a controlledoffshore detonation in 1992 of determinedin Catchingsand Kohler's[1996] refraction survey a World War II marine mine by the Navy (37.6440øN, alignedalong the San Andreasfault. Our Moho velocityof 8.0 122.5312øW),and one of Catchingsand Kohler's refraction 10,726 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

122'45' 122'30' 122'15' 122'

38'

37'45'

37'30'

37'15

37'

122'45' 122'30' 122'15' 122'

-2•0-lbO .....•)" i•)0 200 360"46'0 ...... 560 6t•0 700 800 meters Plate 3. Relocated seismicity(M,• _> 1.5) for San FranciscoPeninsula together with relocatedseismicity (M d _> 1.5) in the San Franciscobay block [Olsonand Zoback, 1998], and NCSN locationsof M d -> 2.0 eventsin the East Bay and Loma Prieta aftershockregion. Seismicityis for the time period 1969-1995. ZOBACK ET AL.' SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,727

P-wave velocity (km/sec)

2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 0.0

-5.0

-10.0

-15.0

-20.0 VELEST ...... PENINSULA ...... GOLDEN GATE S. F. BAY

-25.0

Figure 3. Velocity-depthmodel determinedfrom the VELEST inversioncompared to velocitydepth de- terminationsfrom activeseismic experiments in the Bay area:Peninsula profile [Catchingsand Kohler,1996]; GoldenGate profile [Holbrooket al., 1996];and San FranciscoBay profile [Holbrooket al., 1996]. shots beneath the Golden Gate. Station corrections deter- ing the empiricallyderived equation of Eaton [1992], which mined for the two areas correlate well with surface geology. closelyapproximates Mz• for M d < 3.8 events. The averagestation delay for stationson bedrockis +0.02 s, Resolutionin the northern subareais much poorer because whereasthe averagedelay on stationson Tertiary sedimentary of only one offshore station and potential complicationsre- rocks was -0.14 s. lated, at leastlocally, to a large velocitycontrast across the San Once the new station correctionswere computed,we relo- Andreas fault. Uhrhammer[1981] did a careful study of the cated the NCSN hypocentersin the studyarea usingthe stan- location and focal mechanism solution for one of the largest dard NCSN locationprogram, HYPOINVERSE [Klein,1989]. events in the northern area since 1957, the 4.4 event on the The relocationprocedure resulted in a slightreduction in the coastsouth of San Franciscoin 1979. His map of first-motion overallmean root-mean-square(RMS) travel time residualfor polaritiesindicates that the earthquakeis a right-lateralstrike- the data set (from 0.09 _+0.05 to 0.08 _+0.03 s). The relocation slip event probablyoccurring on the San Andreas. However, resulted in a reduction in standard location errors, as indicated many of the distant (>45 km) first arrivalsshow a distinct by the standardestimate of the horizontal(ERH) and depth lateral refractionof 34ø (implyinga velocitycontrast of 19%, (ERZ) errorsfor eachevent calculated by HYPOINVERSE. with the higher velocity on the NE side of the fault). These There was a 25% reduction in ERH and a 30% reduction in refracted arrivals added noise to the focal mechanism while ERZ aswell as a 50% and 64% reduction,respectively, in their not obscuringit. Uhrhammer usedthese observations to relo- standarddeviations (ERH, 0.46 _+0.33 km from 0.57 _+0.66 cate the 1979 event to lie almost directly on the San Andreas km; ERZ, 0.77 _+0.52 km from 1.11 _+ 1.45 km). Note that fault, a 3.5-km shift to the NE from his preliminaryepicentral these standarderror parametersunderestimate absolute er- location.Our relocationof the 1979 event placesit 600 m SW rors; both ERH and ERZ are about 0.42 times smaller than the of the San Andreas fault and Uhrhammer's location. Hence it principalaxes of the 95% confidenceellipsoid for eachhypo- is likely that muchof the relocatedseismicity distribution in the center. A more reliable measure of the true errors of the northern zone may be shifted ---600 m SW from its actual hypocentrallocations comes from relocationof knownquarry location. blastsand refraction shotswithin the study area. Comparing our locations with the actual locations, we found horizontal errorsthat rangedfrom 0.06 to 2.50 km, with a mean of 0.85 _+ 4. Hypocentral Distribution and Focal 0.47 km. The verticalerrors ranged between 0.20 and 3.10 km, Mechanisms of Relocated Seismicity with a mean of 0.71 _+0.39 km. For the purposesof this study, An epicentral map of well-located earthquakesusing the we defined well-located events as those with ERH < 2.0 km, selection criteria detailed above and locations based on the ERZ < 3.0 km, and RMS < 0.20 s, and recorded on >7 new 1-D velocitymodel and separatestation corrections for stations.Coda (duration) magnitudesM d were computedus- the northern and southern subareas is shown in Plate 3. The 10,728 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

Table 1. Focal Mechanism Parameters for Northern Area

Dura- Plane 1 Plane 2 P Axis T Axis Lati- Longi- tion Number tude, tude, Depth, Magni- of First Azi- Azi- No. Date Time, UT N W km tude Motions Strike Dip Strike Dip muth Plunge muth Plunge

la March 31, 1971 1703:36.21 37ø40.81 122030.57 ' 9.23 3.20 18 160 15 309 77 209 57 45 32 lb March 31, 1971 1703:36.21 37ø40.81 122030.57 ' 9.23 3.20 18 345 45 192 48 172 76 269 2 2 July 25, 1971 0913:28.86 37ø42.60 122030.64 ' 7.46 3.40 32 150 70 38 44 17 49 268 15 3a Nov. 1, 1977 0406:43.23 37ø42.58 122032.38 ' 6.50 3.10 23 320 80 226 70 185 21 92 7 3b Nov. 1, 1977 0406:43.23 37ø42.58 122032.38 ' 6.50 3.10 23 40 45 144 76 14 42 265 19 4 April 28, 1979 0044:44.61 37ø38.72 122028.02 ' 12.90 4.40 38 60 85 152 70 14 18 108 10 5 March 9, 1986 0128:12.62 37ø40.01 122029.92 ' 4.61 3.00 28 335 85 239 40 210 37 96 29 6a Nov. 2, 1986 0346:14.27 37ø38.08 122029.08 ' 7.55 3.00 32 45 90 135 80 360 7 90 7 6b Nov. 2, 1986 0346:14.27 37ø38.08 122029.08 ' 7.55 3.00 32 320 90 230 30 203 38 77 38 7a Feb. 1, 1987 1753:18.76 37042.97 ' 122ø31.16 ' 4.69 3.40 40 335 85 66 80 21 3 290 11 7b Feb. 1, 1987 1753:18.76 37042.97 ' 122ø31.16 ' 4.69 3.40 40 20 40 187 51 53 82 283 5 8 Sept. 19, 1987 1833:59.07 37046.66 ' 122034.94 ' 10.58 3.30 30 55 55 190 45 22 65 124 6 9 Dec. 30, 1987 0109:10.58 37046.72 ' 122034.82 ' 10.32 3.20 49 20 45 161 52 9 69 269 4 10 Feb. 12, 1989 1949:05.49 37049.66 ' 122036.26 ' 4.84 3.10 42 25 25 152 74 37 56 258 27 11 Feb. 12, 1989 1949:49.92 37049.66 ' 122ø35.84 1.78 2.90 24 20 45 173 48 13 76 276 2 12 Feb. 13, 1989 0600:28.16 37050.05 ' 122ø36.35 4.39 3.00 31 25 70 178 22 311 64 107 24 13 Feb. 13, 1989 0801:41.58 37ø50.16 ' 122ø36.39 4.22 3.60 39 60 85 163 21 351 46 132 37 14 Nov. 6, 1989 2337:24.26 37038.96 ' 122ø28.69 4.14 3.10 48 120 65 10 54 340 45 243 6 15 Nov. 24, 1993 1518:08.14 37046.43 ' 122ø35.04 10.49 3.00 36 35 50 162 54 12 60 278 2 16 Jan. 2, 1994 0311:18.24 37046.35 ' 122ø35.06 10.91 3.00 41 25 35 159 64 30 63 267 16 17 Nov. 19, 1991 0207:00.28 37034.64 ' 122ø30.27 7.99 2.40 35 65 75 332 80 19 4 288 18 18 Feb. 16, 1992 1823:22.55 37034.89 ' 122ø30.78 7.84 2.40 43 70 75 335 71 202 3 293 25 19 March25,1996 1117:11.93 37034.65' 122ø25.58 7.58 2.70 60 10 65 149 32 314 64 85 18 20 July 1, 1996 0433:05.09 37042.80' 122032.76' 10.94 3.10 58 35 50 153 61 10 53 272 6 21 April 2, 1988 1043:00.40 37ø36.51' 122022.37' 10.84 2.30 102 0 65 250 54 220 45 123 6 1957 March22, 1957 1944:20.86 37ø40.27 122029.05' 7.27 5.30 31 315 46 91 53 301 66 202 4

depth distributionof seismicityis shownin the crosssections in Both the seismicityand focal mechanismdata suggesttwo Plate 4, in which events that occurred before and after the 1989 discrete and distinct zones of deformation at the northern and Loma Prieta earthquake are depicted with different colors southernends of the 100-kmSan Franciscopeninsula section (locationof sectionlines are givenin Plates5 and 6). Com- of the SanAndreas fault examined.These zones are separated pared with the routine NCSN processinglocations, the relo- on the central peninsulaby a ---15-km-longzone of seismic catedevents show the sameoverall pattern as observedin the quiescenceand are discussedindividually below. standardprocessing, but seismicityis generallymore tightly clustered. In the northern subarea the relocation resulted in a 4.1. Northernmost Peninsula and Golden Gate Platform 0.5- to 1.0-km shift in epicenterstoward the NE; in the north- Epicentersin the northernzone of seismicityoccur in a 3- to ern third of the southern subarea the relocated events shift 5-km-widezone west of the mappedSan Andreasfault on- <500 m to the NE, while in the southern two thirds of the shoreand its newlydefined location and geometryoffshore. In southern subarea the relocated events were shifted 0.5-1.0 km cross-sectionalview (Plate 4a) the breadthof the seismiczone to the SW. and its apparentwestward offset from the San Andreasfault Focal mechanismswere computedfor all M d -> 3.0 relo- canbe seento persistthroughout the entireseismogenic crust, cated events(and a few selectedsmaller eventsas discussed from 3-kmto about13-km depth. Perhaps ---600 m of offsetof below)using only first motion picks from NCSN hand process- seismicitySW from the SanAndreas fault maybe attributedto ing (asopposed to automatedcomputer picks) and newly com- systematiclocation errors related to the largevelocity contrast puted takeoff anglesbased on the revisedseismic velocity acrossthe San Andreasfault describedabove. However, since structure. Mechanisms and 95% confidence solutions were de- the focal mechanismdata indicatethat this seismicityis the terminedusing FPFIT, a leastsquares, grid searchprocedure resultof bothnormal and right-lateral strike-slip faulting on N which minimizes the number of discrepantfirst motions to NW trendingplanes (Figure 4 and Plate 5), manyof the [Reasenbergand Oppenheimer,1985]. The focal mechanisms earthquakesin thisnorthern zone of deformationmay actually for the Md _>3.0 earthquakesin the studyarea are plottedon be occurringon distinct normal faults within the Pilarcitos Plates5 and6 for the northernand southernsubareas, respec- block,rather than on the boundingstrike-slip faults. tively. Each solution is numbered and the focal mechanism Becauseof limited first motionsand azimuthalcoverage, parametersare tabulatedin Tables 1 and 2, and equal area particularlyfor the offshoreand older events,some of the focal projectionsshowing first motionsare givenin Figures4 and 5. mechanismsin thisnorthern area are poorlyconstrained (par- In some casestwo possible,distinct focal mechanismsstatisti- ticularly 1, 3, 8, and 13). Severalevents have first motions callyfit the dataequally well for a givenevent (no overlapin yieldingboth a normaland strike-slipmechanism that fit the 95% confidenceinterval of P and T axes).Both mechanisms data to within a 90% confidencelimit (events3, 6, and 7) and their fit to first motionsare shownin Figures4 and 5, althoughthe strike-slipmechanism in two of the cases(3 and althoughonly one solutionis plottedin Plates5 and 6 (the 7) hasslightly smaller errors. Several mechanisms (e.g., 9, 15, solutionwith the smalleststandard error). 16, 19, and 20) havecompressional arrivals spanning a 120ø- ZOBACK ET AL.' SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,729

710331 1703 860309 128 871230 109 931124 1518 880402 1043 Z= 9.23 M= 3.20 Z= 4.61 M= 3.00 Z=10.32 M= 3.20 Z=10.49 M= 3.00 Z=10.84 M= la o 5 9 15 21 ( !

710331 1703' 861102 346* 890212 1949 940102 311 570322 1944 Z= 9.23 M= 3.20 Z= 7.55 M= 3.00 Z= 4.84 M= 3.10 Z=10.91 M= 3.00 Z= 7.27 M= 5.,! lb 6 1 1 19

710725 913 861102 346 890212 1949 911119 207 Z= 7.46 M= 3.40 Z= 7.55 M= 3.00 Z= 1.78 M= 2.90 Z= 7.99 M= 2.40 2 6b 1 17

771101 406 870201 1753 890213 600 920216 1823 Z= 6.50 M= 3.10 Z= 4.69 M= 3.40 Z= 4.39 M= 3.00 Z= 7.84 M= 2.40 3a 7a 12 18

771101 406* 870201 1753' 890213 801 960325 1117 Z= 6.50 M= 3.10 Z= 4.69 M= 3.40 Z= 4.22 M= 3.60 Z= 7.58 M= 2.70 3b 7b 13 19

790428 44 870919 1833 891106 2337 960701 433 Z=13,68 M= 4.40 Z=10,58 M= 3.30 Z= 4.14 M= 3.10 Z=10.94 M= 3.10 4 8 14 2

Figure 4. Lower hemisphereequal area projectionsof focal mechanismswith first motion data for the northern area; locationof eventsis given in Plate 5. Dilational first motionsare shownby solid circles, compressionalfirst motions are givenby opencircles. An a or b afterthe focalmechanism number indicates that it is one of two possiblesolutions. Captions above each event give date (yearmonth day), time (hour minute),depth (Z, in kilometers),and durationmagnitude.

150ø range in azimuth, clearly establishingthat the normal eventsdown to M d - 1.5 (limit of reasonablefocal sphere faulting is real and not just an artifact of the poor station coverage).Significantly, no thrustmechanisms were identified coveragein this area. Mechanismi9 is for a recentM d = 2.7 anywherein the northern section,even in the vicinity of a event;it is includedbecause it is very well constrainedand is mappedlate Quaternarythrust, the Serrafault, on the NE side the onlymechanism in the southernmost10 km of seismicityin of the San Andreasfault (Figure 1 and Plate 5). the northern zone of seismicity. The closespatial proximity of strike-slipand normal faulting It is difficult to establish whether the normal- and strike- alsoexists for larger-magnitudeevents. The largestearthquake slip-faultingevents in the northern deformationzone have duringthe period of relativelydense instrumentation (post- distinctspatial-depth distributions. Both strike-slip and normal 1968) is a M d = 4.4 right-lateralstrike-slip event located on eventsoccur in closeproximity and similar depth ranges within the San Andreasfault at a depth of 12.9 _+1.0 km near Daly and directly to the southeastof the San Andreas stepover City in 1979 (event4 in Plate 5). As wasnoted previously, the region(Plate 5). Northeastof the stepoverthe mechanismsall first motion distributionpattern for this event indicatedre- shownormal faulting (8-13, 15, and 16), with the best con- fracted first motions due to velocitycontrasts across the San strainednormal faulting mechanisms(8, 9, 15, and 16) all Andreas. Excluding the obvious refracted first motions, located at depths of 10-11 km oppositethe Golden Gate. Uhrhammer[1981] obtainedthe well-constrained(<_+5 ø un- Althoughnot shown,focal mechanisms were computedfor all certaintyin nodalplane attitude)strike-slip solution with the 10,730 ZOBACK lET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

Table 2. Focal Mechanism Parameters for Southern Area

Dura- Plane 1 Plane 2 P Axis T Axis Lati- Longi- tion Number tude, tude, Depth, Magni- of First Azi- Azi- No. Date Time N W km tude Motions Strike Dip Strike Dip muth Plunge muth Plunge

1 Jan. 3, 1970 0251:58.20 37018.87' 12204.74' 6.42 3.70 26 120 40 313 51 37 5 267 82 2 Sept. 23, 1970 0451:27.84 37024.27' 122013.29' 7.24 3.50 25 150 25 319 65 52 20 220 69 3 March 23, 1992 0345:29.35 37023.42' 122013.96' 9.71 3.20 38 45 70 138 81 3 21 270 7 4a Sept. 26, 1973 0404:46.66 37018.62' 122017.98' 7.11 3.20 24 135 70 315 20 225 25 45 65 4b Sept. 26, 1973 0404:46.66 37018.62' 122017.98' 7.11 3.20 24 190 85 289 30 255 33 129 42 5 Nov. 12, 1973 1817:13.48 37014.25' 121058.22' 12.40 4.50 57 215 85 309 50 164 31 269 23 6 March 12, 1974 1245:28.47 37ø20.18' 122ø15.01' 10.55 3.60 39 165 55 300 45 234 6 132 65 7 March 24, 1974 0137:15.39 37ø8.91' 122017.83' 14.54 3.10 34 345 75 250 71 208 25 117 3 8 June 19, 1975 1617:53.80 37022.72 ' 122015.66 ' 9.88 3.40 39 105 60 304 31 202 14 350 73 9 June 20, 1975 0534:00.89 37ø22.61 ' 122015.33 ' 9.55 3.30 36 100 55 325 45 211 6 313 65 10 June 20, 1975 0816:18.10 37022.32 ' 122ø15.41 ' 10.55 3.40 40 305 70 193 44 63 15 172 49 11a June 22, 1975 0012:35.44 37022.39 ' 122015.37 ' 10.27 3.50 27 200 85 101 30 81 42 315 33 11b June 22, 1975 0012:35.44 37022.39 ' 122015.37 ' 10.27 3.50 27 55 5 215 85 123 50 307 40 12 July 27, 1977 2151:17.55 37019.92 ' 12206.85 ' 4.97 3.50 93 120 50 270 44 196 3 95 74 13 Dec. 12, 1981 1511:09.18 37023.36 ' 122015.64 ' 7.95 3.70 39 95 60 334 48 212 7 312 55 14 May 14, 1986 0030:09.48 37022.59 ' 122ø15.12 ' 11.34 3.20 36 115 35 319 57 39 11 265 74 15 June 12, 1987 0543:28.90 37018.44 ' 121052.32 ' 10.27 3.10 59 85 65 328 46 203 11 307 52 16 Nov. 14, 1988 1921:47.37 37022.78 ' 122015.46 ' 9.75 3.60 58 125 50 290 41 208 5 88 81 17 Nov. 14, 1988 1952:58.26 37022.79 ' 122015.63 ' 9.75 3.40 50 125 50 305 40 215 5 35 85 18 May 9, 1989 1855:40.17 37023.50 ' 122010.23 ' 4.61 3.00 65 175 50 302 54 58 2 152 60 19a Oct. 18, 1989 0041:23.82 37ø11.13 ' 12203.73 ' 15.44 5.10 23 350 75 257 80 213 18 304 4 19b Oct. 18, 1989 0041:23.82 37ø11.13 ' 12203.73 ' 15.44 5.10 23 150 40 317 51 53 5 183 82 20 Oct. 18, 1989 0108:08.98 37011.70 ' 12202.88 ' 13.96 3.60 23 75 90 345 70 208 14 302 14 21 Oct. 18, 1989 0418:16.12 37010.66 ' 12203.44 ' 13.97 3.10 29 165 75 260 71 213 3 122 25 22 Oct. 20, 1989 0812:53.94 37011.08 ' 12203.83 ' 14.71 4.00 84 170 75 265 71 218 3 127 25 23 Nov. 7, 1989 2342:37.37 37013.94 ' 12201.74 ' 9.22 4.20 99 135 45 288 48 31 2 128 76 24 Nov. 16, 1989 0459:29.05 37ø11.35' 12202.63 ' 11.58 3.20 56 145 55 269 51 208 2 114 58 25 Dec. 2, 1989 2002:00.49 37014.04 ' 12201.89 ' 9.08 3.90 76 130 40 285 53 26 7 143 76 26 Dec. 26, 1989 2246:41.68 37011.79 ' 12203.20 ' 14.02 3.50 54 80 90 350 90 35 0 125 0 27 Dec. 27, 1989 1610:01.30 37011.70 ' 12203.46 ' 14.74 3.50 65 85 85 355 90 40 4 310 4 28 Jan. 22, 1990 0601:10.55 37019.57 ' 122ø7.51 ' 5.78 3.00 67 90 10 270 80 360 35 180 55 29 Aug. 22, 1990 2124:05.59 37012.49 ' 122ø4.16 ' 12.23 3.60 70 160 45 290 57 43 7 146 62 30 Sept. 23, 1990 1335:47.15 37022.87 ' 122010.64 ' 7.17 3.20 74 265 80 357 80 221 14 131 0 31a Sept. 26, 1990 0253:55.49 37022.97 ' 122010.67 ' 7.03 3.50 63 140 55 288 40 216 8 101 72 31b Sept. 26, 1990 0253:55.49 37022.97 ' 122010.67 ' 7.03 3.50 63 85 40 322 66 29 14 276 56 32 April 16, 1991 0056:59.36 37018.94 ' 12206.74 ' 4.75 3.10 54 115 45 295 45 25 0 281 90 33 June 28, 1991 0926:50.02 3705.62 ' 122019.99 ' 13.80 3.60 70 145 40 325 50 55 5 235 85 34 Sept. 12, 1991 0810:02.46 37ø11.16 ' 12204.42 ' 11.10 3.10 40 125 55 288 36 208 9 69 77

right-lateralplane strikingwithin 2ø of the mapped surface overall level of microseismicityin this area did statistically trace of the San Andreas.However, first motionsfor the larg- increaseover the backgroundin the 20 monthsfollowing the est past-1906earthquake along the SanAndreas fault in north- Lama Prieta earthquake[Reasenberg and Simpson,1992]. ern Californiabefore Lama Prieta, the M L = 5.3 1957 Daly Two notableclusters of seismicactivity occur on the north- City earthquake(labeled 1957 in Figure 4 and Plate 5), are ern San FranciscoPeninsula away from the San Andreasfault. generallyconsistent with other normalfaulting mechanisms in One clusteroccurs along the marginof San FranciscoBay in the northernsection, notably event 14 (Figure 4). A San An- the vicinity of San Franciscoairport (SFO), and the other dreas fault-style right-lateral strike-slipmechanism (with a occurs SW of the San Andreas fault on the Pacific Coast near N30ø-35øWstriking vertical plane) is excludedby the 1957first Point San Pedro (Plate 3). Persistentmicroseismicity in the motions (for example, compare the 1957 mechanismwith vicinityof SFO has occurredsince the onsetof good instru- mechanism4 in Figure 9, and note particularlythat first mo- mental coverageof the San FranciscoBay area in the late tionsfor localstations in the south(shallow takeoff angles) are 1960s[Olson and Zoback,1998]. These events have M d _<2.2, the wrong polarity if the 1957 event was a strike-slipearth- and the bestlocations indicate activity in primarilythe depth quake).Relative relacatian of the 1957 earthquakeusing the range 9-11 km. Mechanism21 is a compositemechanism for 1979 earthquakeas a masterevent places the 1957 event -600 sixevents (Md = 1.5-2.2) over a 5-yearperiod and indicates m to the NE of the mappedtrace of the San Andreasfault at combinednormal and strike-slipmovement on planesstriking a depth of about 6 km [Marsdenet al., 1995]. approximatelyN-S and E-W. Olsonand Zoback [1998] con- Seismicityhas occurred rather uniformlywith time through- cludethat this activityappears to be part of ongoinglow level out this northern zone of seismicity,although clusters of 5-10 deformationof the San FranciscoBay block and cannot be eventswithin a few daysor weeksare not uncommon,partic- related to any specificsurface faulting. ularly in the offshoreareas. There is no discernibledifference Seismicityin the Point San Pedro area occursin a NNW in either the depth or spatialdistribution of seismicitybefore trendingcluster about 3.5 km longby 2.0 km wide. Fifty-seven and after the 1989 Lama Prieta earthquake.However, the Ma _> 1.0 eventshave occurredhere since 1969, the largest ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,731

700103 251 740312 1245 750622 12' 881114 1952 891020 812 900122 601 91O628 926 Z= 6.42 M= 3.70 Z=10,55 M= 3.60 Z=10.27 M= 3,50 Z= 9.75 M= 3.40 Z=15,69 M= 4.00 Z= 5.61 M= 3.00 Z=13.80 M= 3.• 1 3 11 1 22 2i•..-•

700923 451 740324 137 770727 2151 890509 1855 891107 2342 900822 2124 910912 809 Z= 7.61 M= 3,50 Z=14.54 M= 3.10 Z= 4.70 M= 3,50 Z= 5.52 M= 3.00 Z= 9,58 M= 4,20 Z=13.15 M= 3,60 Z=12,72 M= 3,1 1 • 1 23 2 3

e.T

720323 345 750619 1617 811212 1511 891018 41 Z= 9.94 M= 3.20 Z= 9.73 M= 3.40 Z= 7,95 M= 3,70 Z=15.44 M= 5,10 Z=12.16891116 M=459 3.20 Z=900923 7.65 M=1335 3.20 P-/T-axes 3 13 19a

24 Ooß 3

730926 404 750620 534 860514 30 891018 41' 891202 2002 900926 253 Z= 7,23 M= 3,20 Z= 9.26 M= 3,30 Z=11.34 M= 3.20 Z=15,44 M= 5.10 Z= 9.08 M= 3,90 Z= 7.03 M= 3,50 4a 1 191 25 31

730926 404* 750620 816 870612 543 891018 108 891226 2246 900926 253* Z= 7,23 M= 3,20 Z=10.55 M= 3,40 Z=10,27 M= 3,10 Z=13,96 M= 3,60 Z=14.02 M= 3,50 Z= 7.03 M= 3.50 4b 1 1 2 26 31

731112 1817 750622 12 881114 1921 891018 418 891227 1610 910416 56 Z=13.50 M= 4,50 Z=10,27 M= 3,50 Z= 9,75 M= 3.60 Z=13.97 M= 3,10 Z=14.74 M= 3,50 Z= 4.77 M= 3,10 5 11 1 2 •, 2 3

Figure 5. Lower hemisphereequal area projectionsof focal mechanismswith first motion data for the southernarea; location of eventsgiven in Plate 6. An a or b after the focalmechanism number indicates that it is one of two possiblesolutions. See Figure4 for further explanation.Final lower hemisphereprojection showsP and T axes(solid and opencircles, respectively) for all mechanismsand demonstratesthe predom- inanceof NE (fault-normal)compression.

eventbeing Ma - 2.5, howevermost eventsare in the range cantlymore westerlythan the N25ø-28øWstrike of the nodal Ma = 1.0-2.0. This region exhibitsswarm-like activity, with planes. •y of *'-- 57 I•UIU•U...... •-• events occurrin LII• .o .*u.t.• uc- 4.2. Southern Peninsula •een April 1991and July 1992and with 33 of those49 events occurringin a 2-weekperiod in November1991. Focal depths On the southern San FranciscoPeninsula, seismicity occurs of the Point San Pedro eventsrange be•een 6.0 and 8.0 km. in 7- to 10-km-wide bands on both sides of the San Andreas Focal mechanismsfor the •o largestevents (17 and 18) indi- fault, beneath mapped subparallel Quaternary folds and cate dominantly right-lateral strike-slip faulting on NNW thrustswithin the Coast Rangesfault (Plates3 and 4) [Page, trendingplans. Hypocentral locations suggest these events are 1982,1992; Aydin and Page,1984; Hitchcock et al., 1994].Focal probablynot relatedto the SanGregorio fault. Their locations mechanismsindicate that this seismicityis due predominantly •4 km NE of the easternmostmapped strand of the San to thrust or reversefaulting on planesstriking subparallel to Gregoriofault zoneand at depthsof 8 km implyan averagedip the San Andreasfault (Plate 6). Thesemechanisms suggest a of about 63ø NE for this fault, shallower than the 75ø-81 ø NE general pattern of fault-normal (NE) compressionin the dipsof the NNW trendingnodal planesof events17 and 18. crustalzones adjacent to the fault (seestereoplot of the P and Alternately,this seismici• couldbe associatedwith the Pilar- T axesshown in Figure5) consistentwith the surfacegeology. citos fault; however, the N55øW strike of that fault is signifi- On crosssections S1, S2, and particularlyS3 a vertical qui- 10,732 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

a) N1 SGF SAF b)$1 PFSAF

/ o/ o" iø t : _ oo ø -10.- ...... ,-'• ..... •0

o -15.'...... O, • 10. o. 5, lO. 15. 20.

N2 SGF SAF S, AF S2

o oo•O øO

o o

-10:'

O. $. 10. O. 5. 10. 15. 20. N3SGF PF SAF s3 SAF MVF 0.

ß " ! ' øooO

"-' ,' : __•_'•'.1 •0 ! .• •.'• ' : . I to,,• . .• -10. -..... ) •-- o• •...... • o o OJ

-15. , , .' '•...... o. 5, lo. 15,

O. 5, 10. Plate 4. Crosssections of seismicityin the (a) northernand (b) southernzones of seismicity.Earthquakes occurringprior to the 1989M = 7.1 LomaPrieta earthquake are shownby yellowcircles; post-Loma Prieta eventsare shownin red. Symbolsize is scaledto magnitude.Locations of cross-sectionprofiles are givenin Plates5 and6. Data wereprojected from 6 km NW andSE of sectionlines. Triangles indicate surface location of faults:SAF, SanAndreas fault; SGF, SanGregorio fault; and PF, Pilarcitosfault. Green line corresponds to locationof a verticalSan Andreas fault belowsurface trace. The blue line in sectionS3 givessubsurface geometryof the Monte Vista thrustfault (MVF) constrainedto a depthof --•3.0km (solidline) by well data and gravitymodeling [Langenheim et al., 1997]. ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,733

1 2'40 122'30' 122'20'

,,. . \o o•. -.

37'50' .,. -. 1. o 37'50' ; 10,11 N Gok•n Gate 0 8,g• ' Platform o 15,15 "• ' o• ß .,.;.øo ß

o •. cB'o'

'1::•, 0 1906

O0 ß

37'40' 0o.•_ o +- 37'40' •o,.?.,• 14 '• -• o

O '

* • • '• • ,.• •* • e, 0 0 _ ß 17,1: '•. o © o o

,,,• P 0

10 KM

% • 0

122'40' 122'30' 122'20'

-80 .•0.... 1• 4'0 '" 8•) .....1'•0 .... i'60 2• .....2•0 ...... 21•) 3•0 meters Plate 5. Seismicityand focal mechanisms in northernmostpeninsula and GoldenGate platformarea. Events are Ma -> 3.0 eventsexcept for a few smallerevents (see text). Event numbersrefer to Table 1 and to first motionplots shown in Figure4. New offshoregeometry of the SanAndreas and San Gregoriofaults shown by heavyred lines (Plate 2). Thin red lines are Quaternaryfaults from Bortugnoet al. [1992] (note the northernmostsegment of the Serra reversefault superimposedon the green focal mechanismfor the 1957 earthquake).The heavyblack dashed line marksthe locationof the "SanAndreas" graben of Cooper[1973], from Plate 2. Thin black lines mark locations of cross sections shown in Plate 4a. 10,734 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

122'20' 122'10' 122' 121'50'

37'30' 37'30

0

o • MenloPark

• o o Nod•plane dip towards San An as •1la e •18 o •13-14•• . ••o• o 16-170 ••%30½1ag m• -•• m• o • 3 O IgSunnyvale o o 8••. o•.• o••, ? 0 •SanJose

37'20' 37'20' oo 0 oS1o •

37'10' 37'10'

121 '50'

-260 -100 0 1(•)...... 200 300 400 500 600 700 800 meters Plate 6. Seismicityand focal mechanismsof M,• >- 3.0 eventsin the southernsubarea. Event numbersrefer to Table 2 and to first motion plotsshown in Figure 5. Red linesare Quaternaryfaults from Bortugnoet al. [ 1992],dashed red line on NE sideof SanAndreas between sections S2 and S3 representsthe buriedextension of the Monte Vista fault zone basedon gravitymodeling of Langenheimet al. [1997]. Surfaceprojection of Loma Prieta rupture plane determinedfrom geodeticmodeling [Lisowski et al., 1997]indicated by dark blue dashedrectangle at bottom right.

escent zone 1.5-2.0 km wide coincides with the surface trace of dreasfault zone, then the dip of the fault is constrainedto be the San Andreas fault and is defined between ---3-km and >---85ø to depthsof 10 km. 12-km depth, most of the seismogenicpart of the crust.The The microseismicitydrops off between7 and 10 km from the only significant activity located on the San Andreas fault San Andreasfault but increasesagain to the SW in the vicinity proper in this southern zone of seismicityis the March 23, of the San Gregoriofault. In crosssection on both sidesof the 1972, M,• - 3.2 event (event 3 in Plate 6) and its aftershocks San Andreas there is a deepeningof seismicitytowards the located south of Woodside and shown on cross section S1 fault (seeS1 to S3 in Plate 4b and compositesection, Figure 6) (between depthsof 8 and 10 km). Both the location and a suggestiveof a large-scaleflower structure in whichmoderately right-lateral strike-slipfocal mechanismfor this event suggest dippingthrusts at the surfacesteepen with depthto mergewith that the SanAndreas fault is approximatelyvertical to a depth a broadvertical shear zone [e.g.,Sylvester, 1988]. In fact, NE of of 10 km in this region.An approximatelyvertical attitude at the San Andreasfault there is a seriesof mappedsubparallel depth can also be inferred from the 1.25-km-wide"gap" in thrust faults that consistentlydip to the SW, into the San seismicityat ---10-kmdepth on sectionS3 (Plate 4b). If thisgap Andreas. However, on the SW side of the San Andreas fault is interpretedas the maximumpossible width of the San An- the structureis more complexand inconsistentwith a simple ZOBACK ET AL.' SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,735

N40-45ow SAF BerrocaI-Monte Vista-Stanford sw frontalthrust zone NE

-5.

-15. #= right-lateral onNWplane •= right-lateral onNplane finite width ductile shear zone •-' '/=thrustonNW plane N34.5øW right-lateral shear? -10. -5. o. 5. 10. Horizontal Distance from the San Andreas fault (kin)

Figure 6. Interpretativecross section (composite from S1, S2, S3) acrossthe San Andreas fault on the southernSan FranciscoPeninsula (topography along S2 shownin 2x verticalexaggeration). Wavy vertical linesdenote "locked" aseismic San Andreas fault zoneat depth.San Andreasward dips of nodalplanes for Md >--3.0 "fault-normalcompression" thrust/reverse events are shownby heavysolid lines NE of San Andreas. Short dashedlines representdip of north strikingright-lateral strike-slip planes, also consistentwith fault normalcompression. The heavyruled line on the SW sideof SanAndreas fault givesdip of right-lateralplane for San Andreas event 3. The grey long-dashedline NE of San Andreas representsdip of right-lateral strike-slipplane strikingparallel to San Andreas, event 15.

flowerstructure. Here muchof the compressionaldeformation reversefault planesinferred from focal mechanismsare con- at the surfaceis accommodatedby synclinesand anticlines sistentwith the 350-70ø range of surfacefault dips [Sorgand strikingslightly oblique (10ø-15 ø counterclockwise)to the San McLaughlin, 1975; McLaughlin and Clark, 1999] and suggest Andreas. The one mapped late Quaternaryfault on the SW that generallyplanar faults with little or no steepeningwith sideof the SanAndreas, the Butanofault, actuallydips steeply depth persist throughout the seismogeniccrust. The close awayfrom the San Andreas(>70øSW) at the surface[Cum- proximityof some of the thrust/reverseplanes to the vertical mings et al., 1962]. Seismicityon the southernmostcross- "aseismic"zone related to the main SanAndreas fault at depth sectionS3 (primarilyLoma Prietaaftershocks) appears to de- suggestthat these fault planes intersect or merge at a high lineate a steepSW-dipping zone directlyadjacent to the San anglewith the San Andreas.Both planar faults and the high- Andreas.The depth and geometryof this zone is similarto the angleintersection geometry at depth are inconsistentwith sim- geometryof the Loma Prieta mainshockrupture plane, the ple flower structuremodels of thrust/reversefault planessteep- northernend of whichis locatedjust 5 km to the south(surface eningwith depth to mergewith the strike-slipfault zone. projectionof rupture plane shownby dashedrectangle on The focal mechanism data indicate an additional mode of Figure i0). tanure• .1 in the crustalzone u,rccuy1 ß _•1_ adjacent to the San Anm.,1 .... ca• We used focal mechanismsto constrainthe subsurfacege- fault. Focal mechanismsof some the deepestevents seen on ometry of the thrust faulting on the NE side of the fault by crosssection S3 (primarilyLoma Prieta aftershocks)indicate selectingthe NW-striking,SW-dipping nodal plane as the fault right-lateralstrike-slip faulting on steepnorth strikingplanes plane,consistent with the surfacefaulting. The inferreddips at (events20-22, 26, and 27, planesmarked by smalldashes on a depth (heavysolid lines on the compositecross-section across compositecross section shown in Figure 6). One pre-Loma the fault in Figure 6 and histograminset in Plate 6) range Prieta event(event 7) on the NE sideof the SanAndreas also between 40 ø and 55 ø SW for events between 4.7- and 9.5-km had a north striking right-lateral mechanism.This style of depth (exceptevent 2, with a dip of 25ø at 7.6-km depth). faultingis consistentwith NE-compression(fault-normal) re- Becauseof the structuralcomplexity on the SW sideof the San sponsiblefor the thrust/reversefaults, but hasno expressionin Andreas the actual fault plane could not be inferred, so both surfacegeology. possiblenodal planes for the thrust/reverseevents are shownin The largest,pre-Loma Prieta, off-SanAndreas earthquake Figure6 (the thin solidlines). These events also indicate gen- in the southernzone of seismicityis a Ma = 4.5 event in 1973 erallymoderate dips (35ø-60ø). The moderatelydipping thrust/ locatedabout 7 km NE of the San Andreas at a depth of 13.5 10,736 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

km (event5, Plate 6). It doesnot fit the generalpattern of fault lends supportto the conclusionbased on surfacedefor- thrust/reversefaulting in responseto fault-normalcompres- mation and geodeticdata [Prenticeand Schwartz,1991; Segall sion.The mechanismfor this event is not well constrained,but and Lisowski,1990] that the Loma Prieta earthquakefarther indicatesnearly pure right-lateral slip on a fault planestriking southoccurred on a distinctfault plane, not on the San An- subparallelto the SanAndreas fault and dipping50øNE, away dreas.Furthermore, the contrastbetween the high levelsof from the San Andreasfault and oppositeto the dips of the Loma Prieta-triggeredseismicity within the blocksadjacent to surfacethrusts (plane shown by largedashes in Figure6). The the San Andreasfault and the lack of triggeredseismicity lackof pre- or post-LomaPrieta microseismicity in the vicinity directlyon the fault (Figure6 and Plate 4b) suggeststhat the of thislarge event on S3 (Plate 4b) doesnot allowdelineation Coulombfailure strengthof the San Andreasfault is greater of a possiblelarger scalefault on which this M = 4.5 event thanthat of the surroundingcrustal blocks in thisearly portion might have occurred. of the "locked"San Andreasseismic cycle. This high relative A prominent cluster of seismicity2-3 km SW of the San strengthmay be due to high normal stressacross the San Andreasfault in the northernpart of the southernarea shown Andreasresulting from regionalNE compressionand insuffi- on sectionSI (Plate 4b) fallsbeneath the mappedtrace of the cient pore pressurebuildup within the fault zone to counteract Pilarcitosfault. Theseevents occur at depthsof 9-11 km and it. If a systemof pore pressurecells and sealsexists within the yield thrust/reversemechanisms with dips of 31ø-57ø NE for fault zone and is responsiblefor an averageoverall low shear the planedipping towards the San Andreasfault (Figure6). strengthof the SanAndreas fault [Blanpiedet al., 1992;Byeflee, Thus, despitehypocentral locations nearly directly below the 1993;Rice, 1992; Sleep and Blanpied, 1992a, b], then seismicity surfacetrace of the Pilarcitosfault, it is unlikely,given their and focal mechanismdata presentedhere suggestthat these dips,that theyrepresent activity on this approximatelyvertical cellsand their lateral sealsare probablyrestricted to a --•1- to fault as modelledby seismic[Parsons and Zoback,1997] and 1.5-km-widezone at depth. In addition,the concentrationof magnetic[Jachens and Zoback, 1999] data on the northern fault normal deformation in a 7- to 10-km-wide zones on either Peninsula.Unpublished gravity modeling by oneof us(R.C.J.) side of the San Andreas suggeststhat ductile deformationat alsoindicates an approximatelyvertical dip for the Pilarcitos depth may be more distributedover a zone beneath the fault, fault just 10 km north of SI. as is indicatedin Figure 6. Fault-normalcompression in the southernSan Francisco Peninsula is enhancedby the obliquity between the N40ø-45øW striking San Andreas and a N34ø- 5. Tectonicinterpretation and Implications for 35øWoriented plate motion shear.Alternately, the width of Seismic Hazard deformationadjacent to the San Andreasmay be simplya resultof crustalweakening due to more extensivefaulting as 5.1. Strike-Slip Faulting Deformation the blocks on either side accomodate deformation due to local Despite the persistentmicroseismicity in the San Francisco bendsin the fault system. Peninsulaarea, both of the two major activestrike-slip fault Althoughthe presentlack of seismicityon the San Andreas zonesin the studyarea, the San Andreasand San Gregorio andSan Gregorio faults provides no informationon the timing faults,are largelyaseismic during the studyperiod. Geologic of a futureevent, the seismicityand faulting patterns defined in data give ample evidenceof Holoceneslip on both the San this studymay be usefulfor definingrupture segment bound- Andreasand San Gregorio faults and indicateslip rates of aries, particularlyfor the San Andreas fault. Two types of 17-19and --•5 mm yr -1, respectively[Working Group Report for major earthquakesare consideredlikely on the San Andreas NorthernCalifornia Earthquake Potential, 1996]; therefore both fault in the studyarea, a repeat of the M, = 7.8 1906 earth- faults are consideredto be in a "locked"portion of their quake and a repeat of the 1838 M, = 6.8-7.5 event on the seismiccycle. A third strike-slipfault locatedbetween the two, peninsula[Working Group Report on CaliforniaEarthquake the Pilarcitosfault, is alsoaseismic. As wasnoted previously, Probabilities,1990; WorkingGroup on Northern California the Pilarcitos fault has been interpreted as an abandoned EarthquakePotential, 1996]. The normalfaulting and abrupt strike-slipportion of the plate boundaryon the basisof geo- right step("releasing bend") offshore from SanFrancisco de- logic relationships[Page, 1990; Griscomand Jachens,1989]. scribedhere supportsa likely segmentboundary for a M, • Furthermore,the Pilarcitosfault has Quaternarybut no Ho- 7.0 strike-slipevent on the peninsulaas suggestedby the loceneoffset [Bortugno et al., 1992].These facts, together with WorkingGroup on NorthernCalifornia Earthquake Potential its more westerly strike than the San Andreas, lead us to [1996]largely on the basisof Thatcheret al.'s [1997]geodeti- suggestthat its seismicpotential is very low. cally interpreted--•2-m reduction of 1906 slip and "bend" in The aseismiccharacter of the San Andreasfault persiststo the SanAndreas occurring on the GoldenGate platform.The the southernmostportion of the studyarea, directlynorth of north end of rupturein the 1838 earthquakeis unknown,but the LomaPrieta rupture (section S3, Plate 4b), despitethe fact highintensities at MissionDelores in SanFrancisco lead Top- that this portion of the fault was subjectedto some of the pozadaand Borchardt[1998] to locatethe northernend of the largest computed increasesin Coulomb failure stressin the rupture near San Francisco.Further supportfor the offshore Bay area as a resultof that M, = 6.9 event [Simpson,1994; right-stepas a San Andreassegment boundary comes from Simpsonand Reasenberg,1994]. Previously,Reasenberg and observationsof major strike-slipearthquake ruptures that ter- Simpsoh's[1992] analysis of post-LomaPrieta seismicityrate minate near releasingbends such as on the North Anatolian, changessuggested seismicity increases along the SanAndreas southernSan Andreas,and other fault systems[Sibson, 1985; fault in this region;however, their 10-km-widecells along the Wesnousky,1988; Barka and Kadinsky-Cade,1988]. San Andreasfault really sampledthe fault-normalcompres- The 1906earthquake indicates that the rightstepover region sion-relatedseismicity occurring in the crustdirectly adjacent doesnot alwaysact asa segmentboundary. This event not only to the fault. rupturedthrough the offshoreregion but may have nucleated The lack of triggeredslip on this sectionof the SanAndreas there as well. The best estimatefor the epicenterof the 1906 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,737 earthquake based on teleseismicdata (and consistentwith Gate platform.Cooper [1973] imaged a ---4-to 8-km-widebasin localtiming observations and stoppingof astronomicalclocks) betweenthe San Andreas and the San Gregorio faults.Hole et [Bolt, 1968] is <5 km (_+---10km [Boom,1977]) southof the al. [1996] found a maximumdepth of 2 km for this basin.As newlydefined right stepin the SanAndreas fault offshore.The was noted earlier, Cooper [1973] also delineated a distinct, proximityof the 1906 epicenterto the right stepoveroffshore somewhatsinuous 20-km-long, 2- to 3-km-wide"San Andreas" suggeststhat this structuremay have exerted some control on grabensitting on top of the broaderbasin; this youngest graben localizingthe nucleationof that eventand be responsiblefor its is shownby dashedlines in Plates 2 and 5. bilateral rupture character.The 1995Kobe earthquakeepicen- A Pliocene-Quaternarybasin ---3 km wide is exposedon land ter also occurredin a 5-km right step in a right-lateral strike- directly northeast of the San Andreas fault on the northern- slip fault system[WaM, 1996]. In that case the two offset most San FranciscoPeninsula. This sequenceof ---3.0 Ma to rupture planes,defined by aftershocklocations, surface fault- 0.4 Ma shallow marine to estuarine depositsknown as the ing, and geodeticmodeling, had slightdips toward each other Merced formation [Ingram, 1992] may be an older equivalent (---85ø),and the hypocenteroccurred at depth near the inter- of the active pull-apart basin offshore.Jachens and Zoback sectionof these two planesin the stepoverregion. [1999]have interpretated detailed gravity data on the northern Redefinitionof faultingpatterns offshore have implications peninsulato indicate a 2- to 3-km-wide, SE trending trough for the shakinghazard as well. The offshore3-km right stepin filled locallywith more than a kilometer of youngdeposits, and the SanAndreas fault resultsin relocationof the main faulting boundedon the southwestby the onshoreSan Andreas fault strand 3 km closer to San Francisco. This NE shift in the locus and on the northeastby the onshoreextension of the right step of faultingmay be significantfor groundmotion in the western strand. The trough coincidesclosely with the narrow belt of portion of the city. Reducingthe perpendiculardistance from outcroppingMerced formation and shallowsgradually to the the fault from 5 to 2 km would result in an increase of MM southeastover a distanceof---10 km, just as the Merced for- intensityby about one unit, or roughly double the level of mationthins to the southeast.Hengesh and Wakabayashi[1995] groundmotion [Boatwrightand Perkins,1999]. argued that the Merced formation was depositedin a marine basindeveloped within a pull-apartstructure that hasmigrated 5.2. Extensional Deformation: Northernmost Peninsula with the Pacific plate (and currently lies offshore from the and Golden Gate Platform Golden Gate), an interpretation that is consistentwith the Normal faulting,subsidence, and young basin formation off- geophysicaldata. shoreis likely related to the right-steppinggeometry of the San The Mercedsedimentary trough, the basinsobserved in the Andreasand San Gregoriofaults on the Golden Gate platform offshore seismic reflection data, the 1957 M = 5.3 normal and probablyexplains why thesefaults and the crustalwedge faulting earthquake, and the ---30-km-longzone of normal between them lies below sea level. Nodal planes of the well- faultingfocal mechanismssuggest active normal faulting strik- constrained normal faulting mechanismssuggest potential ing subparallelto the San Andreas and extendingto at least fault planesstriking between NNE and NNW, subparallelto 11-km depth. However, the availabledata are insufficientto the strike-slipfaults; however,the low level of seismicityand determine whether most of the extension occurs on one or small number of well-constrainedmechanisms make it impos- more through-going,basin-bounding faults capableof gener- sible to determine if all of this seismicityoccurs on a discrete ating surface-rupturingearthquakes on their own, or if the normal fault zone or zones offshore. Some of the best con- extensionoccurs on distributed small-scalenormal faulting strainednormal faulting eventsoccur at depthsof 10-11 km, with slip triggeredby large strike-slipfaulting earthquakeson suggestinglarger-scale extension (and potentially larger nor- either the San Andreas or the San Gregorio fault. While the mal faults) than would be requiredto simplyaccomodate nor- maximummagnitude of a potential normal fault based on a mal faultingwithin the 3-km San Andreasfault stepover.Nor- fault length of 25-30 km is only 6.0-6.5 [Scholzet al., 1986], mal faults restricted to the 3-km stepover region should proximityof suchfaulting to urban San Francisco(3-10 km intersectat depthsof--• 1.5-2.5 km (assumingdips between 45 ø away) increasesthe importanceof this potential hazard. and 60ø). Further indication of distributed normal faulting in the 5.3. Thrust/Reverse Deformation: Northernmost Peninsula northern section comes from the occurrence of well- The 3.0-0.4 Ma shallow marine Merced formation has been constrainednormal faultingevents up to 15 km NW and 18 km uplifted up to 200 m and compressionallydeformed directly to SE of the actual San Andreas stepover.Master event reloca- the northeast of the San Andreas fault on the northernmost tion for the 1957 M = 5.3 normal faulting event (using the San FranciscoPeninsula. This unit is folded (with dips as high i9/9 3/ = 4.4 San Andreasearthquake as the master event) as 70ø ) and is cut by the Serra thrust/reversefault which is placesthe 1957 earthquake5-6 km southof the stepover.The exposedfor at least 15 km subparallelto and 1-2 km NE of the significantalong-strike extent of extensional tectonism NW SanAndreas fault (Plate 5) [Bonilla,1964, 1994, 1996]. Cooper and SE of the stepoverregion as well as the upper crustalscale [1973] imagedsimilar steeply folded bedsnortheast of the San of extensionare compatiblewith 3-D elasticmodels of defor- Andreas fault just a few kilometers offshore,which he inter- mation adjacentto weak strike-slipfault zones underlain by prets as a continuationof the compressionaldeformation of finite width shearzones [Katzman et al., 1995;ten Brink et al., the Merced observed on land. The youngest documented 1996].Katzman et al.'s modelsof the vertical subsidencein the movement on the Serra fault comes from trenches in which the vicinity of a dilational stepoverin a strike-slipfault demon- fault cuts the late Pleistocene(130-70 ka) Colma formation stratedthat a significantfraction (50%) of the maximumsub- [Hall, 1965, 1966].At one localitythe Serra fault displacesan sidence(and extension)observed in the stepovercan persist undatedA horizonsoil that may be Holocenein age [Bonilla, along strike for 2.5-4.0 times the stepoverdistance. 1994;Hengesh et al., 1996]. Geologicand seismicreflection data indicatethat extension Hengeshand Wakabayashi[1995] proposethat a changein and basin developmenthave been long-lived on the Golden deformationalregime from extensionto compressionimplied 10,738 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA by first deposition,then deformationand uplift of the Merced a convergencerate in the Santa Cruz Mountains region of formation is simply the result of effects of a migrating right between4 and20 mmyr -•. stepin the SanAndreas fault that moveswith the Pacificplate. Unfortunately, the microseismicityis so sparsethat it is Within the right step,extension results in deposition.After the difficultto relate it in detail to specificmapped surface faults. right steppasses by, the previouslyextended region is subjected Hypocentersof many of the eventson the NE side of the San to the regional fault normal compressionin its wake. This Andreas between sectionsS1 and S2 lie directly beneath the hypothesisis inconsistentwith the focal mechanismdata in the mappedfrontal thrustfault system(Figure 6). If the causative northern peninsula,which indicate a combinationof normal fault planes(with dipsin the rangeof 40-55 øat depthsof 5-9.5 and right lateral strike-slipfaulting at depth, evenbeneath the km) were projectedup-dip, their surfacetraces would locate very youngcompressional deformation of the Merced. As was beneaththe highly developedSanta Clara Valley, 3-5 km NE noted earlier, no thrust or reverse faulting mechanismshave of the mapped frontal Monte Vista-Berrocal thrust system been observedanywhere in the northern zone of seismicity, (Figures6 and Plate 6). Gravitygradients within SantaClara even for earthquakesas small as M a = 2.0. An alternate Valley have been interpretedas indicatingpossible NW trend- interpretationof the folding,faulting, and uplift of the Merced ing faultswith up to 130 m of vertical offset of bedrock[Fin- directly adjacentto the San Andreas fault is that they are due laysonet al., 1967], althoughvariations in Franciscanbedrock to a slight,local misalignment between the two offsetstrands of geologymay explainmany of the gradients. the San Andreasfault (---10ø in the immediatevicinity of the The best constrainedfault attitude at depth is that of the right step)resulting in convergencewith continuedmovement Monte Vista thrust fault, part of the Monte Vista-Berrocal on the San Andreas system[Jachens et al., 1996;Jachens and thrust system[Sorg and McLaughlin, 1975; McLaughlin and Zoback, 1999].The closeproximity of the Serrafault to the San Clark, in press].A newly extendednear-surface trace of the Andreasfault (<1.5 km) suggestsit mergeswith or intersects Monte Vista fault basedon geophysicaldata is shownby a red the San Andreas fault at a shallowdepth. Hence the Serra is dashedtrace in the regionbetween S2 and S3 in Plate 6. This probablynot a deep-seatedfault. This possibilityand the ob- thrustplaces Franciscan bedrock over late Cenozoicsediments servednormal faulting focal mechanismsleads us to propose at the surface.Geologic mapping [Sorg and McLaughlin,1975] that this young fault does not have primary seismogenicpo- and interpretationof gravityand aeromagneticdata [Langen- tential, although it may slip aseismicallyduring major San helmet al., 1997]indicate a shallowdip for thisfault (---17ø SW) Andreas events. in the upper 1.3 km which abruptlysteepens to about 600-70ø. Wakabayashiand Moores [1988] have suggested that the NW The geophysicalinterpretation constrains the fault geometry trending Pilarcitosfault representsa thrust fault dippingNE onlyin the upper 3 km. However,simple downdip projection of into the San Andreas fault; hence this fault would have a trend thisfault geometry(see blue line in Plate4b, sectionS3) passes similar to other thrust/reversefaults on the peninsula and throughsome of the seismicity(including Loma Prieta after- couldbe consideredpotentially active. However, as noted pre- shocks)at depthsof 5-7 km in sectionS3. Althoughthe pro- viously,a seismicreflection, seismic tomography and magnetic jected plane of the Monte Vista thrust fault passesnear the data on the northern peninsulaall indicate a approximately aftershock cluster that includes events 23 and 25, its inferred vertical dip for the Pilarcitosfault in the upper 6 km of the 70ø dip is significantlysteeper there than the 400-45ø SW dips crust [Parsonsand Zoback, 1997;Jachens and Zoback, 1999]. of these thrust events. Damage patterns resulting from the 1989 Loma Prieta earthquake were concentrated along 5.4. Thrust/Reverse Deformation: Southern Peninsula mappedthrusts on the peninsula(including the Monte Vista Geologic evidence, topograpy, and the earthquake focal fault). Langenheimet al. [1997] suggesta direct connection mechanismsindicates active thrust/reversefaulting within a betweenthe aftershocks(implying slip at depthon thesefaults) zone 7-10 km wide on either side of the San Andreas fault on and the surfacedamage, but this connectioncan not be dem- planes striking subparallelto the San Andreas. The overall onstratedin detail with the microseismicity. strike of the San Andreas fault in the southern section is Kovachand Beroza [1993] suggestedcontinuity of mapped ---N45øW,roughly 10ø-12 ø westward of the plate motionvector thrustsand microseismicityon the NE sideof the SanAndreas at this latitude [Argusand Gordon, 1991]. Geodeticdata indi- fault in the southernpeninsula as indicatingthe potential for catea total of 37-38 mm yr-• right-lateralslip in the N34ø- M = 6 + thrust event. In fact, a 60- to 70-km-long zone of 35øWplate motion directionover a ---100-km-wideregion cen- youngthrust faults has been mapped along the NE margin of tered on the Bay area and limit the overall convergent the CoastRanges from just southof the southernend of Plate (perpendicular)component to _<3mm yr-• [Lisowskiet al., 6 to just north of the town of Woodside(the Berrocal,Monte 1991; Williamset al., 1994]. However, in the area of the south- Vista, and Stanford fault zones) [Aydin and Page, 1984; ernmost San FranciscoPeninsula the average 10ø-12ø more McLaughlin,1974; McLaughlin and Clark, 1999]. Geomorphic westerlytrend of the San Andreas fault shouldlocally enhance evidence,including sag ponds, fault troughs,and streamdiver- thisconvergent component by ---6.5mm yr-•, assumingthat sions,suggests recent activity on these features [Rogersand the ---17mm yr-• right-lateralplate boundaryshear on the Williams, 1974; Hitchcocket al., 1994]. Our seismicitydata northernpeninsula is accomodatedthrough the bendby slip on indicate that this entire 60- to 70-km long zone of mapped a verticalSan Andreas fault (whichwas not, however,the case frontal thrusts in underlain by thrust faulting microearth- in Loma Prieta). Sinceonly a fractionof the regional_<3 mm quakes.Convergence and uplift estimatesalong this zone are yr-• convergenceis likely accomodated adjacent to the San not well defined,however; the availabledata indicatethey are Andreas (the rest being accomodatedin the East Bay [e.g., consistentat about---_<1 mm yr -• [Jaykoand Lewis, 1996; R. J. Graymetet al., 1994;Jayko and Lewis,1996]) the total conver- McLaughlin,oral communication,1998]. Thus at presentthere gence acrossthe Coast Ranges/SantaCruz Mountains in the is no evidencethat couldrule out the possibilitythat the entire southernmostpeninsula may be in the 6-8 mm yr-• range. 60- to 70-km zone couldfail in a singlecomplex event (possibly Usinggeologic constraints on uplift, Valensise[1994] estimated analogousto the 1952M w = 7.7 Kern County event in south- ZOBACK ET AL.' SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,739 ern California),although smaller thrust events along one of the Two other Quaternarystrike-slip fault zoneson the peninsula, frontal thrustzones (M • 6) may be more likely. the San Gregorio and the Pilarcitosfaults, are also currently Alternately, future thrust eventscould potentiallyoccur on aseismic.The San Andreas and San Gregorio faults have an unknown number of blind faults with no clear surface ex- provenpotential for M - 7 and greater events,whereas the pression.Support for this suggestioncomes not onlyfrom lack Pilarcitosfault appearsto havelow potentialfor slip and prob- of correlationbetween microseismicityand the simple down- ably representsan abandonedportion of the plate boundary. ward projectionof mappedfaults, but also from the lack of Our resultshave several implications for seismichazard. The delineationof the Loma Prieta fault zone in pre-Loma Prieta right-steppinggeometry defined offshorefor the San Andreas seismicity.Careful analysisof microseismicityprior to several lendsstrong support for a proposednorthern segment bound- recentmajor earthquakesin California has failed to delineate ary for a "peninsula"San Andreasrupture [WorkingGroup on the fault plane that ruptured (e.g., the 1983 M L = 6.7 NorthernCalifornia Earthquake Potential, 1996]. Although the Coalinga [Eaton and Rymer, 1990]; 1989 M w = 6.9 Loma 1906 earthquakeruptured through this offshoreregion, it may Prieta [Olsonand Hill, 1997];and 1994Mw - 6.7 Northridge have actuallynucleated within the right stepoverin the San [Haukssonet al., 1995]. Andreasfault, which may explainthe bilateral nature of rup- ture of this event. The 1995 Kobe earthquakenucleated in a similar right step in a right-lateral strike-slipfault and also 6. Conclusions initiated a bilateral rupture [Wald, 1996]. The right step also Seismicityand focal mechanismswithin a 10- to 15-km-wide relocates the main offshore San Andreas fault strand 3 km zone alongthe 100-km-longsection of the SanAndreas fault in closer to downtown San Francisco, which could result in an the San FranciscoPeninsula region indicate a prounounced increaseof predictedintensity by about one unit, or roughly along-strikevariation from a zone of dominantly compres- doublethe level of groundmotion in the westernmostportion sionaltectonism in the southto a zone of dominanatlyexten- of the city [Boatwrightand Perkins,1999]. sional tectonismin the north. Changesin San Andreas fault While the largestand most obviousseismic hazard for this geometryaccompany this changein tectonic style; the fault part of the San FranciscoBay area remainsa major strike-slip makesa broad(•-10ø-12 ø) left (restraining)bend on the south- earthquakeon the SanAndreas fault, our analysisindicates the ern peninsulaand an abrupt3-km right (releasing)step in the possibilityof moderate(M • 6 +) earthquakesdue to slip on north, just offshoreon the Golden Gate platform. subparallelnormal and thrust faults. The 30 km-long-zoneof The northernand southerndeformation zones are separated normal faulting seismicity(including a Mz• = 5.3 earthquake by a pronounced-15-km-long zoneof seismicquiescence both in 1957),together with geologicand seismic reflection evidence on and off the San Andreas fault on the central peninsula. for large-scale,young basin development offshore, suggests the Seismicityin the southernarea (from the north end of the 1989 possibilityof independentsurface rupturing normal faulting M -- 7.1 Loma Prieta ruptureto aboutthe town of Woodside) eventsdirectly offshore from San Francisco.The potential for occurs in 7- to 10-km-wide zones of deformation on both sides a M = 6 + thrust/reversefaulting earthquakein the southern of the San Andreas fault; a vertical San Andreas fault zone is peninsulahas previously been suggestedby Kovachand Beroza characterizedby its nearly completelack of seismicity.Focal [1993]. Our data indicate that a 60- to 70-km-long mapped mechanismsindicate thrust/reverse faulting with NE trending zone of Quaternary thrust faults on the NE side of the San P axes,indicating a pattern of fault-normal compressionsim- Andreas fault is underlainby a continuouszone of thrust and ilar to the patternof stressand deformation observed along the reversefaulting microseismicity. However, specific fault planes San Andreas fault throughoutmuch of California [Mount and at depth correspondingto surfacefaults can not be identified Suppe,1987; Zoback et al., 1987].This pattern of fault normal from the microseismicity.Further refinementof the potential compressionprobably enhanced by the left bend in the fault in reversefault hazard in this region is hamperedby this fact and this area and is consistentwith the young,high topographyof by the possibility,as in case of Loma Prieta, of potentially the Santa Cruz Mountains/CoastRanges. The thrust/reverse active fault zones at depth that may have no clear surface fault planes inferred from focal mechanismssuggest planar expressionor microseismicdefinition. faultsthat intersectthe San Andreasfault zone at a high angle (35ø-55ø),rather than mergingin a flower structuregeometry. Acknowledgments. Travel times for the refraction shots on the In marked contrast, focal mechanismson the northernmost CALNET stationswere provided by Donna Eberhart-Phillipsand peninsulaand offshoreon the Golden Gate indicatea combi- Andy Michael. The paper greatly benefittedfrom careful reviewsby nation of normal faulting and strike-slipseismicity in a diffuse Robert Simpson,Bob Smith,Diane Doser, David Oppenheimer,Uri zx)ncwcbt tot the t•gHt-btC Hg Odlll-•11UlCdb ldUlt. Not a sin e t,•. n,*i,,t•,Wayne T h•tohor, and V iotnri• I •ngonhoim This paper is thrust/reversefocal mechanismwas observeddown to M d = dedicatedto the memory of Stanford University Geology Professor Ben Page, structural geologistextraordinaire, who probably forgot 1.5, the approximatelimit for reasonablefocal spherecover- more about California Coast Range geologythan we will ever know, age.Normal faultingfocal mechanismsoccur within the newly and who alwaysso willingly sharedwhat he knew. defined 3-km right stepoverin the San Andreas fault and at least 15 km along-strikeboth to the SE and NW. Here again, topographyappears consistent with the ongoingtectonism' the References San Andreas fault lies below sea level throughoutmost of the Argus,D. F., and R. G. Gordon, Current Sierra Nevada-North Amer- northern section,consistent with subsidencecaused by exten- ica motion from very long baselineinterferometry; Implications for sion and basin development. the kinematicsof the westernUnited States,Geology, 19, 1085-1088, Most of the seismicityoccurs on secondarythrust and nor- 1991. Aydin, A., and B. M. Page,Diverse Pliocene-Quaternarytectonics in a mal faultsadjacent to the main SanAndreas. The SanAndreas transformenvironment, San FranciscoBay region, California, Geol. fault proper, as elsewherealong the 1906 rupture, is largely Soc. Am. Bull., 95, 1303-1317, 1984. aseismic,suggesting it is in a lockedportion of its seismiccycle. Bakun, W. H., Seismicmoments, local magnitudes,and coda-duration 10,740 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

magnitudesfor earthquakesin central California,Bull. Seismol.Soc. beneath the island of Hawaii, Ph.D. thesis, Mass. Inst. of Technol., Am., 74, 439-458, 1984. Cambridge,1977. Bakun, W. H., Seismicactivity of the San FranciscoBay region,Bull. Ellsworth,W. L., The SanAndreas fault system,California, U.S. Geol. Seismol.Soc. Am., in press,1999. Surv.Prof. Pap., 1515, 153-185, 1990. Baranov,V., A new methodfor interpretationof aeromagneticmaps: Finlayson,D. J., W. R. Hansen,and R. S. Ford, Evaluationof ground Pseudo-gravimetricanomalies, Geophysics, 22, 359-383, 1957. water resources,South Bay, Appendix A: Geology, Bull. 118-1, Barka, A. A., and K. Kadinsky-Cade,Strike-slip fault geometryin Calif. Dep. of Water Resour.,Sacramento, 1967. Turkey and its influenceon earthquake activity,Tectonics, 7, 663- Graham, S. A., and W. R. Dickinson,Apparent offsetsof on-land 684, 1988. geologicfeatures acrossthe San Gregorio-Hosgrifault trend, in Blake, M. C., Jr., D. G. Howell, and A. S. Jayko,Tectonostratigraphic edited by E. A. Silver and W. R. Normark, eds., San Gregorio- terranesof the San FranciscoBay region, in FranciscanGeology of Hosgrifault zone, California,Spec. Publ. Calif. Div. MinesGeol., 137, NorthernCalifornia, Spec. Publ. 43, edited by M. C. Blake, pp. 5-22, 13-23, 1978. Pac. Sect.Soc. of Econ. Paleontol.and Mineral., Los Angeles,Calif., Graymer, R. W., D. L. Jones,and E. E. Brabb, Preliminarygeologic 1984. map emphasizingbedrock formations in ContraCosta County, Cal- Blakely,R., and R. W. Simpson,Approximating edges of sourcebodies ifornia, U.S. Geol. Surv.Open File Rep., 94-622, 1994. from magnetic or gravity anomalies,Geophysics, 51, 1494-1496, Griscom,A., and R. C. Jachens,Tectonic history of the north portion 1986. of the San Andreasfault system,California, inferred from gravity Blanpied,M. L., D. Lockner,and J. D. Byerlee,An earthquakemech- and magneticanomalies, J. Geophys.Res., 94, 3089-3099, 1989. anismbased on rapid sealingof faults,Nature, 358, 574-576, 1992. Hall, N. T., Petrologyof the type Merced Group, San FranciscoPen- Boatwright,J., and J. Perkins,The attenuationof seismicintensity for insula,California, M. S. thesis,126 pp., Univ. of Calif., Berkeley, large strike-slipearthquakes, Bull. Seismol.Soc. Am., in press,1999. 1965. Bolt, B. A., The focusof the 1906 California earthquake,Bull. Seismol. Hall, N. T., FleishhackerZoo to MusselRock (Merced Formation);A Soc. Am., 58, 457-471, 1968. Plio-Pleistocenenature walk, Calif. Div. MinesGeol. Min. Info. Serv., Bonilla,M. G., Bedrock-surfacemap of the SanFrancisco South quad- 19, S22-S25, 1966. rangle,California: San FranciscoBay RegionEnvironment and Re- Hauksson,E. L., L. Jones,and K. Hutton, The 1994Northridge earth- sourcesPlanning Study, Basic Data Contribution26, scale1:31,680, quakesequence in California:Seismological and tectonicaspects, J. U.S. Geol. Surv./U.S.Dep. of Housing and Urban Dev., Washing- Geophys.Res., 100, 12,335-12,356, 1995. ton, D.C., 1964. Hengesh,J. V., and J. Wakabayashi,Dextral translationand progres- Bonilla, M. G., Serra fault zone, San Francisco Peninsula, California sive emergenceof the PleistoceneMerced Basin and implications (abstract),Eos Trans.AGU, 75(44), Fall Meet. Suppl.,681, 1994. for timingof initiationof the SanFrancisco peninsula segment of the Bonilla, M. G., Late Cenozoic folds and thrust faults, San Francisco San Andreas Fault, in RecentGeologic Studies in the San Francisco South quadrangle,U.S. Geol. Surv. Open File Rep., 96-267, 36-38, BayArea, Spec.Publ. 76, editedby E. M. Sangines,D. W. Andersen, 1996. andA. V. Buising,pp. 47-54, Pac. Sect.Soc. of Econ. Paleontol.and Boore, D. M., Strong-motionrecordings of the Californiaearthquake Mineral., Los Angeles,Calif., 1995. of April 18, 1906,Bull. Seismol.Soc. Am., 67, 561-577, 1977. Hengesh,J. V., J. M. Nolan, and J. Wakabayashi,Holocene displace- Bortugno,E. J., R. D. McJunkin, and D. L. Wagner, Map showing ment along the Serra fault, San Franciscopeninsula, California recencyof faulting,San Francisco-SanJose quadrangle, California, (abstract),Eos Trans.AGU, 77(46), Fall Meet. Suppl.,F744, 1996. Reg. Geol. Map Set., map 51, sheet5, map scale1:250,000, Div. of Hitchcock, C. S., K. I. Kelson, and K. C. Thompson,Geomorphic Mines and Geol., Calif. Dep. of Conserv.,Sacramento, 1992. investigationsof deformationalong the northeasternmargin of the Brabb, E. E., and E. H. Pampeyan,Geologic map of San Mateo Santa Cruz Mountains,U.S. Geol. Surv.Open File Rep., 94-187, 51 County,California, U.S. Geol. Surv.Misc. Invest.Set. Map, 1-1257-A, pp., 1994. 1983. Holbrook, W. S., T. M. Brocher, U.S. ten Brink, and J. A. Hole, Byerlee, J., Model for episodicflow of high pressurewater in fault Crustalstructure of a transformplate boundary:San FranciscoBay zonesbefore earthquakes,Geology, 21,303-306, 1993. and the centralCalifornia continental margin, J. Geophys.Res., 101, Catchings,R. D., and W. M. Kohler, Reflectedseismic waves and their 22,311-22,334, 1996. effect on strongshaking during the 1989 Loma Prieta, California, Hole, J. A., W. S. Holbrook, S. L. Klemperer, U.S. Ten Brink, and earthquake,Bull. Seismol.Soc. Am., 86, 1401-1416, 1996. T. M. Brocher,Crustal structure in the SanFrancisco Bay area from Clark, J. C., E. E. Brabb,H. G. Greene, and D.C. Ross,Geology of wide-angle seismicrefraction data (abstract),Eos Trans.AGU, Point Reyespeninsula and implicationsfor San Gregoriofault his- 74(43), Fall Meet. Suppl.,445, 1993. tory, in Tectonicsand SedimentationAlong the CaliforniaMargin, Hole, J. A., H. Thybo, and S. L. Klemperer, Seismicreflections from Spec.Publ. 38, editedby J. K. Crouch,and S. B. Bachman,pp. 67-86, the near-verticalSan Andreas fault, Geophys.Res. Lett., 23, 237-240, Pac. Sect.,Soc. Econ. Paleontol.and Mineral., Los Angeles,Calif., 1996. 1984. Ingram, B. L., Paleoclimaticand paleoceanographicstudies of estua- Cooper,A. K., Structureof the continentalshelf west of SanFrancisco, rine and marine sedimentsusing strontium isotopes, Ph.D. thesis, California, U.S. Geol. Surv.Open File Rep., 1907, 65 pp., 1973. 236 pp., Stanford Univ., Stanford, Calif., 1992. Cordell, L., and V. J. S. Grauch, Mapping basementmagnetization Jachens,R. C., and A. Griscom,Geologic and geophysicalsetting of zonesfrom aeromagneticdata in the San Juan basin, New Mexico, the 1989Loma Prieta earthquake,California, inferred from gravity in The Utilityof RegionalGravity and MagneticAnomaly Maps, edited and magneticanomalies, in TheLoma PrietaEarthquake of October by W. J. Hinze, pp. 181-197,Soc. of Explor.Geophys., Tulsa, Okla., 1989--GeologicSetting and CrustalStructure, edited by R. E. Wells 1985. and J. Vidale, U.S. Geol. Surv.Prof. Pap., 1550-E, in press,1999. Cummings,J. C., The Santa Clara Formation and possiblepost- Jachens,R. C., and M. L. Zoback, The San Andreas fault in the San Pliocene slip on the San Andreas fault in central California, in FranciscoBay region, California: Structure and kinematicsof a Proceedingsof Conferenceon GeologicProblems of the SanAndreas youngplate boundary,Int. Geol.Rev., in press,1999. Fault System,edited by W. R. Dickinson and A. Grantz, Stanford Jachens, R. C., C. W. Roberts, and M. L. Zoback, Total offset and Univ. Publ. Geol. Sci., 11, 191-207, 1968. right-steppinggeometry of the SanFrancisco Peninsula segment of Cummings,J. C., R. M. Touring, and E. E. Brabb, Geologyof the the SanAndreas fault, California,defined by aeromagneticanoma- northernSanta Cruz Mountains,California, in GeologicGuide to the lies (abstract),Eos Trans.AGU, 77(46), Fall Meet. Suppl.,F742, Gasand Oil Fieldsof NorthernCalifornia, edited by O. E. BowenJr., 1996. Calif. Div. Mines Geol. Bull., 181, 179-220, 1962. Jaumfi,S.C., and L. R. Sykes,Evolution of moderateseismicity in the Eaton, J.P., Determinationof amplitudeand durationmagnitudes and San FranciscoBay region,1850 to 1993:Seismicity changes related site residualsfrom short-periodseismographs in northern Califor- to the occurrenceof large and great earthquakes,J. Geophys.Res., nia, Bull. Seismol.Soc. Am., 82, 533-579, 1992. 101,765-789, 1996. Eaton, J.P., and M. J. Rymer, Regional seismotectonicmodel for the Jayko,A. S., and S. D. Lewis(Principal Compilers), Toward assessing southernCoast Ranges,in The Coalinga,California, Earthquake of the seismicrisk associatedwith blind faults, San FranciscoBay May 2, 1983, U.S. Geol. Surv.Prof. Pap., 1487, 97-112, 1990. Region,California, U.S. Geol.Surv. Open File Rept.96-267, 188 pp., Ellsworth, W. L., Three-dimensional structure of the crust and mantle 1996. ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA 10,741

Katzman,R., U.S. ten Brink, and J. Lin, Three-dimensionalmodeling locitystructure beneath San FranciscoPeninsula, California, J. Geo- of pull-apartbasins: Implications for the tectonicsof the Dead Sea phys.Res., 102, 5473-5490, 1997. basin,J. Geophys.Res., 100, 6295-6312, 1995. Prentice, C. A., and D. P. Schwartz, Re-evaluation of 1906 surface Klein, F. W., User's guide to HYPOINVERSE, a programfor VAX faulting,geomorphic expression, and seismichazard along the San computersto solvefor earthquakelocations and magnitudes,U.S. Andreas fault in the southern Santa Cruz Mountains, Bull. Seismol. Geol. Sun2.Open File Rep., 89-314, 44 pp., 1989. Soc. Am., 81, 1424-1479, 1991. Kovach, R. L., and G. C. Beroza, Seismicpotential from reverse Reasenberg,P. A. (Ed.), The Loma Prieta, California,earthquake of faultingon the San Franciscopeninsula, Bull. Seismol.Soc. Am., 83, October 17, 1989--Aftershocksand postseismiceffects, U.S. Geol. 597- 602, 1993. Sun2.Prof. Pap., 1550-D, 310 pp., 1997. Langenheim, V. E., K. M. Schmidt, and R. C. Jachens,Coseismic Reasenberg,P. A., and D. H. Oppenheimer, FPFIT, FPPLOT, and deformationduring the 1989 Loma Prieta earthquake and range- FPPAGE: FORTRAN computerprograms for calculatingand dis- front thrustingalong the southwesternmargin of the Santa Clara playing earthquakefocal mechanisms,U.S. Geol. Sun2.Open File Valley, California, Geology,25, 1091-1094, 1997. Rep., 85-739, 109 pp., 1985. Lisowski,M., J. C. Savage,and W. H. Prescott,The velocityfield along Reasenberg,P. A., and R. W. Simpson,Response of regionalseismic- the SanAndreas fault in central and southernCalifornia, J. Geophys. ity to the static stresschange produced by the Loma Prieta earth- Res., 96, 8369-8389, 1991. quake, Science,255, 1687-1690, 1992. Lisowski,M., M. Murray, and J. L. Svarc,Geodetic measurementsof Rice, J. R., Fault stressstates, pore pressuredistributions, and the coseismichorizontal deformation, U.S. Geol. Sun2. Prof. Pap., weaknessof the SanAndreas fault, in Fault Mechanicsand Transport 1550-A, A81-A95, 1997. Propertiesin Rocks(The BraceVolume), edited by B. Evansand T.-F. Louderback,G. D., Central California earthquakesof the 1830's,Bull. Wong, pp. 475-503, Academic,San Diego, Calif., 1992. Seismol. Soc. Am., 37, 33-74, 1947. Roberts,C. W., and R. C. Jachens,Isostatic residual gravity map of the Marsden, R., M. L. Zoback, D. Dreger, B. Julian, J. Olson, and T. M. San FranciscoBay area, California, U.S. Geol. Sun2.Map GP-1006, Parsons,M5.3 normal faulting event adjacent to the San Andreas scale 1'286,500, 1993. fault on the San Franciscopeninsula (abstract), Eos Trans.AGU, Roecker, S. W., Seismicityand tectonicsof the Pamir-Hindu Kush 76(46), Fall Meet. Suppl.,F423, 1995. regionof centralAsia, Ph.D. thesis,207 pp., Mass.Inst. of Technol., McCulloch, D. S., Regional geologyand hydrocarbonpotential of Cambridge, 1981. offshorecentral California, in Geologyand ResourcePotential of the Rogers, T. H., and J. W. Williams, Potential seismichazards in Santa ContinentalMargin of WesternNorth America and Adjacent Ocean Clara county,California, Spec. Rep., Calif. Div. Mines Geol., 107, 39 Basins--BeaufortSea to Baja California,edited by D. W. Scholl,A. pp., 1974. Grantz, and J. G. Vedder, pp. 353-401, Circum-Pac. Counc. for Scholz,C. H. C. A. Aviles, and S. G. Wesnousky,Scaling differences Energy and Miner. Resour.,Houston, Tex., 1987. betweenlarge interplate and intraplate earthquakes,Bull. Seismol. McLaughlin,R. J., The Sargent-Berrocalfault zone and its relation to Soc. Am., 76, 65-70, 1986. the San Andreas fault systemin the southern San FranciscoBay Segall,P., and M. Lisowski,Surface displacements in the 1906 San region and Santa Clara Valley, U.S. Geol. Sun2.J. Res.,2, 593-598, Franciscoand 1989 Loma Prieta earthquakes,Science, 250, 1241- 1974. 1244, 1990. McLaughlin,R. J., and J. C. Clark, Geologicsetting of the October17, Sibson,R. H., Stoppingof earthquakeruptures at dilationalfault jogs, 1989, Loma Prieta earthquake,in The Loma Prieta Earthquakeof Nature, 316, 248-251, 1985. October1989--Geologic Settingand CrustalStructure, edited by R. E. Simpson,G. D., S.C. Thompson,J. S. Noller, and W. R. Lettis, The Wells and J. Vidale, U.S. Geol. Sun2.Prof. Pap., 1550-E, in press, northern San Gregorio fault zone: Evidencefor the timing of late 1999. Holocene earthquakes near Seal Cove, California, Bull. Seismol. McLaughlin,R. J., W. V. Sliter, D. H. Sorg,P. C. Russell,and A.M. Soc. Am., 87, 1158-1170, 1997. Sarna-Wojcicki,Large-scale right-slip displacement on the eastSan Simpson,R. W. (Ed.), The Loma Prieta, California earthquakeof FranciscoBay regionfault system,California: Implications for loca- October 17, 1989roTeconicprocesses and models,U.S. Geol. Sun2. tion of late Miocene to PliocenePacific plate boundary,Tectonics, Prof. Pap., 1550-F, 131 pp., 1994. 14, 1-18, 1996. Simpson,R. W., and P. A. Reasenberg,Earthquake-induced static Mount, V. S., and J. Suppe,State of stressnear the SanAndreas fault: stresschanges on central California faults, U.S. Geol. Sun2.Prof. Implicationsfor wrenchtectonics, Geology, 15, 1142-1146, 1987. Pap., 1550-F, 55-89, 1994. Murphy, J. M., R. D. Catchings,and W. M. Kohler, Data report for Sleep,N.H., and M. L. Blanpied,Creep, compactionand the weak 1991 active-sourceseismic profiles in the San FranciscoBay area, rheologyof major faults,Nature, 359, 687-692, 1992a. California, U.S. Geol. Sun2.Open File Rep., 92-570, 45 pp., 1992. Sleep,N.H., and M. L. Blanpied,Ductile creep and compaction:A Olson, J. A., and D. P. Hill, Seismicityin the southern Santa Cruz mechanismfor transientlyincreasing fluid pressurein mostlysealed Mountainsduring the 20-yearperiod before the earthquake,in The fault zones,Pure Appl. Geophys.,143, 9-40, 1992b. Loma Prieta, California,Earthquake of October17, 1989:Earthquake Sorg, D. H., and R. J. McLaughlin, Geologic map of the Sargent- Occurrence--PreseismicObservations, U.S. Geol. Sun2.Prof. Pap., Berrocal fault zone between Los Gatos and Los Altos Hills, Santa 1550-C, C3-C16, 1997. Clara County, California, U.S. Geol. Sun2.Misc. FieM Inv. Map Olson,J. A., and M. L. Zoback,Source character of microseismicityin MF-643, scale 1'24,000, 1975. the SanFrancisco Bay block, California, and implicationsfor seismic Spudich,P., The Loma Prieta, California, earthquakeof October 17, hazard, Bull. Seismol. Soc. Am., 88, 543-555, 1998. 1989--Main shock characteristics,U.S. Geol. Sun2.Prof. Pap., Oppenheimer,D. H., Aftershock slip behavior of the 1989 Loma 1550-A. 297 pp.. 1996. Prieta earthquake,Geophys. Res. Lett., 17, 1199-1202, 1990. Sylvester,A. G., Strike-slipfaults, Geol. Soc.Am. Bull., 100, 1666- Oppenheimer,D., F. W. Klein, J.P. Eaton, and F. Lester,The North- 1703, 1988. ern California SeismicNetwork Bulletin January-December1992, ten Brink, U.S., R. Katzman, and J. Lin, Three-dimensional models of U.S. Geol. Sun2.Open File Rep., 93-578, 45 pp., 1993. (Available at deformationnear strike-slipfaults, J. Geophys.Res., 101, 16,205- http://quake.geo.berkeley.edu/ncsn/bulletin.ps). 16,220, 1996. Page, B. M., Modes of Quaternary tectonic movement in the San Thatcher,W., G. Marshall, and M. Lisowski,Resolution of fault slip FranciscoBay region,California, in Proceedingsof the Conferenceon along the 470-km-longrupture of the great 1906 San Francisco EarthquakeHazards in the Eastern San FranciscoBay Area, Spec. earthquakeand its implications,J. Geophys.Res., 102, 5353-5367, Publ. Calif. Div. Mines Geol., 62, 1-10, 1982. 1997. Page, B. M., Evolution and complexitiesof the transform systemin Toppozada,T. R., and G. Borchardt,Re-evaluation of the 1836 "Hay- California, U.S.A., Ann. Tect.,/V, 53-69, 1990. ward fault" and the 1838 San Andreasfault earthquakes,Bull. Seis- Page,B., Tectonicsetting of the San FranciscoBay region,in Proceed- mol. Soc. Am., 88, 140-159, 1998. ingsof the SecondConference on EarthquakeHazards in the Eastern Tuttle, M.P., and L. R. Sykes,Re-evaluation of severallarge historic San FranciscoBay Area, Spec.Publ. Calif. Div. Mines Geol., 113, 1-7, earthquakesin the vicinity of the Loma Prieta and Peninsularseg- 1992. ments of the San Andreas fault, California, Bull. Seismol.Soc. Am., Parsons,T., and M. L. Zoback, Three-dimensionalupper crustalve- 82, 1802-1820, 1992. 10,742 ZOBACK ET AL.: SAN ANDREAS FAULT ZONE, SAN FRANCISCO PENINSULA

Uhrhammer,R. A., The Pacificaearthquake of April 28, 1979,Bull. Williams, S. D. P., J. L. Svarc,M. Lisowski,and W. H. Prescott,GPS Seismol.Soc. Am., 71, 1161-1172, 1981. measuredrates of deformation in the northern San FranciscoBay Valensise,G., Geologicassessment of the relative contributionof region,California, 1990-1993, Geophys. Res. Lett., 21, 1511-1514, strike-slipfaulting, reverse-slip faulting, and bulk squeezingin the 1994. creation of the central Santa Cruz Mountains, California, U.S. Geol. WorkingGroup on CaliforniaEarthquake Probabilities, Probabilites Surv.Prof Pap., 1550-F, 23-47, 1994. of large earthquakesin the San FranciscoBay region,California, Wagner,D. L., E. J. Bortugno,and R. D. McJunkin,Geologic map of U.S. Geol. Surv.Circ., !053, 51 pp., 1990. the San Francisco-SanJose quadrangle, California, Reg. Geol. Map WorkingGroup on NorthernCalifornia Earthquake Potential, Data- Ser.,map 5A, scale1:250,000, Div. of Minesand Geol., Calif. Dep. baseof potentialsources for earthquakeslarger than magnitude 6 in of Conserv., Sacramento, 1991. northernCalifornia, U.S. Geol. Surv.Open File Rep., 96-705,57 pp., Wakabayashi,J., andE. Moores,Evidence for collisionof the Salinian 1996.(Available at http://quake.wr.usgs.gov/hazprep/NCEP) block with the Franciscan subductionzone, California, J. Geol., 96, Zoback, M.D., et al., New evidenceon the state of stressof the San 245-253, 1988. Andreasfault system,Science, 238, 1105-1111, 1987. Wald, D. J., Slip historyof the 1995Kobe, Japan,earthquake deter- Zoback,M. L., and R. C. Jachens,Right-stepping geometry of the San minedfrom strong-motion,teleseismic, and geodeticdata, J. Phys. Andreasand San Gregoriofaults and persistentnormal faulting in Earth, 44, 489-503, 1996. the 1906 San Franciscoearthquake epicentral area (abstract),Eos Wald, D. J., H. Kanamori, D. V. Helmberger, and T. H. Heaton, Trans.AGU, 77(46), Fall Meet. Suppl.,F742, 1996. Sourcestudy of the 1906 San Franciscoearthquake, Bull. Seismol. Soc. Am., 83, 981-1019, 1993. R. C. Jachensand M. L. Zoback, Mail Stop 977, U.S. Geological Wells, R., and J. Vidale (eds.), The Loma Prieta, California,earth- Survey, 345 Middlefield Road, Menlo Park, CA 94025. quakeof October17, 1989--Geologicsetting and crustal structure, ([email protected];[email protected]) U.S. Geol. Surv.Prof Pap., 1550-E, in press,1999. J. A. Olson, Trimble NavigationCompany, Sunnyvale, CA 94401. Wesnouskry,S. G., Seismologicaland structural evolution of strike-slip ([email protected]) faults, Nature, 335, 340-343, 1988. Williams, S. D. P., Current motion on faults of the SanAndreas system in central California inferred from recent GPS and terrestrialsurvey measurements,Ph.D. thesis,282 pp., Univ. of Durham, Durham, (ReceivedDecember 15, 1997;revised August 18, 1998; England, 1995. acceptedOctober 13, 1998.)