TECTONICS, VOL. 18, NO. 6, PAGES 1064-1083 DECEMBER

1999

Lithospherefolding: Primary responseto compression? (from central Asia to Paris basin) S. Cloctingh Facultyof Earth Sciences,Vrije Universiteit,Amsterdam

E. Burov•

Directionde la Recherche, Bureaude Recherches G6ologiques et MiniSres, Orl6ans, France

A. Poliakov UMR 5573, CNRS, UniversityMontpellierII, Montpellier,France

Abstract. We examinethe role of lithospherefoldingin the (escaping) of the lithosphere, and (3) underthrusting, large-scaleevolutionof the continentallithosphere. Analysis subduction,when shorteningis accommodated by localized, of the recordof recentverticalmotionsandthe geometryof stable, downward escape of the lithosphere, i.e., by basin deflectionfor a number of sites in Europe and underthrusting of one block or plate alonglargethrustfaults. worldwide suggeststhat lithosphericfolding is a primary The dynamicsof lithospherefoldingformsan importantbut responseof the lithosphere to recentlyinducedcompressional hitherto largely underestimatedcomponent in models stressfields. Despite the widespreadopinion, folding can developed to study the evolution of the lithospherein persist during long periods of time independentlyof the compressional regimes:until recentlylarge-scalefolding was presenceof many inhomogeneitiessuchas crustalfaults and acceptedonly for a few areasof tectoniccompression suchas inheritedweaknesszones.The characteristic wavelengthsof the northeastern Indian Oceanor centralAustralia[Turcotte foldingaredetermined by thepresenceof younglithosphere in and Schubert,1982; Fleitout and Froidevaux, 1982; Lainbeck, largepartsof Europeand centralAsia andby the geometries 1983]. However, a numberof recent data setssuggestthat of the sedimentbodiesactingas a load on the lithosphere in lithosphericfoldingmay be a muchmorewidespread modeof basins. The proximity of these sites to the areas of active deformationthan previouslythought,thoughit may take less tectonic compressionsuggeststhat the tectonicallyinduced obviousformsthanin thewell-known"distinct"cases[Ziegler horizontalstresses areresponsible for the large-scale warping et al., 1995; Cloetinghand Burov, 1996]. For this reason,a of the lithosphere. Wavelengths andpersistence of foldingare better understanding of processes controlling surface controlledby many factorssuchas rheology,faulting,time expressionsof folding is required to addresssome of the after the end of the major tectonic compression,nonlinear recently raised questions (e.g., commonly inferred effects,and initial geometryof the foldedarea.In particular, relationshipsbetweenthe foldingwavelengthand the age of the persistenceof periodical undulationsin central Australia the continentallithospheredo not explain some data from (700 Myr sinceonsetof folding)or in theParisbasin(60 Myr) recentlyindicatedareasof compressional instabilities[Ziegler longafterthe endof the initial intensivetectoniccompression et al., 1995]). requiresa very strongrheologycompatiblewith the effective It shouldbe notedthatat a small-scale, foldingor buckling elastic thickness values of about 100 km in the first case and of layered mechanicalstructuresis a typical responseto 50-60 km in the second case. horizontal shortening. For example, folding is largely observedin sedimentarylayers,in outcropsexposingductile shear bands, in clays, and in other superficial structures 1. Introduction [Johnson,1980; Smith,1975]. Sincethe mechanismof folding is largely scale-independent, it is quite reasonableto suggest The lithospherecan undergotectonicshorteningin three that folding may be reproducedon a lithosphericscale[e.g., principal ways: (1) volumetricshorteningby distributedor Biot, 1961; Stephensonand Lambeck,1985; Stephensonand localizedthickeningof the lithosphere dueto compression, (2) Cloetingh,1991]. Detailedsmall-scalestudiesof foldingin the folding, when shorteningis accommodatedby unstable, sedimentarycover,clays,soils,and othersuperficialmaterials subperiodical,vertical upward and downward bending [Smith, 1975, 1977, 1979] often consider sophisticated nonlinearrheologiesin many aspectssimilar to those of the 1Nowat Department of Tectonics, University of Pierre andMarie deep lithosphericmaterials.Thus it would be logical to Curie, Paris assumethat the transitionfrom a small scaleto a large scale will not changethe characteristic responseof the modeled Copyright1999by theAmericanGeophysical Union. system.The only really importantdifference,which might be expected,is associatedwith the effect of gravity, which is Papernumber 1999TC900040. 0278-7407/99/1999TC900040512.00 negligible for small systemsand significantfor large ones. 1064

CLOETINGH

ET AL.: LITHOSPHERIC

FOLDING

1065

Thepotential gravityenergy scales asogh 2 perunitlength, in the northeasternIndian Ocean [Geller et al., 1983; Stein et whereh(x) is surfaceelevation,O is the density,andg is the accelerationdue to gravity, which requiresa nonlinear increasein tectonic force needed to maintain continuously growingtopographic uplift. Thusthe influenceof the gravity forceon foldingis negligiblefor the systems with h < 10-20m but becomesimportantstartingfrom h > 50-100 m. For this reason,small-scalefoldsmay be of very large amplitude,a few timeslargerthanthe layerthickness andcomparable to or exceedingthe wavelengthof folding. At lithosphericscale, upward elevationsdo not exceed 5 kin, which is smallerthan

the thicknessof the folded layersand lessthan 10% of the characteristic lithospheric foldingwavelengths (30-600 kin). Most lithospheric-scale studiesusedmore simplifiedmodels thanthe small-scale studies;for example,a simplifiedelastic rheologywas typically used as a first approximationof the lithospheric materialproperties. Thisapproximation maywork

al., 1989], continentalfoldingwas subsequently recognizedin the lithosphereof centralAustralia [Stephenson and Lambeck, 1985; Beekman et al., 1997], central Asia [Nikishin et al., 1993; Burov et al., 1993; Burg et al., 1994; Cobbold et al., 1993; Burov and Molnar, 1998], Arctic Canada [Stephenson et al., 1990], Iberia [Walthamet al., 1999; S. Cloetinghet ai., Late Cenozoic lithosphere folds in Iberia?, submitted to Tectonophysics, 1999, hereinafterreferredto as Cloetinghet al., submittedmanuscript,1999] and the Paris basin [Lefort and Agarwal, 1996; J.-P. Brun, personal communication, 1994]. A surprisingaspectof the outcomeof thesestudieswas that the occurrenceof folding was not restrictedto lithosphere with intrinsic

zones of weakness but also occurred in areas

characterizedby the presenceof cold and presumablystrong lithosphere(seeFigure 1 andTable I). More recently, evidence has also been put forward to for the flexural modelsbut predictsunrealistically high supportthe occurrenceof a componentof lithospherefolding stresses for bucklingmodels.Partlyfor this reason,modelsof in someof the major extensionalbasinsof Europe,suchasthe lithospheric foldingwereput asidefor quitea longtime,just North Sea basin [van Weesand Cloetingh, 1996; van Balen et until the late seventies, whena numberof authors[McAdoo al., 1998], the Pannonianbasin[Horvdthand Cloetingh,1996; and Sandwell, 1985; Zuber, 1987; Turcotte, 1979; Fleitout van Balen et al., 1996; Fodor et al., 1999; Bada et al., 1998], and Froidevaux,1982, 1983] showedthat the applicationof and the Gulf of Lion's margin of the westernMediterranean viscousor more realisticyield-stress rheologiescan partly [Kooi et al., 1992; Chamot-Rookeet al., 1999]. These basins resolvethe "high-stress" problemsmet in the elasticfolding are locatedon lithospherewith low rigidity, and the observed models.However,solutionof the "high-stress" problemjust wavelengthsseem at first hand to be at odds with these freed spacefor anotherone, the "low-stress" problem.The rheologicalcontexts(seeFiguresI and2). latteris associated with the currentwide-spread opinionthat However, the discrepancywith the theoreticalpredictions thrustfaults,eitherpreexistingor accompanying lithospheric and the observationsof the compressional deformationin the shortening, may weakenthe compressed layerso muchthat it former extensionalbasinsis not surprising,since in most of will not be ableto supportstresses necessary to initializeand the "problematic"casesthe geometryand other assumptions maintain folding (the latter requiresa large competence of the linear theory are poorly satisfied.The linear theory of contrast betweenthe foldedlayerandtheembedding). folding developedin the sixties to seventiesby Biot [1961], Startingwith earlypaperson the occurrence of lithospheric Ramberg [1961], Smith [1975], Fletcher [1974], and Johnson folding in the intracratoniclithosphereof central Australia [1980] predicts developmentof sinusoidalundulationsof a [Larnbeck,1983] or on the folding of the oceaniclithosphere horizontally shortened competent layer embedded in less

time / age [My] 400

800

'

'

W. Goby (crust)

1200

I

I

1600

I

I

2000

I

LO

....... :_.,q., :,,..,• •.......... ,....... *......... •,upper crustal foldirig •,.-. ••

•250•,% ...... Fer_clhana '• "•W.. •obyimantle)

I-250 •• • g

750

7•

o

4bo

ioo

ooo

time / age [My]

Figure 1. The observed wavelength of foldingasa functionof thethermalage(i.e., mechanical mantlethickness) calculated according to themodelfromBurovet al. [1993].Numbers correspond to theonesusedin Table1. Squares showthecasesof "regular"folding,whereasthe starsmark "irregular"cases.Differenttheoreticalcurvescorrespond to crustal,mantle (suppomng thepresence of thedecoupled rheology), and"welded"folding.

1066

CLOETINGH

ET AL.: LITHOSPHERIC

FOLDING

Table 1. Estimatesfor Wavelengthsof Folding, Effective Elastic Thickness,Thermal Age, and the Onset of Folding Thermal Age,

X.,km

EET,km

Ma

200-250 (1) 500-600(2) 200 (3) 200 (present); > 400-500 (preserved)(4) 300-360(5) 100-200 (6) 200 ? (7) 200-250 (8) 350-400(9) 300 (10) 40 (11) 50 (12) 600 (13)

40 50 30 25 *

60 (14)

to,

Onsetof Folding,Myr

Present Stateof Folding

60 400-600 200 > 700

8 60 60 400-700

activedeformation preserved preserved preserved

Type B N B B

15 >_30 20-35 15 6-9 10-30 20-25 20-25 > 100

175-400 > 100 300 175 < 20 30 < 20 20 > 1200

8-10 60 6 8-10 4-6 6-8 6-8 6-8 1200

activedeformation preserved activesubsidence activedeformation activedeformation activedeformation activedeformation activesubsidence preserved ?

B B/N N B/N N B/N N N B

< 10

65

8-35

activedeformation

B/N

EET, effective elastic thickness.Numbersin bracketsrefer to data sources:1, Indian Ocean [Cochran, 1989; Curray and

Munasinghe,1989], (2) Russianplatform[Nikishinet al., 1997]; 3, Arctic Canada[Stephenson et. al., 1990]; 4, central Australia[Lambeck,1983; Stephenson and Lambeck,1985;Beekmanet al., 1997]; 5, westernGoby [Nikishinet al., 1993; Burov et al., 1993]; 6, Parisbasin[Lefort and Agarwal, 1996]; 7, North Sea basin [van Weesand Cloetingh,1996]; 8, Ferghana andTadjikbasins[Burget al., 1994;BurovandMolnar, 1998];9, Pannonian basin[Horv•thand Cloetingh,1996]; 10, Iberian continent[Stapelet al., 1997; Vegaset al., 1998; Cloetinghet al., submittedmanuscript,1999]; 11, southern TyrrhenianSea [Mauffret et al., 1981]; 12, Gulf of Lion [Kooi et al, 1992]; 13, Transcontinental Arch of North America [Ziegleret al., 1995]; 14, Norwegiansea[DoreandLundin,1996; Vagneset al., 1998]."B" standsfor regularfoldingstyle, "N" standsfor "irregular"foldingstyle,and"B/N" standsfor the casesdisplayingbothtypesof behavior.

*Valueisforrecent reheating at200Ma. competentsurroundings. This deformationis characterized by a singledominantwavelengthbetween3 and 50 km thickness of the competentlayer. Accordingto the linear theory, this dominant wavelength is time-independentand is mainly controlledby the thicknessof the folded layer and much less by otherparameterssuchas competence contrastor shortening rate. The linear analysis holds for many observedcasesof folding but not for all. A number of very recent (generally small-scale)studies[Davis, 1994; Zhang et al., 1996; Mercier

were basedon simplifiedpressure-,strain-,and temperatureindependent rheologies, whereas the numerical studies employing nonlinear brittle-elasto-ductile rheology did not consider developed (large strain) stages of deformation. Consequently,a more thoroughstudy is neededto verify the previouspreliminaryresultsandthe hypothesison foldingand faultinginteractions. Based on the above mentioned new observational data sets

and theoreticaldevelopments,the primary goal of our studyis et al., 1997; Bhalerao and Moon, 1996; Hunt et al., 1996; Lan to demonstratethat lithospheric folding is much more and Hudleston, 1996] have shown that the wavelengthsof widespreadthan is currently assumed.We propose that foldingmay be significantlydifferentfrom thoseinferredfrom folding representsa rather typical initial and intermediate the linear theory,if time-dependence and "non-sinusoidality" stageof tectoniccompression, which may take quitedifferent of deformation,realistic(elasto-plasto-ductile) rheologies,and forms, which complicatesits identification.For this reason, layer geometriesare taken into account.In particular,without the casesof"irregular" folding were not recognizedbefore.To any lateral inhomogeneities, dependingonly on the boundary explorethis concept,we will discussin the followingthe role conditions,time, rheology,and geometry,the wavelengthof of the nonlinearrheology, faulting, and other factorssuchas unstableplate deflectionscan significantlyvary along the sedimentloadingand prefoldingplate geometryin modifying shortenedplate, or even one single "megafold"can be formed the inferredwavelengthsof lithospherefolds. In addition,we [Hunt et al., 1996;Burg and Podladchikov,1999]. Regarding investigate further the role of coupled versus decoupled the frequentlydiscussed influenceof the preexistingfaultson rheologiesin the responseof the lithosphereto large-scale the developmentof folding,recentnumericalexperiments on compressional deformation.We will discriminate betweenthe folding of brittle-elasto-viscous lithosphere[Gerbaultet al., "typical," "regular," or "distinct" folding characterizedby 1999] as well as previousanalogueexperimentson plasto- periodical deformation, which can be explained by the elasticlithosphere[Shemenda,1989, 1992;Burg et al., 1994; commonlineartheory,and"irregular"folding,whichdoesnot Martinod and Davy, 1994] have shownthat (1) lithospheric fit in the conventional theoreticalscheme;for example,it may be aperiodical,polyharmonic,or have a much shorter or folding and faulting can develop simultaneouslyand (2) preexistingcrustal-scale faults do not preventbut participate longerwavelengthcomparedto the "linear"predictions. in the developmentof folding instabilities.The instabilities,in turn, can provokethe formationof new faultsat the inflection 2. Folding as a Mechanism of Shortening: points of the folds. This disagreementwith the usual Origin of Different Forms of Folding expectationscan be explainedby the underestimated role of the gravityandfrictionin the behaviorof the large-scalefaultFolding is commonly associated with periodical and-fold systems.Yet most of the previousanaloguestudies deformationof layeredstructures,and thus the periodicityis

CLOETINGH ET AL.: LITHOSPHERICFOLDING

1067

a) Central Asia (neotectonicmovements) 80OE

70øE

90øE

DZUI

.AZ^r 700 Ma andstrongdiabaselowercrustalrheology; Table2). Notethecrustalandmantlefaultinganda largerwavelength of thedeformation (500 km). Shortening is at a rateof 1.5 cm/yr.

1077

Strain

rate

Accumulated plastic strain -20

-40 -60

-80 100

120

•me = 3.8 m. 0.36

0.0

deviatoric

shear

0.72

dev

stress

CTIi

-20

-40 oI•O -60

-80 100 120

ti.me = 3.8 m. 0.0

1.0e+09

2.0e+09

dev

deviatoric shear. y,•stress --.

o'II

-20

Moho • -60 -80 -100

4.6e+04

1.0e+09

2.0e+09

Plate 1. Numericalexperimentdemonstrating nonlinear,irregularfolding:foldingand faultingexperiment(Ferghana/Tadjik basintype lithosphere,centralAsia, quartzcrustalrheology,olivine mantlerheology(Table 2), thermalage 175 Ma. The faultsappearbeforethe foldingdevelops,but then the two processes, faultingand folding, can co-existin sucha way that folding is accommodated by faulting.Becauseof the weaknessof the lower crust,the uppercrustis completelydecoupled from the mantle and interactswith it only by flow in the lower crust.The participationof the flow changesthe wavelength andamplitudeof folding,which finishesby the developmentof a singledownwarped"megafold."

1078

CLOETINGH

ET AL.' LITHOSPHERIC

[Hunt et al., 1996]. As notedin section3.2, a new databasein-

cludingsuchirregularfoldingcasesis now availablefor large areasof the Europeanforelandand its marginsas a resultof intensive geophysicaland geological studies carried out duringthe last decade,which have significantlyenhancedthe currentunderstanding of the basin(de)formationmechanisms operatingin theseareas.

Plate 1 demonstrates the caseof inverse"megafolding", the result of the prolongationof the numerical experiments considering simultaneous development of foldingandfaulting in a youngFerghana/Tadjik typebasin(Figure2c, [Burovand Molnar, 1998], which, as was mentionedabove,possibly resultsfrom unstabledownwarping,and may be associated with deep mantle faulting. Actually, for Ferghana-Ta4jik settingsour numericalexperiments reproducethe formationof active crustaland mantlefaults with characteristic spacing correspondingto the thicknessesof the brittle crustal and

mantledomains.Surprisingly,the appearance of faultsdoes not significantlyinfluencethe wavelengthof folding: both processes continueto coexistfor a long time, so that faulting serves as a mechanism of folding in the brittle domain [Gerbault et al., 1999]. Also observedin some analogue experiments[Martinod and Davy, 1994], this "continuous" behaviorof faulted lithospherecan be explainedby fault lockingdueto gravityandfriction:aftersomesliding(uplift) onthefault,thepotentialgravityenergyandthusworkagainst

FOLDING

fasterthan the others,resultingin a lossof periodicityand the formationof a mega-fold(Plate 1), which can finish up by initialization of subductionand mountainbuilding. In the case presentedin Plate 1, large horizontal shorteningalso resulted in the decreaseof the observedwavelengthof mantlefolding from the initial value of 250 km to approximately150 km, and the increaseof the crustalwavelengthfrom 40-50 to 100 km. Note that the experimentsshown in Figure 5 present a "developed"casewith respectto that demonstrated by Burov and Molnar [1998]. Consequently, these results can be regarded as a possible scenario of the evolution of the deformationin the recentlyinvertedbasins. As anotherpossibleexampleof irregularfolding, Figure 5 demonstrates a case of demicoupled/demi decoupled lithosphere(similar to Figure 3b but with coldergeothermof 400 Ma). In the intermediatecases(betweenthe coupledand decoupledstate)the mantlelithospherecan be in someplaces coupledor decoupledwith the uppercrust,dependingon the stressand strainrate. In this case,somepartsof the plate may deform

in

a biharmonic

mode

whilst

others

will

exhibit

longer-wavelength monolayerfolding. 5.3. Experiments on Preserved Folding

As was discussedby Bird [1991] and Avouac and Burov [1996], large-scaleundulationsof the lithospherein the absenceof sufficientcompressioncannotbe preservedfor a frictionto be doneby the forcesof horizontalshortening long time (> 10 Myr), exceptfor very strong(especiallylow become too high, and the fault locks and transmitsthe crustal) lithospheric rheology. Otherwise, they will be horizontal stress as a continuum medium. As soon as the flattenedowing to the gravity-drivencrustalflow associated compressioncontinues,one of the folds finally startsto grow with the omnipresent largecrust-mantledensitycontrastat the

5% shortening, no erosion o

•.

ooo 500, -250

0

250

500

•

25% shortening, no erosion

ooo

400•,•••,,.•.• •/ -200

0

200

25% shortening, erosion 4000 i

i

-200

0

200

400 x (km)

Figure 5. Numericalexperimentdemonstrating nonlinear,irregularfolding.In somecases,the instabilitiescanbe quitechaotic. Figure 5 demonstrates differentcasesof irregularfoldingwith wavelengthand amplitudevarying alongthe plate on differentstagesof deformationdue to partialcrust-mantle couplingand strainlocalizations(400 Ma lithospherewith weak quartz-dominated crustalrheology;Table 2). After 5% shortening,(top), after 25% shortening,(middle), after 25% shortening(bottom),strongzero-orderdiffusionalerosion[Avouacand Burov, 1996] tunedto keep meanelevationsat the level of 3000 m. Erosionreducesthe contributionof gravity-dependent terms(middlewavelength)and accelerates local deformations.Strongerosion,insufficientlycompensated by the tectonicdeformation(bottom), wipes out most of the topography. Yet, if the erosionis tunedto theaverageelevationrates,it maydramatically accelerate folding.

CLOETINGH ET AL.' LITHOSPHERIC FOLDING

Moho boundary.Near the Moho, some parts of the folded crust occur at the same depth as the folded densemantle, resultingin remarkablepresstiredifferences(5 MPa per I km of Moho depression plus a contribution from the hydrostaticallydisbalancedpart of the stirfacedepression). These pressuredifferencesin most casesare sufficient to overcomethe yieldingstressof the lowercrustat Moho depth levels [Bird, 1991]. As a result, the crust would flow and flatten the bent layer. The occurrenceof long - timescale preservationof folding after cessationof the compression (Figure 6) in the presumablystrongAustraliancraton,Parisian basin or Russianplatformpointsto a quite strongrheology yielding high EET values (> 60 km) and consequently confirming experimentalestimateslisted in Table 2. Yet the amplitudeof the verticaldeflectionin the Parisbasinis quite small, allowingfor othermechanisms suchas simpleflexure due to the load by the Alpine system.However,the periodic deformationat the same wavelengthis more or less well expressedto the northwestof the area,which would not be the case if the northeasternpart of the basin was simply flexed downby the load of the Alps. Most of the presentlyobservedareasof folds coincidein time with the Alpine collision 60 Myr ago. It is thus reasonableto assume that the most typical characteristic timescaleof the gravity collapseof the large-scalefoldsin the intermediate-age lithospheres is limitedby thistime,thoughin the casesof very weak quartz-dominated lower crustthe folds

1079

may disappear withinthe following8-15 Myr (analogously to the estimatesmade by Bird [1991] for the mountains).The experimentof Figtire6 (old 1000 Ma lithospherewith diabase lowercrust;Table2) demonstrates the preservation of foldinginduced deformation for 10 Myr (potentially > 50 Myr) following the cessationof the tectoniccompression. In this experimentwe stoppedthe horizontal compressionafter the first 14 Myr of shorteningand then studiedthe asymptotic behaviorof the systemin the next 10 Myr. Duringthis period the amplitude of folds decreasedby less than 10%, which yieldsat least50 Myr decaytime as an asymptoticprediction. The presenceof the crustalfaults, of course,may accelerate the gravity collapseof the folds,restiltingin the creationof the inverted basins. For example, the experimentof Figtire 6 predictsreactivationm•d inverted activity of some faults after the cessationof the tectoniccompression. 6. Discussion

Up till now, we have discussedvarious parameters characterizingthe rheologicalstateof the folded lithosphere and the associateddeformationsignature.The initiation of folded basin formation dependson the interplay of forces operating on the continental lithosphere and the spatial distributionof the lithosphericstrength.In platform settingsa large distance away from plate boundaries, far-field lithosphericstressesshould be relatively constant[Zoback,

posco.-mpression stage -40 -80 -120

compression stopped at .::.•..•.:-,..•'•.•...•':,•',:•-..- • .....

•.-::.•:•:•.:•'•.•!•;•" . ..:. ............... ."':.. -.

..:.".•.•': ::':'•"•'• • ..... ß .''..":' '•:•. ':' ....... "• ........ :.

...•....•;:•.

..... :•":"""t•' •.::':' ,•:..•:•:½•...:•.. ..........

-80 -120

::

.....j

,•:• •