Top Down Processes Controlling Dike Initiation, AMC Depth and Extrusive Patterns at Spreading Centers

W.R. Buck

Corresponding author: buck@ldeo.columbia.edu
Lamont-Doherty Earth Observatory of Columbia University, Rt. 9W Palisades, NY 10964

Abstract:
Far more basalt is emplaced at spreading centers than at terrestrial volcanoes, yet many of our ideas about magmatic processes come from studies of volcanoes. Differences between spreading centers and volcanoes may be fundamental. For example, because terrestrial volcanic magma chambers lie within the lithosphere they can build sufficient over-pressures to force magma to breakout of the chambers and open dikes. Also, the topographic relief of volcanoes may lead to severe under-pressure of magma as dikes reach the surface at the sides of the volcano. The under-pressures may be sufficient to cause collapse of the magma chamber roof and production of caldera.

At fast- to intermediate-rate spreading centers there is evidence that axial magma chambers sit at the base of the lithosphere and overlie partially molten crust. These sub-lithospheric magma chambers should not easily build magma over-pressure. To initiate a dike at a plate-spreading center, fault activity may be needed to break through the ductile “buffer zone” above the magma chamber, bringing magma in contact with stressed brittle rock. If an episode of one or more dikes relieves extensional stresses near the spreading axis then the repeat time for dike episodes equals the time to re-stress the axial lithosphere to the point of fault failure. Then the time to re-stress the axial lithosphere and trigger new dikes depends linearly on the depth to the axial magma chamber and inversely on the spreading rate. For reasonable elastic and fault failure parameters the model is consistent with observed time intervals if the region of axial lithospheric stress relief during a dike episode is a few to ten kilometers wide.

Axial summit troughs on the EPR and “split linear volcanoes” on the Juan de Fuca Ridge may be analogous to collapse caldera on terrestrial volcanoes. It is possible that these spreading center collapse features are produced by lithospheric stresses and not magma under-pressure. A likely source of stress is the density structure of the shallow crust. Since the stresses related to plausible density variations are small it is possible that bending related fractures only propagate to allow collapse when basalt is extruded over the surface. At fast spreading ridges the collapse trough may be filled in by extrusives, while the deeper trough at intermediate spreading ridges the build-up of a volcanic ridge may be too slow to fill the region of collapse. The history of collapse filling affect the distribution of extrusives and AMC depth.

Keywords:
Dikes, sills, axial magma chambers, extrusives, crustal structure

Contributions to Integration and Synthesis:
Several datasets could be used to test the process models outlined in this work. First, on the 9°N section of the EPR earthquake data can be used to test the model prediction that both the frequency and depth of earthquakes should increase with time after the end of a dike episode. Future monitoring on this and other ridge segments, particularly intermediate and slower spreading ridges are critical to testing the model predictions. We also need to know whether dikes intrusion episodes on all spreading centers always end in extrusion, indicating complete release of extensional stress. It is quite possible that dike episodes are very different on slow- and ultra slow spreading ridges.

The magmatic collapse of linear caldera models can best be tested with data on the structure of the upper crust and by considering temporal variation of axial magma chamber sills. Studies of the structure of extrusives and rotation of layer 2 seen in tectonic windows are of particular importance. The link between collapse and segmentation also needs to be developed in future model studies.

Figures:
Figure 1. Conceptual differences. Buck_fig1.jpg
Figure 2. Dike model predictions and data. Buck_fig2.jpg
Figure 3. Collapse model –no filling. Buck_fig3.jpg
Figure 4. Collapse with trough filling. Buck_fig4.jpg