Journal of Conference Abstracts

The Relationship Between Buoyant Mantle Flow and the Mantle Bouguer Anomaly Patterns Observed Along the Mid-Atlantic Ridge from 33°N to 35°N

Laura S. Magde Woods Hole Oceanographic Institution, Woods Hole, MA 02543-1542, USA

lauram@copper.whoi.edu

David W. Sparks Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY, 10964, USA

Robert S. Detrick Woods Hole Oceanographic Institution, Woods Hole, MA 02543-1542, USA

Introduction

In the absence of seismic observations, gravity data are often used to constrain variations in crustal thickness and mantle density along the mid-ocean ridges (e.g., Kuo and Forsyth, 1988). Some of the most dramatic features are the Mantle Bouguer Anomaly (MBA) "bull's eyes" observed beneath segment mid-points along the Mid Atlantic Ridge (MAR). These anomalies have been attributed to along-axis crustal thickness variations due to focused mantle upwelling (e.g., Lin et al., 1990). This study uses a three-dimensional mantle convection model (combining both buoyant and plate-driven flow) to quantitatively investigate the contribution of focused mantle upwelling to the MBA observed between the Oceanographer and Hayes transforms. Our results suggest that buoyant flow driven by temperature variations and a small retained melt fraction (<2%) enhances overall crustal production and can explain the long wavelength variations in MBA and crustal thickness observed in this area. However, our models do not predict the magnitude of crustal thinning associated with the non-transform offsets. This suggests that the MBA bull's-eyes associated with individual spreading segments along the MAR are not formed by separate mantle diapirs, but result from the combined effects of three-dimensional mantle upwelling and melt migration.

Study Area

The Oceanographer-to-Hayes section of the MAR extends from 33°N to 35°N and is composed of 5 segments (OH-1 through OH-5) spreading symmetrically at a half-rate of 1.5 cm/yr. The two prominent long-wavelength features in the observed MBA are the regional increase from north to south (i.e. the anomaly becomes more positive) and the signal associated with the 200 km spacing of the transforms (Detrick et al., 1995). Shorter-wavelength features include gravity lows or bull's-eyes centered over segments OH-1, OH-2 and OH-3, but not over the two smaller segments (OH-4 and OH-5). Within individual segments, there is also an asymmetric along-axis gradient, with more positive MBA and deeper topography near the larger offsets. If the entire MBA anomaly is attributed to differences in crustal thickness, predicted variations of >4 km would be associated with each of the transforms, while more modest variations in crustal thickness of 1-3 km would be predicted at the smaller non-transform offsets (Detrick et al., 1995). A near-axis seismic line along segment OH-1 indicates crustal thicknesses of 10 km at the center of the segment, 6 km at the southern segment boundary, and 4 km at the Oceanographer transform (Sinha and Louden, 1983).

Methods

Mantle convection and the resulting crustal production beneath this portion of the MAR are simulated in a region 512 x 1028 km and 300 km deep. We use a 65 x 129 x 65-node finite difference model. The passive plate-driven flow is calculated for the Oceanographer-to-Hayes segmentation geometry using a propagator matrix method (Sparks et al., 1993). Viscosity is a function of depth, with a smooth increase by a factor of 20 centered about a depth of 200 km. Buoyant flow is confined to a uniform viscosity asthenosphere bounded by this viscosity increase with depth and the 1150°C isotherm. Buoyancy is supplied by temperature variations (thermal expansion coefficient = 3 x 10^-5 °C^-1), mantle depletion (10% depletion is equivalent to a temperature increase of 80°C), and retained melt (density difference between solid and melt = 500 kg/m³). We use a solidus with a slope of 3.75°C/km and a melting rate of 0.33% per km of adiabatic upwelling. The retained melt fraction is calculated assuming a balance between melt production and vertical flow (Jha et al., 1994). Crustal thickness at a given point on a segment is taken to be the rate of melt production in the plane perpendicular to the axis, divided by the spreading rate.

We ran experiments for mantle potential temperatures between 1350° and 1400°C, asthenospheric viscosities between 1020 and 1019 Pa-s, and mantle permeabilities which produce maximum retained melt fractions of between 0 and 4.0%. The strength of the buoyant upwelling increases as viscosity decreases and as retained melt fraction and mantle temperature increase. The predicted MBA is calculated using a Fourier-transform method from both the mantle density and crustal thickness variations. The predicted topography was calculated assuming local isostasy.

Results

All of the simulations predict large topographic lows and MBA highs associated with the Oceanographer and Hayes transforms. Between these two transforms is a broad MBA low, centered over the southern portion of segment OH-1 and the northern part of OH-2. While the models predict a small MBA high associated with each of the three longest non-transform offsets, the predicted MBA is not separated into the three distinct lows observed over segments OH-1, OH-2 and OH-3.

The amplitude of the predicted variations in MBA, topography variations and crustal thickness depend on the prescribed model parameters. Simulations with more buoyant flow (generated either by decreasing the asthenospheric viscosity or increasing the amount of melt retention) display more along-axis variation in crustal thickness, gravity and topography. The variation in MBA over the transforms increases markedly in simulations that have more than about 2% maximum retained melt fraction, or asthenospheric viscosities below 5x10¹⁹ Pa-s. The predicted average crustal thickness decreases with decreasing mantle temperature and increasing mantle viscosity.

These models suggest that there is a component of buoyant upwelling beneath the MAR that results in enhanced overall crustal production and increased long-wavelength (> 100 km) variations in crustal thickness and mantle density. However, this buoyant flow does not generate individual mantle diapirs rising beneath each segment, as has been proposed to explain the MBA bull's eyes observed along the MAR (Lin et al., 1990; Parmentier and Phipps Morgan, 1990).

Fig. 1: Cartoon showing mantle flow and melt migration in a vertical section beneath a slow-spreading ridge. Black arrows show the mantle flow observed in the numerical models. There is a long-wavelength variation in upwelling velocity aligned with the large transform offsets, but mantle diapirs are not seen beneath individual segments. There is very little variation in the initial depth of melting. However, even the small non-transform offsets cool the lithosphere sufficiently to create a sloping upper surface of the melting region. If melt migrates along this sloping interface, it will be focused toward segment centers (gray arrows). The resulting along-axis variations in crustal thickness play an important role in forming the short-wavelength features of the MBA bull's eyes. However, they do not directly reflect mantle convection patterns. The observed MBA results from the combined effects of three-dimensional mantle upwelling and melt migration.

Implications

Our results suggest that the MBA bull's-eyes are not formed by individual mantle diapirs, but result from the combined effects of three-dimensional mantle upwelling and melt migration. Variations in the height of the melting column and the amount of melt produced, due to cooling from above, are important at large-offset transforms, but not at the smaller non-transform offsets. We believe the most likely explanation for the large crustal thickness variations observed at the non-transform offsets is along-axis melt migration which is not included in our simple model for melt extraction. If melt migrates along the sloping upper surface of the melt region (Sparks and Parmentier, 1994), then even small variations in temperature structure at non-transform offsets may lead to significant focusing of melt towards segment centers (Figure 1).

References

Detrick, R.S., Needham, H.D. & Renard, V., J. Geophys. Res. 100, 3767-3787 (1995).

Jha, K., Parmentier, E.M. & Phipps Morgan, J., Earth Planet. Sci. Lett. 125, 221-234 (1994).

Kuo, B.-Y. & Forsyth, D.W., Mar. Geophys. Res. 10, 205-232 (1988).

Lin, J., Purdy, G.M., Schouten, H., Sempéré, J.-C. & Zervas, C., Nature 344, 627-632 (1990).

Parmentier, E.M. & Phipps Morgan, J., Nature 348, 325-328 (1990).

Sinha, M.C. & Louden, K.E., Geophys. J. R. Astron. Soc. 75, 713-736 (1983).

Sparks, D.W. & Parmentier, E.M., in Magmatic Systems, ed. M.P. Ryan, Academic Press, 55-76 1994).

Sparks, D.W., Parmentier, E.M. & Phipps Morgan, J., J. Geophys. Res. 98, 21,977-21,205 (1993).