A Large Impact on Earth’s Geography

Title: On a Possible Giant Impact Origin for the Colorado Plateau

Author: Xiaolei Zhang

Lead Author’s Institution: Department of Physics and Astronomy, George Mason University

Status: Submitted to Earth and Planetary Science Letters [open access]

Author’s note: The discussion in this astrobite was written from the point of view that the hypothesis in the paper is correct – which of course remains to be seen by future studies.

Science requires collaboration, and often that collaboration leads to projects that span across different fields of science. This paper caught my eye mainly because the Colorado Plateau is one of my favorite places to travel, and though I’m not a geologist, its unique and beautiful rock formations have always piqued my interest. As I read this paper, I was surprised to find that this author also was not a geologist by trade. Xiaolei Zhang is an astrophysicist, but decided to carry on her late father’s unfinished project to uncover the formation of the Colorado Plateau and the possibility of giant-impact-plate tectonic motion.

By looking at planetary geology through an astronomer’s lens, it is clear that though discoveries of exo-solar planets are exciting, sometimes discoveries about humanity’s first known planet can be just as intriguing. There are still many things to learn about our home planet Earth. For example, geographical features on Earth can tell us about its astronomical history— its track record with meteors and other space debris.

Today’s paper discusses evidence Earth collided with a Mars-sized object roughly 750 Million years ago to form today’s Colorado Plateau. To put this in perspective, the impact that contributed to the mass extinction of dinosaurs was likely caused by an asteroid only about 10 km across.

This shows a bird’s eye view of the Colorado Plateau region, which could have been formed from a glancing impact with a Mars-sized planet.

The Colorado Plateau is the region around the Four Corners region of the southwestern United States. It is about 337,000 square kilometers and consists mostly of a high-desert plateau of a shallow bowl shape. The Plateau is a surprisingly stable region; the entire area does not have fault lines or features that indicate strong plate movement, which is a bit unusual given that its surroundings have these features. Sedimentary record shows that the entire North America continent, including the Colorado Plateau region, was very flat in early times, but now it has many mountains and sharp geographic features surrounding the area. These geographical peculiarities point towards an impact with an astronomical object significantly different from Earth forming this plateau.

The features discussed above indicate that this region was very likely formed by an impact, and since we are able to know the age of some of the features originating from the impact, we know at what time this collision happened. Zhang discusses that this impact likely happened 750 million years ago.

At this point in time, out solar system would have been very stable in its structure, so if an impact occurred, where could this object have come from? Zhang suggests the impacting object came from a spiral density wave as the solar system orbited the Milky Way; as the solar system orbited the Milky Way, it could have passed through a spiral arm of the galaxy. If this happened, a free-floating Mars exoplanet could have been drawn into the solar system and then crashed into Earth.

Zhang discusses multiple geographical features that provide evidence for the giant-impact theory about the Colorado Plateau. She also provides quantitative arguments to support the claim of the Mars impact. To review, these arguments include

  • Despite much igneous and orogenic activity at its boundaries, the plateau has remained structurally stable.
  • The type of rock throughout this area indicate evidence of shock metamorphism.
  • The basalt throughout the lowest level of rock also shows shock geographic formation. These features can also be dated around the time of the theorized impact.

This rare paper combines both geology and astrophysics spectacularly. By using the geological evidence and providing an astronomical theory, Zhang shows in a unique way how our planet got some of its features. Using interdisciplinary studies like this, we could learn much more about other planets within our solar system and newly-found exoplanets.

Zhang presents a quantifiable theory of a giant impact formation of the Colorado Plateau; though others have qualitatively mentioned this theory, this paper is the first to give a quantifiable estimate of size based on collisional dynamics. Zhang successfully completed this geological project by approaching it from her astrophysicist point of view, which is an incredibly important part of interdisciplinary science.  To me, this project exemplifies the collaborative nature of scientific work; not only does it span across different studies, it spans beyond the lifetime of just one person.

About Mara Zimmerman

Mara is working on her PhD in Astronomy at the University of Wyoming. She has done research with Heartbeat binary stars, and currently works on modeling the composition of debris disks. When not at her computer writing code, she likes hiking, baking scones, and making puns.

Discover more from astrobites

Subscribe to get the latest posts to your email.


  1. I’m a little disappointed by how un-critically this paper was read. Zhang is not a geologist, and in this case that’s a major liability, as she ignores decades of geological knowledge about how things like mountain ranges form. The confidence with which she characterizes rock types, again without consulting geologists, is extremely concerning. That would be like me, as an astrophysicist, looking at the internal anatomy of a bird and trying to guess its evolutionary history *without talking to a biologist*. And on top of the likely incorrect geological suppositions, Zhang even gets her *astrophysics* horribly wrong. It would take as many pages as the paper itself to fully document all the errors, but here a few major ones: a) it is unlikely in the extreme for a rogue planet to collide with Earth simply given the mean free path of small rocks in the Milky Way mid-plane, b) Zhang’s impact velocities are unphysical for her proposed scenario, c) Zhang’s speculation on the fate of the impactor is unphysical and inconsistent with her proposed origin (again the velocities are wrong), d) egregious misuse of crater scaling relations (if it gives you an unphysical answer that’s telling you something about its suitability), and e) an impact like this would indeed resurface the planet with a lava ocean and *unquestionably* extinguish all life, at a time when we know biodiversity was increasing exponentially.

    If anything, this paper is a great example of interdisciplinary science done incorrectly and arrogantly with disregard for the contributions of experts in relevant fields. And definitely an argument for not posting papers on arxiv as a single author without having other scientists critique the paper first. Zhang really should have included others in this work; they would have made a big difference.

    • (1) One of the founding fathers of the continental drift theory, Alfred Wegener, was trained as a meteorologist and an astronomer, not a geologist. For the establishment of a cross-disciplinary field, someone has to be willing to crossover from one discipline to another.

      (2) During the past few decades, the plate-tectonic paradigm propagated the idea that mountains are formed at the plate boundaries (i.e., at either converging, or diverging, plate boundaries). However, the mountains surrounding the Colorado Plateau are what we call intracratonic: the Rocky Mountains at the Plateau’s boundaries are within a single craton, and the formation of these mountains cannot be accounted for by the traditional plate tectonic theory (unless one cooks up unnatural, ad hoc scenarios). This well-known problem of the origin of Rocky Mountains was discussed in the book Annals of the Former World by John McPhee. Giant impact theory appears to be the most natural way to account for such continental mountainous features.

      (3) As shown by Zhang 1996, Astrophysical Journal 457, 125, the mean free path at the spiral density wave crest near the Galactic radius at the Sun’s orbit is around 1 kilo-parsec. Every spiral-arm crossing a stellar system would experience one encounter/scattering on average. Added to that, there is the formation of massive stars and their demise as supernovae. The blastwave of the supernovae could also perturb the neighboring stellar systems’ orbits.

      (4) The impact velocities assumed in the paper, of 10 km/sec relative to the Earth, is the average of what had been inferred for other impact events, i.e., the Barringer crater; as well as the K/T event studied by Alvarez et al. (the one that supposedly wiped out the dinosaurs).

      (5) The impactor, when first captured by our solar system, is not supposed to be in an equilibrium configuration with the then-mature solar system. The dissipative interactions with the solar system’s other members are expected to slowly change the eccentricity of the impactor’s orbit, making it rounder with time (since the minimum energy configuration of an orbit with a given amount of angular momentum is a circle — see Lynden-Bell and Kalnajs 1972). The dissipative exchange could also lead to mean orbital radius decay. These all could take many orbital cycles: as a matter of fact, there is a phase offset between the known major extinction peak epoch and the spiral arm crossing epoch, which could be partly accounted for by this slow secular orbital re-shaping effect for the impactor.

      Similarly, post impact the impactor is likely to be trapped inside the solar system for many more orbits, but could eventually escape due to the sling-shot effect of the gravitational tidal force from other planets (NASA makes use of this effect to save fuel for a journey into the outer solar system). If the escape effort fails, there is also the chance for the post-impact rogue planet to impact onto another solar system planet (i.e., Venus has a young crust, less than 500 million years old).

      (6) The crater scaling relation of Dence et al. (1977) is a classic result, which holds for many orders of magnitudes of crater size. This result was also quoted in the classic monograph Major Impact and Plate Tectonics by Price (2001). The current work used this result in the original context without modification.

      (7) As shown in the current paper, the energetics of the impact exactly accounted for the splitting of the supercontinent Rodinia at 750 million years ago, and the dislodge of the Colorado Plateau from the surrounding craton (without this structural severing, there will be no uplift of the entire Plateau area during the Cenozoic, after other major events mentioned in the paper). The energetics of the impact also is of the right order to account for the Great Unconformity which is observed all over the world (presumably due to impact vaporization of the upper crust).

      (8) As for the issue of major life extinction, there is in fact evidence that the Great Unconformity, which is likely a result of this giant impact event (arguments given in the paper), is related to the Cambrian explosion of biodiversity

      (see https://news.wisc.edu/evidence-for-a-geologic-trigger-of-the-cambrian-explosion/ ).

      So, rather than extinguishing life, this event may be the harbinger of the complex lifeforms that later developed on Earth (i.e., at 750 Ma when the impact event occurred, the lifeforms then were very primitive, which accounts for the placement of this period in the Proterozoic eon).

      I am currently in the process of expanding the paper into a full length paper from its current Letters length. In this updated version much more geological evidence and references will be given to support the proposed scenario. The author has so far done a thorough search of literature starting from those published more than 100 years ago (including the early classic book Pre-Cambrian Geology of North America by van Hise published in 1909). Several dozen binders of published research papers have been procured.

      • 1) If you want to delve into a field other than your own, it’s a good idea to bounce your ideas off of people already in the field, to see if they pass basic muster. Reading papers in the field is not the same thing, no matter how many you read.

        2) You can easily explain inland Rocky Mountain orogeny without “ad-hoc unnatural” scenarios by invoking shallow subduction of the Farallon Plate during the Late Cretaceous and accretion of land onto the Western North American continental shelf, such as the Farallon Islands, Vancouver Island (from the Kula Plate), and the Juan de Fuca Plate in San Francisco (one of several remnants of the Farallon Plate). And as Paul Spudis notes in a much more reasonable impact supposition (he does not invoke a giant impactor), https://www.airspacemag.com/daily-planet/could-colorado-plateau-be-ancient-impact-scar-180956994/, there is no evidence of shock-heated metamorphic rock in the Colorado Plateau. Given his geological credentials, I’d believe his claims over yours, absent rigorous peer-reviewed proof in the paper of shock-heated minerals, other than just supposing that it looks like there might be on visual inspection (Figure 5).

        3-5) Mean free path depends on interaction cross-section. You need to deliver a rogue planet into the inner solar system for it to even have a chance of dynamically relaxing into near-Earth orbit. The average stellar separation in the Disk is about 1 parsec, so even if you generously reduce that to 0.1 pc in a spiral arm, requiring a rogue planet to get close enough to Jupiter to get scattered in means you get a mean free path of about 1 trillion AU–or about 5 million light years. That’s why we call it a *collisionless* system. Even if you suppose that rogue planets outnumber stars 100-to-1 (highly questionable), you still get a mean free path of order tens of kpc–when the spiral arms are at best only a few kpc wide. And even then, you need the rogue planet to get scattered into the inner solar system just right to have an impact velocity of only a few km/s. In reality, an encounter with an interstellar rock (like `Oumuamua) should involve impact velocities *much higher* than typical solar system impacts.

        This makes the interaction unlikely, the impact velocity suspect, but more importantly, the scenario for *after* the impact is completely unphysical, because the escape velocity of the Earth is about 11 km/sec. If the impactor is not to be captured into Earth orbit, then 11 km/sec is an absolute minimum impact velocity. Even if the impactor has a much higher impact velocity (and therefore smaller radius), the likelihood of being scattered out of the solar system is much lower than the likelihood of being scattered into an eccentric and/or inclined but still-bound orbit. So from a purely dynamical perspective, the supposed impact velocity is not possible without accretion or capture of the impactor, the interaction in the first place is highly unlikely, and the supposed exit scenario requires a sequence of unlikely events. From a purely statistical perspective, this makes this hypothesis an extraordinary claim that must similarly require extremely extraordinary evidence.

        6-8) The crater scaling relation given in Dence 1997 is based on observed craters, which necessarily come exclusively from a population of small impactors hitting much larger bodies. Use of their scaling relation gives an impact diameter of order the planet’s radius for this collision because that’s an inappropriate extrapolation of the empirical relation–on those scales, the interaction ceases to behave like a rock hitting water, and instead is better-modeled by fluid interactions of two bodies. At the energy scales involved in a giant impact, the solid strength of the impactor and planet are not really significant, which is why there should not be a bounce–both behave like fluids. Even for smaller impacts, the collision is best modelled with two viscous fluids. Similarly, this brings up the point which must completely rule out a giant impact for the Colorado Plateau, and which was hinted by Greg Roelofs: the Roche limit for the Earth and a rigid Mars-sized impactor is over 17,000 kilometers. The impactor should be completely tidally-disrupted over the course of even a grazing collision, and much of the impact-facing side of Earth’s crust should also be disrupted by the impactor’s gravity–thus why this would be a life-ending event. There is an excellent reason that the only terrestrial giant impact in the solar system’s history for which we have good evidence resulted in a resurfacing event and the formation of the Moon. With an impactor this size, there are no glancing-blows: all collisions are cataclysmic.

        Ultimately, this hypothesis requires a great deal of special coincidences, weakening of several geophysical and astronomical theories including basic dynamics, and ends up being unlikely in the extreme.

        Far more likely: the Great Unconformity is the result of a long period of erosion followed by the submergence of much of North America in the late-Cambrian/early-Ordovician, resulting in a long period of sediment deposition, causing a large gap in sediment chronology. The mountains and folding of the Rockies and the orogens near the Colorado Plateau are the result of conventional subduction of ancient plates at shallow angles and widening of the continental margin through accretion, followed by as-yet-poorly-explained uplift of the Plateau.

        Personally, I’ll stick with Occam’s Razor in the absence of extraordinary evidence.

        • Mr. Paradise,

          You didn’t point out in your reply that I was the one pointing you to Paul Spudis’s link in a recent private email (and this link I had also included in my original arXiv paper as well). Pual Spudis mentioned that he was also not the one coming up with this idea originally, the idea of it was conveyed to him by his old professor at Arizona State University, Carleton Moore. Generally this idea has been around for a few decades, but nobody has given any quantitative details. I was not even aware of this paper when I first did my research and field studies for this project (since the Spudis article was a popular article, not a refereed scholarly article, and Spudis did not speculate on impactor size at all, nor any impact quantitative. So you are saying you’d believe his claims without quantitative supporting evidence, than a serious quantitative calculation, just because I had come from the field of astrophysics? Isn’t this kind of unquestioning behavior what underlies the current unhealthy academic climate of blindly following often erroneous fashionable trends in physics, in astrophysics, as well as in many other fields?).

          For the idea of the mean-free-path near the solar neighborhood, as I later also pointed to you in a private email, you can find it


          that paper’s pdf file is now free for the public to download. I rightly appreciate that when this paper was first published, you were quite young so I can’t expect you to have read it then (Prof. Colin Norman called this work “The most significant advance in galactic dynamics in the past decade” at the time when this paper was published). But now, you are a graduate student working for your PhD., there is all the reason to look through this classic paper, to understand that the spiral arm is a gravitational collisionless shock, and that means there is a local gravitational instability at the arm location, and within such an instability the spiral arm width is the effective mean-free-path of the stellar systems (not for direct collision, for which you will need to take into account cross section, but for scattering and perturbations that were caused by the arm as a whole).

          To emphasize: The galactic disk with a self-organized density wave mode (such as spirals and bars), is NOT a collisionless system. That tight coupling of the basic state of the galactic disk to the density wave pattern is what enabled the secular evolution of galaxies along the Hubble sequence.

          Subduction, as I previously commented, does not produce features as the tightly sandwiched pattern of alternatively-turned sedimentary laters observed for the Uncompahgre formation, as shown in Figure 4 of my article. A similar photo was also given in Blakey and Rooney 2018, Ancient Landscapes of Western North America, p. 58. This kind of tightly sandwiched pattern is evidence for strong and instantaneous shear force, the hallmark of giant impact, and it could not be modeled by fluid interactions.

          For Great Unconformity as an erosional feature, past geologists have wondered about that: where went the erosional remnant? Whereas as impact detritus, we have a lot of evidence, and my revised paper will quote that in great length.

          I have indeed talked to geologists as well, and the revised paper will thank them individually (starting from my father Prof. Hongren Zhang, who was a trained geologist, as well as his colleagues and collaborators on the giant-impact and plate tectonics project in China. That event they had studied, of Jurassic period in the Pacific, also assumed a Mars-sized impactor, though my father, being mainly a geologist rather than a dynamicist, did not given the details of impact dynamics calculations. Their papers were published in refereed journals starting in 1998, and my father’s last paper on this topic was published posthumously in 2016).

          I am not claiming escape is the only route. The follow-on impact into Venus, Jupiter and Saturn are also possible. Also, take a look at the following link for Uranus:


          Anyway, I am not sure it is productive to continue this discussion online, so please look for further updates on this project in my revised article.

  2. Such an impact should have far reaching effects – atmospheric, oceanic and identifiable debris outside of impact area. Not easy for me to accept premise without examination of these variables.

  3. Mars has a diameter of ~6800 km. The Colorado Plateau has an area of ~386000 km^2, which corresponds to a circular patch of ~1400 km diameter. This is far less than Mars-sized. Then throw in a ratio of ~20:1 for crater:impactor diameters.
    Mars-sized becomes even more unlikely.

    • See Figure 6 for impact geometry. Your argument is for small object, high speed impact cratering.

  4. 20 to 1 ratio of impactors to crater diameters is correct for meteors traveling 30 to 40 thousand mile an hour – essentially asteoids. Must assume this impactor is in earth similar orbit with similar speed.

  5. C-mon! Mars-sized object would shatter half of our planet and likely create another moon. If you say 1000 km object, i would think “hmmm, very unlikely”, 100 km object, yeah, maybe, possibly… But even that would be already very obvious in global geological record, would it?

  6. A Mars-sized exoplanet impactor? Well, if you are pushing dubious ideas, you might as well go for broke.

  7. The paper clarifies (as this summary should have!) that the proposed impact was a glancing blow.

    However…the velocities specified for the impactor, both before and after, are vastly less than the Solar System’s escape velocity (by a factor of two or three). This seems entirely inconsistent with the idea of a “rogue exoplanet,” i.e., one that fell into Sol’s potential from afar. I certainly see no possible way it could have gotten back out again, at least not without a ludicrous coincidence (i.e., passing close enough to one or more other planets to get a gravitational assist, or falling into the Sun without passing through and without triggering other solar changes).

    The geometry of the glancing blow in Figure 6 (16 km depth) also seems unphysical to me, implying a seriously large “bounce” effect; based on simulations I’ve seen, planets are more like jelly blobs when they interact. Even without direct touching, there appears to be no mention of tidal effects at all. (Caveat: I haven’t read the whole paper, but a search for the word “tidal” turns up nothing.) I don’t have a lot of intuition for this, but if you have two planetary bodies in extreme proximity, even for just a few seconds, I’d naively expect a huge amount of crustal stress, compounded with massive oceanic inundation, and maybe even sufficient torque to tweak Earth’s axial tilt.

    Finally, the abstract speaks of the correlation between mass extinctions and Earth’s passage through the spiral arms (density waves), which I understood to have been debunked on statistical grounds by now.

    • For entry and exit velocity issue, see reply to Adiv Paradise.

      As for the “bounce” effect, it is real: the simulations of giant impacts conducted so far nearly all used fluid/gas models with no strength of the solid Earth included. This is fine for the scenario of the formation of the moon, which occurred during the early phase of the Solar system (i.e., 4.5 billion years ago), when the Earth has not solidified. But it is not fine for the time period of 750 million years ago, when this particular impact event is supposed to have occurred. The strength of the Earth is what gives the bounce, and this strength also accelerates the Earth, per Newton’s third law, as the analysis in the paper shows.

      Also, take a look at Figure 4, that sandwiched pattern of sedimentary layers (i.e., the horizontal normally oriented layers in between the two vertically upturned sedimentary layers), could never be produced by a normal mountain building event at the plate boundary (i.e., subduction, which happens a few inches a year at the most), according to the usual plate tectonic theory. Such features, occurring on such a large scale, can only be the result of a rapid, high-intensity shear force, and very localized at that (note that Uncompahgre Formation is indeed at the Colorado Plateau’s boundary which is expected to experience such shear force).

      This shear force of the impact dislodged the entire Plateau from the surrounding craton, without it the Cenozoic uplift of the Plateau would not have been possible.

      As for the correlation of the spiral density wave and major extinction periods: The correlations are taken purely at face value, i.e., the period of spiral arm crossing is what it is, and the period of major extinction events (or major groups of them) is what it is. So one can doubt how well one knows each of these periods, but once known, there is no extra manipulation needed to arrive at the apparent correlation.


Leave a Reply