4D model solves Yellowstone’s geographical mysteries

From geysers to tectonic plate movements, corn syrup has the answers. Trey Ratcliff

A decade ago, with Ross Griffiths of the Australian National University, we aimed to build a 4D model which could replicate the Earth’s tectonic processes. Now, our research has helped us understand how some of America’s most mysterious geographical formations took shape.

We wanted to build a new lab apparatus for simulating subduction zones. These are regions of the Earth where cold, dense oceanic plates meet less dense continental plates and descend into the Earth’s mantle. The mantle is the viscous layer between the Earth’s crust and outer core. Subduction is the dominant force driving movement of such plates, and one of the features that makes Earth unique in the solar system. Our goal was to represent pressures and movements in the mantle as it responded to the sinking oceanic plates, as well as to thermal evolution, which is movement of heat from the core to the surface.

Many areas of science and engineering have a long tradition of employing analog - rather than computerised - models, to represent larger scale processes. Just as toys like planes, boats and dolls can be modelled to represent larger versions using scale factors, laboratory experiments may be used to model the mantle system by employing certain dynamic scale factors. To our 3D analog model we added a fourth dimension - time.

We used a fiberglass plate to represent the Earth’s subducting oceanic plate and high viscosity corn syrup to represent the Earth’s mantle. Three features make corn syrup a nice analog for the mantle. Like the mantle, the fluid is inertia-free. This means that when forces are removed the flow stops immediately. Corn syrup has a temperature-dependent viscosity, so heat makes it runny while cold makes it strong and unyielding. It also scales well to the mantle (and can provide a nice dessert after the work is complete).

We first used our apparatus to study a number of general aspects and mechanisms of subduction zones, without focusing on any particular region. But all that changed when we joined a diverse group of geoscientists, interested in determining the cause of some strange magma activity in the Pacific Northwest over the last 20 million years.

The focus then became an odd volcanic track in central and eastern Oregon in the US, called the High Lava Plains. The goal was to connect the dots between the High Lava Plains and two other notable geographical features - the Columbia River Flood Basalts and the Snake River Plain. This area around the Yellowstone park is the epicentre of much debate in the earth sciences.

A number of scientists have said that the Columbia River Flood Basalts and the Snake River Plain are the expressions of a surfacing mantle plume. A mantle plume is made of hotter, less dense rock, buoyed upward through the mantle toward the Earth’s surface. They resemble the kinds of plumes that we see every day: hot air balloons, lava lamps, pasta sauce on a stove and steam plumes from industrial towers are all common examples, where less dense fluid or air rises through denser material.

In the mantle, plumes take on a classic shape: a large, bulbous, leading volume called the “plume head” followed by a thin tail, called the “plume conduit”. The surfacing of plume heads has been linked with huge outpourings of basalt called “large igneous provinces”, while tails are commonly proposed as the cause of linear volcanic tracks like the Hawaiian Island chain in the Pacific.

Map of Cascades-Yellowstone subduction system showing general locations of the Steens Columbia River Flood Basalts (SCRB) and the opposite trending volcanic tracks (the High Lava Plains (HLP) and the Snake River Plain (SRP)) C. Kincaid, H. Gao & K. Druken/Nature Geoscience

But for our area, a plume was not the most obvious suspect. One issue is that the Columbia River Flood Basalts and the Snake River Plain are not laid out in a linear pattern, as would be expected if a plume surfaced beneath the westward-drifting North American continental plate. Instead, the two features seem broken, with the Snake River Plain track displaced well south of the flood basalts.

Nor does the High Lava Plains volcanic track fit the classic plume model. It is odd for two reasons: it appears west of the flood basalts, and its magma output grows younger toward the west. We would intuitively expect the opposite. As the plate moves west, the head of the plume should arrive at the surface first, causing the flood basalts, while the tail should cause a track of volcanic activity that grows younger toward the east, just like the Snake River Plain.

A schematic representation of the plume-subduction lab apparatus. K. Druken, C. Kincaid/Nature Geoscience

To accurately represent this complex subduction system, we needed to include a number of features. One of these was a phenomenon called “rollback subduction”, which is when the sinking oceanic plate rolls back underneath itself as it descends into the mantle.

We also had to represent the extension of the overriding plate. The overriding plate is made up of buoyant continental crust and heavy chilled mantle. The dense oceanic plate - which has only a sliver of crust - sinks beneath the overriding plate. Between the sinking and overriding plates lies what is referred to as the “mantle wedge”, the source region for volcanic arcs that roughly parallel the trench in most subduction zones.

The project’s pièce de résistance was the representation of the mantle plume, where we could control the plume’s buoyancy and position. We did this by using a modified pressure cooker where corn syrup was heated and then injected through a heated pipe into the base of the tank. By controlling its temperature, we controlled the density, viscosity and rise rate of the plume.

Pouring more corn syrup in the setup. Kelsey Druken

Most prior models inject plumes into a simplified fluid environment. Models which incorporate both subduction and a plume show this simple view is not sufficient. Rollback motion produces vigorous flow from the back to the front of the subducting plate. This is known as a “torroidal flow”. This is like a super large eddy in the mantle. Our results showed that torroidal flow had a big impact on the buoyant plume feature.

In the initial stages, we released the plume into a tank of still fluid, where it evolved in textbook fashion: a large, leading head formed, which rose to the surface, stalled and spread out. Behind this was a narrow conduit through which plume fluid was continuously fed from the base to the surface of the tank. But as soon as we turned the plates on it was apparent this simple morphology was no longer in play.

Circulation currents driven by the plates immediately deformed both head and tail, dramatically limiting the ability for continued upward, buoyant motion. As the torroidal flow moved around the edge of the sinking plate, it effectively took a bite out the plume head, drawing the material into a thin lens and dragging it towards the edge of the subducting plate. The remains of the plume surfaced in a pattern that matches the location and timing of the High Lava Plains. The unchewed part of the plume head rose and surfaced as a big pulse of heated, melting substance, similar to the flood basalts.

The plume tail tilted over beneath the westward-drifting overriding plate, leaving patterns of heating and melting that grew younger to the east, just like the Snake River Plains. The torroidal flow not only bit the head in half, it also pushed the tail well to the south, matching the real feature. This is how corn syrup helped us explain how a single plume could be responsible for the Columbia River Flood Basalts, the southerly deflection of the Snake River Plains and the bizarre, westward-younging High Lava Plains track.