Collisional subduction: how it works

HAL Id: hal-02333652

https://hal.archives-ouvertes.fr/hal-02333652

Submitted on 25 Oct 2019

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Collisional subduction: how it works

Paul Pitard, Anne Replumaz, Francesca Funiciello, Laurent Husson, Claudio Faccenna

To cite this version:

Paul Pitard, Anne Replumaz, Francesca Funiciello, Laurent Husson, Claudio Faccenna. Collisional subduction: how it works. Earth and Planetary Science Letters, Elsevier, 2018. �hal-02333652�

Collisional subduction: how it works

1

2

Paul Pitard

, Anne Replumaz

, Francesca Funiciello

, Laurent Husson

and

3

Claudio Faccenna

4

aISTerre (Institut des Sciences de la Terre), Université Grenoble Alpes, CNRS, F-38000 5

Grenoble, France 6

2Dipartimento Scienze, Università degli Studi, Roma TRE, Largo S. Leonardo Murialdo 1, 7

00146 Rome, Italy 8

*E-mail: [email protected] 9

10

Abstract

11

Since several decades, the processes allowing for the subduction of the continental 12

lithosphere less dense than the mantle in a collision context have been widely explored, but 13

models that are based upon the premise that slab pull is the prominent driver of plate tectonics 14

fail. The India-Asia collision, where several episodes of continental subduction have been 15

documented, constitute a case study for alternative views. One of these episodes occurred in 16

the early collision time within the Asian plate between two blocks, in front of the Indian 17

lithosphere during its northward subduction that followed the oceanic subduction of the 18

Tethys ocean. But the mantle kinematics and dynamics that induce such counter-intuitive 19

behavior, wherein subduction is not driven by slab pull, is not yet well understood. By means 20

of analogue experiments, we explore the southward subduction dynamics of the Asian 21

lithosphere below Tibet, where it is not attached to any dense oceanic slab (collisional 22

subduction). During oceanic subduction (analogue of Tethys subduction), the main 23

component of slab motion is driven vertically by its negative buoyancy, while the trench rolls 24

*Manuscript

Click here to view linked References

back. The convective pattern consists in a pair of wide convective cells on both sides of the 25

slab. But when the indenter (Indian analogue) attached to the ocean starts to bend and plunge 26

in the mantle, trench motion reverses. Its advance transmits the far field forces to two upper 27

plates (Asian analogues). The more viscous frontal plate thickens, and the less viscous 28

hinterland plate, which is attached to the back wall of the box, subducts. During this 29

transition, a pair of sub-lithospheric convective cells is observed on both sides of the Asian 30

analogue slab, driven by the shortening of the frontal plate. It favors the initiation of the 31

backwall plate subduction. Such subduction is maintained during the entire collision by a 32

wide cell with a mostly horizontal mantle flow below Tibet, passively advecting the Asian 33

analogue slab. We conclude that collisional subduction of the Asian lithosphere is driven by 34

the northward transmission of far field forces due to the underlying global convection cell 35

underneath the Indian plate, transmitted by the forward advance of the Indian indenter.

36 37

1. Introduction

38

Continental subduction has been proposed for decades as a key process occurring during 39

the long-lasting collision between India and Asia, with a northward motion of India at a rate 40

in excess of 50 mm/yr since ~50 Ma (e.g. Patriat and Achache, 1984; Molnar and Stock, 41

2009; van Hinsbergen et al., 2005), allowing the subduction of the continental lower crust and 42

lithospheric mantle while the upper crust thickens and forms the Tibetan plateau (e.g.

43

Mattauer, 1986; Tapponnier et al., 2001). However, only recently have some evidence of such 44

process been collected. The fact that the continental lithosphere could subduct has first been 45

hypothesized from the seismicity within the continental collision framework of the Pamir, 46

which reveals two subduction zones (Chatelain et al., 1980; Burtman and Molnar, 1993).

47

However, the nature of the lithosphere, continental or oceanic, remained speculative, as well 48

as the slab length. Global P-waves tomography allows to discuss the maximum depth extent 49

of the two distinct slabs (Negredo et al., 2007): the Indian slab deepens northward almost 50

vertically down to ~600 km beneath the Hindu Kush, and the Asian lithosphere is seemingly 51

deepening southward down to ~400 km beneath the Pamir (Fig. 1). The maximum depth of 52

seismicity is compatible with continental lithosphere subduction (Negredo et al., 2007).

53

Recent seismic studies of the Pamir-Hindu Kush region show details of the lithospheric 54

structure of such slabs (Mechie et al., 2012; Kufner et al., 2016), revealing a strong evidence 55

of their continental nature by showing the subduction of light continental lower crust on top 56

of denser continental upper lithosphere down to ~400 km depth (Schneider et al., 2013).

57

Although less documented, the Asian lithosphere in central Tibet is inferred to subduct 58

southward down to 300 km, with no related seismicity (Kind et al., 2002; Replumaz et al., 59

2014). During the early Tibetan collision stage, a first episode of subduction of the Asian 60

lithosphere likely occurred beneath Qiangtang, as recorded by Cenozoic volcanics between 50 61

and 30 Ma in the Qiangtang block (e.g. DeCelles et al., 2002; Goussin et al., submitted) (Fig.

62

1). This slab detached, and sunk, and is possibly revealed by a positive P-wave speed anomaly 63

in the lower mantle (Replumaz et al., 2010) (Fig. 1).

64

Beneath the Himalaya, it has been proposed that at present-day the Indian lower crust, 65

attached to its lithospheric mantle, is bent and is underplated below southern Tibet (Nabelek 66

et al., 2008) (Fig. 1). A possible additional process that may help the subduction of 67

continental plates it eclogitization. During the bending of the Indian plate beneath the 68

Himalaya, its density could be close to mantle density, which favors underplating below 69

southern Tibet (Hetenyi et al., 2007).

70

However, a dynamic explanation of continental subduction is still lacking, mostly because 71

it is in principle less dense than the mantle (in the absence of secondary processes like 72

eclogitization). Deciphering its contribution to the formation of the Tibetan plateau where this 73

under-explored process is possibly important (Mattauer, 1980; Tapponnier et al., 2001;

74

Replumaz et al., 2014; Goussin et al., submitted). But above all, this observation calls for to 75

go beyond the paradigm of slab pull as a unique driver of plate tectonics. Alvarez (2010) 76

already pointed to the fact that protracted collisions require support from mantle tractions;

77

sustained continental subduction is even more difficult to sustain under the premise that slab 78

pull is the prominent, if not unique, tectonic engine.

79

A low-density contrast between the continental lithosphere and the mantle facilitates the 80

subduction of the continental lithosphere attached to a dense oceanic slab (Capitanio et al., 81

2010; Bajolet et al., 2013), but it cannot trigger the subduction of the Asian lithosphere which 82

is not attached to a dense slab (Fig. 1). To date, only two models attempt to quantitatively 83

explain this process. First, a three-dimensional analogue model analyses the subduction 84

dynamics of a light lithosphere, analogue of the Asian plate, below Qiangtang. This model 85

explores the geodynamic contexts wherein such atypical subduction may occur (different 86

density of plates, attached or not to an oceanic plate) without exploring the underlying mantle 87

dynamics (Replumaz et al., 2016). The second is a 3D numerical model of the Hindu Kush–

88

Pamir orogenic system specifically exploring the role of the complex geometry of the regional 89

subduction zone, slab-curvatures in particular (Liao et al., 2017). Here, we present new 90

analogue models of the Asian lithosphere subduction, focusing on the subduction induced 91

mantle circulation active in this geodynamic environment. We opt for a very simple initial 92

geometry, featuring two linear subduction zones of opposite vergences (Fig. 2), similar to the 93

supposed geometry of in Central Tibet in the early collision time (Fig. 1). Our models 94

highlight the deep mantle contribution to the collisional subduction in a simple context of 95

frontal collision, equivalent to the collision between India and Asia. They provide an 96

explanation for the subduction of a light and buoyant continental lithosphere into the mantle 97

in a collision context, emphasizing the role of far field forces instead of subduction related 98

forces, typically slab pull. Most importantly, this process obviously rules out slab pull as the 99

omnipotent driver of plate tectonics and calls back to the more ancient conveyor belt view 100

advocated by Holmes (1929).

101 102

2. Model setup and experimental procedure

103

The model is composed of thin sheets of silicone putty as analogs for the lithospheres, 104

lying on top of low-viscosity glucose syrup that simulates the rheology of the asthenospheric 105

mantle. Silicone putties are visco-elastic materials, quasi-Newtonian at experimental strain 106

rates (Weijermars, 1986) and the glucose syrup a Newtonian fluid. The properties of analog 107

materials are listed in Table 1. The viscosity and density of silicone and syrup are constant 108

and are considered as average effective values.

109

The experimental setup consists in a Plexiglas tank filled with 11 cm of glucose syrup, 110

corresponding to the upper mantle in nature (i.e., 660 km), on top of which three silicone 111

plates rest, and represent the lithosphere (Fig. 2). The indenter plate is made of two adjacent 112

silicones: a silicone unit which is denser than the underlying syrup, that is analogue to the 113

Tethys oceanic lithosphere (referred to as ocean in the following), and an adjacent silicone 114

unit which is conversely less dense than the syrup and is analogue to the Indian continental 115

lithosphere (referred to as indenter). The trailing edge of the low-density silicone is attached 116

to a piston which is advancing forward at constant velocity. At the back of the box, the 117

backwall plate is made of a light silicone which is analogue of northern Tibet, the trailing 118

edge of which is fixed to the back wall of the tank. A third silicone plate, less dense and less 119

viscous than the backwall plate, analogue to the southern Tibet lithosphere, is disposed 120

between the indenting plate and the backwall plate, and is referred to as upper plate.

121

The experiments are scaled for gravity, length, density, viscosity and velocity following 122

previous studies (e.g. Weijermars and Schmeling, 1986; Funiciello et al., 2003). The scaling 123

factor for length is 1.7 x 10^-7 so that 1 mm of the model corresponds to 6 km in nature (for 124

example, the upper mantle thickness is 110 mm in model and 660 km in nature). The density 125

and viscosity ratios between the oceanic silicone putty and the mantle glucose syrup are 1.05 126

and 1.5 x 10³ respectively (Table 1). The piston advances at a rate of 0.54 cm min^-1 which 127

corresponds to a convergence rate of 5.7 cm yr^-1 in nature, mimicking the northward motion 128

of India at a rate in excess of 50 mm/yr (e.g. Patriat and Achache, 1984; Molnar and Stock, 129

2009; van Hinsbergen et al., 2005); 1 minute in experimental time correspond to 0.55 Ma in 130

nature.

131

Each experiment is monitored over its entire duration by top and lateral snapshot views 132

taken at regular time intervals, which enables us to quantify the amount of subducted 133

lithosphere, the thickening of the plates, and the flow field (Fig. 2). The glucose syrup, 134

analogue to the mantle, is preliminary seeded with bright reflecting air micro-bubbles that 135

behave as passive tracers when enlightened from the bottom of the box by a narrow laser 136

beam located along the centerline of the system (Fig. 2). These bubbles have a diameter less 137

than 1 mm, so that their possible influence on the density/viscosity is negligible (Funiciello et 138

al., 2006). Post-processing of side views is made using PIVlab, a time-resolved digital particle 139

image velocimetry tool of Matlab, which provides sparse velocity vectors using the bright 140

passive tracers seeding the mantle (Thielicke and Stamhuis, 2014). The velocity field is 141

measured in 2D, along the vertical center plane of the model and shows the poloidal 142

component of the flow. The toroidal component of the mantle flow off the center plane of 143

measurement is not recorded and is inferred from earlier studies (e.g. Funiciello et al., 2004;

144

Piromallo et al., 2005; Husson et al., 2012a).

145 146

3. Experimental results: mantle flux related to collisional subduction

147

In this paper we present in detail the result of an ad hoc model (Fig. 3), which captures and 148

summarizes the overall behavior of an experimental series of 18 models (15 without micro- 149

bubbles in the syrup, 3 of them published in Replumaz et al. (2016), 3 models done with 150

micro-bubbles), which has been designed with a particular focus on mantle flow during 151

collisional subduction (Fig.2).

152

The evolution of this model follows 3 stages. During the first stage, the oceanic part of the 153

indenter plate subducts under its own weight into the syrup (Fig. 3A). During that stage, the 154

dip of the slab rapidly increases because the trench rolls back at a lower rate than the forward 155

motion of the piston; this triggers regional scale mantle circulation. The mantle flux shows a 156

classic pair of convective cells on both sides of the oceanic slab extending along the entire 157

box depth (Fig. 4A). The convective cell under the indenter plate (cell 1a) has the highest 158

velocity (mean value ~3.5 10^-5 m/s, Fig. 4B), sheared from the top by the forward advancing 159

indenter (uindenter = 5 10^-5 m/s) and further excited on its lateral side by the vertical descent of 160

the oceanic slab (vocean = 7 10^-5 m/s). The convective cell below the upper and backwall plates 161

(cell 1b) has a mean velocity magnitude of ~1.5 10^-5 m/s, more than twice slower, excited 162

along its lateral side by the descent of the dense slab, and to the top by slab roll back (at a 163

slower horizontal velocity, uroll back = 2 10^-5 m/s). During this stage the flow is dominated by 164

the poloidal component, as revealed in our 2D view.

165

The second stage is a transition phase, occurring 15’ after the beginning of the model.

166

During this stage the oceanic slab reaches the bottom of the box while the continental indenter 167

starts to curve and plunge into the syrup (Fig. 3B). The inception of the positively buoyant 168

indenter in the subduction zone reverses the motion of the trench which starts to advance 169

toward the upper plate at a horizontal velocity of uindenter = 3.6 10^-5 m/s. Trench advance of the 170

indenter plate forces the backwall plate to subduct. During this stage the mantle convective 171

pattern reshuffles and becomes more articulated, with 2 pairs of convective cells related to 172

both slabs (Fig. 4B). The pair of convective cells related to the slab belonging to the indenter 173

plate is asymmetric: a wide cell is observed below the indenter similar to the one of the first 174

stage (cell 1a) while a smaller convective cell appears below the upper plate, in the deeper 175

half of the box where the oceanic slab still shows a vertical component of the subduction 176

velocity (cell 1b). The second pair of convective cells surrounds the backwall slab. These 177

asymmetric cells are located only in the shallowest half of the box. The velocity is one order 178

of magnitude lower than in cell 1a (mean value ~5 10^-6 m/s), and their life span is very short.

179

The convective cell underneath the upper plate (cell 2a) responds to the surface drag caused 180

by the shortening of the upper plate starting at this stage, and is driven by trench advance that 181

forces contraction (Fig. 4B). This shallow convective cell generates a downward flow at the 182

margin between the upper and backwall plates. This downward flow could initiate the 183

downward motion of the backwall plate at the interface between the two plates. The cell 184

which is located below the backwall plate (cell 2b) is less developed, as no shortening of the 185

backwall plate occurs at that stage, and only responds to the descent of the backwall plate in 186

the mantle. During this stage the flow is also dominated by the poloidal component.

187

The third stage of model evolution presents a continuous trench advance of the continental 188

indenter slab and a continuous subduction of the backwall plate. The subduction of the 189

indenter enhances slab curvature during the rest of the experiment (Fig. 3C&D). The 190

convective pattern changes radically, with no more pair of cells but a dominant horizontal 191

motion of the mantle, which is dragged from above by the indenter (Fig. 4C&D). It is 192

advancing at a mean horizontal velocity of uindenter = 3.8 10^-5 m/s. The horizontal flow under 193

the upper and backwall plates has a maximum velocity of ~2.5 10^-5 m/s close to the indenter, 194

decreasing gradually toward the back wall (Fig. 4C). The flow underneath the upper and 195

backwall plates is entirely forward in the central section of the box, and is confined in the 2D 196

plane between the advancing indenter slab and back wall of the box. Importantly, this implies 197

that the return flow is entirely toroidal, i.e. sideways from the subducting slab. During the 198

transition, as the indenter advances, the toroidal flow overcomes the poloidal flow observed in 199

the previous stages which was driven by the oceanic subduction. A residual poloidal cell 1a is 200

observed at the bottom of the box, with a downward vertical component of velocity along the 201

oceanic slab (mean value ~5 10^-6 m/s). The convective cell observed below the piston is a 202

boundary effect of our experimental setup that we let aside. During this stage, the slab that 203

belongs to the backwall plate is passively transported (advection) by this horizontal mantle 204

flow, not perturbed by the slab. When the length of the slab increases, the flow is slightly 205

perturbed by the slab (Fig. 4D). The flow is faster upstream of the slab than downstream so 206

that a dynamic mantle push occurred on the backwall slab (Fig. 5). The backwall slab velocity 207

of subduction is constant, uback = 2.3 10^-5 m/s.

208

Between 15’ and 60’ min, the indenter slab reaches ~5 cm depth in the box, while the 209

backwall slab reaches about half of it (~2.5 cm depth with a ~4 cm long slab), corresponding 210

to a mean vertical component of velocity of vback ~ 10^-5 m/s, half that of the indenter. Plate 211

thickening is propagating northward during the experiment, first with the thickening of the 212

indenter plate, followed by thickening of the upper plate (~75% of shortening at the end of the 213

experiment), and finally by that of the backwall plate (for detailed measurement and analysis 214

of the thickening process, see Replumaz et al., 2016).

215 216

4. How is the collisional subduction driven?

217

Despite evidence for subduction of positively buoyant continental lithosphere, inferring the 218

dynamics of such a counter-intuitive process is not clear yet. Indeed, when such a plate 219

subducts, its buoyancy is positive and is not a driving force. The slab buoyancy is 220

proportional to the density contrast (Δρ) between the silicone plate and the glucose syrup (ρ = 221

1428 kg/m³), gravity acceleration (g = 10 m/s²), and slab thickness (h) and length (l) (e.g.

222

Turcotte and Schubert, 1982). Considering a plate lighter than the syrup (ρ = 1402 kg/m³), the 223

slab buoyancy is positive, increasing with slab length. For the backwall slab, the maximum 224

value of the slab buoyancy is ~0.08 N/m (h = 8 mm, l = ~4 cm maximum). This positive 225

buoyancy cannot drive the subduction of the backwall plate, for it is even a resisting force.

226

Another engine has to drive the collisional subduction. In our models, this engine is the 227

push of the piston (Fig. 2). The piston drives the indenter plate at a constant velocity. The 228

push of the piston is transmitted to the indenter plate which shows a high horizontal motion 229

during the entire evolution of the experiment (Fig. 4). This push mimics far-field tectonic 230

forces. When the indenter plunges into the mantle, the trench begins to move forward, 231

transmitting the push of the indenter to the upper plate. First the upper plate absorbs this 232

motion by thickening (stage 2), then it transmits the motion to the backwall plate, which is 233

forced to subduct. Because the indender is forced to move forward at a prescribed rate, the 234

tectonic force adjusts to overcome the resisting forces and sets a force balance that allows for 235

the convergence to occur at the prescribed rate. At the lithospheric scale, the main resisting 236

forces acting on the backwall slab are the bending resistance and the lithosphere/lithosphere 237

interface resistance between the back wall and the upper plates (e.g. Funiciello et al., 2003;

238

Replumaz et al., 2016). In our experiment, the push of the upper plate on the backwall plate 239

hinge instantaneously adjusts to overcome these resisting forces, thereby allowing for bending 240

and plunging in the mantle.

241

The buoyancy force of the indenter slab is also positive, as for the backwall slab, with a 242

maximum value of ~0.17 N/m (ρ = 1402 kg/m³, h = 1.1 cm, l = ~6 cm). But as the indenter 243

slab is attached to the oceanic slab, with a strong negative contrast compare to the mantle 244

(buoyancy force value of -0.6 N/m with ρ = 1508 kg/m³, h = 1.1 cm, l = 7 cm), the cumulative 245

buoyancy for the indenter/oceanic slab remains negative throughout the entire course of the 246

experiment (-0.43 N/m minimum). The negative buoyancy of the oceanic slab, name slab pull 247

force, is sufficient to pull it in the mantle (e.g. Turcotte and Schubert, 1982). Evolving in a 248

self-consistent way with a balance between buoyancy and viscous forces of the entire system, 249

this slab pull also generates trench retreat, due to the roll back of the slab under its own 250

weight (Funiciello et al., 2003). In our experiment, trench retreat is impeded by the forward 251

advance of the piston (Fig. 3). When the indenter reaches the trench, the forward horizontal 252

motion becomes dominant, despite a cumulative negative buoyancy for the indenter/oceanic 253

slab. The tectonic force becomes dominant upon the buoyancy force, which promotes trench 254

advance and the transmission of the tectonic force to the upper and backwall plates. They 255

provide an explanation for the subduction of a light and buoyant continental lithosphere into 256

the mantle in a collision context, emphasizing the role of far field forces instead of subduction 257

related forces, slab pull in particular.

258 259

5. Comparison with natural system

260

Recently, mounting evidence for subduction of positively buoyant continental lithosphere 261

is available. For example, seismic profiles image the subduction of light continental lower 262

crust on top of denser continental upper lithosphere down to ~400 km depth in Pamir 263

(Schneider et al., 2013). At lithospheric scale, partial dynamic explanation of such process 264

exists, as the densification of the Indian lower crust by eclogitization during its bending 265

beneath the Himalaya, leading to a low density contrast between the continental lithosphere 266

and the mantle (Hetenyi et al., 2007). At the mantle scale, simple models have been proposed, 267

showing for example that such a low density contrast between the continental lithosphere and 268

the mantle facilitates the subduction of the Indian continental lithosphere attached to a dense 269

oceanic slab (Capitanio et al., 2010; Bajolet et al., 2013), even after partial or complete 270

breakoff between the Indian plate and the oceanic plate (Capitanio and Replumaz, 2015). Yet, 271

it remains insufficient to trigger the subduction of the Asian continental lithosphere, which is 272

not attached to a dense slab (Fig. 1). Recent 3D numerical geodynamic models of the Hindu 273

Kush–Pamir orogenic system show that the Asian lower crust is also able to subduct not 274

attached to a dense slab, but in a very peculiar context with a complex initial subduction-zone 275

geometry corresponding to this region, to explain the curvature of the Pamir slab (Liao et al., 276

2017).

277

Our analogue experiments consist of a simplified initial geometry of two linear subduction 278

zones of opposite vergences (Fig. 2), similar to the supposed setting of the lithospheres in 279

Central Tibet in the early collision time (Fig. 1). They show that the subduction of continental 280

lithosphere lighter than the mantle is dynamically sustainable, provided that an additional 281

contribution forces the system to subduct. In our experiments, far field forces are responsible 282

for sustained trench advance during collision, and subsequently force the subduction of 283

continental units into the mantle (Fig. 5). This forced subduction takes advantages of weaker 284

zones, such as suture zones, which in our experiments, are materialized by a discontinuity in 285

the upper plate system. This setting resembles the case of the Tibetan plateau, which is made 286

of several blocks separated by different suture zones (Fig. 1). The differential thickening that 287

occurs between 2 plates of different viscosities, generates shallow lithospheric convective 288

cells, which in turns favor the initiation of the subduction of the more viscous back plate (Fig.

289

4B). Such rheological contrast is observed in Tibet, where the South Tibetan lithosphere is 290

thought to be exposed to a higher heat flux and more abundant Cenozoic magmatism (e.g.

291

Chung et al., 2005; Kapp et al., 2005; Ding et al., 2007), and is therefore possibly weaker.

292

In the analogue experiment, the piston produces a constant kinematic boundary condition 293

which in principle does not resemble any dynamic natural condition. At the scale of the 294

collision zone, this setting mimics the far field forces that sustain the northward motion of 295

India. Here, the enigmatic label far field simply refers to the forces that contribute to plate 296

tectonics but that originate remotely from plate boundaries. They are not many in the plate 297

dynamics budget (e.g., Forsyth and Uyeda, 1975), and essentially refer to mantle drag from 298

the underlying mantle convection. Active mantle drag is essential to drive plates against or 299

away from each other, and provides a dynamic explanation to the observed kinematics of 300

plate boundaries (Husson, 2012). It may contribute to propel subducting plates (Alvarez, 301

2010), as it is the case for India-Eurasia (e.g. Becker and Faccenna, 2011), but is equally 302

efficient in pushing the upper plate towards a convergent zone, as it is for example the case 303

for the South America-Nazca convergent system (Husson et al., 2012b). Alternative models to 304

active mantle drag also explain the fast convergence of India with the support from forces 305

arising from the South, by the inception of the slab into the lower mantle that would excite a 306

larger convection cell (Faccenna et al., 2013), by reinforced push from the Reunion plume 307

(van Hinsbergen et al., 2005), or by a weaker, lubricating asthenosphere that restrain the 308

resistance to the northward motion of India (Cande and Stegman et al., 2011).

309

The force balance at the plate boundary constantly evolve and should modulate plate 310

convergence at all times (e.g. Iaffaldano et al., 2011; Clark, 2012), but regardless the 311

northward motion of India continued at a rate in excess of 50 mm/yr since the beginning of 312

the collision (e.g. Patriat and Achache, 1984; Molnar and Stock, 2009; van Hinsbergen et al., 313

2005), which supports the hypothesis of a primordial role for far-field forces. The explanation 314

is to be found at a larger scale. The Indian collision is embedded in a much wider convergent 315

system, namely the Tethyan subduction zone, from the Mediterranean to the Banda arc, 316

driving the dynamics of the global system and constraining the force balance at the boundary 317

between India and Eurasia. This makes the assumption of constant convergence rate during 318

the collision an acceptable simplification. Indeed, our model adequately reproduces the 319

horizontal motion of the indenter, analogue of the motion of India.

320 321

6. Conclusion

322

We explore the subduction dynamics of the continental lithosphere, a process which is 323

increasingly documented below Tibet. Of course, the positive buoyancy of such a slab is ruled 324

out as the driving force because it refrains subduction. In our experiments, it is the horizontal 325

push of the piston, whose force is transmitted first to the indenter, and second to the upper 326

plate, which is the main driving force that we call the tectonic force. At the lithospheric scale, 327

the horizontal tectonic force is higher than the bending of the backwall plate and the 328

lithosphere/lithosphere interface resistances, allowing for its bending. At the mantle scale, the 329

push of the mantle flow on the backwall slab overcomes the positive buoyancy of the slab, is 330

higher than the resistance at the lithosphere/asthenosphere interface, and allows for a 331

dynamically sustainable subduction (Fig. 5).

332

Applied to the convergence between India and Eurasia, the constant push of the piston is 333

an acceptable kinematic analogue of the far field forces generated by the mantle convection 334

cell underneath India, fueled by the African superplume. We show that the subduction of the 335

Asian lithosphere in front of the advancing Indian plate is driven by the viscous strength of 336

the lithosphere which allows the transmission of tectonic forces remotely from the collision 337

front, and forced the Asian lithosphere to plunge in the mantle. Such forced subduction by 338

external far field forces occurs along a suture, a pre-existing discontinuity in the upper plate.

339

We emphasize that far field forces, here active mantle drag, are fundamental to explain the 340

protracted motion of India towards Eurasia, instead of subduction related forces, slab pull in 341

particular. Likewise, we contend that similar dynamics likely apply to other convergent 342

systems.

343 344

Acknowledgments

345

This work has been supported by a grant from LabEx OSUG@2020 (Laboratoires 346

d’Excellence, Observatoire des Sciences de l’Univers de Grenoble; Investissements d’avenir–

347

ANR10 LABX56) and the ANR DSP-Tibet (Agence Nationale de la Recherche de France, 348

Probing Deep and Surface Processes in Central Tibet: ANR-13-BS06-0012-01).

349

350 351 352

References

353

Alvarez, W., 2010. Protracted continental collisions argue for continental plates driven by 354

basal traction, Earth and Planetary Science Letters, 296, 434–442.

355

Bajolet, F., Replumaz, A., and Lainé, R., 2013, Orocline and syntaxes formation during 356

subduction and collision: Tectonics, v. 32, p. 1529–1546, doi:10.1002/tect.20087.

357

Becker, T.W., and Faccenna, C., 2011, Mantle conveyor beneath the Tethyan collisional belt:

358

Earth and Planetary Science Letters, v. 310, p. 453–461, doi:10.1016/j.epsl.2011.08.021.

359

Burtman, V.S. and Molnar, P., 1993. Geological and geophysical evidence for deep 360

subduction of continental crust beneath the Pamir. Spec.Pap.-Geol. Soc. Am. 281, p76 361

Cande, S. C., and D. R. Stegman, 2011, Indian and African plate motions driven by the push 362

force of the Reunion plume head, Nature, 475, p. 47-52, DOI: 10.1038/nature10174 363

Capitanio, F.A., Morra, G., Goes, S., Weinberg, R.F., Moresi, L., 2010. India-Asia 364

convergence driven by the subduction of the Greater Indian continent. Nature Geoscience, 365

3, 136–139, doi: 10.1038/NGEO725.

366

Capitanio, F.A., Replumaz, A., and Riel, N., 2015, Reconciling subduction dynamics during 367

Tethys closure with large-scale Asian tectonics: Insights from numerical modeling:

368

Geochemistry, Geophysics, Geosystems, v. 16, p. 962–982, doi:10.1002/2014GC005660.

369

Chatelain, J.L., Roecker, S.W., Hatzfeld, D., Molnar, P., 1980. Microearthquake seismicity 370

and fault plane solutions in the HinduKush region and their tectonic implications. Journal 371

of Geophysical Research, 85, 1365–1387.

372

Chung, S.-L. et al. Tibetan tectonic evolution inferred from spatial and temporal variations in 373

post-collisional magmatism. Earth-Sci. Rev. 68, 173–196 (2005).

374

Clark, M. K., 2012, Continental collision slowing due to viscous mantle lithosphere rather 375

than topography, Nature, 483, doi:10.1038/nature10848 376

Davy, P., and P. R. Cobbold, 1991, Experiments on shortening of a 4-layer continental 377

Lithosphere, Tectonophysics, 188, 1– 25.

378

DeCelles, P.G., Robinson, D.M., and Zandt, G., 2002, Implications of shortening in the 379

Himalayan fold-thrust belt for uplift of the Tibetan Plateau: Tectonics, v. 21, 1062, 380

doi:10.1029/2001TC001322.

381

Ding, L., Kapp, P., Yue, Y. & Lai, Q., 2007. Postcollisional calc-alkaline lavas and xenoliths 382

from the southern Qiangtang terrane, central Tibet, Earth and Planet Science Letters, 254, 383

28-38. E. R.

384

Faccenna, C., Becker, T.W., Conrad, C.P., and Husson, L., 2013, Mountain building and 385

mantle dynamics: Tectonics, v. 32, p. 80–93, doi:10.1029/2012TC003176.

386

Forsyth, D. & S. Uyeda, 1975, On the relative importance of the driving forces of plate 387

motion, Geophys. J. R. Astron. Soc., 43, p. 163–200 388

Funiciello, F., C. Faccenna, D. Giardini, and K. Regenauer-Lieb (2003), Dynamics of 389

retreating slabs: 2. Insights from three-dimensional laboratory experiments, Journal of 390

Geophysical Research, 108(B4), 2207, doi:10.1029/2001jb000896.

391

Funiciello, F., Faccenna, C., Giardini, D., 2004. Role of lateral mantle flow in the evolution of 392

subduction system: insights from 3-D laboratory experiments. Geophys. J. Int., 157, 1393–

393

1406.

394

Funiciello, F., Moroni, M., Piromallo, C., Faccenna, C., Cenedese, A., Bui, H.A., 2006, 395

Mapping the flow during retreating subduction: laboratory models analyzed by feature 396

tracking. J. Geophys. Res. 111, B03402 03410.01029/02005JB003792.

397

Goussin, F., C. Cordier, S. Guillot, P. Boulvais, P. Roperch, A. Replumaz, K. Schulmann, G.

398

Dupont-Nivet, Z. Guo, Metasomatized mantle and growth of the Tibet Plateau, submitted 399

Nature Geoscience 400

Hetenyi, G., Cattin, R., Brunet, F., Bollinger, L., Vergne, J., Nabelek, J. and Diament, M., 401

2007. Density distribution of the India plate beneath the Tibetan plateau: Geophysical and 402

petrological constraints on the kinetics of lower-crustal eclogitization. Earth and Planetary 403

Science Letters, 264, doi: 10.1016/j.epst.2007.09.036 404

Holmes A., Radioactivity and earth movements, 1929, Transactions of the Geological Society 405

of Glasgow, 18, p. 559–606.

406

Husson, L., B. Guillaume, F. Funiciello, C. Faccenna, L.H. Royden, 2012a, Unravelling 407

topography around subduction zones from laboratory models, Tectonophysics, 526, 5-15, 408

DOI: 10.1016/j.tecto.2011.09.001 409

Husson, L., C.P. Conrad, C. Faccenna, 2012b, Plate motions, Andean orogeny, and volcanism 410

above the South Atlantic convection cell, Earth and Planetary Science Letters 317, 126- 411

135, DOI: 10.1016/j.epsl.2011.11.040 412

Husson, L., 2012, Trench migration and upper plate strain over a convecting mantle, Physics 413

of the Earth and Planetary Interiors, 212, 32-43, doi:10.1016/j.pepi.2012.09.006 414

Iaffaldano, G., Husson, L., Bunge, H.-P., 2011. Monsoon speeds up Indian plate motion.

415

Earth and Planetary Science Letters 304, 503–510, doi:10.1016/j.epsl.2011.02.026 416

Kapp, P., Yin, A., Harrison, T. M. & Ding, L. Cretaceous-Tertiary shortening, basin 417

development, and volcanism in central Tibet. Geol. Soc. Am. Bull. 117, 865–878 (2005).

418

Kind, R., et al., 2002, Seismic images of crust and upper mantle beneath Tibet: Evidence for 419

Eurasian plate subduction: Science, v. 298, p. 1219–1221, doi:10.1126/science.1078115.

420

Kufner, S.-K., et al., 2016, Deep India meets deep Asia: Lithospheric indentation, 421

delamination and break-off under Pamir and Hindu Kush (Central Asia), Earth and 422

Planetary Science Letters, v. 435, p. 171–184, doi:10.1016/j.epsl.2015.11.046.

423

Liao, J., T. Gerya, M. Thielmann, A. G. Webb, S.-K. Kufner, A. Yin, 2017, 3D geodynamic 424

models of continental subduction with opposite polarities in the Hindu Kush–Pamir 425

orogenic system, northwestern Tibetan Plateau, Earth and Planetary Science Letters, 480, 426

p. 133-146, DOI: 10.1016/j.epsl.2017.10.005 427

Mattauer, M., 1986, Intracontinental subductions in tertiary mountains belts of Asia - 428

relationships with strike-slip faults, Bulletin de la Societe Geologique de France, 2, p. 143- 429

157.

430

Mechie, J., Yuan X., Schurr B., Schneider F., Sippl C., Ratschbacher L., Minaev V., Gadoev 431

M., Oimahmadov I., Abdybachaev U., Moldobekov B., Orunbaev S. and Negmatullaev S., 432

2012. Crustal and uppermost mantle velocity structure along a profile across the Pamir and 433

southern Tien Shan as derived from project TIPAGE wide-angle seismic data. Geophysical 434

Journal International, 188, 385-407, doi: 10.1111/j.1365-246X.2011.05278.x 435

Molnar, P., and J. M. Stock (2009), Slowing of India’s convergence with Eurasia since 20 Ma 436

and its implications for Tibetan mantle dynamics, Tectonics, 28, TC3001, 437

doi:10.1029/2008TC002271.

438

Nábĕlek, J., Hetényi G., Vergne J., Sapkota S., Kafle B., Mei Jiang, Heping Su, John Chen, 439

Bor-Shouh Huang and the Hi-CLIMB Team, 2009. Underplating in the Himalaya-Tibet 440

Collision Zone Revealed by the Hi-CLIMB Experiment. Science, 325, 441

doi:10.1126/science.1167719 442

Negredo, A.M., Replumaz, A., Villasenor, A., and Guillot, S., 2007, Modeling the evolution 443

of continental subduction processes in the Pamir-Hindu Kush region: Earth and Planetary 444

Science Letters, v. 259, p. 212–225, doi:10.1016/j.epsl.2007.04.043.

445

Patriat, Ph., Achache, J., 1984. India–Eurasia collision chronology has implications for crustal 446

shortening and driving mechanisms of plates. Nature 311, 615–621.

447

Piromallo, C., T. W. Becker, F. Funiciello, C. Faccenna, 2005, Three-dimensional 448

instantaneous mantle flow induced by subduction, Geophysical Research Letters, 33, DOI:

449

10.1029/2005GL025390 450

Replumaz, A. Negredo A.M., Guillot S., van der Beek P. and Villaseñor A., Crustal mass 451

budget and recycling during the India/Asia collision, Tectonophysics, 492, 99-107, doi:

452

10.1016/j.tecto.2009.10.007, 2010 453

Replumaz, A., F. A. Capitanio, S. Guillot, A. M. Negredo, and A. Villaseñor, The coupling of 454

Indian subduction and Asian continental tectonics, Invited Focus Review, Gondwana 455

Research , 10.1016/j.gr.2014.04.003, 2014 456

Replumaz, A., F. Funiciello, R. Reitano, C. Faccenna, M. Balon, Double subduction of 457

continental lithosphere, a key to form wide plateau, Geology, DOI:10.1130/G38276.1, 458

2016 459

Schneider F.M., X. Yuan, B. Schurr, J. Mechie, C. Sippl, C. Haberland, V. Minaev, I.

460

Oimahmadov, M. Gadoev, N. Radjabov, U. Abdybachaev, S. Orunbaev, S. Negmatullaev, 461

2013, Seismic imaging of subducting continental lower crust beneath the Pamir, Earth and 462

Planetary Science Letters, 375, 101–112, DOI:10.1016/j.epsl.2013.05.015 463

Tapponnier, P., Xu, Z.Q., Roger, F., Meyer, B., Arnaud, N., Wittlinger, G. and Yang, J. S., 464

2001. Oblique stepwise rise and growth of the Tibet plateau. Science, 294, 1671-1677.

465

Thielicke, W. and Stamhuis, E.J. (2014): PIVlab – Towards User-friendly, Affordable and 466

Accurate Digital Particle Image Velocimetry in MATLAB. Journal of Open Research 467

Software 2(1):e30, DOI: http://dx.doi.org/10.5334/jors.bl 468

Turcotte, D. L., and G. Schubert (1982), Geodynamics Application of Continuum Physics to 469

Geological Problems, John Wiley, New York.

470

van Hinsbergen, D.J.J., Steinberger, B., Doubrovine, P.V., and Gassmoeller, R., 2005, 471

Acceleration and deceleration of India-Asia convergence since the Cretaceous: Roles of 472

mantle plumes and continental collision: Journal of Geophysical Research, v. 116, B06101, 473

doi:10.1029/2010jb008051.

474

Weijermars, R. (1986), Flow behavior and physical-chemistry of bouncing putties and related 475

polymers in view of tectonic laboratory applications, Tectonophysics, 124(3-4), 325-358, 476

doi:10.1016/0040-1951(86)90208-8.

477

Weijermars, R., Schmeling, H., 1986. Scaling of newtonian and non newtonian fluid 478

dynamics without inertia for quantitative modelling of rock flow due to gravity (including 479

the concept of rheological similarity). Phys. Earth Planet. Inter. 43, 316–330.

480 481

Indenter Ocean Upper plate Back wall plate Mantle

Exp. 1

20·25·1.1 cm³ 20·7·1.1 cm³ 20·10·0.8 cm³ 20·20·0.8 cm³ 45·55·11 cm³

ρ = 1402 kg/m³

µ = 2 × 10⁵ Pa∙s

1508

3.3 × 10⁴

1397

2.9 × 10⁴

1402

2 × 10⁵

1428

30

Dimensionless parameters Equivalence model-nature

L° Characteristic Length:

Lmodel/Lnature

1.7 × 10⁻⁷

t° Characteristic time:

(tmodel/tnature) = (Δρgh)lith nature / (Δρgh)lith model × (ηlmodel / ηlnature)

1 minmodel → 0.55 Mynature

U° Characteristic velocity:

(Umodel/Unature) = tnature / tmodel × Lmodel /Lnature

1 cm h⁻¹model → 0.18 cm y⁻¹nature

482

Table 1: Materials properties and experiments scaling (see also Bajolet et al., 2013;

483

Replumaz et al., 2016).

484 485

486

Figure 1:

487

A/ Topographic map of the collision zone between India and Asia. The Indian crust is 488

separated from the Asian crust by the Tsangpo suture (TS, in purple). In central Tibet, 489

Cenozoic volcanic rocks between 50 and 30 Ma (black dots) are observed north of the 490

Qiangtang block (Qg), located south of the Jinsha suture (JS, in orange). KS: Kunlun suture, 491

Pa: Pamir, HK: Hindu Kush.

492

B/ Three-dimensional schematic view of the lithospheric structure (middle-lower crust and 493

lithospheric mantle) below Tibet in the early collision time. The Indian lithosphere (IN) 494

subducted northward facing the Asian slab (AS) beneath Qiangtang (Qg). Both slabs detached 495

from the continents and have sunk down to the lower mantle, as inferred from global 496

tomographic cross section (Replumaz et al., 2010, 2014). The same color code used for the 497

different plates is applied to experimental data.

498

C/ Three-dimensional schematic view of the lithospheric structure at present-day, showing 499

to the southwest, the Indian lithosphere (in purple) subducting northward beneath the Hindu 500

Kush (HK) down to the transition zone (~600 km), facing the Asian lithosphere (in green) 501

subducting beneath Pamir down to 300–400 km. In central Tibet, the Asian lithosphere (Ti, in 502

green) is subducting southward down to 300 km along the Kunlun suture (KS), facing the 503

Indian craton (CR) underplating south Tibet.

504

505

Figure 2 : Experimental setup, three-dimensional view of the initial stage, and top and side 506

views during experiment.

507

508

Figure 3: Side view photos of experiment 1, showing successive position of the ocean 509

(underlined in black), indenter (in pink), upper and back wall (in green) slabs. Trench rolls 510

back of the ocean during the first stage is shown by a black arrow, trench advance of the 511

indenter during the rest of the experiment is shown by a pink arrow. Slab lengths at 60min 512

shown by arrow along the slabs. Bright dots are air micro-bubbles used as passive tracers, 513

lighted from below.

514 515

516

4A 517

518

4B 519

Figure 4:

520

A/ Poloidal (2D in the vertical plane) velocity field of the glucose syrup, showing 521

convective cells (white number). Images of the micro-bubbles are recorded by the lateral 522

camera and post-processed to obtain a time-resolved digital particle image velocimetry of the 523

micro-bubbles, providing sparse velocity vectors of the mantle flow (black arrow), not 524

available for the plates (red cross).

525

B/ velocity field (black arrow), velocity magnitude (color scale), and flow streamlines 526

(white isocontours).

527 528

529 530

Figure 5: schematic view of mantle flux related to collisional subduction in the middle of 531

the experimental box.

532