CRUSTAL SEISMOLOGY HELPS CONSTRAIN THE NATURE OF MANTLE MELTING ANOMALIES: THE GALAPAGOS VOLCANIC...

CRUSTAL SEISMOLOGY HELPS CONSTRAIN THE NATURE OF MANTLE MELTING

ANOMALIES: THE GALAPAGOS VOLCANIC PROVINCE

AGU Chapman ConferenceFt. William, Scotland, 31/08/2005

V. Sallarès (1), Ph. Charvis (1), E. Flueh (2), J. Bialas (2)

(1) IRD-Géosciences Azur, Villefranche-sur-mer, France

(2) IFM-GEOMAR, Kiel, Germany

STUDY AREA

2O Ma

15 Ma

12 Ma

Projects:

0 Ma

SALIERI-2001

IRD-GéoAzur IFM-GEOMAR

IFM-GEOMAR IRD-GéoAzur

PAGANINI-1999

G-PRIME-2000

WHOI U. Hawaii

Objectives

• To determine the velocity structure and crustal thickness of the GVP-volcanic ridges & estimate their uncertainty Joint refraction/reflection travel time tomography Monte Carlo-type analysis

OBJECTIVES

• To determine upper mantle density structure based on velocity-derived models

Gravity and topography analysis

• To connect seismic parameters (H, Vp) with mantle melting parameters (e.g. Tp, damp melting, composition) Mantle melting model

• To contrast model predictions with other observations Geochemistry, temperature, mantle tomography…

Cocos

Carnegie

20 Ma

Cocos

Carnegie

RESULTS

~19 km

~19 kmVeloc. Grad.

3-4 km

Cocos

Carnegie

15 Ma

RESULTS

~18.5 km

Cocos

12 Ma

Carnegie

RESULTS

~16.5 km

^^

~13 km

G-PRIME-2000

<Vp, L3>~7.10-7.15 km/s

h~6 km

RESULTS

Overall H-Vp anticorrelation

Cocos

Carnegie

Cocos

CarnegieGHS

RESULTS

Mantle? Gravity and topography analysis

Cocos

Carnegie

Cocos

CarnegieGHS

RESULTS

Mantle? Gravity and topography analysis

Cocos

Carnegie

Cocos

CarnegieGHS

RESULTS

Mantle? Gravity and topography analysis

)()()()()(

)(xhxhZ

xhxxhx

cw

cmcwmwm

−−ΔΔ+ΔΔ=Δ ρρρ

Airy+Pratt+Crustal dens. correction:

Crustal structure Nature of the anomaly

MANTLE MELTING MODEL

Crustal thickness, Vp [Tp, active upwelling (x=w/u0), composition]

● 2-D steady-state model for mantle corner flow (Forsyth, 1993)

● Include deep damp melting (Braun et al., 2000)

● Active upwelling confined to beneath the dry solidus (Ito et al., 1999)

)()(),(),( 0 zzuzFzxwzxm χΓ=∂∂=&

MANTLE MELTING MODEL

∫∫==Rc

m

c

m dxdzzxmuu

MH ),(

00

&&

ρρ

ρρ

Connection H melting parameters

M Total volume of melt production . [*My-1*km-1] (melt fract./weight)rm, rc mantle, crustal density

Connection Vp melting parameters

F Mean fraction of meltingZ Mean depth (P) of melting

∫∫=R

dxdzzxmFM

F ),(1 && ∫∫=

R

dxdzzxmzM

Z ),(1

&&

Vp (F,P)

Korenaga et al., 2002

Pyrolite

Estimate H, Vp as a function of Tp, x, Mp, dz, a, composition,

through P, F

H-Vp Diagrams

NATURE OF THE GHS

Hotter

Active convection

MPd=15%/GPa, MPw=1%/GPa, a=0.25, dz=50 kmMPd=15%/GPa, MPw=1%/GPa, a=1, dz=50 kmMPd=15%/GPa, MPw=2%/GPa, a=0.25, dz=50 kmMPd=20%/GPa, MPw=1%/GPa, a=0.25, dz=50 km70% pyrolite + 30% MORB

Compositional anomaly?

SUMMARY

Summary

• All GVP-aseismic ridges show a systematic, overall L3 velocity-thickness anti-correlation

This is contrary to the predictions of the thermal plume model Need to consider a fertile anomaly, possibly a mixture of depleted pyrolitic mantle + recycled oceanic crust

• Velocity-derived density models account for gravity and topography data without need for anomalous upper mantle density

Upper mantle density anomaly is undetectable at distances >500 km from GHS (or 10 My after emplacement)

OTHER OBSERVATIONS

• Major element geochemistry

Fe8 > 13 for individual samples at Galapagos platform

Fe8 higher than “global MORB array” at the edges of CNSC

Positive Na8 – crustal thickness correlation along CNSC, associated to deep, hydrous melting (Cushman et al., 2004) smooth Fe8 signature along most of CNSC?

Match with other observations?

• Temperature

GHS-lavas erupt 50-100ºK cooler than Hawaiian lavas cooling during ascent through lithosphere (Geist & Harpp 2004)

Excess temperature estimations: 215ºK (Schilling, 1991) <200ºK (Ito & Lin 1995)

130ºK (Hooft et al., 2003) 30-50ºK (Canales, 2003) <20ºK (Cushman et al., 2004)

OTHER OBSERVATIONS

• Isotopes geochemistry

Sr-Pb-Nd isotope and trace element signatures consistent with derivation from recycled oceanic crust (e.g. Hauff et al., 2000; Hoernle et al., 2000; Schilling et al., 2003)

Sm-Nd and U-Pb isotope systematics indicate that the age of recycled crust is 300-500 My only (Hauff et al., 2000), which seems to be too short for lower mantle recycling(?)

• Mantle tomography

P-wave tomography with temporary local network (Toomey et al., 2001) has resolution to 400 km only

Receiver functions (Hooft et al., 2003) show thinner than normal transition zone

P and Pp waves finite-frequency tomography (Montelli et

al., 2004) show anomaly only at upper mantle (S-wave?)

OTHER OBSERVATIONS

P- and Pp- finite-frequency tomography

660 km-discontinuity

?

ISSUES

Issues

• If there is a regional chemical heterogeneity, why not upper mantle density anomaly?

• Why is volcanism so focused while global tomography anomaly appears to be much broader? Why is melt not driven to CNSC?

• Why is the GHS apparently a continuous, stable, long-lasting melting anomaly?

• How can the dense, fertile mantle rise to the surface in the absence of a significant thermal anomaly?

• Where does recycled oceanic crust comes from?

FUTURE WORK

Future work?

• Seismological petrology + gravity & topography analysis

Estimate seismic crustal and upper mantle structure with error bounds

Compare H-Vp diagrams for other LIPs

Determine Vp(P,F) for source compositions other than pyrolite• Increase geochemical data/melting experiments adequate to distinguish between thermal/hydrous/chemical origin

• Improve understanding of mantle dynamics

• Test consistency of geochemical predictions with alternative models