Absolute asthenosphere viscosity from Caribbean dynamic topography, mantle structure and magmatismYi-Wei Chen1 (yiweichen.tw@gmail.com), Lorenzo Colli1, Dale E. Bird1, Jonny Wu1 and Hejun Zhu2 1Dept. of Earth and Atmospheric Science, University of Houston2Dept. of Geoscience, University of Texas at Dallas

1

3

4 9

2A Honey Sandwich or a Peanut Butter Sandwich?

Independent constraints

10

8

Conceptual evolution

Flow velocity varies with different viscosities 7 Residual Topography (i.e. Dynamic topography)

A gravity-constrained crustal model (using Oasis Montaj)

Lithospheric age of the Caribbean 5 Sedimentary Thickness (This study) Moho Depth (This study)6

Topography = Thermal subsidence + Dynamic topography + Isostatic topography

Pressure Difference & Velocity Profile

References:1. Müller et al. (2008) Geochemistry, Geophysics, Geosystems 9.4.2. Weismüller et al. (2015) Geophysical Research Letters 42.18: 7429-7435.3. Magnani et al., (2009) Journal of Geophysical Research: Solid Earth 114.B24. Paulson and Richards. (2009) Geophysical Journal International 179.3: 1516-1526.5. Schaber et al. (2009) Geochemistry, Geophysics, Geosystems 10.11.6. Boschman et al. (2014) Earth-Science Reviews 138: 102-136.7. Gazel et al. (2011) Lithos 121.1-4: 117-134. 8. Hoggard et al. (2016) Nature Geoscience 9.6: 456.9. Hu et al. (2016) Nature 538.7625: 368. 10. Sandwell et al. (2014) Science 346.6205: 65-67.

High Pressure Low Pressure

Lesser Antilles Arc

CentralAmerican Arc

Arc extinct

Alkaline basalts withGalapagos signature

The leading edge of slow anomalies

Velocity profile with our preferred asthenospheric viscosity of 5*1018 Pa s

Galapagos plume

Slab window

Dynamic uplift~300m

Tilted Lithosphere

Before 8.5 Ma

Lower Mantle

Lithosphere

Asthenosphere

8.5 Ma

0 Ma

Tortuguero <<0.1 Ma? Guayacan/Azules: 4.5-2 Ma

Victoria dikes: 6.5 MaPearl Lagoon/Cukra Hill: <0.7 Ma

Alkaline basalts (Gazel et al., 2011)

210-2 -1

δln Vs (%)

20ºN

10ºN

0º80ºW90ºW 70ºW 60ºW

Beata Ridge

Slab Windowopened at 8 Ma

AlkalineBasalts:5 cm/yr

Depth: 300 km

Azimuthal anisotropy, proxy for flow direction

A A’

B

B’

Leading edge of the slow anomalies: 15 cm /yr

1000

800

600

400

200

0

1000

800

600

400

200

0

A A’

B B’

Dep

th (k

m)

Dep

th (k

m)

Lesser Antilles Arc

Subducted Atlantic Slab

Subducted Proto-Caribbean Slab ?

South Americacontinental root

Subducted Caribbean slab

Asthenosphere

Asthenosphere

The Galapagos Beata Ridge

The AtlanticThe CaribbeanThe Pacific

Robust Residual topography (This study) Hoggard et al., 2016

Residual topography (This study)

slab

win

dow

Less

er A

ntill

es tr

ench

Res

idua

l Top

ogra

phy

(RT)

(km

)

-2

-1

1

2

0

h = 300 m

d = 2500 km

P= ρgh

ρ = 3300 g/ccg= 9.8 m/s2

0

40

80

120

160

2000 5-5 10 15 Velocity (cm/yr)

Dep

th b

elow

lith

osph

ere

(km

)

Leading edge of

3*10185*101810195*10195*1020

the hot material

v(y) = * * y * (y - H) + u *

Pressure gradient Velocity as a function ofdepth

Viscosity values from 1 Thickness from 9

from 9

ρgß 12η

yH

Best-fit regression line:

Alkaline basalts from 9

0 500 1000 1500-500-1000-1500 2000 2500 Distance (d) fromslab window (km)

RT = α + ß*dα = 0.476 ± 0.012 ß = 1.4*10-4 ± 9*10-6

The uncertainty of velocity profile from the uncertainty of pressure gradient

Our preferred velocity profile with our proposed viscosity

Density (g/cc)Thickness (km)

Crust

Mantle

SedimentWater

Moho (see 6 )

Basement (see 5 )

Topography

Free air gravity anomaly

- GTOPO30 (USGS, 1998)- Terrainbase (Row et al., 1995)

- Refraction / Reflection- Global Sedimentary Thickness (Divins, 2003)

- Refraction- Gravity inversion

Oceanic Mantle:age-dependent(see 3 )= 3.25-3.31

Continental Mantle= 3.3

Crust of oceanic plateaus= 2.85

Sediment: thickness-dependent = 2.15-2.6

Water = 1.03Sea level

GravityFree air gravity anomaly ∑Mass = ∑ ( Thickness x Density)

Topography & Bathymetry (km)

-8 50

Residual Topography (m)

-150 8500

20ºN

10ºN

0º80ºW90ºW 70ºW 60ºW

BR

Cross-section in 8

Moho depths (km)

-31-55 -43 -19

Slope-intercept Synthetic

Rough & Acoustic Basement Boundary

?

?

80ºW90ºW

20ºN

10ºN

0º

2010 30 40

20

10

30

40

Inversion Moho (km)

Ref

ract

ion

Moh

o (k

m) 1 t

o 1 lin

e

Slope-intercept

Synthetic

Seismic refraction

Sediment Thickness (km)

0 5 10 15

20ºN

10ºN

0º

80ºW90ºW 70ºW 60ºW

seismic reflection

seismic refraction

reflection + refraction

wells

DSDP

IODP

Cocos-Nazca spreading center Basalt samples of

the Caribbean crust(~100-80 Ma)

Cayman trough

spreading center

Age (Ma) (from Muller et al., 2008)

0 50 100 150

20ºN

10ºN

0º

80ºW90ºW 70ºW 60ºW

Mantle flow

Dynamic topography Isostatic topography

Thick Crust Thin Crust

3.25 3.27 3.29 3.31Density

(g/cc)

4

Dep

th (k

m)

12

20

0 100Age (Ma) 200

Thermal subsidence(Richards et al., 2018)

++=

Galapagos

Cocos ridge

Beata Ridge

Colombiabasin

Aves ridge Lesser Antilles

LNR

UNR

Cayman Trough

Carnegie ridge

Venezuelabasin

Sandra riftPFZ

NPDB

SCDB

Topography and Bathymetry (km)

0-4-6 -2 2 4

20ºN

10ºN

0º

80ºW90ºW 70ºW 60ºW

S. AmericaContinent

N. America Continent

Continental Terranes (Boschman et al., 2014)

3. Calculate flow velocities with different viscosity values.

4. Compare to independent velocity constraints from geology.

2. Convert it to an average pressure gradient.

1. Quantify the dynamic topography across the Caribbean region.velocity profile

low viscosityupper mantle high

Pressure lowPressure

velocity profile

Geological constraints

Geological constraints

high viscosityupper mantle

warped lid

Strategy:

Viscosity (Pa s)1018 10211019 1020

PSR at Indonesia (Hu et al., 2016)

(Han et al., 2019)

Isostatic rebound at Western US

GIA at Iceland

PSR at Samoan islands

PSR: Postseismic Rebound

GIA:Glacial Isostatic Adjustment

(Auriac et al., 2013)

(Bills et al., 2007)

GIA at Canada & Scandinavia

(Paulson & Richards, 2009)

Global Geoid (Schaber et al., 2009)

Viscosity: 5*1018 Pa s (This study)

(2*103 mPa s) (2*105 mPa s)

←The Caribbean is bounded by subduction zones and continental terrains (transparent yellow polygons); therefore, the subducted slabs and the continental roots hinder the asthenosphere to flow freely, but confine the flow within a narrow gateway beneath the Caribbean, a scenario similar to the simplified 2D model shown in 2 .

We used seismic refraction, reflection and borehole data to build the velocity model of the upper most crust as a function of depth. The velocity model and the 2-way travel time from seismic reflection help us constrain the sediment thickness. Sub-tracting the sediment thickness from the bathymetry, the basement depth is ob-tained.

The thermal age of the Caribbean lith-osphere was was rescaled from 100 Ma to 80 Ma based on the basalts.

← To obtain dynamic topog-raphy, we subtract the total topography from thermal subsidence and isostatic to-pography. Thermal subsid-ence is obtained from the plate cooling model of Rich-ards et al. (2018) using the age model from Müller et al. (2008) (see 4 ). Isostatic effect was corrected using seismic-constrained sedi-mentary thickness (see 5 ) and gravity constrained Moho (see 6 ). The residual topography (see 7 ) is gen-erated by mantle flow and is our best-estimate for dy-namic topography.

Due to the limited refraction constraints of the Moho (red lines and blue open boxes), we used free air gravity anomaly to constrain the Moho. Our regional gravity-con-strained Moho model fits local Moho depth estimates from refraction studies (see inset).

Independent constraints come from (1) back arc magmatism which shows clear Galapagos isotopic signatures, initiating at ~6.5 Ma near the slab window (black circle) which opened at ~8 Ma and propagating at a rate of 5 cm/yr northward. (2) Full waveform tomography shows a slow velocity anomaly in the western Caribbean (west of Beata Ridge). If we assume the anomaly is hot mantle material flowing through the slab window, the propagation rate is ~15 cm/yr. The two cross sections suggest that the asthenosphere is ~200 km thick.

The dynamic topography across the Caribbean region constrains the pressure gradient that drives Galapagos hot mantle material flowing eastward through the slab window. Given the driving pressure gradient and the thickness of the asthenosphere, flow speeds depend only on the viscosity of the as-thenosphere. A value of ~5*1018 Pa s best fits the propagation rates of the back-arc basalts and the imaged seismic structure.

The concept of a weak asthenosphere sandwiched between mobile tectonic plates and mechanically strong sub-asthenospheric mantle is fundamental to understand plate tectonics and mantle convection. However, the quantitative estimation of absolute asthenospheric viscosity is highly uncertain and it varies by 1 to 3 orders of magnitude (see 1 ). Wide-range viscosity estimations come mainly from two reasons: (1) Most observations can be satisfied with proper viscosity contrasts –instead of the absolute viscosity– between asthenosphere and sub-asthe-nospheric mantle. (2) The viscosity contrast trades off with the thickness of the asthenospheric layer.

Here, we estimate absolute viscosity of asthenosphere for the first time from the pressure gradient and the asthenospheric flow velocity at the Caribbean, using the well-established analytical solution (see 8 ). Funnelled by subduction zones and continental lithospheric roots (see 3 ), the Caribbean region provides a unique tectonic setting reminiscent of a plane channel (see 2 ) that closely approximates the conditions of the analytical solution. The asthenesphric flow at Caribbean from the Pacific towards the Atlantic has been a long-standing hypothesis. We in-ferred the flow velocity from regional magmatism (see 9 ) and tomographic imaging with the onset time of the flow at ~8.5 Ma constrained by the opening of Panama slab window (see 10 ). This flow is mainly pressure-driven (i.e. Poiseuille flow), since the Caribbean has been fixed in mantle reference frame since Eocene. The pressure gradient (see 7 ) is calculated from the dynamic topography (see 3 ) across the Carib-bean, obtained by removing the isostatic effect of sediments (see 5 ) and crust (see 6 ). Independent constraint of asthenosphere thickness of ~200km is suggested by tomography ( see 9 ).

The importance and improvement of this study are three folded. First, we provided a better quantified pressure gradi-ent based on a proper isostatic correction and uncertainty es-timations. With reasonable assumptions, we are the first study that use the stain rate (i.e. flow velocity) directly from the asthenosphere, instead of an average strain rate of the upper mantle from surface constraints. Finally, we show for the first time an on-site estimation of the viscosity where the asthenosphere thickness is independently constrained. Our results suggest the viscosity of the asthenosphere at the Ca-ribbean is ~5*1018 Pa s, which is in line with estimates for non-cratonic and oceanic regions, but is significantly lower than post-glacial rebound estimates for cratonic regions. This further supports the notion that the stronger asthenosphere estimates are relatively limited to cratonic regions.