Lecture 13

Layered Convection

Layered Convection

-  Mantle convection results in chemical differentiation

-  This depends on the degree of partial melting, which in turn depends on temperature, composition and water content

-  Process of chemical differentiation in upper mantle well understood

-  Presence of distinct chemical reservoirs separated for more than a billion years

-  Their exact locations and mechanisms of survival poorly understood

MORB vs OIB -  base of lower mantle that is a slab graveyard -  delamination of continental lithosphere -  layered mantle convection

Layered Mantle Convection

-  jump in P-wave velocity at 660 km -  descending slabs sometimes stopped at 660 -  geochemical studies: - abundance of 40Ar

- 3He/4He higher than ~10 times the atmospheric value

Vol. 7, No. 4 April 1997

GSA TODAYA Publication of the Geological Society of America

ABSTRACT

Two new global high-resolutionmodels of the P-wave and S-waveseismic structure of the mantle werederived independently using differentinversion techniques and different datasets, but they show excellent correlationfor many large-scale as well as smallerscale structures throughout the lowermantle. The two models show thathigh-velocity anomalies in the lowermantle are dominated by long linearfeatures that can be associated with thesites of ancient subduction. The imagessuggest that most subduction-relatedmantle flow continues well into thelower mantle and that slabs may ulti-mately reach the core-mantle boundary.The models are available from anony-mous ftp at maestro.geo.utexas.edu indirectory pub/grand and at brolga.mit.edu in directory pub/GSAtoday.

INTRODUCTION

Since forming about 4.5 Ga, planetEarth has been cooling by means of rela-tively vigorous convection in its interiorand by conductive heat loss across thecold thermal boundary layer at the topof the mantle (mainly the oceanic litho-sphere). The primary force driving convec-tion is the downward pull of gravity onthe cold, dense lithosphere resulting indownwellings of slabs of subducted litho-sphere. Understanding the nature of the

Figure 1. Cross sections of mantle P-wave (A) and S-wave (B) velocity variations along a sectionthrough the southern United States. The endpoints of the section are 30.1°N, 117.1°W and 30.2°N,56.4°W. The images show variations in seismic velocity relative to the global mean at depths from thesurface to the core-mantle boundary. Blues indicate faster than average and reds slower than averageseismic velocity. The large tabular blue anomaly that crosses the entire lower mantle is probably thedescending Farallon plate that subducted over the past ~100 m.y. Differences in structure between thetwo models in the transition zone (400 to 660 km depth) and at the base of the mantle are probablydue to different data sampling in the two studies.

A

BFARALLON SLAB

2700 km depth

2770 km depth

Global Seismic Tomography: A Snapshot of Convection in the Earth Stephen P. Grand, Department of Geological Sciences, University of Texas,

Austin, TX 78712Rob D. van der Hilst, Department of Earth and Planetary Sciences, Massachussetts Institute of Technology, Cambridge, MA 02139Sri Widiyantoro, Research School of Earth Sciences, Australian National University, Canberra, ACT 0200, Australia

Tomography continued on p. 2

120°W 100°W 80°W 60°W 40°W

40°N

20°N

1997AnnualMeetingCall for Papers

page 13

Electronic Abstracts Submission

page 14

Registration Issue June GSA Today

ORZ w

Mrmdz+

Z 0

wrwdz =

Z S0

Mrcdz =

Z M+H

Mrcdz+

Z 0

M+Hrwdz f or oceans

dedt

= A0t2exp✓�

H(1� t/tp)2

RT

◆

✓dPdT

◆

eq=

S1 �S2

V1 �V2

∂r∂t

+— · (rv) = 0

∂T∂t

+ v ·—T = k—2T +H

rCp

∂v∂t

+(v ·—)v =1r

— ·s+g

si j =�Pdi j +hei j

Ph ⌘G(Dr

r )

argh

2

Rayleigh number : vigour of convection; Ra ~ 106 Critical Rayleigh number : point at which convection

initiates; Rac ~ 103

ORZ w

Mrmdz+

Z 0

wrwdz =

Z S0

Mrcdz =

Z M+H

Mrcdz+

Z 0

M+Hrwdz f or oceans

dedt

= A0t2exp✓�

H(1� t/tp)2

RT

◆

✓dPdT

◆

eq=

S1 �S2

V1 �V2

∂r∂t

+— · (rv) = 0

∂T∂t

+ v ·—T = k—2T +H

rCp

∂v∂t

+(v ·—)v =1r

— ·s+g

si j =�Pdi j +hei j

Ph ⌘G(Dr

r )

argh

Ra =graDT d3

kh

2

Enhanced for: - 3D convection - higher Rayleigh number

Nusselt number : ratio of heat transported via convection to heat transported via conduction

∂T

∂z

=∂∂z

T0er f

✓z

2p

kt

◆�=

T0ppkt

e

�z

2/4kt

∂T

∂t

=k

rC

p

1r

2∂∂r

✓r

2 ∂T

∂r

◆+

A

rC

p

T =A

6k

(a2 � r

2)

�k

dT

dr

=Ar

3

T =�Q

b

b

2

k

✓1a

� 1r

◆

0 =k

r

2∂∂r

✓r

2 ∂T

∂r

◆+A

T = T

a

✓z

L

+•

Ân=1

2np

sin

✓npz

L

◆exp

✓�n

2p2kt

L

2

◆◆

Re =rud

h⇠ 10�19 �10�21

Nu =Qd

kDT

⇠ 10

4

10

EPS 122: Lecture 20 – Convection

the mantle

Phase changes at 400 and 660 km

increasing Rayleigh number

� mantle becomes stratified

avalanches of material into lower mantle

The many views of

Yuen

EPS 122: Lecture 20 – Convection

What about the plates?

One layer or two?

Subduction �?� Downwelling

Ridges �?� Upwelling

Mantle avalanches (Tackley, 1993)

Transition zone water filter model

410

660

Source: Bercovici & Karato, Nature (2003)

Whole-mantle convection and thetransition-zone water filterDavid Bercovici & Shun-ichiro Karato

Department of Geology and Geophysics, Yale University, PO Box 208109, New Haven, Connecticut 06520-8109, USA

...........................................................................................................................................................................................................................

Because of their distinct chemical signatures, ocean-island and mid-ocean-ridge basalts are traditionally inferred to arise fromseparate, isolated reservoirs in the Earth’s mantle. Such mantle reservoir models, however, typically satisfy geochemicalconstraints, but not geophysical observations. Here we propose an alternative hypothesis that, rather than being divided intoisolated reservoirs, the mantle is filtered at the 410-km-deep discontinuity. We propose that, as the ascending ambient mantle(forced up by the downward flux of subducting slabs) rises out of the high-water-solubility transition zone (between the 660 kmand 410 km discontinuities) into the low-solubility upper mantle above 410 km, it undergoes dehydration-induced partial meltingthat filters out incompatible elements. The filtered, dry and depleted solid phase continues to rise to become the sourcematerial formid-ocean-ridge basalts. The wet, enriched melt residue may be denser than the surrounding solid and accordingly trapped at the410 km boundary until slab entrainment returns it to the deeper mantle. The filter could be suppressed for both mantle plumes(which therefore generate wetter andmore enriched ocean-island basalts) as well as the hotter Archaeanmantle (thereby allowingfor early production of enriched continental crust). We propose that the transition-zone water-filter model can explain manygeochemical observations while avoiding the major pitfalls of invoking isolated mantle reservoirs.

The two main sites of mantle upwelling in the Earth aremid-ocean ridges, such as the East Pacific Rise, whichdraw mantle materials up from shallow depths, andintraplate hotspots, or ocean islands, such as Hawaii,which are possibly due to deep-seated, upwelling plumes.

The lavas sampled from these two environments, mid-ocean-ridgebasalts (MORB) and ocean-island basalts (OIB), respectively, are ingeneral chemically distinct; that is, MORBs are depleted in incom-patible elements including uranium and thorium, while OIBs arerelatively enriched in these elements. This suggests that MORB andOIB originate from different source regions.

The terrestrial heat flow also suggests the presence of differentsource regions. In particular, the approximately 36 TW of heatoutput from the mantle cannot be produced by a mantle entirelycomposed of MORB source materials, because it has so fewradioactive heat sources; the remaining heat sources are thusthought to be hidden at greater depths in isolated reservoirsunsampled by mid-ocean ridges, but possibly sampled by plumes.In total, these and other observations suggest that the mantle isseparated into at least two chemically distinct reservoirs, onedepleted by melting processes at ridges, the other relatively enrichedand untapped.

Although for many years the 660-km-deep discontinuity wasbelieved to be the barrier between these reservoirs, recent tomo-graphic studies suggested that subducting slabs cross this boundary1,2

and sink possibly as far as the core–mantle boundary3. Othermechanisms for retaining distinct reservoirs have been proposed,including poor mixing (the “plum pudding” model4) and placingthe boundary between reservoirs near the bottom of the mantle atthe D 00 layer5,6. One of the most recent mantle-layering models7

places the barrier between reservoirs at about 1,600 km depth; theadvantages of this model as opposed to others are discussed in ref. 7.However, whereas the 1,600-km barrier is inferred by the depth atwhich slabs lose much of their tomographic signature (and are thuspresumably stopped) the boundary is not observed in global seismicvelocity models8,9 against which tomographic models are treated asperturbations. This lack of evidence has been attributed to largeundulations in the boundary (due to impinging slabs and con-vective “lava-lamp” type motion10), causing an effectively broad,diffuse and hence undetectable boundary. Lack of observation of

these undulations in tomography are attributed to the somewhatfortuitous cancellation of chemical and thermal effects in causingseismic velocity anomalies11.Layered-mantle models also suffer from dynamical inconsisten-

cies, regardless of whether the boundary between the layers is at660 km, 1,600 km, or 2,700 km depth. Perhaps most importantly,because any enriched lower layer has most of the heat-producingelements, the depleted overlying layer is heated almost entirelyalong its base11. Basally heated convective systems are characterizedby upwelling and downwelling currents with comparable heattransport. However, various analyses of mantle plume heat flowsuggest that they account for less than 10%of the net heat flux out ofthe mantle12, the remainder being mostly transported by slabs. Ingeneral, evidence for slab-dominated convective circulation is verydifficult to reconcile with layered-convection models that have adepleted layer basally heated by an underlying enriched one11.Although the conflict between geochemical and geophysical

constraints has engendered great activity in the mantle dynamicscommunity, the problem remains unresolved, and to some extent—given the ongoing appeal to mantle layering—is no further alongthan 20 years ago. We propose an alternative hypothesis that avoidsthe requirement for distinct chemical reservoirs and mantle layer-ing, and explains the observations through a mechanism we termthe “transition-zone water filter” (Fig. 1).

Water in the transition zoneThe mantle transition zone is the region between the olivine–wadsleyite phase boundary at a depth of 410 km, and the ring-woodite–perovskite/magnesiowustite transition at 660 km depth,with various other phase transitions in between13. Transition-zoneminerals have unusual properties relative to upper-mantle minerals,including water solubility14 and diffusivity of various atomic andelectronic species15,16. In particular, high-pressure mineral-physicsstudies have shown that transition-zone minerals at average mantletemperatures have anomalously high water solubility, on the orderof 1 wt% (and as much as 3 wt%) whereas the solubility of water inupper- and lower-mantle minerals is less than 0.1–0.2 wt% (refs 17–19). The actual water content in the mantle can be inferred frompetrological observations. In particular, volcanic rocks indicate(given estimates for their degree of melting) that the MORB source

hypothesis

NATURE |VOL 425 | 4 SEPTEMBER 2003 | www.nature.com/nature 39© 2003 Nature Publishing Group

Discussion paper 5