Coulomb Theory Applied to Accretionary and Nonaccretionary Wedges: Possible Causes for Tectonic Erosion And/Or Frontal Accretion Serge E

Home , Accretionary wedge

Coulomb theory applied to accretionary and nonaccretionary wedges: Possible causes for tectonic erosion and/or frontal accretion Serge E. Lallemand, Philippe Schnürle, Jacques Malavieille

To cite this version:

Serge E. Lallemand, Philippe Schnürle, Jacques Malavieille. Coulomb theory applied to accretionary and nonaccretionary wedges: Possible causes for tectonic erosion and/or frontal accretion. Journal of Geophysical Research, American Geophysical Union, 1994, 99 (B6), pp.12033-12055. hal-01261600

HAL Id: hal-01261600 https://hal.archives-ouvertes.fr/hal-01261600 Submitted on 25 Jan 2016

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. B6, PAGES 12,033-12,055,JUNE 10, 1994

Coulomb theory applied to accretionary and nonaccretionary wedges: Possible causes for tectonic erosion and/or frontal accretion

Serge E. Lallernand URA CNRS 1760 "G6ophysiqueet Tectonique",Universit6 Montpellier II, Montpellier, France

Philippe Schntirle URA CNRS 1315, Laboratoirede G6ologieStructurale, Universit6 Pierre et Marie Curie, Paris, France

JacquesMalavieille URA CNRS 1760 "G6ophysiqueet Tectonique",Universit• Montpellier II, Montpellier, France

Abstract. Basedon observationsfrom both modemconvergent margins and sandboxmodeling, we examine the possibleconditions favoring frontal accretionand/or frontal and basal tectonic erosion.Mean characteristic parameters (!.t, !.t*• and [) areused to discussthe mechanical stability of 28 transectsacross the frontal part of convergentmargins where the Coulomb theory is applicable.Natural observations reveal that "typical accretionary wedges" are characterizedby low taperswith smoothsurface slope and subductingplate, low convergencerates and thick trench sediment,while "nonaccretionarywedges" display large taperswith irregularsurface slopes and roughsubducting plate, high convergencerates and almostno trenchfill. Sandboxexperiments were performed to illustrate the effects of seamounts/ridgesin the subductionzone on the deformationof an accretionarywedge. These experiments show that a wedge of sandis first trappedand pushedin front of the seamountwhich actsas a movingbulldozer. This is followed by a tunnelling effect of the subductingseamount through the frontal wedge material, which resultsin considerablesand reworking. At an advancedsubduction stage, the d6collementjumps back from a high level in the wedgeto its formerbasal position. We concludethat a high trench sedimentationrate relative to the convergencerate leadsto frontal accretion.In contrast,several conditionsmay favor tectonicerosion of the upperplate. First, oceanicfeatures, such as grabens, seamountsor ridges, may trap upper plate material during their subductionprocess. Second, destabilizationof the upperplate material by internalfluid overpressuringcausing hydrofracturing is probablyanother important mechanism.

Introduction "critical taper" at failure for a given rock density within the wedge. These parametersare the internal and effective-basal Following work by Davis et al. [1983] on the mechanicsof friction angles and pore fluid pressurein the wedge. In this fold and thrust belts, Dahlen [1984] derived a rigorous paper we apply this theory both to accretionary and solution for the "critical taper" that a submarine sediment nonaccretionarywedges using 28 well-constrainedgeometries wedge of noncohesivefrictional material on a basal plane of of convergentmargins. weaknesswill attain, when sufficiently compressedfrom its We use the classificationof convergentmargins as defined back end. The rock is treatedas a frictionalplastic (Coulomb- by yon Huene and Scholl [1991, 1993], dependingon the type) material without cohesion.A cohesiveplastic material occurrenceor absenceof an accretionarywedge (or prism or without frictional shear strength has been previously complex)at the toe of the activemargins. Such a complexis investigated by Chapple [1978]. This material model thus generally well imaged on seismic records because of its neglectselastic strains and strainhardening and softeningand particularfold andthrust structure. The complexcontinuously the limiting condition will be given by the effective-stress- grows by incorporationof new imbricate slices of trench fill dependentinternal friction of Coulomb-Mohr[Mandl, 1988]. and oceanic material or by underplatingalong accreting Since that time, numerous authors have applied the margins (Figure 1). On the other hand, the wedge either "Coulomb theory" to accretionarywedges to estimate some maintainsits initial volume or consumesitself by tectonic characteristicparameters [e.g., Dahlen et al., 1984; Zhao et erosion at nonaccretingmargins (Figure 1). Both sediment al., 1986]. Three parametersare sufficient to calculate the accretionat the toe and crustconsumption further back may occursimultaneously as is documentedalong some transects of Copyright1994 by the AmericanGeophysical Union. the Japanand Peru Trenches [von Huene and Lallemand, 1990]. Several authors have demonstrated that considerable loss of Paper number 94JB00124. upper plate material has occurred along some convergent 0148- 0227/94/94 JB-00124505.00 marginsthrough tectonic erosion processes [e.g., von Huene

12,033 12,034 LALLEMANDET AL.:ACCRETION•Y ANDNONACCRETIONARY WEDGES

NON-ACCRETIONARY MARGINS (21,000 KM) and Lallemand, 1990; von Huene and Scholl, 1991; Lallemand 100% SEDIMENT UNI)ERT!tRUST et al., 1992a]. For example, the Japan and Peru submarine wedgesprobably decreased their volumeby about50 % during OCEAN FLOOR the Neogene.Lallemand [1992] compiledthe subductionzones SEDIMENT where great subsidencetogether with volcanic front retreat _'C7___-_-; were documented.These "erosional"margins include northern Japan,Peru, Izu-Bonin, Mariana, Tonga and Guatemala- Costa Rica. Although the loss of upper plate material appearsquite IGNEOUSOCEAN CRUST v certain, especially in the light of the great subsidencerecorded along these margins, the mechanism responsible is still poorly known. Several explanationshave been proposedsuch as the wedging of positive relief featuresinto the subduction zones(Japan, Peru, Tonga) causingfrontal erosion[yon Huene INTERMEl)iATE-ACCRETIONARY MARGINS ( 16,000 KM ) and Lallemand, 1990; Lallemand et al., 1990], the 80% SEDIMENT UNi)ERTHRUST fragmentationof the upper plate by hydrofracturingleading to basal erosion [yon Huene and Lee, 1983; Platt, 1990] or a normal geological mode for a convergentmargin where the OCEANFLOOR ]xRWPRISM trench fill does not exceed 500 m in thickness [Le Pichon and Henry, 1992]. SEDIMENT/xCCRETIO•:::::::::.:...... ,.....• To investigate the possible causes of tectonic erosion and/or sediment accretion, we review some of the parameters V V V V V V V V V of active margins such as the convergence rate and the ••' IGNEOUSOCEAN CRUST occurenceof grabensor seamountson the subductingplate. We also use laboratory sandboxmodels to illustrate possible mechanismsof basal and frontal erosion of the overriding plate. TYPICAL-ACCRETIONARY MARGINS (7,000 KM) 70% SEDIMENT UNDERTHRUST The Coulomb Theory Applied to Accretionary Wedges OCEANFLOORAccRETIONAR¾ pRISM Previous Studies As a first approximation, we assume that accretionary prisms behave as homogeneous wedges formed of noncohesiveCoulomb material frictionally sliding on a rigid '•• IGNEOUSOCEANCRUVST v v v v base (subductingplate). This theory is scale-independent.The maximum depth of applicability is given by the downward increasein temperatureand correspondsto the transitionfrom Figure 1. Classificationof convergentmargins as proposed by von Huene and $choll [1993, p. 168] with their cumulative a pressure-dependent,time-independent Coulomb behavior to length of occurrence. In this paper we call "intermediate a pressure-independent, temperature-dependent plastic accretionary wedges" what von Huene and Scholl called behavior [Davis et al., 1983]. This transition occurs at "accretionarywedges with small prisms(5-40 km wide)" and variable depths, dependingon the thermal gradient, generally "typical accretionarywedges" what they called "accretionary at 15 + 5 km depth. wedgeswith largeprisms (> 40 km wide)." We use the exact solution of Dahlen [1984] for a critical

after Dahlen 1984 Figure 2. Cross-sectionalsketch of a submarinenoncohesive critical wedge showing the Cartesian coordinatesx, z and the angleso•, [i, W0, andWb. Strengthin the wedgeis proportionalto the effectivestress Oz, shownschematically by the shadedarea on the right. Modified after Dahlen [ 1984]. LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES 12,035 taper. Let ct and 13be the topographicslope angle and the solutions(I.t, I.t*b = tan•*•, •.) whichaccount for thestability relatedd6collement angle (Figure2). Wband W0 are the angles of a given wedge. Furthermore,it is very difficult to measure betweenthe maximum compressivestress c h and the baseand these parameters in situ. Table 1 summarizes the few seafloor, respectively. If the wedge is uniform and measurementsthat are presently available. noncohesive,then the orientation of o• is everywhere the Variables I.t and I.t*• are deduced from structural same(Wo is a constant).Dahlen [1984] demonstratedthat the considerationsof conjugatethrust faults occurringat the toe of some wedges.Hafner [1951] has shown that the Coulomb following threeequations ((9), (17) and (19) in his paper) are sufficient to determine the exact critical taper and the slip criterion is satisfied along two conjugateslip planes inclined lines orientationin every part of the stability field diagram as aboutthe c• axis at angles0 = +(45ø - •/2). Thus the sumof the a functionof surfaceslope ct andbasal dip 13(Figure 2). basal step-up angles of newly formed conjugate thrusts correspondsto 90 ø - • (Figure 3). From these data, we can (1) easilyderive I.t = tan • andW•, which allow us to calculateI.t*• 1 sinct* 1 according to equation (3). Davis and von Huene [1987] used (2) this method for the first time to derive theseparameters from ß0 = • arcsin(sin,)- • ct* Aleutian accretionaryprism data. We also applied this method 1 . ,sin•b*. (3) for the Oregon and Nankai accretionaryprisms, which exhibit • =• arcsm•,sin•) • - 2 pop-up structures at the toes of their accretionary wedges with the effective angle of basal friction, assuming that forethrust and backthrust are conjugate and newly formed. The data we used were multichannel seismic •b*= arctan[gb t• - }•)1 ' lines NT 62-8 [Moore et al., 1990a] for the Nankai wedge and WO76-4 [Moore et al., 1990b] for the Oregonwedge. Most of and the "modified"slope angle, the values of pore fluid pressureratio •. were derived from (1-P•dP) tantx). already publishedwell measurements[Davis et al., 1983]. or*= arctan ((1- •.) Our mean values of I.t and I.t*b are both based on three measurementsand are obviously only representativefor some Variables • and •b are the internaland basal friction angles. accretionarywedge toes. As these values differ considerably The internal and basal friction coefficientsare Ix = tan •) and Ixt, from the meanvalue for Taiwan of 1.03 used,for example,by = tan •t,- Variables•. and•.b are the internaland basal pore fluid Davis et al. [1983], assumingthat Byedee'slaw is valid at the pressureratios corresponding to (Ps- Pscano•)l (o, - P•anoor); Ps base (I.t•= 0.85), we will adopt the value of I.t = 0.52 averaged is the pore fluid pressureand o, is the lithostatic pressure amongthree in situ measurementswith an uncertaintyof about along an axis normal to the seafloor.Variables Pw and p are 20 %, rather than any other empirical value deducedfrom laboratory experiments [Byerlee, 1978]. Furthermore, the the densitiesof water and rocks respectively. methodused by Davis and yon Huene [1987] allowsIa.*• to be Discussion About Some Characteristic Parameters determineddirectly without knowing I.t•, •. and •.•. In the same The reliability of the characteristicparameters previously way, the mean value of •. = 0.88 is averagedfrom five well describedneed to be discussed,as there exists an infinity of measurementswith an uncertaintyof about 10 % (Table 1).

Table 1. Summaryof Available CharacteristicParameters MeasuredValues I• •* b=•b(1-•b)/(1-•.)

Taiwan _ - 0.675 :!: 0.05 * Aleutian 0.45 :!:0.09 t 0.28 + 0.07 t •. 0.87 * Nankai 0.50 :!: 0.!0 0.20 + 0.10 - Guatemala _ - highõ Oregon 0.62 + 0.10 0.23 :!: 0.10 0.85 :!:0.03" Barbados _ - • 1q M akran _ - -- 1 ** Mean values 0.52 :!: 0.10 0.24 :!: 0.09 0.88 :!: 0.08

VariablesI.t, I.t*b and •. weremeasured in accretionarywedges and partly published by Daviset al. [1983]. * From well measurements[Suppe and Wittke, 1977], * Fromstructural considerations of conjugate thrust faults at the toe of the wedge[Davis and yonHuene, •987], * Fromwell data[Hottman et al., 1979], õ This wasdeduced from well data[Aubouin et al., 1982], •1From well data [Moore and von Huene, 1980], qlFrom well measurements (packer) [Moore et al., 1982], ** From well data,private communicationto Davis et al. [1983]. The valuesin bold were calculatedby the authorsfrom structuralconsiderations of conjugatethrust faults at the toe of wedgesusing the methodof Davis and yon Huene [ 1987]. 12,036 LALLEMAND ET AL.: ACCRETIONARYAND NONACCRETIONARYWEllES

'4•0• •OXeJ Trenchf'fil

D6collement sin{*b- 1St0+ =•b 45 =o 90ø_•/2 -• 2xlt b= arcsin ( sin{ ) - {*b I=t,l I **hi

Aleutian 22 ø 44 ø 14 ø 0.45 0.28

Oregon 23ø 35ø 6 ø 0.62 0.23

Nankai 24 ø 39 ø 7 ø 0.50 0.20

Figure 3. Sketchshowing the methodused by Davis and yon Huene [1987] to determine,for the Aleutian Trench, the "internal"and "basaleffective" coefficient of friction at the toe of a wedge.We have usedthe same method to calculatethese parameters in the Oregon and Nankai trencheswhere pop-up structuresare observed at the toe of the wedges.

Figure4 is an exampleof a stabilityfield diagrambased on we have studied 28 well-constrainedtransects of "typical equations(1), (2) and (3), using mean valuesof characteristic accretionary,""intermediate" and "nonaccretionary"wedges naturalparameters (see Table 1). The meanrock densityhas that are listed in Table 2 (see locationon Figure5). We chose been taken as 2300 kg/m3. The tectonic regimes nine transectsrepresentative of typical accretionarywedges (compressive,extensive) of criticaltapers along the envelope (according to the literature). Five of these are located in the are labeled on Figure 4. Two restrictions can be made Lesser Antilles area to examine latitudinal variationsof taper concerningthis averagedstability field. First, mean values within a given subductionzone. We have classified 11 other werededuced from parameters measured in typicalaccretionary transectsas "intermediate"accretionary wedges characterized wedges.Consequently, the proposedmean envelopeis not by small accretionary prisms. Five of these transects are necessarilyrepresentative of nonaccretionarywedges. Second, located along the Japan Trench to examine lateral variations. the error bars for each parameterare significant.We have Finally, eight transectswere selectedamong wedges without indicated the uncertaintyon each characteristicparameter accretionary prism and generally characterized by tectonic within smaller diagrams(Figure 4) drawn for some realistic erosion. Five of these are located in the Tonga - northern setsof values (-4 ø < (x < 14ø, 0 ø < [3 < 20ø). The uncertainties Kermadec Trench. are thosementioned in Table 1, i.e. [t + 20%, [t* b+ 40% and)• We have plotted the mean representativetapers for each of + 10%. We note that the locusof the stabilityfield in an ((x, the 28 transectson a mean stability field diagram (Figure 6) similar to those of Figure 4. Table 2 summarizesthe different 13)diagram is not very sensitiveto variationsof friction[t and frontal tapersconsidering mean o• and • measuredfrom the g't,, but is highlysensitive to variationof porefluid pressure trench back to a vertical line correspondingto a d6collement •.. depth of 10 km below seafloor. We have used publisheddepth sections(sources are given in Table 2) for estimatingthe dip Lessonsfrom natural wedges angle of the d6collement. Because of the time/depth conversionuncertainty, we considerthat the error on [3 can Taper variations reach _+20%. Inasmuchas the stability field is elongatedin the Before examining the mechanical control of accretion or [I direction, variations in • do not severely affect the erosion,let us considerthe presentdistribution of convergent- conclusions. margin taperson a mean stabilitydiagram. For that purpose, We observethat typical accretionarywedge tapersare close

Figure 4. Stabilityfield diagramfor criticalwedges (surface slope (x versusbasal dip [3)using mean measured frictionangles and pore fluid pressure ratio, i.e., [t = 0.52;[t* b = 0.24;)• = 0.88;Pw = 1030kg/m3 and p = 2300 kg/m3 (meandensity for a 0- to 10-kmwedge thickness [Lallemand et al., 1992a]).Slip lines are plotted along the stability field envelopewhich coincideswith critical tapers.The box outlines the set of realistic valuesof ((x, [3)for naturalwedges. We haveplotted the error estimatesof the threecharacteristic parameters: g + 20%; g*b + 40%; )• + 10% in the top threediagrams. Solid circlescorrespond to the meantapers of the 28 natural transectslisted in Table 2 and labeled in Figure 6. Note the high sensitivityof the area of the stability field to the pore fluid pressurevariations and converselythe low sensitivityof the coefficientsof friction. LALLEMANDET AL.:ACCRE•ONARY AND NONACCRETION•Y WEDGES 12,037 12,038 LALIY_•AND El' AL.:ACCRETIONARY AND NONACCRETIONARYWEDGES

r. o LALLEMANDEl' AL.:ACCRETIONARY AND NONACCRETIONARY WEDGES 12,039

••• •'• '"ø '"ø••, 0 ¸• , , , .9,o••'•' •'• o

A A A

I I ! I I I I •

ZZZZZZZZ 12,040 LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES

I I I I I ! I I I 2ø 4ø 6ø 8ø 10o 12ø 14ø 16ø 18ø

14 ø-- Non-accretionary wedges 12 ø-- IUNSTABLEEXTENSIONAL REGIME I

l 0 øl

N3o ON7 8 o-- Intermediate accretionary wedgesx 6 ø_ oN2 118 x '1•18 N1O

x15 •14 18 A5 x 12 19 STABLE] A Typical accretionary wedges

Figure 6. Enlargedview of the observedstability field (see Figure4 for location),with the mean tapersof the 28 transectsplotted and labeled(see values in Table 2). The threegroups of wedgesare clearlydistinct on thatdiagram except some wedges such as Nankai, Aleutian and Costa Rica. See the text for furtherexplanation. to each other (0.6ø < at < 1.1ø and 1.5ø < 13< 4.1ø) exceptthe observed tectonic regimes: compressional, stable or Nankai and Aleutian wedges where basal tectonic erosion is extensional(see Figure 4). Unless we obtain measurementsof suspectedfurther back from the trench [Lallemand, 1992. friction angles or pore fluid pressuresin those areas, we can Seven tapers representative of typical accretionary wedges only concludethat tapersare large and very scattered. form a cluster around the lower stability mean envelope.This The lateralvariations of eachtaper among the five transects probably means that their friction angles and pore fluid crossingthe Japan Trench margin (14 to 18; Figure 8) are pressureratios are similar. Small lateral variations are noted significant but remain close to the upper stability field along the LesserAntilles subductionzone (A1 to A5; Figure7) envelope.At smallerwavelength (10 km insteadof 80 kin), we despite the increasing size of the accretionary wedge from observelarge variationsof at (0.1ø < at < 8.6ø) along eachof north to south due to the sediment input from the Orinoco the five Japanesetransects, attesting either that characteristic river. At smaller wavelength (20 km insteadof 150 km), we parametersvary along a transector that the wedges may be always observe very smooth topographic slopes (0 ø < at < stable. This can be explainedby the long history of tectonic 1.9ø) along each of the five Lesser Antilles transects. erosionthat shapedthe JapanTrench marginduring Neogene "Intermediate"and nonaccretionarywedges are scatteredin time, which was followed during Quaternary time by some regionsof higherat and 13,the largesttapers corresponding to accretion [yon Huene et al., 1982; yon Huene and Lallemand, nonaccretionary wedges. Most of them are located in the 1990; Lallemand et al., 1992a]. Lateral variations from one vicinity of the upper extensionalenvelope, two of them (12 transect to another are of the same order of magnitude as and I9) are clearly locatedin the mean stability field, and three variations along a single transect.We note that transectI7, of them (N3, N4 and N7) are located in the unstable which was used as a reference profile for demonstrating extensionalfield. Such a dispersioncan be interpretedin two tectonic erosion processes [e.g., yon Huene and Culotta, ways. If we assumethat characteristicparameters are similar in 1989], can be distinguishedfrom the other transectsby a every subduction zone, then we can conclude that half of higher surfaceslope angle at the toe of the prism. intermediate to nonaccretionary wedges are stable/passive Similar remarkscan be made concerningthe five transects when tectonic erosion occurs and half are characterized by of the Tonga-northemKermadec Trench margin (N2 to N6; extensional faulting. Alternatively, if we assume that the Figure 9) but with a larger scatteringof mean tapers, even characteristic parameters of "intermediate" and greater when considering variations along each of them at nonaccretionary wedges are significantly different, then smaller wavelength (0.2 ø < at < 25.9ø!). This very steep several stability envelopescan be adjusted to accountfor the extreme surface slope was measured along a 4-km-long LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES 12,041

margins exhibit scattered and larger tapers with large variations in surface slope along one transect. This rough morphology may reflect some tectonic activity such as basal removal of upper plate material, which steepensthe surface slope. Depending on the tectonic regime of such margins, 18øN deduced from seismic interpretationsor diving observations, • Barbuda it is possible, for example, to provide some estimatesof the fluid pressureratio X within the wedge, making some basic assumptions(e.g., keepingthe two otherparameters g andg*b constant; insets of Figure 4). Increasing X narrows the stability field and consequentlybrings the compressionaland extensional domains closer.

A4 Guadeloupe Other variations of specific parameters N In addition to the tapers, we compare the following parametersin Table 2: the trench sediment thicknessT t, the orthogonal convergence rate v, the vertical offset of oceanic faults V o and an estimate of the oceanic seafloor roughness along the 28 selectedtransects. We define an "oceanicseafloor Martinique roughness"coefficient (Re) which is a factor proportionalto the total volume of oceanicpositive and negativefeatures such as grabens, ridges or seamounts within 200 km from the trench axis, integrated over the three sets of transects(Lesser 14øN Antilles, Japan and Tonga) and then normalized. Figure 10 summarizesthe mean specificparameters (ct, [3,Tt, v, Vo, Re) for each group of convergent margins. The product T t x v • Barbados• correspondsto the "modeminput sedimentaryflux" (arbitrary unit in Table 2; 10'• km2/Ma). The correlationbetween the different groups of wedges for each specific parameter is

12øN

62øW 60øW 58 øW

Figure 7. Location of the five transectsoff Lesser Antilles. segment of transect N4. Again, it indicates either that characteristicparameters vary along a transector that the 40ON wedgemay be stable.The steepnessof the surfaceslope for the two transectsN3 and N4 is probablyrelated to the recent JAPAN subductionof the LouisvilleRidge [Lallemandet al., 1992b]. Extensional faulting has been documented north of the collision zone [Pelletier and Dupont, 1990] and is in agreementwith the tapers'distribution on the meanstability diagram(Figure 6). Pelletier and Dupont [1990] have shown that the tectonicregime in the KerrnadecTrench changes near 32øS from erosion in the north to accretion in the south. We thus decided to classify the southernKermadec transect I3 in the "intermediate"accretionary wedge group. The surfaceslope of this transect is far smoother than those of N2 to N6. Fromthe observedvariations of 28 convergentmargins, we concludethat marginscharacterized by large accretionary wedgesare homogeneousregarding their small tapers(cluster of pointson Figure6). The locationof the clusteralong the lowerstability mean envelope is compatiblewith the process of continuousfrontal accretion. The wedgesswing between a Seamount compressionaltectonic regime and stability close to the Kashima criticaltaper. Surface slopes are generallysmooth compared with other margins.On the other hand, nonaccretionary Figure 8. Locationof the five transectsoff northeastJapan. 12,042 LALLEMANDET AL.:ACCRETIONARY AND NONACCRETIONARYWEDGES

candidatesas primary causes for frontal accretionor tectonic 16øS erosion. Second, we also observe good correlationsbetween the type of the margin and two other parameters,the taper angle and the oceanicfaults vertical offset. Thus we will try to clarify in this paper the type of interactionwhich accountsfor these observations.

LessonsFrom Sandbox Modeling 20øS Tongatapu Sandbox modeling is especially useful because we can measuredirectly the internal and basal friction anglesas well as the step-up angles of faults. Furthermore,we can observe the dynamic evolution of the wedge and its internal deformation during experiments. A glass-sided rectangular deformation box roughly reproduces the geometrical conditions and kinematics of a subductionzone. A polyvinyl chloride (PVC) medium which simulatesthe oceaniccrust with its sedimentarycover is pulled at a constantrate. Sand units are progressivelystacked against I Osboum 24osa rigid backstop, analogousto the island arc or continental basementagainst which sedimentsare accreted,to generatean accretionaryprism. It was demonstratedby Malavieille et al. [1991] that a 30ø+ 5ø dipping backthrustforms within the sand when the backstopinclination differs from this critical angle, which is close to the angle of friction of dry sand. Isotropic models are built by forming horizontal sand layers, which include colored passive marker beds, on the moving plate. Dry quartz sand of aeolian origin, with a grain size of less than 300 gm, is used as an analoguefor the oceanic sediments that make up accretionary wedges. The sand is essentially cohesionless and its deformation is time- independent.This material satisfactorily simulates the brittle Coulomb behavior of shallow crustal rocks in laboratory experiments[e.g., Hubbert, 1937; Hotsfield, 1977; Davis et al., 1983; Mc Clay and Ellis, 1987; Mandl, 1988; Mulugeta, 1988]. To a first approximation,we assumethat sedimentary rocks of accretionaryprisms behaveas a single layer with an internal friction angle of 27 + 5ø (this study). Their cohesion, Co, is negligible compared with common shear stresses Figure 9. Location of the five transectsoff Tonga Islands recordedin nature(about 107 to l0 s Pa). For the sandused in and the one off Southern Kermadec Islands. our experiments,cohesion C s = 20 Pa is alsonegligible and the internal angle of friction, 0 = 300 (g = tan 0 = 0.57), is similar. Thus the main difference with nature is the absence of obvious,but there is no clear correlationwith T t x v. Typical pore fluid and hydrostatic pressures.The general Coulomb accretionarywedges are characterizedby thick trenchsediment criterion for shear traction x at failure becomes (average:2.4 km), low convergencerates (average: 3 cm/yr), almost no oceanic fault scarpsand a small seafloorroughness x = C0+ (o,•-Pf) tan O (4) coefficient.Non-accretionary wedges are characterizedby thin whereC o is thecohesion, On is thenormal stress, and Pf is trench sediment (average: 0.4 km), high convergencerates the fluid pressure(= 0 in sandboxexperiments). Variables (average 11.7 cm/yr), larger oceanicfaults offsets(0.7 km) and a high seafloor roughnesscoefficient. In contrast, and •,* simplify into t• and• N equations(2) •d (3) b•ause variationsin input sedimentbudgets do not clearly correlate •= •=0 •d p• =0 mo. with any type of accretionarywedge. Densityof spri•led sandis about1.7 g/cmz •d themean These observations are compatible because low densityof accretedrocks is about2.3 g/cmz. The deformation convergencerates allow greater trenchdeposit, assuming a in these experiments occurs in a normal gravity field. constant trench sedimentation rate. Thick trench fill favors Consideringa me• cohesionof 10-15 MPa in incompletely frontal accretionand a wide accretionarywedge (with a gentle lithified sediments [Hoshino et al., 1972] •d a measured surface slope) would deflect the subductingplate less than a cohesion of 20-170 Pa at the interface between the dry sand margin characterizedby a larger taper [Karig et al., 1976]. •d •e v•ous basalmediums, the scalingof the experiments First, we can reasonablypostulate that the type of the margin [Hubbert, 1937; Ramberg, 1981] is such that 1 cm in the (i.e., accretionaryor nonaccretionary)does not influence the model is more or lessequivalent to 1 • • nature.•e scaling convergencerate v or the occurrenceof seamountsand ridges factorof 10• is calculatedby the ratio (C/pg) •,• / (C/pg) on the oceanicplate V o. Consequently,v and Vo both are good • •x•i•n•- LALLEMAND El' AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES 12,043

Intermediate TypicalA.W. A.W. NonA. W.

11.7 cm/yr 8.8 cm/yr IOrthogonalconvergence rate= vI

6.9 ø

13.1 ø [Dipangle of d6collement =

2.4km I Trenchsediment thickness = Tt I

0.7 km

Oceanicfaults vertical offset = V91

13 15 I Oceanicseafloor roughness coefficient = R cI

7.2 7.9

Modemsediment input flux = Tt x vI

Mean valuesobtained from twentyeight transects Figure 10. Mean parameterscharacterizing each group of wedgesas describedin Table 2. AbbreviationA. W. meansaccretionary wedges. See the text for further explanations.

Five experiments have been conductedusing the same apparatusand sand, but different boundaryconditions. Several authorshave discussedresults from sandboxmodeling when EXP.,I simulating an accretionarywedge [Davis et al., 1983; Dahlen et al., 1984; Malavieille, 1984; Mulugeta, 1988; Colletta et al., 1991; Huiqi et al., 1992; Malavieille et al., 1993]. We briefly illustrate this type of experiment and propose some simple mechanical explanations for the results. In contrast, EXP.2 I experimentssimulating the wedge deformationwhen ridge or seamountsubduction occurs have not been intensivelystudied [Biagi, 1988; Malavieille et al., 1991; Robion, 1991; High bosol friction Lallemand et al., 1992b; Lallemand and Malavieille, 1992]. Low bosol friction We examinein detail someof this secondtype of experiment.

Mechanics of Accretionary Sand Wedges '<• ' 10 cm ' We have tested different values of basal friction for the Figure 11. Two modeledcritical wedgesobtained after the samethickness of accretedsand. We haveused for thatpurpose sameshortening and sameincoming sand thickness but using several basal media such as polished PVC which is two differentvalues of basalfriction. Variable gb = 0.33 in characterizedby a low coefficientof basalfriction (!.t b = 0.33 experiment 1 and 0.50 in experiment 2. For the same andCb = 170 Pa);mylar sheet (!.t b = 0.42 andC• = 160 Pa) and horizontald6collement, the topographicslope is steeperfor a higher basal friction. two typesof gummedpaper which are characterizedby higher 12,044 LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES

coefficientsof basalfriction (gb = 0.46 - 0.50 and Cb = 110 - The first effects of asperity subductionare the indentation 130 Pa). of the wedge and consequentactive compression/shorteningin Figure 11 illustrates two wedges obtained after the same front of the indenter, the uplift of the wedge above the shorteningand same incoming sand thicknessbut using two subducting feature and the blocking of the basal d6collement different values of basal friction. For the same horizontal propagation.The secondeffect which is of great interestwhen d6collement, the topographicslope is steeper for a higher looking at potential tectonic erosion is the path of the basal friction. This is easily explained by Coulomb theory d6collement in the vicinity of the asperity. The asperity is using equation(30) of Davis et al. [1983]: fixed on the basal medium, so that the d6collement is necessarilyforced upwards within the sand. This phenomenon was describedat the Tonga Trench by Ballance et al. [1989] as (5) "tunnelling process." We observeon crosssection 4 (Figure 12) that a sandwedge where the dimensionlessintegral K is a functionof I.t and g•, is isolatedin front of the indenter,which is limited arcwardby which is close to 1.8 accordingto the values obtainedby a 30 ø dipping slip plane (ol is supposedto be horizontal).The numericalcalculation with g = 0.57 (K = 1.9 for • = 0.33 and slip plane acts as a ramp while the new d6collement propagatesforward and the wedge is passivelypushed arcward K = 1.65 for g b = 0.50). This equation is apparentlynot by the subducting seamount. A similar sand wedge is not correctly dimensioned,but this is becauseDavis et al. [1983] formed in the ridge experiment (Figure 13) becauseits flank used the small-angle approximation in which sinct • ct and already dips at about 30ø . Neither does it appear in the slice sin[}= [5.Variable • = 0 in theexperiments, so that ct = •/(1 + experiment(Figure 13) becausethe arcwardflank of the slice K). The predictedvalues of 0t = 6.5ø and 10.8ø, respectively, dips only a few degreesand the d6collementjust follows the for the previouslow and high coefficientsof friction are close basal plane of weakness. to the measured values of 6.5 ø and 10 ø. From the top of the asperity (Figures 12 and 13), a new The style of deformation also differs when basal friction d6collementforms subparallel to the basal plate becauseit is changes. For example, backthrustsare more abundantwhen inherited from the first stages of asperity subduction.We the basal friction is low. The reasonis that the c•l axis dips noted that this newly formed "d6collement"(called the "top d6collement" in the following discussion) was forced to toward the foreland at an angle •g• that increases with propagatearcward at the same rate as the asperitysubduction. increasingcoefficients of basal friction [Davis and Engelder, Its level within the sand wedge thus appearsto be controlled 1985]. When the basal friction is very low, •g• approaches by the height of the asperity before subduction of the zero. The potential slip planes are then almostsymmetrical basementslice (Figure 13), except in the slice experiment, about the horizontal and backthrusts are thus common. When where the "top d6collement"is initiated as a thrust dipping the basal friction is high, •gb increases.The forward verging about 25 ø. plane then dips at 0 - •4/•,and the backwardverging plane at 0 We observe in experiment5 (Figure 13) that the top + •, with 0 = 45ø -•)/2 = 15ø [Hafner, 1951]. The shallower d6collement becomes inactive after about 10 cm of slice dip may be favored becauseof bedding strengthanisotropy subduction [Lallemand et al., 1992b] as evidencedby the [Davis and Engelder, 1985]. generationof a new accretionarywedge at the toe of the older deformedwedge. As long as the top d6collementis active,the Mechanics of Sand Wedges During entire thicknessof incomingsand is underthrustedbeneath the Seamount/Ridge Subduction deformedwedge. As soon as the d6collementjumps from a high level in the wedge to the basalplane of least mechanical Tectonic erosionis a processwhich is very difficult to test resistance,a new thrustappears within the incomingsand unit using sandboxmodels becauseof the limited possibilitiesof and generatesa new accretionarywedge. The path of the basal experimental boundary conditions. On the other hand, it is d6collementshields a domain where sandis trappedand then easy to examine wedge deformation related to asperity subductedin the wake of the asperity. subduction.This simple approachneglects factors such as The jump of the d6collementto the bottomof the wedge sedimentanisotropy or fluid pressurethat stronglyinfluence occursonly in experiment5, probablybecause the height of the processin nature,but we will see in the followingsection the basement asperity was less than in the two other that the basic conclusions obtained by the experimental experimentsand becauseof a sufficientamount of penetration approach are consistentwith the natural observations.We use (subduction).One may argue that the d6collementuses the theseexperiments as simple illustrationsof naturalprocesses. least energeticpath, but the complexity of the systemmakes Three experiments have been conducted with 2 cm of estimatesof energy requirementsovercomplicated. Such an continuousaccretion of sandabove a low frictionalmedium energetic approachhas been already discussed[Lallemand, = 0.33). The accretionarywedges were first built and then 1992; Schniirle, 1994]. Further experimentsneed to be indentedby variouspositive features,a smoothtruncated cone performedin orderto quantifythe amountof penetrationwhich simulating a seamount,a rigid body normal to the sidewalls is enough,according to a given heightof asperity,to produce with a roundedcross section simulating an aseismicridge and a such a jump. seaward vergent basementslice also normal to the sidewalls simulating an active basement thrust slice. The first Summary of the Results of Experimental Modeling experiment(seamount subduction) is illustratedin Figure 12 [Biagi, 1988]. The two other experiments(ridge and slice Five experiments were performed to illustrate the subduction) are illustrated in Figure 13 [Lallemand et al., deformationof a sand wedge when accretionand asperity 1992b]. subductionoccur. We concludethat the following: [Exp.3]

Figure12. (a)Perspective view of thefinal stage of theseamount subduction experiment 3. Note the two lobesthat develop in thewake of thesubducting seamount involving the incoming sand layer. These reflect theblocking of theforward propagation of the dScollement against the seamount. (b) Interpretedcross sectionsof the final stage of seamountsubduction experiment after the model had been cut and photographed at 2 cmintervals to observelateral variations of theinternal deformation. Note the drastic shortening of the wedgein frontof the indenter (seamount) andthe lateral ramps which accomodate thedevelopment ofthe two lobes (frontal imbricates). 12,046 LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WE•ES 12,047

1. Critical tapers stabilize along the lower stability field Positive Feature Effects envelopewhen accretionoccurs without asperitysubduction We are fully aware that the application of empirical laws (experiments 1 and 2). In contrast, tapers increase when asperitysubduction occurs (experiments 3 to 5) in responseto deducedfrom experimentsto natural casesis not rigorous,but the higher basal friction (sand against sand) along the top these experiments provide realistic explanations of various observations. Figure 14 represents four examples of ddcollement.The surfaceslope is locally overcritical(0t = 30ø subducting seamountsand associatedwedge deformation. It which is the angle of repose of sand) in response to the has been demonstratedby Lallemand et al. [1989] that the continuous removal of sand from the front of the wedge down-faulted part of the Kashima seamount, presently (tunneling effect). These experimentalresults are in good subductingin the Japan Trench, pushesforward a wedge of agreementwith previousobservations. upper plate material. The step-up angle of the thrust that limits 2. In the wake of a subductingasperity, a top ddcollement the sandwiched material is close to 30 ø. The Bougainville forms within the wedge at a height relative to the basal plate seamount in the New Hebrides Trench was also intensively which is almost the same as the height of the asperity. The surveyed.Two dives crossedthe plate boundaryshowing that a newly formed ddcollementis active over a large distance wedge of material belonging both to the apron of the beforeit jumps back to the basalplate (arrestingthe tunneling seamountand to the upper plate was pushedforward [Collot et effect). This distanceseems to increasewith the height of the al., 1992]. Using magnetic anomalies,a subductingseamount asperity.This phenomenongenerates a supercriticalcliff at was detected at the junction between the Japan and Kuril the front of the overriding wedge in responseto continuous Trenches. A gigantic scar reflects the collapse related to its reworking of the frontal part of the wedge. Furthermore,the recent subductionand the trace of the plate boundarycan be inactive flat top-ddcollement can be reactivated during followed 1.3 km above the trench floor at the level of a compressionalphases of accretionafter the downwardjump to midslope terrace [Lallemand and Chamot-Rooke, 1986]. the basal ddcollement. Finally, Ballance et al. [1989] have pointed out a subducting 3. After the downwardjump of the ddcollement,some sand seamount:the Motuku Guyot just north of the Louisville Ridge is trappedunder the basal ddcollement(shielded domain) and in the Tonga Trench, using a multichannel seismic line. The draggedpassively with the asperity.This processcontributes presence of the Motuku Guyot is marked by a prominent to the erosion of the overriding wedge and to the dramatic reflecting horizon 10 km long and a high contrast in seismic wedge deformationwhich continuesfor a long time after velocities. The surface trace of the d6collement can be penetrationof the asperity. followed 1 to 2 km above the trench floor in the vicinity of the subductingseamount. Mechanical Control of Accretion and Erosion' Other examples of subductingseamounts are published in Natural Cases the literature, all showing significant deformation of the wedge both in front and in the wake of the indenter[e.g., Fryer One of the basic observationsin our experimentsis that in and Smoot, 1985; Fryer and Hussong, 1985; Pontoiseet al., order to have tectonic erosion, the ddcollement is required to 1986; Lonsdale, 1986, Lallemand and Le Pichon, 1987; migrate upwards within the wedge material. How, then, is the Collot and Fisher, 1989; Lallemand et al., 1990]. d6collement forced upwards in both the natural and Von Huene et al. [1994] published a Hydrosweep experimental examples? We suggest two possible bathymetric map off Costa Rica where three stages of explanations: seamount subduction are beautifully documented.Figure 15 1. A subductingridge, seamountor horst first forces the shows an extract of this map. The first stage (1 on Figure 15) ddcollementto migrate upwards.Then, after a certain amount is equivalent to the examplesshown on Figure 14. It shows a of convergence,the ddcollementjumps downwardprobably to giganticscar in the wake of the subductedseamount without an a level of least mechanicalresistance. The path of the final smoothed d6collement isolates a shielded domain of rock accretionarywedge at its base, attesting that the level of the ddcollement is shallower than the trench seafloor (see the which is draggedpassively with the subductingfeature. schematiccross section 1 on Figure 16). This first seamount 2. There exists a level characterizedby a lower effective has alreadyundergone 20 km of subduction.The secondstage friction higher in the wedge. Dahlen [1984] has shown that a is illustratedby anotherseamount (2 on Figure 15), which has decreasein I.tb will producethe sameeffect as an increasein )•b already performed 30 km of subduction counted from the since they only occur in the combination I.tb(1- kb). trench. A morphological scar atteststo its subduction.At this Consequently,an upward migration of the ddcollementto a stage, a new accretionary wedge formed in its wake. This higher level will be possible if the frictional resistance is demonstratesthat the ddcollementjumped from a high level in lower or the pore fluid pressurehigher. the wedge to the bottom after between 20 and 30 km of

Figure 13. (a) Experiment 4 after Lallemand et al. [1992b]. The first illustrated stage is just before subductionof the ridge. The secondand final stageillustrates the shorteningand uplift of the wedge after ridge subduction.The active ("top") ddcollementparallels the basalplate and its level equalsthe height of the ridge. The subductingsand remainsuncleformed in the wake of the ridge. (b) Experiment5 after Lallemand et al. [1992b]. The first illustratedstage correspondsto the initiation of the thrustingof the basementslice before its incorporationbeneath the accetionarywedge. A pop-upis createdwithin the sandjust abovethe slice. The secondand final stageshows the specific wedge morphologyacquired during slice subduction.The wedge was first shortenedand uplifted; a flat thrust correspondsto the propagationof the ddcollementduring slice subductionand a new accretionarywedge develops. 12,048 LALLEMANDLT AL.:ACCRETIONARY AND NONACCRETIONARY WEDGES

0/ ß ' ' I • I • I • • • , P844. •"6øN• •' k S 42 'L•'•0 • 5 10 1520 25 30 •, 35 TRENCHJAPAN40 45 • 50 5560

8 •012 • • • KASHIMASEAMOUNT

NEW HEB•IOES

TRENCH - •0'•0t 0 5• 10• 15• 20• 25 • 30 • 35• 40 • 45 • 50 • 55• 60• Line 100

10 - B•6AI••_ 124]

0 5 10 15 20 25 30 oJ I ! • • I ,42 • KURIL

8 10.• SEAMOUNT

0 5 10 15 20 25 30 35 40 45 50 L 3-10 I

TONGA '•'•-•'•"•-• TRENCH ,,. •O'•j•'•TUKU •ZœL•"•'•.....7 .,-?" GUYOT

D LALLEMANDEl' AL.:ACCRETIONARY AND NONACCRE'TION•Y WEDGES 12,049

275•30 ' 9•30 '

95 mm/y

COSTA-RICA MARGIN

COCOS PLATE

Figure15. Hydrosweepbathymetric map of theCosta Rican margin extracted from a largermap acquired duringthe R/V Sonne 76 cruise [von Huene et al., 1994]. Several seamounts onthe Cocos plate are subducted beneaththe margin. Their morphological traces, parallel to theplate convergence, areeasily recognized on thismap. Three stages of seamountsubduction, labeled 1 to 3, areschematically illustrated oncross sections in Figure 16. subductionas illustrated on Figure 16. Finally, a third scar has not been offscraped,at least not until this depth of related to a subducted seamount is observed 50 km back from subduction(3 on Figure 16). the trenchwhere the margin'sthickness exceeds 10 km (3 on In additionto the upperplate materialwhich is dragged Figure 15). This observationdemonstrates that the seamount downwardwith the seamountsduring subduction, we have seen

Figure14. Fourexamples of subductingseamounts and associated deformation of the wedge. Cross sections,without vertical exaggeration, are based on interpretedmultichannel seismic lines except for the Kuril Trenchwhere the location of the seamountis deducedfrom a magneticmodel. The •ectionsare located on thebathymetric maps to theleft. The dashed circles superimposed onthe bathymetric maps are the surface projectionof thesubducting seamounts top. The dashed and solid lines with black triangles represent t..he deformationfront. Isobaths are each 250 m exceptfor theTonga Trench box where they are at 500 m. The JapanTrench example (Figure 14a) is after Lallemand etal. [1989]. The New Hebrides example (Figure 14b) is afterFisher et al. [1991]and Collot et al. [1992].The Kuril Trench example (Figure 14c) is afterLallemand and Chamot-Rooke[1986] and Cadet et al. [1985].The Tonga Trench example (Figure 14d) is afterBallance et al. [1989] and Gnibidenkoet al. [1985]. 12,050 LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES

before seamount subduction

Figure 16. Schematiccross sections without vertical exaggeration of threestages of seamountsubduction basedon both morphologicalobservations (Figure 15) and sandboxmodeling results. The first stagebefore seamountsubduction corresponds to the arealocated between seamount 1 and2 on Figure15. that a great thickness of frontal upper plate material is fractures [yon Huene, 1984; yon Huene and Lallemand, 1990], passivelytransported in the wake of the trailing flank of the hence the level of maximum pore fluid pressurewill move up seamountsover several tens of kilometers.This phenomenon through the wedge. Moore [1989] noted that with high ratesof contributesto the oversteepeningof the topographicslope underthrusting, as along nonaccretionary wedges, sediment especially at the toe of the wedgesas observedalong the descendingbeneath the d6collementzone is probably buried previouslydescribed nonaccretionary wedges and most of the faster than it can dewater. From Table 2 we observe that the intermediatewedges (Figure 6). Some scarsmay reflect the critical convergence rate which marks the occurrence of emergenceof presentlyinactive ancient "top ddcollements" accretion is 6 + 1 cm/yr. For fast convergencerates, the (Figures16 - number3 - and 17). Sucha prominentreflection process of upward migration of the d6collement occurs parallelingthe subductingplate and emergingat the baseof a progressively beneath the part of the wedge which is scarpis observed,for example,in the northernJapan Trench underthrusted by overpressuredsediments, in other words, 3.5 km abovethe presentddcollement [von Huene and Culotta, mainlythe frontalpart of the wedge[Platt, 1990] (Figure17). 1989]. Also, the averageporosity of a thick sectionof incoming Figure 10 also shows that nonaccretionarywedges are sediment is less than for a thin section (because of characterizedby a high amplitudeof horstsand grabens. compaction).Consequently, the fluid potentialwill be higher Consideringthe former discussion,it seemslogical that the in regions characterized by thin trench sediment/fast horstsbehave like any other positive oceanicfeatures. Their convergence rates, i.e., nonaccretionary and intermediate flanks are even steeperthan thoseof seamountsor ridges,so margins(Figure 10). This explanationis in agreementwith that the slumpedmasses from the toe of the wedgeand/or the the Coulombwedge model, because the low taperpredicted by trench fill (dependingon the trench sedimentationrate) are the modelwhen the basalfriction is low is only valid when easily trappedin the grabensand passivelydragged with the frontal accretion is active. The difference here is that basal and subductingplate (Figure 17). This processis clearlyimaged frontal removal occurs instead of frontal accretion. As a seismically in the northern Japan Trench [yon Huene and consequence,the taper adjusts itself to the downward flux of Culotta, 1989], the Kurile Trench (P. Schnurleet al., Tectonic material,with respectto the Coulombwedge model, to become regimeof southernKurile Trenchas revealedby multichannel eitherstable or evenovercritical (Figure 6). seismiclines, submitted to Tectonophysics,1994), the Tonga Trench [Ballance et al., 1989] and the Peru and New Hebrides Trenches [yon Huene, 1986]. Conclusions

Lower Effective Basal Friction Higher in the Possible Causes for Frontal Accretion Wedge? We havepointed out in thispaper that typical accretionary Along nonaccretionaryor intermediatemargins such as wedgesare characterizedby a low convergencerate, a thick Japan,fragmentation by hydrofracturingof the upperplate trenchsediment thickness, a low taperand a relativelysmooth was invoked by yon Huene and Lee [1983] and Platt [1990]. subductingseafloor (Figure 10). Consequently,we conclude, Rock at the base of the upper plate may disaggregateas to a first approximation,that a low convergencerate allows overpressuredpore fluid invades and permeatesthrough the deposition of a thick trench fill, which nourishes the LALLEMANDET AL.: ACCRETIONARYAND NONACCRETIONARYWEDGES 12,051

Anywedge j' z•uplift subsidenceoldd6collementneoformed

0 10 Km •7 remnantarp 0 I .. material• ------•-'""•••

•'" ct•• upward •' trenchretreat jumping of m thed6collement

Nonor Intermediateaccretionary wedges I steepening subsidence of frontal __ • • slope 00I 5I Km hydrofracturings•'J.•$ v _?•."-'--•_..•••---•• •,,•"•.•-• fillinggrabens of I cousedby •.•.•; •. •'•;.••• ' I subductin•. • dewatering. .. s's's s •.• s' s;.s•• s- s• • •'•••-'"' •is • '• .••_:.•••..•'"'"--'""•'-•'"'•"•••"••'••'••••••••• I upper p•a•e •. •' '. •. s •- •. • •.•• • ....•-" '""•'•-•'•'•'•••...... L5 Km / material s• s.• s•••• "'-'-•...... : • .....•...... •: [rench re[rear ••• • •""""'-"••'-'•"'•=•••...... _ • progressiveupward t , migrationofd&collement •:•#•-..... ' • convergenceratemodified afteryon Huene andCulotta E198 9 ] Figure17. Sketchesshowing mechanisms of tectonic erosion along convergent margins. (a) A subducted positivefeature is responsible forthe dragging of upperplate material (dotted area). Th!s situation may occur alongany type of wedge.(b) This figure was modified after von Huene and Culotta [197,9]. Horst and graben structuresare subducting as happens generally along nonaccretionary or intermediate accretionary wedges. Grabenstrap upper plate material slumped from the front of thewedge. The fast convergence rate, generally typicalof nonaccretionaryor intermediate accretionary wedges, is responsible fora delayin dewateringof underthrustedsediment and thusfavors the upwardmigration of the d6collement.Again, the dottedarea correspondsto theremoval of upperplate material. accretionarywedge by partitioningof sedimentat somelevel seamountsand ridges. The bestexample is probablythe Tonga within the trenchfill. Many authorshave alreadydescribed trenchwhich shortenedby about 100 km in the wake of the this mechanismof sedimentpartitioning [e.g., Le Pichonand subductedpart of the Louisvilleridge [Pelletier and Dupont, Henry, 1992]. The low taperis purely a consequenceof 1990; Lallemandet al., 1992b]. This erosionalprocess related continuous frontal accretion maintaining the taper at the to subductingseamounts and ridges is, a priori,independent of undercriticallimit with respect to the local characteristic the convergencerate, henceof the type of margin. A fast parameters(lower envelope on Figures4 and6). However,the convergencerate will just acceleratethis type of erosional existence of accretion does not exclude erosional processes process.This is lessobvious when consideringthe grabens, beingactive as suggestedby Le Pichonand Henry [1992]. as shownon Figure 18. Severalexamples, especially from "intermediateaccretionary The secondcause is probablythe convergencerate driving wedges"(northern Japan and Peru, for example),illustrate two possiblemechanisms that produce hydrofracturing of the concomitant frontal accretion and basal tectonic erosion. upperplate materialby overpressuring(Figure 10). First, a convergencerate fasterthan the dewateringrate of subducted Possible Causes for Frontal and Basal Tectonic sedimentwould produceoverpressuring which may lead to Erosion some form of sediment stoping. Second, a fast convergence Figure 18 suggestspossible mechanisms of tectonic rate doesnot allow the depositionof a thick trenchfill b•ause erosion and interactionsbetween the various parameters.It of time considerations. Because of a low degree of accounts for the observations made both on the studied compaction, the thin section of trench sediment is transectsand from the experiments. characterizedby a high porosity,hence a high fluid content, The first and most simple causeof tectonicerosion is, which resultsin overpressuringalong the ddcollement. without doubt, the subductionof oceanicfeatures, which drag We also suggestthat tectonicerosion processes become upperplate materialwithin grabensor in the vicinityof self-maintainedafter they are activefor a sufficienttime. We 12,052 LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES

cFASTRATE OF-• lesstime for meanporosity decrease .• ( HIGHPOROSITY OF ONVERGENCEJ trenchfill deposition (THINTRENCH HLL) becauseless compaction !•UBDUCI•DSEDIMENT.•/

FLUID OVERPRESSURES (HIGH DIPANGLEh NEAR THE DECOLLEMENT • •,.0FDECOLLEMENT•

('•TF_.t•SURFACE'• basalremoval ofmaterial SLOPEJ '• atthe toe ofthe wedge TECTONIC EROSION

ßo%• SEAMOUNTS I SUBDUCTING RIDGES

Figure 18. Diagramsummarizing the variousinteractions between several parameters which lead to tectonic erosion [after Lallemand, 1992]. See the text for further details.

observe (see Table 2, for example) that nonaccretionary benefited from three constructive reviews from John Platt, margins are characterized by a high dip angle of their M.M.P. Bott and an anonymous reviewer. Contribution d6collement producing an increase of the amplitude of the CNRS-INSU-DBT 661. outer bulge and thus a larger bending of the subductingplate before it enters the subduction zone. This generatesnormal References faulting in the upper part of the oceanic lithosphere,hence horsts and grabens,which contributeto the processof erosion Addicott, W.O., and P.W. Richards,Plate-tectonic map of the as described earlier. Because of the continuous basal removal Circum-Pacific region, southwest quadrant, scale of material at the toe of the wedge, the surface slope also 1'10,000,000, Circum-Pac. Counc. for Energy and Miner. steepens.A second consequencewill be the increase of the Resour., Am. Assoc. of Pet. Geol., Houston, Tex., 1981. taper, so that the subducted sediment will be buried more Aubouin, J., et al., Leg 84 of the Deep Sea Drilling Project, Subduction without accretion: Middle America Trench off rapidly beneath an increasing pile of upper plate material. Guatemala, Nature, 297, 458-460, 1982. Last, it produces fluid overpressureswithin the subducted Ballance, P. F., D. W. Scholl, T. L. Vallier, A. J. Stevenson, sequenceof sedimentand hydrofracturing. H. Ryan, and R. H. Herzer, Subductionof a late Cretaceous If the self-maintenance of sufficiently mature erosional seamountof the Louisville Ridge at the Tonga Trench: A margins is real, then one could expect to find margins model of normal and accelerated tectonic erosion, characterizedby a long history of continuousremoval of upper Tectonics, 8, 5, 953-962, 1989. plate material. Such subductionzones may exist along the Izu- Biagi, R., G6om6trie et cin6matique des prismes d'accr6tion Bonin and Mariana trenches[Lallemand, 1992]. Arresting this s6dimentaires' Mod61isation analogique, Mdm. DEA, processof erosionwould probablyrequire a long periodof low Montpellier Univ., Montpellier, France, 1988. convergence rate. Byeflee, J., Friction of rocks,Pure Appl. Geophys.,116, 615- Conditionsfor tectonicerosion are accumulatingalong the 626, 1978. Cadet, J.-P., et al., De la fossedu Japonh la fossedes Kouriles: Tonga Trench becausethis subductionzone recordsthe highest Premiersr6sultats de la campagneoc6anographique franco- known convergencerate, the highest offset of oceanic normal japonaise KaYko (Leg III), C. R. Acad. Sci. Paris, 301, faults and becausethe Louisville Ridge was recentlycaught up 287-296, 1985. in the subductionzone. It is therefore not surprisingthat the Chapple, W.M., Mechanics of thin-skinned fold-and-thrust Tonga margin is typical of an erosional margin and that a belts, Geol. Soc. Am. Bull., 89, 1189-1198, 1978. great amount of upper plate material has been removed Coletta, B., J. Letouzey, R. Pinedo, J.-F. Ballard, and P. Ba16, [Pelletier and Dupont, 1990; Ballance et al., 1989; Lallemand Computerized X-ray tomography analysis of sandbox et al., 1992b]. models: Examples of thin-skinned thrust systems, Geology, 19, 1063-1067, 1991. Acknowledgments. Numerous discussions with Dan Collot, J.-Y., and M. A. Fisher, Formation of forearc basins Davis, Roland yon Huene, Xavier Le Pichon,Claude Rangin, by collision between seamountsand accretionarywedges: Manuel Pubellier, Sylvain Calassou and Philippe Robion An example from the New Hebrides island arc, Geology, allowedus to improveearlier versions of the manuscript.Jean- 17, 930-934, 1989. FranqoisBrouillet and Andr6 Bertbethelped us to draw someof Collot, J.-Y., S. E. Lallemand, B. Pelletier, J.-P. Eissen, G. the figures. This paper was first submitted in June 1992. It Glaqon, M. A. Fisher, H. G. Greene, J. Boulin, J. Daniel, LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES 12,053

and M. Monzier, Geology of the d'Entrecasteaux-New Hottman, C.E., J. H. Smith, and W. R. Purcell, Relationship Hebrides arc collision zone: Results from a deep amongEarth stresses,pore pressure,and drilling problems submersible survey, Tectonophysics, 212, 213-241, offshore Gulf of Alaska, J. Pet. Technol., 31, 1477-1484, 1992. 1979. Cooper,P., and B. Taylor, Seismicityand focal mechanismsat Hubbert,M. K., Theory of scalemodels as appliedto the study the New Britain trench related to deformation of the of geologic structures,Geol. Soc. Am. Bull., 48, 1459- lithosphere, Tectonophysics,164, 25-40, 1989. 1520, 1937. Dahlen, F. A., Non-cohesivecritical Coulomb wedges: An Huiqi, L., K. R. McClay, and D. Powell, Physicalmodels of exact solution, J. Geophys. Res., 89, 10125-10133, thrustwedges, in ThrustTectonics, edited by K.R. McClay, 1984. pp. 71-81, Chapmanand Hall, London,1992. Dahlen, F. A., J. Suppe,and D. M. Davis, Mechanicsof fold- Isacks, B.L., and M. Barazangi, Geometry of Benioff zones: and-thrust belts and accretionary wedges: Cohesive Lateral segmentation and downwards bending of the Coulomb theory, J. Geophys. Res., 89, 10087-10101, subductedlithosphere, in Island Arcs, Deep Sea Trenches, 1984. and Back-Arc Basins,Maurice Ewing Ser. vol. 1, edited by Davey, F.J., K. Lewis, J.R. Childs, and M.A. Hampton, M. Talwani, and W.C. Pitman III, pp. 99-114, AGU, Convergent margin off east coast of North Island, New Washington,D.C., 1977. Zealand, Parts I and II, in Seismic Images of Modem Iwasaki, T., H. Shiobara, A. Nishizawa, T. Kanazawa, K. ConvergentMargins Tectonic Structure,edited by R. von Suyehiro, N. Hirata, T. Urabe, and H. Shimamura,A Huene, AAPG Stud. Geol., 26, 49-53, 1986. detailed subduction structure in the Kuril trench deduced Davis, D., and T. Engelder,The role of salt in fold-and-thrust from ocean bottom seismographic refraction studies, belts, Tectonophysics, 119, 67-88, 1985. Tectonophysics,165, 315-336, 1989. Davis, D. M., and R. von Huene, Inferences on sediment Karig, D.E., J.G. Caldwell, and E.M. Parmentier,Effects of strength and fault friction from structuresat the Aleutian accretion on the geometry of the descendinglithosphere, Trench, Geology, 15, 517-522, 1987. J. Geophys.Res., 81, 6281-6291, 1976. Davis, D., J. Suppe,and F. A. Dahlen, Mechanicsof fold-and- Kawamura,T., and Y. Aoki, The Nankai Trough margin, record thrust belts and accretionarywedges, J. Geophys.Res., 55-8 in Seismic Images of Modem ConvergentMargins 88, 1153-1172, 1983. Tectonic Structure, edited by R. von Huene, AAPG Stud. Fisher, M. A., Tectonic processesat the collision of the Geol., 26, 54-56, 1986. d'Entrecasteaux Zone and the New Hebrides island arc, J. Kroenke, L.W., C. Jouannic,and P. Woodward, Bathymetry of Geophys. Res.,91, 10470-10486, 1986. the Southwest Pacific, 1st ed., scale 1:6,440,000, CCOP- Fisher, M. A., and H. G. Greene, The New Hebrides Trench, SOPAC, Honiara, Solomon Islands, 1983. Lines 1 and 12, in SeismicImages of Modem Convergent Kulm, L.D., et al., Western North American continental MarginsTectonic Structure, edited by R. von Huene,AAPG margin and adjacent ocean floor off Oregon and Stud. Geol., 26, 43-48, 1986. Washington, Atlas 1, 32 sheets, Ocean Drill. Program Fisher, M. A., J.-Y. Collot, and E. L. Geist, Structure of the Reg. Atlas Ser., Mar. Sci. Inst., Woods Hole, Mass., collision zone between Bougainville Guyot and the 1984. accretionwedge of the New Hebridesisland arc, Southwest Ladd, J.W., and R.T. Buffler (Eds.), Middle America Trench off Pacific, Tectonics, 10, 887-903, 1991. Western Central America, Atlas 7, 21 sheets, Ocean Drill. Fryer, P., and D.M. Hussong,Seamarc II studiesof subducting Program Reg. Atlas Ser., Mar. Sci. Inst., Woods Hole, seamounts,in Formation of Active Ocean Margins, edited Mass., 1985. by N. Nasu et al., pp. 291-306,TERRAPUB, Tokyo, 1985. Lallemand, S.E., Transfert de mati•re en zone de subduction Fryer, P., andN. C. Smoot,Processes of seamountsubduction (Earth Material Transfer in SubductionZones) - Travaux in the Mariana and Izu-Bonin Trenches, Mar. Geol., 64, publi6sdepuis 1990, Mem. Sci. Terre Univ. Pierre et Marie 77-90, 1985. Curie, Paris, 92-27, vol. l, 189 pp., 1992a. Gnibidenko, H. S., G.I. Anosov, V. V. Argentov, and I. K. Lallemand, S.E., Transfert de mati•re en zone de subduction Pushchin,Tectonics of the Tonga-KermadecTrench and (Earth Material Transfer in SubductionZones) - R6flexions Ozboum Seamount Junction area, Tectonophysics,112, sur les cons6quencesde l'6rosion tectonique,Mem. Sci. 357-383, 1985. Terre Univ. Pierre et Marie Curie, Paris, 92-27, vol.2, 96 Hafner, W., Stress distributions and faulting,Geol. Soc. Am. pp., 1992b. Bull., 62, 373-398, 1951. Lallemand, S., and N. Chamot-Rooke, Sur la cause du Hamburger, M., R.K. Cardwell, and B.L. Isacks, d6crochementsenestre entre les fossesdu Japon et des Seismotectonicsof the northern Philippine Island arc, in Kouriles: Subduction-collision d'un ancien volcan sous- The Tectonic and Geologic Evolution of SoutheastAsian marin (in French with English abstract), C. R. Acad. Sci. Seas and Islands, Part 2, Geophys.Monogr. Ser., vol. 27, Paris, Ser. 2, 303, 1443-1448, 1986. edited by D.E. Hayes, pp. 1-22, AGU, Washington,D.C., Lallemand, S., and X. Le Pichon, Coulomb wedge model 1983. applied to subductionof seamountsin the Japan Trench, Hamilton, W., Tectonicsof the Indonesianregion, U.S. Geol. Geology, 15, 1065-1069, 1987. Surv. Prof. Pap., 1078, 308 pp., 1979. Lallemand, S.E. and J. Malavieille, L'6rosion profonde des Hasegawa,A., N. Umino, A. Takagi, S. Suzuki, Y. Motoya, S. continents, Recherche, 249, 1388-1397, 1992. Kameya, K. Tanaka, and Y. Sawada,Spatial distributionof Lallemand, S., R. yon Huene, and R. Culotta, Subduction of earthquakesbeneath Hokkaido and northern Honshu, Japan the Daiichi Kashima seamount in the Japan trench, (in Japanesewith English abstract),J. Seismol.Soc. Jpn., Tectonophysics, 160, 231-247, 1989. 36, 129-150, 1983. Lallemand, S., J.-Y. Collot, B. Pelletlet, C. Rangin, and J.-P. Hotsfield, W. T., An experimental approach to basement- Cadet, Impact of oceanic asperitieson the tectogenesisof controlled faulting, Geol. Mijnbouw, 56, 363-370, 1977. modem convergent margins, Oceanol. Acta, 10, 17-30, Hoshino, K., H. Koide, K. Inami, S. Iwamura, and S. Mitsui, 1990. Mechanical properties of JapaneseTertiary sedimentary Lallemand, S. E., P. Schnurle, and S. Manoussis, rocks under high confining pressures,Rep. 244, 200 pp., Reconstruction of subduction zone palcogeometries and Geol. Surv. of Japan,Kawasaki, 1972. quantification of upper plate material losses caused by 12,054 LALLEMAND E-• AL.: ACCREFIONARY AND NONACCREFIONARY WEDGES

tectonic erosion, J. Geophys. Res., 97, 217-239, 1992a. Yokoyama, H. Nakatani, T. Yokota, and K. Tanaka, Lallemand, S. E., J. Malavieille, and S. Calassou, Effects of Crustal structureof the island-arcslope and the trenchoff oceanic ridge subduction on accretionary wedges: Sanriku and mechanismof subduction(in Japanese),Mar. Experimental modeling and marine observations, Sci., 10, 207-214, 1978. Tectonics, 11, 1301-1313, 1992b. Nagumo,S., J. Kasahara,and S. Koresawa,OBS airgunseismic Le Pichon, X., and P. Henry, Erosion and accretion along refraction surveynear sites 441 and 434 (J-1A), 438 and subduction zones: A model of evolution, Proc. K. Ned. 439 (J-12), andproposed site J-2B: Legs56 and57, DSDP, Akad. Wet.,95, 297-310, 1992. 1, Initial Rep. Deep Sea Drill. Proj., 56-57, 459-462, Lewis, S.D., and D.E. Hayes, A geophysical study of the 1980. Manila trench, Luzon, Philippines, 2, Forearc basin Pelletier,B., and J. Dupont,Effets de la subductionde la ride structural and stratigraphic evolution, J. Geophys. Res., de Louisville sur l'arc des Tonga-Kermadec,Oceanol. 89, 9196-9214, 1984. Acta,10, 57-76, 1990. Lonsdale, P. F., A multibeam reconnaissanceof the Tonga Platt, J.P., Thrust mechanics in highly overpressured Trench axis and its intersectionwith the Louisville Guyot accretionarywedges, J. Geophys.Res., 95, 9025-9034, Chain, Mar. Geophys. Res., 8, 295-327, 1986. 1990. Ludwig, W.J., et al., Sediments and structureof the Japan Pontoise, B., et al., La subduction de la ride de Louisville le trench, J. Geophys. Res., 71, 2121-2137, 1966. long de la fosse des Tonga: Premiers r6sultats de la Malavieille, J., Mod61isation exp6rimentale des campagneSeapso (leg V), C. R. Acad. Sci. Paris, 303, chevauchementsimbriqu6s' Application aux chaines de 911-918, 1986. montagnes,Bull. Soc. Gdol. Fr., 7, 129-138, 1984. Pr6vot, R., and J.-L. Chatelain, Sismicit6 et risque sismiqueh Malavieille, J., S. Calassou,C. Larroque,S. Lallemand,and J.- Vanuatu,Rapport 5-83, 551 pp., ORSTOM, Noum6a,New F. Stephan, Experimental modelling of accretionary Caledonia, 1983. wedges,Terra Abstr., 3,367, 1991. Ramberg,H., Gravity,Deformation and the Earth'sCrust, 2nd Malavieille, J., C. Larroque and S. Calassou, Mod61isation ed., 452 pp., Academic,San Diego, Calif., 1981. exp6rimentaledes relations tectonique/sddimentationentre Rangin, C., and M. Pubellier, Subductionand accretionof bassin avant-arc et prisme d'accrdtion, C. R. Acad. Sci. PhilippineSea Plate fragmentsalong the Eurasianmargin Paris, 316, 1131-1137, 1993. in Tectonics of Circum-Pacific Continental Margins, Mammerickx, J., and S.M. Smith, Bathymetry of the edited by J. Aubouin and J. Bourgois, pp. 139-164, southeast Pacific, scale 1:6,440,000, Scripps Inst. of TERRAPUB, Tokyo, 1990. Oceanogr., Univ. of Calif., La Jolla, 1978. Reyners,M., A microearthquakestudy of the plate boundary, Mandl, G., Mechanics of tectonic faulting: Models and basic North Island, New Zealand, Geophys.J. R. Astron. Soc., concepts,in Developmentsin Structural Geology, vol. 1, 63, 1-22, 1980. 407 pp., Elsevier, New York, 1988. Robion, P., Etude de la d6formationd'un prisme d'accr6tion Maritime Safety Agency, Bathymetric chart of south-west soumish de l'6rosiontectonique: Approche exp6rimentale Nippon, Map 6314, scale 1:1,000,000, Tokyo, 1983a. et num6rique,Mdm. DEA, 63 pp., ParisVI Univ., 1991. Maritime Safety Agency, Bathymetric chart of north-east Sakurai, M., A. Mogi, J. Chujo, and T. Miyazaki, Nippon, Map 6312, scale 1:1,000,000, Tokyo, 1983b. Multichannel seismic reflection profiling off Joban Maritime SafetyAgency, Bathymetricchart of Hokkaido,Map District Northeast Japan, Rep. Hydrogr. Res., 16, 1-17, 6311, scale 1:1,000,000, Tokyo, 1983c. 1981. McCaffrey, R., Active tectonics of the eastern Sunda and Schntirle, P., Contribution h la compr6hension des Banda arcs, J. Geophys. œes.,93, 15163-15182, 1988. m6canismesd'6rosion tectonique et h la quantificationdes Me Clay, K.R., and P.G. Ellis, Analogue models of flux de mati•re dans les zones de subduction, Mem. Sci. extensional fault geometries, in Continental Extensional Terre Univ. Pierre et Marie Curie, Paris, 94- 05, 173 pp., Tectonics,edited by M.P. Coward, J.F. Dewey, and P.L. 1994. Hancock, Spec. Publ. Geol. Soc. London, 28, 109-125, Scholl, D.W., J. McCarthy, and H. Ryan, Forearc margin, 1987. Central Aleutian Ridge, in Seismic Images of Modern Moore, J. C., et al., Offscrapping and underthrustingof ConvergentMargins Tectonic Structure,edited by R. von sedimentat the deformationfront of the BarbadosRidge: Huene, AAPG Stud. Geol., 26, 10-13, 1986. Deep Sea Drilling Project leg 78A, Geol. Soc. Am. Bull., Shipley, T.H., and G.F. Moore, Sediment accretion, 93, 1065-1077, 1982. subduction,and dewatering at the base of the trench slope Moore, J. C., Tectonics and hydrogeologyof accretionary off Costa Rica: A seismic reflection view of the prisms: Role of the d6collementzone, J. Struct. Geol., 11, d6collement,J. Geophys.Res., 91, 2019-2028, 1986. 95-106, 1989. Speed, R.C., et al., Lesser Antilles Arc and adjacentterranes, Moore, J.C., and R. von Huene,Abnormal pore pressureand Ocean Drill. Program Reg. Atlas Set., Atlas 10, 27 sheets, hole instability in forearc regions: A preliminary report, Mar. Sci. Inst., Woods Hole, Mass., 1984. report, 29 pp., Ocean Margin Drill. Proj., Menlo Park, Suppe, J., and J. H. Wittke, Abnormal pore-fluid pressuresin Calif., 1980. relation to stratigraphyand structurein the active fold-and- Moore, G. F., T. H. Shipley, P. L. Stoffa, D. E. Karig, A. thrust belt of northwestern Taiwan, Pet. Geol. Tailvan, 14, Taira, S. Kuramoto, H. Tokuyama, and K. Suyehiro, 11-24, 1977. Structure of the Nankai Trough accretionaryzone from von Huene, R., Structuraldiversities along modern convergent multichannel seismic reflection data, J. Geophys. Res., margins and the role of overpressuredpore fluids in 95, 8753-8765, 1990a. subduction zones, Bull. Soc. Geol. Fr., 7, 207-219, 1984. Moore, J. C., D. Orange,and L. D. Kulm, Interrelationshipof von Huene, R. (Ed.), Seismic images of modern convergent fluid venting and structuralevolution: Alvin observations margins tectonicstructure, AAPG Stud. Geol., 26, 50 pp., from the frontal accretionaryprism, Oregon,J. Geophys. 1986. Res., 95, 8795-8808, 1990b. von Huene, R., and R. Culotta, Tectonic erosion at the front of Mulugeta, G., Modelling the geometry of Coulomb thrust the Japan Trench convergent margin, Tectonophysics, wedges, J. Struct. Geol., 10, 847-859, 1988. 160, 75-90, 1989. Murauchi, S., T. Asanuma, H. Taguchi, H. Kinoshita, I. von Huene, R., and S. Lallemand, Tectonic erosion along the LALLEMAND ET AL.: ACCRETIONARY AND NONACCRETIONARY WEDGES 12,055

Japanand Peru convergentmargins, Geol. Soc. Am. Bull., Wadge, G., and J.B. Shepherd, Segmentationof the Lesser 102, 704-720, 1990. Antilles subduction zone, Earth Planet. Sci. Lett., 71,297- von Huene,R., and H. Lee, The possiblesignificance of pore 304, 1984. fluid pressuresin subductionzones, AAPG Mem., 34, 781- Westbrook, G.K., J.W., Ladd, P., Buhl, N., Bangs, and G.J., 791, 1983. Tiley, Cross-sectionof an accretionary wedge: Barbados von Huene,R., and D.W. Scholl,Observations at convergent Ridge complex, Geology, 16, 631-635, 1988. margins concerning sediment subduction, subduction Wiens, D.A., Bathymetric effects on body waveforms from erosion, and the growth of continental crust, Rev. shallow subduction zone earthquakes and application to Geophys., 29, 279-316, 1991. seismicprocesses in the Kurile trench,J. Geophys.Res., von Huene, R., and D.W. Scholl, The return of sialic material 94, 2955-2972, 1989. to the mantleindicated by terrigenousmaterial subducted at Zhao, W.-L., D. M. Davis,F. A. Dahlen,and J. Suppe,Origin convergent margins, Tectonophysics, 219, 163-175, of convex accretionarywedges: Evidence from Barbados, 1993. J. Geophys.Res.,91, 10246-10258, 1986. yon Huene, R., M. Langseth, N. Nasu, and H. Okada, A summaryof Cenozo¾ctectonic history along the IPOD S.E. Lallemand and J. Malavieille, URA CNRS 1760 Japan trench transect, Geol. Soc. Am. Bull., 93, 829-846, 1982. "G6ophysiqueet Tectonique",Institut des Sciencesde la Terre, yon Huene,R., L. Kulm, J. Miller, and D. Hussong,The Peru de l'Eau et de l'Espace, Universit6 Montpellier II, case 060, continental margin, record sections2 and 3, in Seismic place E. Bataillon, 34095 Montpellier c6dex 05, France. (e- Imagesof Modern ConvergentMargins Tectonic Structure, mail: [email protected]) edited by R. von Huene, AAPG Stud. Geol., 26, 37-40, P. Schntirle, Laboratoire de G6ologie Structurale, 1986. D6partementde G6otectonique,boite 129, Universit6Pierre et von Huene, R., et al., Morphotectonicfeatures of the Costa Marie Curie, T 26-0, El, 4 place Jussieu,75252 Paris c6dex 05, France. RicanPacific margin surveyed during the Sonne76 cruise, in "Geologicand TectonicDevelopment of the Caribbean Plate Boundary in SouthernCentral America", Dedication: Richard Weyl (1915-1988), Spec. Publ., Geol. Soc. Am., (ReceivedMay 24, 1993; revisedDecember 6, 1993; accepted in press, 1994. January 10, 1994)