Western US Mantle Dynamics and Volcanism

What is the source of late Cenozoic volcanism in the Western United States? Is the Yellowstone hot spot track produced by a deep-seated mantle plume or by some other process acting only at shallow levels? Why do the major Miocene to present aged volcanic fields exist along major lithospheric boundaries?

We hope to place physical and chemical constraints on possible scenarios for these questions through the use of numerical models. For example, we are exploring the possibility that small-scale asthenospheric convection (driven by melting processes) may be either entirely or at least partly responsible for these and other volcanic features through out the Basin and Range. We have found that melting-related buoyancy in the upper mantle plays an important role in governing small-scale (on the order of 50-100 km) convective processes in the asthenosphere.

Also check out these links to some of our collaborator's web sites:


Following is a geologic map of the western U.S. (from Philip B. King and Helen M. Beikman, US Geological Survey), showing important place names relevant to this discussion. Note that the St. George volcanic field (see figure below) lies along the northern edge of the Colorado plateau (which is also the location of a Proterozoic hingeline) and probably reflects an ancient transition from stronger lithosphere to weaker lithosphere.

Abbreviations are as follows:
  • YS = Yellowstone volcanic field
  • SRP = Snake River Plain
  • SF = San Francisco volcanic field
  • CRFB = Columbia River flood basalts
  • B&R = Basin and Range


Here is the same map, with volcanic rocks of Miocene age and younger shown in black for emphasis:


Here is another map, showing the approximate boundaries and ages of major Precambrian provinces. Note that the Jemez Lineament coincides with the Yavapai-Mazatzal province boundaries.


Shown here is a plot of the major silicic time progressive volcanism (approximate countours of age in Ma) that occurred in the western U.S. during the Cenozoic (original plot from Gene Humphreys). The pre-middle Miocene volcanism is generally termed the "mid-Tertiary ignimbrite flare-up" and the two more recent episodes are the Yellowstone-Snake River Plain (YS-SRP) and Newberry trends. This plot makes it fairly clear that time-progressive volcanism is an ubiquitous feature of the Cenozoic geology. It also shows how small more recent volcanic episodes have been in comparison to the mid-Tertiary ignimbrite flare-up.


Some thoughts so far:
From the above figures, we see that the late Cenozoic St. George volcanic field and Jemez lineament occur along pre-existing major lithospheric boundaries, while the Snake River Plain appears to march straight into Archean Wyoming cratonic lithosphere. This is not the only difference between these volcanic fields: the onset of SRP volcanism in the late Cenozoic is clearly time-progressive and tends to trace out a fairly nice hotspot track relative to North American plate motion, while the St. George and Jemez volcanics exhibit no time progression.

Although St. George and Jemez are also roughly aligned with plate motion, the major lithospheric structures they follow have the same alignment. We might then ask: are the alignments principally controlled by lithospheric structure, plate motion, or a combination of both? The answer to this question might be sought by looking for other volcanic fields along major lithospheric boundaries that are not parallel to plate motion. One volcanic field that shares the edge of the Colorado plateau with St. George and Jemez is the San Francisco field, which is located around Flagstaff, Arizona (SE of the Grand Canyon). In fact, scattered Miocene to present volcanic rocks can be found around the entire circumference of the Colorado plateau, with the exception of its northern edge. The Jemez lineament not only lies along the edge of the Colorado plateau, but has the added benefit of being along a major Precambrian suture zone.

Following is an image from Gene Humphreys of the seismic velocity anomalies at 100 km depth beneath the western U.S. (red=slow, blue=fast):

Note the correlation between late Cenozoic volcanism and slow upper mantle seismic velocities.


Shown here is a compilation of various seismic studies conducted across the Snake River Plain (from Humphreys, 2000), which is in the wake of the Yellowstone hot spot. Note especially the red and blue stuff, which represents fast and slow variations in seismic velocity respectively. This varies greatly over length scales of ~200 km.

This is a schematic cartoon showing the expected parabolic shape of a plume head impinging on the base of the lithosphere (from Ribe and Christensen). Compare this with the sies mic signature above, and note that no big parabolic wake exists under the Snake River Plain.

This is a movie from Bill Moore's plume-lithosphere interaction model showing a 3-D isotherm. This is a calculated plume shape, showing the same parabolic broadening as the above cartoon. Also, note the small-scale convective instabilities developing in the plume head. Could this be the reason for small-scale structure observed across the YS wake? This is one possibility we are studying.

The strain-spot: an alternative scheme with no deep-seated plume involved. This is small-scale convection, driven mainly by melt buoyancy. Such convection is organized into rolls by shear between the lithosphere and mantle, accounting for the linear trends. Time progression can be made to occur by thinning along a zone of focused extension, lengthening the rolls and inducing a sort of "hot spot" propagation.

The strain-spot scheme for Yellowstone. Shear is induced by plate motion, and melting occurs due to extension and the presence of a very hot mantle. Hot spot-like propagation occurs if the eastern edge of the basin and range has focused extension which moves eastward in a wave with the hot spot reference frame. Note that a plume may also work in concert with this mechanism, perhaps by simply injecting hot material beneath the lithosphere, making the system more unstable to small-scale convection.




Do time-progressive linear volcanic trends usually come in pairs? No! But here you see the Yellowstone and Newberry volcanic trends starting from the same area, and then propagating out in opposite directions. Newberry propagates westward into Oregon, while Yellowstone heads east. This is highly suggestive of a common origin for the two trends.

Scheme for a strain-spot that gives rise to the Newberry time-progressive volcanic trend in Oregon. In this case, extension is partitioned along the western edge of the basin and range, and shear flow is controlled by corner flow through the Juan de Fuca subduction zone wedge.

Okay, enough cartoons...let's look at some calculations. Here is the boundary and initial conditions for the simple strain-spot model. Extension is localized over a limited region and the plate is forced to move through it. Melting and related buoyancy is included. The little red nub is a plume that I can turn on or off if desired. The depth of this box is 400 km and the background shows the initial temperature field.

A baseline case: let's see what happens when there is no melting, just thermal buoyancy. This is a 3-D isotherm, and you can see that weak small-scale convection begins to occur downstream, in initially perpendicular to the strain direction because this flow induces a lateral temperature gradient, but after some time a cross-roll change occurs and the flow takes the form of rolls aligned with plate motion (termed "Richter rolls").

Now let's do the same thing, but turn on melting and related buoyancy effects. This is another isotherm, but you can see that things become a lot more vigorous, and extreme variations of temperature can be induced over very short length scales. The small-scale convection takes the form of Richter rolls much sooner.

Same case as above figure, only this time showing the resultant melt field. Melt occurs along the upwelling limbs of the rolls, as expected. The melt zone thickens in depth downstream , and the model gives a spacing between rolls that is roughly 2-3 times the melt zone thickness. This gives a nice agreement with the seismic tomography.

Shown here is a slice looking along a roll from one of my calculations. The color field represents temperature and the velocity field is also plotted. In this simulation, the lithosphere is extremely stiff and has a very high viscosity. This high viscosity contrast eliminates the drips and makes the convection more compact.

This is what the initial temperature and velocity field profiles look like from the side, with upstream to the right and downstream to the left. Note the upwelling under the middle caus ed by localized extension.

Have we given up on the plume? Not yet. Let's see what happens when we have the same melting effects in the exact same model, but add the plume on the bottom. This is an isotherm for this case, and you get a big plume head thermal effect at the top, but no small-scale stuff like you see in Bill Moore's model above.

This is the melt field produced in the above plume model. This shows, perhaps, why the small-scale stuff seen in Bill Moore's model goes away. The whole head gets partially melted and equally buoyant, which in this model seems to suppress any sort of small-scale convection in the plume head. Reducing melting effects may help a lot in this case, getting the small-scale convection in the plume wake.




Shown here is a topographic map of the western us, with SKS splitting (dark red lines, from Derek Schutt) and GPS velocities (black arrows) showing the relative surface velocity fie ld across the great basin. Note in particular that the differences in surface velocity are greatest along the edges of the basin and range province, while very little deformation occurs internally, consistent with the proposed strain-spot mechanism requir ing focused extension at the edges of the basin and range.

This is a map showing the time progression of the Yellowstone hot spot over the last 16 million years. Closer examination of the details reveals some sloppiness in its ideal hot sp ot behavior, including a slight bend and change in rate of propagation. If the strain-spot model is correct, then this does not record the passage of the North American plate over a deep-seated mantle plume, but rather records the history of extension alo ng the edges of the basin and range province.

What's up next? I am currently running many more of these models and varying different parameters to see how things scale. Either way it seems that the basin and range mantle has to be very hot to sustain these melting processes. We are also looking at other non-time progressive linear volcanic trends, like the Jemez lineament in New Mexico, and the St. George trend in Utah. These also seem to point in the direction of plate motion, and so they may be the same sort of phenomenon.




So is there a plume under Yellowstone? We're officially still on the fence with this issue. A few things seem more certain: 1) you need a hot mantle under the lithosphere to induce small scale-convection...whether this is supplied by a plume wake or just regionally hot mantle doesn't really matter, 2) Partial melting should occur in the upwelling limbs of the small-scale convective rolls, since lack of this extra buoyancy source produces temperature variations that are too small (around 25 K in amplitude) to explain the seismic velocity variations in the western US asthenosphere, 3) Compositional effects on buoyancy due to depletion must not be too high (upper limit of around 1% decrease in density) since this inhibits the formation of small-scale convective rolls.

So where do we stand? The more models I do of this process, and the more I learn about its connection to other observations through out the Western U.S., the less I tend to think a plume is a necessity in the dynamics. The only thing we need a plume for is to make a nice time progression, but as shown above, there are other ways to do this. Other time-progressive volcanic events, quite similar in many respects to the Yellowstone Newberry system, have been well documented in the geological record. In particular, the mid-Tertiary ignimbrite flare-up is a much larger scale disturbance, and records time progression from S. N.M. to S. Nevada in the southern basin and range, and another time progression from the N. great basin to the same location at the same times. Only this time-progressive event occurred prior to the mid-Miocene, which is when Yellowstone-Newberry fired up. Is there a link here? Perhaps.

Movie: Click here for a movie (mpeg format) of the formation of shear induced melt lineations under an extending region neighboring a thick craton.