Late Neogene and Quaternary Tectonics Associated with Northward Growth of the San Andreas Transform Fault, Northern California

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 93, NO. B5, PAGES 4797-4819, MAY 10, 1988

Late Neogene and Quaternary Tectonics Associated With Northward Growth of the San Andreas Transform Fault, Northern California

HARVEY M. KELSEY

Department of Geolo•ty, Western Washin•tton University, Bellin•tham

GARY A. CARVER

Department of Geology, Humboldt State University, Arcata, California

Strain patterns within the forearc at a convergent margin adjacent to a passing fault-fault-trench triple junction show a systematic evolution. We describe late Neogene and Quaternary deformation in northern coastal California in the last 3 m.y. associated with the migration of the Mendocino triple junction. Deformation in the North American forearc near Humbolt Bay consists of a 30-km-wide zone of on-land contraction with a narrower zone of translation to the east. The zone of subaerial contraction is part of the relatively wide, southernmost extent of forearc contraction within the Cascadia subduction zone. Net Quaternary northeast-southwest contraction across forearc thrust faults is at least 7.9 km, and minimum fault slip rates on the six major thrust faults are 0.8-2.3 mm/yr. Net right slip within the zone of translation is a minimum of 3 km. South of the forearc, translation is predominant in the North American plate and is accommodated along two major right-lateral fault zones of the San Andreas transform boundary, the Lake Mountain, and Garberville fault zones. Discontinuous exposures of late Neogene sediments along these faults near Covelo and Garberville, respectively, indicate that crustal contraction, expressedby thrust faults and associatedfolding, predated translational deformation at each site. Inception of contraction near Covelo was related to inception of internal deformation of the Gorda plate at 3 Ma. This contraction, due to coupling of the convergent plates, migrated northward in the forearc in advance of and at the rate of migration of the triple junction. Present-day deformation near Humboldt Bay thus reflects contraction associatedwith subduction of the Gorda plate and translation farther east associatedwith oblique convergenceat the plate boundary. As the triple junction migrates, the strike-slip faults of the San Andreas transform boundary will extend to the north-northwest into the present forearc, and forearc strike-slip faults will become faults of the San Andreas system. Contractional structuresnear Humboldt Bay will then ceaseto be active and will be preservedin the geologic record in the same way as contractional structures are preserved farther south.

INTRODUCTION inson and Snyder, 1979a; McLaughlin and Nilsen, 1982]. Fur- This paper has two purposes. The first is to describe late thermore, a landward jump of the transform margin occurred Neogene and Quaternary deformation in north coastal Cali- about 10 Ma, when dominant right-lateral motion shifted fornia in the past 3 m.y. The second is to relate the pattern of from the San Gregorio-Hosgri fault system east to the San deformation preserved by the geology to plate interactions Andreas fault system[Graham and Dickinson,1978]. associated with the Mendocino triple junction. The Men- Another source of triple junction instability is a change in docino triple junction was the product of collision of the relative plate motion [Dickinson and Snyder, 1979a-I. Such a Pacific-Farallon ridge with the North American plate at about change occurred at the Mendocino triple junction, starting at 28 Ma [Atwater, 1970]. The collision created two triple junc- about 2.5 Ma [Silver, 1971] or 3.0 Ma [Wilson, 1986-1,when tions that migrated away from each other as the San Andreas Gorda-Pacific plate relative motion ceased to be parallel to transform boundary extended in length. The Mendocino triple the Mendocino fracture zone. This change in relative motion junction is at the northern end of this transform boundary is recorded by bends in the magnetic anomalieson the south- (Figure 1). ern Gorda plate [Raft and Mason, 1961]. Change in relative Crustal deformation in north coastal California during the plate motion has been accommodatedby deformation of the past 3 m.y. is in part the result of the geometricinstability of southern Gorda plate as it slides along the Mendocino frac- the Mendocino triple junction as it migrates northward. If the ture zone and is subducted under North America [Silver, fault-trench segment of the triple junction that bounds the 1971; dachens and Griscom, 1983; Wilson, 1986]. North American plate does not describe a small circle, then The pattern of internal deformation of the Gorda plate, the junction is unstable because it cannot retain its geometry including rupture along northeast trending, left-lateral faults as it moves [McKenzie and Mor•lan, 1969]. This type of insta- parallel to the trendsof the bent magneticanomalies, indicates bility creates extensional basins near the triple junction and north-south contraction [Silver, 1971; McPherson et al., 1981; landward jumps of the transform margin [Dickinson and Smith et al., 1981; Wilson, 1986]. Therefore, in the last 3 m.y., Snyder• 1979a]. The Neogene structural basins of the northern subduction of the internally contracting Gorda plate, rather California Coast Ranges provide evidence for the hypoth- than extension, is probably the major factor influencing crust- esized extension that accompanies the northward propagation al deformation on land near Cape Mendocino. of the San Andreas transform margin [Blake et al., 1978; Dick- As a consequenceof the northward elongation of the Pacific-North American transform boundary, there is no sub- Copyright 1988 by the American Geophysical Union. ducting slab beneath the North American plate south of the Paper number 7B2052. triple junction (called a slab window by Dickinsonand Snyder 0148-0227/88/007B-2052505.00 [1979b]). In northern coastal California, deformation is more

4797 4798 KELSEYAND CARVER:QUATERNARY TECTONICS, NORTHERN CALIFORNIA

125 ø 124 ø 123ø 122 ø 42 ø \ ' I I i FIGURES 2 E• 5

MT. SHASTA LOST MAN FAULT

MAD 41 RIVER 41 ø FZFAU FIGU RE 3 ONE LITTLE REDDING LASSEN PEAK

EATON ROUGHS ONE)I• FAULT ZONE MENDOCINO I TRIPLE JUNCTION LAKE MOU NTAI N 40 ø FAULT ZONE NORTH 40 ø I ROUND PACIFIC VALLEY AMERICAN PLATE PLATE \• BARTLETT

39 •ZONE :59 ø

POPE CR.t FAU•LT• SACRAMENTO __-'•-• CEDAR • HEALDBURG NAPA •ROUGHS • ROGERS CR. FAULT • FAULT ZONE FAULT ZONE CORDELIA FAULT ZONE GR E EN VALLEY FAULT 38 CONCORD FAULT 38 ø -HAYWARD FAULT 0 20 40 60 KI LOM ETERS I I I I ! I I SAN GREENVILLE FRANCI FAULT SEALCOVE FAULT FAULT I I i ZONE 125ø 124 ø 125 ø 122ø Fig. 1. Generalizedtectonic map of northernCalifornia showing plate geometry and major faults. Modification of a compilationby DePoloand Ohlin [1984]. In additionto ourmapping for thenorthern segment, sources include Jennings [1977],Kelsey and Cashman [1983], McLaughlin [1974], Steffen et al. (unpublishedreport, 1983), Upp [1982-1, Wagner and Bortugno[1982], and Wright et al. [1982]. KELSEYAND CARVER: QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4799 complex and heterogeneous immediately north of the slab troduced by Angelier [1984] and utilized by Angelier et al. window. In this region, where the subductedGorda plate un- [1985], Norris and Cooper [1986], and Frizzell and Zoback derlies the accretionary margin of North America, crustal de- [1987]. formation includes componentsof both contraction and later- The type of strain recorded by mesoscalefaults is apparent al translation. Farther south, above the slab window, defor- in outcrop by offset relations and slickensidestriae or grooves mation appears to be mainly by translational displacementon on fault surfaces.Where fracturesand faults occur together in relatively straight, north-northwest trending strike-slip faults groups of similar orientation (clusters), we assume the same that occur as three distinct fault zones(Figure 1). origin for the fractures and the faults. Mesoscale fractures and faults are ubiquitous in the late INVESTIGATIVE APPROACH Neogene sediments of the study area. Generally, these structures increasein density within 200-300 m either side of mega- Mapping of Deformation scale faults, with greatest density near the fault trace. We In order to document late Neogene and Quaternary crustal therefore consider mesoscalefaults as genetically related to the deformationin the study area (Figures2 and 3), we selectively megascalefault. mapped areas of Quaternary and Neogene cover sediments. Pairs of fault setsare common at many measuring sites.The However, we did not do detailed field studiesof the Neogene sets are separatedby an acute angle. In caseswhere the two Wildcat Group sedimentsin the Eel River basin becausepre- fault setsformed synchronously(each offsetsthe other), have vious studies [Ogle, 1953; Woodward-ClydeAssociates, 1980] opposite sensesof displacement,and show displacement per- are available. pendicular to the line of intersection of the fault sets, the two Late Neogene and Quaternary sediments cover about 20% fault setsare a conjugatefault pair [Hobbs et al., 1976]. of the area (Figures 2 and 3) and are concentrated near the We found two types of conjugate faults in Neogene sedi- coast. The Franciscan assemblage [Blake and Jones, 1981] ments near megascalefaults. In type one, all faults were verti- forms the basementrocks. Where cover rocks are absent,post- cal or subvertical, and all indicators of fault slip were horizon- Miocene deformation in the Franciscan could not be con- tal or subhorizontal. In type two, all faults dipped less than clusively documented.However, on the basis of Quaternary 60ø, all indicators of fault slip showed a downdip sense,and landformsand fault offsets(detailed below) it is most probable offsets showed a reverse movement sense. that active faulting also occurs east of the area of Neogene Conjugate strike-slip shearsare associatedwih transcurrent cover. fault systems[Harding, 1973; Wilcox et al., 1973]. However, We mapped by aerial photographic interpretation and field the orientation of conjugate shears does not provide an un- reconnaissancethose fault zones that do not cut Neogene sed- equivocal direction for the principal strain axes, despite iments but nonetheless appear to be active. We define a fault models [Wilcox eta!., 1973] that show such a relation under zone as a zone that comprises individual fault traces that are idealized conditions. Orientation of the strike-slip conjugate close together and share a similar trend. In order to map a faults relative to the principal strain for the region varies de- fault zone as probably active, each fault within the zone had pending on postfaulting tectonic rotation and on where the to meet the following criteria: (1) each fault must have a vis- conjugate fault developed along a segmented fault zone ually obvious trace defined by a prominent topographic lin- [Sibson, 1986]. Therefore conjugate strike-slip faults are used eament or trough hundreds of meters wide, (2) the trace must to document megascaletranscurrent fault deformation, but we be accompaniedby a linear trend of springs,indicating that will not use these faults to assessaxes of principal strain. the structure locally controls groundwater flow, (3) the trace In the portion of the study area north of the latitude of the must display at least two of the following features:aligned sag Little Salmon fault (Figure 2), the cover sediments are Plio- ponds, aligned notched ridges, aligned linear drainage seg- cene and Pleistocene,and only one deformation is present. In ments, or ridgetop depressions,and (4) at least one of the this area we could establish a correlation of low-angle conju- traceswithin the fault zone must have offset a Neogene unit or gate faulting (at the mesoscale)with thrust faults (at the mega- a distinctiveFranciscan melange block. scale).Similarly, high-angle conjugatemesoscale faults are as- Faultsshown by dashedlines on Figures2 and 3 are major sociatedwith high-angle,dominantly strike-slip megascale faults that cut Mesozoic basement rocks but do not meet the faults. criteriafor classificationas probablyactive faults. Though suf- In the studyarea near the Russfault and farther southward ficient geomorphicevidence is lacking,the dashedfaults none- (Figure 3) the Neogene sedimentsappear to have two gener- thelessmay havebeen active in the Quaternary. ations of deformation,based on offsettingrelations of the megascalefaults. Regional folding is associatedwith the first de- MesoscaleStructures formation. Associated with the major faults that cut Neogene and youngersediments is a broad(400-600 m wide)zone of frac- ClassificationofCrustal Strain tures. These fractures can be divided into three classes: those Crustal strain is dominated either by contraction (thrust with evidence of shear origin, called mesoscale(outcrop-scale) faults, reverse faults, folds) or lateral translation (high-angle faults; those with evidence of extension origin, called exten- faults that in all caseshave strike-slip displacements that are sion fractures; and those of indeterminate origin, which we much greater than any observed dip-slip displacements). call fracturesbecause they cannot be further classified. Crustal extension is not a regionally significant type of strain Analysis of mesoscalefaults was used as a field technique to in northern coastal California. However, many outcrops show differentiate megascale strike-slip faults from megascale re- extension,and extensionalstrain is significantat fault bends verse faults. This field technique has been further developed and stepoversin the strike-slip fault zones. Intermontane val- elsewhere [Kelsey and Cashman, 1983]. Somewhat related leys in the coast ranges (Figures 2 and 3) are geomorphic techniques,directed toward paleostressanalysis, have been in- expressionsof such localized extension. 4800 KELSEYAND CARVER' QUATERNARY TECTONICS, NORTHERN CALIFORNIA

124 o

LATE N EOGENE AN D QUATERNARY

41o$0 ' PRE- 41ø$0 '

STRIKE-SLIP FAULT 0 5 I0 15 20 :>5 30 I I I I I I I THRUST FAULT KILOMETERS \ FAULT WITH VERTICAL OFFSET PROBABLY ACTIVE FAULT FAULT-ACTIVITY UNKNOWN

•Ns

41o00 ' 41ø00 '

\ • BAY N ' "

• . .?' ....•:..•,

'• . ' ' . "':'". -.'..'.... ß NsO -•, . .::.;.•.[:/-..'...'..•...... E 40o$0 ' ß . '...... '...... ;;....;:.•:. 40ø$0 ' CAPE', TOWN .:'.:'"'-:'•':::'"•:;'" MTJ FAULT '..... I I 124 ø Fig.2. Tectonicmap of north coastal California north of Cape Mendocino. Offshore data from Clarke [1987]. Faults withsolid lines are probably active (for criteria, see text); faults with dashed lines do not meet criteria of probablyactive faults.Faults: LMF, Lost Man fault; GFZ, Grogan fault zone (offshore only); BCL, Bridge Creek lineament; BLF, Big Lagoonfault; Trinidad fault zone (offshore only); BF, Blue Lake fault; MF, McKinleyville fault; MRF, Mad River fault; FHF,Fickle Hill fault: SCL, Snow camp lineament; CBT, coastal belt thrust. Folds: Gold Bluffs syncline; FHA, Fickle Hillanticline; HHA, Humboldt Hill anticline; TBA, Table Bluff anticline; AA, Alton anticline; RS, Redway syncline. Intermontanevalleys: MM, Murphy'sMeadows; LV, Larrabee Valley; BV, Burr Valley. Water bodies: AB, Arcata Bay; HB,Humboldt Bay. Localities: O, Orick; T, Trinidad;MB, Moonstone Beach; A, Arcata; E, Eureka; B, Bridgeville; D, Dinsmore.Geologic units: dot pattern, all lateNeogene and Quaternary sediments (mid-Miocene to present; excludes Miocenerocks of Kings Range terrane of McLaughlin etal. [1982]); no pattern, early Cenozoic and pre-Cenozoic rocks of Franciscanassemblage (includes Kings Range terrane) and Klamath Mountain province; NS, isolated patches of late Neogenesediments, agedesignation uncertain; MTJ, approximate location of Mendocino triple junction.

CONTRACTIONAL DEFORMATION Neogenebut probablynot activetoday, is documentednear Late Neogene and Quaternary sedimentsrecord crustal Garbervilleand near Covelo,California. contractionin the Humboldt Bay area of north coastalCali- fornia.Contemporaneous deformation is occurringalong the Mad River Fault Zone Mad Riverfault zone and in thelower Eel Rivervalley (Figure The Mad River fault zone [Carveret al., 1983] is a set of 2). Crustalcontraction, probably active in thelatter part of the northwesttrending, northeast dipping imbricate thrust faults KELSEYAND CARVER: QUATERNARYTECTONICS, NORTHERN CALIFORNIA 4801

•24 o

:'•:.:..;. LATEAN D QUATERNARYNEOGENE • PRE-QUATERNARY STRIKE-SLIP FAULT '• - -- THRUST FAULT •, FAULTWITH • VERTICALOFFSET PROBABLY ACTIVE FAULT %% __--__FAULT-ACTIVITY \ UNKNOWN

40 ø 30' CAPE 40ø30 '

CAPE TOWN .1 \ I 'FIGURE 9 FAULT

FIGURES \ o 6•?• N,, -K Ns r"

rn N •' .,)SCD 40 ø 00' 40o00 '

• • •' •SV •! •., \ ROUND

FIGURE

LAYTONVI LLE

124 o

Fig. 3. Tectonic map of northern terminus of San Andreas fault zone. Offshore data from Clarke [1987]. See caption to Figure 2 for abbreviations and explanations that are not noted below. Faults: CBT, coastal belt thrust; GFZ, Garberville fault zone; DCF, Dean Creek fault; LF, Laytonville fault; TF, Temblor fault. Folds: FHA, Fickle Hill anticline; HHA, Humboldt Hill anticline;TBA, Table Bluff anticline;AA, Alton anticline; RS, Redway syncline.Inter- montane valleys: HV, Hettenshaw Valley; KHV, Kettenpom-Hoaglin valleys; SCD, Salt Creek depression;SV, Summit Valley; RV, Round Valley; SO, Stoten opening;EV, Eden Valley. Localities:G, Garberville; Z, Zenia; K, Kettenpom; LM, Lake Mountain; P, Piercy; C, Covelo.

that extend from the coast approximately 40 km inland to the anticlinal limbs dipping 200-30ø and southern limbs nearly vicinity of Maple Creek (Figures 2 and 5). The fault zone vertical and locally overturned.Fold axesparallel the trend of contains five principal thrust faults (Trinidad, Blue Lake, the thrusts (N35ø-40øW) and plunge gently north-northwest. McKinleyville, Mad River, and Fickle Hill faults) and numer- Fold limbs are cut by the thrust faults. The faults offset the ous minor thrust faults. Road cut exposuresand exploratory Pleistocene marine and fluvial sediments of the Falor Forma- trenches [Woodward-Clyde Associates, 1980] show that the tion (ageof the Falor Formation is discussedbelow) [Manning faults dip between 20ø and 40ø (Table 1). The Blue Lake anti- and Ogle, 1950] and offset late Pleistocenefluvial and marine cline and Fickle Hill anticline constitute major folds within terracesnear the coast [Carver et al., 1986]. the zone (Figure 2). The folds are asymmetrical,with northern Net dip slip on each of the faults can be estimatedby the 4802 KELSEYAND CARVER: QUATERNARY TECTONICS, NORTHERN CALIFORNIA

TABLE 1. Summary of Fault Deformation in the Mad River Fault Zone Based on Displacements of the Falor Formation Basal Contract

Fault Vertical Dip-Slip Slip or Fault Age," Displacement, Displacement, Rate, Zone Location Ma Dip m m mm/yr

Trinidad fault Canon Creek 0.7 40 ø 575 900 1.3 Blue Lake fault Canon Creek 0.7 40 ø 750 1200 1.6 Blue Lake fault Korbel 0.7 35 ø 950 1600 2.3 McKinleyville fault Simpson Timber Company Road 4500 0.7 35ø 300 525 0.8 Mad River fault Simpson Timber Company Road 5400 0.7 35ø 325 575 0.8 Fickle Hill fault Jacoby Creek 0.7 25ø 350 825 1.2 Mad River fault zone Jacoby Creek-Korbel Section 0.7 35ø 2500 4400 6.4

"Age is the age of initiation of faulting assumingfaulting commencedwith termination of Falor deposition, or a little more than a million years after eruption of the Huckleberry Ridge tuff. See discussionin text.

amountof offsetof the basaldepositional contact of the Falor actualshortening rate is at leasttwice that calculatedfrom the Formation on the Franciscanassemblage. Net dip slip for maximum age of the Falor sediments.In Table 1 we assume eachfault rangesfrom 525 to 1600 m (Table 1). Using the that faultingcommenced at 0.7 Ma, or approximately1.2 m.y. rangeof dip and the amountof dip-slipdisplacement for the afterdeposition of the HuckleberryRidge ash. Given this age five thrust zones,net horizontal shorteningacross the Mad for fault initiation, dip-slip displacementof the basal Falor River fault zone in Pleistocenetime due to low-anglefaulting contacton individual faults in the Mad River fault zone yields must be about 3.6 km along a northeast-southwestaxis. fault dip-slip rates ranging between0.8 and 2.3 mm/yr (Table A minimum shorteningrate for the Mad River fault zone 1). Summingthese slip rates acrossthe fault zone yieldsa net (assumingthat all shorteningis due to low-angle faulting) is slip rate of 6.4 mm/yr, or (assumingan averagethrust fault dip provided by a maximum age for the Falor Formation. This of 35ø) 5.2 mm/yr of NE-SW horizontal shorteningduring the maximum age comesfrom a volcanicash near the deposition- last 0.7 m.y. This net 3.6 km of NE-SW horizontalshortening al base of the Falor, which is correlated by its geochemistryto from documented thrust faults is a minimum estimate. Ad- the 1.8-2.0 Ma Huckleberry Ridge tuff from the Yellowstone ditional contraction of unknown magnitudemust exist due to caldera [Sarna-Wojcickiet al., 1987]. Minimum horizontal folding associatedwith the mapped thrust faults as well as shorteningrate acrossthe Mad River fault zone since1.8-2.0 faulting on other unidentifiedNeogene contractionalstruc- Ma is thereforeapproximately 1.8-2.0 mm/yr. turesthat likely occurin Franciscanrocks to the northeast. In all measured stratigraphic sectionsthe top of the Falor Deformation of the late Pleistocene marine terraces pro- Formation has been removed by faulting, but the minimum vides an independentmeasure of shorteningrate acrossthe thicknessis in excessof 940 m. The formation includespoorly Mad River fault zone [Carver et al., 1986]. From one to seven consolidatedinterbedded clays, silts, sands,pebbly sands,and raised marine terraces straddle the Trinidad, McKinleyville, gravels deposited in conditions ranging from shallow open Mad River, and Fickle Hill faults. The older terracesare pro- marineto beach,estuarine, and fluvialenvironments. The lack gressivelyoffset by greateramounts (Figure 4). Burkeet al. of angularunconformities with significantstratigraphic relief [1986] assignedages to the terraces(Figure 4) by correlation indicatesthat Mad River fault zone faulting and folding were of the altitude distribution of the terraceson the upfaulted not activeduring deposition of the Falor sediments.Therefore (northeastern)fault blocksto worldwidehigh sealevel stands we feelthat it is likelythat Mad Riverfault zone deformation defined by datedmarine terraces at New Guinea[Bloom et al., did not accompanyFalor depositionbut ratherit was the 1974;Chappell, 1983]. Terrace age assignments are according developmentof faultand fold structures that terminated Falor to methodsand assumptionsdiscussed by Bull [1985].From deposition. the terracedisplacement and agedata, dip-slip displacement Time span of Falor depositionis not well constrainedat the rates for the four faults of the Mad River fault zone in the last youngerlimit. Depositsat MoonstoneBeach south of Tri- 0.2m.y. are 0.9, 0.9, 1.2, and 0.8 mm/yr, respectively (Figure 4). nidad(MB, Figure2), which are either upper Falor Formation Cumulativedip-slip displacement rate duringthe late Pleisto- or a youngeroverlying unit (W. Miller III, personalcommuni- ceneis thereforeabout 3.8 mm/yr. This minimumrate (which cation, October 1986), yield age estimates of somewhere be- excludes the Blue Lake fault (BL, Figure 2) that does not cut tween0.35 and 1.0 Ma basedon measured23øTh/23'*U and Pleistocenesurfaces) is equivalent to a horizontal shortening 234Th/238Uratios on the coral Balanophylliaelegans (B. of about 3.3 mm/yr, which is somewhatless than the mini- Szabo, written communication to W. Miller III, April 1985). mum shorteningrate of 5.2 mm/yr based on offset of Falor- These ages for the youngest Falor sedimentssuggest a deposi- Franciscan contacts. tional interval for the Falor Formation of about 0.9-1.5 Ma Thrust faults of the Mad River fault zone are active. Not (assumingthat deposition commencedat about the time of the only do thesefaults offset late Pleistocenemarine terraces with volcanic ash). scarpsof 15-90 m, but one of the faults offsetsa Holocene From the above estimates, tectonic activity in the Mad fluvial strath terrace in the lower Mad River valley by 7 m. River fault zone was initiated about 0.9-1.5 m.y. after deposi- Mesoscopicfaults increasein intensity toward the thrust tion of the volcanic ash, or 0.4-1.0 m.y. before present. faults of the Mad River fault zone. These faults, measured Averaging this time interval, we assume that tectonic activity within 200 m of the individual major thrust faults, are low commenced about 0.7 Ma. angle and form a conjugateset (Figure 5, sites A and B). If horizontal shortening spanned the last 0.7 m.y., then the Where beddingin the Falor Formation is horizontal,the con- KELSEYAND CARVER' QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4803

225

2OO

KIN LEYVI LLE 8• TRI N I DAD FAULTS MAD RIVER FAU • 150 E

/ Li.I 1•'5 / /FICKLEHILL FAULT

<[ 100

/ / n 75 / ß MAD RIVER FAULT / ß TRINIDAD FAULT 0 Mc KINLEYVILLE FAULT 50 ß FICKLE HILL FAULT

0 50 100 150 200 250 TIME BEFORE PRESENT,(ko)

Fig. 4. Dip-slipdisplacement magnitudes for faultsof the Mad Riverfault zone (y axis)plotted against terrace ages (x axis).Dip-slip displacement calculated from vertical terrace offset and an assumedfault dip of 25ø. Dip-slip estimates for the Fickle Hill and Mad River faults are minimum estimatesbecause the downfaulted side of the terrace is buried beneath Holocenealluvial deposits.Terrace age determinationsdiscussed in text. jugate setis symmetricabout a verticalplane (Figure 5, siteB). Thesefolds probably originated above faults that did not orig- Where Falor Formation bedding is inclined, conjugate sym- inally ruptureto the surface,and the foldspropagated as the metry of faultsabout a verticalplane is restoredby rotation of fault grew stratigraphicallyupward [Suppe,1985]. The prox- these features with bedding as bedding is rotated back to imity of the HumboldtHill anticlineto the northwestend of horizontal(Figure 5, siteB unrotatedversus B rotated).At site the Little Salmonfault providesan exampleof a likely fold- B, not only is conjugatesymmetry restored by rotation, but thrust fault pair. the mean fault pole directionsfor the fault pole clustersare The Little Salmon fault is the major thrust fault of the both definedwith a smaller•z95 confidence limit [Fisher, 1953] southern Humboldt Bay area [Ogle, 1953] (Figure 2). The after rotation (Table 2). character of this fault has been investigatedby Woodward- From the structural data we infer that these mesoscale low- ClydeAssociates [1980]. The dip of the Little Salmonfault anglereverse faults formed prior to foldingrelated to move- rangesfrom 20ø to 30ø . Progressivelysmaller amounts of dis- ment on the major fault. The mesoscalefaults formed only placementon successivelyyounger beds indicate that the duringincipient movement on the masterfault, at a time when Little Salmonfault has undergonerepeated displacement since strain was more evenlydistributed across the 500- to 1000-m- the late Pleistocene. Total cumulative vertical displacement wide fault zone and before strain was concentrated on the near Humboldt Bay is approximately2 km, whichimplies that master faults of Figure 5. net northeast-southwest directed contraction due to thrusting just on the Little Salmonfault is at least4.3 km. Correlation Lower Eel River Valley and HumboldtBay Area of dated horizons from drill hole borings through the hanging The lower Eel River valley and the adjoining Humboldt wall and footwall of the fault indicates a minimum slip rate for Bayarea show abundant evidence of northeast-southwestcon- the fault in the last 0.8 m.y. of 1.3 mm/yr. Correlation of other traction. The late Tertiary and Quaternary Wildcat Group horizonsyields slip rates as high as 2.2 mm/yr. Other faults [Ogle,1953; Woodward-ClydeAssociates, 1980] is deformed associated with the Little Salmon fault are the Goose Lake in a seriesof west-northwestto northwest trending folds. Most fault (Figure2) and the Bay Entrancefault [Woodward-Clyde notable of theseis the Eel River syncline(Figure 2), which has Associates,1980]. In a generalsense, the Little Salmon fault been a site of depositionsince at least mid-Miocene[Ogle, and associated structures have accommodated northeast- 1953]. Other folds includethe Alton anticline,Table Bluff southwest directed contraction in the same manner as faults anticline,and Humboldt Hill anticline(Figure 2) [Ogle, 1953]. and folds of the Mad River fault zone farther to the north, Late Pleistocene marine and fluvial terraces are cut on the thoughthe directionof maximumshortening has a slightly flanks and noses of all these folds, and these surfacesdip in a more northerly orientation. fashionsympathetic to, but lesssteeply than, the Wildcatsedi- Contraction in the Eel River valley-Humboldt Bay area ments,suggesting the foldshave been growing throughout the appearsto be active,based on warpingof late Pleistocene Pleistocene.Low-angle reverse and thrustfaults cut the flanks surfaces. The last movement on the Goose Lake fault zone of thesefolds [Woodward-ClydeAssociates, 1980] (Figure 2). was lessthan 6300 years B.P. [Woodward-ClydeConsultants, 4804 KELSEYAND CARVER; QUATERNARY TECTONICS, NORTHERN CALIFORNIA

124 ø o I •o 2o I •o I • I I , I C (U N ROTATED)'

A (U N ROTATED):

D (UN ROTATED): B ( U N ROTATED): n:144 n:68

+ .•+ + + + + + + + 41 o + + 41 o

+ ++ +++

+ +

+ +, ++

4. + • -•-•+'•!

+ i•. +++

+ 4. + 4. + + • + +

(R(•ATED) \ \ ß .. . '":'" '"!"'-::ß 'i:?;.::::.:.. . , \ " " ß '"''.: . '.'." :::i.' . CAPE <•.' ...;..! ;'i'.: •Y MENDOCINO ..'..'"'.':'" • •

124 ø

Fig. 5. Map of the Mad Riverfault zone and the northernterminus of the EatonRoughs fault zoneshowing pole-to-planelower hemisphere spherical projections of mesoscalefracture and fault plane data. Unrotated data are raw datafrom the outcrop. Rotated data are rotated with bedding as bedding is rotatedto horizontal.Original bedding was horizontalat sitesA andC. Seefurther discussion in text. BCL, Bridge Creek lineament; BMF, Bald Mountain fault; HB, Humboldt Bay: MC Maple Creek; MM, Murphy's Meadows.

1980], on the basis of a •'•C date on charcoal in faulted lake the Garbervilleunit (northeasternoutcrop belt) (Figure6). deposits. Trenches excavated across the Little Salmon fault Biostratigraphicstudies (diatoms and mollusks)by Menack showfolded and faulted valley fill sedimentsof Holoceneage. [1986] indicate that the Kimtu is older than the Garberville (lateearly to middleMiocene versus late Mioceneto Pliocene, Garberville Sediments and Dean Creek Fault respectively),but the upper ages of both units are un- The basic structural geology of the Garberville sediments constraineddue to erosion.Our mappingindicates that struc- (Figure 6) was first recognizedby MacGinitie [1943]. In a turallythe two unitsare difficultto distinguish.Each outcrop more thorough study, Menack et al. [1985] and Menack belt is in depositionalcontact on top of the Franciscanto the [1986] divided thesesediments into two informallynamed southwest and in fault contact with the Franciscan to the units,the Kimtu unit (thetwo southwesternoutcrop belts) and northeast(Figure 6). Each belt has a regionalstrike to the KELSEYAND CARVER: QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4805

TABLE 2. The o•95Confidence Limits for FisherMeans of posited" in a matrix of Temblor debris, a situation that can Clusters of Poles to Mesoscale Faults: Figure 5, Site A: most reasonably be explained if an overlying Franciscan unit Unrotated Fault Data Versus Rotated Fault Data erosionally downwasted (through a low-angle fault) into a O•95Confidence 0•95Confidence lower horizon of Temblor sediment, and (3) two three-point Pole Limit for Limit for solutions on the faulted Temblor contact taken across stream Cluster Unrotated Cluster Rotated Cluster valleys indicate fault dips of 6ø and 18ø. The amount of dip slip on the fault is unknown. The main body of Temblor SW quadrant 3.9ø 3.3ø sedimentspinches out to the north, possibly becausethe entire NE quadrant 4.4ø 3.9ø thickness of exposed sediments(elsewhere at least 400 m thick) has been thrust under the Franciscan melange along the Tem- blor fault. The Temblor fault postdates deposition of the Tem- northwestand dips moderatelyto the northeast.Beds are lo- blor Formation, but its timing cannot be further constrained. cally foldedand overturnednear the northeastfault contactin each unit. We presumethat this folding and local overturning TRANSLATIONAL DEFORMATION are due to movementon the northeast bounding faults (Figure North of the latitude of the Mendocino triple junction, 6). translational deformation occurs along a narrow zone within None of the three northeast bounding faults are exposed, the forearc, both to the east and the north of the zone of and fault dip is uncertain from direct observation.One of contraction. South of the approximate latitude of the triple these faults, the Dean Creek fault, truncates the Garberville junction, the San Andreas fault systemconsists of a broad, 65- unit (Figure 6). Mesoscalestructures increase in density to 80-km-wide belt of translational deformation. toward the Dean Creek fault both along the South Fork Eel River and the Alderpoint Road. Pole-to-planespherical pro- Grogan-Lost Man Fault Zone jectionsof mesoscalefaults at thesesites (sites A and B, Figure North of Big Lagoon, the Plioceneand Pleistocene(pre- 7) show distinctclusters of poles.These fault polesform two dominantlyPleistocene) Prairie Creek Formation [Kelseyand moderate to low-angle conjugate pairs, symmetric about a Cashman,1983; Cashmanet al., 1988] (top of Figure 2) out- verticalplane, only after the faultsare rotated with beddingas cropsin a 125 km2 coastalarea. The sedimentsare the on- beddingis rotatedto horizontal. shorepart of an extensivesequence of sedimentsdeposited on The patternof mesoscalefaults adjacent to the Dean Creek the continental shelf at the mouth of the Klamath River. A fault is similar to the fault pattern in the Mad River fault zone. prominentstructure in the unit is the northwesttrending, The conjugatepattern at Garbervilleis more diffusedue to an slightlynorthwest plunging Gold Bluffssyncline (GBS, Figure apparentsecond generation of faulting along the Garberville 2). The northeastlimb of the synclineis steeperand cut off by fault zone (discussedbelow). Despite the added complexity,a faulting. pattern of low-anglefaulting is associatedwith the Dean Three major megascalefaults, as well as a numberof minor Creek fault, and we infer that the Dean Creek fault is a thrust megascalefaults, offset the Prairie Creek Formation(Figure or low-anglereverse fault that placesFranciscan melange over 2). Two of these faults, the Grogan fault and the Sulphur Garberville sediments. As in the Falor Formation, the meso- Creek fault, are each exposedin a single exposure,and the scalelow-angle faults probably formed during initial contrac- Lost Man fault is not exposed at all. However, abundant tional deformation before the Dean Creek fault had substan- mesoscale faults and fractures occur in Prairie Creek sediment tial displacement.Though the other two faults that cut the in proximity to the Lost Man fault and to a lesserextent near two outcrop belts of the Kimtu unit are probably the same the Groganfault. These features are predominantlyhigh angle ageand samestructural style, they are poorly exposed.Stream and form conjugatesets [Kelsey and Cashman,1983]. Analysis bank exposuresof the fault cutting the southwesternmost of these faults and fractures in conjunction with mesoscale, Kimtu unit (Figure 6) suggestthat it is a low-angle reverse deformation-related normal faults, thrust faults, and folds led fault. Kelsey and Cashman[1983] to infer that the Lost Man and Groganfaults were right-lateral high-angle faults that havean Temblor Formation and Temblor Fault appreciable,but lesssignificant, component of reversedip-slip The Miocene Temblor Formation of Clark [1940], exposed offset as well. Because of extensive erosion, sense of fault offset near Covelo, California (Figure 8), is a sequenceof fluvial cannot be inferred from outcrop distribution alone. However, gravels,sands, and coal, interbedded with clayey,silty sands of the outcropdistribution of Prairie Creek sedimentsin relation estuarineand marineorigin. The formationis largelyobscured to the two faults constrainsright-lateral offset to lessthan 10 by vegetationand landsliding;exposures of fine-grainedsedi- km. ment are rare, and mesoscopicfault and fractures are seldom In addition to being associatedwith high-angle mesoscale observed. structuresthe Grogan and Lost Man faults have linear fault The structuralsetting of the Temblor Formation is similar traces over rugged topography. These characteristicsare in to that of the Garberville sediments. The formation strikes contrast to the imbricate thrust faults of the Mad River fault predominantlynorth-northwest, is depositedon the Francis- zone that have irregular traces over equally rugged topogra- can to the southwest, and is in fault contact with the Francis- phy and are associatedwith low-angle,mesoscopic faults and can to the northeast along the Temblor fault. We infer the fractures. The thrust faults of the Mad River fault zone trend fault contact to be low angle because(1) the outcrop pattern N35ø-40øW, whereasthe Grogan and Lost Man faults trend of the fault is highly sinuousand topographicallyconforms to N25øW and N10ø-15øW, respectively. an irregular,low-angle fault plane,(2) in summitareas of low Though we interpretboth the Grogan and Lost Man faults relief (where landslide depositscould not accumulateunless as predominantlyhigh-angle strike-slip faults, a reverseslip there was a topographicinversion during erosion),dense crys- componentis presentas well. The Grogan fault in the one talline blocks from the Franciscan melange appear "de- locality whereit is exposedshows a sharp fault contactwith 4806 KELSEYAND CARVER' QUATERNARY TECTONICS, NORTHERN CALIFORNIA

123o50 ' 123045 '

KJfm 40"10' N 40"10'

0 4000 8000 I , I •FE iE• I • I o t 2 I•,,,I,,,,l•l,,l•,, ,I GARBERVILLE KM UNIT KJfm

Tg \

KJfm• IIII

,,•i REDWAY • \ 5• 'I-H-

40"05' 40"05'

123045 '

Fig. 6. Geologicstructure map of the NeogeneGarberville sediments (Tg), in the vicinityof Garberville,California. Themap is designedto showoutcrop extent and structure in unitTg, thepre-Cenozoic geology is highlygeneralized. The Ty/KJfmcontact is takenin part fromR. J. McLaughlinet al. (manuscriptin preparation,1988), 1:100,000 Garberville 1ø sheetgeologic map. Ty, FranciscanYager sediments (Paleogene); KJfm, Franciscan melange. an attitude of N28'•W, 54"NE, and small fractures in sands right slip in the Tertiary [Kelsey and Hagans, 1982; Blake et beneaththe fault are mostly low angle. Regionally(within 1 al., 1985]. km of the fault), however, mesoscopicstructures associated South of Orick, California(O, Figure 2), Quaternarytrans- with the Grogan fault are high angle [Kelsey and Cashman, lational motion on the Grogan and Lost Man faults cannot be 1983]. There is thereforethe possibilitythat dip-slip motion demonstrated because Pliocene or Pleistocene cover sediments may be as importantas strike-slipmotion for the Grogan fault are absent.The south-southeasternextent of the Grogan fault where it cuts Prairie Creek sediments. as a possibly active Quaternary structure is unknown. The Both the Grogan and Lost Man faults are major fault zones tectonicmap (Figure 2) showsthat the Eaton Roughs-Lake that appear to be reactivated Mesozoic tectonic features. The Mountain fault zone is the easternmostactive Quaternary Grogan, in particular, has accommodatedupward of 75 km of structure south of latitude 40ø30'N. It is possible,however, KELSEYAND CARVER' QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4807

12:3ø 45' I I A (UN ROTATED): A (ROTATED): B (U N ROTATED): N n=112 N n :112 N n=30 40 ø t5' -- 40 ø t5'

C(UN ROTATE D)'•..''.•' '•'..''.,...'."..ß.'...':.•'•'•".';•

i•;...•..i.'..:.:..!.•,_ S (ROTATED): ß F•..i...'.. '•.:GARBERVILLE

'<'::::• -•'• ':•:f:;.;'"". 0 I , •KM• • 5I N

D (U N ROTATED): D (ROTATED)' N n=67 N n=67

40 ø -- 40 ø

c (ROTATED)=

I I

Fig. 7. Pole-to-planelower hemispherespherical projections of mesoscalefracture and fault planedata for sitesalong the faulted marginsof the Garberville sediments(Tg). Unrotated data are raw data from the outcrop. Rotated data are rotated with bedding as bedding is rotated to horizontal. See discussionin text. that the southern part of the Grogan fault, or related faults, cal isolated tabular graywackebody, ShowersRock. The trace has accommodated dextral translational strain in the Quater- of the Eaton Roughs fault zone separatesthese bodies (Figure nary. 9). The northern 12 km of the Eaton Roughs fault zone consist Eaton Roughs Fault Zone of two straight tracesthat trend north-northwe;stinto the most The Eaton Roughs fault zone is defined by numerous un- landward (easterly)exposures of the Falor Formation (Figures draineddepressions, linear drainage segments, and instances 2 and 3). The Falor sedimentsare vertically offset (and per- of unstable ground that together define a north-northwest haps laterally offset as well) by at least one trace of the Eaton trending straight trace for 66 km from Maple Creek south- Roughs fault, demonstratingthat there has been movement on southeast to its merger with the Lake Mountain fault zone the fault since 0.4-1.0 Ma. [Kelsey and Allwardt, 1983] (Figure 2). The fault zone is W e investigateddeformation at the transitionof the Mad named after Eaton Roughs, a distinctive tabular body of re- River fault zone and the Eaton Roughs fault zone by ana- sistant Mesozoic graywacke sandstone north of Dinsmore, lyzing the mesoscalefaults in the easternmostFalor sediments California, which is offset right laterally 3.3 km from an identi- near Maple Creek compared to those in the western Falor 4808 KELSEYAND CARVER' QUATERNARY TECTONICS, NORTHERN CALIFORNIA

125ø20 ' 123o15 ' I

Qal

;0 0 KJf I i i z COVELO• iii ( Mt

I-- I-- rrl .-<

\ Kdf ß.. Oal ß ß

Kdf 39045 ' ••/ 74::::0

0 4000 8000 • •5 AuT I , I , I • I • I • • FEET '- • ,,,,, ,, , ,, N .... KM...... )

I ', I 123020' 123015' Fig.8. Geologicmap of theMiocene Temblor Formation (Mt) nearCovelo, California. The map shows outcrop extentand structure in unitMt' thepre-Cenozoic geology is highlygeneralized (KJf, Franciscan assemblage). Also shown aretwo tracesof theLake Mountain fault zone. One of thesetraces right laterally offsets unit Mt andthe Temblor fault. sedimentsnear the coast.As already discussed,the Falor sedi- C and D, Figure 5) are mainly high angle, and restoration of ments near the coast are offset by megascale,imbricate thrust thesesediments to horizontal beddingdoes not restorethe faultsand mesoscalelow-angle faults (Figures 2 and 3). Meso- faults and fracturesto a low-angle,conjugate pattern (site D, scalefaults and fracturesin the easternFalor sediments(sites Figure 5). To the contrary, restoration of these features to KELSEYAND CARVER: QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4809

123ø40 '

0 I 2 3 4 5 I I i I I I KILOMETERS

EATON ROUGHS DSTON E BLOCK

40o30 ' 40ø30 ' DINSMORE ß

UNDRAINED • DEPRESSIONS'"'•

BURR VALLEY

LARABEE VALLEY

I 3ø40 ' Fig. 9. Generalizedgeologic map of the Van Duzen River basin near Dinsmore,California, area showingthe 3.3-km right-lateral offset of the lithologicallysimilar Eaton Roughsand ShowersRock sandstoneblocks. Entire map area is underlainby Franciscanassemblage sandstone or melanget•nits. Alluvial valleysand undraineddepressions of probably tectonic origin are also shown. horizontal bedding creates a random pattern of the poles to younger than inception of thrust faulting on the Mad River planes, whereas the unrestored structural data show one or fault zone. If our interpretation is correct, high-angle faults of two clusters of poles to high-angle fault planes (Figure 5). the Eaton Roughs fault zone near Maple Creek must cut ear- From this we infer that the high-angle faults formed after lier low-angle structures of the Mad River fault zone. Unfortu- folding. We found insufficient data to determine if the two nately, these relations are not exposed in the field. Farther clusters represented a conjugate set. However, nearly horizon- northwest (toward the coast) along the Mad River fault zone, tal grooves and striations observed on some high-angle meso- the two deformations are not superimposed.Hence right slip scale fault planes near Maple Creek suggestthat these struc- on the Eaton Roughs fault zone is probably occurring simul- tures are part of a systemof strike-slip faulting. taneouslywith thrusting along the coastalportion of the Mad River fault zone. Tectonic Transition at Northwest End A large component of translational displacementon the of Eaton Roughs Fault Zone Eaton Roughs fault zone is probably being accommodatedby Based on the mesoscale structural data, we infer that a contraction on the Mad River fault zone. Net horizontal transition in tectonic style occurs from contraction near the shortening across the Mad River fault zone, from basement coast along the Mad River fault zone to translation inland offset, is at least 3.6 km (seeabove). Lateral offset of the north- along the Eaton Roughs fault zone. The transition occursnear northwest trending Eaton Roughs from Showers Rock is Maple Creek (Figures 2 and 5). about 3.5 km, suggestingthat the Eaton Roughs fault zone in Falor sedimentsalong the Mad River fault zone are locally part accommodatesnortheast-southwest directed contraction. overturned in tight folds near the thrust faults. This over- However, it is unlikely that all translational motion is dissi- tuning is evident in the Falor near Maple Creek as well, where pated in contraction on the Mad River fault zone, for three high-angle mesoscalefaults are superimposedon the folds. reasons. First, there are at least two strands (and possibly From this relation we infer that inception of strike-slip fault- more) of the Eaton Roughs fault zone (Figure 3), and it is ing on the Eaton Roughs fault zone near Maple Creek is likely that right-slip displacementhas occurred on multiple 4810 KELSEYAND CARVER:QUATERNARY TECTONICS, NORTHERN CALIFORNIA strands. Second, 8 km north of Maple Creek, a subsiding to the Eaton Roughs fault zone. The elongate, roughly rhom- basin called Murphy's Meadows (Figures 2, 3, and 5) occurs at boidalshape of the Salt Creekdepressidn and its positionin a reasonable location for a pull-apart basin involved in the the fault zone are both features similar to other megascale right step of deformation from the Eaton Roughs fault zone to depressions observed along dilational fault jogs in trans- faults of the Grogan fault zone. Third, the Prairie Creek sedi- current fault systems [Sibson, 1986]. ments to the north-northwest have been offset by right-lateral The Lake Mountain fault zone is the northward continu- translational strain that must be related to similar strain to ation of the Bartlett Springs fault zone, which was recognized the south-southeast. This strain is either being taken up on the by McLau•thlin and Nilsen [1982] and McLau•thlin et al. Eaton Roughs fault zone or possibly on a southern (and [1985a] to be a major fault zone of the San Andreas fault poorly exposed)portion of the Grogan fault. system.Earthquake focal mechanism solutions from seismicity Transfer of strain north-northwestward from the Eaton along the Barlett Springs fault zone suggestright slip [Dehl- Roughs fault zone to the Grogan-Lost Man fault system is inger and Bolt, 1984]. The fault zone offsets Quaternary sedi- attractive because northwest of Murphy's Meadows there are ments near Lake Pillsbury [DePolo and Ohlin, 1984]. Clark a series of prominent linear features on the west slope of the [1983] mapped 0.9 km of right-lateral offset for a distinctive Redwood Creek basin that progressively step to the right to Franciscan melange block along the Bartlett Springs trend the Grogan fault, which offsets Prairie Creek sedimentsnorth and noted a series of small basins along the trend that he of Orick, California [Harden et al., 1982] (Figure 2). Along interprets as pull-apart basinsrelated to right slip. one of these linear features, the Bridge Creek lineament (BCL, Our mapping of the Miocene Temblor Formation near Figure 2), a Pleistocene alluvial fill up to 50 m thick, is in- Covelo indicates that these sedimentsmay be offset right lat- tensely fractured and faulted. The fractures are uniformly high erally about 2 km along a lineament at the southernend of the angle, form a conjugate set, and have the same pattern as Lake Mountain fault trend (Figures 3 and 8). Though ex- those adjacent to the Lost Man fault [Kelsey and Cashman, posures are poor, the low-angle Temblor fault is apparently 1983], suggesting that faulting and some translational dis- offset along the lineament. A megascale fold, seen in regional placement have occurred along the Bridge Creek lineament outcrop pattern and documented by bedding orientation in during the Pleistocene. the eastern outcrop of Temblor sediment, probably developed Lake Mountain Fault Zone during right slip (Figure 8). Fragments of Temblor sediment no more than 500 m 2 in area are smeared out in the Francis- The Lake Mountain fault zone was originally mapped by can rocks near the fault trend between the two offset bodies of Herd [1978] as extending from near Covelo north- Temblor sediments (Figure 8). Offset of the Temblor places a northwestward all the way to Maple Creek (Figure 3). We maximum age of post-middle Miocene on the timing of fault- have subdivided the fault zone into the Lake Mountain fault ing. However, becausethe offset occurs along the Lake Moun- zone and the Eaton Roughs fault zone becausethe fault pat- tain trend, we infer that the offset is as young as latest Tertiary tern and the fault trend of the two fault zones are distinctively or Quaternary. different.The changein pattern occursnear Kettenpom, Cali- fornia (K, Figure 3). Garbert;ille Fault Zone Evidence for both Holocene activity and right slip is more abundant for the Lake Mountain fault zone than for the The Garberville fault zone is a discontinuous series of Eaton Roughs fault zone. Herd [1978] first recognizedthe north-northwest trending lineaments that extend south- striking geomorphicevidence for youthful faulting along the southeast from Bull Creek, through Garberville, to just north Lake Mountain fault zone at Lake Mountain (LM, Figure 3). of Laytonville (Figure 3). At its southern end the Garberville Abundant linear sag ponds, aligned drainages,topographic fault zone mergeswith the Maacama fault zone. In this region trenches,and saddlesdefine four separatenorth trending, sub- our fault lineament mapping overlaps by 16 km with that of parallel faults at Lake Mountain. Both to the north-northwest Pampeyan et al. [1981], and there is good agreement on the toward Zenia (Z, Figure 3) and to the south-southeasttoward location of major fault traces.There is no evidenceof Holo- Round Valley (Figure 3), similar geomorphicfeatures are pres- cene surfacerupture along the Garberville fault; however, at ent. Unlike the Eaton Roughs fault zone that consistsof one least one surfacerupture has occurred along the Maacama in or two well-defined continuous fault traces, individual faults Holocene time [Upp, 1982] and active right-lateral fault creep on the Lake Mountain trend are more numerous, extend from has been recorded on the northern section of the Maacama 1 km to more than 30 km in length, and both left step and fault zone [Harsh et al., 1978]. right step toward the north-northwest(Figure 3). North-northwest of Garberville, the Garberville fault zone The transition from the Lake Mountain fault zone to the can be traced as a 4-km-long, north-northwest trending zone Eaton Roughsfault zone near Kettenpom entailsa changein (•, 200 m wide) of sag ponds, notched ridges, and aligned fault zone trend from N20ø-25øW to N30ø-35øW, respectively. springs(Figure 3). Intermittently, a well-definedtrace is vis- The two fault zones overlap in the vicinity of the Kettenpom- ible. Farther north, the Garberville fault zone in some manner Hoaglin valleys(KHV, Figure 3) and in the encompassingSalt probably connectsto or offsetsthe Russ fault north of Cape Creek depression(SCD, Figure 3). The Salt Creek depression Mendocino or the Capetownfault (Figure 2), though a lack of is an elongate,40 km2 topographiclow. Drainagein the Salt late Neogene or Quaternary cover sedimentsand an absence Creek depressionis anomalouslytoward the north-northwest of well-defined lineaments prevent mapping this transition on (the master stream, the North Fork Eel River, flows south- aerial photographs.Near Garberville, the fault zone truncates southeast),and the depressionhas not undergonethe regional the depositional base of the Garberville unit along a set of uplift characteristicof the surroundingarea. We feel that the overlapping fault splays (Figure 6). One of these fault splays depressionis of tectonic origin and is due to localized exten- trends N70øW and dips 72øNE. sion in a 40 km 2 area associatedwith the changein fault trend Due to the alignment of the Garberville fault zone with the and right step of the Lake Mountain fault zone northeastward Maacama fault zone to the south, it is likely that the Gar- KELSEYAND CARVER:QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4811

berville fault zone is an active, right-lateral fault system. The vicinity. The principle compression axis in the North Ameri- Garberville fault zone subparallels the coastal belt thrust can plate near Cape Mendocino is north-south or northeast- [Jones et al., 1978; Underwood, 1982] for most of its length. At southwest based on seismic records for 1972 [Simila et al., two locations where the coastal belt thrust intersects one 1975] or about N40øE as derived from historical geodetic data strand of the Garberville fault zone, apparent right-lateral [$nay et al., 1987]. Simila [1980, p. 134] states that the "inland offset of the coastal belt thrust of a few kilometers is implied shear zones near Cape Mendocino indicate strike-slip fault- (Figure 3). These offsets could be the result of right-lateral ing" based on seismicity from 1968 to 1978. These observa- motion on the Garberville fault zone, but we have no field tions suggesta strike-slip component to the Russ and Cape- evidence to support this interpretation. An alternative inter- town faults, in addition to the vertical component indicated by pretation (R. J. McLaughlin and S. Ellen, oral communication, the marine terrace elevations. A third component of crustal 1987) is that the Garberville fault zone steps westward north- deformation is regional tilting, indicated by the southward tilt west of Garberville and is aligned with steep faults in the of the marine terrace remnants north of the Russ fault [Ogle, Franciscan coastal belt of that area (Figure 3). Therefore the 1953]. If compression is north-south or northeast-southwest, nature of the northward continuation of the Garberville fault then any strike-slip component on the east-westtrending Russ zone is speculative. and Capetown faults may be left lateral. However, movement Near Garberville, mesoscale fractures and faults are well on the Mendocino fracture zone due west offshore is right exposed in the Garberville unit at two river bank outcrops 100 lateral [McEvilly, 1966]. The opposite sensesof shear would and 400 m from the main fault trace. Analysis of these features imply weak coupling of the overriding and subducting plates shows high-angle fault clusters (Figure 7, sites C and D), and near Cape Mendocino. If this is the case, the interaction of the faults at site C form a conjugate pair. We infer that the high- two plates near Cape Mendocino is notably different from the angle mesoscalestructures near the Garberville fault zone are strong plate coupling that is inferred near Humboldt Bay (see genetically related to the fault zone and that the fault zone is discussion below). Based on existing geodetic, seismological, high angle. If fracturesare rotated to horizontal bedding (sites and geologic data, therefore, it is inconclusive whether the C and D, rotated versus unrotated, Figure 7), there is not a Cape Mendocino area is primarily deforming by lateral or notable difference in the degree of clustering of pole-to-plane compressionalstrain. We suspectthat the former is dominant, data. However, for the high-angle conjugate fault set (site C, but it is unclear whether lateral strain on mapped faults in this Figure 7), rotation to horizontal bedding increasesthe propor- region is dextral or sinistral. Additional work on the neotec- tion of low-angle (less than 45 ø) faults, thereby reducing the tonics of the area just north and east of Cape Mendocino is conjugate symmetry. Though field data are equivocal in this clearly needed. instance, we feel that the limited evidence suggeststhat the high-angle mesoscalestructures formed after folding. In con- San Andreas Fault Zone trast, mesoscale structures associated with the Dean Creek We did not investigate the San Andreas fault zone because fault are moderate to low angle and formed prior to folding the fault zone is offshore in northern California. A possible (see discussion above). exception is an onshore fault segmentat Point Delgada where The Dean Creek and Garberville faults, though subparallel tectonic ground rupture occurred during the 1906 earthquake in trend, are therefore inferred to have different dips and a [Brown and Wolfe, 1972]. Based on the dating of adularia different sense of movement based on associated mesoscale veins that crossthe supposedfault trace, however, McLaughlin structures. Consequently, these faults probably represent two et al. [1985b] concluded that tectonic offset may not have generations of faulting, the first in a contractional strain occurred in 1906 at Point Delgada. regime, the secondin a translational strain regime. Net right-lateral offset along the combined San Andreas- An alternative possibility is that these low-angle and high- Hosgri-San Gregorio fault trends since post-early Miocene is angle mesoscale fracture sets represent a single deformation approximately 440 km (Fox et al. [1985], compiled from nu- event (i.e., the same tectonic stress field). Though mesoscale merous studies). Net right-lateral slip on the Eaton Roughs- faults of different types can be due to a single tectonic stress Lake Mountain fault zone, and possibly the Garberville fault state (Angelier et al. [1985] and Frizzell and Zoback [1987] zone, is minor (< 5 km) compared to late Neogene right slip cite field examples), we feel that this is unlikely at Garberville on the San Andreas. However, right-slip rates on these three because the two mesoscalefault types are for the most part fault zones in the last 0.1 m.y. may be comparable. spatially separated and best developed adjacent to either one fault or the other. SPATIAL PATTERN OF CONTEMPORANEOUS DEFORMATION

Deformation in northern coastal California consists of a Russ and Capetown Faults forearc contractional zone, characterized by thrust faults and Based on aerial photographs, both the Russ [Ogle, 1953] associated folds, and a zone of lateral translation, and Capetown faults appear to offset by approximately 70-m characterized by right-lateral strike-slip faults (Figure 10). The remnants of the most prominent late Pleistocene marine ter- northern portion of the translation zone is within the forearc race surfaces at the coast near False Cape and Cape Men- and overlies the subducted Gorda slab. The southern portion docino, respectively (Figures 2 and 3). Mesoscale structures of the translation zone overlies the slab window and is the adjacent to the Russ fault are well exposedin a massive,sub- northernmost segmentof the transform boundary between the horizontal 2-m-thick sand bed in Wildcat Group sediments North American and Pacific plates (Figure 10). Of the three [Ogle, 1953] along Wildcat Grade. The structures are moder- distinct right-lateral fault zones that make up the northern- ately to steeply dipping (65ø-90ø), but no pattern is apparent, most part of the transform boundary, the two western fault probably becauseof more than one generation of faulting. zones (but not the San Andreas fault zone) continue to be well We are uncertain of the sense of motion on these two faults defined tens of kilometers northward into the forearc. because of the paucity of Neogene cover sediments in the The eastern boundary of the translational zone is taken to 4812 KELSEYAND CARVER' QUATERNARY TECTONICS, NORTHERN CALIFORNIA

125 ø 124 ø 42 ø I

.:./:...:•ß 1.FOREARCCONTRACTION

• 2.FOREARCTRANSLATION •$. MTdREGION

i•4.ß TRANSLATIONSANANDREAS

41 ø 41 ø

MENDOCINO TRIPLEJUNCTION

40 ø 40 ø

0 20 40 60 I I I I '•

125 ø 124 ø t23 ø Fig. 10. Tectonic map of California showing the extent of contractional and translational zones. Offshore data from Clarke [1987]. Region 1, zone of forearc contraction; region 2, zone of forearc translation; region 3, zone of translation within the San Andreas transform zone; region 4, Mendocino triple junction region. See text for further explanation. LMFZ, Lake Mountain fault zone; ERFZ, Eaton Roughs fault zone; GF, Grogan fault; BMF, Bald Mountain fault; BLF, Big Lagoon fault; LMF, Lost Mountain fault; BMFZ, Bald Mountain fault zone. South edge of Gorda plate after Jachens and Gristore [1983] and Wilson [1986]. be the easternmost limits of the combined Lost Man-Grogan- faults appear on acoustic reflection profiles to be steeply east Eaton Roughs-Lake Mountain fault zones, which together dipping to vertical [Clarke, 1987] (Figures 10 and 11). The define the eastern limit of distinct linear features and land- faults extend to or near the seafloor and offset Quaternary forms associated with youthful faulting. Farther to the east, sediments.Basement offsets suggestedby the acoustic reflec- pronounced north-northwest linear features persist in the tion data are 100-1300 m up to the northeast [Clarke, 1987]. landscape, but they are not associated with the Quaternary Right slip is suggestedalong the faults by their steep dips, landforms characteristic of recently active faults. The northern relatively straight traces, variation in amount of apparent extent of translation is unknown, but translational defor- offset along the traces, and en echelon folds associated with mation on land narrows toward the coastline to a 10-km-wide the fault traces [Clarke, 1987]. belt at the mouth of the Klamath River (KRM, Figure 2). The contact between the contraction and translation zones On the basis of structural mapping on the continental shelf is moderately well defined both onland and offshore. The [Clarke, 1987], we correlate mapped faults on land to offshore contraction-to-translation transition on the continental shelf is structures (Figures 10 and 11). Faults that are correlatable within a north-northwest trending transition zone between the with the Lost Man, Grogan, Bald Mountain, and Big Lagoon Trinidad fault zone and the Bald Mountain fault zone KELSEYAND CARVER: QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4813

128 ø 127 ø 126 ø 125 ø 124 ø I 43 ø 43 ø

NORTH AMERICAN PLATE

OREGON 42 ø 42 ø CALIFORNIA

GOR PLATE

41 ø 41 ø

ENDOCINO FRACTUR

MENDOCINO IPLE JUNCTION 40 ø 40 ø

oI I .oI 60I PACIFIC PLATE KM

28 ø 127 ø 126 ø 125 ø 124 ø Fig. 11. Tectonicmap of northcoastal California (from Figure 2) shownin contextof the Pacific,Gorda, and North Americanplate boundaries and the deformation of themagnetic anomalies in the southGorda plate. Magnetic data and southedge of Gordaplate from Figure 6 of Jachensand Griscom [1983]. Offshore structure data from Clarke [1987].

[Clarke, 1987] (Figure 11). From the latitude of Big Lagoon offsetsNeogene or youngersediments. The fault appearsto be (BL, Figure 2), the transition boundary heads north- high anglewithin 5 km of the coast[Cashman et al., 1986-1, northwestward,then northward, and by the latitude of the but the sinuousoutcrop trace of the fault suggeststhat it may Klamath River, it is subparallelto the coastlineabout 30-35 not be high anglefarther inland (Figure 2). km offshore. To the south, the contractional zone is bounded by the On land, this transitionalcontact defines a semicirculararea Mendocinotriple junction region(Figure 10).This area shows of contraction centered about Humboldt Bay. The onshore evidence of both late Pleistocene faulting and broad crustal and offshoredata combinedtherefore suggest that the zoneof tilting.The sparsedata providedfrom seismicity[Simila et al., contractionresulting from Gorda-North American plate is 1975; Simila, 1980] and from field study of the Russ and widestjust north of Cape Mendocinoin the vicinityof Hum- Capetownfaults suggest that translationdominates the defor- boldt Bay (Figure 10).The transitionbetween contraction and mational style. translation is well defined for the Eaton Roughs-Mad River Maps showingepicentral distribution of seismicityfor the fault zone. However, the northern end of this on-land transi- area roughlyencompassed by Figures2 and 3 for the period tion is poorly defined.Little is known about either the Big 1980-1982[Cockerham, 1984; Walter, 1986] showa patternof Lagoonor Bald Mountainfaults. The Big Lagoonfault offsets shallow depth earthquakesthat, in a broad sense,reflect and foldslate Neogeneand Quaternarysediments near Big mappedtranslational faults on the northernpart of the San Lagoon.The straight trace suggests that the Big Lagoon fault Andreastransform zone. The most striking correspondence is ishigh angle (Figure 2), though mesoscale structural data near thetwo pronounced north-northwest linear trends of epicen- the fault are not available.The Bald Mountainfault nowhere ters that correspondto the Maacama and Bartlett Springs 4814 KELSEY AND CARVER: QUATERNARY TECTONICS,NORTHERN CALIFORNIA

fault zones.Near latitude39ø50'N, seismicity is alsoprevalent faunal age of this unit. Faunal agesof the Temblor and Gar- along the southernmostmapped traces of the Garberville and berville sedimentsoverlap in the middle Miocene, but Tem- Lake Mountain fault zones.However, seismicity does not re- blor depositionterminated in the Mioceneand depositionof flect thesetwo mappedfault tracesfarther north (north of the Garberville unit persisted until at least the late Pliocene. latitude 40øN) in the zone of forearc translation. Thus thesesedimentary packages tend to becomeyounger to Seismicity is frequent but scattered beneath the contrac- the north.

tional zone south and southeastof Humboldt Bay [Walter, DISCUSSION 1986]. Most earthquakesoccur at a depth of 10-30 km. This broadzone of seismicitydies out to the north and east[Smith Pattern of Deformation and Knapp, 1980; McPhersonet al., 1981], and the coastal Late Neogene and Quaternary crustal deformation in portion of the Mad River fault zone shows a low level of northerncoastal California is dividedinto two strainregimes: seismicity.In general, seismicityin the contractional zone contraction and lateral translation. Lateral translation occurs cannot be related to mappedsurface faults, despite much evi- along the San Andreas fault system above the slab window denceof Quaternaryfaulting. The scatteredhypocenters in the south of the subductedGorda plate. Strike-slipfaulting also contractionalzone may in part reflectdiffuse seismicity on the occursin the forearc region north of the triple junction as a shallowdipping thrust faults. However, the depthof the hypo- linear zone (8-15 km wide) arcward of the contractingwedge centers [Walter, 1986] suggeststhat most of the broad seis- of sediment at the continental margin. This thin zone, which micity in the contraction zone reflectssubduction and internal includes the Eaton Roughs-Grogan-Lost Man fault zone, is breakup of the Gorda plate at depth.Because the thrust faults typical of transcurrent fault zones within the forearc region do not manifestfrequent seismicity (they have been histori- landward of obliquely convergentsubduction zones [Fitch, callyquiescent), they must release stress more infrequently in 1972; darrard, 1986]. The forearc strike-slipfault systemcon- large, shallow focus earthquakes.The recurrenceinterval of nectsto the south with faults of the San Andreasfault system. faultswith late Pleistocenescarps in the vicinityof Humboldt The forearcfaults do not exhibitthe abundantmicroseismicity Bay [Woodward-ClydeAssociates, 1980] supportsthis conten- characteristic of those translational faults to the south above tion. the slab window. As the slab window migratesnorthward, Seismicitydecreases to the east of Humboldt Bay and the theseforearc strike-slip faults will probably becomethe active contraction-to-translationtransition occurs above the region faultsof the San Andreastransform plate boundary,as sug- where,at depth, the south-eastwarddip of the Benioff-Wadati gestedby Kelseyand Cashman[1983]. Geologicdata, present- zonenotably steepens [Cockerham, 1984; Walter, 1986]. ed above, show that the forearc strike-slipfaults have offset PleistoceneFalor and Prairie Creek sedimentsand they are probably active faults despite the lack of microseismicity TIMING OF DEFORMATION along their traces. Faunal ages aid in constrainingthe timing of deformation As dictated by plate convergence,contraction occursin the of the Neogenesedimentary units. The southernmostunit, the forearc region immediately landward of the subduction zone. Temblor Formation, is middle Miocene on the basis of fossil The Dean Creek thrust fault and the Temblor fault also teeth and oysters[Clark, 1940]. Deposition of the Garberville recordcontraction, but thesefaults are presentlysouth of the sediments(the next Neogeneunit to the north) spansthe inter- Mendocino triple junction, near the northern margin of the val late-early-to-middle Miocene for the Kimtu unit and late slab window, and well within the San Andreasfault system. Miocene to Pliocenefor the Garberville unit [Menack, 1986]. We infer that this contractionpredates San Andreasright-slip Menack 1-1986]cites evidence for minor syndepositionaluplift motion. and tilting of just the Garberville unit, but the generaluniform Herd [1978] suggestedthat right-slip faults of the San An- dip of both the Kimtu and Garberville units of 15ø-40ø north- dreas fault system north of San Francisco Bay defined the west (Figure 6) suggeststhat the major episode of contrac- sliver-shapedHumboldt plate, which is moving northward tional deformationoccurred after the last phaseof deposition relativeto stableNorth America.Based on our fault mapping in the late Pliocene(about 2-3 Ma) but beforeonset of trans- and limited offset data, we do not see evidencefor a rigid lational deformation related to the Garberville fault zone. The plate. The sliver appearsto be internally deformingby right slip on the Maacama and Garberville fault zones at rates upper Garberville unit is a regressive sequence I-Menack, 1986], and we infer that deposition in the Garberville unit approximatelyequal to right slip on the east margin of the sliver along the Bartlett Springs-Lake Mountain-Eaton ceaseddue to uplift and contraction of the basin in a fashion Roughs fault zones. Perhaps deformation across the Hum- similar to the tectonic demise of the Falor basin. boldt plate betweenlatitudes 39ø45'N and 40ø15'N(Figure 3) Falor Formation deformation probably did not commence is similar to that suggestedby Prescottand Yu [-1986] for the until about 0.7 Ma for reasons discussed above. Contractional region immediately north of San Francisco Bay, where plate deformation of the Falor Formation is at least as recent as motion is distributed over a wide zone. Thus the Humboldt latest Pleistoceneand probably is active. plate, better called the Humboldt deformation zone, is an The late Neogene and Quaternary tectonic history of the elongatezone internally deformingby right slip and also being Temblor and Garberville sediments is similar and consists of displaced by right slip from stable North America. There are (1) a period of contractional deformation following deposition compellingarguments based on lithosphericstrength [Vink et in which the units were folded and thrust under Franciscan al., 1984] and subductiongeometry [darrard, 1986] that conti- sediments, and (2) subsequent right-lateral deformation. De- nental slivers such as the Humboldt deformation zone can formation in the Falor is constrainedto within the last 1 m.y. becomedetached and migrate great distancesas displacedter- Major deformation of the Garberville sediments commenced ranes. The Humboldt deformation zone may therefore be an after about 2-3 m.y. Deformation in the Temblor Formation incipientterrane sliverjust starting (or continuing)its voyage is not geologically constrained, except as younger than the northward. KELSEYAND CARVER: QUATERNARY TECTONICS,NORTHERN CALIFORNIA 4815

Mechanism of Upper Plate Deformation contractile zone must migrate northward at the rate of migra- Near Humboldt Bay tion of the Mendocino triple junction in the last 3 m.y. In a The southernmost part of the zone of forearc contraction at longitudinal frame of referencethis migration was about 30.3 Humboldt Bay spatially coincides with the southeastern por- arc min of latitude/m.y. (about 56 km/m.y.) [œngebretsonet al., tion of the subducted Gorda plate (Figure 11). This is the 1985]. portion of the Gorda plate that is undergoing maximum inter- Based on the above triple junction migration rate and infer- nal distortion, based on both an elastic model of dachens and ences on the mechanism of deformation, we suggest that Griscom [1983] and a kinematic strain rate model of Wilson sequential episodes of contractional and translational defor- [1986]. The internally deforming downgoing slab is under- mation of the Temblor Formation and the Garberville sedi- neath the wide southernmost portion of the upper plate zone ments can be spatially and temporally correlated with the of contraction (Figure 11). We feel that this spatial overlap is passage to the west of the Mendocino triple junction and the best explained by coupling of the downgoing and deforming northward migration of the slab window. The timing for this Gorda plate with the overriding North American plate. This kinematic model is based on the inference that, first, Falor coupling is reflected in surface faulting by a rapid rate of Formation sediments at the northern edge of the contraction North American plate contraction in the last 1 m.y. (a mini- zone started to deform sometime between 0.7 and 1.0 Ma, mum of 5.2 mm/yr of NE-SW shortening on the Mad River when the Mendocino triple junction was 56-63 arc min of fault zone and a minimum of 1.3 mm/yr of NE-SW shortening latitude (104-117 km) south of the latitude of the Falor For- on the Little Salmon fault system) and in the fact that micro- mation (we assume here that North America is fixed in the seismicity in recent years (1974-1977, 1981-1982) is high in the geographic reference frame). Second, the Eaton Roughs fault vicinity of Humboldt Bay but dies off rapidly to the east of zone propagated northward and offset the Falor Formation longitude 124ø [Smith and Knapp, 1980; Cockerham, 1984]. sometime within the last 0.1-0.01 m.y. The inferred translational deformation on the Russ and Cape- In a similar fashion the Garberville sediments and the Tem- town faults may also reflect upper plate coupling to the lower blor Formation must have started to deform when the Men- plate, in which case the translational strain is right lateral, docino triple junction was about 104-117 km south of each of consistent with the sense of motion between the Gorda and these sedimentary units, respectively. At a migration rate of Pacific plates along the Mendocino fracture zone to the west about 56 km/m.y., deformation would have started approxi- [Couch, 1980]. Conversely, if coupling is not strong near Cape mately 1.8-2.1 m.y. before the Mendocino triple junction ar- Mendocino, the documented northeast-southwest contraction rived at the latitudes of these units. Given that the Garberville in the North American plate requires that any translation and Temblor sediments were opposite the triple junction at along the Russ or Capetown faults be left lateral. 0.4 and 1.2 Ma [Engebretson et al., 1985], these units started to deform at about 2.2-2.5 and 3.0-3.3 Ma, respectively(Table Seismic Hazard in Humboldt Bay 3), given our tectonic model. Figure 12 is a schematic portray- Gorda-North American plate coupling and the resulting al of the model, showing one possible reconstruction of triple significant amounts of shortening in the North American plate junction location along the California coast at the times when have implications for seismic hazard in the Humboldt Bay the Temblor, Garberville, and Falor sediments, respectively, area. A minimum value for the total northeast-southwest di- started to deform. rected shortening in the contraction zone is 7.9 mm/yr in The 3.0-3.3 Ma time for inception of contractional faulting about the last 0.7-1.0 m.y. This rate is between 25 and 100% in the Temblor Formation is close to the 3 Ma time when of the total convergence rate between the Gorda and North Gorda-Pacific relative plate motion changed, initiating sub- American plates in the last 5 m.y. or less, depending on the duction of a deforming Gorda plate. Contraction therefore time period and whether one employs the convergence values should not be an important deformational event for Neogene and azimuths of Riddihough [1984] or Engebretson et al. sediments south of the Temblor Formation. [1985]. In either case, contraction in the North American Neogene sediments of the Little Sulphur Creek basins occur plate has accommodated a significant amount of the strain farther south at latitude 38ø45'N. The sediments are exposed associated with subduction. This in turn must decrease the along the trend of the Maacama fault and were deposited in a amount of potential stressbuildup that could occur within the pull-apart basin [McLaughlin and Nilsen, 1982]. These basins subduction zone on the megathrust. Stress released in thrust formed and were syntectonically filled in an extensional envi- fault events within the forearc in the southernmost portion of ronment related to the passage of the Mendocino triple junc- the Cascadia subduction zone near Humboldt Bay may be tion to the west at approximately 3-4 Ma [McLaughlin and substantial [Carver and Burke, 1987]. Though geologic and Nilsen, 1982]. The extensional regime probably was due to seismotectonic data suggest a potentially great seismic hazard instabilities inherent in the migration of the triple junction for the rest of the Cascadia subduction zone [Heaton and [Dickinson and Snyder, 1979a], at a time prior to the relative Hartzell, 1987; Atwater, 1987], it remains unclear whether the motion change between the Gorda and the Pacific plates. By thrust faults of the Humboldt Bay area pose a greater or lesser the time the triple junction migrated to a latitude about 59 arc seismic hazard than rupture along a portion of the megathrust min south of the Temblor sediments and 15 arc min south of further north along the Cascadia subduction zone. the Sulphur Creek basins (39ø00'N), however, relative plate motion had changed, and deformation to the northeast of the Kinematic Model for Deformation Associated triple junction was contractional rather than extensional. With Triple Junction Migration If the relatively wide part of the forearc contraction zone Reconstruction of Deformation in Northern around Humboldt Bay reflects subduction of an internally California in the Last 3 m.y. deforming Gorda plate, then this wide contractile zone should Based on our kinematic model, we present a speculative have existed at the inception of internal deformation of the reconstruction of the timing and distribution of crustal defor- Gorda plate, about 3 Ma [Wilson, 1986]. Further, the wide mation in north coastal California in the last 3 m.y. (Table 3). 4816 KELSEYAND CARVER.'QUATERNARY TECTONICS, NORTHERN CALIFORNIA

TABLE 3. Timing of Deformation in North Coastal California the wake of contractional deformation by about 0.9-1.0 m.y., Related to Northward Migration of Triple Junction based on the timing of inception of translational deformation

Start of Start of in the Falor Formation near Maple Creek. This translational Sedimen- Latitude Estimated Contractional Strike-Slip deformation, which is superimposed on initial contraction, tary of Age of Faulting, Faulting, commences in a forearc setting but persists as the slab window Unit Unit Unit" Ma Ma migrates northward into the position previously occupied by the downgoing slab. Projections of the edge of the Gorda Falor 40ø53'N early 0.7-1.0•' 0.1 - .01• plate under North America, inferred from gravity data and Formation Pleistocene Garberville 40ø08'N late-early 2.2-2.5•' 1.3-1.6c' plate motions [Jachens and Griscom, 1983; Wilson, 1986] (Fig- sediments Miocene to ures 10 and 11), indicate that the slab edge is presently ap- Pliocene proximately below the Temblor Formation. We therefore infer Temblor 39ø45'N middle 3.0-3.3 c' 2.1-2.4 c that the slab window arrives about 3 m.y. after contraction Formation Miocene starts, or 2 m.y. after strike-slip faults start to be superimposed on top of the contraction in the forearc. "Ages of units discussed in text. /'Timingof deformationis constrainedby geologicevidence (see The superposition of strike-slip faulting on reverse faulting, text). observed at both Garberville and near Covelo, will occur in 'Timing of deformation is interpreted from proposed tectonic the Mad River fault zone in the next 1-1.5 m.y. as the Eaton model, no geologic age constraints are available (see text). Roughs fault zone propagates through and dominates over the thrusts of the Mad River fault zone. Similarly, the Garberville fault zone may propagate through the contractile fault zones The wide forearc contractile zone at the south end of the (Little Salmon fault, Goose Lake faults) that presently exist Cascadia subduction zone migrates northward in front of the south of Humboldt Bay. Alternatively, right-slip motion on triple junction and precedes arrival of the triple junction to the Garberville fault zone may ceaseas the neighboring Eaton the west by about 2 m.y. Translational deformation follows in Roughs fault zone becomes dominant. Because of the northerly trend of the subduction zone that favors right stepping of

125 ø 124 ø the transform boundary [Dickinson and Snyder, 1979b], we i favor the latter option. 41 ø _ 41 o A further implication of this tectonic scenario is that the LOR FM contractional structures observed near Covelo and Garberville probably represent a small fraction of the total number of thrust faults and associatedfolds that must have developed in the forearc as the subducting and deforming Gorda plate migrated northward. Only where Neogene sediments are not stripped away by erosion is evidence for this contraction preserved. In the Mad River fault zone, where erosion is less advanced, numerous folds and imbricate thrust faults are ex- t:OMa -• ( y • GARBERV)LLE posed. Deformation in the Garberville sediments supports the '• • :":'"'i;•EDIMENTS notion that the King Range terrane of McLaughlin et al. 40* 40* [1982] was accreted to North America farther to the south, $:0.8 Ma -• • • \ "x,. TEMBLOR perhaps at the latitude of the Transverse Ranges, and brought northward by the San Andreas fault and obductively accreted in the latest Neogene, as suggested by McLaughlin et al. [1985b]. The Garberville sediments do not record a mid- Miocene contractional event of the magnitude that should be observed if the King Range was obductively accreted at the latitude where it is now. However, the sediments do show strike-slip deformation that is reasonably compatible with a late Neogene strike-slip docking of a terrane traveling to the north-northwest.

;59* 39 ø CONCLUSIONS Strain patterns within the forearc at a convergent margin adjacent to a passing triple junction show a systematic evolution with time. The pattern of late Neogene and Quaternary deformation in northern coastal California (Figure 10) is con- • •KM sistent with a model of deformation that takes into account (1) , , the northward migration of the Mendocino triple junction and I I the San Andreas transform boundary, (2) internal deformation 125* 124* of the Gorda plate in the last 3 m.y., and (3) substantial con- Fig. 12. Map of west coast of California showing three previous traction in the overriding North American plate in the last 3 positions of Mendocino triple junction. The reconstructed positions m.y. due to subduction of the internally deforming Gorda are for the three approximate times (3.1., 2.3, and 0.8 Ma) at which plate. the Temblor Formation, Garberville sediments, and Falor Formation, respectively, started to deform in responseto the northwardly migrat- The model of deformation is based on mapping of defor- ing triple junction. Migration rate from Engebretsenet al. [1985]. mational structures in northern coastal California (Figures 2 KELSEYAND CARVER: QUATERNARY TECTONICS, NORTHERN CALIFORNIA 4817

and 3); these structures show evidence of crustal contraction Gorda-North American plate boundary. As the Mendocino or translation or both. North of the slab window (as defined triple junction migrates farther northward and the San An- by Dickinson and Snyder [1979b]), where the Gorda plate is dreas transform boundary extends in length, the strike-slip subducting beneath North America, deformation in the North faults of the San Andreas will extend to the north-northwest American forearc consists of a 30-km-wide zone of on-land into the present forearc, and forearc strike-slip fault zones contraction with a narrower zone of translation farther to the such as the Eaton Rough fault zone will become faults of the east (Figure 2). The zone of on-land contraction is the sub- San Andreas fault system. Concurrently, the contractional aerial portion of the relatively wide southernmost extent of structures near Humboldt Bay (Mad River fault zone and contractional deformation in the forearc of the Cascadia sub- Little Salmon fault) will cease to be active and will be pre- duction zone (Figure 10). Thrust faults and associated folds of served in the geologic record in the same way as contractional the Mad River and Little Salmon fault zones are the principle structures are preserved in Neogene sediments farther south- contractional structures. Strike-slip faults of the Eaton Rough near Covelo and Garberville. fault zone are the translational structures farther to the east. Net northeast-southwest directed contraction across the Mad Acknowledgments. T. A. Stephens contributed substantially to River fault zone in the Quaternary is at least 3.6 km, and field mapping in the Mad River fault zone. Kelsey's mapping in the Temblor Formation was facilitated by logistical help from the U.S. minimum fault slip rates on the five thrust faults in the zone Geological Survey and the field assistanceof F. Hochstaeder. J. C. range from 0.8 to 2.3 mm/yr. The net slip rate across the Mad Young provided valuable field data for both the Falor and Prairie River thrust fault zone is at least 6.4 mm/yr (Table 1). A Creek formations. Soil investigations by Bud Burke contributed to minimum slip rate estimate for the Little Salmon fault during neotectonic interpretations. C. M. de Polo generously gave permission to use a fault map compilation that forms the basis of Figure the Pleistocene is about 2 mm/yr, and net northeast-southwest 1. R. J. McLaughlin lent us a preliminary copy of his mapping on the directed contraction across the Little Salmon fault is at least Garberville and Covelo 1 ø sheets. The trace of the coastal belt thrust 4.3 km (computed from data of Woodward-Clyde Associates in Figure 3 is taken with permission from unpublished mapping of R. [1980]). Net right slip on the Eaton Roughs fault zone is J. McLaughlin, S. Ellen, A. Jayko, and M. C. Blake. S. H. Clarke uncertain, but limited geologic evidence suggeststhat it is of provided comment on our onshore-offshore fault correlations. R. F. Burroester aided with statistical analysis of fault data. R. L. Wheeler the order of 3 km. guided us to a more rigorous presentation of mesoscopicfault data. Farther south above the slab window (Figure 10), trans- Research was supported in part by U.S. Geological Survey Earth- lation presently is the predominant type of deformation. This quake Hazard Reduction Program contract 14-08-0001-22009. S. M. translation is accommodated along two major right-lateral Cashman, R. J. McLaughlin, D. J. Merritts, R. S. Stein, J. C. Tinsely, and R. S. Yeats reviewed the manuscript. fault systems of the San Andreas transform boundary, the

Lake Mountain and Garberville fault zones (Figure 3). How- REFERENCES ever, discontinuous exposures of late Neogene sediments along Angelier, J., Tectonic analysis of fault slip data sets, d. Geophys. Res., these fault zones near Covelo and Garberville, California, re- 89, 5835-5848, 1984. spectively, indicate that a period of crustal contraction, ex- Angelier, J., B. Colletta, and R. E. Anderson, Neogene paleostress pressed by thrust faults and associated folding, predated the changes in the Basin and Range, Geol. $oc. Am. Bull., 96, 347-361, 1985. present translational deformation (Figures 6 and 8). The ear- Atwater, B. F., Evidence for great Holocene earthquakes along the lier period of contractional deformation originated in the outer coast of Washington state, Science,236, 942-944, 1987. southern outcrop area of Neogene sediments (the Temblor Atwater, T., Implication of plate tectonics for the Cenozoic tectonic Formation near Covelo), and contractional deformation start- evolution of western North America, Geol. $oc. Am. Bull., 81, 3513- ed at a later date in the Garberville sediments. Translational 3536, 1970. Blake, M. C., Jr., and D. L. Jones, The Franciscan assemblage in movement along the Lake Mountain and Garberville fault northern California: A reinterpretation, in The Geotectonic Evolu- zones truncates the earlier contractional structures. tion of California, Rubey vol. 1, edited by W. G. Ernst, pp. 306-328, Constraints on the timing of initiation of contraction in the Prentice-Hall, Englewood Cliffs, N.J., 1981. Temblor and Garberville sediments (Table 3) suggest that in- Blake, M. C., R. H. Campbell, T. W. Dibblee, D. G. Howell, T. H. ception of contraction was related to both migration of the Nilsen, W. R. Normark, J. C. Vedder, and E. A. Silver, Neogene triple junction and to a major change in plate motion between basin formation in relation to plate-tectonic evolution of the San Andreas fault system, California, Am. Assoc. Pet. Geol. Bull., 62, the Pacific and Gorda plates at about 3 Ma [Wilson, 1986]. 344--372, 1978. Late Neogene contraction is not observed in Neogene sedi- Blake, M. C., Jr., A. S. Jayko, and R. J. McLaughlin, Tec- ments south of the Temblor sediments because Neogene sedi- tonostratigraphic terranes of the northern Coast Ranges, Califor- ments south of the Temblor were deformed prior to 3 Ma. nia, in Tectonostratigraphic Terranes of the Circum-Pacific Region, Earth $ci. Ser., vol. 1, edited by D. G. Howell, pp. 159-171, Circum- Assuming coupling of the Gorda plate to the overriding plate, Pacific Council for Energy and Mineral Resources, Houston, Tex., internal deformation of the Gorda plate promoted contraction 1985. in the North American plate. This contractional welt migrated Bloom, A. L., W. S. Broecker, J. M. A. Chappell, R. K. Matthews, and northward in the forearc of the North American plate in ad- K. J. Mesolella, Quaternary sea level fluctuations on a tectonic vance of the Mendocino triple junction and at the rate of coast: New 23øTh/23'*U dates from the Huon Peninsula, New Guinea, Quat. Res., 4, 185-205, 1974. migration of the triple junction. The position of the tripe junc- Brown, R. D., Jr., and E. W. Wolfe, Map showing recently active tion at the times of initial contraction in the Temblor and breaks along the San Andreas fault between Point Delgada and Garberville sediments (Figure 12) is determined based on the Bolinas Bay, California, U.S. Geol. Surv. Misc. Field Invest. Map, present configuration of the triple junction relative to active 1-681, 1972. Bull, W. B., Correlation of flights of global marine terraces, in Tec- deformation farther north in the Humboldt Bay area and on tonic Geomorphology, edited by M. Morisawa and J. T. Hack, pp. the assumption that this deformation is genetically related in 129-154, Allen and Unwin, Winchester, Mass., 1985. spaceand time to the present position of the triple junction. Burke, R. M., G. A. Carver, and S. Lundstrom, Soil development as a Present-day deformation in the vicinity of Humboldt Bay relative dating and correlation tool on marine terraces of northern therefore consists of contractional deformation associated with California (abstract), Geol. Soc. Am. Abstr. Programs, 18, 91, 1986. Carver, G. A., and R. M. Burke, Late Holocene paleoseismicityof the subduction of the Gorda plate and translational faulting far- southern end of the Cascadia subduction zone, Eos Trans. AGU, ther to the east associated with oblique convergence at the 68, 1240, 1987. 4818 KELSEYAND CARVER:QUATERNARY TECTONICS, NORTHERN CALIFORNIA

Carver, G. A., T. A. Stephens, and J. C. Young, Quaternary thrust Hobbs, B. E., W. D. Means, and P. F. Williams, An Outline of Structur- and reverse faulting on the Mad River fault zone, coastal northern al Geolo•ty, 571 pp., John Wiley, New York, 1976. California (abstract), Geol. Soc. Am. Abstr. Pro,trams, 15, 316, 1983. Jachens,R. C., and A. Griscom, Three-dimensional geometry of the Carver, G. A., R. M. Burke, and H. M. Kelsey, Deformation of late Gorda plate beneath northern California, J. Geophys. Res., 88, Pleistocene marine terraces along the California coast north of 9375-9392, 1983. CapeMendocino (abstract), Geol. Soc. Am. Abstr. Pro,trams, 18, 93, Jarrard, R. D., Terrane motion by strike-slip faulting of forearc sli- 1986. vers, Geology, 14, 780-783, 1986. Cashman, S. M., P. H. Cashman, and J. D. Longshore, Deformational Jennings, C. W. (Compiler), Geologic map of California, scale history and regional tectonic significance of the Redwood Creek 1:750,000: Calif. Div. of Mines and Geol., Sacramento, 1977. schist, northwestern California, Geol. Soc. Am. Bull., 96, 35-47, Jones,D. L., M. C. Blake, Jr., E. H. Bailey, and R. J. McLaughlin, 1986. Distribution and character of upper Mesozoic subduction com- Cashman, S. M., H. M. Kelsey, and D. R. Harden, Geology and plexes along the west coast of North America, Tectonophysics,47, descriptive geomorphology of the Redwood Creek basin, Hum- 207-222, 1978. boldt County, California, Geomorphic Processesand Aquatic Hab- Kelsey, H. M., and A. O. Allwardt, Evidence for a major Quaternary itat in the Redwood Creek Basin, Northwestern California, edited fault zone in the Central Belt Franciscan melange, northern Cali- by K. M. Nolan, H. M. Kelsey, and D.C. Marrou, U.S. Geol. Surv. fornia (abstract), Geol. Soc. Am. Abstr. Proqrams, 15, 316, 1983. Prof. Pap., 1454, in press, 1988. Kelsey, H. M., and S. M. Cashman, Wrench faulting in northern Chappell, J. M. A., A revised sea level record for the last 300,000 years California and its tectonic implications, Tectonics,2, 565-576, 1983. on Papua New Guinea: A numerical calculation, Quat. Res., 9, Kelsey, H. M., and D. K. Hagans, Major right-lateral faulting in the 265-287, 1983. Franciscan assemblageof northern California in late Tertiary time, Clark, J. T., Geology of the Bartlett Springs trend, northern Coast Geolo•ty, 10, 387-391, 1982. Ranges, California, M.S. thesis, 68 pp., Univ. of Tex. at Austin, MacGinitie, H., Central and southern Humboldt County, Bull. Calif. Austin, 1983. Div. Mines Geol., 118, 643-645, 1943. Clark, S. G., Geology of the Covelo district, Mendocino County, Manning, G. A., and B. A. Ogle, Geology of the Blue Lake Quadran- California, Univ. Calif. Berkeley Publ. Geol. Sci., 25, 119-142, 1940. gle, California, Bull. Calif. Div. Mines Geol., 148, 35 pp., 1950. Clarke, S. H., Geology of the California continental margin north of McEvilly, T. V., Sea floor mechanicsnorth of Cape Mendocino, Cali- Cape Mendocino, in Geolo•ty and Resource Potential of the Conti- fornia, Nature, 220, 901-903, 1966. nental Mar•7in of Western North America and Adjacent Ocean McKenzie, D. P., and W. J. Morgan, Evolution of triple junctions, Basins--Beaufort Sea to Baja California, Earth Sci. Ser., vol. 6, Nature, 224, 125-133, 1969. edited by D. W. Scholl, D. W. Grantz and J. G. Vedder, pp. 15A1- McLaughlin, R. J., The Sargent-Berrocalfault zone and its relation to 15A8, Circum-Pacific Council for Energy and Mineral Resources, the San Andreas fault system in the southern San Francisco bay Houston, Tex., 1987. region and Santa Clara Valley, California, J. Res. U.S. Geol. Surv., Cockerham, R. S., Evidence for a 180-km-long subducted slab be- 2, 593-598, 1974. neath northern California, Bull. Seismol. Soc. Am., 74, 569-576, McLaughlin, R. J., and T. H. Nilsen, Neogene non-marine sedi- 1984. mentation and tectonics in small pull-apart basins of the San An- Couch, R., Seismicity and crustal structure near the north end of the dreas fault system, Sonoma County, California, Sedimentology,29, San Andreas fault system, Studies of the San Andreas Fault Zone 865-876, 1982. in Northern California, edited by R. Streitz and R. Sherburne, Spec. McLaughlin, R. J., S. A. Kling, R. Z. Poore, K. L. McDougall, and E. Rep. Calif. Dir. Mines Geol., 140, 139-151, 1980. C. Beutner, Post-middle Miocene accretion of Franciscan rocks, Dehlinger, P., and B. A. Bolt, Seismic parameters along the Bartlett northwestern, California, Geol. $oc. Am. Bull., 93, 595-605, 1982. Springsfault zone in the Coast Rangesof northern California, Bull. McLaughlin, R. J., H. N. Ohlin, D. J. Thormahlen, D. L. Jones, J. W. Seismol. Soc. Am., 74, 1785-1798, 1984. Miller, and C. D. Blome, Geologic map and structure sections of DePolo, C. M., and H. N. Ohlin, The Bartlett Springs fault zone: An the Little Indian Valley-Wilbur Springs geothermal area, northern eastern member of the California plate boundary system (abstract), Coast Ranges, California, U.S. Geol. Surv. Open File Rep., 85-285, Geol. Soc. Am. Abstr. Pro,trams, 16, 6, 486, 1984. 24 pp. and two map sheets,1985a. Dickinson, W. R., and W. S. Snyder, Geometry of triple junctions McLaughlin, R. J., D. H. Sorg, J. L. Morton, T. G. Theodore, and C. related to San Andreas transform, J. Geophys. Res., 84, 561-572, E. Meyer, Paragenesisand tectonic significanceof base and pre- 1979a. cious metal occurrencesalong the San Andreas fault at Point Del- Dickinson, W. R., and W. S. Snyder, Geometry of subducted slabs gada, California, Econ. Geol., 80, 344-359, 1985b. related to San Andreas transform, d. Geol., 87, 609-627, 1979b. McPherson, R. C., S. W. Smith, and N. I. Severy,The Humboldt Bay Engebretson,D.C., A. Cox, and R. G. Gordon, Relative motions seismic network, 1974-1980 (abstract), Earthquake Notes, 52, 41, between oceanic and continental plates in the Pacific basin, Spec. 1981. Pap. Geol. Soc. Am., 205, 59 pp., 1985. Menack, J. S., Geology of the Neogene sedimentary rocks in Gar- Fisher, R. A., Dispersion on a sphere, Proc. R. $oc. London, 217, 295-305, 1953. berville, California, M.A. thesis, 231 pp., Cornell Univ., Ithaca, Fitch, T. J., Plate convergence,transcurrent faults, and internal defor- N.Y., June 1986. mation adjacent to southeastAsia and the western Pacific, J. Geo- Menack, j. S., B. Roth, and J. Barron, Tertiary terrane tectonics of phys. Res., 77, 4432-4460, 1972. northwestern California constrained by timing of Neogene shelf Fox, K. F., R. J. Fleck, G. H. Curtis, and C. E. Meyer, Implications of deposits(abstract), Eos Trans. AGU, 66, 1090, 1985. the northwestwardly younger age of the volcanic rocks of west- Norris, R. J., and A. F. Cooper, Small-scale fractures, glaciated sur- central California, Geol. Soc. Am. Bull., 96, 647-654, 1985. faces and recent strain adjacent to the Alpine fault, Geoloqy, 14, Frizzell, V. A., Jr., and M. L. Zoback, Stress orientation determined 687-690, 1986. from fault slip data in Hampel Wash area, Nevada, and its relation Ogle, B. A., Geology of the Eel River Valley area, Humboldt County, to contemporary regional stressfield, Tectonics,6, 89-98, 1987. California, Bull. Calif. Div. Mines, 164, 128 pp., 1953. Graham, S. A., and W. R. Dickinson, Evidence for 115 kilometers of Pampeyan, E. H., P. W. Harsh, and J. M. Coakley, Preliminary map right slip on the San Gregorio-Hosgri fault trend, Science, 199, showing recently active breaks along the Maacama fault zone be- 179-181, 1978. tween Laytonville and Hopland, Mendocino County, California, Harden, D. R., H. M. Kelsey, T. A. Stephens,and S. Morrison, Geo- U.S. Geol. Surv. Map, MF-1217, 1981. logic map of the Redwood Creek basin, U.S. Geol. Surv. Water Prescott, W. H., and S. Yu, Geodetic measurement of horizontal de- Resour. Div. Open File Rep., 81-496, 1982. formation in the northern San Francisco Bay region, California, J. Harding, T. P., Newport-Inglewood trend, California--An example of Geophys.Res., 9, 7475-7484, 1986. wrench style of deformation, Am. Assoc.Pet. Geol. Bull., 56, 97-116, Raft, A.D., and R. G. Mason, Magnetic survey off the west coast of 1973. North America, 40øN latitude to 50øN latitude, Geol. Soc. Am. Bull., Harsh, P. W., E. H. Pampeyan, and J. M. Oakley, Slip on Willits 72, 1267-1270, 1961. fault, California (abstract), Earthquake Notes, 49, 22, 1978. Riddihough, R., Recent movementsof the Juan de Fuca plate system, Heaton, T. H., and S. H. Hartzell, Earthquake hazards on the Cas- d. Geophys.Res., 89, 6980-6994, 1984. cadia subduction zone, Science, 236, 162-168, 1987. Sarna-Wojcicki,A.M., S. D. Morrison, C. E. Meyer, and J. W. Hill- Herd, D. G., Intracontinental plate boundary east of Cape Men- house,Correlation of upper Cenozoic tephra layers between sedi- docino, Geology, 6, 721-725, 1978. ments of the western United States and eastern Pacific Ocean and KELSEYAND CARVER: QUATERNARYTECTONICS, NORTHERN CALIFORNIA 4819

comparison with biostratigraphic and magnetostratigraphic age nents: A source of displaced terranes, J. Geophys.Res., 89, 1(,,072- data, Geol. Soc. Am. Bull., 98, 207-223, 1987. 10,076, 1984. Sibson, R. H., Earthquakes and linear infrastructure, Philos. Trans. R. Wagner, D. L., and E. J. Bortugno, Geologic map of the Santa Rosa Soc. London, Ser. A, 317, 63-79, 1986. quadrangle, scale 1:250,000, map 2A, 5 sheets,Calif. Div. of Mines Silver, E. A., Tectonics of the Mendocino triple junction, Geol. Soc. and Geol., Reg. Geol. Map Serv., Sacramento, 1982. Am. Bull., 82, 2965-2978, 1971. Walter, S. R., Intermediate-focus earthquakes associated with Gorda Simila, G. W., Seismological evidence on the tectonics of the northern plate subduction in northern California, Bull. Seismol.Soc. Am., 76, section of the San Andreas fault region, Studies of the San Andreas 583-588, 1986. Fault Zone in Northern California, edited by R. Streitz and R. Wilcox, R. E., T. P. Harding, and D. R. Seeley, Basic wrench tec- Sherburne, Spec.Rep. Calif. Div. Mines Geol., 140, 131-138, 1980. tonics, Am. Assoc. Pet. Geol. Bull., 557, 74-96, 1973. Simila, G. W., W. A. Peppin, and T. V. McEvilly, Seismotectonicsof Wilson, D. S., A kinematic model for the Gorda deformation zone as the Cape Mendocino, California, area, Geol. Soc. Am. Bull., 86, a diffuse southern boundary of the Juan de Fuca plate, J. Geophys. 1399-1406, 1975. Res., 91, 10,259-10,270, 1986. Smith, S. W., and J. S. Knapp, The northern termination of the San Woodward-Clyde Associates,Evaluation of the potential for resolv- Andreas fault, Studies of the San Andreas Fault Zone in Northern ing the geologic and seismic issues at the Humboldt Bay Power California, edited by R. Streitz, and R. Sherburne, Spec. Rep. Calif. Plant Unit no. 3, appendices, Woodward-Clyde Consultants, DiLl. Mines Geol., 140, 153-164, 1980. Walnut Creek, Calif., 1980. Smith, S. W., R. C. McPherson, and N. I. Severy, The Eureka earth- Wright, R. H., D. H. Hamilton, T. D. Hunt, M. L. Traubenik, and R. quake of 1980, breakup of the Gorda plate (abstract), Earthquake J. Shlemon, Character and activity of the Greenville structural Notes, 52, 44, 1981. trend, Proceedings, Conference on Earthquake Hazards in the Snay, R. A., M. W. Cline, and E. L. Timmerman, Project REDEAM: Eastern San Francisco Bay Area, edited by E. W. Hart, S. E. Models for historical horizontal deformation, NOAA Tech. Rep., Hirschreid, and S.S. Schulz, Spec. Publ. Calif. Div. Mines Geol., 62, NOS 125 NGS 42, 76 pp., 1987. 187-196, 1982. Suppe, J., Principles of Structural Geology, 537 pp., Prentice-Hall, En- glewood Cliffs, N.J., 1985. G. A. Carver, Department of Geology, Humboldt State University, Underwood, M. B., The Garberville thrust--A contact of probably Arcata, CA 95521. Miocene age within the Franciscan Complex, northern California H. M. Kelsey, Department of Geology, Western Washington Uni- (abstract), Geol. Soc. Am. Abstr. Programs, 14, 635-636, 1982. versity• Bellingham, WA 98225. Upp, R. R., Holocene activity on the Maacama fault, Mendocino County, California, Ph.D. thesis, 112 pp., Stanford Univ., Stanford, (Received June 19, 1987; Calif., 1982. revised January 7, 1988; Vink, G. E., W. J. Morgan, and W. Zhao, Preferential rifting of conti- acceptedNovember 12, 1987.)