An Apple Pie From Scratch, Part Va: Tectonics: Construcing a Plate Tectonic History

USGS

It’s easy to imagine that, once a planet forms, its interior becomes a sort of independent realm, governed by its own laws and ignorant of the cosmos outside. In this model, the crust and its machinations are analogous to the thin scum that forms on a pot of soup; inconsequential byproducts, with no bearing on the whole.

But in reality, a better metaphor for the evolution of a planet’s interior might be a stage play: There’s a lot going on behind the scenes, and for a complex production the stagehands may outnumber the actors. But what happens on stage drives everything else, and a perceptive audience member can infer the nature of the whole beast from what they are shown.

Crustal tectonics is not a sideshow. It is an engine converting heat from the interior into motion, and what happens on a planet’s surface can affect the behavior of the interior as deep as the core. As a planet ages and cools, this engine operates in different modes which can have a profound effect both on the landforms and conditions on the surface and on the hospitability of the planet to complex life.

Simulating Plate Tectonics

Back to Part IVc

Earth Today

To start off, a quick tour: On average, the Earth’s radius is 6370 kilometers, but only the top 35 km is the crust (varying locally between 5 km at young sections of ocean and 70 km at old sections of continent). Underneath, the mantle stretches for over 2800 km to the boundary with the core.

Kelvinsong, Wikimedia

However, the distinction between the crust and mantle is based on chemical composition, as evidenced by the behavior of seismic waves passing through different materials. In terms of physical properties and motion, the uppermost region of the mantle is rigidly attached to crust and moves with the tectonic plates and so is often grouped with it in the lithosphere, 80 km thick (again, varying locally between 5 and 200 km). Below that, the asthenosphere stretching to 220 km depth is more ductile and fluid (though not liquid) and not attached to the plates, but does tend to be dragged along with them. The lower mantle—the mesosphere—stretches the rest of the way to the core-mantle boundary at 2980 km depth, is more uniform, and is largely composed of bridgmanite (MgSiO₃). Though it is at high temperature and can act as a fluid at large timescales (by continuously deforming), the high pressure keeps it mostly solid throughout.

Below that, the core is composed largely¹ of a 9:1 mix of iron and nickel, with around 3% lighter elements (mostly oxygen and sulfur, though possibly also a fair bit of hydrogen according to some recent research²). The outer core, stretching to 5150 km depth, is liquid in spite of the high pressure thanks to a temperature of 4000 to 5400 K (hotter with depth). Though the inner core is even hotter—as high as 6000 K—it is under even higher pressure, around 3,500,000 atm, such that it is forced back into a solid.

As mentioned in the last post, most new heat is generated in the mantle and crust by decaying radioactive isotopes, but the core retains a good deal of primordial heat and is the best-insulated part of the planet, so has the highest temperature. As it cools, iron from the liquid outer core crystallizes onto the surface of the inner core. The liquid just outside the inner core thus becomes iron-depleted and so lighter and more buoyant than the overlaying liquid. It rises to the core-mantle boundary and there loses its heat to the cooler mantle and gains back its iron by mixing with surrounding liquid, and so becomes less buoyant and sinks back down towards the inner core, thus forming a convection cell. The liquid metal is conductive, and so this motion—along with the Earth’s rotation—forms the planet’s magnetic field.

"Advection" is transfer of heat by magma reaching the surface. Bkilli 1, Wikimedia.

The mantle, now heated by the core, convects as well, though much slower given its high viscosity (again, the mantle is solid, not liquid, but through continuous plastic deformation—think of something like squeezing putty—it can gradually move and flow like a fluid). Its own radiogenic heat strengthens this current, and so once the plume of rock reaches the top of the mantle it breaks through the crust and reaches the surface. As the pressure drops the rock melts, and lava flows out of the seam opened in the crust. It quickly cools and solidifies, and is pushed aside by more lava bursting out and piling up around the seam. A ridge of new rock forms straddling this seam; the mid-ocean ridges we can see in many of our oceans today.

As this fresh rock is pushed further from the ridge, it cools and becomes denser, causing it to “float” lower on the underlying mantle, like a ship laden with cargo. This sinking helps to pull more material away from the ridge, and it’ll all move together as one tectonic plate. Eventually the rock will encounter another tectonic plate, and if that other plate is younger and less dense, then the rock will be pushed below it, forming a subduction zone. The rock sinks into the mantle, adding more pull to keep the rest of the plate behind it moving. It sinks lower, breaking up and mixing into the mantle as it goes, and eventually reaches the core-mantle boundary, closing the loop of mantle convection.

Surachit, Wikimedia/USGS

I’ve been emphasizing the pushing force from the mantle plume here to keep a narrative thread, but in reality the relative importance of “ridge-push” from mid-ocean ridges and “slab-pull” from subduction zones is a long-running debate, with the current consensus leaning towards the latter being far more important in most cases. I’ll explain more in a moment, but the key point is that subduction causes ridge spreading, not the latter, and in fact it also shapes the entire pattern of the mantle convection cell. Heat from the interior forces the mantle to convect, and would even if there was no tectonic motion on the surface at all (we'll discuss the implications of that next time), but the tectonic plates control where that convection happens.

If the original seam of lava bursting through the crust is underwater—as is usually the case—then the rock forming at the mid-ocean ridge will have some water mixed in with it. As it travels from mid-ocean ridge to subduction zone, it will also tend to accrue layers of sediment from material settling out of the water. These sediments will trap more water, and some—like carbonate minerals—will themselves be formed of other volatiles (carbon, hydrogen, sulfur, chlorine). As the plate subducts and warms up in the mantle, these volatiles will melt out of the rock and rise upwards, puncturing the overlaying plate and bursting through the surface as volcanoes. The volatiles themselves will mostly be released as gasses, but as they melt and rise, they bring some of the surrounding silicate rock with them. Heavy elements like iron will tend to sink out of the mix as it rises, and so once this magma reaches the surface, it will form iron-depleted felsic rock that is less dense than the mafic rock formed at mid-ocean ridges. Islands of this material will pile up on the overlaying plate along the subduction zone, forming an island arc.

This shows subduction-related volcanism along a continental coastline, but it's essentially the same mechanism as for island arcs. KDS4444, Wikimedia

Given enough time, the individual volcanoes can produce enough rock to form a plate of their own; a section of continental crust, as opposed to the oceanic crust that forms at mid-ocean ridges. This lighter and thicker crust floats higher on the mantle, causing much of its surface to rise above the oceans.

Note that continental crust extends further down into the mantle as well. USGS

Once formed, continents grow by various mountain-forming events called orogenies. When a continent encounters plates of oceanic crust, the oceanic crust always subducts under the continental crust. As it does, more subduction volcanism causes more rock to build up on the continent along the plate boundary. If the oceanic plate already carries some volcanic islands formed of felsic rock on it, these will collide with the continent and fuse with it, forcing some of the rock to fold up and form high mountain ranges.

Eventually subduction can pull two large continents together. They can continue to be pulled together for some time, pushing up a large plateau of folded rock, and one plate may even partially subduct under the other. But continental crust is too light to sink into the mantle, and so before long the plates will stop moving and fuse together. The thickened section of crust will be eroded from above and below, somewhat reducing the total amount of felsic rock (some rock is also pulled off the overlaying plate at subduction zones), but this is a far slower loss process than subduction and cannot destroy a continent completely, so continents are largely permanent once formed. Sections of the continents are billions of years old, while the oldest oceanic crust—in the eastern Mediterranean—is no more than 280 million years old, and most of the oceans are far younger.

NOAA

Thus, our modern planet is a patchwork of old continents and young oceans. Most of the tectonic plates contain sections of both types of crust, but the sections of oceanic crust are ephemeral attachments to the more persistent continental cores.

With these processes in mind, the boundaries between plates can all be classified into 3 types:

Red dots indicate earthquake epicenters. Note that in the continent-continent case (right) the continental crust will not completely subduct

Convergent boundaries, where plates are moving towards each other. These can be ocean-ocean subduction zones, such as the Marianas trench in the Eastern Pacific forming the Marianas island arc; ocean-continent subduction zones, such as the Peru-Chile trench off the western shore of South America, forming the Andes mountain range; or continent-continent collisions, such as the collision of India into Eurasia, forming the Himalayan plateau. The subduction obviously pulls in the subducting plate as the end is dragged down into the mountain, but there's also a degree of suction pulling the overlaying plate towards the subducting plate.

Exactly how subduction starts isn't fully understood, but once it gets going, it may slow or temporarily stall, but it seems the only way to stop subduction completely is a major collision between two continents; smaller collisions will only cause temporary interruptions, as we'll discuss in a bit more detail later.

domdomegg, Wikimedia

Divergent boundaries, where plates are spreading apart; mostly the mid-ocean ridges, such as in the Atlantic. But mantle plumes can also burst through continental crust, forcing the continent to spread apart and forming a rift valley, with volcanic ridgelines to either side (sometimes these plumes rise spontaneously on their own, sometimes continents are pulled apart by subduction zones and the mantle rises up to fill the gap). In fact, divergent boundaries almost always first form within continents, because despite being thicker than ocean crust, continental crust is also more brittle, and so will tear apart sooner under stress.

Before long, the continent is split apart, and a new ocean basin forms with a mid-ocean ridge. At this point the divergent boundary is essentially a passive feature; as subduction causes plates to pull apart, hot rock from the mantle rises up to fill the gap, solidifying onto the edges of the adjacent plates more-or-less symmetrically; the boundary thus appears to "move" over time based on the relative motion of the two plates (if, say, one plate is static while the other moves west at 10 cm/year, the divergent boundary will drift west by 5 cm/year due to this symmetrical plate growth). In addition to numerous mid-ocean ridges in the modern ocean, East Africa is undergoing rifting, though it’s unclear if this will ultimately split the continent apart or stall out before that happens.

Transverse boundaries, where plates are moving in different directions past each other—or in the same direction but at different rates. Friction between these plates can cause earthquakes and land deformation, such as near the San Andreas fault in western North America, but no crust is being formed or subducted, and there is no significant volcanism. However, it's fairly common for predominantly convergent or divergent boundaries to have small sections of transverse motion within them, and even large portions of boundaries can switch back and forth between transverse and their more typical motion.

In an idealized case such as the one I first described, a plate has a divergent boundary on one side and convergent on the other, such as is the case for the Nazca plate, forming a sort of conveyor belt of crust forming and then subducting at about the same. But this need not always be a case. The African plate is surrounding mostly by divergent boundaries, meaning it is growing on all sides, while the Filipino plate is surrounded mostly by convergent subduction zones and will be completely consumed before long. The total area of the plates must, of course, remain constant, but the different motions of all the plates can allow area to effectively be traded between plates that are not in contact (note, for example, how the western movement of the South American plate is shrinking the Nazca plate while also causing the African plate to grow, thus effectively transferring area from the former to the latter). Even where two plates are in contact, their complex shape or rotation of the plates can cause different types of boundaries to exist along the same border.

Major plates and the relative motion at their boundaries. USGS

No one plate can get too large, though; the older the oceans, the greater the chance they’ll subduct, and the larger the continent, the greater the chance that it will rift apart.

The key point here is that tectonic plates are not merely passengers on mantle convection. In a broad abstract sense the thermodynamics of the planet’s interior does provide the energy for plate motion, but once the plates start moving it is they that dictate the motion of rock through the mantle, not the other way around.

Patterns of Plate Motion

Now that we have a handle on the basic mechanisms of plate tectonics, let's zoom out a bit and look at how these basic processes of subduction and rifting work out in terms of the motions of plates and continents. To be clear, a lot of the details of exactly how plate tectonics operates on a fundamental level and what patterns are possible are not fully known. Most of the established "rules" of plate tectonics³ are known based not on a full understanding of the mechanisms involved but simply what patterns we see in Earth's past, and that record is fairly murky; we have a fairly decent picture of the last 600 million years or so, but before that the geological record becomes fairly patchy and we're not even entirely sure when plate tectonics began (though as we'll discuss in the next post, it may have been a gradual transition through increasingly plate-tectonics-like regimes).

One consistent pattern we do see is the repeated appearance of supercontinents, assemblages of all or most of the world's major landmasses into a single large continent. You've likely heard of the supercontinent Pangea 250 million years ago, but before that was another supercontinent Pannotia 600 million years ago, Rodinia about 1 billion years ago, and Nuna 1.8 billion years ago, with some vague hints in the geological record of another potential supercontinent Kenorland 2.5 billion years ago.

Altogether this suggests a supercontinent cycle where the continents rift apart and then join back together. The exact pacing depends on how you count Pannotia, which is somewhat the odd one out, never quite including all major continents and perhaps better to consider as just a late stage in Rodinia's breakup; if we exclude Pannotia, that implies a consistent 750-million year cycle, but if we include it and also consider that projections of Earth's future tectonic motion⁴ generally predict assembly of a new supercontinent in about 200-250 million years, that may imply we've shifted from a slower cycle into a quicker 400-500 million year supercontinent cycle. The other past few cycles for which we have detailed records are also not without their quirks: Rodinia started rifting apart before it was fully assembled and Pangea was continuously spalling off microcontinents from its eastern coast all throughout its tenure.

But even if plate motion doesn't always neatly conform to an ideal supercontinent cycle, it still provides a convenient framework for constructing our own continents: Every continental world should be in some stage of the breakup, assembly, or tenure of a supercontinent, and it's generally easiest to start designing a tectonic history during a supercontinent's tenure, when there are the fewest pieces in motion.

Broadly speaking there seem to be two main types of supercontinent cycle: extroversion, where continents rift apart and then circle around the planet to collide back together to form a new supercontinent on more-or-less the opposite side of the world from the original supercontinent; and introversion, where the continents move apart for a while, but then reverse course and collide back together and form a new superconinent about the same positions as the old supercontinent.

Earth in cross-section through different types of supercontinent cycles. Bradley 2011

Though the exact causes for one or the other cycle or supercontinent cycles in general aren't totally clear, it may help to look at some idealized examples of what these cycles look like and what features continents will have throughout them.

Extroversion

Let's start there with a simple supercontinent, flanked by subduction zones.

For these diagrams, the top area is a map with red, green, and blue lines showing divergent, transverse, and convergent boundaries (with "teeth" on subduction zones pointing to the overlaying plate), arrows indicating plate motion, brown areas indicating mountain formation, and maroon dots showing volcanism; and the bottom area shows a not-to-scale cross-section through the crust in the same area, with thicker, lighter continental crust and thinner ocean crust.

If there's a mid-ocean ridge in the ocean consistently forming new ocean crust to feed into these subduction zones, this situation could be quite stable, and indeed some past supercontinents seem to have sat more-or-less in stasis for hundreds of millions of years (though this may also reflect shifts in the nature of Earth's tectonics over time; again, more on this in the next post). But a large supercontinent insulates the mantle below it, and the trapped heat causes plumes of hot rock to form that eventually rise and burst through the surface in the form of hotspots—isolated volcanoes—and flood volcanism—vast regions of volcanic activity and lava flows (it’s also possible⁵ the plumes have less to do with insulation and more with how patterns of subduction around the supercontinent affect mantle convection below it, but at any rate these plumes do consistently form under supercontinents). These plumes do not themselves cause rifting, but they form weak points in the crust which allows for subductions zones outside the continent to pull it apart. They also push the continental crust up slightly, creating a slight slope between the supercontinent core and surrounding subduction zones that the continents can slide down.

Source

New rifts usually form initially at a triple junction, where initial tearing of the crust at a weak point spreads outwards in 3 directions.

Eventually the spreading rifts from neighboring triple junctions join together into a single zig-zag shaped rift, which you can still see in, for example, the overall shape of the modern Atlantic coasts.

The third rifts from the triple junctions that don’t join with others become failed rifts (also called aulacogens) and cease spreading, though they remain as weak points in the crust and can be "reactivated" later if there's ever further rifting. They also tend to erode easier than the surrounding crust and so often become parts of later river valleys, or may even form shallow inlets or ocean channels.

Rifts formed in the Pangea breakup, which major associated rivers noted as well. Source.

The rift zones will appear first as volcanically active rift valleys, but eventually split open into new ocean basins with mid-ocean ridges down their middles. Rifting often isn’t clean or simultaneous; Africa split from North America long before South America, and North America and Eurasia are still partially connected today over the Bering Strait.

USGS

But rifting can accelerate quite rapidly once underway, and this seems to be one of the few situations where outward pushing from the ridge significantly contributes to plate motion.

None of this is to say that large, supercontinent-induced mantle plumes are required for rifting, incidentally—similar plumes might occasionally form under smaller continents, and competing subduction zones can also pull a continent apart; as the crust thins and tears, this will tend to encourage some upwelling of hot mantle rock anyway—but it helps contribute to the particularly vigorous rifting of supercontinents during breakup.

Anyway, at this point most continents will show a fairly typical asymmetric profile, with modern South America being a good example:

The coast facing the direction of motion, with the subduction zone, forms an active margin, with a deep ocean trench just offshore, a volcanic orogeny onshore that pushes up a high mountain range, and a steep coastline between them. The leeward coast forms a passive margin, with well-eroded topography, vast coastal plains, and wide continental shelves (though as we'll note later, it's common for there to be some old eroded mountains there as well).

Though I'm mostly keeping to linear motion with two continents in my examples here for simplicity, more typically you'll often have several continents rifting away in all directions, forming a ring of landmasses and island arcs dividing a young, expanding interior ocean surrounded by passive margins from an older, shrinking exterior ocean surrounded by active margins.

But of course reality is rarely so neat; you may have some continents like Africa surrounded by passive margins that are essentially "left behind" in the interior; you may have precocious continents that rift away from the supercontinent early—or never joined it at all—and so end up isolated in the exterior ocean; and as we'll discuss new subduction zones may form in the interior ocean, forming more complex patterns.

But returning to our simple case, these continents can only move apart so far; as the exterior ocean shrinks, eventually the continents will circle around the planet and meet up again on the far side.

How long the exterior ocean survives often depends on the state of the mid-ocean ridges within it. A situation like we see here, with a central mid-ocean ridge feeding ocean crust directly into surrounding subduction zones, can be quite stable, with the surrounding continents moving slowly or perhaps even stalling out entirely. But in many cases there's likely to be some slight mismatch in the speed of ocean crust subduction or continental motion, which will cause the mid-ocean rift to drift towards one side or the other, until it eventually subducts.

At this point there's no new crust forming to accomodate the subduction around the edges of the exterior ocean. It's not impossible for new rifting to start in the ocean—a new divergent boundary appears to be currently forming in the Indian ocean as India and Australia are being pulled in different directions—but more typically the competing pull of the continents on the same plate will cause them to be pulled together, ultimately closing the former exterior ocean as the last of its crust is subducted and the continents collide.

I've set up a very symmetrical collision here, but of course real collisions will vary and the exact details of how different sections of crust move about in the final stages of collision can be quite complex, but what typically ultimately happens is that one continent partially subducts under the other. As the two continents push into each other, a tall mountain range is pushed up between them, but volcanic activity stops without a supply of volatiles from subducting ocean crust.

The buoyancy of the continental crust prevents it from sinking more than a couple hundred km into the mantle, so eventually subduction finally stops (and the already subducted ocean crust detaches from the continent and sinks into the mantle). The two continents suture together, forming a single landmass with a common plate that will now move as one, but fractures and faults formed during the collision will remain as weaknesses where new rifts may form if the continent is ever pulled apart again, and even without fully rifting plates often aren't completely rigid and can have a bit of internal sliding and stretching long after collisions.

The mountain range also remains, and will gradually erode down over millions of years. This is a point worth emphasizing because it's often missed in more simplistic approaches to mapping with tectonics: not all mountain ranges are associated with current tectonic boundaries. All of Earth's continents have remnants of mountains formed in past collisions that are far from any plate boundaries and have long since stopped seeing any significant relative motion.

At any rate, in our simple 2-continent world, that completes a supercontinent extroversion cycle: areas that were once the interior of the old supercontinent are now on the coasts, and the old coastlines are now mostly trapped in the interior.

Introversion

Let's follow our recently-formed supercontinent through another supercontinent cycle. For new rifting to start, we need new subduction to start, but we don't fully understand how this actually happens. As the ocean crust around the edges of the continent's passive margin ages, it will become denser and less buoyant, but also stiffens, so some additional process is needed to fracture the crust. Broadly speaking⁶ there are three concepts for how this occurs (excluding cases we'll discuss in a bit where one subduction zone induces formation of another), all of which might happen in different cases:

We'll discuss the induced cases later. Stern and Gerya 2018

Passive margin collapse: In some cases the ocean crust may break away from the continental crust and start subducting directly along the coastline, but this seems to be quite rare and may require preexisting faults from previous tectonic stresses.
Transform collapse: Probably the most common mechanism; where there is a transform fault between two sections of crust of different age, the older side may eventually sink and subduct under the other. This may happen at plate transverse boundaries, but smaller transform faults inevitably form in growing oceans as different sections of the divergent boundary spread at different rates; the geometry of plate motion over a spherical surface makes these faults inevitable, so they can expected to be present in every ocean.
Plume head margin collapse: Large plumes of upwelling hot rock around young hotspots may push up some crust while tugging down on crust around the plume, eventually starting subduction around its edges.

Once subduction starts, the subducting slab will pull on the crust around it, fracturing the crust or stressing preexisting faults, such that the subduction zone spreads outwards. Some researchers even suggest that formation of wholly new subduction zones from scratch is rare, and instead new subduction zones mostly form by spreading out of old zones.

Regardless, subduction seems to be more likely as the ocean crust ages, and in the time it's taken for our continents to circle the planet, the adjacent crust must have gotten pretty old, so new subduction zones will likely appear around the coasts eventually, turning the former passive margins into new active margins.

We're now in essentially the same situation we started with in the first supercontinent cyle, so we can rift the supercontinent apart again and get going. New rifts between continents often form close to where they previously sutured together, but generally not along the exact same lines, so as continents join and then rift back apart, there's essentially always some exchange of landmass between them.

As I alluded to earlier, this tends to place the old mountains formed in assembly of the supercontinent along the passive margin of this new continent, while new mountains form along the active margin. These older mountains will be significantly eroded down by now and will have wider coastal plains between them and the sea than the newer mountains, but it's still worth noting that, contrary to the common tendency of many inexperienced mappers and simplistic map generators to make continents with high interiors sloping down to low coasts, the more common (but not ubiquitous) tendency is for continents to have low interior basins surrounded by coastal mountain ranges of various ages. This also incidentally helps form large river basins like the Amazon or Mississippi, as these large interior basins have to drain out through the few gaps between mountain ranges.

Anyway, as the continents spread apart, what was the interior ocean of the last supercontinent cycle becomes the exterior ocean of this one, and a new interior ocean forms between them. If we did another extroversion cycle, we could keep endlessly bouncing these continents between either side of the world. But this time, let's imagine that a new subduction zone forms a bit quicker within the interior ocean before a new supercontinent forms.

The interior ocean is now being subducted on only one side, so the mid-ocean ridge quickly subducts into the new subduction zone. The eastern continent here is then directly subducting the plate of the western continent, pulling them back together (presuming the exterior ocean still has mid-ocean ridges that can feed new ocean crust into the outside subduction zones; if not, this would likely cause the western continent to rift in two, with one half continuing west into the exterior ocean and the other half heading east).

Naturally this eventually leads to a collision, likely a somewhat more vigorous one with one continent directly subducting the other. The continents likely won't collide back together along the same line they rifted along, but nevertheless the interior of the old supercontinent is mostly the interior of the new supercontinent, the coasts remain the coasts, and the exterior ocean survives while the interior ocean closes.

Even aside from global introversion cycles, this process of rifting, spreading, forming a new interior subduction zone, and colliding back together, called a Wilson cycle, is also fairly common at smaller scales (and even a pair of extroversion cycles can be considered a long Wilson cycle). Parts of modern North America and Europe have collided and rifted apart twice over the last couple billion years, and several projections of future tectonic motion predict they may be on course for a third collision.

Fabrichter, Wikimedia

To a certain extent, supercontinent cycles may just be an inevitable result of the geometry of plate motion and tendencies of the processes involved: aging oceans will prompt the formation of new subduction zones, which will pull continents together, and so long as continents suture together faster than they rift apart—which generally seems to be the case except in the early stages of supercontinent breakup, when the new continents are all moving apart into a large exterior ocean—they will inevitably join into a single supercontinent eventually.

But convection of the mantle may also play a role: upwelling of the mantle pushes the crust slightly up, around the position of the former supercontinent, and downwelling in the mantle elsewhere may pull the mantle slightly downwards. Though the geometry of subduction zones and other plate boundaries still determines the exact movement of plates (and may influence where downwelling occurs), this subtle gradient in the crust may help bias their motion enough to encourage supercontinent assembly. Shifts in mantle convection may help influence whether an extroversion or introversion cycle occurs, but it may also come down to chance and circumstance in terms of how quickly new subduction zones form in the interior ocean and how long mid-ocean ridges survive in either ocean.

But really any particular supercontinent cycle will tend to have versions of both⁷ cycles in separate ocean basins, and some researchers also distinguish a third in-between pattern, orthoversion, where continents do a sort of half-turn before colliding somewhere between their starting point and the opposite side of the world without fully closing either interior or exterior ocean.

Of the last several cycles, Nuna to Rodinia was probably mostly an extroversion cycle but many of the details are still unclear; Rodinia to Pannotia might be counted as either an extroversion or orthoversion cycle; and Pannotia to Pangea was largely introversion but again with elements of orthoversion.

Competing models of future tectonic motion largely come down to different predictions of the nature of the current supercontinent cycle: the Novopangea model predicts continued extroversion, closing the Pacific; the Pangea Ultima model predicts introversion, closing the Atlantic; the Amasia model predicts orthoversion, with continents mostly shifting north and closing the Arctic; and the Aurica model predicts a mix of introversion and extroversion, with both the Atlantic and Pacific closing and Eurasia rifting apart to form a new ocean.

White lines indicate divergent zones and thick black lines along coasts show subduction zones. Davies et al. 2018

At any rate, now that we've got a broad sense of the most common patterns of global tectonic motion, I'd like to go over a few more unusual cases we might encounter on more local scales. I can't predict every eventuality that might turn up, but the essential rule to keep in mind throughout is that, again, subduction is active and rifting and ridge spreading is passive. Subduction zones dictate the movement of plates, though the relative importance of different subduction zones seems to depend on mantle processes that aren’t fully understood. The best interpretation right now is that mantle convection forces the supercontinent cycle to occur, but subduction determines all the details.

Flat-Slab Subduction

Usually the mountain range along an active margin is fairly narrow and steep, such that it might almost appear as a sheer wall on the continent's coastline in large maps, but in some cases a continent may experience flat-slab subduction, where a subducting plate doesn’t immediately dive into the mantle but slides underneath the continent for some distance before finally turning downwards.

This causes uplift and deformation far into the interior of the continent, forming a much broader mountain range, and volcanic activity either shifts further inland or may stop entirely. Even inland of the mountain range itself, much of the continent's interior may be uplifted. This is partially how the Rockies in North America formed, pushing up much of the continent in the process to form the gradient across much of the interior, and more recent flat-slab subduction has contributed to some of the broader sections of the Andes in South America.

Exactly why this happens isn't totally clear, though there are two main proposals: first, that it's caused by (relatively) rapid motion of the continent⁸ towards the subduction zone, such that it's essentially overrunning the subduction zone and ocean crust behind it faster than it can subduct; second, that it's caused by subduction of young crust too buoyant to subduct normally, or even subduction of a mid-ocean ridge⁹, which in particular can leave open a "window" in the subducting ocean crust through which hot mantle rock can rise and push up the continent.

EOS.org, modified from Finzel et al. 2016.

For the most part this doesn't change much about the continent's motion, but in North America's case, subduction of mid-ocean ridges and so breaking of the subducting slabs—though also the somewhat oblique relative motion of the North American and Pacific plates—also seems to have caused subduction to stall out, with only transverse motion on much of the west coast, but this is probably temporary and subduction will resume here eventually; though whether this is necessarily always the case when mid-ocean ridges are subducted is not fully settled.

Slab Rollback

Converse to the rapid motion that may cause flat-slab subduction, this is caused by

slow subduction, with ocean crust being subducted slower than it is produced. This means that the age of the ocean crust at the time it subducts is increasing over time. Thus the subducting crust is becoming progressively denser and less buoyant, and so it tends to sink quicker and steeper into the mantle. This produces a degree of suction on the overlaying plate, pulling it towards the subducting plate.

Niu 2014

The general result is that parts of the coastline tear away from the continent and migrate out to sea with the subduction zone. Volcanism and uplift still continue along the subduction zone, forming an offshore island arc. Behind this arc, the continental crust is stretched and thinned out, forming a back-arc basin. (though note that "back-arc basin" is sometimes also used to refer to the interior lowlands behind more regular coastal orogenies as well).

Fractures in the crust turn into hotspots, forming more volcanic islands. New oceanic crust can even form in the basin, but there is no mid-ocean ridge and the basin won't ever open up into a new ocean.

This situation might arise whenever a continent has subduction zones on opposite coasts, such as during an introversion cycle—when a new subduction forms on the interior ocean side but the old subduction zone is still active on the exterior ocean side—or in the later parts of an extroversion cycle if new subduction zones start forming around the interior ocean before the continents have collided. As the continent moves, the leading edge will have a regular active margin with a coastal range, but the trailing edge will be pulled away from the mid-ocean ridge feeding it ocean crust, and so tend to experience slab rollback.

But slab rollback can also emerge in other cases of slow continent motion or restricted ocean crust motion. The east coast of asia is experiencing slab rollback today due to the slow motion of Eurasia and eastward shift of mid-ocean ridges in the Pacific ocean, forming an offshore island arc in the form of Japan and the neighboring Kuril and Ryuku islands.

If the continent’s motion shifts such that slab rollback stops (e.g. if it reverses course or collides with another continent on its leading edge) the subduction zone and island arc will be pushed back into the mainland and transition into a more typical active margin, though a fair bit of ocean crust from the back-arc basin may be scooped onto the continent in the process, which may contain some resources like chromium that are relatively rare in continental crust.

Island Arc Accretion

When new subduction zones form along a continent coastline, they need not form exactly on the coast; though they typically start near the coast, they may arc out into the ocean as they spread along preexisting faults and weaknesses. They'll then form a new island arc that can reach far out to sea.

Again, arcs are usually connected to continents at one end or the other, this is just a conceptual diagram.

But the ocean crust between the island arc and continent will continue to age, so sooner or later another subduction zone is likely to form, particularly if the continent starts moving in the direction of the island arc. This naturally pulls the island arc into a collision with the continent.

The new continental crust formed in the island arc will join the continent, pushing up collision mountain ranges as it does. Subduction continues on the new continental coastline, such that the combined collision and continued coastal volcanism forms a single broad coastal mountain range.

This tendency to form offshore island arcs which then get pulled in to accrete onto the continent is a fairly common way continents grow over time, to the point that a large portion of modern continental crust first formed as island arcs. North America has recently accreted several island arcs along its west coast (which is another contributing factor to the broad mountain ranges there), and Asia is in the early stages of accreting the Philippines and Marianas arcs.

But while we're at it, there are some slightly more complicated cases of accretion worth mentioning as well:

Subduction Jumping

In this case, imagine that rather than an island arc with its own subduction zone, a continent is pulling in a microcontinent, a smallish landmass that has rifted off some other continent.

The continent and microcontinent will collide, closing the subduction zone and forming a collisional mountain range. However, if the microcontinent is fairly thin, subduction may not stop at this point: instead, the subduction zone will "jump" across the continent, starting up on the new coastline. As with much of the mechanisms of subduction, the exact details of how this occurs aren't fully understood, but it seems the stress of the collision may cause the ocean crust to fracture behind the microcontinent, allowing it to subduct.

This can also happen with oceanic plateaus (thick sections of ocean crust formed by seafloor volcanism), with the result that subduction effectively moves offshore, but the plateau will likely be pushed up into the continent eventually.

Polarity Reversal

In this case, imagine that there initially is no subduction on the continent's coast, and instead an island arc or microcontinent is subducting the ocean crust attached to the continent.

As per usual, this eventually results in a collision. But as in the last case, subduction likely won't stop: the stress of collision fractures the crust behind the arc, and new subduction starts on the continent's new coastline. The oddity here is that subduction switches direction, so that what was once the subducting plate now becomes the overlaying plate.

You can also imagine a final scenario where an island arc and continent (or a second island arc) are both subducting a small ocean plate between them.

Exactly how this plays out can vary, but ultimately with the same end result: the two landmasses collide and subduction continues on the new coastline of the continent.

Subduction jumping and polarity reversal can also occur when island arcs or microcontinents collide, though which will occur in which circumstance can vary; in general the side with older ocean crust is more likely to subduct, but this isn't a universal rule.

The prevalence of these mechanisms means that essentially the only way to permanently stop a subduction zone is with a major collision between two large continents. The degree to which this is always strictly true for every case is debateable, but in broad terms when thinking of tectonic motion over hundreds of millions of years, it's a decent rule to follow.

Rotation and Arc Formation

I've already alluded to this a couple times, but it's worth highlighting explicitly: long ocean island arcs might also form progressively as two landmasses rift apart. If their rift cuts through a subduction zone, rather than splitting apart, the subduction zone will often tend to grow with the new ocean, bridging the two continents:

The geometry of motion on a spherical surface means plates can't move perfectly apart, and indeed the more common tendency is for a more hinge-like motion due uneven forces. This rotational motion is part of why subduction zones and island chains so often have their characteristic arc shape; on a flat world with plate tectonics, tectonic boundaries and island chains would tend to be straighter.

These island arcs can become quite long and stretch far from the continent, but because subduction usually starts near coastlines, you shouldn't expect to see island arcs completely isolated in the middle of the ocean.

Subduction Invasion

As continents move apart during supercontinent rifting, gaps will necessarily appear between them. The formation of island arcs as described above will tend to fill these gaps, potentially forming a complete ring of subduction dividing the interior and exterior ocean, as I showed for the extroversion case. But this isn't always the case; sometimes instead transverse boundaries can appear in these gaps, or the overall pattern of motion can just be a bit more complicated than landmasses neatly moving apart in a ring. This can allow for parts of the exterior ocean to start subducting the interior ocean—"invading" it—in a couple different ways:

First, a transverse boundary separating the oceans might develop into a subduction zone by transform collapse. Usually the interior ocean will tend to have younger crust than the exterior ocean, and so this new subduction zone will subduct the exterior ocean, but this may not always be the case if new crust is still forming in the exterior ocean.

Second, a landmass in the exterior ocean may collide with one of the subduction zones. This may be a large continent that rifted away from the supercontinent early and has since reversed course, but even a small landmass could trigger polarity reversal.

In either case, this new subduction zone could start rapidly expanding into the interior ocean, and may spread to the coasts of nearby continents. Indeed, subduction invasion events may be a major element of what causes introversion cycles (though it's hard to tell because they can be difficult to reconstruct from geological evidence).

Subduction invasion from the Pacific into the Atlantic is currently underway around the Caribbean and Scotia seas, which is what leads some researchers to suggest we're bound for an introversion cycle.

Triple-Junction Ocean Plate Formation

For this case, consider an ocean composed of three plates all diverging from each other, forming three mid-ocean ridges. For simplicity's sake we might imagine this as an exterior ocean surrounded by subduction zones, but that doesn't necessarily have to be the case. If it is an exterior ocean, then this is a fairly stable configuration so long as all the ocean plates are diverging at about the same pace.

Where the mid-ocean ridges intersect at a triple junction, new ocean crust will usually form on all three sides, attached to the existing plates. However, if the plates are not all moving directly apart but instead rotating around the triple junction to some extent, this twisting motion might cause a gap to open at the triple junction (the exact geometry is a bit tricky to explain and involves some extra elements like a short-lived subduction zone, so I refer you to this paper¹⁰ for details).

Mantle rock will rise to fill the gap, forming new ocean crust, but it may not be attached to any of the existing plates, and so will form its own new ocean plate. As the surrounding ocean plates continue to move away, new mid-ocean ridges will form around this new section of crust, adding more crust to its sides and causing it to grow. The old plates, meanwhile, continue to subduct at the ocean's edges. This asymmetric motion—the new plate remaining relatively still while the old plates move towards the subduction zones—causes the mid-ocean ridges to also drift out towards the edges of the ocean.

Eventually the old plates will subduct completely, along with the mid-ocean ridges. This may leave the ocean with no source of new ocean crust and so force the surrounding subduction zones to converge. So the formation of a new ocean plate like this can destabilize a previous stable ocean basin and force its closure.

This is generally believed to be how the Pacific ocean formed, though without much data on ocean plate motion older than around 200 million years ago, we can't say much about how common it is in general. In the Pacific's case, most of the surrounding plates have subducted, but one still survives in the Nazca and Cocos plates to its southeast, which could conceivably allow the ocean to survive if the Pacific plate starts moving quicker to the west and the mid-ocean ridge shifts towards the ocean's center.

(I had trouble finding a good animation of this one, but this one is fairly decent for at least the early stages; pay attention to the left side of the map).

Continental Megashear

This is a slightly tricky one but I thought it could be worth mentioning: Some ambiguities in the geological record (specifically, paleomagnetic data from Australia that doesn't neatly match up with typical reconstructions of Pangea's formation) have led some researchers to suggest¹¹ that Pangea initially formed in a different configuration, "Pangea B", with much of the northern part of the supercontinent farther west than typically reconstructed. The two halves of the supercontinent would then have slid into their more familiar "Pangea A" configuration over 20 million years along one enormous 6,000-km-long transverse boundary, sometimes called a megashear system.

Muttoni et al. 2003

This model has never gained wide acceptance, with recent studies¹² finding more evidence to support the conventional reconstructions and alternate explanations for the anomalous data. However, as far as I can tell, these studies don't seem to question the basic plausibility of such a megashear event in principle, just it's occurence in this particular case. This may just be due to our uncertainties about the mechanisms of plate tectonics, and if we discard the Pangea B model that doesn't leave us with any other clear examples from Earth's past to go by (though smaller shear events are fairly common), but at any rate a megashear scenario is perhaps still worth considering as a possibility if the configuration of subduction zones favors it.

Tethys-Type Oceans

Our last two cases are essentially just combinations of several of the mechanisms we've discussed, but worth working through as examples. In this case, imagine a large, arc-shaped continent has subduction along part of the inside face of that arc, which might happen due to a collision that only closes part of a subduction zone, or just the appearance of new subduction on the coast of a preexisting continent.

In this case, the subduction zone in the west is directly pulling on the land on the east side of the sea, but because they're connected by the arc of the continent they can't directly collide together. This could perhaps lead to a megathrust scenario as described above, but more typically the result is that part of the continent on the sea's eastern coast tears away (because as mentioned, continental crust is more brittle and will tear before ocean crust) and moves west to collide with the island arc.

If this is a fairly small microcontinent, then subduction will jump across it and carry on. There's now a mid-ocean ridge in the sea, but as we saw with the later stages of introversion, subduction on one side of the ocean will eventually consume the mid-ocean ridge, bringing us essentially back around to where we started.

Stopping the subduction zone for good requires a large continent to break away (or enter the situation from elsewhere) and collide with the subducting shore. As per usual, this forms a collisional mountain range. But if one of the continents has recently had several collisions with microcontinents or island arcs (which doesn't necessarily require this particular Tethys-type setup), that will leave numerous weaknesses and faults which may begin to shift again in this collision. Combined with the fairly rapid collision that tends to result from one-sided subduction like this, that can cause a particularly broad and complex mountain plateau to form.

As the name implies, the classic example of this is the Tethys ocean to the East of Pangea, which had a subduction zone along its northern boundary, which progressively pulled in several microcontinents from the southern regions of Pangea before finally pulling in India and forming the Tibetan plateau (I couldn't find a good video of this whole process specifically, but you can see it if you scroll back up to the global tectonics video and pay attention to the ocean to Pangea's east from about 400 million years ago onwards).

Complex Collisions

Finally, the examples I've given for landmass collisions have all been fairly straightforward, with straight boundaries, but this may of course not always be the case. If two continents with more complex coastlines converge, some areas may start colliding while others are still divided by sections of ocean.

What happens then can be quite complex, as the collisions cause different sections of crust to fracture and shift, and the small isolated subduction zones can tear off pieces of crust around them, undergo slab rollback, or invade neighboring sections of ocean. One particular curious pattern is the potential for slab rollback and back-arc spreading in these small spaces to allow a subduction zone to move forward into collision with another landmass, rather than pulling a landmass in to collide with it. Stretching of the crust and hotspot activity can form numerous islands in the back-arc basin.

The Mediterranean is the classic example of this, representing the early stages of a collision between Africa and Eurasia, with local subduction and collision forming its mix of islands, peninsulas, and ridgelines, with slab rollback in the Tyrrhenian, Aegian, and Levantine seas and the Pannonian basin.

(Note that the start of this video moves back in time)from the present day, then forward again in the latter half.)

Other Considerations

Plate speed depends on a couple different factors. Subducting plates move fastest due to slab pull, and subduction even helps pull overlaying plates through slab suction. Ridge push has a secondary effect, which is stronger for plates spreading away from a former supercontinent thanks to the slight slope created by mantle plumes—but to be clear, a continent can only be moved by ridge push if there is subduction elsewhere helping it along. Countering these forces is drag from the mantle, which has a stronger effect on continents because they have deeper lithospheric “roots”. As points of reference:

India, on a mostly-oceanic subducting plate, topped out at 20 centimeters/year in the late Cretaceous, and even after colliding with Asia is still moving north at around 6 cm/yr.
The Nazca plate, a subducting oceanic plate, is moving east¹³ at 10 cm/yr, and the Pacific plate west at around 8 cm/yr.
North and South America, large continents with large active margins, are both moving west at 3 cm/yr.
When it initially starting rifting from Africa, North America only moved at 1 cm/yr for the first 25 million years, before abruptly accelerating¹⁴ to 3.5 cm/yr in a 6-million-year-window.
Eurasia, a huge continent with a small active margin, is only moving about 1.5 cm/yr. Despite being a “convergent” boundary, the eastern coast is actually spreading at 1-2 cm/yr to accommodate the motion of the Americas (which is to say, oceanic crust is still subducting but the boundary is moving away from the mid-ocean ridge), which is causing slab rollback and back-arc extension.
Africa and Antarctica, continents surrounded by divergent boundaries, both move at less than 1 cm/yr.

So as general guidelines:

Situation	Subducting Ocean	Recent Subduction Collision	Active Margin Continent	Passive Margin Continent
Plate speed (cm/year)	10-20	5-10	2-5	<1

1 cm/yr is about 0.9° of rotation around the Earth every 10 million years (i.e. 0.9° latitude or longitude on the equator), so as a convenient rule of thumb you might say that an active margin continent moving west on the equator for 100 million years should tend to cover about 20-50° longitude. For planets of other sizes, you should divide that conversion factor by the planet's radius in Earth radii; but whether or not we should expect planets of different size or even different age to have similar plate speeds is a largely unresolved question.

There appears to have been¹⁵ a general increase in average plate speed since the onset of plate tectonics, roughly doubling over the last 2 billion years, which may be associated with the period of the supercontinent cycle decreasing by the same factor, but at any one time overall variation will be larger than this general trend. Plates can accelerate very quickly when a new subduction zone is formed or a mid-ocean ridge is subducted.

Note that the speed of a plate’s movement is not necessarily the speed at which it’s boundaries are moving, save for in the case of a continent’s leading-edge active margin; an ocean plate can be moving at high speed but bounded by static convergent and divergent margins (though no boundary is totally still today).

The area of the continents appears to have generally increased over time, though exactly how quickly and whether it's continuing to grow remains unclear¹⁶. Today, it appears that the Earth is gaining an average¹⁷ of 0.6-0.9 cubic kilometers of continental crust per year—at an average thickness of 35 km, that’s 0.017-0.025 km²/year, or 17-26 thousand km²/million years, compared to a current continental crust area (including submerged continental shelves) of ~200 million km². That’s around 9-15% growth over the last billion years; not a lot, but worth bearing in mind (all of this assuming average continental crust thickness has remained constant, which might not be the case). This may go faster on larger planets because of square-cube scaling between the area of the crust and the volume of the mantle, but it's not clear.

Finally, sea levels tend to rise and fall relative to the continents with the supercontinent cycle. During a supercontinent’s tenure, the oceans are mostly composed of old, low-lying oceanic crust and the continents are lifted up by mantle plumes, combining to cause low sea levels. Once the continent breaks up, the old oceanic crust is replaced with younger, more buoyant crust, reducing the volume of the oceans, and the continents move off the mantle plumes and so sink slightly, so sea levels rise over the next hundred million years or so, until the oceanic crust ages and continents begin reassembling. Global climate and glaciations also affect sea level, but the relationship of those to tectonic events is harder to parse; we’ll tackle that in a later post.

Simulating Plate Tectonics

To get a map with a realistic tectonic history, there’s a few different approaches to consider, all of which involve creating a new map. I don’t suggest trying to retcon a tectonic history onto a map you’ve already made—sorry—for much the same reason I would suggest tossing out a set of pool balls onto a table and then trying to devise a pool game that would arrive at that state. I also don't suggest trying to just slap down plate boundaries at random and then draw continents to fit them; most attempts I've seen of that create plate boundary patterns that simply could never plausibly arise in reality, and in general continents are not passive passengers on plates, they play a key role in deciding how plate boundaries develop and where plates move.

The Delegation Approach

Most map generators simply aren't very good at replicating the features of plate tectonics, relying on fractal generation algorithms that produce terrain that may vaguely resemble real coastlines and mountains at a glance but doesn't stand up to any real scrutiny. Rather than a diverse world, you end up with pretty much the same ragged coasts, rolling hills, and gently rising highlands everywhere on the planet.

As it stands, the only full map generator I've seen with a decent implementation of plate tectonics is that in the alpha version of Songs of the Eons, a strategy game in development. It doesn't perfectly replicate all the features that emerge from plate tectonics (sadly it lacks fold-and-thrust belts, probably due to resolution limitations) but it pretty reliably produces a reasonable distribution of plate boundaries and associated mountain features, and even simulates glacial action to an extent.

Elevation map produced by Songs of the Eons--from their website.

But there are some other map generators that bear mentioning as well, for better or worse:

Tectonics Explorer is a web app designed to show plate tectonics in action, with a fairly detailed model for plate motion over a simulated globe, including formation of orogenies and even distinct rock types. It is not, however, designed as a map generator: the setup options are very limited, the resolution fairly coarse, and it isn't designed to run for long periods through whole supercontinent cycles; mountains don't erode down once formed and very bizarre plate patterns can form. The code is public though so perhaps it could be expanded upon in the future.
Tectonics.js is tantalizingly close to a decent simulator of tectonic motion, but with a couple fatal flaws: rather than attempting to simulate persistent subduction zones, the model simply deletes ocean crust past a certain age and pulls in surrounding crust; no distinction is made between subducting and overlaying plate, and no associated orogenies are created; mountains form only by collisions. Given the emphasis I've been placing on the importance of subduction zones as the drivers of plate motion, I can't really recommend this model in its current form.
Undiscovered Worlds is an in-development map generator that doesn't appear to have any detailed simulation of plate tectonics, but does mke some attempt to replicate many of the major features and appropriate terrain, and also has one of the better attempts at climate simulation I've seen in a map generator of this type. But given its early development state, it's not packaged in any easy executable and so may be difficult to get running at all.
Some other online map generators like Azgaar's or the Mewo2 fantasy map generator work decently enough on small scales, but don't make any real attempt to form tectonic features and so aren't appropriate for continental or global maps. I also wouldn't pay much attention to Azgaar's climate generation.

The Quick and Dirty Approach

If you want more control over how the resulting continents look, then we can try to make our own tectonic history by hand. But if you don’t want to spend too long on it, then we can use supercontinent cycle as a shortcut by working forward from the last supercontinent to work out a reasonable modern distribution of continents and plate boundaries. Most people seem to want a world roughly resembling modern Earth—late in the breakup of the last supercontinent—so I’ll start with one of those.

Basically, we just take a vaguely positioned supercontinent, slash it down the middle with a zig-zagging rift, and move the resulting fragments directly away.

My supercontinent. Red is the rift, arrows indicate motion of the continents.

Break apart the fragments a bit, rotate them, maybe crash a couple of them into each other, and we should be good.

The arrows within the continents show the rotation of each fragment.

We should end up with an interior ocean surrounded by passive margin coasts—that are vaguely mirrored on opposite sides—and an exterior ocean surrounded by active margins.

Blue lines are active margins, grey lines are passive margins.

If we follow modern Earth’s example we might designate the active margins on one side of the ocean as leading-edge and those on the other side as trailing-edge with slab rollback. We can also place a couple more subduction zones between the continents, perhaps even invading the interior ocean. The exact appearance of these coastlines is something I'll mostly leave to another time, but in short:

Leading-edge active margins should have fairly smooth coastlines with high coastal ranges.
Trailing-edge active margins should have more jagged coastlines with offshore island arcs.
Passive margins may either have similarly jagged coastlines with numerous long inlets from river estuaries reaching into a low coastal plain, or my sometimes be smoother due to uplift by the mantle.

Blue lines are leading-edge active margins, orange lines are trailing-edge active margins, purple lines are new subduction zones.

Finally, in the continent interiors we can place high mountain ranges on active margin coasts and continent collision boundaries and low mountains randomly and along the interior coasts as remnants from older collisions.

Dark brown is new high mountain ranges, beige is old lower ranges

If we wanted a world during supercontinent assembly, we have to decide whether to do an extroversion or introversion cycle. For extroversion, we keep moving the continents out—rotating and crashing them into each other as they go—until they start meeting on the opposite side of the world—throwing in some trailing-edge active margins on the coasts of the interior ocean as the crust there ages and more subduction zones break into it from the sides.

Orange lines are new active margins, and brown areas new collisional mountain ranges. I haven't adjusted to coastlines to reflect this, just moved the landmasses.

For introversion, we place subduction zones in the interior ocean, but turn them into leading-edge active margins and leave the exterior coasts as trailing-edge active margins (this needn’t happen on all sides of the ocean, just one will do). Carry the continents back towards each other, but with some rotation as they’re pulled about by the new subduction zones; the coasts shouldn’t meet back up neatly in the same places they split

Blue lines are new leading-edge active margins, orange lines are former leading-edge active margins that have become trailing-edge. Again, coastlines aren't adjusted.

If we want a supercontinent, we just carry forward either case until the continents have all met back up and closed the oceans between them, leaving high mountain ranges along the sutures and active margins on most—or all—of the coasts. The supercontinent should still be moving somewhat relative to the oceans, so we’ll have a leading edge and trailing edge, and we can also have microcontinents moving around the edges if we don’t want all the land area in one piece.

New supercontinent, carried forward from the introversion model. Blue lines are leading-edge active margins and orange lines are trailing-edge, after the supercontinent has formed.

There are some ways we could try to better account for the motion of continents across a spherical surface and how that distorts their appearance on a flat map, or try to better account for the features built up over multiple supercontinent cycles, but if we want to get more detailed there are better tools for the job.

The Obsessively Detailed Approach

If we really want a complete, detailed, and realistic history for the whole planet, then we need to fully simulate the history back not just from the last supercontinent, but all the way back to the supercontinent before it—around 500 million to 1 billion years of history. This is far enough back to cover the formation of almost all major geological features on Earth today. And while we're doing that, we'll be tracking not just the movement of the continents, but all the plates, and the boundaries between them.

So we want to start with a supercontinent, break it apart, let it reassemble into a new supercontinent, and then break that apart and continue to the stage in the cycle we want to reach, all the time following the rules we established in the last section. As we do this, we’ll be recording the positions of a few key geological features:

Orogenies, distinguishing between Andean-type orogenies, where a subduction zone forms along the coast of a continent; Laramide-type orogenies, where a rapid motion or subduction of a mid-ocean ridge causes flat-slab subduction; Ural-type orogenies, where a continent simply collides with another continent, microcontinent, or island arc; and Himalayan-type orogenies, where several collisions happen in a row or there's a particularly rapid collision between continents with one-sided subduction. We shouldn’t have to worry much about slab rollback and back-arc spreading in trailing edge subduction zones, as these will shift back into more regular orognies in the long run—we just have to note the ones still active at the end.

For reference, here are the widths of some modern examples of these orogeny types:

Location	Orogeny Type	Width (km)
Ecuador	Andean	80
Chile/Argentina	Andean	120
Peru	Andean	200
Chile/Bolivia	Laramide	750
US southwest	Laramide	1300
Russian Urals	Ural	50
Switzerland/Italy	Ural	180
British Columbia	Ural/Andean	600
Nepal/China	Himalayan	1200

Failed Rifts, as these may rift again later, and even if they don’t they tend to form low-laying regions that become the core of major river valleys later on, like the modern Mississippi.

Hotspots, which form local volcanic features and remain mostly (but not always) stationary relative to the mantle—Hawaii is a prime example. There is some evidence¹⁸ that these are mostly associated with sections of subducted crust sitting above the core-mantle boundary, but they can appear elsewhere. As a guideline, we’ll expect them to mostly appear either near the rifting center of a supercontinent, roughly on the opposite side of the planet, or in areas of weakened crust like trailing edge back-arc basins, but they’re also possible anywhere else. They’re small features that tend to be erased by other processes in the long run, so we don’t really have to worry about them until the last 100 million years or so of the sequence. Definitions vary but there are around 60 hotspots active today, though many are very small or appear at continent boundaries where we already expect there to be volcanic activity—so really 10-20 should be fine.

Large Igneous Provinces, which are massive regions of flood basalts created by giant mantle plumes. Iceland is the closest equivalent we have today, but it’s nowhere near the same scale (I’ll describe them and their effects in more detail when I discuss mass extinctions). They appear in roughly the same areas as hotspots, though with no particular association with trailing edges—LIPs are more consistently associated with large mantle plumes. The largest known—the Central Atlantic Magmatic Province—formed along the North America-Africa rift at the start of Pangea’s breakup. We’ll only concern ourselves with the largest examples, and also only those on the continents—though they appear frequently in the oceans as well, they don’t typically create islands or any other particularly notable feature. Again, definitions vary but there have been about a dozen large continental LIPs in the last half-billion years, maybe 5 of which were associated with the breakup of Pangea. Also note that LIPs often leave hotspots behind after they conclude.

To do all this, I’ll be using a program called GPlates, a free program designed to allow paleogeographers to map the motions of Earth’s plates in the past and does a pretty decent job with fictional plates as well. In principle everything I’m going to do here could also be achieved with a beach ball and some cut-out pieces of paper, but using software just seems easier for me. I would not recommend attempting this on a flat map.

I've now added a supplemental tutorial on using GPlates in a separate post; I still recommend the tutorial written up here as a secondary source. Refer to those for the specifics of using the program; what I want to lay out here is a more conceptual outline of how to simulate tectonics in semi-realistic way; at least, well enough to produce realistic "modern" geography.

Remember that you can click each of these images for a bigger version. All are created by me in GPlates and exported to Mollweide projections.

I’ll start out with an ideal supercontinent containing 10 major cratons (marked in grey). These are particularly old sections of continental crust at the core of all of Earth’s continents that are thicker and tougher than the rest and so are rarely broken or even much deformed—so we’ll keep them intact through the exercise, and give them rather flat interiors at the end. Each craton has some accreted continental crust around it, and for now there are no additional bodies outside the supercontinent.

Overall this supercontinent occupies about 1/4 of the planet’s surface area in order to about match Earth’s land area around a billion years ago, but there’s no particular reason that should be the case for all worlds (though more than about 40% leaves the continents very little space to maneuver and there's some reason¹⁹ to think plate tectonics will simply lock up with more than 50% continental area). We don’t need to worry about the plates in the oceans or any features of the continents aside from the placement of the cratons, because all of these will be near-completely erased and overwritten with new plates and features over the course of this exercise.

From here I'll move forward in 50-million-year steps, moving the continents, creating new oceans, and splitting and fusing plates as required by the rules laid out so far—though still with some space for artistic interpretation in the exact speed and direction of plate movement and placement of new rifts. Though we haven't marked out any boundaries in the oceans, we only need a few subduction zones to start out with and before long we'll have completely replaced the old oceans with new crust.

But first, a little bookkeeping: I’ll give each of the cratons a letter designation, and refer to any continents they form by the combination of the cratons in them; so this first landmass carries the admittedly awkward title of supercontinent ABCDEFGHIJ. As new ocean plates appear I’ll designate them with lowercase Roman numerals (i,ii, iii, iv, etc.) and microcontinents (sections of continental crust without a craton) I’ll designate with numbers (1, 2, 3, 4, etc.).

I also want to break up the sequence of events into geological periods, but with more memorable names. I’ll define the time starting with the breakup of this supercontinent as the Cuvieric Period.

I’m going to be taking a very “let’s see how it goes” approach, but just to keep everything organized I’m going to decide now that our first supercontinent cycle should be an extroversion cycle. Using my powers of prophecy, I can see that this means most of the action is going to occur on the other side of the planet, awkwardly on the edges of the above map—so for the sake of consistency I’m going to shift our view 105° west. I can also see that the entire process is going to take 850 million years.

Let’s get started.

850 mya: Early Cuvieric

We start off much as we did before in the quick-and-dirty approach, with a major rift bisecting the supercontinent (marked in red). But this time I’ve made sure to avoid the cratons, and mark out the third-arm failed rifts. I’ve also added subduction zones around the rim of the supercontinent to help pull it apart (marked in blue) and noted the Andean-type orogenies caused by them (marked in black).

Presumably there would be LIPs and hotspots associated with this rifting, but the landforms created won’t last 850 million years so we don’t have to worry about them this time around. It’s just something to bear in mind when we work out the history of life on this world.

800 mya: Mid Cuvieric

The two supercontinent halves—ABCDEF and GHIJ—drift apart, leaving a mid-ocean ridge between them (marked in red, like the rifts) and transform boundaries on their edges (marked in green). New ocean crust forms between them (shaded in lighter blue than the Pre-Cuvieric ocean crust). They do not break apart evenly—on such a long border on a spherical planet that’s not really possible, but in this case especially the subduction zone on GHIJ’s north shore helps rotate the continents as they split.

Arrows indicate relative plate motion at boundaries, with length roughly correlating to speed. All these closeup maps are in orthographic projections; i.e., like looking at a globe.

But already, GHI is rifting away from J, with associated failed rifts—the main rift itself is an extension of a “failed” rift from the initial rifting event. Such multi-stage rifting appears to be typical of supercontinent breakups, with all the resulting continents moving away from a specific rifting center.

As the continents move, the subduction zones on their flanks churn away, and I’ve added volcanic island arcs along those in the oceans. I’ll continue to do so as new subduction zones appear, though to be clear these aren’t meant to be accurate depictions of the types of islands that would appear—just abstract representations that there are islands there.

750 mya: Late Cuvieric

The continents continue to drift apart, and now a triple-junction of mid-ocean ridges has formed between them. I’ll add one more rift to split ABCD from EF—again along a prior failed rift—and call it good so far as the primary supercontinent breakup is concerned.

But already there are other forces causing rifting: The subduction zone on GHI’s east coast has spread south, creating an enclosed Tethys-like ocean basin with a subduction zone on one side and no mid-ocean ridge to feed it. As such, a section of land—Microcontinent 1—is being torn away from shore opposite the subduction zone.

By this point, new ocean crust and motion of the continents away from the mantle plume under the rifting center has caused a significant rise in sea level, and we can expect inland seas on many of the continents—though I won’t bother showing that here.

With the supercontinent rifting coming to an end and other tectonic processes starting to play out, we’ll call this the end of the Cuvieric and start another period, the Anningic.

700 mya: Early Anningic

The continents continue to rift apart, moving at increasing speed. EF and J—which is now crossing over the north pole—are speeding towards each other due to the subduction between them, and Microcontinent 1 is moving particularly fast towards a collision with the other shore of GHI.

650 mya: Late Anningic

Microcontinent 1 collides with GHI at 660 mya, forming a continent-continent convergent boundary (marked in blue) and a Ural-type orogeny (marked in black). The subduction zone “jumps” across the thin landmass, consuming the new oceanic crust created behind 1 as it travelled.

EF is speeding towards its own collision with J, and the pull of J’s subduction zones is causing it to rotate clockwise. This compresses the oceanic crust near EF's north shore, helping form a new subduction zone that spreads down its east coast; with these large 50-million-year timesteps we don't have to worry too much about which exact mechanism forms the new zone, but in this case likely transform collapse. The mid-ocean ridge to EF's east continues to spread, but now will drift west due to the subduction. EF and J are thus bound for a typical Wilson cycle and rejoining.

As EF drifts away from ABCD, the subduction zone on its west coast spreads as well. EF is still moving westward, even though it is subducting crust on it east side.

600 mya: Early Owenian

At 620 mya, EF finally collides with J. We’ll take this collision to conclude the Anningic and start another period, the Owenian.

As the continents collide and slow down, subduction pulls E and F apart, and the subduction zone on E’s east coast has consumed the mid-ocean ridge there and begins pulling a section of land, Microcontinent 3, away from J. To the south, E’s subduction zone spreads and links up with F’s subduction zone, which allows the whole assemblage—EFJ—to drift south.

Meanwhile, the subduction zone on GHI’s interior ocean has consumed the mid-ocean ridge there, and begins pulling another microcontinent—Microcontinent 2—away from the far shore.

550 mya: Mid Owenian

The complicated EF-J collision continues. E and F have now both solidly collided with J, forming Ural-type orogenies, but a remaining section of subduction zone pulls a small landmass, Microcontinent 4, from E’s shore. In the east, 3 begins to collide with E.

In the south, Microcontinent 2 swiftly moves towards collision with GHI, but GHI itself is moving towards collision with ABCD.

At sea, the mid-ocean ridge between EFJ on one side and ABCD and GHI on the other has grown so broad that spreading along its whole length is no longer possible due to the planet's spherical geometry. A subduction zone forms along its eastern edge—north of GHI—which starts pulling GHI and EFJ together. EFJ has also consumed the mid-ocean ridge to its south, so we’ve solidly moved from supercontinent breakup to supercontinent assembly. Though we said it would be an extroversion cycle, as is typical it’s actually more of a mix: an extroversion between ABCD and GHI, and an introversion or perhaps orthoversion between EF and J and between EFJ and ABCDGHI

With continents over both poles, we can expect the climate to be cooling—perhaps there’s even an ice age. This, along with aging of the new oceans, will pull sea level down from its late-Cuvieric high.

That said, there might be a brief warm spell associated with our first LIP in northeast GHI (marked in orange), as well as a possible extinction event.

500 mya: Late Owenian

The EF-J collision event has now largely concluded; E and F are solidly attached to J, and Microcontinents 3 and 4 have collided with E and J, respectively. The trapped sea between E and J would probably be filled in before long, but I’ll leave it in place as a weak point in the crust. I could also have the subduction zone jump these microcontinents and continue pulling chunks away—this is probably more realistic—but given that these regions will soon be involved in the collision with the southern continents, I’ll leave them for now.

As EFJ moves south and its island arcs collide with ABCD’s, the islands accrete to the overlaying plate and form a section of continental crust—eventually a whole new continent could be formed this way, but that won’t be the case here.

In the south, Microcontinent 2 collides with GHI at 530 mya, compressing the already-deformed lands of former Microcontinent 1. If this process continued uninterrupted, H might eventually be torn away and finally close the subduction zone—but again, that won’t be the case here.

Note here that we might infer the presence of an old mid-ocean ridge from before our simulation began due to relative plate motion around the pre-Cuvieric ocean.

450 mya: Early Huxleyic

ABCD finally collides with GHI at about 460 mya, closing the Owenian and starting the Huxleyic period. As with the EF-J collision, it’s a messy process, and subduction zones pull G—and its accreted microcontinents—away from HI.

Meanwhile, the oceanic crust formed at the start of the Cuvieric is now pushing 400 million years old. A subduction zone spreads down the west coast of GHI, eventually separating the continent from the oceanic crust and forming our first ocean-only plate, Ocean i, which continues to move east.

A subduction zone also spreads along EFJ’s eastern coast, so the continent is pulled in 2 directions. E splits from FJ, taking a large part of J’s former plate with it. FJ continues south towards the already-starting collision with the assembling supercontinent, and E drifts east out to sea.

Just for fun, I’ve also dropped a small LIP in FJ.

400 mya: Mid Huxleyic

The ABCD-HI collisions continues, with widespread mountain formation and land deformation, including a proper Himalayan-type orogeny between ABCD and G, where 2 microcontinents are sandwiched.

FJ also collides with the supercontinent’s north, forming an impressive orogeny of its own thanks to the trapped sections of former island arcs.

E continues east, though it’s rotating clockwise thanks to subduction between it and ABCD. It also experiences a massive LIP, potentially causing a large mass extinction.

At sea, the triple junction of mid ocean ridges spreads apart to create another plate of new oceanic crust, Ocean ii. Though it has no subduction zones of its own, this plate will spread at the expense of surrounding plates as they are subducted near the coasts. At this point, almost all remaining oceanic crust from before the Cuvieric has been subducted (save for a couple patches near the south pole and between the colliding continents), so we have a complete image of the tectonic forces at play in our world.

One consequence of this is that to keep Ocean i moving in the right direction to feed all the subduction zones on GHI’s west coast, it has to rotate clockwise and so start subducting part of the ocean connected to ABCD.

350 mya: Late Huxleyic

The ABCD-GHI collision concludes and FJ more solidly attaches to the assemblage. We can definitely say that we’ve entered the tenure of a new supercontinent, ABCDFGHIJ. We’ll mark the occasion by closing the Huxleyic and starting the Marshian period.

We could let the situation continue to play out until E joined the supercontinent, but just to make things a little more interesting we’ll have a minor rifting event that separates D and Microcontinent 5 from the supercontinent’s east coast.

As these new landmasses move east, the subduction zone on E’s east coast pulls the distant ocean crust west. The result is a new subduction zone splitting the 2 sections of crust apart, forming Ocean iii. Meanwhile, subduction zones spread along the north coasts of E and FJ, splitting them from Ocean iv. There is now a complete ring of subduction zones separating the assembling supercontinent from the surrounding superocean—which is exactly what should happen as the oceans age and the continents stop moving relative to each other.

As Ocean I moves south, the mid-ocean ridge between it and Ocean iv begins to subduct under the supercontinent’s west coast, so we’ll put a Laramide-type Orogeny there.

300 mya: Early Marshian

Internal mountain building has largely ceased on ABCDGHI, though it continues along subduction zones and a sliding collision in the south with Microcontinent 5. The whole landmass drifts slowly north and rotates clockwise. The vast interior will be fairly arid, with a large desert at the center.

D drifts north and E drifts south, accreting island arc crust along its south coast. A small section of the ocean crust attached to Microcontinent 5 gets pinched out between E and Ocean I, forming a small ocean plate, but it won’t last long enough to be worth giving a name.

Ocean ii continues to grow, and the very last of the pre-Cuvieric oceans have either been consumed by subduction or trapped within the supercontinent. Some small fragments will have been lifted onto the continents and survive there as ophiolites.

250 mya: Late Marshian

Microcontinent 5 impacts E, sandwiching some island arcs, and other island arcs from ocean i impact E’s south coast, creating a long, complex orogeny. The subduction zone jumps across plate 5, and E is pulled towards a collision with the supercontinent.

D continues north, and has a moderate LIP.

The mid-ocean ridge between Ocean iv and Oceans i, ii, and iii is now so broad that, as has happened before, it can no long spread across its entire length. Ocean iv begins subducting below Ocean i.

200 mya: Early Copian

E finally collides with ABCFGHIJ at about 220 mya, forming a vast Himalayan-type orogeny, and D begins colliding with the same landmass, so for short time we have a complete supercontinent ABCDEFGHIJ.

But not for long. A rift crosses the interior, and with it comes a massive LIP, flooding the region with lava. ACG in the south begins splitting from BFHIJ in the north. So ends the Marshian and begins the Copian period.

The northward motion of BFHIJ encourages the formation of a new subduction zone on its north coast to subduct the old crust inside the now-separated Ocean v. Ocean i is also rapidly subducting, and the subduction of its mid-ocean ridge with Ocean ii causes a Laramide-type orogeny along ACG’s west coast.

Rifting also breaks apart E. The new landmass contains much of the crust from former Microcontinent 3, back in the Owenian, so we’ll give it the same name. It may look like an odd snaking continent here, but in reality it would probably appear as a long archipelago like Indonesia.

Now that E has stopped, a subduction zone spreads along its north coast, separating it from Ocean vi. This subduction also pulls away a section of D, Microcontinent 6.

150 mya: Late Copian

The supercontinent has now solidly split in two, but like last time it doesn’t spread evenly; subduction zones in the east cause BFGHIJ to rotate clockwise and ACEG to rotate counterclockwise.

Also like last time, the rifting doesn’t conclude in one step. G begins rifting from ACE, with an associated LIP.

To BFHIJ’s north, Ocean i has subducted completely and Ocean v will soon follow. Their associated island arcs attach to BFHIJ’s coast, forming Ural-type orogenies and completing a typical round of island arc accretion.

In the east, D finishes colliding with BFHIJ and Microcontinent 6 speeds south towards ACE. In this process, another oceanic plate is cut off, Ocean vii. This whole eastern ocean is now a maze of subduction zones, island arcs, and mid-ocean ridges, but it’s all being inexorably pulled west towards collisions with the continents.

In the south, Microcontinent 3 collides with G to its west and islands from Ocean vi to its east. Take special note here how odd this appears in the Mollweide projection, as opposed to the view over the south pole below; always keep a map's distortion in mind when working near the poles.

Finally, as we approach the end we should start adding hotspots (marked with purple dots)—some near rifting centers, some randomly across the globe. For now, though, there's just one left over from the Early Copian LIP.

100 mya: Early Andrewsian

At this point I’m trying to construct more of a continuous geological history rather than doing everything in 50-million year steps. As such, shortly after G rifts from ACE, CE rifts from A at 110 mya—closing the Copian and beginning the Andrewsian period.

But the supercontinent breakup isn’t over—an LIP within BDFHIJ foretells another rifting event.

Microcontinent 6 also collides with CE around this time, and the mid-ocean ridge behind it is quickly consumed. This pulls CE towards BDFHIJ, heralding a second Wilson cycle for E and J.

Oceans vi and vii are also being quickly subducted, and Oceans iii and iv are following them. But Ocean ii’s subduction on the other side of the world helps keep their mid-ocean ridges balanced in the center.

To the south, 3 continues to collide with G over the south pole, and subduction of the ii-iii mid-ocean ridge causes a Laramide-type orogeny on its north coast.

Again like last time, at this point we should expect high global sea levels and inland seas—though perhaps that may be moderated by the presence of a large, likely-glaciated mountain range over the south pole. You'll also note some more hotspots, and also that the hotspots are remaining static as the plates move over them.

50 mya: Late Andrewsian

DJ rifts from BFHI at 80 mya, and both this rift and the A-CE rift have several associated hotspots—as well as one final LIP in CE’s south.

G is moving pretty swiftly at this point, and as it shifts north from the south pole it creates something of a pocket between it and BFHI that Ocean ii is pushing into—such that there is a small section where the oceanic crust attached to BFHI and G is subducting under Ocean ii, starting subduction invasion into the young interior ocean. The collision with Microcontinent 3 also continues, and has now enclosed a section of ocean, where slab rollback could lead to interesting patterns of land and sea.

Ocean vi has almost completely subducted away, and there is also a small section of Ocean vii remaining but it’s so small that I was too lazy to bother labelling it.

I did bother to add the vi-vii rift in this one, and I've noted an early rift pulling microcontinent 4 towards CE.

0 mya: Ostromian

At 30 mya, BF rifts from HI—the last rifting in the breakup of the supercontinent. At the same time, CE barrels into DJ, forming a broad Himalayan-type orogeny. We’ll use these events to mark the close of the Andrewsian and opening of our final period, the Ostromian.

Just before the collision, Microcontinent 4—from back in the Owenian—is pulled from DJ’s south coast and collides with the island arc to CE’s north, then is carried back north into DJ. This, in combination with slab rollback in the enclosed sea, should lead to an interesting, Mediterranean-like coastline.

Microcontinent 3 continues to collide with G, but by 0 mya there is still a substantial enclosed sea remaining.

Ocean ii continues to push into the pocket between G and HI. It isn’t a separate plate yet, but it will be before long so we can go ahead and call it Ocean viii. There’s also a small new section of subduction between A and G due to G’s counterclockwise rotation.

And because this is the last section of tectonic history, I’ve also pulled out some sections of the east coast of DJ, as it’s been moving slowly and we can expect some back-arc extension.

And that finally brings us to the “present day”. Here's a chart I threw together to help us keep track of the periods I've established and drift of the continents throughout them:

Widths are not proportional to continental area because I only realized I should have done that after I finished.

And indeed we can pretty well see a supercontinent cycle of breakup (Cuvieric, Anningic), assembly (Owenian, Huxleyic), tenure (Marshian), and breakup once more (Copian, Andrewsian, Ostromian). As with Earth, it's not yet clear if current plate motion is heading towards another extroversion cycle, or if subduction invasion at the edges of the interior ocean will pull the continents back into introversion, or perhaps some combination of the two.

Now, we can’t really go on calling our continents “HI” and “BF”, so I’ve come up with some interim names for our 6 continents (I’m still counting CE and DJ as separate, given the high mountain range dividing them) and 4 major oceans until we figure out what sort of names their eventual inhabitants might use.

As a final touch for this post, we can use our recorded history to get a rough sense of the modern topography. As we’ve moved along, I’ve been marking out the areas affected by orogenies and LIPs, and now we have a record for all the major mountain-building events of the last 850 million years (You can use gplates to generate these age maps for you):

This map is in an equirectangular projection, which will be more useful for modifying it later on.

Mountain range height doesn’t directly correlate to age, but there is a relationship; Based on the heights and ages of many of Earth’s current ranges, average mountain height tends to begin around 2500 meters and drop about 5 m every million years, with peak height starting around 4000 m and dropping 8 m/myr. But there are plenty of exceptions and local variation and initial height is particularly broad (and really an exponential decay might be a better model, but we're working in generalities here). With this as a guide and a bit of artistic interpretation, I’ve put together a rough pass of Teacup Ae’s global topography:

These are average elevations, mind; we'll add plenty of peaks and valleys later on.

I’ll talk a bit more in a later post about the exact considerations regarding different mountain types, erosion, and uplift, but for now this gives us a general idea of the planet’s terrain and will help in working out the climate.

And that about does it. There’s still a lot of cleaning up and fleshing out to be done: In addition to further work on internal topography, the coastlines need to be refined, the island arcs expanded, and other islands added. Bear in mind that the boundaries shown aren’t actually coastlines, but boundaries between continental and oceanic crust (roughly). Even when sea level is low, much of the margins of the continental crust sections will be submerged continental shelves—more on passive margins. Though, in this specific case, I’ve tallied up the areas of crust at the start and end of the sequence and found that continental area didn’t increase as I was aiming for (I should have added more land during island arc collisions). It actually decreased, though only by about 0.2%. So I might expand out these shelves a bit so I can retain more land area.

But for now, it’s a pretty good result.

Orogenies are black when active then fade to white as they age, LIPs are orange areas, hotspots are purple dots. Some of the borders can be a bit janky during collisions because I only bothered updating them every 50 million years.

An alternate view showing tectonic boundaries (red for divergent, blue for convergent, green for transverse) and distinct plates in different colors.

I’ll admit, this is a pretty tedious method to get a world map; what you see here is the result of a solid week of long afternoons (though it probably would have gone a lot faster if I was less obsessive about matching up all the lines just right). But I’m pretty happy with the result: A world like Earth, with a similar diversity of landforms, and a solid geological history to work off of (we’ll see the utility of that over the next few sections) but different in interesting ways:

Except between Hutton and Lyell, there are no land connections between the continents as there are between most continents on Earth. Hutton and Steno might still have a land bridge in the depths of an ice age, but at such points it would be covered with ice sheets anyway. There are only thin seas separating Steno, Holmes, and Hutton, and (once I’ve expanded the islands) Lyell and Wegener, so there will be some exchange of plant and animal life and intelligent Teacupites. Still, natural and social divergence of the continents will be easier than on Earth—especially on Agassiz, which is completely isolated from the other continents.
Hutton has a European-like west coast and Asian-like east coast, but a much thinner (though mountainous and probably arid) interior than Eurasia, which could bring these diverse regions into closer contact than in our world.
Steno and Agassiz both have sections near the poles that we should probably expect to develop ice caps, but also extend into temperate climate zones. We don’t have that full gradient on any one landmass on Earth today.
Holmes is an “island continent” somewhat smaller than Australia, but less arid and closer to other landmasses and likely centers of civilization.

Next post, we’ll take a quick detour into looking into alternate modes of tectonics that we might see on more alien worlds, and then we’ll move on to planetary climates and biomes.

In Summary

Convection of heat in Earth’s interior drives plate tectonics, but the motion of particular plates is mostly determined by patterns of subduction.
Plates spread apart at divergent boundaries—rifts and mid-ocean ridges—come together at convergent boundaries—subduction zones and collisional mountain ranges—and slide past each other at transverse boundaries.
Ocean crust subducts and is completely replaced on ~250 million year timescales
Continental crust is formed at subduction zones and never subducts, though it can be deformed and compressed by collisions.
Earth undergoes a cycle of supercontinent formation, tenure, breakup, and reformation in a new form roughly every 750 million years.

Though this may have accelerated to a 400-500 million year cycle more recently

Supercontinent cycles can be extroversions, with continents spreading and closing the exterior ocean, introversions, with continents reversing direction and closing the interior ocean, or a combination of both, or perhaps orthoversions, with continents rotating into a collision with both interior and exterior oceans surviving.
The position of subduction zones is the primary determining factor in the motion of plates, though the influence of mantle plumes plays a role as well.
Mountains form at orogenies due to subduction zone volcanism or continental collisions, and then gradually erode down afterwards.
Regular leading-edge, slab-rollback trailing-edge, and flat-slab subduction zones produce different types of active margins.
Subduction jumping, polarity reversal, and subduction invasion allow subduction zones to survive until closed by a major continent collision.
Broad, Himalayas-like plateaus are formed by repeated impacts of small continental fragments followed by a large, high-speed continent collision.
Complex, Mediterranean-like seas are formed by trapping of oceanic crust between converging continents.
LIPs, hotspots, and failed rifts are other tectonic features that can affect global geography.

Notes

If you want to see a model of Earth's development from the very beginning, this video is a pretty decent representation, but with the major caveats that any geographic or atmospheric data before about 2 billion years ago is very speculative, and the landmasses shown don't necessarily represent all the landmass that existed at the time; just what we have direct evidence for.

Buy me a cup of tea (on Patreon)

Part Vb

1 Alfè, D., Gillan, M. J., & Price, G. D. (2007). Temperature and composition of the Earth's core. Contemporary Physics, 48(2), 63-80.
2 Yuan, L., & Steinle-Neumann, G. (2023). Hydrogen distribution between the Earth's inner and outer core. Earth and Planetary Science Letters, 609, 118084.
3 Scotese, C.R., (2015) Plate Tectonics Driving Mechanisms: Some Simple Rules that Explain Why the Plates Move the Way They Do. Researchgate
4 Davies, H. S., Green, J. M., & Duarte, J. C. (2018). Back to the future: Testing different scenarios for the next supercontinent gathering. Global and Planetary Change, 169, 133-144.
5 Korenaga, J. (2007). Eustasy, supercontinental insulation, and the temporal variability of terrestrial heat flux. Earth and Planetary Science Letters, 257(1-2), 350-358.
6 Arculus, R.J., M. Gurnis, O. Ishizuka, M.K. Reagan, J.A. Pearce, and R. Sutherland. 2019. How to cre-ate new subduction zones: A global perspective. Oceanography 32(1):160–174, https://doi.org/10.5670/oceanog.2019.140.
7 Yoshida, M., & Santosh, M. (2014). Mantle convection modeling of the supercontinent cycle: Introversion, extroversion, or a combination?. Geoscience Frontiers, 5(1), 77-81.
8 Humphreys, E., Hessler, E., Dueker, K., Farmer, G. L., Erslev, E., & Atwater, T. (2003). How Laramide-age hydration of North American lithosphere by the Farallon slab controlled subsequent activity in the western United States. International Geology Review, 45(7), 575-595.
9 Antonijevic, S. K., Wagner, L. S., Kumar, A., Beck, S. L., Long, M. D., Zandt, G., ... & Condori, C. (2015). The role of ridges in the formation and longevity of flat slabs. Nature, 524(7564), 212-215.
10 Boschman, L. M., & Van Hinsbergen, D. J. (2016). On the enigmatic birth of the Pacific Plate within the Panthalassa Ocean. Science Advances, 2(7), e1600022.
11 Muttoni, G., Kent, D. V., Garzanti, E., Brack, P., Abrahamsen, N., & Gaetani, M. (2003). Early permian pangea ‘B’to late permian pangea ‘A’. Earth and Planetary Science Letters, 215(3-4), 379-394.
12 Domeier, M., Font, E., Youbi, N., Davies, J., Nemkin, S., Van der Voo, R., ... & Torsvik, T. H. (2021). On the Early Permian shape of Pangea from paleomagnetism at its core. Gondwana Research, 90, 171-198.
13 "Plate Tectonics", OpenLearn
14 Brune, S., Williams, S. E., Butterworth, N. P., & Müller, R. D. (2016). Abrupt plate accelerations shape rifted continental margins. Nature, 536(7615), 201-204.
15 Condie, K., Pisarevsky, S. A., Korenaga, J., & Gardoll, S. (2015). Is the rate of supercontinent assembly changing with time?. Precambrian Research, 259, 278-289.
16 Korenaga, J. (2018). Crustal evolution and mantle dynamics through Earth history. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 376(2132), 20170408.
17 Hawkesworth, C., Cawood, P. A., & Dhuime, B. (2019). Rates of generation and growth of the continental crust. Geoscience Frontiers, 10(1), 165-173.
18 Kerr, R. A. (2013). The deep Earth machine is coming together.
19 Lenardic, A., Moresi, L. N., Jellinek, A. M., & Manga, M. (2005). Continental insulation, mantle cooling, and the surface area of oceans and continents. Earth and Planetary Science Letters, 234(3-4), 317-333.

Previous versions:

Before July 2024

Comments

AndrewJanuary 8, 2020 at 4:41 AM
Awesome!
ReplyDelete
Replies
NajahfreemanMarch 28, 2020 at 8:19 PM
I understood most of ther article, got to the Gplates part. Dd all the tutorial you linked. Came back here and now i'm completely lost. Like I can't move past your original supercontinent. I just don't know how to do what you have done. Is your supercontinent just a bunch of features that you arelady planned ? Do you draw each 50 Million years something new ? How do you keep track of failed rift ? I have so many questions. I'm so sad i was pumped and ready to simulate my world tectonic with my new found knowledge (very well explained even for a non native speaker), but now i'm stuck :(
ReplyDelete
Replies
AnonymousMay 10, 2020 at 7:08 AM
Is it possible to start at 850ma and then move back to 0ma in Gplates? I've messed around with it a bit and all the rotations/positions are based on 0ma. In yours you've started at 850ma, but I can't figure out how to make movements adjust from there, and not work "backwards" from a map of what the world will be.
I'm not sure if this is a limitation of the software.
ReplyDelete
Replies
NajahfreemanMay 29, 2020 at 11:07 AM
Hello !
I hope you are doing well. Wanted to know if you had a vague idea to when your Gplates tutorial would be ready. I kinda gave up on Gplating my world for now as I'm sure I'm making stupid mistakes and not doing it properly so I'd rather wait to see how you do it.
Thank you :)
ReplyDelete
Replies
EthannatJuly 10, 2020 at 9:18 PM
This is quite a comprehensive lesson. Thank you for taking the time to educate us - I'm fascinated by this science and very appreciative for the worldbuilding advice.
ReplyDelete
Replies
SamuelSeptember 14, 2020 at 9:28 PM
You mentioned that you don't recommend trying to import a tectonic history on an existing map, but do you have any advice for somebody who already has a map that they are attached to?

I don't want to start over from scratch (because I worry that this method doesn't leave me with enough control over the final shapes of the continents, which is rather important to me - see: attachment to existing map). But just drawing in tectonic plates and assuming that the tectonic history works out to get their present position and landforms is apparently not a reasonable solution.
ReplyDelete
Replies
UnknownNovember 29, 2020 at 7:09 PM
I'm pretty late on finding this blog admittedly. But these tutorials have been so fascinating and helpful to read! I recently started a new worldbuilding project, but I was wondering- Would you ever back track tectonic history? Say you have the contemporary stage of a world's plates and might speculate some history? How far back would you be able to go potentially for that?

I'm asking because I did make a very interesting landmass and plate set up for the contemporary stage of the world, but hadn't really gone into tectonic history to get a lot of geographic and resource based features better painted.
ReplyDelete
Replies
AnonymousDecember 7, 2020 at 7:55 PM
If I'm modeling my planet's tectonics over an extremely long time with the extremely in-depth method that you did, do I necessarily need to start with a supercontinent? Based on your post on alternate tectonic schemes, wouldn't it be possible for a watery planet switching from drip and plume tectonics to plate tectonics to have multiple large plateaus that transition to continental interiors?
ReplyDelete
Replies
JoelDsaundersJuly 21, 2022 at 3:42 PM
Can someone make a video on this? I find videos easier to follow.
ReplyDelete
Replies
AnonymousOctober 9, 2022 at 3:41 AM
In the Cuvieric era, why does the subduction zone continue north past the F orogeny when there's already a subduction zone at J? Why wouldn't the transform boundary continue all the way to the northern edge of the F orogeny, since that's the direction the continent's travelling in anyways? Is either scenario possible?
ReplyDelete
Replies
AnonymousDecember 14, 2022 at 5:19 PM
How do "shattered" masses of microplates, such as (old) Europe, (old) central Asia, the Caribbean, and Indonesia, form?
ReplyDelete
Replies
NathanSJanuary 10, 2023 at 6:49 PM
What kind of speed are you moving things in the worked example? Because they seem to be moving a lot slower than the suggested speeds, especially with how long the collisions take to play out? And what does the 'compressed the oceanic crust' at 650 million years look like in practice? The mid ocean ridge overlapping EF's ocean crust? starting to get close to doing that?
ReplyDelete
Replies
AnonymousJanuary 15, 2023 at 2:23 PM
1.How can I decide the speed of plate? Is it random?
2.If oceanic crust meets the border of plates, then what happens?
3.The time when the plates merge/divide is random?
ReplyDelete
Replies
AnonymousJanuary 17, 2023 at 11:18 AM
See: https://cdn.discordapp.com/attachments/826031364595384370/1063480500108410940/20230114_001805.gif
White: Border of Plates, Gold:Oceanic crust
Is it possible? If not, what happens?
ReplyDelete
Replies
AnonymousJanuary 17, 2023 at 2:41 PM
There is two plates, divided by divergent boundary and transform boundary.
In this picture, oceanic crust(Yellow) is next to right plate.
ReplyDelete
Replies
AnonymousMarch 9, 2023 at 1:37 AM
In the case that oceanic crust is pulled in different directions as a result of subduction, would this open up a new ridge in the ocean? In gplates I've only seen continental rifting, in this circumstance would oceanic crust rift or create a new plate?

Sorry if it's in the post I've read it several times in the past but cant find the information im looking for at 1am.

Thank you for these lovely tutorials. Between your blog and Artifexian's video's I've learned so much.
ReplyDelete
Replies
KurodaniApril 29, 2023 at 8:13 AM
In the section of the speed of tectonic plates you write: "For planets of other sizes, divide these values by the radius in Earth radii". Does that mean that tectonic plates move faster on smaller planets and slower on larger planets? Say an active margin continent on a planet with a radius that is about 60% of Earth's radius. If I interpret you correctly then the movement should be 2.7/0.6=4.5 cm/year. Wouldn't that mean that the supercontinent reforms a lot faster on a smaller planet? That seems strange to me, but I'm a total novice when it comes to tectonic movement. Sorry if I misunderstood...
ReplyDelete
Replies
NoahJune 18, 2023 at 5:52 PM
I'm looking at the video of the worlds history showing the plates and boundaries, and I'm wondering how you are able to make them work so dynamically together. I've only just recently started to use the topology tool in Gplates creating line and boundary topologies & only manage to use half stage rotation for the dynamic mid-ocean ridges but cant figure out how to make dynamic collisions, like where your plates and boundary's look like their being absorbed by the overriding plate mine will just overlap. is their a trick to making dynamic Collisions or have I misread your Gplates tutorial?
ReplyDelete
Replies
PascalSeptember 29, 2023 at 5:36 PM
Is it possible for a extroversion cycle to not fully "eat up" the old ocean before forming the next super continent ? My North pole simply has a ring of subduction zones around it for some reason, and a portion of ocean on the southern hemisphere is also still old ocean for some reason, or did i go wrong somewhere ?
ReplyDelete
Replies
AnonymousOctober 25, 2023 at 10:01 PM
I am somewhat confused regarding the highlands on the coast of Brazil and southern Africa; I've always just assumed that big Atlantic ocean-style rifts made these big plateau/highland regions on continental coasts of divergent boundaries, but nothing seems to be supporting this assumption. Sure rift valleys and such form along divergent boundaries but they seem to be unable to form a large mountainous region like that? Other divergent boundaries between continental landmasses don't seem to have this kind of topography either, like Florida or most of Australia. They're also not orogenies, and mostly not LIPs, actually the only thing I can find that overlap with these features seem to be cratons? Aren't cratons supposed to be too old/stable for mountains, such that if they ever did have mountains they'd practically be completely eroded away by the modern day? What's going on here?
ReplyDelete
Replies
ShineNovember 25, 2023 at 3:10 AM
Despite its limitations, do you suggest using Eons if you don't want to spend a lot of time going into detail? Do you have any suggestions for attempting to solve the limitations with eons with external editing?
ReplyDelete
Replies
FuarApril 20, 2024 at 2:15 AM
Hello. I have a question about beginning a tectonic simulation (using GPlates). How do you go about shaping and placing cratons? Mostly placing them. Do you think about what you want the supercontinent to look like and then drop the cratons around or do you place cratons randomly and then draw the continental crust around it? I'm struggling to get a good shape that's around 25% of my planet's surface area. For realism does a supercontinent need to have any specific features? I assume it can't just be a massive blob.
ReplyDelete
Replies
AnonymousApril 27, 2024 at 8:06 PM
Question about mountain typs overlapping; so I have two continents that collided to form a Ural type orogeny, with part of an Andean type overlapping with part of the Ural type shortly after formation. Note the Outer Arc would form on the back slope side.

Further down(southern part)the chain a purely Andean portion would be overlapped with a Ural(outer Arc on the Foreslope side).

How would these two portions work?

Would I just put an Ural type for the southren portion of the chain or would the outer Arc still exist? And for the northern part should I combine the two, for lack of better words, with each parts respective erosion?
ReplyDelete
Replies
BasilJune 11, 2024 at 8:25 PM
Hi! i have a question about cratons. If there is land that isn't part of a craton, does that assume that there will be some kind of geologic activity there? since if there wasn't that would just be a part of a craton, or is it more of a loose thing just for the simulation?
ReplyDelete
Replies
AnonymousJune 18, 2024 at 4:25 AM
I've been referring to your blog for 2 years now and I just now realized why you called it An Apple Pie from Scratch. I would think, "What does any of this have to do with Apple Pie?"
ReplyDelete
Replies
AnonymousOctober 31, 2024 at 7:53 AM
Just a note, the Australian plate moves at 6-7 cm/yr, not about 1.
ReplyDelete
Replies
Lena SynnerholmDecember 22, 2024 at 6:25 PM
I think the error you call “Atypical Continent” is due to insufficient sense of scale. Islands like Hainan often have their highest point in the middle of the landmass. This may have been overgeneralized to also apply to continents. Unfortunately, plate tectonics does not work like that.
ReplyDelete
Replies
AnonymousMarch 10, 2025 at 4:01 PM
Is it possible to do any of this without Gplates and are there any Gplates alternatives for mobile? If I try to do it on a flat map, I end up with polar distortion
ReplyDelete
Replies
AnonymousMarch 14, 2025 at 12:48 PM
I accidentally wrote something in a reply, I said ‘I really want to replicate this on mobile but I don’t have gplates’ and if I were to try, it would result in significant polar distortion.
ReplyDelete
Replies
AnonymousMarch 22, 2025 at 12:51 PM
I don’t have access to Gplates, is there any way I can do this on mobile without Gplates? I really want you to respond.
ReplyDelete
Replies
RiceyJune 10, 2025 at 9:07 PM
How do you simulate Tethys style collisions and slab rollback on Gplates? I noticed you did that in this simulation and the slab rollback in Artifexian's world, but it still isn't really clicking with me.
ReplyDelete
Replies
AnonymousAugust 13, 2025 at 2:06 PM
I'm interested in what you think soil conditions on Holmes/BF will be by 0 mya. It seems kind of like Australia, there being seemingly no major orogenies for the past 200 million years, but instead its gone from tropical to temperate in the past 150 milliion years instead of the reverse. Do you think it will still be mainly lateritic or is ~50 million years enough time for it to build up some decent soil even without mineral inputs from any big mountain ranges? More generally I've just been learning about how uniquely poor Northern Australian soils are and am curious on your thoughts on how the same conditions could come about in conworlds.
ReplyDelete
Replies

Add comment