I'm intrigued by the topos of varying season lengths brought to us in the Game of Thrones universe. I think it makes for a brilliant and versatile backdrop for the story, and it is something I'd like to explore in different settings also.

Here is what we know about the seasons of the Game of Thrones world, from observations and anecdotal evidence in the books and series:

  • The seasons vary greatly in length, and happen somewhat infrequently.
  • There was a summer that lasted for at least 9 years (the one coming to an end during the story told).
  • Supposedly there were winters that lasted for 10 years.
  • There still exists the notion of a "year" - suggesting that even within a long winter/summer/... there are still some regular yearly fluctuations to be observed.
  • It seems the cycle is hard to predict - otherwise the Citadel/the maesters would probably have figured it out. This doesn't mean that it has to be random, but suggests that the cycle is at least long/complicated enough that a prediction is not obvious to observants with mediaeval means.
  • For all I know the whole season-length variation theme is mostly prominent on a single continent (Westeros) - but it could be global also, with other locations simply being less extremely affected by it due to generally milder climates.

I'm looking for a science-based answer to how these big variations in season length could come to be.

  • An answer should cover as many of the above points as possible, while being rooted in science. (I'm aware that the GoT universe allows for magic, but I'm not interested in that.)
  • I'm also looking for answers that will keep up the varying season lengths in perpetuity - and not just a series of disconnected freak incidents.

My orbital mechanics knowhow is a bit rusty, but I could imagine that some combination of elliptic oscillation and precession, maybe combined with an eccentric orbit could be behind this? Could this work to the extremes shown above? Can somebody confirm or deny this possibility? Or there could be entirely different natural phenomenons responsible of course...

  • 3
    $\begingroup$ It might be worth it to look into the mechanics of ice ages, which you could argue are like long-period seasons ... possibly there's effects you could use on a timescale that's between ice ages and "normal" seasons. $\endgroup$ – etarion Feb 12 '19 at 15:05
  • $\begingroup$ This is a more detailed question but, yeah, it's a duplicate. The other one has 20 answers. $\endgroup$ – Cyn says make Monica whole Feb 12 '19 at 18:46
  • $\begingroup$ Blast, I even searched before writing this... Well I can't really think of a way to differentiate this question from the other one. Bummer. $\endgroup$ – fgysin reinstate Monica Feb 13 '19 at 8:49

15 Answers 15


The year is related to the motion of the planet around its star. When the star are back to a certain position, one says 1 year has passed.

To account for season length which are in the range you ask, you could have a sort of variable star with small variation of intrinsic luminosity on an irregular period.

When the luminosity is at its peak, you have the Y years long summers, when it is at is bottom, you have the X years long winter.

  • 2
    $\begingroup$ See also Vernor Vinge's A Deepness in the Sky, which has an extreme version of this with the OnOff star creating a cycle of a few hot years followed by a slow decline into a decades-long winter. $\endgroup$ – thirtythreeforty Feb 12 '19 at 14:29
  • $\begingroup$ @thirtythreeforty it must be noted that such "slow decline" would take a couple of days or weeks max, depending on how cold is cold enough and how good of an insulator your atmosphere is. $\endgroup$ – John Dvorak Feb 12 '19 at 18:26
  • $\begingroup$ Sure, if the star did turn completely off suddenly. In Deepness, the star's output gradually declined over a few decades. $\endgroup$ – thirtythreeforty Feb 12 '19 at 18:27
  • $\begingroup$ Ah, okay. That makes sense. Your daylight temperatures will be locked to the sun output and your nighttime temperatures to the length of day. $\endgroup$ – John Dvorak Feb 12 '19 at 18:28
  • $\begingroup$ Defining a year as one orbit of the host star is a relatively new concept. People in a medieval analogue are not going to be aware that their planet is orbiting their star. Pre-heliocentric civilisations defined their year as 'the passing of a full cycle of the seasons', so if the seasons are not directly associated with their astronomical year, their definition of a 'year' will be very different. $\endgroup$ – Arkenstein XII Feb 12 '19 at 19:40

I'd suggest something along the lines of El Niño/La Niña - complex, hard to predict phenomena that can minimize or intensify global weather patterns. For Earth, they can last 2 to 7 years, and effect all seasons across multiple continents. Granted, they don't completely delete seasons, but it's not hard to imagine their effect being more pronounced and/or longer lasting on a planet with different enough geography and oceanic currents. There also isn't anything to say that a world can't have more than one of this sort of phenomenon, and their overlapping cycles could make changes even more intense and less predictable.

  • $\begingroup$ Excellent answer. In addition to the El Nino Souther Oscillation there is the PDO (Pacific Decade Oscillation) and the NAO with periods on the order of decades as well as some weather cycles with periods of weeks. $\endgroup$ – MongoTheGeek Feb 12 '19 at 21:08

The seasons are driven by complex interactions, and while most answers already given here include a piece of the puzzle, none paint a complete picture, so I'll try to integrate some of the major factors.

The seasons on any planet (not just Earth) are governed primarily by the following characteristics:

  • The tilt of the planet's rotational axis relative to its orbital plane (obliquity of the ecliptic)

This is the main driver of seasons on Earth and affects both the duration of sunlight received in a day (the hemisphere more exposed to the sun will receive more hours of sunlight) and the directness of that sunlight (which affects absorption and reflection). The degree of tilt affects the relative strength of the seasons; for instance, if Earth's obliquity of the ecliptic was at or above 45 degrees, then when one hemisphere was tilted toward the sun it would experience dramatically longer, hotter days, while the opposite hemisphere would endure equally long nights.

This plays almost no role in Earth's climate as our axis of rotation is stabilized by our Moon. If we lacked that stabilizing feature, Earth's axis would slowly wobble due to a variety of factors (sloshing of the ocean, flex in the crust, etc.) such that the tilt at one point in the orbit would not be the same at another point in the orbit. This could lead to significant variations in the length of seasons which may be difficult to predict.

  • The eccentricity of the planet's orbit around its star (variation in distance from the star)

This directly affects how much solar energy is available to be absorbed by the planet. Earth has a nearly circular orbit, so we see less than 7% variation in available sunlight between perihelion and aphelion and this mechanism is therefore not a significant factor in our climate. A more elliptical orbit would result in a much warmer average temperature when close to the parent star, as well as an increased orbital speed at that closer distance. Conversely, the average temperature would be much lower and the orbital speed slower at the greatest distance from the star.

This doesn't mean you'd have hotter, shorter summers and longer, colder winters, because summer and winter are mainly defined by axial tilt. A more eccentric orbit would affect the relative strength of the seasons. If the eccentric orbit was somehow coupled with variation in the planet's axial tilt, (e.g. the axial tilt wobbles once every 2 orbits, such that at perihelion on year 1 it might be summer in the north, and at the same point in the orbit on year 2 it's winter), then you might see more dramatic variations season length.

  • Variation in the star's luminous output (solar cycles, star type)

Stars are not static objects, and their luminous and spectral output varies depending on their size, type, age, and other factors our scientists have yet to fully understand. Our own Sun operates on a roughly 11-year cycle of sunspot activity which has been shown to influence the seasons. A more wild star, or a multi-star system, would have an even greater impact on the intensity and possibly the length of seasons.

  • Variation in the absorptivity, heat capacity, emissivity, and reflectivity of the planet (ability to capture, store, emit, and reflect heat from the parent star)

Earth is roughly 75% water, which absorbs sunlight readily and stores heat easily. Our clouds reflect light readily, but can also prevent heat from escaping back into space. As clouds move across the planet's surface, and as the planet rotates on its axis, different materials are exposed to the sun, the local reflectivity and absorptivity changes, wind and evaporation come into play, and we get the complex interactions that we call weather. This variation in the planet's ability to absorb and release heat is possibly the second-greatest driver of the seasons on Earth, and it is so staggeringly complex that many people still don't agree on how it works (climate change, anyone?).

Venus is locked in an endless, brutal summer, not because it is closer to the Sun, but because its own weather readily absorbs heat but does not release it easily. Mars, by contrast, is entombed in eternal winter largely because it has almost no atmosphere to speak of, and thus can't hold onto what little sunlight it receives.

Mixing and matching the above factors could yield a variety of unusual climates and seasons, without the need for bizarre celestial circumstances or magic. Although, it is fair to say that using the weather to justify unusual weather may require a bit of handwavium since that subject is so complex.

  • 1
    $\begingroup$ Axial tilt is far more significant than eccentricity on Earth. It's quite easy to get an eccentric orbit where solar distance is the dominant factor and axial tilt mainly determines day length. $\endgroup$ – Mark Feb 12 '19 at 21:15

Our own sun has an 11 year cycle of sunspot activity. This means it is 5 1/2 years from minimum to maximum activity. It could be some solar cycle that causes the season lengths.

The reason we have seasons is because of the tilt of the earth's axis. Earth's orbit is like 99.9% circular, and doesn't contribute as significantly to the seasons. This almost implies that a more elliptical orbit could be catastrophic. It could be possible with precession though, if it was close to the length of the year it could cause long seasons. It would more likely be very regular and periodic though in that case. Earth's precession takes 26000 years per cycle.

  • 1
    $\begingroup$ Earth's orbit has an eccentricity of 0.0167, meaning it is only 98.33% circular. $\endgroup$ – Skyler Feb 12 '19 at 19:48
  • $\begingroup$ @Skyler wow practically an oval then $\endgroup$ – jreese Feb 12 '19 at 19:51
  • $\begingroup$ @Skyler how do you calculate that anyway? For eccentricity = 0.0167, it gives an aspect ratio of 0.99986. Where do you get 98.33% circular from? $\endgroup$ – jreese Feb 13 '19 at 14:04
  • $\begingroup$ ...from a bad calculation that assumed an eccentricity of 1 was 0% circular. $\endgroup$ – Skyler Feb 13 '19 at 14:34

Many good answers here, but for a real world example, you want to look at Kepler 413-b.

The tilt of the planet's axis varies by about 30 degrees over an 11 earth year cycle. (61 of its years). Astronomers believe the phenomenon is caused because it orbits a binary star on a tilted orbital plane. If you take the same phenomenon on a planet that orbits wider and slower in a system's habitable zone, you'd have winter and summer swap about once every 30 years, but during the years where the axis lines up with the star you would not experience seasonal change at all resulting in several years of winter at the poles and several years of summer near the equator.


Your planet does not have seasons based on a "standard" orbit around a single central star. For example:

  • If you have a binary star system, sometimes you will pass from "close to Star A" to "close to Star B", for an extended Summer. Or, with a "hot" and a "cool" star, the season lengths will vary based on the ratio between how fast your planet orbits the pair, and how fast your Stars orbit each other
  • If your planet's orbit is perturbed by massive objects in the solar system (e.g. nearby gas giants) then it may not be an oval - sometimes you are pushed into a closer orbit (longer summers), sometimes you are thrown out into a wider one (longer winters)
  • Similar to the above, if your Seasons are instead partially determined by tidal heating then the other objects in your solar system can cause unusual patterns per the ratio of their orbits, repeating over centuries instead of months (e.g. a 400 year cycle of long and short seasons - long enough that records can be made, but too long to be "in living memory")
  • 1
    $\begingroup$ While the orbit being perturbed by other massive objects in the solar system is certainly possible, this is unlikely to be stable. In the long term, it's much more likely to result in the planet being perturbed into some alternate, stable orbit, being ejected from that solar system, or colliding with some other large body in the system. It's unlikely for something which is being perturbed within a human-noticeable time-frame to have had an orbit that's stable long enough for life to have developed and flourished. $\endgroup$ – Makyen Feb 12 '19 at 20:09
  • $\begingroup$ It's possible that the perturbations are a new thing as a result of a massive body entering the system, but that doesn't bode well for the long-term stability of the planet's orbit. $\endgroup$ – Makyen Feb 12 '19 at 20:09

El Niño

El Niño and La Niña are terms which describe the biggest fluctuation in the Earth's climate system and can have consequences across the globe. The fluctuation sees changes in the sea-surface temperature of the tropical Pacific Ocean which occur every few years.

These events are due to strong and extensive interactions between the ocean and atmosphere. They are associated with widespread changes in the climate system that last several months, and can lead to significant human impacts affecting things such as infrastructure, agriculture, health and energy sectors.

In England we basically only get winter in El Niño years, otherwise it remains mild and damp and snows only on high ground.

If you give them a weather cycle similar to El Niño but is more pronounce and runs on a cycle over several years it's possibly not unreasonable for northern climes to have winter snows that don't melt for years on end or completely miss a winter here and there.


The binary star hypothesis is seconded by astronomers. But the configuration is a bit unusual to produce

enter image description here
Motion of the planet above and beyond the orbital plane of the stars in the Sitnikov problem. x-axis shows time in units of revolution of the primaries. y-axis gives distance from planet above and beyond the orbital plane. The plot shows motion of the planet for three different initial conditions. In the beginning, the motion is very similar, but after about 33 revolutions of the primaries, first differences appear and after 60 revolutions the three initial conditions develop into totally different types of motion. The red line shows a planet moving very far away from the stars in one direction; the blue line shows the same behaviour with a planet moving away in the other direction. In both cases a very long and possible endless winter would begin. The black line shows a planet that continues to have a more or less regular sequence of seasons.

The planet would have to be

enter image description here Configuration of the Sitnikov problem. m1 and m2 are two stars and m3 is a planet, moving perpendicular to the orbital plane of the stars.

Now that we know which explanations do not work, we want to introduce the one astronomical configuration which is able to successfully account for the chaotic seasons of Westeros: the “Sitnikov Problem” (Sitnikov 1961). It is a sub-case of the elliptical, restricted three-body problem. Due to the complexity of the possible kinds of motions in this system, it is – despite its simple mathematical formulation – a well and long studied, unique dynamical model in celestial mechanics.

Let there be three masses m1, m2 and m3, where m1 and m2 are called “primaries” (i.e. the stars) and m3 is a smaller (and in the mathematical model massless) body (i.e. a planet). If we let the two stars orbit under their mutual gravitational influence, they will both move around their common barycenter. Now, let us assume a coordinate system that has its origin in that center of mass. If we put the planet exactly in the barycenter, it will be influenced by the gravity of the two stars (but because it is thought as massless, it itself will not disturb the motion of the stars) and follow an orbit along a straight line through the center of mass and perpendicular to the barycenter of the stars (see Figure 1).

This "Sitnikov problem" is concluded with

from the many explanations brought forward by other authors to explain the seasons in the world of “Game of Thrones”, none is able to satisfyingly address all details. There is only one astronomical explanation that works and that is the celestial mechanical case of the Sitnikov problem. Two stars and a planet that orbits along a straight line perpendicular to the orbital plane of the stars can build a planetary system, that has all the properties to reproduce any desired sequence of seasons. Not only that, but due to the nonlinear properties of the Sitnikov configuration and the mathematical theorem of Moser it is proven, that there is no sequence of seasons, however strange and obscure, that can not be found on a planet in such a system.

We thus have solved once and for all the problems surrounding the question if and when “Winter is coming”.

From Florian Freistetter & Ruth Grützbauch: "Sitnikov in Westeros: How Celestial Mechanics finally explains why winter is coming in Game of Thrones", 2018 (arXiv)

  • $\begingroup$ +1. However, from the article, However complicated and chaotic the seasons of Westeros are (or will be in the future of“Game of Thrones”): one can find a planet moving in a Sitnikov configuration that has exactlythat type of seasons. The erstwhile astronomer/students didn't consider planetary rotation at all, which would lead to completely frozen (aka "uninhabitable") northern and southern hemispheres in turn. The solution was clever, but not perfect, which isn't surprising since the author George Martin couldn't have cared less about the veracity of celestial mechanics. $\endgroup$ – JBH Feb 12 '19 at 22:01

I second the "Binary Star" hypothesis.

If there is a system with two start that are wildly different in size, one of these stars will be the center of that system, orbited by everything (including the second star).

if the two stars were roughly equal size, they would orbit each other, but a massive and a dwarf star would have the massive star as a clear center.

now, this planet would track its' "year" by passage around the primary star, but the relative location of the secondary star will make a great difference to the overall temperature of said planet, making for extended summers if it is close, and extended winters when it is far away (especially is the planet is on the colder end of the "Goldilocks zone").

highly eccentric orbits for both the planet and the child star would mean differing relative speeds, which would make the "extended summer" and "extended winter" be of varying lengths throughout the centuries as well

  • 1
    $\begingroup$ Consider our own solar system, but with Jupiter just enough larger that it would be a star rather than a mere planet. Seasons on Mars would be greatly affected by the second star. $\endgroup$ – Monty Harder Feb 12 '19 at 19:41
  • $\begingroup$ @MontyHarder Something like that indeed. Perhaps with more elliptical orbits for less predictable patterns, but that is the base idea i posited $\endgroup$ – ThisIsMe Feb 13 '19 at 8:27

A Binary star system in which they orbit each other relatively close to each other (thus from ground observers it appears as one star, hence the GoT universe) But the stars have vastly different solar outputs.

Star A: High solar radiation.
Star B: lower solar radiation.

Binary with eclipses

Add in an elliptical orbit and the stars orbiting relatively slow you have the possibility of being aligned to the high output star for a multitude of years with a variance of cool and warm allowing recognition of a year. Then a shift starts to occur as you face the lower output star which blocks the higher radiation of Star A.

enter image description here

The people noting "winter is coming" are keen to the changes that happen when one star is passing the other, or in other words, when Star B fully eclipses Star A. this goes on for a 5 year period until the transition happens again where "Summer is coming" occurs. The transitional periods multiplied by the ellipse orbit allows for the infrequent 8-10 year timeframes between 'winter' and 'summer'

  • $\begingroup$ While a potentially interesting solution, I'm not sure it quite works. As a "year" is one orbit of the star, at least part of the year you'd see the other (non-blocked) side of the binary, unless the orbital periods are matched. Matching orbital periods with different orbital distances is difficult/impossible. You might posit it only blocks the bright sun for part of the year, and long winters are due to blockage during to (normal, tilt-based) summer. But the opposite would be blockage during normal winters, which shouldn't lead to long summers, but rather even harsher winters. $\endgroup$ – R.M. Feb 12 '19 at 16:28
  • $\begingroup$ (Not to say that this can't be made to work, but you'd need to work more heavily on the exact dynamics of the system.) $\endgroup$ – R.M. Feb 12 '19 at 16:30
  • $\begingroup$ In order to have both stars inside the orbit of a habitable planet, as shown in the figure, they would have to be orbiting each other very fast. If the brighter star is occluded by the other at some point in time, it will very soon be completely out from behind the other star from the viewpoint of the planet. I don't think there's any way this could be a multi-year cycle. $\endgroup$ – David K Feb 12 '19 at 18:22

The best explanation I have seen for the unpredictability of seasons in the Game of Thrones universe, is that their world has rapid Milankovitch Cycles.

As has been suggested in other answers, but not entirely explained, the Earth's ice-ages are driven by the interaction between several varying cycles. The most important of these are Precession, Axial Tilt, and Eccentricity.

Because the three cycles have very different periods, they align or misalign to create what would be a highly unpredictable pattern to those who do not have an understanding of the astronomical causes.

On Earth, these cycles take tens of thousands of years, so they are not obvious on the scale of human lifetimes, but if they each had much shorter periodicity, they would act in effect as a highly unpredictable seasonal cycle.


Lets start with what we can rule out. Here on earth, the seasons depend on the axis tilt of earth to sun. That There is a bit of wobbling, but not enough for summers that last years. So what else can be responsible for varying season lengths?

  • Different distance of sun and earth. Not very reasonable, why would the orbit change back and forth? (Elliptical orbit is no explanation, as 1 year = 1 full orbit)

  • Different sun activity. Our sun is not a static shining light. It is a violent nuclear reaction. So maybe it could be explained by different activity of the sun.

  • Global warming / cooling. There could be a lifecycle of species that have effects on climate change that the citadel knows nothing about.


A planet with no moon would wobble on its axis much more than Earth does. This could create rapid climatological changes that people living on that planet might perceive as changes to the length or quality of their "seasons".

Earth's seasons are regular and predictable because our axial tilt doesn't wobble that much, so as the Earth revolves around the sun the amount of sunlight received at any latitude during any month tends to remain constant. A planet with a rapid and notable wobble to its axial tilt would not have that regularity and predictability.

Specific to the Game of Thrones world, it is my understanding that their odd seasons began when their moon was destroyed by a cosmic impact. That science isn't great, because their moon was only shattered and not utterly destroyed, but it may be Martin's intention that the season "problem" is related to the moon "problem".


In my climate science course the explanation of ice ages in the quaternary period was epicycles of precession of:

  • the tilt of the Earth (going back and forth between various amounts of tilt)
  • the eccentricity of the orbit (how close the orbit is to a circle)
  • the direction of the Earth's tilt relative to its eccentricity

Right now the northern hemisphere has its winter at the point in its orbit where it's closest to the sun. Hence, the winter is slightly less intense. Every few dozen thousand years, these "line up" and the northern hemisphere has winter at the farthest point in its orbit. The winter gets more extreme, enough for an ice age. The focus is the northern hemisphere because the southern hemisphere has so much ocean moderating the temperature and preventing these extremes.

(there are more details and my recollection is hazy, so please forgive some technical errors)

I think a faster, more extreme version of this phenomenon could work. So every ten or so years, the tilt and eccentricity line up. On top of that, the eccentricity and tilt would be varying in intensity, causing more or less extreme versions of the long summer+winter pattern. Within these epicycles, the planet is still orbiting and still causing annual seasons.

I don't know if this more extreme version is physically possible.

This doesn't solve the "maesters can't predict it", since Wikipedia says that axial precession has been analyzed accurately for a long time. Maybe there are lots of feedback loops: trapped methane, albedo, forest cover, etc., that would vary the length of each period within each cycle. Maybe the rates of precession of each phenomena are changing from other astronomical bodies.

This could combine with the El Nino idea; maybe the particular area gets a warm current from somewhere (like Europe does) but some of the orbital cycles will sometimes stop or reverse this current (if the axial precession is severe enough, there might be noticeable changes in Coriolis forces driving the current!).


Volcanic Activity

At least one proposed cause of Earth's own Little Ice Age is a spike in major volcanic eruptions

A planet without axial tilt, or very minor axial tilt would normally appear to not have seasons. Significant changes in volcanic activity would introduce artificial 'seasons' that could last many years.

An increase in volcanic activity would indicate the start of winter. The length and severity of each winter would depend on the amount and type of materials released in the eruptions.

On the converse, a decrease in volcanic activity would bring about summer.

The Long Spring and Brutal Summer

Imagine a planet with such volcano moderated seasons on the inner edge of its star's Goldilocks zone. When the volcanic activity is at a minimum, the planet would boil, huge tropical storms would rage in some areas, while others suffer drought. When the activity is at a maximum, the planet would freeze.

The ideal situation then is to be in the Long Spring: Some steady volcanic activity, not enough to bring about winter, but enough to prevent summer.


Not the answer you're looking for? Browse other questions tagged or ask your own question.