JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO.

B5, PAGES 4826-4838, APRIL 10, 1986

Extensionof ContinentalLithosphere' A Model for Two Scalesof Basin and Range Deformation
M.T. ZUBER I AND E.M. PARMENTIER
Dept. of Geological Sciences,Brown University, Providence,Rhode Island
R.C. FLETCHER

Centerjbr Tectonophysics, TexasA&M University, College Station, Texas

We develop a model for deformationin an extendingcontinentallithosphere that is stratifiedin density and strength, assuminga rheology consistentwith seismic focal depths and experimentalflow laws. The model demonstrates that necking instabilitiesat two wavelengths will arise due to the presenceof a strong upper crust and upper mantle separated by a weak lower crust. The magnitudes of the instabilitiesare directly related to strengthcontrasts within the lithosphere,while the dominantwavelengths of neckingare controlledmainly by the thicknesses of the stronglayers. The resultsare applied to the Basin and Range Province of the western United States where two scales of deformation can be recognized, one corresponding to the spacingof rangesand the other to the width of tilt domains.A Bouguer gravity anomaly and associated regionaltopography with a wavelengthcomparable to the width of tilt domainshas also been recognized.For plausibledensityand strengthstratifications, our resultsshow that the horizontalscale of shortwavelengthneckingis consistent with the spacingof individualbasinsand ranges,while that of the longer wavelengthnecking is consistent with the width of tilt domains. We thus suggestthat Basin and Range deformationmay be controlledby two scalesof extensional instability.Extensionin the weak lower crust in this model is laterally displacedfrom regions of upper crustal extension. The resultant horizontal shearing in the lower crustmay be a mechanism for the initiationof low-angleextensional detachments.

mation occursin the upper crust, where strengthis controlledby frictional propertiesand the mode of faulting [Byerlee, 1968]. The continental lithosphere is stratifiedin both densityand At greaterdepths,dislocation creep, which is chiefly a function strength. In this study, we examine the implicationsof this of strain rate and temperature, is the dominant deformation stratification for large-scaleextensional deformationof the litho- mechanism.This resultsin strengththat decreases sharplywith sphere. In a previous study, Fletcher and Hailer [1983] treated depth. In constructing Figure 1 we assume,as in other studies the lithosphere as a strongsurfacelayer of uniformstrength and of continental deformation [Glazner and Bartley, 1985; Bird, densityoverlyinga weakersubstrate of the samedensity.Their 1978], that flow in the continentalcrustcan be approximated by study showedthat unstableextension,or boudinage, resultsin a quartztheologyand in the upper mantleby an olivine theol-

INTRODUCTION

assumption for the the concentration of extension into regions with a regularspac- ogy. While this is probablya reasonable ing determined primarily by the thickness of the stronglayer. upper mantle, it must be considereda lower limit for the crust. a majorconstituent of the continental crust,is stronger A more realistic strengthstratificationof the continentallitho- Feldspar, greaterthan 300C[Tu!!isand Yund, spherewould consistof a strongupper crust and upper mantle thanquartzat temperatures separated by a weak lower crust, with the densityof the crust 1977, 1980]. However, uncertainties in the flow law of feldof the mechanics of polylessthan that of the mantle. If a strongsurfacelayer necksat a spar, as well as in our knowledge given wavelength,then another strong region at depth may phaseflow and the modal mineralogyof the lower crust, conof introducea secondwavelength of necking.In this study, we strain us to assumethat quartz theology is representative results evaluatethe conditions requiredfor the growth of two wave- crustal strength.As shown in Figure 1, experimental lengthsof necking instabilityand apply this hypothesis to the extrapolatedto lower crustal conditionsindicate that olivine is thanquartzat similarpressures andtemperatures, Basin and Range Province, which, as we will presently muchstronger as is apparent by the sharp increase in strength at the crust-mandescribe, exhibitstwo scales of periodicdeformation. tle boundary. In fact, despite highertemperatures at depth,the StrengthStratificationof ContinentalLithosphere uppermantleis even stronger than the uppercrust, while the Studiesof the theologyof the continental lithosphere [e.g., lowercrustis a regionof very low strength. For the conditions assumed in Figure 1, the uppermantle Kirby, 1983; Brace and Kohlstedt, 1980] suggest a variationof undergo brittledeformation; however,strength in ductile strength with depth like that illustratedin Figure 1, which was should to the geothermal gradient andthe acticonstructed following Brace and Kohlstedt[1980]. Brittle defor- flow is highlysensitive vationenergyand preexponential frequency factor in the flow law. A smallchangein any of theseparameters would shift the depth of the boundarybetweenbrittle and ductile deformation INowat Geodynamics Branch, NASA Goddard Space Flight Cenand couldeliminatethe regionof brittle fracturein the mantle. ter. Because of the uncertainties in lithosphere theologywe do not Copyright1986by the American Geophysical Union. wish to emphasize absolute magnitudes, but rather relative differences in strength. Brittle deformation in the uppermantleis Paper number 4B5282. 0148-0227/86/004B-5282505.00 not requiredfor a large strength contrast at the Moho, sincethe
4826

ZUBER ET AL ' EXTENSION OF CONTINENTAL

LITHOSPHERE

4827

LITHOSPHERIC

STRENGTH

lineament patterns; however, in many cases the zones are not

associated with identified structural features.

-,soo

o':5
o
E

(MPa)
500
!

-,ooo

-500

ooo
1

compress

tension

The spacingsof the rangesand tilt domainsare summarized in Figure 2. Tilt domain spacings,shown in the top histogram, were measured between antiformal axes and synformal axes, normal to strike, from Stewart's [1980] Figure 1. Distances between domain boundariesrange from as little as 50 km to
almost 500 km, but most cluster around 200 km. This is in con-

trast to the distancesbetween individual ranges, shown at the bottom of Figure 2. This histogram, reproducedfrom Fletcher and Hallet [1983], shows that rangesare separatedby an average of about 30 km, or about 1/6 the spacing of the tilt
domains.

In a previous study of large-scaleextensionin the Basin and Range, Fletcher and Hallet [1983] treated the lithosphereas a strong plastic layer overlying a weaker viscoussubstrateof the same density. Using flow laws for a range of rock types, they found that the lithosphere extendsunstablyproducingregionsof enhanced and reduced extension. This study showed that the 60 Fig. 1. Strengthof the continentallithospherewith a crustalthickness dominant wavelengthof the necking which arisesdue to unstaof 30 km, ;x= 10 -ss - anda geothermal gradient of 15 K km ble extensionis consistentwith the spacingof individual basins Flow laws were taken from Brace and Kohlstedt {19801. Major contrasts andranges. in strengthoccur (1) between the brittle upper crust and ductile lower On the basis of our results for extensionof a strengthand crust, (2) at the crust-mantleboundary, and (3) within the mantle. Such density stratified lithosphere, we suggestthat tilt domains may strengthcontrastsmay lead to the growth of instability during uniform also be the surfaceexpression of boundinage-like deformationin extension. the Basin and Range, with a wavelength greater than the spacingof ranges.We suggest that the developmentof this secductile strengthof olivine is significantlygreater than that of ond, longer wavelengthof instability may be related to a strong quartz at the same PT conditions. region of the upper mantle separated from the strongupper crust A zone of low strengthin the continental lithosphere is also by a weak lower crust. Kinematic models of basin and range consistent with seismicresults.In a studyof the distributionof structure have been proposed which address the relationship earthquakefocal depths in continentalregions not associated between the ranges and tilt patterns [Anderson et al., 1983; with recent subductionzones, Chen and Molnar [1983] found Zoback et al., 1981; Stewart, 1980, 1978, 1971]. To our the upper crust and upper mantle to be seismicallyactive and knowledge, however, no mechanical models have been the lower crust to be essentiallyaseismic.They interpretedthe proposedto explain the distribution of large-scale tilted fault aseismiclower crest as a weak region which deformsby ductile blocks. In the present study we examine the possibility that flow. Thus both experimental and observationalevidence indicate that the continental lithospherewill consist of a strong SPACING OF TILT DOMAINS uppercrustand mantleseparated by a weak lower crust.
8 Mean-194Km
N-27

Basin and RangeProvince

As with other regionsof continentalextension,the Basin and Range exhibits high heat flow, extensive volcanism, regional uplift, and widespreadnormal faulting. Details of the structure and geophysics of this area are summarizedby Eaton [1982], Zoback et al. [1981], Stewart [1978], Thompsonand Burke [1974], and others.Modem Basinand Rangetopography, which is regularly spacedand has a crestto trough amplitudeof about 1 km, initiated at about 13 Ma [Zoback et al., 1981] and trends generallyN-S in the northernpart of the province and NW-SE in the south.The area is also markedby a broad-scale regional tilt patternof major Cenozoicfault blocks, which strike generally parallel to the basin and range structure. Stewart [1980] recognizedthese alternatingregionsof consistent tilt directions,
called tilt domains, which are continuous over distances of

cr
8

I00

200

3,00

x(Km)

SPACING OF RANGES

:::::::::..I..::::::::::::::::: Mean--:51 Km ::::::::::::::::::::::::: N 27

4
0 '.

!ii}ii}i i 71 er and Hallet (1983)

:0
, ,

50-500 km along strike. The tilt domains are separatedby "antiformal" and "synformal"axes, which are shown schematically in Figure 8. Within a given tilt domain, fault blocks dip away from antiformal and toward synformal axes. Transverse zones, which are characterizedby an absenceof major tilted blocks and by changesin fault patternsand topographic grain, strike parallel to the extensiondirectionand separateregionsof differing tilt. In some places, transverse zones follow regional

6b x(Km)

Fig. 2. Spacings of large-scale deformational featuresin the Basin and Range Province. The spacings of tilt domains (top) are distances between successiveantiformal boundaries and synformal boundaries measured from Figure I of Sten'art11980]. Spacings betweensuccessive ranges(bottom)are reproduced from F/etcherand Ha/let 11983].

4828

ZUBER ET AL.'

EXTENSION OF CONTINENTAL LITHOSPHERE

unstableextensionof continentallithosphereleads to two scales of deformation, one of which correspondsto the spacing of ranges and the other to the spacingof tilt domains. We show that stressesassociatedwith the long wavelength deformation result in differencesin resolved shear stresson oppositelydipping high-angle normal faults forming individual basins and ranges. Greater slip on faults with higher resolved shear stress could result in the observedrotation or tilting of high-anglefault blocks. Recently, Froidevaux [1985] has suggested a boudinagelike thermal structureof the upper mantle on the basisof gravity anomaly patterns. The 200 km wavelength gravity anomaly identified on the maps of Hildenbrand et al. [1982] coincides approximatelywith spacingof tilt domainsand thus may also be relatedto the longerwavelengthof deformation(seeFigure 8).
MODEL FORMULATION

Instabilityin the lithosphere,extendinghorizontallyat a uniform or mean rate ;xx, occurs due to the amplification or growth of small, random disturbancesalong initially planar interfaces.As the disturbances grow, the interfaces deform sinusoidally with an amplitude A. We assume(1) two-dimensional flow, and (2) the amplitude of the disturbance to be small compared to its wavelength.Mathematically,this is regardedas the superposition of an unstablesecondary or perturbing flow on the mean or primary flow which describes the uniform extension of the medium. This linearized formulation describes only the incipientstages of unstable extension.
Linearization

In the linearized problem, stresses and strain rates for the

total flow are written

We examine a model of the continentallithosphere which, as shown in Figure 3, is composedof three layers and a substrate. The crust consistsof two layers of thickness h and h2, each with a uniform density (@, @=@2)and strength(x, x>x2). The mantle consists of a layer of thickness h3 overlying either a semi-infinite half-space with uniform strength and density (J model) or a half-space in which the strengthdecreases exponentially with depth (C model). In the former case, the strongmantle region is represented as a layer of uniform strength,and the strengthdecreases discontinuously to a lower uniform value in the substrate.In the latter, mantle strengthis everywherecontinuous but falls to zero at depth. Models with either uniform or exponentially varying strength were chosen because analytic solutionscan b obtained for the flow in each layer, thus avoiding the need for a fully numericalsolution. Neither model is an exact representation of the rheologicalstructureshown in Figure 1. Variations in the strengthof the lithospherein regionsof uniform compositionare not abrupt, as suggested by the use of discrete layers. However, the rapid decreasein strengthwith depth associated with ductile flow results in a relatively rapid transition from strong to weak. As will be shown later, both models result in similar dominant wavelengthsand instability growthrates.

(1)

where an overbardesignates the mean flow and a tilde the perturbing flow. For thermally activated creep the relationship between the principal stresses Ol and o3 and the strain rate e
has the form

e = A (ol - o3) nexp(-Q/RT)

(2)

whereA is the frequencyfactor, Q is the activationenergy,R is the gas constant, T is the temperature, and n is the stress exponent. The principalstresses and the strainrate are relatedby the the viscosityIt as
o,-o3 = 4 (3)

By substituting (3) into (2) to eliminatethe strainrate, the viscositycanbe defined

1/4) A (O 1 --03)l-n exp(Q/RT)

are written

(4)

In an isotropicviscousfluid the stress-strain rate relationships

XeU
!

EXTENSION

MODEL

% = 2[t%- p6,j

(5)

wherep is the hydrostaticstress.By combiningthe expressions for the normal stresses in (5) and the incompressibility condition (e,,= 0), the strength (x) is defined
x = (Oxx-Ozz)/2 = 2g:xx
(6)

In the presentstudy, Ozz=0 in excessof the hydrostaticstress since the mean extension is entirely in the x direction. As described by Fletcher and Hallet [1983], the relationship betweenthe perturbingstresses and strain rates is obtainedby substituting (1) and (6) into (5), expanding,and retainingterms

to firstorder in
Oxx= (u/n) (;xx/;xx) - P
Fig. 3. Model of a density and strength-stratified lithosphere. The lithosphereconsistsof a strong upper crust, weak lower crust, and a strongupper mantle over a weaker mantle substrate. Densitydifferences occur at the free surface and crust-mantleboundary. Horizontal extensioncauses initially planarlayersto deform into sinusoidal shapes.

Oz = (u/n) (zz/xx) - P
Ox: (xz/xx)

(7)

ZUBER ETAL ' EXTENSION OFCONTINENTAL LITHOSPHERE

TABLE 1. Dimensionless Parameters

4829

Parameter

Definition

Sl S2 S3 S4
R
R2 R3

()1-- @o)ghl/%l (02- @l)ghl/%2 (03- @2)ghl/%3 (04- @3)ghl/4

x/x2
x/x3 Xl/X4

law fluid in which the stressexponent nr--> oz. This approximatesa perfectlyplastic material in which deformationcan be discontinuous across sliplines, which may representincipient faults. A perfectlyplasticmaterialmay ,thus be a representative rheologicalmodel for a region which deforms by pervasive faulting [Chappie, 1978]. In this study, we use a value of

n= 10 4 for reasons discussed in Appendix 1. Thelower crust

and the mantle are modeled as uniform power law fluids with n2= n3- n4= 3, which is characteristic of deformationdue ,todislocationcreep. To examine the general characterof the deformationin the

strength jump (J) model, we assume that the lower crustis 100 times weaker than the upper crust, that the upper mantle is twice as strongas the uppercrust, and that the mantle substrate is 50 times weaker than the upper crust. England [1983], in a A solution for the perturbing flow is foundby requiring that the study of continental extension, also assumed that the upper to lithospheric strength. perturbing stresses and strainratessatisfy the equilibrium equa- mantle made the greatestcontribution tionsandtheincompressiblity andcompatibility conditions. Strength estimates are based on experimental results and on the observedstressreleasedin shallow earthquakes (tens of megaGrowth Rate Factor

-Jhl

pascals). In thecontinuous strength (C) model, themantle layer

of 5 km and the e-folding depth of the substrate The amplitudeof vertical deformationat an interface has a thickness is taken to be 6 km in accordance with Figure 1. A,(k',t) is proportional to the exponential of the meanhorizontal strain ;xxt as

A,(k',t) oc exp[(q-1);xxt]

(8)

Figure 4 showsthe growthrate factor q as a functionof the dimensionless wave number k' for three mantle strength stratifications: curves a and b are for lithosphereswith strong

uppercrustal and uppermantlelayersfor the J and C models,

whereq is the growthrate factorand t is time. The growthrate factordetermines the rate of amplification of a disturbance with dimensionless wave numberk' (=2jrh/,). If q>l, the lithosphere will extend unstably, andthe instability will growfastest
at the wave numberat which q is maximized. At each value of the wave number there are four values of q (equal to the number of interfaces); however, in generalonly one value will be

respectively,and curve c is the sameas curve a but without the strong mantle layer. In both curves a and b the growthrate spectrumhas two peaks:that at the larger wave number representsa shortwavelength instabilitywhich arisesdue to necking of the stronguppercrustand that at the smallerwave numberis a longer wavelengthinstabilitywhich arisesdue to the presence of the strongregion of the upper mantle. The J and C models

t physical behavior at large wave numbers as dispositiveand thuscontribute to the growthof instability. The havedifferen, determination of q requiresa solutionof a linear systemof cussedin Appendix 2 for the simple case of a single strong equations to satisfythe velocityand stress matching conditions layer. However, as shownin Figure4, both modelspredictsimilar p,hysical behavior for wave numbers of interest (k'<Jr) and at eachinterfaceasdiscussed in Appendix1. The similarity of thesetwo modThe problemin its mostgeneral form is described by a num- similar dominantwavelengths. that the existence of two wavelengths of instabilty ber of dimensionless parameters which are listed in Table 1. els suggests
S-S4 are ratios of buoyancy to strength at each of the interfaces.In the presentproblem, S2 and S4 are set to zer9
becausethe crust and mantle are of uniform density. In the J

model, R-R3 are the ratios of the strengths of the various regions. In the C model,the strength at the mantlelayer-substrateinterface is continuous so R3= 1. Here fl0w is determined by the parameter or,the ratioof theviscosity decaydepthin the substrate to the stronglayer thickness.The values of these parameters determine whether the lithosphere will extendunstably, and if so, the dominant wavelength at whichnecking will
Occur.

20o
\
q

To consider the implications of strength and density stratification in the continentallithosphere,we determinethe

growthrate spectrum q(k')for g rangeof modelparameters. Theresults arethenapplied, to discuss theobserved wavelengths ofdeformation inthe Basin find Range Province.
RESULTS

--- J model
-C model

Fig. 4. Growthrate of the instability q as a function of dimensionless

with stronguppercrustaland uppermantlelayers(R = 100, sphere like that shown inFigure lYnwhich a strong upper crUs[model =0.5, R = 50); curve b, C model with strongupper crustaland upper
.

We consider a strength stratification for the continental litho-

wave number k' for three mantle strength stratifications:curve a, J

density isassumed within the crus.t (@c = 2950 kgm -3)and man- gle strong surface layer(R = 100,R2 =R = 1). For all models S: S=0.05 S-,:S4=0, !l= l04,and 11,=11=114=3. tle (@m=3200'kg m-3). Theupper crust is treated asa po;Wer

andmantle 19yer areseparated by a weak lower crust. ,Uniform mantlelayers(R =

100,R, =0.5, c = !.2): curvec, J modelwith a sin1.2,

4830

ZUBER ET AL.' EXTENSION OF CONTINENTAL LITHOSPHERE

maximum within the strong mantle layer, and its magnitude decaysexponentiallywith depth. The deformationpenetrates to a depthcomparable to its wavelength. Our interpretation of theseresultsin the contextof Basin and Rangedeformation is that the shorterwavelength instabilitymay control the spacingof individual ranges,and the longer waveMoho lengthmay controlthe spacings of long wavelengthtopography b) long X and the tilt domains. In the short wavelengthcase, regionsof enhanced extension and thinning correspondto basins, and regionsof apparent thickening correspond to ranges.In the long wavelength case, the peaksand troughsin the long wavelength c )comb,ned Moho displacement profiledefineregionsin which fault blockschange tilt directionsas determinedby the senseof horizontal shear. Regions of alternatingpositive and negative surface gradient correspond to Stewart'stilt domains. We will now considerthe influence of lithospherestructure and theology on the the predictedwavelengthsof deformation Fig. 5. Displacement fieldsassociated with {a) short,(b) long, and (c) focusing primarily on the J model. In an extending layered combined wavelengths of neckinginstability. The shortwavelength of medium, an increasein density with depth acrossan interface deformation clearly showsthe neckingof the brittle uppercrust. Note reducesthe vertical componentof the perturbingflow and acts thepresence of shearing withinthelowercrustin Figure5c. to confine deformationto within regions of uniform density. This effect is most notableat shortwavelengths where, as illustrated in Figure 5, the relative magnitudeof deformationat the does not depend on details of the strengthstratification.The surface and at the crust-mantleboundary is significantly growth rate spectrumfor a single strong layer overlying a subdifferent for the long and short wavelengthdeformation.The strate with uniform strength(Figure 4 curve c) has a magnitude long wavelength deformation has a large amplitudeat the crusteverywheregreaterthan that for the multilayerJ model, and the mantle boundaryrelative to that at the surface. In contrast,the dominant wave number occurs at a point intermediatebetween short wavelength deformation nearly vanishes at the the short and long wavelength peaks. This suggeststhat the crust-mantle boundary.This result, which predictsa flat Moho presence of a strong subsurface layer acts to damp out on a scaleof tensof kilometers,holdsfor any reasonable range instability over a range of wave numbers with the greatest of crust-mantle density contrasts (i.e., @ .... t<@ .... fie). The damping occurringat the minimum betweenthe short and long densityincreaseacrossthe crust-mantle boundaryalso acts to wavelength peaks. The long wavelength instability is thus decrease the overall magnitude of the growth rate spectrum, but driven by necking of the surface layer rather than by the tenfor the range of physically realistic conditionsthe effect is dency of the mantle to neck. The mantle layer flexes in small. responseto necking of the surface layer but does not itself The amplitudes of the peaksin the growth rate spectrum are undergonecking. Further discussion of the mechanism of long directlyrelatedto the strengths of the stronglayers.The effects wavelength deformation is presented in Appendix1. of variations in lithosphericstrength on the behavior of the Figure 5 shows the crustal displacementfields associated growthrate functionin the absence of buoyancyforcesand with with the short, long, and combined dominant wavelengthsfor h = h2= h = 15 km for the J model are summarized in Table 2. the J model. These diagrams were constructed by calculating displacements at equally spacedpoints on an initially rectangular grid and connectingthe tips of the displacement vectors. In the linearized theory, the amplitudes of the long and short a) W(k z) b) W(k',z) -I 0 wavelength deformation are each proportional to their initial amplitude. Thus the relative amplitudesare not determinedby the growth rate factors alone. The amplitudesshown in Figure crust 5 were chosen to clearly show the character of deformation at 30 mantle each wavelength. In all cases the mean extension has been removed, so that areas of apparentcompression are in reality areasof minimum extension.In the short wavelengthcase, the 6O 6O effect of necking of the strong upper crust, as well as the changein deformationalstyle at the brittle-ductiletransitioncan be clearly observed. In the longerwavelength case, the upward 90 90 deflection of the crust-mantle boundary is the most obvious deformational feature. Flow in both the crust and underlying
CRUSTAL DEFORMATION FIELD
Illllil!\\\\II llllllllll\\\\ I !

Zl i11 I

mantlecan be represented in termsof the amplitude of the vertical velocity W(k',z) at a given wave number. W(k',z), normalized to unity at the surface(z--0), is shownin Figure 6 for the
two dominant wave numbers in the J model. For the short

wavelengthdeformation,the vertical velocity at the base of the crust and in the underlyingmantle is small, as reflectedin the small relief on the crustmantle boundaryin Figure 5a. The vertical velocity of the long wavelength deformation reaches a

Fig. 6. Verticalvelocity asa function of depthfor (a) shortand(b) long wavelengths of instability.Amplitudes are normalizedto a surfacevalue of unity. Note that for the short wavelength casedeformationis a maximum in midcrustal regions, whilefor the long wave-

lengthcasedeformation peaksin the lowercrustand strongupper

mantle.

ZUBER ET AL..' EXTENSION OF CONTINENTAL LITHOSPHERE

4831

TABLE 2.
R1
100 100 100 100 100 100 50 50 50 25 25 25 25

Effect of Strength Contrasts on the DominantGrowthRate Factor

R2
0.5

R3
50 50 100 100 50 100 50 50 50 25 50 25 50

qai
91.13
105.14
117.85

qd 2
167.19 167.94 167.94 167.19 169.40 169.42 85.42 86.18 87.65 43.43 43.43 44.17 44.17

kal'
0.60 0.60 0.60 0.60 0.65 0.65 0.60 0.65 0.70 0.65 0.60 0.70 0.65

kd2'
2.30 2.30 2.30 2.30 2.30 2.30 2.30 2.30 2.30 2.30 2.30 2.30 2.30

ai,km
157 157 157
157 145 145 157 145 135 145 157 135

a2,km
41 41 41
41 41 41 41 41 41 41 41 41

1 1 0.5 2 2 0.5 1 2 0.5 0.5 1 1 1 0.5 1 2 1 0.5

1
1

97.13 119.71

140 65 59 38 70 91 82 26

25 20 20 20 20 10
10
10

25 50 50 50 20 50
50
100

35 64 39 93 41 33 48.90 41.33 35.34 43.34

51.06 34.54

145
135
145

41
41
41

44.17 34.94 35.70

0.70 0.65 0.65

2.30 2.30 2.25

145

42

37.23
35.68

0.70
0.70

2.25
2.25

135 135 145 135 145

145 135

42 42 42 42 42
42 44

10

0.5

100

10 10 10
5

2 2 1
1

100 50 10
5

23.89 29.12 32.05 25.59 38.33 33.86 19.20 10.35

4.48

2 1

2 1

2 1

17.86 18.61 18.61 17.86 20.13 20.12 18.58 9.91 4.52

--

0.65 0.70 0.65 0.65 0.70 0.75 0.75 0.80 1.10

1.55

2.25 2.25 2.25 2.25 2.15 2.15 2.25 2.15 1.85

126

44

2.00

126 118 86 61

42
44 51 --

R, R2,andR3aredefined in Table1. qd,kd',andkaarethedominant growth ratefactor,wavenumber, andwavelength, respectively.

The function is most sensitiveto changesin the strengthof the uppercrustrelativeto that of the other regions.An increasein the relative strengthof the upper crust increases the amplitudes of both short and long wavelength peaks but increasesthe shorterwavelengthpeak to a greaterextent. A relative increase in the strengthof the strongupper mantle region decreases the amplitudes of both peaks,but the decrease is lessfor the shorter wavelength peak. The variation in the dominant wavelengths due to the differences in assigned layer strength is not significant. The presence of a weak lower crust between a strongupper crust and mantle is a prerequisitefor a secondwavelengthof instability. As shown in Figure 4, a single stronglayer over a uniform viscoussubstrate has a growth rate spectrumwith a single maximum. A strength-stratified crust overlying a uniformly strongmantle is an insufficient conditionfor two scalesof instability; the mantle must containa regionof high strength,though as shownby the growth rate spectrum for the C model, a discontinuityin strength within the mantleis not required. If the lower crust and mantle deform by steady state creep the value of n is probably about 3, but variations within the range 1-3 have a negligible effect on instability growth rate. For a single layer embeddedin a weaker medium both with n---1, the growth rate is less than unity in the absenceof an unstabledensity stratification[Smith 1977]. On this basis it might be expectedthat no necking instability would occur in a strongupper mantle with a linear theology. However, a strong upper mantle region with n3-I beneatha surface layer with large n does show a weak long wavelengthinstability, which supports the contention that this instabilityis driven by necking of the surface layer (see Appendix 1). If brittle deformation occursin the upper mantle, as suggested in Figure 1, then flow in this region may be best described by a large stress exponent.

Increasingn in the strongpart of the mantle enhancesthe instability of the longerwavelength of deformation. Dominant wavelengthsare determined primarily by layer thicknesses. Smith [1979] showedthat a single layer with large n embeddedin a weaker medium with n= 1 extends unstably with a wavelength to layer thickness ratio of ./h--4. While our resultsagree with Smith's for the limiting case, the presenceof another strong layer (with finite n) and the inclusion of buoyancy forcesat the interfacesmarkedly changethis ratio. Table 3 is a compilationof the dominantwavelengths of instabilityfor a rangeof layer thicknesses with R = 100, R2= 0.5, and R3= 50 for the J model. Buoyancyeffectshave been includedhere because the dimensionlessparametersS-S4 are a function of layer thickness.Results show that ./h ranges from approximately 2.8 to 3 for the shorterwavelength instability and from about 11 to 15 for the longer wavelength instability. However, ./(h + h2+ h3)'-4for the longerwavelength. Increasing the thicknessof the surface layer, keeping the thicknesses of the other layersconstant,increases both dominant wavelengths.A decreasein the thicknessof the strong upper mantle increasesthe longer dominant wavelength but has no apparenteffect on the shorterwavelength.A decreasein the thickness of the weak lower crust may slightly decrease the dominantwavelengthfor the longer wavelengthpeak, but again the shorterwavelengthpeak is unaffected.These results show that the lower crustal and upper mantle layers influence the behavior of the growth rate spectrumto a lesser extent than doesthe surface layer.
APPLICATION TO THE BASIN AND RANGE PROVINCE
We will now consider the observed structure of the Basin

and RangeProvincein termsof theseresults and examinepossi-

4832

ZUBER ETAL.' EXTENSION OFCONTINENTAL LITHOSPHERE

TABLE 3.Dominant Wavelengths for aRange ofLayer Thicknesses

h,km
5 10

h2,km
5 5

h3,km
5 5

kdl'
o si
0 5o

kd2'

kd!,km
126
137
121 56 60 58 126

ka,,km
_

kd2/hl
2.78

2..'6
224 2 24
2 22 2 26 2 26

58

28
28
28 14 14 14 28

14

12.56

2.80 2.80 2.83 2.78 2.78 2.78 2.83 2.86 2.86 2.86 2.83 2.83 2.83
2.86

10 10 5 5 5 10 15 15 15 10
10 10

10 5 5 10 10 10 10 15 10 10
20 15

5 10 10 5 10 10 10 10 15 15
10 15

0 46
052
0 56

.b6
12.08
11.22

052
0 54

12.08 11.64 12.57 13.09

13.66

2 26
2 20

0 5o
0.48

2:20
2.20 2.20 2.22 2.22 2.22 2.20 2.16 2.16
2.16

196 205

43 43

0.46 0.50 0.52

0.48

188
121

43
28

12.57 12.08 13.09 12.57 13.09 13.65

131 126 196 273

28 28 43 58

0.50
0.48

15 20 20 20 15

15 15 20 15 15

15 15 15 20 20

0.46 0.44

2.91
2.91 2.91
2.86

286
273

58
58

14.28
13.66

0.46
0.48

2.20

196

43

13.09 13.66 13.09

13.66

15
15 20 20 20 20 10 10 10 25

20
20 20 10 20 10 10 20 20 20

15
20 20 10 10 20 20 10 20 20

0.46
0.48

2.18 2.18
2.16

205
196 273
273

43
43 58
58

2.88

2.88
2.91

0.46 0.46
0.42

2.16
2.16

13 66 14 96 12 57 12 08

291
291

299

58

0.50

2.16 2.22 2.22

251
121

58
28

291
283 283 283 2 96

0.52 0.46
0.50 0.46 0.44 0.46
0.48

137 126 341 357

28 28 74 74

13 66
12 57 13 66 14 28

2.22
2.12 2.12 2.12 2.16

25
25 20 20 20 25

25
20 20 25 25 25

20
25 25 20 25 25

2 96
2 96 291

0.44 0.46
0.46

2.16 2.16 2.12

341 262 286 273 341

74 58 58 58 74

13 66 13.09
i4.28

291
291 2 96

13.66 13.66

hi, h2, andh3 are the thicknesses of the strong uppercrust,weaklowercrust,andstrong uppermantlelayers.kd' and kd are the dominant
wave numberand wavelength.

ble structuraland geophysical implicationsof the style of defor-

tively. The differenceis greaterfor the long wavelength insta-

longer wavelengths of deformation penetrate to mation i'esulting fromthe superposition of the longandshort bility because depthandtherefore sense the different mantle strength wavelengthinstabilities.A 15 km thickness for the strongupper greater
crust is reasonable both from the flow law of quartz [Brace and Kohlstedt, 1980] and the depth to which earthquakes are observed in the Basin and Range [Eaton, 1982; Smith, 1978].
stratification in the models. In both models, the scale of the

short wavelength disturbance is overestimated and that of the

long wavelength disturbanceis underestimated.Better agreethicker strong mantle layer. Strong crustal layer thicknesses of about 10 and 11 km and mantle layer thicknesses of about 18 and 8 km would be requiredto match the observations for the J and C models, respectively. However, given the major uncer-

For a total crustal thickness of 30 km [Eaton, 1982;Smith, ment would beobtained fora thihner brittle surface lfiyer anda
1978; Thompsonand Burke, 1974] the thicknessof the weak

lowercrestis alsoabout15 km. The thickness of the regionof highstrength in theupper mantle for boththeJ andC models is
based on the flow law of olivine [Brace and Kohlstedt, 1980].

In the former case, a 15 km thick layer is chosen to represent a region of high strength in the mantle. In the latter case, a mantle layer thickness of 5 km and an e-folding depthof 6 km are chosen on the basisof Figure 1.

tainties in lithosphere rheology discussed earlier, weemphasize

the existenceof two wavelengths of instability rather than the parameter valueswhich providea numericalbest fit. Since contrastsin strengthaffect mainly the amplitudeand not the wavelength of instability, uncertaintiesin our choicesof S values

Two Scalesof Deformation

does not strongly influence these estimates oflayer thickhesses.

Amplitudes are moredifficultto estimate on the basisof the

The shape of the growthrate spectrum depends not on the

actual strengthof the layers but on their strengthratios R, R2, and R3. For the variations in strengthassumedin Table 3 and Figure4 for the J model(R - 100, R2= 0.5, R3- 50), layerthicknesses of about 15 km correspond to wavelengths of deformationwhich reasonablymatch the observedspacings of ranges and tilt domains. The dominant wave numbersof the peaksin the growth rate spectrum in Figure 4 correspond to distances of 42 and 40 km between ranges and 170 and 157 km

present linearized model because theydepend directly on initial

amplitudes. It is, however,possible to establish a lower limit of the dominant growthrate factorrequii'ed to produce the approximate structural relief of the rangesand the largesttilt domains.

For a lowerlimit of totalBasin andRange exthsion of 10%

and initial verticalsurface perturbations in the rangei0-100 m, dominant growthrate factorsof qd--24 and 47 will resultin vertical surface deformation of the order of 1 km. As summarized

strength contrasts required betweenantiformal boundaries for the J and C models,respec- in Table 2, the minimumlithospheric

ZUBER ET AL.: EXTENSION OF CONTINENTAL LITHOSPHERE

4833

for qa in this range are consistentwith the strength profile shown in Figure 1. We have not considered the effect of fluid pressure, which would act to reduce the effective frictional strength of the upper crust;however, for a hydrostatic value the decreasein average crustal strengthwould not be great enough to suppress the growthof eitherwavelength of instability.

Crustand Mantle Deformation

Figure 5c is a plot of the superposition of the crustal displacement fields and showsthe combineddeformationfrom the two wavelengths of instability. A contrastin the style of deformation between the plastic surface layer and the ductile lower crust is clearly visible. Regions of maximum extension are of particularinterestsince faulting shouldlocalize in these areas.
At the surface, localized extension occurs in the short wave-

length basins. The near-surface region at the center of the diagram exhibitsenhanced thinningbecause the extensionfor both the long and short wavelengthcomponents is a maximum there. This model thus predicts localized highly extended regions spacedof the order of the long wavelengthinstability. In the lower crust, areasof greatestextensionoccur primarily beneath rangesand are horizontally removed from regions of near-surface extension.

Fig. 7. Asymmetry of resolved shear stresson conjugate(dip=60 ) faults (heavy lines) associated with surfacetopographywhich dips down to the right. The resolvedshear stressis greateron the fault plane which dips in the direction of the surface topography. See text tk)r implications.

Another interestingaspectof Figure 5c is the style of deformation in the vicinity of the brittle-ductiletransition.The superposed long and short wavelength deformation results in enhancedhorizontal shearingnear the base of the strong upper crust. Localized shearingat midcrustaldepthscould be a mechanism for the formation of low-angle normal faults. The location of horizontal shearingwithin the crust in Figure 5c is consistent with suggestions by Andersonet at. [1983] and Wernicke [1981] that these features root far from regions of nearsurface, thin-skinned deformation. In addition, seismic reflection

preexistingstructuralweaknesses, such as reactivatedMesozoic thrusts.Also, the linearizedmodel developedhere can treat only the initial growth of structures and cannot accountin detail for small scale features formed in areas of large horizontal extension. Finally, becausethe model is two-dimensional, it cannot explain the presence of the transversezones. Stewart [1980] suggests that thesezones mark the northwardmigration of Basin and Range deformation with time and are likely related to tectonic stressesassociatedwith changes in the plate boundary geometry of the western margin of North America during the
Cenozoic.

profiles in the Basin and Range [Allmendingeret at., 1983, Deformation 1985] and in the Scottish Caledonides [Smythe et at., 1982] Gravity studies [Eaton et al., 1978] show that most of the show low-anglereflectors which penetrate to midcrustaldepths. Andersonet al. [1983] have suggested that the zone of decou- regional topographyin the Basin and Range is in approximate pling for detachments in regionssuch as the Sevier Desert and isostaticequilibrium. The regional Bouguergravity signature,as Raft River Valley may be the brittle-ductiletransition. In this depicted in maps of the filtered Bouguer gravity field of the model the brittle-ductile transition is not an actual surface of United States[Hildenbrand et al., 1982], exhibitsa N-S striking mechanicaldecoupling;instead,shearingis vertically distributed 200 km wavelengthundulation which correlateswith the long wavelengthregional topographyof the province [Eaton et al., in the upperpart of the ductilelower crust. that these long wavelength Shear stresses associated with the long wavelengthdeforma- 1978]. Eaton et al. have suggested tion provide an explanationfor the tilt domains. In each tilt gravity anomaliesare at least partially due to temperaturevariadomain, high-anglefault blocks associated with the short wave- tions in the mantle. Froidevaux [1985] estimates that about 7 lengthdeformationshow a consistent directionof tilting or rota- km of relief on the crust-mantleboundary would be required to tion. Assumethat conjugate,high-anglenormal faults, dipping explain the amplitude of the observed anomalies by crustal at 550-60, form initially in response to the uniform horizontal thickness variations alone. Since seismic reflection studies in extension.Shear stresses due to the growing long wavelength the Basin and Range [Allmendingeret al., 1983, 1985] show a deformation will cause differences in resolved shear stress on relatively flat Moho, he concludesthat density variations in the these oppositely dipping faults. As illustrated in Figure 7, a mantle must be responsiblefor the observedlong wavelength counterclockwise shear couple on horizontal planes at depth in gravity anomaly and estimatesthat about a 50 km vertical disthe crust occurs in areas where the surface slopes down to the placementof mantle isothermswould be requiredto accountfor right. Shear stressesof this sense increase the resolved shear the observed anomaly. The short wavelength deformation is consistentwith large stresson high-anglefault planesdippingto the right and reduce it on fault planesdipping to the left. The converseis true if the vertical relief at the surfacecomparedto that at the crust-mantle surfacedips to the left. Therefore, motion on faults dipping in boundary. However, if the Moho also has relatively small relief the downslopedirection of the surface is preferred. Thus, in at long wavelengths,ductile extensionalone cannot accountfor Stewart's [1980] terminology,synformaland antiformalbounda- the amplitude of the observed gravity anomaly. As shown in ries will coincide with crestsand troughs, respectively,of the Figure 6, the vertical velocity for the long wavelengthdeformation at the Moho is approximatelyequal to the maximum value long wavelengthsurface deformation. The presentmodel cannotbe appliedto explain all aspects of in the underlying mantle. Thus the vertical relief on initially finite deformation, for example, features which form along horizontal surfaceswould be nearly a maximum at the Moho. if

Relation of Gravity Anomalies to Extensional

4834

ZUBER ET AL.; EXTENSION OF CONTINENTAL LITHOSPHERE

44 I1 1 I1 I1) 10'6
--Syn

sphere, the dominantwavelengthsof necking are most strongly controlledby the layer thicknesses. The relative strengths of the layers controlthe amplitudes of the instabilities but only weakly affect the dominantwavelengths. In the lower crust and mantle where dislocationcreep dominates deformation,varying the stress-exponent in the range of 1-3 has a negligibleeffect on instabilitygrowth rate. In contrast to the case of a single layer in a weaker viscousmaterial both with n= 1 which always extends stably, a strong mantle layer with n3= 1 exhibits a weak long wavelengthinstabilitybecause mantle deformationis driven by unstableextensionof the upper
crust.

antiformalboundary

synformal boundary

Fig. 8. Tilt domainsand Bouguergravity in the Basin and Range

Province. Dashed and solid lines correspondto synformal and antitkrmal boundaries traced from St(,n'trt's [1980[ Figure 1. Dotted and hatched regions correspondto gravity highs and lows, respectively, taken from the map by Hihh'nbraml ctal. [1982] for all wavelengths less than 250 km. Anomalies range from approximately +35 to -35 mGals. Note that the longestand most continuous synlbrmal and anti- surface. This is consistent with the results of recent COCORP formal boundariestire associaledwith with gravity lows and highs, reflection profiling which show a relatively flat Moho in the respectively.

The two wavelengths of necking instability that result from strengthstratification of the lithospheremay explain the formation of ranges and tilt domains in the Basin and Range Province. For a discontinuous strength(J) model, thicknesses of the strongupper crust, weak lower crust, and strongupper mantle regionsof the continentallithosphere of about 15 km, which are in agreementwith experimental flow laws and seismic results, producestructures with wavelengths consistent with the spacings of ranges and tilt domains. Regions of localized extension at depth which arise due to necking of the strongupper crust are laterally displacedfrom regions of near-surfaceextension. The resultinglocalizationof horizontal shearingin the upper part of the weak lower crust predicted by this model may represent incipientlow-angleextensional detachments. For short wavelength deformation, the model predicts smaller vertical relief at the crust-mantleboundary than at the

the long wavelength gravity anomalyis due to densityvariations in the mantle, this suggests, in constrast to the view of Froidevaux [1985], that convective flow generatedby density differencesin the mantle may be requiredto accountfor the magnitude of the anomaly. However, the long wavelength necking instabilitymay determine the horizontalscaleat which this convective motion subsequently occurs.

The large-scaletilt pattern, represented by synformaland antiformal boundaries,and the pattern of long wavelengthgravity anomalies are shownin Figure 8. Althoughthe surfacestructures are more complicatedthan the gravity field, the longest, most continuous synformaland antiformalboundaries occur in

areas of negativeand positive gravity anomaly respectively. Seismic reflection studieshelp to define the shallow structure createdby faulting or brittle deformation but cannotresolve deeper ductile deformation which produces no well-defined
reflectors. There is thus no direct evidence that tilt domains at

Basin and Range. However, for long wavelength deformation, the model predicts larger vertical relief at the crust-mantle boundary than at the surface, and the vertical displacement which occursat the Moho is nearly equal to the maximum value that occurs in the mantle. If the long wavelength gravity anomaly in the Basin and Range cannot be attributedto variations in crustal thickness,then horizontal temperaturevariationsdue to uplift of isothermsin the necking lithosphere may provide the requireddensity variations. However, our models predict that if relief on the Moho is small at long wavelength, unstableextension of the mantle alone cannot account for the amplitude of deformationrequired. This suggests that convective motions in the mantle, possiblywith a wavelengthcontrolledby the necking instability, may be responsiblefor the long wavelength gravityanomaly.

APPENDIX l: SOLUTION OF THE PERTURBING MULTILAYER FLOW PROBLEM

the surfacecorrespond to deeperductiledeformation. However, the spatialrelationship of long wavelength gravityanomalies to We wish to determine the perturbing flow in an extending surfacestructures suggests that thesestructures may be related mediumconsisting of three viscouslayersover a viscous subto ductiledeformationat depth. strate, each with a uniform strength.The equationfor the flow in a single viscous layerwith uniformeffective viscosity is
CONCLUSIONS

D4W - 2k2(2/n1)D2W + k4W-0

(AI)

An extendingcontinentallithosphere with a density and strength stratification that is consistent with seismicity observa- where D=d/dz, k (=2r[/k) is the wave number, and W is the by tionsand experimental rock deformation studies is unstable with streamfunctionsatisfied respectto neckingand will deform with two wavelengths of cos kx instability.Shortand long wavelengths of deformation arisedue (A2) to the presence of the strongupper crust and upper mantle fi =-k -Dwsinkx regions of the lithosphere. In simplelayered models of the litho-

ZUBER ET AL.'

EXTENSION OF CONTINENTAL LITHOSPHERE

4835

where a and

are the horizontal and vertical components,

respectively, of the perturbing velocity field. Equation(A1) has thegeneral solution[FletcherandHailer, 1983]

A I - Al(0)exp[(q-1);xxt],where q is the growth rate factor. By substituting the form of the perturbedinterface and normalizing with respect to the meanstrainrate ;xx, we rewrite (A7) as
(-1-q) A, + '&(0,dl)--0
(A8)

W(k,z) = (Acos13kz + A2sin 13kz)exp(otkz) + (Aacos 13kz + A4sin 13kz)exp(-otkz) (A3)

We wish to solve for the coefficients

and the values of the

where ct = (1/n)/2 and 13 = (1 - 1/n)/2. This solution is valid

for all of the layersand the substrate, but eachregionhas a different set of constants A-A4 (upper crust), B-B4 (lower crust), C-C4 (upper mantle), and D-D4 (mantle substrate).In the
viscous substrate the stresses and velocities must be bounded as

growth rate parameter,which requiresthe simultaneous solution of the matching conditionsand the growth rate equations(A8). The coupledproblem is solved by the method of static condensation[Bathe, 1982]. We beginwith the 18x18 matrix system Kq) -- qMq) (A9)

z---->,thereforeD3 and D4 vanish. The perturbingvelocity and stress components within the uppercrustare

/q = [(Acos[3kz+ A2sin [3kz)exp(ctkz)

+ (A3cos [3kz+ A4sin [3kz)exp(-cttkz)] coskx

fi = -ct{[(A + 6A2)cos[3kz + (A2-6A) sin[31kz]exp(ctkz) - [(A3- 6A4)cos[3kz

where K is the matrix of matching conditions and vertical velocities, q is the matrix containing the eigenvalues(growth rate factors), M is the matrix containingthe coefficientsof the eigenvalues,and q) is a vector of the coefficientsand amplitudes.This system can be rewritten

aa Ka] I(])l = q rma i1 I(]) 1

ca Kc d L0
wherema is a 4x4 identitymatrix. From (A10)
(A4)

(A10)

+ (A4 + 6A3)sin[3kz]exp(-ctkz)} sinkx

Ozz = :ctk[(Acos [3kz+ A2sin [3kz)exp(ctkz) -(A3cos [3kz+ A4sin [3kz)exp(-ctkz)] coskx
Oxz = -xk{[(A + 61A2) cos[3tkz + (A2- 6A) sin[3kz]exp(ctkz) + [(A3- 6A4)cos[3kz

Kcaqba + Kccq)c = 0
and

(A11)

qbc = -K-Kaqb

(A12)

Substitutionof (A12) into (AI0) transformsthe system into the standard eigenproblem

+ (A4 + 6A3)sin[3kz]exp(-ctkz)} sinkx

(Ka-qMa)q)a - 0
where

(AI3)

where 6= (n- 1)/2.

The linearization to firstorderin kA requires that
Oxz(X,dl)= 2'rkAI sin kx
(A5)

Ka= Kaa-KacKcc-Kca

(A14)

We solve (AI3) by a QR orthogonalization algorithm[Dahlquist where A I is the maximum amplitudeat the ith disturbedinter- and BjOrk, 1974]. To determinethe coefficients and amplitudes, face at depth d,. The difference in density acrossan interface we eliminate A from (AI0) and substitutethe growth rate factor obtained above. The resulting linear system can then be requires solvedby standard methods. (A6) (Jzz(X,dl) = (l-I - Q,)gAicoskx The magnitude of the growthrate factorobtained from (A13) is a strongfunctionof the stress exponentin the surfacelayer.
The normal stresses are continuous, and the shear stress van-

ishes at the free surface, while at the other interfaces both the stresses and velocitiesare continuous. This provides 14 match-

ing conditions from whichthecoefficients canbe obtained. The rate of amplitudegrowth at each of the interfaces introducesa system of four differentialequations

In this study, we adopted a large n to approximate a perfectly plasticmaterial,and it is usefulto examinethe consequences of this choice of rheology. Figure A l, which plots qd for both dominantwavelengths as a functionof n, showsthat increasing n in the stronglayer enhances the instabilityof both the long and short wavelength features. Limiting behavior is not

/1-"'-g2xxAi + /q,(0,d)

approached until n103 forthelong wavelength instability and wavelength instability, which agrees with (A7) nl104 for theshort
the results of Fletcher and Hallet [ 1983].

where , =dA/dt. The number ofvalues ofthe growth rate factor

obtainedat a given wave numberequalsthe numberof equations in (A7), though in generalonly one value of q will be greater thanone and thuscontribute to the growthof instability. The shapeof the perturbed interfacewill changewith time as

A decreasein n shifts the maxima to longer wavelengths and decreases the magnitude of the growth rate spectrum at all wave numbers.The amplitudeof the shorterwavelength peak is decreased significantlyrelative to the longer wavelengthpeak. As noted in the resultssection,a strongmantle layer stabilizes

4836

ZUBER ET AL.' EXTENSION OF CONTINENTAL LITHOSPHERE

150,

LIMITING
125

BEHAVIOR

short .'

IOO

75

long X

25

crustallayer, andS =6, which corresponds to a layer with finite strength, andwith 0t= 0.1 andR = 10. In the C model, the short wavelength(k' large) components in the perturbation are greatly dampedrelative to the J model. For S--0, the range of instability is 1.4 - k' - 2.9, with q reaching a maximum of about85 at ka' - 1.9 or X/h= 3.3. Additionalregionsof instabilityare separated by regionsof damping, and the magnitudesof subsidiarypeaks decreasesharply. The effect of finite strengthin the layer is to decreaseinstabilityor increasedampingat all wave numbers.For S =6, the region of instabilityis reducedto a small interval 1.5 --<k' - 1.9 about
the maximum at ka'= 1.6 or X/h=3.9. The maximum in the growthratefunctionis qa= 20. In the J model, for the strong layer case, intervals of insta-

o' .......

.......

Fig. A1. Magnitudeof the rate growth rate factor q as a functionof the stressexponentin the surface layer n for the conditionsS = 1.2, S = 0.05, S_, =S4 =0, R = 100, R_, = 0.5, andRs= 50. Limiting behavioris
maxima, respectively.

bility are centered at ka '= (n + 1) yr/2, n=0, 1, 2,..., and for all not reached untilnr-->104 and l0s for the short andlongwavelengthpeaksqa= 20. The value R-- 10 usedwas not expected to yield

the closestcorrespondence in qa and ka' with the C model, but was simply chosenarbitrarily in order to comparethe overall prothe systemwith respect to neckingover a limited rangeof wave behaviorof the two models. The characterof the structures numbers, with the maximum stabilizationoccurringat the rela- ducedin eachinstabilityintervalof the J model is different:ka' tive minimum in q. This is suggested because,as illustratedfor in the range 0 - ka' - t results in a structurewhere vertical at the surfaceand the layer-substrate interfaceare the J model in Figure 4, the growth rate spectrumfor an displacements extending,strongsurfacelayer over a uniform viscoussubstrate oppositein sign, while ka' in the range t _ ka' - 2t results whereverticaldisplacements at the surfaceand the envelopesand is everywheregreaterthan that for an extending in a structure medium with two strong layers. If the strongupper mantle acts layer-substrate interface have the samesign. The effect of finite to stabilizethe flow, then both wavelengths of deformation must layer strength in the J model is to reducein the magnitude of arise in response to neckingof the surfacelayer. An exception the instability.Although,as in the C model, the regionof instato this occursif the upper mantle, insteadof the uppercrust, is bility in each interval is shiftedtoward lower k' values(longer the effectdiminishes in successive intervals. characterized by large n. A layered medium in which wavelengths), In the C model, the presence of a single, prominent nl = n2= n4--3 and n3---> ochas a single long wavelengthinstability which resultsin pinch and swell deformation of the mantle. maximumover the rangeof wave numbersprovidesan obvious In the J model, if both the surfaceand upper mantle layer have choice for the wavelengthof maximum instability. In the J a large n, the growth rate spectrum exhibitsa singleshortwave- model, we have assumed that maxima which fall in the first to featuresof length maximumand an inflectionpoint at longerwavelengths. interval of instability, 0 - ka' - t, correspond We have not been able to identify conditionswhich result in interest,and this requiressome rationalization.From a physical two discrete maxima. However, in the C model two wavestandpoint,the maxima at large ka' correspond to very short lengthsof instabilitywill occur if both n and n3 are large, but wavelengthfeatures, which are not observedin natural strucit is not clear whether deformation in this case is driven by ture. In the multilayeredproblem, the vertical stratification in simultaneous neckingof the stronglayersor simply by necking strengthresults in two maxima in the range 0 - ka' - t. of the uppercrust. Analysisof the variationsin physicalproperties of the lithosphere (strength, density, stress exponent, layer thickness)
APPENDIX 2: COMPARISON OF MODELS WITH CONTINUOUS

AND DISCONTINUOUS STRENGTH STRATIfiCATION

C-Model

J-Model

In this appendix we comparemodelsin whichthe strength of the substratebeneath a single strong layer (1) decreases exponentially with depth (C model) or (2) decreases discontinuously to a lower uniform value at the bottom of the layer (d model). The variation of strengthwith depth for each model is illustrated in Figure A2. Plots of the growth rate factor as a functionof wave number are shown for the continuous strength (C) model and the strength jump (d) model in Figure A3. The C model is charac-

terizedby two parameters, ct, the ratio of the depth in which the effective viscosityin the substrate falls off by 1/e to the layer thickness h, andS--(Player- @o)gh/;, the ratio of the lithostaticstressat the base of the layer to the layer strength.The J model is characterized by S and the ratio of layer strength to substratestrength, R. Results for the J model are obtained from an analytical solutionwhile those for the C model are. determined numerically.The two cases examinedin eachmodel are for S =0, which corresponds to the limit of a very strong

Fig. A2. Models of a lithospherewith a continuousdecreasein strength (C model) and a jump in strength (J model) across the brittleductile transition.

ZUBER ET AL.' EXTENSION OF CONTINENTAL LITHOSPHERE

4837

all behavior of the models is similar within the first (small wave
80 C-Model

40

-40 -80

'I
J-Model

number)intervalof instability. The falloff in the magnitudes of higher-order peaksin the C model may thus justify the exclusion of these peaks in the J model. We have used a strength jump to approximatea continuous decreasein strengthwith depth in the lithosphereand are therefore interestedin that part of the behavior of the J model which best approximatesthe behavior of the C model; this occursin the first instabilityinterval.
Acknowledgments. This research was supported by NASA grant NSG-7605. Comments by Leigh Royden and George Thompson improved the paper. We are grateful to Bob Simpson for providing us with mapsof the Bouguergravityfield in the United States.
REFERENCES

IO
q

6
k

e
Allmendinger, R.W., H. Farmar, E. Hauser, J. Sharp, D. Von Tish, J.
Oliver, and S. Kaufman, Phanerozoic tectonics of the Basin and

Fig. A3. Magnitude of thegrowth ratefactor q asa function of wave

number k' for the C and J models with (t=().l and R = 10. For both

modelsS=0 (stronglayer) and =6 (finite strength layer), S2=0,

ill ...... = 104, and n,,,/ ........ = 1.

describedin the results sectionclearly associate the lower and higherwave numberpeaksin this range with the presence of a strong upper crust and upper mantle. Only variationsin the physicalproperties of the surfacelayer affect the characterof the maxima for kd'>J1;. As nlmer--- oc, the viscous resistance to deformationdecreases dramatically, and the disturbance producedat an interfacedecaysslowly away from the interfaceand is partially reflectedat the other interfaces.In this manner, deformationis controlledby a resonance effect [Smith, 1979] in which the secondaryflow at an interface drives deformationat the other interfaces.It is this effect which is responsible for the presence of the higher-order peaksin the J model. The magnitudesof successive higher-ordermaxima decrease exponentially when n in the surfacelayer has a finite value. Hence, the large wave number peaks in the J model are a consequence of the assumption of perfectly plastic behavior in theuppercrust. It is not possibleto develop a comparativeunderstanding of
the nature of the instabilities of the two models because a

Range-Colorado Plateau transitionfrom COCORP data and geologic data, in Reflection Seismologyand the Continental Crust: A Global Perspective, GeodynamicsSet., edited by M. Barazangi and L. Brown, AGU, Washington,D.D., in press,1985. Allmendinger, R.W., J.W. Sharp, D. Von Tish, L. Serpa, L. Brown,
S. Kaufman, J. Oliver, and R.B. Smith, Cenozoic and Mesozoic

structure of the eastern Basin and Range Province, Utah, from COCORP seismic-reflection data, Geology, 11,532-536, 1983. Anderson, R.E., M.L. Zoback, and G.A. Thompson, Implicationsof
selected subsurface data on the structural form and evolution of some

basinsin the northern Basin and Range Province, Nevada and Utah,
Geol. Soc. Am. Bull., 94, 1055-1072, 1983.

Bathe, K.-J., Finite Element Proceduresin EngineeringAnalysis, pp. 573-586, PrenticeHall, EnglewoodCliffs, N.J., 1982. Bird, P., Inititiation of intracontinental subductionin the Himalaya, d. Geophys. Res., 83, 4975-4987, 1978. Brace, W.F., and D.L. Kohlstedt, Limits on lithosphericstressimposed by laboratory experiments, J. Geophys. Res., 85, 6248-6252, 1980. Byerlee, J.D., Brittle-ductiletransitionin rocks, J. Geophys.Res., 73,
4741-4750, 1968.
1978.

Chappie,W.M., Mechanics of thin-skinned fold and thrustbelts, Geol.

Soc. Am, Bull., 89, 1189-1198,

simple analyticalsolutioncannotbe obtainedfor the C model

[cf. Fletcher and Hailer, 1983]. Therefore, in order to examine

the similarity in behaviorof the C and J models, one must rely on generalnumericalresultsfor a range of parameters.In both models, kd/h increasesas S! increasesand falls in the narrow rangerc -< kd/h 4 for the C modeland 4 -< kd/h 8 for the J model. This ratio has a weak dependence on c in the C model and on Ri in the J model. For both models, the magnitudeof the growth rate function decreases as S increases.In the C
is unstable for all S i. The latter is due to the fact that the

Chen, W.P., and P. Molnar, Focal depthsof intracontinental and intraplate earthquakes and their implications for the thermal and mechanical properties of the lithosphere, J. Geophys. Res., 88, 4183-4214, 1983. Dahlquist,G., and A. Bj6rk, NumericalMethods, pp. 216-217, Prentice-Hall, Englewood Cliffs, N.J., 1974. Eaton, G.P., The Basin and Range province: Origin and tectonic significance, Annu.Rev. Earth Planet. Sci., 10, 409-440, 1982. Eaton, G.P., R.R. Wahl, H.J. Prostka, D.R. Mabey, and M.D. Kleinkopf, Regional gravity and tectonic patterns:Their relation to late Cenozoicepeirogenyand lateral spreading in the WesternCordillera,
Geol. Soc. Am. Mem. 152, 51-91, 1978.

England, P., Constraintson extension of continental lithosphere,J. Geophys. Res., 88, 1145-1152, 1983. Fletcher, R.C., and B. Hallet, Unstableextensionof the lithosphere:A mechanicalmodel for Basin and Range structure,J. Geophys. Res.,
88, 7457-7466, 1983.

model, instability occurs where SI (2Or) -I, while theJ model Froidevaux, C.,

strengthcontrasts at both the upper and lower interfacesdrive the instability, and necking will take place regardlessof the magnitude of S. The magnitude of the growth rate functionin Astr. Soc., 59, 463-478, 1979. the C model depends on both S and or, while for the J model, Hildenbrand, T.G., R.W. Simpson, R.H. Godson, and M.F. Kane, q has a weak Si dependenceand a strong R l dependence. Digital colored residualand regional Bouguergravity maps of the conterminousUnited States with cut-off wavelengthsof 250 km and Because there is no physicalconnection betweenR i and or, it is 1000 km, U.S. Geol. Surv. Geophys.Invest. Map, GP-953-A, 1982. not possibleto obtain a simple relationship that describes what S.H., Rheology of the lithosphere, Rev. Geophys., 21, valuesof theseparameters give the sameqd or kd. However, Kirby, 1458-1487, 1983. both modelsyield qd valuesof the sameorderof magnitude for Proffett, J.M., Jr., Cenozoicgeologyof the Yeringtondistrict, Nevada,

Basin and Range large-scale tectonics:Constraints from gravityandreflection seismology, J. Geophys. Res., in press,1985. Glazner, A.F., and J.M. Bartley, Evolution of lithosphericstrengthafter thrusting,Geology, 13, 42-45, 1985. Goetze, C., and B. Evans, Stressand temperaturein the bending lithosphere as constrained by experimental rock mechanics, Geophys. J. R.

similar R and ot -I. The above comparison indicates that

thoughthe natureof the instabilities may be different, the over-

and implicationsfor the nature and origin of Basin and Range faulting, Geol. Soc.Am. Bull., 88, 247-266, 1977.

4838

ZUBER ET AL.: EXTENSION OF CONTINENTAL LITHOSPHERE

Smith, Robert B., Seismicity, crustalstructure,and intraplatetectonics Tullis, J., and R.A. Yund, Hydrolytic weakening of experimentally of the interior of the western Cordillera, Mem. Geol. Soc. Am., 152, deformedWesterly granite and Hale albite rock, J. Struct. Geol., 2,
111-144, 1978. 439-451, 1980.

Smith, Ronald B., Formation-of folds,boudinage, andmullions in

non-Newtonian materials, Geol. Soc. Am. Bull., 88, 312-320, 1977.

Wernicke, B., Low-angle normal faults in the Basin and Range Province: Nappe tectonicsin an extendingorogen, Nattire, 291, 645-648,
1981.

Smith, RonaldB., The folding of a stronglynon-Newtonian layer, Am.

J. Sci., 279, 272-287, 1979.

Smythe, D.K., A. Dobinson,R. McQuillan, J.A. Brewer, D.H. Matthews, D.J. Blundell, and B. Kelk, Deep structureof the Scottish Caledonides revealedby the MOIST reflectionprofile, Nature, 299,
338-340, 1982.

Zoback, M.L., R.E. Anderson,and G.A. Thompson,Cainozoicevolution of the state of stressand style of tectonismof the Basin and Range province of the westernUnited States, Philos. Trans. R. Soc.
London Ser. A, 300, 407-434, 1981.

Stewart, J.H., Basin and Range structure: A systemof horstsand grabens producedby deep-seated extension,Geol. Soc. Am. Bull., 82,
1019-1044, 1971.

R. C. Fletcher,Centerfor Tectonophysics, Texas A&M University,

CollegeStation,TX 77843. E. M. Parmentier,Departmentof Geological Sciences,Brown University, Providence,R102912. M. T. Zuber, Geodynamics Branch, Code 621, NASA/Goddard SpaceFlight Center,Greenbelt,MD 20771.

Stewart, J.H., Basin-rangestructurein western North America: A

review, Mem. Geol. Soc. Am., 152, 1-31, 1978.

Stewart, J.H., Regionaltilt patterns of late Cenozoicbasin-range fault

blocks, western United States, Geol. Soc. Am. Bull., 91, 460-464,
980.

Thompson, G.A., and B.B Burke, Regional geophysicsof the Basin and RangeProvince,Ann. Rev. Earth Planet. Sci., 2,213-238, 1974.

Tullis, J., and R.A. Yund, Experimental deformation of dry Westerly granite,J. Geophys. Res., 82, 5705-5718, 1977.

(ReceivedSeptember 20, 1984; revised July 31, 1985; accepted August 1, 1985.)