Ultracataclasite structure and friction processes of … and...ELSEVIER Tectonophysics 295 (1998) 199–221 Ultracataclasite structure and friction processes of the Punchbowl fault,

ELSEVIER Tectonophysics 295 (1998) 199–221

Ultracataclasite structure and friction processes of the Punchbowl fault,San Andreas system, California

Frederick M. Chester Ł, Judith S. ChesterDepartment of Geology and Geophysics, Texas A&M University, College Station, TX 77843, USA

Received 10 October 1996; accepted 28 May 1998

Abstract

The Punchbowl fault is an exhumed, 40C km displacement fault of the San Andreas system. In the Devil’s Punchbowl,the fault contains a continuous ultracataclasite layer along which the Punchbowl Formation sandstone and an igneousand metamorphic basement complex are juxtaposed. The fabric of the ultracataclasite layer and surrounding rock indicatethat nearly all of the fault displacement occurred in the layer. By analogy with nearby active faults, we assume thatthe Punchbowl fault was seismogenic and that the ultracataclasite structure records the passage of numerous earthquakeruptures. We have mapped the ultracataclasite layer at 1 : 1 and 1 : 10 to determine the mode of failure and to constrain theprocesses of seismic slip. On the basis of color, cohesion, fracture and vein fabric, and porphyroclast lithology, two maintypes of ultracataclasite are distinguished in the layer: an olive-black ultracataclasite in contact with the basement, and adark yellowish brown ultracataclasite in contact with the sandstone. The two are juxtaposed along a continuous contactthat is often coincident with a single, continuous, nearly planar, prominent fracture surface (pfs) that extends the length ofthe ultracataclasite layer in all exposures. No significant mixing of the brown and black ultracataclasites occurred by offseton anastomosing shear surfaces that cut the contact or by mobilization and injection of one ultracataclasite into the other.The ultracataclasites are cohesive throughout except for thin accumulations of less cohesive, reworked ultracataclasitealong the pfs. Structural relations suggest that: (1) the black and brown ultracataclasite were derived from the basementand sandstone, respectively; (2) the black and brown ultracataclasites were juxtaposed along the pfs; (3) the subsequent,final several kilometers of slip on the Punchbowl fault occurred along the pfs; and (4) earthquake ruptures followed thepfs without significant branching or jumping to other locations in the ultracataclasite. By comparison with rock frictionexperiments, the slip localization along the pfs in the ultracataclasite implies rate weakening behavior with a criticalslip distance similar to laboratory values, and thus relatively small nucleation and breakdown dimensions for earthquakeruptures. Of the various mechanisms proposed to explain the low strength of the San Andreas and to produce dynamicweakening of faults, those that require or assume extreme localization of slip are most compatible with our observations. 1998 Elsevier Science B.V. All rights reserved.

Keywords: structure; faulting; friction; cataclasis; earthquakes

Ł Corresponding author. E-mail: [email protected]

0040-1951/98/$19.00 1998 Elsevier Science B.V. All rights reserved.PII: S 0 0 4 0 - 1 9 5 1 ( 9 8 ) 0 0 1 2 1 - 8

200 F.M. Chester, J.S. Chester / Tectonophysics 295 (1998) 199–221

1. Introduction

Large deformations of the upper portion of theEarth’s crust are primarily achieved through seis-mic faulting. Plate boundary faults, such as the SanAndreas fault, achieve large displacements via nu-merous seismic slip events. The San Andreas faultand possibly other large displacement faults slip un-der a resolved shear stress that is low compared tosimple quasi-static models of faulting based on ex-perimental rock mechanics (e.g., Hickman, 1991). Inspite of the extensive study of seismic faulting onthe San Andreas system, the physical and chemicalprocesses responsible for the nucleation, propaga-tion and arrest of earthquake ruptures are poorlyunderstood (e.g., Sibson, 1989; Scholz, 1990; Brune,1991; Segall, 1991). In fact, the geophysical liter-ature contains many underconstrained models anduntested hypotheses for the mechanics of seismicfaulting (e.g., Melosh, 1979, 1996; Byerlee, 1990,1993; Brune, 1991; Rice, 1992; Sleep and Blanpied,1992; Brune et al., 1993).

Experimental and field studies of the formationand growth of brittle faults lead to the conclusionthat a fault does not form and grow as an isolatedshear crack (e.g., Gay and Ortlepp, 1979; Cox andScholz, 1988; Cowie and Scholz, 1992; Reches andLockner, 1994). Rather, a fault grows through a com-plex breakdown process at the fault tip that involvescoalescence of fractures and shears to form through-going principal slip surfaces (Sibson, 1986). Concen-tration of shear stress at a fault tip produces arraysof isolated cracks that extend, interact, and linkto form fracture networks and throughgoing shears.This growth process is associated with a breakdownin strength from the intrinsic yield strength of theintact rock to the residual friction on the throughgo-ing principal slip surfaces (Cowie and Scholz, 1992).Continued slip on the newly formed principal slipsurfaces is associated with various wear processesthat produce layers of breccia, cataclasite and gougebounded by tabular zones of damaged rock (Sibson,1986). Subsequent slip could modify the internalstructure of a fault zone in a number of ways (e.g.,Hull, 1988; Wojtal and Mitra, 1988; Means, 1995).

Growth of an individual earthquake rupture in-volves a breakdown process at the rupture tip sim-ilar to the process of fault growth (e.g., Rudnicki,

1980; Swanson, 1992; Scholz et al., 1993). The di-mensions of the breakdown zone for an earthquakerupture on an existing fault are probably less thanthose for new faults because an existing fault alreadyhas a damaged zone (Cowie and Scholz, 1992).The structural signature from the passage of a seis-mic rupture probably is a narrow zone or zones ofconcentrated shear demarcating the rupture surfacewithin a broader zone of distributed fracturing. Theoccurrence of repeated earthquakes on an existingprincipal slip surface implies that the surface is re-strengthened during interseismic periods (e.g., Rice,1983). The degree of restrengthening may influencethe path of subsequent earthquake ruptures. If anexisting principal slip surface remains weak relativeto the surrounding rock, then a subsequent earth-quake rupture may occur along that surface ratherthan branch or jump to a new location. In this case,a principal slip surface could accommodate numer-ous slip events and ultimately accumulate very largedisplacement. In contrast, if a principal slip sur-face is greatly restrengthened by some process, suchas neomineralization, subsequent ruptures may oc-cur elsewhere within the fault zone. In general, thedegree to which an earthquake rupture follows a pre-existing principal slip surface and the thickness of anearthquake damage zone relative to the thickness ofan entire fault zone are unknown.

Exhumed faults display a static view of the cu-mulative structure produced during fault growth andsubsequent episodes of fault activity at various en-vironmental conditions. Although deciphering thekinematics and separating out the alteration resultingfrom near-surface weathering can be difficult, thesefaults provide an important view of the physicaland chemical processes operating in the seismogenicregime (Sibson, 1977; Bruhn et al., 1990; Chesteret al., 1993). Exhumed faults show considerablevariation in internal structure, but most large dis-placement faults are tabular zones of concentratedshear bordered by damage zones of fractured andbrecciated rock (Flinn, 1977; Brock and Engelder,1977; Wallace and Morris, 1986; Chester and Lo-gan, 1986; Sibson, 1986; Bruhn et al., 1990; Little,1995). Concentrated shear often is recorded by thereorientation and destruction of primary structures,development of cataclastic foliations, and presenceof extremely comminuted material such as gouge or

F.M. Chester, J.S. Chester / Tectonophysics 295 (1998) 199–221 201

ultracataclasite. We refer to a zone of concentratedshear (cataclasite and ultracataclasite) as the faultcore (Chester et al., 1993). At the macroscopic scale,a fault core represents the principal slip surface of thefault (Sibson, 1986). A fault zone may display a corenear one or both boundaries of the damaged zone,a single core centralized in the damaged zone, oran anastomosing network of several cores within thedamaged zone (Wallace and Morris, 1986; Rutter etal., 1986). Often, the fault core contains mesoscopicscale principal slip surfaces, i.e., features record-ing further concentration of shear at the mesoscopicscale (Chester et al., 1993; Arboleya and Engelder,1995). Thus, the definition of a principal slip sur-face depends on scale of observation (Arboleya andEngelder, 1995).

The Punchbowl fault, a deeply exhumed, largedisplacement fault of the San Andreas system, dis-plays either a single principal slip surface or pairedprincipal slip surfaces within broader zones of dam-aged host rock (Chester and Logan, 1986; Chesteret al., 1993; Schulz and Evans, 1998). The intensityof deformation progressively increases toward thefault cores. The cores may contain foliated catacla-sites, but always contain a single, continuous layerof ultracataclasite. The ultracataclasite consists al-most entirely of matrix particles less than 10 µmin diameter with porphyroclasts of vein fragments(Chester and Logan, 1987). Fabric analyses indicatethat nearly all of the shear displacement of the faultwas accommodated within the fault cores, and thatmost of the displacement was localized to the ultra-cataclasite layer. Principal slip surfaces are alwayspresent within the fault cores, suggesting localizationat the mesoscopic and microscopic scales as well.

The purpose of this paper is to characterize themesoscopic structure of the primary ultracataclasitelayer of the Punchbowl fault in the Devil’s Punch-bowl Los Angeles County Park (Fig. 1). At thislocation the Punchbowl fault juxtaposes fracturedbasement rock and arkosic sandstone of the Punch-bowl Formation along a continuous and distinct, 0.3m thick ultracataclasite layer (Figs. 2 and 3). Pre-sent-day exposures of the Punchbowl fault recordshear at 2 to 4 km depth (Chester and Logan, 1986).The depth has been estimated using post-Plioceneuplift and erosion rates for the San Gabriel Moun-tains (Oakeshott, 1971; Ehlig, 1975; Morton and

Matti, 1987), thickness of the sedimentary sequencein the Devil’s Punchbowl basin cut by the fault,and mineral assemblages and microstructures of thefault rocks (Chester et al., 1993). We can not saydefinitively that the Punchbowl fault slipped seismi-cally; however, active faults in the Central Trans-verse Ranges are seismogenic. These active faultscut rock types similar to or the same as those cutby the Punchbowl fault and are situated in the sametectonic setting. By comparison, we assume that thePunchbowl fault represents a mature, formerly seis-mogenic, fault zone of the San Andreas system.Thus, we use the mesoscopic structure of the ultra-cataclasite to infer the distribution of displacementduring the final stages of faulting, and to constrainthe characteristics of earthquake rupture surfaces andrupture processes.

2. Geology of the Punchbowl fault

The Punchbowl fault is an inactive, exhumed faultof the San Andreas transform system in the centralTransverse Ranges of southern California (Fig. 1). Itis located approximately 5 km southwest of and isparallel to the active strand of the San Andreas fault.The Punchbowl fault is truncated to the northwestand to the southeast by the San Andreas, and canbe considered an abandoned strand of the San An-dreas System (Dibblee, 1967, 1968; Barrows et al.,1987; Matti and Morton, 1993). The fault has a sin-uous trace, is often steeply dipping to the southwest,and in some locations displays anastomosing faultstrands that bound slices of exotic rock types. Thefault cuts crystalline rock of the San Gabriel base-ment complex, and along much of its length juxta-poses basement rocks and the Punchbowl Formationof the Devil’s Punchbowl basin (Noble, 1954).

The Devil’s Punchbowl basin is a small, elon-gate basin located between the Punchbowl and SanAndreas faults which is filled with more than a 1-km-thick sequence of the Punchbowl Formation (Noble,1954; Woodburne, 1975). The basin probably orig-inated as a pull apart basin during the early phasesof Punchbowl faulting in the Middle Miocene. ThePunchbowl Formation consists of a cobbly to pebblyarkosic sandstone with interbeds of siltstone and abasal breccia. The Punchbowl Formation overlaps


Fig. 1. Geologic map of the Punchbowl fault in the vicinity of the Devil’s Punchbowl Los Angeles County Park, northeast San GabrielMountains, California. The study area is in the Devil’s Punchbowl Park, at the southeastern end of the Punchbowl basin.

the angular unconformity with the Paleocene SanFrancisquito Formation along the northeastern edgeof the basin showing that the sediments onlappedrelief to the east (Weldon et al., 1993). The south-western side of the basin is truncated by the Punch-bowl fault and the marginal sedimentary deposits arenot preserved. A less than 0.5 km wide outcrop of

the basal breccia extends approximately 11 km tothe southeast of the basin along the Punchbowl fault(Noble, 1954). This basal breccia may record sedi-mentation associated with the fault-controlled mar-gin of the Devil’s Punchbowl basin (Weldon et al.,1993). The basal breccia both overlies and is cut bysubsidiary faults of the Punchbowl system suggest-


Fig. 2. Photographs of the Punchbowl fault zone and ultracataclasite layer. (a) View of the Punchbowl fault zone looking northwestfrom the Devil’s Chair overlook. White aplite of the San Gabriel basement complex (left) and Punchbowl Formation sandstones (right)are juxtaposed along the ultracataclasite layer. Note that the lithologic contacts in the basement have been rotated near the contact withthe ultracataclasite layer as a result of distributed shear within the fault zone. Location of the slip-parallel exposure mapped at a scaleof 1 : 10 is indicated. (b) A portion of the slip-parallel exposure of the ultracataclasite layer. Basement at the top and sandstone atthe bottom. The contacts between the layer and cataclastic host rocks are sharp, and the ultracataclasite is texturally distinct. Note thewedge-like protrusion of ultracataclasite into the Punchbowl Formation sandstone.


ing contemporaneous deposition and fault movement(Chester, unpubl. mapping, 1995).

In the Devil’s Punchbowl Los Angeles CountyPark, which covers the eastern portion of the Devil’sPunchbowl basin, the Punchbowl fault system iscomposed of two fault strands that bound a sliceof fractured and faulted basement up to 0.5 km inthickness (Fig. 1; Noble, 1954). The slice of frac-tured basement between the northern and southernfault strands contains a heterogeneous assemblageof Precambrian biotite gneiss and quartzofeldspathicgneiss with alternating leucocratic and melanocraticbands, and massive to foliated Cretaceous plutonicrocks including quartz diorite, tonalite, granodiorite,and biotite monzogranite (Cox et al., 1983). Locallywithin the Devil’s Punchbowl Park, the monzogran-ite is a brilliant white aplite that complexly intrudesmelanocratic rocks. The southern fault strand juxta-poses similar rock types (Cox et al., 1983), is notalways well developed, and is segmented and dis-continuous (Chester, unpubl. mapping, 1995). Thenorthern fault strand is much better developed andis defined by a continuous layer of ultracataclasitealong which the Punchbowl Formation and fracturedbasement are juxtaposed (Fig. 2).

A total right-lateral separation on the Punchbowlfault system of 40 to 50 km is indicated by offsetof the correlative San Francisquito and Fenner faults(Dibblee, 1967, 1968). This separation is consis-tent with the offset of the Devil’s Punchbowl basinfrom the inferred sediment source terrane (Weldonet al., 1993). The partitioning of total displacementbetween the northern and southern fault strands isunknown. The distribution of rock types along thePunchbowl fault in the Devil’s Punchbowl suggeststhat at least 10 km of separation occurred on thenorthern strand. Less than 10 km of separation wouldrequire an offset extension of the Fenner fault and as-sociated Pelona Schist to occur in the slice betweenthe northern and southern strands (Fig. 1). On thebasis of internal structure and continuous character,we think that the northern strand accommodated amuch greater percentage of the total slip than thesouthern strand. The timing of movement on thePunchbowl fault system is less clear. Although thefault appears to have been active during the forma-tion of the Devil’s Punchbowl basin, at least half ofthe total displacement occurred during the Pliocene

and Pleistocene, after the deposition of the Punch-bowl Formation was complete (Barrows et al., 1985;Matti et al., 1985; Weldon et al., 1993). Analysis ofthe subsidiary fault fabric and folds in the Punch-bowl Formation suggests that the slip vector for thenorthern strand plunged approximately 30º to thesoutheast during later stages of faulting, which isconsistent with the reverse fault geometry (Chesterand Logan, 1987).

3. Method

Two exposures of the ultracataclasite layer ofthe northern fault strand were mapped in detail.One exposure is located directly below the Devil’sChair overlook in the Devil’s Punchbowl Los Ange-les County Park, and the other is across the smallcanyon approximately 100 m to the northwest ofthe Chair. The orientation of the outcrop below theChair is approximately perpendicular to the aver-age slip vector (slip-perpendicular) inferred for thePunchbowl fault (Chester and Logan, 1987). Theother outcrop surface is approximately perpendicularto the layer and parallel to the slip vector (slip-par-allel). At each location, the ultracataclasite layer islocated in a gully and was partly exposed by erosion.We removed additional dirt and talus with a shoveland broom to completely expose approximately 30m of the layer. Two sections of the layer, totalling16 m length, were mapped at a scale of 1 : 10 usinga decimeter grid as a guide (Fig. 4). The grid wasplaced on the outcrop and repeatedly moved as map-ping progressed. The grid was positioned accordingto a reference line that was attached to the outcrop.The projection for each strip map is normal to thefault plane. An additional map was made at a scale of1:1 of a section of the slip-parallel outcrop (Fig. 5).The section mapped was excavated to a flat surfaceusing a sharp chisel; debris was removed with an airblower, and the surface was rinsed with water. Thearea excavated was approximately twice the size ofthe map (Fig. 3). The 1 : 1 map was made by tracingthe visible structure on plate glass that was firmlyaffixed to the outcrop surface.

The maps show the locations of lithologic con-tacts, veins, small faults, and fractures. Fine- andmedium-grained sandstone units are distinguished in


Fig. 3. Excavation of the ultracataclasite layer mapped at a scale of 1 : 1. Basement at the top and sandstone at the bottom. Note theporphyroclasts of basement rock in the ultracataclasite layer and the subsidiary fault in the sandstone. The prominent fracture surface islocated in the center of the layer and identified by arrows. The less cohesive brown ultracataclasite occurs in a layer along the basementside of the pfs. The location of the 1 : 1 map (Fig. 5) is shown.

the Punchbowl Formation. Melanocratic and leuco-cratic units are distinguished in the basement. Twomain ultracataclasite units are distinguished on thebasis of color. Color was classified for freshly brokensurfaces of the ultracataclasite by comparison witha rock color chart. The two ultracataclasite units aresubdivided further on the basis of fracture, vein, andcohesion characteristics.

The mineralogy of the host rocks and ultracat-aclasite were determined through petrographic andX-ray diffraction analyses of samples collected inthe Devil’s Punchbowl area (Table 1).

4. Structure of the Punchbowl fault zone

4.1. Punchbowl Formation

The damaged zone in the Punchbowl Formationis typically about 15 m thick. The zone is distin-

guished by the increase in mesoscopic fracture andsubsidiary fault density above regional levels. Sed-imentary layering and other sedimentary structuresare evident within the zone, often to within 1 m ofthe ultracataclasite layer. All along the contact withthe ultracataclasite layer, the sandstone is penetra-tively fractured and faulted, and displays cataclastictextures. The cataclastic sandstone always is textu-rally distinct from the ultracataclasite and forms asharp contact with the ultracataclasite layer.

In the region of the slip-parallel strip map, adiscontinuous layer of medium-grained cataclasticsandstone exists between the fine-grained sandstoneand the ultracataclasite layer (Fig. 4). Shear alongthe contact between the two sandstone units is in-dicated by a thin accumulation of reddish-brownultracataclasite (Fig. 5). Subsidiary faults cut bothsandstone units, however, the contact between thefine- and medium-grained sandstones is not offset.Many of the faults in the medium-grained sandstone


Table 1Average mineralogy of the host rock and ultracataclasite of the Punchbowl fault zone, Devil’s Punchbowl County Park, California

Minerals Rock unit

Punchbowl Formation a Ultracataclasite b Crystalline basement c

Quartz abundant abundant abundantPlagioclase feldspar abundant common abundantPotassium feldspar common trace minorCalcite, dolomite, siderite trace trace minorHornblende, other amphiboles trace trace traceIllite, mica minor trace minorKaolinite trace none traceChlorite trace trace minorChlorite–smectite mixed layer trace trace minorSmectite minor abundant traceIllite–smectite mixed layer trace trace traceLaumontite common trace minorClinoptillolite trace minor noneAnalcime trace minor trace

Relative mineral abundances determined by X-ray diffraction analysis: abundant >25%; common >10%; minor >3%; trace >0%; noneD 0%.a 12 analyses of 9 samples.b 13 analyses of 11 samples (4 samples of olive black and 7 samples of dark yellowish brown ultracataclasite).c 18 analyses of 18 samples.

have a conjugate geometry consistent with exten-sion parallel to the slip direction of the Punchbowlfault. The subsidiary fault and fracture geometriesin the mapped exposures are similar to extensionaland contractional duplexes described in detail bySwanson (1988, 1990).

The contact between the ultracataclasite layer andPunchbowl Formation sandstone displays both rel-atively straight and somewhat undulatory segments,as well as small wedges of ultracataclasite protrud-ing into the sandstone (Fig. 2b). The undulationsbetween straight sections often appear as embay-ments into the sandstone. The two largest wedgesobserved are displayed in the slip parallel exposure.At this location, each wedge is approximately 0.3 min length and is oriented at a low angle to the ul-tracataclasite layer. These wedges are not associatedwith intersections between larger subsidiary faultsand the ultracataclasite layer (Fig. 2b).

4.2. Basement complex

The slice of basement between the northern andsouthern fault strands is pervasively fractured andfaulted. A progressive increase in deformation inten-

sity towards the ultracataclasite layer is not alwaysevident. However, primary structures in the base-ment, such as gneissic layering and lithologic con-tacts, often are progressively reoriented with prox-imity to the ultracataclasite layer. At many locations,distributed shear of the basement along the ultracata-clasite layer produced a zone of foliated cataclasitesup to 5 m thick. The lithology of the porphyroclastsin the foliated cataclasites matches the lithology ofthe adjacent basement host rock. Where foliated cat-aclasites are derived from melanocratic protoliths,the contact between the foliated cataclasite and theultracataclasite, and between the foliated cataclasiteand the basement host rock, can be gradational.Where non-layered leucocratic rocks are in contactwith the ultracataclasite, the zone of cataclasites isthinner and foliations are absent. At these localitiesthe contact between the ultracataclasite and base-ment is distinct.

In the region of the slip-parallel strip map, a3-m-thick zone of foliated cataclasites in the base-ment borders the ultracataclasite layer. Foliations aresubparallel to the ultracataclasite layer. An intrusivecontact between the white aplite and older Precam-brian gneiss, as well as other lithologic contacts, can

F.M.C

hester,J.S.

Chester

/Tectonophysics295

(1998)199–221

207

Fig. 4. Structure of the ultracataclasite layer mapped at a scale of 1 : 10. (a) Slip-perpendicular exposure is shown in two panels: A–B and B–C. (b) Slip-parallel exposure is shownin four panels: D–E, E–F, F–G, and G–H. Location of the photographs shown in Fig. 2b and Fig. 3 are shown.


Fig.

4(c

ontin

ued)

.


Fig.

4(c

ontin

ued)

.


Fig. 5. Structure of the ultracataclasite layer mapped at a scale of 1 : 1.

be traced from the surrounding host rock into thezone of foliated cataclasites. The contact geometriesare consistent with drag folding and right lateralshear.

The contact between the ultracataclasite layer andbasement rock is irregular. In addition to larger wave-length, low amplitude undulations, the contact dis-plays small wavelength, high amplitude protrusionsof ultracataclasite into the basement (Fig. 4). In

many cases the protrusions are associated with inter-sections between melanocratic layers or subsidiaryfaults and the ultracataclasite layer.

4.3. Ultracataclasite layer

The roughness of a surface may be characterizedby a dimensionless ratio of amplitude to wavelengthwithin a geometric frequency interval (e.g., Power


and Tullis, 1991). In the vicinity of the Devil’s Chair,the ultracataclasite layer is exposed for over 100m. For wavelengths between 5 and 50 m, the layerconstitutes a continuous surface with a roughnessthat is on the order of 10�3. Within the excavatedand mapped region, the ultracataclasite layer variesbetween 0.15 and 0.55 m in thickness, and theaverage roughness of the layer boundaries is onthe order of 10�1 to 10�2.

The ultracataclasite layer forms a continuousboundary between the two host rocks; cataclasitederived from the Punchbowl Formation was neverfound on the southern (basement) side of the ultra-cataclasite layer. Similarly, cataclasite derived fromthe basement is not present on the northern (Punch-bowl Formation) side of the ultracataclasite layer.

Except for the rare pebble- to cobble-size prophy-roclasts, approximately 75% of the ultracataclasiteis composed of matrix grains that are <10 µm indiameter, 20% is veins and vein fragments, and 5%is single crystal porphyroclasts (Chester and Lo-gan, 1987). The veins are about 100 µm thick, andvein fragments are often about 100 µm in diame-ter. Rarely are the veins and porphyroclast greaterthan 1 mm in size. The ultracataclasite is rela-tively uniform in composition and mineralogicallydistinct from the surrounding Punchbowl Formationand basement rock. In general, the ultracataclasite isdepleted in potassium feldspar, laumontite, chlorite,and illite=mica, and enriched in smectite, clinop-tilolite, and analcime (Table 1). The contrastingmineralogy of the ultracataclasite reflects hydrationreactions promoted by extreme comminution (e.g.,Chester and Logan, 1986; Evans and Chester, 1995).

5. Internal structure of the ultracataclasite layer

5.1. Color

The ultracataclasite layer is composed of an oliveblack (5Y 2=1) unit and a dark yellowish brown(10YR 4=2) unit in both mapped segments. Eachunit is remarkably uniform in color. The olive blackultracataclasite (hereafter referred to as the black ul-tracataclasite) always is found in contact with thecrystalline basement on the south side of the layer,and the dark yellowish brown ultracataclasite (here-

after referred to as the brown ultracataclasite) alwaysis found in contact with the Punchbowl Formationsandstone on the north side (Fig. 4). The contact be-tween the black and brown ultracataclasites is sharp,continuous, and nearly planar; it is far less irregularthan the contacts between the ultracataclasites andhost rocks. Relationships noted at several exposuresof the ultracataclasite layer between the two mappedsegments, and at several additional exposures ofthe layer elsewhere in the Devil’s Punchbowl Park,suggest the contact between the brown and blackultracataclasites is at least 100 m long, and likelymore than 1.5 km long.

Inclusions of either ultracataclasite were neverfound in the other ultracataclasite. The relativelyfew pebble- to cobble-size porphyroclasts that arepresent in the ultracataclasite tend to occur in groupsnear the boundaries of the layer. Porphyroclasts ofthe Punchbowl Formation are confined to the brownultracataclasite and porphyroclasts of basement rockare confined to the black ultracataclasite (Fig. 4).Within the region mapped, the brown ultracataclasiteconstitutes less than half of the layer, and variesbetween 0.005 and 0.2 m in thickness.

Subsidiary faults in the damaged zones of thePunchbowl Formation and basement complex alsocontain ultracataclasite material. In contrast to thePunchbowl fault ultracataclasite layer, the subsidiaryfault ultracataclasite layers are highly variable incolor. The subsidiary fault ultracataclasite layers inthe Punchbowl Formation display hues of red andyellow, whereas those in the basement display huesof green and red. The differences in color may re-flect differences in composition and mineralogy. Themineral phases, major and trace element chemistry,and the stable isotope geochemistry of the fault rocksand adjacent host rocks are being studied and will bereported elsewhere.

5.2. Fractures

Fracture surfaces within the cataclasites of thefault core and within the damage zones on eitherside of the core often display slip lineations. Incontrast, fractures cutting the ultracataclasite layerare relatively smooth and undecorated, and do notdisplay plumose structures, slip lineations or anyother feature that would indicate a possible direction


of shear. Fracture surfaces generally do not offsetinternal contacts, such as between the brown andblack ultracataclasite, or boundaries of the ultra-cataclasite layer. Except for one prominent fracturesurface discussed below, the fractures are less thana meter in length. Of these, the longer fractures areslightly wavy, display a preferred orientation parallelto the layer, and are consistent with the ‘Y-, R1- andP-shear’ orientations of the Riedel array (e.g., Loganet al., 1979). There also is a distinct set of shorterfracture surfaces that display a quasi-conjugate ge-ometry with bisector normal to the ultracataclasitelayer, i.e., approximately in the ‘R2- and X-shear’orientations.

The prominent fracture surface (pfs) is relativelyplanar, perfectly continuous, and extends the entirelength of both the slip-parallel and slip-perpendicularexposures (Fig. 4). In many locations, the pfs iscoincident with the contact between the brown andblack ultracataclasites. All other structures in theultracataclasite either merge with or are truncatedby the pfs. The pfs is far more planar than theboundaries of the ultracataclasite layer and displays aroughness on the order of 10�3. The pfs is somewhatmore undulatory in the slip-perpendicular than in theslip-parallel exposure.

5.3. Veins and cohesion

The cohesion of the ultracataclasites varies sig-nificantly and systematically throughout the layer.All of the black ultracataclasite and approximatelyhalf of the brown ultracataclasite are cohesive. Mi-croscale veins and vein fragments that are barelyvisible in outcrop are common in the cohesive ul-tracataclasites. Intact veins generally are oriented atlow angles to the fault. However, most veins are dis-tended or broken and strung out into layers subparal-lel to the fault. Most of the cohesive ultracataclasitesdisplay a lower density of fracture surfaces overallbut a greater density of fractures oriented at largeangles to the layer. In these regions the ultracatacl-asite weathers to blocky fragments. Other portionsof the ultracataclasite, particularly the black ultracat-aclasite along the boundary of the layer, display ahigh density of anastomosing fracture surfaces andtends to weather into flakes. Locally, the flaky, cohe-sive black ultracataclasite displays small crenulations

with amplitudes and wavelengths on the order of 5mm (Fig. 4).

The layering in the cohesive ultracataclasites, de-fined by veins, fragments of veins, and fracturesurfaces, is subparallel to the interior and boundarycontacts of the ultracataclasite layer. The concordantrelationship is most obvious along undulations of thecontact between the brown and black ultracatacla-sites and along undulations of the boundaries of theultracataclasite layer. However, truncation of the lay-ering is evident locally, particularly along the pfs andat the boundaries of the wedge shaped protrusions ofultracataclasite into the Punchbowl sandstone catacl-asites (Fig. 4).

The remaining portion of the brown ultracatacl-asite is significantly less cohesive, does not containveins or vein fragments, and tends to part along anas-tomosing surfaces to produce very small flakes. Theless cohesive brown ultracataclasite only occurs as athin, discontinuous layer along the pfs. In excavat-ing the outcrop for the 1 : 1 mapping, we found thatthe less cohesive ultracataclasite could be cut easilywith a sharp hand-held chisel, whereas the cohesiveultracataclasite had to be fractured with a chisel andhammer. In addition, when rinsing the surface withwater, the less cohesive ultracataclasite became pastywhen wet. The less cohesive brown ultracataclasiteis a very distinct layer that occurs between the blackultracataclasite and the pfs (Figs. 3 and 5).

6. Distribution of slip

6.1. Localization of displacement

Field observations and experimental studies sup-port the general assumption that particle size anddegree of sorting tend to decrease with increasingdisplacement along a fault if all other parametersare constant (e.g., Engelder, 1974; House and Gray,1982; Sibson, 1986; An and Sammis, 1994). Thesedata suggest that the ultracataclasite in the core ofthe Punchbowl fault records a very large magnitudeof shear displacement, and that the abrupt changein texture at the contact between the ultracatacl-asite and bounding cataclasites records an abruptchange in the magnitude of shear strain. Extremelocalization of displacement to the ultracataclasite


layer also is suggested by the geometry of litho-logic contacts and other primary structures in thehost rock surrounding the ultracataclasite layer. Theprimary lithologic contacts in the cataclasites andfractured rock bounding the ultracataclasite layerare offset and reoriented by distributed shear. How-ever, in spite of the cataclastic deformation, primarylithologic contacts between basement lithologies inthe Devil’s Punchbowl, such as between the whiteaplite and Precambrian gneiss, are easily traced intothe fault zone and up to the ultracataclasite layer.The geometry of these contacts indicates that thesum of all fault parallel displacement in the catacla-sites and fractured rock bounding the ultracataclasitelayer totals less than approximately 100 m. Thus,nearly all of the ten-plus kilometers of displacementon the northern strand of the Punchbowl fault wasaccommodated by shear within the ultracataclasitelayer.

The fracture surfaces represent further localiza-tion of deformation within the ultracataclasite layer.Although we can not unequivocally demonstrateshear displacement on these fracture surfaces be-cause of the lack of offset markers and surface dec-oration, several features indicate that at least someof the surfaces are sites of shear. The mesoscalefractures are preferentially oriented subparallel tothe ultracataclasite layer, and the fabric is similarto the Riedel shear fracture fabric documented inmany experimental and natural brittle deformationzones (e.g., Logan et al., 1979; Swanson, 1988; Ar-boleya and Engelder, 1995). The mesoscale fabricalso is similar to the microscale shear band fab-ric of the Punchbowl ultracataclasite documentedby Chester and Logan (1987). Unlike the mesofrac-tures, however, shear offsets at the microscale aredemonstrated by the cutting and offset of veins, veinfragments, and color banding. The orientations andsense of shear on the microscopic shear bands areconsistent with both the orientations and sense ofshear expected for the Riedel array. It is possiblethat the presence of mesofractures is accentuatedby near surface phenomena associated with uncov-ering and weathering. Nonetheless, the similarity ofthe mesoscale and microscale fabrics suggest thatin the very least, the mesofractures follow planesof weakness in the ultracataclasite resulting fromthe microscale layering and shear bands. As such,

the mesoscale fracture fabric reflects the syntectonicstructure of the ultracataclasite if not the geometryof actual shear surfaces.

We infer that the pfs represents a late-stage,throughgoing, mesoscale principal slip surfacewithin the ultracataclasite because (1) the pfs ispresent and continuous in all exposures of the ul-tracataclasite, (2) all contacts, layering, and fracturesurfaces in the ultracataclasite either merge withor are truncated by the pfs, (3) it forms the con-tact between different rock types such as betweenthe brown and black ultracataclasites or between thebrown ultracataclasite and Punchbowl Formation, (4)the pfs displays a lower roughness in the slip-parallelthan in the slip-perpendicular exposures, consistentwith mesoscale corrugations of the surface alignedwith the inferred direction of slip on the Punchbowlfault, and (5) the pfs is spatially associated with thethin layer of less cohesive ultracataclasite (Figs. 3and 5).

6.2. Timing and processes of formation of the brownand black ultracataclasites

The general structure of the fault zone, includingthe continuous ultracataclasite layer, must have beenestablished fairly early in the fault history becausethe ultracataclasite layer has accommodated almostall of the shear displacement on the fault. Probablymost of the early-formed features in the ultracata-clasite layer were destroyed during subsequent de-formations when the large shear strain was imposed.The formation and juxtaposition of the brown andblack ultracataclasites may represent two of the moreancient events recorded in the fault core. We do notknow if the black and brown ultracataclasites werederived solely from the basement and PunchbowlFormation, respectively. However, significant mixingof comminuted basement and Punchbowl Formationis not consistent with the restricted distribution ofporphyroclasts.

We infer that the brown ultracataclasite waslargely derived from the Punchbowl Formation sand-stone because (1) the brown ultracataclasite is alwaysin contact with the sandstone, (2) the brown ultracat-aclasite contains only porphyroclasts of sandstone,(3) porphyroclasts of sandstone are restricted to thebrown ultracataclasite, and (4) the brown ultracata-


clasite is very similar in appearance to the ultracat-aclasites in the subsidiary faults of the PunchbowlFormation. Similar reasoning implies that the blackultracataclasite must have originated from comminu-tion of the basement rock. The similar texture andmineralogy of the two ultracataclasites suggests thatthey formed at fairly similar conditions.

A simple model for the formation and juxtapo-sition of the brown and black ultracataclasites in-volves (1) generation of the black ultracataclasitelayer along a section of the fault within the basementand generation of the brown ultracataclasite layeralong a section of the fault within the PunchbowlFormation, followed by (2) large displacement onthe pfs to place the black and brown ultracataclasitesin contact (Fig. 6). The contact between the twoultracataclasites must have formed prior to the very

Fig. 6. Schematic illustrating a simple model for the formation and juxtaposition of the dark yellowish-brown and olive-blackultracataclasites. During the early stages of faulting some segments of the fault were wholly contained in the Punchbowl Formation orin the basement. In these segments the ultracataclasite is derived from a single host rock type. At late stages of faulting, after largedisplacement on the fault, the Punchbowl Formation and the basement are juxtaposed. Translation of the brown and black ultracataclasiteswith the host rock places the ultracataclasite in contact.

last increment of displacement along the pfs becausethe contact either merges with or is truncated by thepfs.

We interpret that the less cohesive brown ultracat-aclasite formed during the last stage of deformationin the fault core. Cohesion of the ultracataclasite ispartly a result of mineralization and the vein ce-mentation. Almost all of the rocks in the fault corecontain microscale veins that record several episodesof cementation. Veins and vein porphyroclasts werereoriented to form layers parallel to the clay folia-tion by offset on distributed networks of microscaleshear bands and particle-scale flow in the clay matrix(Chester and Logan, 1987). The cementation eventsmust have been syntectonic because the veins andmicroscopic shear bands are mutually cross cutting.The fact that the less cohesive brown ultracataclasite


generally lacks veins and vein fragments, suggeststhat it represents the most recently reworked anduncemented material of the fault core.

6.3. Mode of failure during the last phase of faulting

The relations mapped indicate that all of the ul-tracataclasite underwent an earlier phase of deforma-tion involving cementation and microscale shearingduring which cohesion developed, and a later phaseinvolving displacement along the pfs in the interiorof the ultracataclasite layer. This later phase pro-duced a narrow layer of reworked, less cohesiveultracataclasite, which we interpret to be the productof abrasive wear from slip on the pfs.

Although the contact between the black andbrown ultracataclasites could have formed throughdistributed shearing flow, we suggest a simple modelin which both the formation of the contact and allsubsequent displacement on the fault occurred bylocalized slip on the pfs. Such a model is consistentwith the observations that the pfs (1) does not cutacross and offset the contact between the brown andblack ultracataclasites, and (2) either coincides withthe contact or, as seen in the excavation mapped ata scale of 1 : 1, is separated from the contact by theless cohesive brown ultracataclasite. Slip on the pfswould have reworked the cohesive ultracataclasite byabrasive wear. At sites of abrasion and removal ofultracataclasite, the pfs would continue to form thecontact between the two ultracataclasites, and layer-

Fig. 7. Figure illustrating wear and wear product accumulation associated with slip along the undulatory pfs. Along some segments ofthe surface, such as at restraining bends where high normal stress leads to high wear rates, the cohesive ultracataclasite is removed at thesurface through abrasion. Along other segments, such as at releasing bends where low normal stress leads to low wear rates or separation,thin layers of abrasive wear product tends to accumulate along the surface. Wear product should accumulate as thin lenticular layers, andaccumulation results in migration of the slip surface away from the contact with the cohesive ultracataclasite.

ing in the ultracataclasites would be truncated by thepfs. At sites of wear product accumulation, the pfswould progressively migrate away from the contactbetween the two ultracataclasites as the wear productaccumulated (Fig. 7).

Regardless of the process that formed the con-tact between the two ultracataclasites, the presentgeometry would only be preserved if no signifi-cant mixing of the two ultracataclasites occurredsince juxtaposition (Fig. 8). After the contact wasformed, shear displacement on any surface that cutacross the contact would have generated discontinu-ities and duplications of the contact. If displacementhad occurred on an anastomosing set of localizedslip surfaces, or on slip surfaces that branched orcut across the ultracataclasite layer, then one wouldexpect to see interfingering and stacking of faultbounded slices of brown and black ultracataclasitesrather than a single continuous contact between thetwo units (Fig. 8b). The geometry of the contacts be-tween the ultracataclasite and surrounding rock andthe fact that the basement and sandstone are confinedto their respective sides of the ultracataclasite layerare additional evidence that shears did not cut in andout of the layer. Although the wedge structures at theboundary of the ultracataclasite layer could recordsome flow of the ultracataclasite into the surroundingrock, there is no evidence of mixing flow betweenthe brown and black ultracataclasites. Features suchas ultracataclasite filled cracks or injection structuresthat record flow or mobilization of ultracataclasite


Fig. 8. Schematic showing the internal structure of the ultracataclasite layer expected for four endmembers as a function of sliplocalization and degree of mixing at the mesoscopic scale. (a) Slip on a single, stationary, throughgoing surface will lead to little mixing.This mode of failure results in juxtaposition of contrasting host rock and ultracataclasite across the localized slip surface. This structure ismost similar to that of the Punchbowl ultracataclasite layer (see text). (b) Contemporaneous or alternating slip on multiple, anastomosing,throughgoing surfaces leads to mixing across the layer. Fault slices of host rock and ultracataclasite may be distributed and juxtaposedthroughout the layer. For the case of closely spaced surfaces and large slip, the layer may appear homogeneous at the mesoscopic scale.(c) Ultracataclasite components may be juxtaposed but remain segregated by distributed flow within the layer if streamline or laminarflow is maintained. (d) Components may mix if flow is turbulent or non-laminar. Fluidization and injection of ultracataclasite is anextreme example of non-laminar flow that could lead to mixing of components. As in the case (b), flow processes could homogenize theultracataclasite.

(e.g., Brock and Engelder, 1977; Lin, 1996) also arenot observed (Fig. 8d).

Some distributed fracturing and flow of the ultra-cataclasite and host rock would have been necessaryto accommodate movement along the nonplanar pfs.However, it appears that such deformation was minorduring the final phase of faulting. Thus, only a smallfraction of the ultracataclasite layer was activelyundergoing shear, and most of the ultracataclasiteretained the cohesion and microfabric formed duringearlier deformation events.

The amount of displacement on the fault since for-mation of the contact between the brown and black

ultracataclasites must be at least equal to the lengthof the contact measured parallel to the slip direction.Field relations suggest that this length is greater than1.5 km. A similar estimate is derived from the outcroppattern of the Punchbowl Formation. The PunchbowlFormation extends along the north side of the Punch-bowl fault for approximately 10 km to the southeastof the study location and probably to 2 km in depth(Fig. 1). The outcrop pattern and the right-lateral, re-verse-oblique slip on the Punchbowl fault (Chesterand Logan, 1987) suggest that the sandstone and base-ment were in contact for at least the final 2 km of slipalong the section of the fault mapped.


7. Friction and earthquake mechanisms

The mesoscopic structure of the Punchbowl faultultracataclasite layer records the mode of failure dur-ing the final several kilometers of slip on a largedisplacement fault of the San Andreas system. As-suming that the fault slipped seismically, the finalseveral kilometers of slip would have generated upto 103 large magnitude earthquake slip events, or aneven greater number of smaller magnitude events.Thus, the structure of the ultracataclasite layer re-flects deformation associated with the passage ofnumerous earthquake ruptures and implies that: (1)during the later stages of faulting, repeated earth-quake ruptures occurred along the same well definedpfs; (2) slip along this pfs generated a frictional wearproduct from a pre-existing, layered, cohesive ultra-cataclasite; and (3) newly formed surfaces and splaysdid not accommodate a significant amount of slip.

To date, few features have been identified that areunique to seismic slip or aseismic creep. Pseudo-tachylytes are the best evidence for paleoseismic slipevents (e.g., Sibson et al., 1975; Swanson, 1992).Other features that may indicate seismic slip are im-plosion breccias, fluidized cataclasite, and injectionstructures. These latter structures are thought to re-sult from sudden pressurization or depressurizationprocesses (Scholz, 1990; Lin, 1996). At the otherextreme, crystal fiber growth on a slip surface, cat-aclastic foliations, and layering often are attributedto slow, aseismic creep processes (Sibson, 1986;Groshong, 1988; Miller, 1996). However, becausecreep compaction may be the typical mode of fail-ure during interseismic periods of the seismic cycle(Sibson, 1989; Chester et al., 1993), the presence offeatures formed by creep cannot be used as evidenceprecluding seismic slip. The existence of structuresrecording both localized slip and distributed flowmay indicate repetitive deformation sequences char-acteristic of seismic cycling (e.g., Hobbs et al., 1986;Power and Tullis, 1989). Except for foliation andlayering, none of the features diagnostic of seismicor aseismic slip are present in the Punchbowl faultzone.

Some characteristics of internal fault structureassociated with unstable slip have been suggestedby rock deformation experiments investigating faultfriction. Faults often are modeled in experiments by

sliding two blocks of intact rock along a contactingsurface or a thin layer of simulated gouge (e.g., Lo-gan, 1975; Beeler et al., 1996). In simulated gougeexperiments, the imposed displacement initially isaccommodated in the layer by distributed slip ona network of Riedel shears (Logan et al., 1979).With greater displacement, localized shears parallelto the layer develop within and at the boundariesof the layer and subsequently accommodate the im-posed displacements (Engelder et al., 1975; Loganet al., 1979, 1992). Abrasive wear along a localizedshear surface produces extremely fine-grained ma-terial that tends to accumulate along irregularitiesof the surface. At relatively large displacements, asachieved in rotary shear experiments on simulatedgouge (e.g., Beeler et al., 1996), the mode of failureis quite similar to that in the Punchbowl ultracatacla-site layer. That is, the displacement in an experimentis accommodated along distinct layer-parallel shearsurfaces, and apparently only a narrow zone withinthe layer is undergoing active shear at any one time(Marone and Kilgore, 1993; Beeler et al., 1996).Fine-grained wear product accumulates as the shearsurface slowly migrates through the layer.

A process that weakens a fault is necessary for adynamic earthquake rupture to nucleate (e.g., Rice,1983). Experiments suggest that the extreme local-ization of slip to discrete surfaces, such as occursat large displacements in simulated gouge experi-ments, favors the small critical slip distance and rateweakening friction behavior that promotes dynamicinstability (Marone and Kilgore, 1993; Beeler et al.,1996). Thus, the localization of displacement in thePunchbowl fault to the pfs associated with narrowzones of less cohesive ultracataclasite would appearconsistent with rate weakening behavior and seis-mic slip. Theoretical models suggest that both thenucleation patch size of an earthquake rupture andthe dimensions of the breakdown zone at the earth-quake rupture tip scale with the critical slip distance(Rudnicki, 1980; Dieterich, 1992). The breakdownprocess at the rupture tip involves distributed frac-ture and fracture linkage to form the rupture surface.Assuming that the zone of distributed fracture, occur-rence of discontinuities, and roughness of the rupturesurface increase with the dimension of the break-down zone, we infer small nucleation and break-down dimensions for ruptures in the Punchbowl fault


zone. The preservation of preexisting structures nearthe pfs, such as the contact between the black andbrown ultracataclasite, suggests that the breakdownprocesses did not significantly disrupt a large volumeof rock.

Repeated seismic rupture on a single slip surfacein the Punchbowl ultracataclasite layer is oppositeto the kinematics of pseudotachylyte bearing faultswhere slip surfaces are rarely reruptured (e.g., Gro-cott, 1981; Swanson, 1988, 1989, 1990). Pseudo-tachylyte bearing faults are effectively welded uponsolidification of the friction melts and have strengthsthat probably approach or exceed the host rock.Thus, subsequent ruptures tend to form in the hostrock. Sidewall ripouts have been described for pseu-dotachylyte bearing faults, and represent a mecha-nism for bypassing a segment of a fault that seizesduring slip (Swanson, 1989). Although strength re-covery also must have occurred along the slip surfaceof the Punchbowl fault, the repeated rupture of thepfs and the absence of sidewall ripouts implies thatthe strength contrast between the slip surface andwall rock remained pronounced.

Processes that lead to weakening during dynamicslip may be important for the occurrence of greatearthquakes (e.g., Brune, 1991). In addition, suchdynamic weakening mechanisms may be necessaryto explain the low shear strength of the San Andreasfault. A number of processes have been proposedthat could lead to dynamic weakening includingthermal pressurization of pore fluids (e.g., Sibson,1973), liquefaction, fluidization of granular material(e.g., Melosh, 1996), and reduction of normal stressby interface separation during dynamic slip (Bruneet al., 1993). Frictional heating may lead to sig-nificant temperature changes in cases where rapid,large slip occurs in narrow zones (Brune et al., 1969;Mase and Smith, 1987). Under extreme conditionsof seismic slip, temperature changes can be greatand cause local melting of rock. For seismic faultingin fluid saturated rock, heating should cause a localincrease in pore fluid volume (e.g., Sibson, 1973). Inconditions where permeability and changes in poros-ity are small, heating can lead to pressurization offluid, a local reduction in effective stress, and dy-namic weakening (Mase and Smith, 1987). Changesin porosity and permeability in the Punchbowl ul-tracataclasite layer during the last phase of faulting

were probably small, as most of the ultracataclasiteretained its cemented and cohesive character. Largeshear on surfaces or within thin layers of less co-hesive ultracataclasite sandwiched between cohesiveultracataclasite, as observed along the Punchbowlfault, could be consistent with thermal pressurizationduring seismic slip. Such a structure also may beconsistent with interface separation during seismicslip.

Acoustic fluidization can occur in cohesionless,granular material with sufficient concentration ofhigh frequency acoustic energy (Melosh, 1979,1996). Fluidization also could occur in a fluid sat-urated granular material through liquefaction. Forthe case of seismic faulting, acoustic fluidization orliquefaction in a fault core should produce struc-tures recording flow within the layer. Flow structuresgenerated since the juxtaposition of the black andbrown ultracataclasites generally are absent in thePunchbowl ultracataclasite layer. In fact, deforma-tion involving distributed flow of granular materialas occurs during liquefaction, fluidization, Coulombplasticity of thick gouge zones (Byerlee and Sav-age, 1992), and rolling on space filling bearingswith compatible kinematic rotations (Herrmann etal., 1990), probably was not important during thefinal several kilometers of slip on the Punchbowlfault.

8. Conclusions

In the Devil’s Punchbowl area, the Punchbowlfault contains a single, continuous ultracataclasitelayer centrally located in a broader zone of extremelyfractured rock. The ultracataclasite layer constitutesa macroscopic scale principal slip surface that ac-commodated 10 km or more of slip and juxtaposedthe Punchbowl Formation lithic sandstone and ig-neous and metamorphic rocks of the San Gabrielbasement complex. We characterized the mesoscopicstructure of the ultracataclasite by mapping two ex-posures at a scale of 1 : 10 and an excavation at 1 : 1.Different types of ultracataclasite were distinguishedon the basis of color, cohesion, fracture and vein fab-ric, and porphyroclast lithology. The ultracataclasiteis internally layered, and the geometry and cross-cutting relations of layer contacts record the mode


of failure during the final stages of fault activity. Byanalogy with modern faults of the San Andreas in theTransverse Ranges, we assume that the Punchbowlfault slipped seismically, and that the structure of theultracataclasite layer probably records deformationresulting from the passage of numerous earthquakeruptures.

Structural relations indicate that the ultracatacl-asite underwent an earlier phase of deformationinvolving cementation and microscale flow duringwhich cohesion developed, and a later phase involv-ing displacement along a prominent fracture surface(pfs) in the interior of the ultracataclasite layer andproduction of a thin layer of reworked, less cohesiveultracataclasite. Prior to the latest stage of slip alongthe pfs, a brown ultracataclasite and black ultracat-aclasite associated with the Punchbowl Formationand the basement, respectively, were juxtaposed. Theamount of displacement on the fault since forma-tion of the contact between the brown and blackultracataclasites is greater than 1.5 km.

We infer that the contact between the brown andblack ultracataclasites was formed by localized slipon the pfs. With subsequent slip, attrition wear andwear product accumulation occurred along the pfs.At sites of attrition, the layering in the ultracatacla-site was truncated at the pfs and the contact betweenthe brown and black ultracataclasite coincides withthe pfs. At sites of wear product accumulation, thepfs is separated from the contact between the brownand black ultracataclasites by a layer of less cohesiveultracataclasite that represents the accumulation ofreworked ultracataclasite. There is no evidence forsignificant mixing of the brown and black ultracat-aclasites as would occur by offset on anastomosingshear surfaces in the layer or by mobilization andinjection of one ultracataclasite into the other.

The internal structure of the Punchbowl fault im-plies that earthquake ruptures were not only confinedto the ultracataclasite layer, but largely localized tothe pfs. During the latest phase of deformation, thebulk of the ultracataclasite retained cohesion andlayer contacts were relatively undisturbed. As such,the earthquake ruptures must have followed the pfswithout significant branching or jumping to otherlocations in the ultracataclasite. By comparison withlaboratory studies of rock friction, the localization ofdisplacement in the Punchbowl ultracataclasite im-

plies rate weakening behavior with small critical slipdistance, and thus small nucleation and breakdowndimensions for ruptures.

Of the various mechanisms proposed to explain ofthe low strength of the San Andreas and to producedynamic weakening of faults, those that require orpredict wide zones of less cohesive granular materialthat flows and mixes appear incompatible with ourobservations. Mechanisms that assume or are consis-tent with extreme localization of slip, such as thermalpressurization of pore fluids and possibly interfaceseparation waves, are more consistent with structuresproduced during the final phase of movement on thePunchbowl fault.

Acknowledgements

It was under John Logan’s guidance that we bothbegan our research on crustal faulting. His influenceon our research is obvious, particularly regarding theimportance we place on combining field and exper-imental approaches to geologic problems. We arepleased to dedicate this paper to John, and thankhim for his support and friendship over the years.Discussions with J.P. Evans were helpful. Reviewsby Sue Agar and Mark Swanson motivated addi-tional study and helped us to clarify our thinking.This research was supported by the U.S. Geologi-cal Survey (USGS), Department of the Interior, un-der awards 1434-94-G-2457 and 1434-HQ-96-GR-02709, and by a National Science Foundation (NSF)award EAR92-05973.

References

An, L.J., Sammis, C.G., 1994. Particle size distribution of cata-clastic fault materials from southern California: A 3-D study.Pure Appl. Geophys. 143, 203–227.

Arboleya, M.L., Engelder, T., 1995. Concentrated slip zones withsubsidiary shears: Their development on three scales in theCerro Brass fault zone, Appalacian valley and ridge. J. Struct.Geol. 17, 519–532.

Barrows, A.G., Kahle, J.E., Beeby, D.J., 1985. Earthquake haz-ards and tectonic history of the San Andreas fault zone, LosAngeles County, California. Calif. Div. Mines Geol. Open FileRep. 85-10LA, 139.

Barrows, A.G., Kahle, J.E., Beeby, D.J., 1987. Earthquake haz-ards and tectonic history of the San Andreas fault zone, Los


Angeles County, California. In: Hester, R.L., Hallinger, D.E.(Eds.), San Andreas fault-Cajon Pass to Palmdale. Pac. Sect.Am. Assoc. Pet. Geol. Vol. Guideb. 59, 1–92.

Beeler, N.M., Tullis, T.E., Blanpied, M.L., Weeks, J.D., 1996.Frictional behavior of large displacement experimental faults.J. Geophys. Res. 101, 8697–8715.

Brock, W.G., Engelder, T., 1977. Deformation associated withthe movement of the Muddy Mountain Overthrust in theBuffington Window, southeastern Nevada. Geol. Soc. Am.Bull. 88, 1667–1677.

Bruhn, R.L., Yonkee, W.A., Parry, W.T., 1990. Structuraland fluid-chemical properties of seismogenic normal faults.Tectonophysics 218, 139–175.

Brune, J.N., 1991. Seismic source dynamics, radiation and stress.Rev. Geophys. Suppl. 29, 688–699.

Brune, J.N., Henyey, T.L., Roy, R.F., 1969. Heat flow, stressand rate of slip along the San Andreas fault, California. J.Geophys. Res. 74, 3821–3827.

Brune, J.N., Brown, S., Johnson, P.A., 1993. Rupture mech-anisms and interface separation in foam rubber models ofearthquakes — a possible solution to the heat-flow paradoxand the paradox of large overthrusts. Tectonophysics 218, 59–67.

Byerlee, J., 1990. Friction, overpressure and fault normal com-pression. Geophys. Res. Lett. 17, 2109–2112.

Byerlee, J., 1993. Model for episodic flow of high pressure waterin fault zones before earthquakes. Geology 21, 303–306.

Byerlee, J.D., Savage, J.C., 1992. Coulomb plasticity within thefault zone. Geophys. Res. Lett. 19, 2341–2344.

Chester, F.M., Logan, J.M., 1986. Implications for mechanicalproperties of brittle faults from observations of the Punchbowlfault zone, California. Pure Appl. Geophys. 124, 79–106.

Chester, F.M., Logan, J.M., 1987. Composite planar fabric ofgouge from the Punchbowl fault, California. J. Struct. Geol. 9,621–634.

Chester, F.M., Evans, J.P., Biegel, R.L., 1993. Internal struc-ture and weakening mechanisms of the San Andreas fault. J.Geophys. Res. 98, 771–786.

Cowie, P.A., Scholz, C.H., 1992. Physical explanation for thedisplacement–length relationships of faults using a post-yieldfracture mechanics model. J. Struct. Geol. 14, 1133–1148.

Cox, B.F., Powell, R.E., Hinkle, M.E., Lipton, D.A., 1983. Min-eral resource potential map of the Pleasant View RoadlessArea, Los Angeles County, California. U.S. Geol. Surv. Misc.Field Stud. Map MF-1649-A, scale 1 : 62,500.

Cox, S.J.D., Scholz, C.H., 1988. On the formation and growth offaults: an experimental study. J. Struct. Geol. 10, 413–430.

Dibblee, Jr., T.W., 1967. Areal geology of the western Mojavedesert. U.S. Geol. Surv. Prof. Pap. 522.

Dibblee, T.W., Jr., 1968. Displacements on the San Andreas faultsystem in the San Gabriel, San Bernardino, and San JacintoMountains, southern California. Proc. Conf. on GeologicalProblems of the San Andreas fault system. Stanford Univ.Publ. Geol. Sci. 11, 260–276.

Dieterich, J.H., 1992. Earthquake nucleation on faults with rate-and state-dependent strength. Tectonophysics 211, 115–134.

Ehlig, P.L., 1975. Basement rocks of the San Gabriel Mountains,

south of the San Andreas fault, southern California. Calif. Div.Mines Geol. Spec. Rep. 118, 177–186.

Engelder, J.T., Logan, J.M., Handin, J., 1975. The sliding char-acteristics of sandstone on quartz fault-gouge. Pure Appl.Geophys. 113, 69–86.

Engelder, T., 1974. Microscopic wear grooves on slickensides:indicators of paleoseismicity. J. Geophys. Res. 79, 4387.

Evans, J.P., Chester, F.M., 1995. Fluid–rock interaction in faultsof the San Andreas system: Inferences from San Gabriel faultrock geochemistry and microstructures. J. Geophys. Res. 100,13007–13020.

Flinn, D., 1977. Transcurrent faults and associated cataclasis inShetland. J. Geol. Soc. London 133, 231–248.

Gay, N.C., Ortlepp, W.D., 1979. Anatomy of a mining-inducedfault zone. Geol. Soc. Am. Bull. 90, 47–58.

Grocott, J., 1981. Fracture geometry of pseudotachylyte genera-tion zones: A study of shear fractures formed during seismicevents. J. Struct. Geol. 3, 169–178.

Groshong Jr., R.H., 1988. Low-temperature deformation mech-anisms and their interpretation. Geol. Soc. Am. Bull. 100,1329–1360.

Herrmann, H.J., Mantica, G., Bessis, D., 1990. Space-fillingbearings. Phys. Rev. Lett. 65, 3223–3226.

Hickman, S.H., 1991. Stress in the lithosphere and the strengthof active faults. Rev. Geophys. Suppl. 29, 759–775.

Hobbs, B.E., Ord, A., Teyssier, C., 1986. Earthquakes in theductile regime? Pure Appl. Geophys. 124, 309–336.

House, W.M., Gray, D.R., 1982. Cataclasites along the Saltvillethrust, U.S.A., their implications for thrust-sheet emplacement.J. Struct. Geol. 4, 257–269.

Hull, J., 1988. Thickness-displacement relationships for defor-mation zones. J. Struct. Geol. 10, 431–435.

Lin, A., 1996. Injection veins of crushing-originated pseudo-tachylyte and fault gouge formed during seismic faulting. Eng.Geol. 43, 213–224.

Little, T.A., 1995. Brittle deformation adjacent to the Awaterestrike-slip fault in New Zealand: Faulting patterns, scalingrelationships, and displacement partitioning. Geol. Soc. Am.Bull. 107, 1255–1271.

Logan, J.M., 1975. Friction in rocks. Rev. Geophys. Space Phys.13, 358–361.

Logan, J.M., Friedman, M., Higgs, N.G., Dengo, C., Shimamoto,T., 1979. Experimental studies of simulated gouge and theirapplication to studies of natural fault gouge, Proc. Conf. VIII,Analysis of Actual Fault Zones in Bedrock. U.S. Geol. Surv.Open-file Rep., pp. 305–340.

Logan, J.M., Dengo, C.A., Higgs, N.G., Wang, Z.Z., 1992.Fabrics of experimental fault zones: Their development andrelationship to mechanical behavior. In: Evans, B., Wong, T.(Eds.), Fault Mechanics and Transport Properties of Rocks.Academic Press, London, pp. 33–68.

Marone, C., Kilgore, B., 1993. Scaling of the critical slip dis-tance for seismic faulting with shear strain in fault zones.Nature 362, 618–621.

Mase, C.W., Smith, L., 1987. Effects of frictional heating onthe thermal, hydrologic, and mechanical response of a fault. J.Geophys. Res. 92, 6249–6272.


Matti, J.C., Morton, D.M., 1993. Paleogeographic evolution ofthe San Andreas fault in southern California: A reconstruc-tion based on a new cross-fault correlation. In: Powell, R.E.,Weldon II, R.J., Matti, J.C. (Eds.), The San Andreas Fault Sys-tem: Displacement, Palinspastic Reconstruction, and GeologicEvolution. Geol. Soc. Am. Mem. 178, 107–160.

Matti, J.C., Morton, D.M., Cox, B.F., 1985. Distribution andgeologic relations of fault systems in the vicinity of the cen-tral Transverse Ranges, southern California, U.S. Geol. Surv.Open-file Rep. 85-365.

Means, W.D., 1995. Shear zones and rock history. Tectono-physics 247, 157–160.

Melosh, H.J., 1979. Acoustic fluidization: a new geologic pro-cess? J. Geophys. Res. 84, 7513–7520.

Melosh, H.J., 1996. Dynamical weakening of faults by acousticfluidization. Nature 379, 601–606.

Miller, M.G., 1996. Ductility in fault gouge from a normal faultsystem, Death Valley, California: A mechanism for fault-zonestrengthening and relevance to paleoseismicity. Geology 24,603–606.

Morton, D.M., Matti, J.C., 1987. The Cucamonga fault zone,geologic setting and Quaternary history. U.S. Geol. Surv. Prof.Pap. 1339.

Noble, L.F., 1954. Geology of the Valyermo Quadrangle andvicinity. U.S. Geol. Surv. Map GQ-50.

Oakeshott, G.B., 1971. Geology of the epicentral area. Calif.Div. Mines Geol. 196, 19–30.

Power, W.L., Tullis, T.E., 1989. The relationship between slick-enside surfaces in fine-grained quartz and the seismic cycle. J.Struct. Geol. 11, 879–894.

Power, W.L., Tullis, T.E., 1991. Euclidean and fractal models forthe description of rock surface roughness. J. Geophys. Res.96, 415–424.

Reches, Z., Lockner, D.A., 1994. Nucleation and growth of faultsin brittle rocks. J. Geophys. Res. 99, 18159–18173.

Rice, J.R., 1983. Constitutive relations for fault slip and earth-quake instabilities. Pure Appl. Geophys. 121, 443–475.

Rice, J.R., 1992. Fault stress states, pore pressure distributions,and the weakness of the San Andreas fault. In: Evans, B.,Wong, T.F. (Eds.), Fault Mechanics and Transport Propertiesin Rocks. Academic Press, London, pp. 475–503.

Rudnicki, J.W., 1980. Fracture mechanics applied to the earth’scrust. Annu. Rev. Earth Planet. Sci. 8, 489–525.

Rutter, E.H., Maddock, R.H., Hall, S.H., White, S.H., 1986.Comparative microstructures of natural and experimentallyproduced clay-bearing fault gouges. Pure Appl. Geophys. 124,3–30.

Scholz, C.H., 1990. The Mechanics of Earthquakes and Faulting.Cambridge Univ. Press, New York, NY, 439 pp.

Scholz, C.H., Dawers, N.H., Yu, J.Z., Anders, M.H., Cowie, P.A.,

1993. Fault growth and fault scaling laws: Preliminary results.J. Geophys. Res. 98, 21951–21961.

Schulz, S.E., Evans, J.P., 1998. Spatial variability in microscopicdeformation and composition of the Punchbowl fault, southernCalifornia: Implications for mechanisms, fluid–rock interac-tion, and fault morphology. Tectonophysics 295, 225–246.

Segall, P., 1991. Fault mechanics. Suppl. Rev. Geophys. 29, 864–876.

Sibson, R.H., 1973. Interactions between temperature and fluidpressure during earthquake faulting, a mechanism for partialor total stress relief. Nature 243, 66–68.

Sibson, R.H., 1977. Fault rocks and fault mechanisms. J. Geol.Soc. London 133, 191–213.

Sibson, R.H., 1986. Brecciation processes in fault zone: Infer-ences from earthquake rupturing. Pure Appl. Geophys. 124,159–175.

Sibson, R.H., 1989. Earthquake faulting as a structural process.J. Struct. Geol. 11, 1–14.

Sibson, R.H., Moore, J.M., Rankin, A.H., 1975. Seismic pump-ing — a hydrothermal fluid transport mechanism. J. Geol. Soc.London 131, 653–659.

Sleep, N., Blanpied, M.L., 1992. Creep, compaction, and theweak rheology of major faults. Nature 359, 687–692.

Swanson, M.T., 1988. Pseudotachylyte-bearing strike-slip duplexstructures in the Fort Foster Brittle Zone, S. Maine. J. Struct.Geol. 10, 813–828.

Swanson, M.T., 1989. Sidewall ripouts in strike-slip faults. J.Struct. Geol. 11, 933–948.

Swanson, M.T., 1990. Extensional duplexing in the York Cliffsstrike-slip fault system, southern coastal Maine. J. Struct.Geol. 12, 499–512.

Swanson, M.T., 1992. Fault structure, wear mechanisms and rup-ture processes in pseudotachylyte generation. Tectonophysics204, 223–242.

Wallace, R.E., Morris, H.T., 1986. Characteristics of faults andshear zones in deep mines. Pure Appl. Geophys. 124, 107–125.

Weldon II, R.J., Meisling, K.E., Alexander, J., 1993. A specula-tive history of the San Andreas fault in the central TransverseRanges, California. In: Powell, R.E., Weldon II, R.J., Matti,J.C. (Eds.), The San Andreas Fault System: Displacement,Palinspastic Reconstruction, and Geologic Evolution. Geol.Soc. Am. Mem. 178, 161–198.

Wojtal, S., Mitra, G., 1988. Nature of deformation in some faultrocks from Appalachian thrusts. Geol. Soc. Am. Spec. Pap.222, 17–33.

Woodburne, M.O., 1975. Cenozoic stratigraphy of the TransverseRanges and adjacent areas, southern California. Geol. Soc.Am. Spec. Pap. 162, 91.

Documents

Ultracataclasite structure and friction processes of … and...ELSEVIER Tectonophysics 295 (1998) 199–221 Ultracataclasite structure and friction processes of the Punchbowl fault,