Mitchell McMillan
geodynamics and geomorphology

Research topics

My research investigates how deep-seated geodynamic processes affect the evolution of Earth’s surface, including potential feedbacks between climate, tectonics, and lithology. My main interest lies in orogenic plateaus, e.g., the arid, high elevation, highly deformed mountain belts in the Central Andes, Tibet, and other locations. I primarily use geologic mapping, eolian geomorphology, geodynamic modelling, and low-temperature thermochronology to explain the deformation of the crust and topography of these interesting regions. A recent research direction is understanding the behavior of aqueous fluids in the deep crust above subduction zones, especially in the context of metamorphic reactions. A set of reactions loosely referred to as eclogitization are expected to produce large volumes of dense, high pressure rocks, which could become gravitationally unstable. Deep crustal fuilds may therefore be significant drivers of the geodynamics and topography of mountain belts.

Expand the headings below to read in more detail about these topics.



Lithospheric Dripping

My PhD project at the University of Toronto involves studying lithospheric dripping. Lithospheric dripping is the idea that part of the lower crust or mantle lithosphere (the “plate” in plate tectonics) can become so dense that it sinks and drips off into the mantle. This process is called the Rayleigh-Taylor instability and is applicable to a wide range of phenomena, from stellar nebulae to mushroom clouds, to a drop of water suspended on top of oil. If plate tectonics describes how the rigid oceanic plates move laterally, pulling the continents along for the ride, lithospheric dripping is one way to describe how weak, putty-like continental plates (especially mountain belts) deform. Unfortunately, lithospheric dripping is near-impossible to observe directly, and we have to look closely at the sedimentary, volcanic, and geophysical evidence for a given region to discern whether dripping may have occurred. In doing so, it helps to compare these observations to scaled experiments of lithospheric dripping (whether mathematical solutions, numerical models, or laboratory experiments).

My approach is to (1) synthesize the literature on lithospheric drips, including the evidence for each dripping event as well as alternative hypotheses (slab fragments, delamination, and others) and relevant modelling studies; (2) do some geodynamic modelling of lithospheric dripping, especially as it applies to my field area in the Puna Plateau; and (3) study a location that may have experienced lithospheric dripping during the Miocene, the Antofalla region of the Puna Plateau.

Research plan developed for my PhD thesis.

A review of the literature on lithospheric dripping reveals a surprising number of proposed drips: I found 26 locations on Earth for which researchers have proposed the lithospheric dripping hypothesis. I specify Earth here because dripping has also been proposed to explain some circular terranes on Venus called coronae. Even well-known landscapes like the Grand Canyon may be partially due to lithospheric dripping. To make sense of all these events, I synthesized numerical modeling studies to propose that lithospheric dripping occurs as two distinct types. The first type involves upper-crustal compression and basin formation; the second type involves upper-crustal extension and topographic uplift. The main factor that determines which type of drip prevails is the strength of the crust relative to that of the mantle lithosphere, with strong crusts (>10x stronger than the lithosphere) favoring the first type. This analysis is not new, as numerical modellers have long recognized these two distinct outcomes; I found, however, that more complex geodynamic models also produce these two distinct outcomes (with some minor complications), and that both types of dripping are likely to be found on Earth. This result is interesting, because many papers that engage with the lithospheric dripping hypothesis only consider the first type of drip (strong crust). The second type (weak crust) has received relatively little attention by geologists, maybe because weak crusts are relatively rare and require specific tectonic conditions such as back-arc volcanism or continental rifting to weaken the crust sufficiently.

Illustration of strong-crust (top) and weak-crust dripping (bottom).

Evidence from the Southern Puna suggests that the weak-crust type of dripping may have occurred during the Miocene. This makes the drip hypothesis difficult to test here, because relatively few geodynamic models have investigated the effects of this type of drip. We know that weak-crust drips should cause upper-crustal extension above the drip, along with crustal thickening and perhaps partial melting (see the illustration, above). There are important outstanding questions, however: what happens after the drip detaches? Does the crust rebound by deforming compressionally, or does the thickened crust collapse under its own weight? What happens at the edges of the drip? Is compression strong enough to significantly deform upper-crustal rocks? What about if we impose a background of tectonic strain, say moderate compression due to oceanic slab subduction? These are all questions that apply to the purported drip in the Southern Puna Plateau, but that cannot be definitively answered by referencing existing geodynamic models. For my thesis, I am constructing a few models designed to explore these weak-crust drips more thoroughly, refining the dripping hypothesis and providing falsifiable predictions that can be compared to geologic data in the Southern Puna.

A very simplified three-dimensional model of lithospheric dripping (constant densities and Newtonian viscosities) compares favorably to a laboratory experiment of the Rayleigh-Taylor instability, recreating features such as “line” or “curtain” plumes, linear instabilities that connect the more significant “drip” downwellings. What this model and laboratory experiments lack are the strong gradients in viscosity and density that characterize Earth’s lithosphere.

Teasing apart the evidence for lithospheric dripping involves a thorough understanding of the timing of tectonic deformation, sedimentation, and volcanism in a region. The Antofalla region of the Southern Puna is situated near the center of an inferred dripping event (mainly based on patterns of volcanism - the dark rocks dotted throughout the landscape), but the geologic structures in the region had not been mapped in detail. Over two field seasons, we mapped the geology and major structures, building a picture of the tectonic events that led to various episodes of folding and faulting. We also collected samples for low-temperature thermochronology, which allows us to estimate exactly when major episodes of compression or extension occurred. My talk at AGU 2020 presents our first pass at interpreting all this data. We see evidence for very early compressional deformation in the area that began ~50 Ma, about 20 Myr earlier than previously thought. We also found evidence for an episode of transient, but significant, extension that created the modern Antofalla depression. The extension that we infer also occurred earlier than previously thought in this area (~16 Ma vs. only a few Ma).

Summary of Cenozoic tectonic events in the Antofalla region (Puna Plateau) inferred from geologic mapping, cross-sections, and low-temperature thermochronology. Black arrows show the location and timing of compressional events. Pink lines trace the edge of compressional deformation as it propagates outward from the core of major mountain ranges (Quebrada Honda and Calalaste) into the sedimentary basin. The blue shape illustrates an episode of transient extension inferred from sedimentary relationships and low-temperature thermochronology.

The next step is to bring these lines of research together by modelling the Antofalla dripping hypothesis specifically, comparing numerical predictions to geologic observations and data gathered in the field. This will involve examining all the geologic observations in the region collected by various workers, including my new data and comparing key observations (such as the spatial extent and timing of sedimentation, volcanism, and tectonic deformation) with those predicted from a geodynamic model tailored specifically for the region. Construction of such a specific geodynamic model has its own challenges, as it is difficult to know what the starting conditions of the model should be (How thick was the Puna Plateau’s crust 20 Myr ago? How hot and weak was it?). The project will most likely involve building a suite of models that are similar and realistic, but with small changes in initial conditions to test whether lithospheric dripping hypothesis is sensitive to any of these changes. If most of the models produce drips that match geologic observations, it strengthens the dripping hypothesis here.

Wind Erosion

The Salina del Fraile depression is a large (~300 sq. km) depression about 1 km deep, with a conspicuous rhomboidal shape. In some respects, it looks like a classic pull-apart basin created by strike-slip and normal faulting (i.e., tectonic extension) but we didn’t observe any such faults in the field. Instead, the major structures are clearly compressional, not extensional, and the floor of the depression is made up of older rocks. A series of anticlines (mapped as black arrows) seem to define the overall shape of the depression. This suggests that, instead of being lowered by faulting, the depression formed when rocks exhumed by the anticlines were eroded.

Map of wind-related landforms and their associated wind directions in the Salina del Fraile. Inset shows the wind directions from NASA climate reanalysis data over a 10 year period; the orientation of the strongest winds agrees very well with the wind directions inferred from various landforms. Strange landforms such as smoothed mesas and elongated ridges are also apparent.

Folds that exhume erodible strata in their cores, leading to them being hollowed out by erosion, are familiar phenomena. I mapped one such landform near Santa Fe, New Mexico as an undergrad student. There are also well-known ones in the Negev desert. The problem here? There’s no evidence of rivers or glaciers in this hyperarid region, especially ones that would be powerful enough to create such a large depression. My supervisor Lindsay Schoenbohm and I developed an alternate hypothesis: that the depression was carved by wind erosion. Once we made this inference, other oddities about the landform came into clearer focus, including the appearance of conspicuously smoothed topography, including large mesas and ridges, and gravel megaripples, large ripples created by strong winds that are made of gravel instead of sand. It seems that several million years of “normal” wind erosion (0.1 mm/yr) is enough to excavate a 1 km deep depression and to keep topographic slopes smooth, linear, and relatively free of debris. While we struggle to find similar landforms created by wind erosion on Earth, we did find that Mars has some smoothed topographic features, including large mesas and buttes that somewhat resemble the ones on the Puna. To read more, check out our paper published in the Journal of Geophysical Research: Earth Surface.

Although this project was originally a tangent from my lithospheric drip research, it actually has important implications there as well. The fact that we can now rule out a major episode of left-lateral transtension on a N-S trending fault (the original pull-apart basin hypothesis) means that the kinematic history of deformation is less complicated than previously thought. Multiple phases of ~E-W compressional deformation are evident from folded strata and unconformities exposed in the walls of the Salina del Fraile, and a later period ~NW-SE extension is evident from the pervasive normal faults and horst-and-graben structures in the SE portion of the map. Such a sequence of deformation may (or may not) be produced by a lithospheric drip.

Streambank Erosion