Let's consider an O'Neill cylinder with a radius of 3.2km and length of 20km. In the classic O'Neill design, we have three large axis-aligned windows in the shell, alternating between purlins of habitable surface. The windows provide light, and presumably affect thermal transfer considerably.

Before we start things spinning, we pump the inner volume full of breathable air. I believe O'Neill proposed 1/2 atmosphere, 20% partial-pressure O and 30% N, but I don't have my copy of High Frontier handy at the moment.

We'll assume that the designers of the habitat have set up the mirrors and cooling system needed to maintain as comfortable a shirt-sleeve environment as possible.

We spin the cylinder at ~0.5rpm to produce 1g of centripetal acceleration on the inner surface, and then we wait for everything to reach some sort of equilibrium-ish state.

The tangential velocity at the outer radius is ~630km/hr (nearly 400 miles/hour), whereas at the axis it's nil.

Because the spin gravity is fictional, acceleration is only imparted to the atmosphere by frictional interactions with the inner surface of the cylinder. Let's assume it's not smooth, but populated with short buildings (probably two or three stories, max), small trees, people, etc.

So, near the inner surface, you have the outer edge of a vortice of atmosphere spinning at ~400mph, and a relatively calm eye at the axis.

There are a lot of oddities about this vortice though. I'm imagining it like a smoothie in a blender, except the whole blender is spinning, with paddles extending from the walls of the blender instead of a blade at one end.

Because there's no true gravity, the air molecules are just hanging out in space until they interact with the inner surface and are imparted with a tangential velocity. The outer edge of the vortice will end up moving at roughly the same tangential velocity as the inner surface, so just like you don't notice the air in your car moving with you at 60mph, the residents shouldn't be bothered by super-hurricane-force winds. But if you consider how ballistic trajectories work in this system, I think there'll be the sensation of a steady "downward"/spinward breeze as the accelerated mass of air meets the curve of the ground.

Then we have to consider thermal transfer and convection, but again, the gravity is fictional, so I'm not sure we can think about this system in the same way we'd think about atmosphere on Earth. Normally you would think of hot air rising, and cooler air sinking, but without real gravity, that goes out the window.

Instead you have the frictional interaction at the outer edge of the atmosphere imparting linear velocities to masses of air. If the blender analogy applies, I think there's going to be some significant amount of pressure differential between axis and outer radius, but I have no idea how much.

So because of the pressure differential, you'd still have warm air "rising" as the higher pressure system seeks a lower pressure environment, and I think you'd have masses of cooler air "sinking" back into the higher-pressure outer radius, with all the weird apparent deflections that you get in a rotating reference frame: anything moving in toward or out from the axis will appear to be deflected anti-spinward.

Then there's the wind shear to account for: going from a theoretically-calm axis to 400mph winds at radius will be no joke.

So between denser, cooler air masses deflecting anti-spinward as they make their way out from the axis, and that steady spinward breeze from frictional linear accelerations at the surface level, it definitely seems like we're going to at least have buffeting spinward and antispinward breezes at the surface. Unless I have any signs reversed in my very-provisional mental modeling! :)

As for thermal inputs, you have radiative insolation through the windows, you have conductive heat transfer through the shell, and you have heat generated by the friction between atmosphere and the inner walls, buildings, trees, etc. (In another question Carlos Zamora suggests convective systems developing between the windows and the land purlins, but he may not be taking the "blender effect" into consideration...)

My question: what in the world is this crazy weather system going to be like, experientially, for shirtsleeve humans living on the surface? There's certainly going to be some crazy wind shear as you move from axis to radius (no human gliders in this scenario, I think, and no fluffy white clouds). I think the surface would be habitable, even if it might always be good kite-flying weather.

The harder the science you can appeal to in the answer, the better, but I don't think anybody has actually studied this question with any amount of rigor, so I'll happily take flights of fancy and imagination as well. :D

I've read all the discussions I could find about atmospheres in one of these contraptions. I see lots of suggestive hints, lots of questionable assumptions, but no clear answer:

Update: @Matthew makes a really good point about convection: it's just various densities of gas seeking an equilibrium. When I first was preparing the question, I was weirding myself out switching between the two reference frames and considering the transition from rest to spinning. As the habitat hits its target angular velocity and the contents come to equilibrium, velocities are transferred from walls to nearby gasses via friction, and then knock-on until everything near the walls is moving in a relatively orderly fashion along with the walls, so everything close to the walls is acting more or less as if they were under gravity, with convection currents and the whole nine yards.

  • 1
    $\begingroup$ Gentry Lee & Arthur C. Clarke guessed on some of these questions later in the Rama series. Spin up & down causes massive winds, cooling settles air closer to the "ground", and even when warm, the closer to the central axis the lower the pressure, so fewer molecules to deal with due to "gravity". Not science based, so not an answer. $\endgroup$ Oct 24, 2019 at 20:08
  • $\begingroup$ I'm new here. Is the science-based tag still too restrictive for what I'm trying to get at? $\endgroup$
    – David
    Oct 24, 2019 at 21:39
  • $\begingroup$ I think science-based should get you decent answers. The tag hard-science will likely get you bad/few answers with a lot of arguments, so I think you're good for avoiding it. Going without any science tag will get you a lot of conjecture, supposition, "maybe"s, and some good answers. IMO, you're good to go, even if you don't get the answer(s) you're looking for. Personally, IDK if anyone has researched large masses of air in a massively large cylinder. It should be interesting if anyone else has heard of such research. $\endgroup$ Oct 24, 2019 at 21:49
  • 1
    $\begingroup$ I'm content with amateur physicists and meteorologists riffing off of what they know. :D $\endgroup$
    – David
    Oct 24, 2019 at 22:09
  • 3
    $\begingroup$ Since no one has actually built an O'Neill Cylinder, I don't know if we can say for sure (outside of some serious computer modeling, at least). However, I want to point out that assuming convection won't work because the gravity is "fake" doesn't necessarily hold. You're talking about phenomena that happen because of density separation. It so happens, we have lots of experience with density-based separation devices that work via "fake" gravity. They're called centrifuges. $\endgroup$
    – Matthew
    Oct 28, 2019 at 21:37

4 Answers 4


Overall, you'll get convection and Coriolis effects from air movement just as you would on a rotating planetary surface. The biggest difference, relative to Coriolis, is that the axis of rotation is parallel to the ground and there's a big velocity difference over a relatively small height range.

On Earth, the Coriolis effect on a strong updraft near the equator is negligible, because you're lifting the air perhaps a kilometer, with a starting radius of 6400 km and a rotation rate of 24 hours. In your O'Neill cylinder, you're lifting air a kilometer from a starting radius of 3.2 km, and a rotation rate of half a minute -- so as you note, there's roundly 200 km/hr difference in rotation speed from ground to 1 km altitude.

So you have thousands of times the Coriolis effect you'd have on Earth. Will that produce tornadoes every time the ground gets a little warmer than the axis?

Probably Not.

Why? An updraft will never manage to build up enough momentum to turn into a (horizontal) whirlwind; it'll start turning spinward almost instantly. Instead, you'll get tiny eddies, surely nothing bigger than a minor dust devil (a meter or two diameter and a couple m/s or so rotation velocity), and to a person on the "ground" they may be perceptible only as a wind gust that blows your hair one way and your pants legs the other. In the end, air can still rise, so you can develop clouds and rain (assuming enough humidity and temperature differential over height) -- but you won't get violent Coriolis storms because there's too much Coriolis effect.

  • 1
    $\begingroup$ I'm preparing an answer myself, and my conclusion is that the entire habitat itself is a violent storm. :D $\endgroup$
    – David
    Oct 29, 2019 at 14:41
  • $\begingroup$ I like the idea of the eddies; like little horizontal dust devils. I think the effect would be more pronounced at the edge of the windows. Perhaps not good kite-flying weather after all? $\endgroup$
    – David
    Oct 29, 2019 at 16:41
  • $\begingroup$ You'll get more eddies where the air is heated, which is where it contacts the heated "ground" -- the windows will require constant heat input to keep them from frosting over (radiating to space except where the mirror provides sunlight). No reason to spend energy heating them above air temperature. $\endgroup$
    – Zeiss Ikon
    Oct 29, 2019 at 16:55

Yes, there will be wind shear, but it is irrelevant:

On Earth, there is a 1656km/hr difference over a distance of 10,000 km between how fast the air is moving at the poles and how fast it moves at the equator. Because it is so spread out, the sheer is so minor that this coriolis effect is only noticeable at the global scale. On your O'Niell cylinder, you will have a 630km/hr difference in air velocities between the center and the perimeter, but that velocity difference is only spread out over 3.2 km. This means your coriolis effect is about 1200 times more pronounced as on Earth, but the nature of how an O'Niell cylinder works will make this added wind shear zero-out to the observer in most cases.

The reason hurricane/tornado winds are so dangerous is not because they are fast, but because they are moving at a different speed than we are. Let's pretend you build a 1km tall tower inside of the O'Niell cylinder. The base will be moving at 630km/hr and the top will be moving around a smaller circle; so, it will only be moving at 433km/hr. Incidentally, the speed of the air at the ground will be about 630km/hr, and the air at at 1km will also be about 433km/hr; so, you will not experience in change in the air velocity relative to your own as you go up. The only entities inside the cylinder that will experience this sheer will be anything free floating that has yet to sync up with the cylinder's rotation.

The only apparent wind you will get will be from what ever whether patterns are formed by the external heating and cooling of the cylinder resulting in updrafts across the gradient.

  • $\begingroup$ "The only entities inside the cylinder that will experience this shear will be anything free floating that has yet to sync up with the cylinder's rotation." Exactly the conclusion I've been coming to. (This is the scenario that first made me wonder about this question.) $\endgroup$
    – David
    Oct 29, 2019 at 16:13
  • $\begingroup$ @David Speaking of unsynced... if you place airlocks at the center of the ends, supply ships should be able to come in under minimal shear and either use elevators to off-load cargo, or ride the wind gradient down; so, even being unsynced should not be a major concern. $\endgroup$
    – Nosajimiki
    Oct 29, 2019 at 17:22
  • $\begingroup$ "if you place airlocks at the center of the ends"... which is exactly how Rama did it, and I believe Howard Taylor also (similar structures show up a few times in Schlock Mercenary). Actually, you pretty much have to do that, because station-keeping relative to something that isn't just spinning, but moving, is ludicrously hard and requires constant fuel expenditure. Hatches away from the axis will tend to throw anything exiting them away from the cylinder (although for some types of egress — just not ingress — this can be a good thing). $\endgroup$
    – Matthew
    Oct 29, 2019 at 20:48
  • $\begingroup$ Interesting! Probably impractical to fly a vehicle into the main volume of the habitat, compared to handling freight in a microgravity section of the habitat, then transferring, but for the right setting it'd be pretty cool. $\endgroup$
    – David
    Oct 29, 2019 at 21:21

It really depends how hot the floor is. If the structure absorbs and transmits a lot of stellar energy inwards as heat at ground level then you're going to get a lot of convection and the thunderstorms etc... that go with it. If you introduce relatively little heat directly at ground level, but primarily heat the atmosphere from the top down, using what I think of as the "central filament" design (per Arthur C. Clarke's Rama), then you have a lot more control over the emission wavelength and it's interaction with the internal landscape and you can create less violent weather.

  • $\begingroup$ O'Neil's design has most of the heat is coming from ground level via the mirrors and windows. Since we're end-on to the sun, there might be a significant heat gradient between window and land purlins. Very interesting! Clarke's central filament is, unfortunately, "sufficiently advanced technology" for the scenario of "things we can actually build". $\endgroup$
    – David
    Sep 2, 2021 at 17:56
  • $\begingroup$ @David We could build a lighting grid along the spin axis with modern technology, it has very little stress of any kind on it after all and what little there is is symmetrically distributed. I'm not sure if we could pump enough photons out of it fast enough to do the job or not though. $\endgroup$
    – Ash
    Sep 2, 2021 at 22:16
  • $\begingroup$ Heh, OK, if you want to make it simple like that! I was imagining some kind of plasma conduit sort of thing. I wonder if LED could pump out enough light for the job? I'd hate to be the guy who has to change the bulbs though. $\endgroup$
    – David
    Sep 4, 2021 at 22:53
  • $\begingroup$ @David I'm not sure but with modern LEDs it is possible, and they can be spectrum tuned as well. Yeah no that's not a job for anyone who has the slightest issue with heights or co-ordination, so that's a hard no for me. $\endgroup$
    – Ash
    Sep 5, 2021 at 1:08

Thinking about blenders got me thinking about tornadoes.

If we consider the largest recorded tornadoes, the 2013 El Reno tornado was 4.2 km wide with max wind speeds of 302 mph. An O'Neill Cylinder with those specifications only produces 0.4g acceleration.

With an O'Neil cylinder we are literally talking about bottling up a supertornado and selling lots in the walls of the storm. 🤯

Of course, because the lots and the people who live in them are moving with the storm, everything is hunky-dory. Maybe?

By examining the characteristics of tornadoes and mesocyclones, we can start to make some interesting suppositions for the builder of these monstrosities.

  • Tornadoes emit sound:
    • high frequency (Abdullah, 1966): atmospheric tornadoes have been described as making "a peculiar whining sound like the buzzing of a million bees", which is drowned out by a roar that begins when the tornado makes contact with the ground.
    • low frequency (Bedard, 2005): tornadoes produce identifiable inaudible infrasonic signatures
  • The pressure drops rapidly in the core of the tornado (as I suspected from the blender analogy). An individual who gave an account of seeing the center of a relatively small tornado above his head reported a "strong gassy odor and it seemed I could not breathe".
  • Tornadoes emit electric signals and fields (Leeman, 2008)
  • The same account describes the center as very still: "Everything was as still as death."
  • The same account describes lightning(!) in the center: "the whole was made brilliantly visible by constant flashes of lightning which zigzagged from side to side", though Wikipedia notes that "Tornadic storms do not contain more lightning than other storms" (citation needed).

The comparison is complicated, of course, since it's hard to separate out the effects from interactions between a tornado and the larger atmosphere. (It's easier to separate out effects from interactions with the ground, as plenty of tornadoes have been observed in mid-air, with distinct characteristics.)

Update to clarify: As with anything regarding rotating reference frames, it's all about your perspective. The good news is that, as I mentioned, the residents are moving with the storm, so the inner surface experience is probably more along the lines of a blustery fall day. The danger of a tornado is in how the vortex interacts with stationary objects on the ground, i.e. in a different inertial reference frame. We've captured this vortex in a bottle, but it's still a vortex, so comparing to mid-air tornadoes is a useful exercise./Update

The most important potential characteristics for a resident in the colony might be:

  • The bees. Is this sonic phenomenon inherent to the tornado itself, or is it an interaction w/ the external atmosphere? Some observers have connected the noise with the small sub-vortices that seem to be emitted from the outer edge of the main tornado vortex, so this may only be a feature of terrestrial tornadoes. Nobody knows, though, so this is still fair game for creative license.
  • The brown note. Infrasound emissions can produce irritability, disturbed sleep, and fatigue.
  • The electric field. We already know that the colony acts as a giant Faraday cage, requiring exterior signals to be repeated inside. Would the noise inside put a damper on radio communication? Would a cell phone work?
  • The lightning(?!). Would the residents be subject to a constant light show at the axis? Hopefully this effect would be amenable to preventative engineering!

Some very interesting non-obvious effects of the setting, useful to storytellers, but perhaps less useful to O'Neill and Bezos' idyllic visions. :)

I'm also interested in the effects at the axis. Old stories about O'Neill cylinders are full of people flying with wings, etc. in the free fall at the center. One book I read even had a mountain at one end that you could climb up to the axis. Sounds like the flying might work, perhaps with oxygen, but don't stray too close to the 400mph walls of the vortex. And watch out for the lightning!

  • 1
    $\begingroup$ Unless there's something draining air pressure from the axis of the cylinder (Leak Drill!! Get to shelter, put on your emergency bubbles, and break out the patch kits!), the pressure sink that's, um, central to tornado formation is absent. Instead, is there reason to believe the airflow (in general, ignoring convection and its Coriolis effects) would be turbulent enough to notice the 640 km/hr, any more than we notice the 1600 km/hr wind we live in at the Earth's equator? $\endgroup$
    – Zeiss Ikon
    Oct 29, 2019 at 15:52
  • $\begingroup$ I'm not sure what you're getting at. I'm pretty sure the pressure differential between center and circumference of a vortex is due to the rotation; you see the same effect in a blender or a sink drain. The only thing missing on the habitat is axial acceleration to shape it into a cone. As for turbulence: as on the surface of the earth, the residents on the inner surface of the cylinder are moving at the same velocity as the local atmosphere. I'm assuming it's good kite flying weather most of the time, but hopefully nothing worse than that. Finally, why the downvote? $\endgroup$
    – David
    Oct 29, 2019 at 16:09
  • $\begingroup$ "I'm assuming it's good kite flying weather most of the time." That's a lot different from saying everyone in the cylinder is living in the wall cloud of an F5 tornado. What does the damage and creates the other effects of a tornado is a combination of bringing near-stratospheric air pressures to ground level, and rubbing a 300 mph wind along the ground. Neither of which occurs in our O'Neill habitat. $\endgroup$
    – Zeiss Ikon
    Oct 29, 2019 at 16:18
  • $\begingroup$ As with anything on this matter, it all depends on your reference frame. From an observer free-floating at the axis, it does rather look like people are living on the wall of a tornado. Drift too far away from the axis and you're going to hit the inner wall of the vortex and experience some nasty effects. I'll update my answer to clarify. $\endgroup$
    – David
    Oct 29, 2019 at 16:29
  • $\begingroup$ There is no inner wall of the vortex. Air viscosity will ensure that there's a more-or-less smooth transition from the "rotating on its axis" air at the zero-gee line, to the "still with respect to the floor" air at the outer hull. Free fall from the axis and you'll pick up horizontal velocity from Coriolis effect, but the air isn't free falling, it's mixing. $\endgroup$
    – Zeiss Ikon
    Oct 29, 2019 at 16:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .