I, like everybody I suppose, have read the explanations why the colour of the sky is blue:

... the two most common types of matter present in the atmosphere are gaseous nitrogen and oxygen. These particles are most effective in scattering the higher frequency and shorter wavelength portions of the visible light spectrum. This scattering process involves the absorption of a light wave by an atom followed by reemission of a light wave in a variety of directions. The amount of multidirectional scattering that occurs is dependent upon the frequency of the light. ... So as white light.. from the sun passes through our atmosphere, the high frequencies become scattered by atmospheric particles while the lower frequencies are most likely to pass through the atmosphere without a significant alteration in their direction. This scattering of the higher frequencies of light illuminates the skies with light on the BIV end of the visible spectrum. Compared to blue light, violet light is most easily scattered by atmospheric particles. However, our eyes are more sensitive to light with blue frequencies. Thus, we view the skies as being blue in color.

and why sunsets are red:

... the light that is not scattered is able to pass through our atmosphere and reach our eyes in a rather non-interrupted path. The lower frequencies of sunlight (ROY) tend to reach our eyes as we sight directly at the sun during midday. While sunlight consists of the entire range of frequencies of visible light, not all frequencies are equally intense. In fact, sunlight tends to be most rich with yellow light frequencies. For these reasons, the sun appears yellow during midday due to the direct passage of dominant amounts of yellow frequencies through our atmosphere and to our eyes.

The appearance of the sun changes with the time of day. While it may be yellow during midday, it is often found to gradually turn color as it approaches sunset. This can be explained by light scattering. As the sun approaches the horizon line, sunlight must traverse a greater distance through our atmosphere; this is demonstrated in the diagram below. As the path that sunlight takes through our atmosphere increases in length, ROYGBIV encounters more and more atmospheric particles. This results in the scattering of greater and greater amounts of yellow light. During sunset hours, the light passing through our atmosphere to our eyes tends to be most concentrated with red and orange frequencies of light. For this reason, the sunsets have a reddish-orange hue. The effect of a red sunset becomes more pronounced if the atmosphere contains more and more particles.

Can you explain why the colour of the sky passes from blue to orange/red skipping altogether the whole range of green frequencies?

I have only heard of the legendary 'green, emerald line/ flash'

enter image description here

that appears in particular circumstances

Green flashes are enhanced by mirage, which increase refraction... is more likely to be seen in stable, clear air,... One might expect to see a blue flash, since blue light is refracted most of all, and ... is therefore the very last to disappear below the horizon, but the blue is preferentially scattered out of the line of sight, and the remaining light ends up appearing green

but I have never seen it, nor do I know anybody who ever did.


6 Answers 6


The sky does not skip over the green range of frequencies. The sky is green. Remove the scattered light from the Sun and the Moon and even the starlight, if you so wish, and you'll be left with something called airglow (check out the link, it's awesome, great pics, and nice explanation).

Because the link does such a good job explaining airglow, I'll skip the nitty gritty.

So you might be thinking, "Jim, you half-insane ceiling fan, everybody knows that the night sky is black!" Well, you're only half right. The night sky isn't black. The link above explains the science of it, but if that's not good enough, try to remember back to a time when you might have been out in the countryside. No bright city lights, just the night sky and trees. Now when you look at the horizon, can you see the trees? Yes, they're black silhouettes against the night sky. But how could you see black against black? The night sky isn't black. It's green thanks to airglow (or, if you're near a city, orange thanks to light pollution).

Stop, it's picture time. Here's an above the atmosphere view of the night sky from Wikipedia:

And one from the link I posted, just in case you didn't check it out:

See, don't be worried about green. The sky gets around to being green all the time.

  • 1
    $\begingroup$ But not noticeably green like red or blue is, and no green is visible in between the red sunset and the blue sky. Why? $\endgroup$ Commented Mar 11, 2019 at 13:35
  • 1
    $\begingroup$ @GrantGryczan First off, it is noticeable. If you go far from a city on a clear night, you can see a green hue. Also, as for the green at sunset, see the answer below this one $\endgroup$
    – Jim
    Commented Mar 12, 2019 at 11:59
  • 1
    $\begingroup$ But not as noticeable as red or blue. That's my point. You don't have to go far from a city to see red or blue. Yet, between those two hues, we're missing green! $\endgroup$ Commented Mar 16, 2019 at 17:24
  • $\begingroup$ In my experience, tree silhouettes are visible out in the countryside not because of the greenish sky background, but because they occlude the stars. (Stepping out of a tent to go to the bathroom in the middle of the night far from the nearest artificial lights can be remarkably disorienting.) $\endgroup$
    – Vikki
    Commented Apr 5, 2021 at 23:35
  • 1
    $\begingroup$ @Vikki-formerlySean I grew up in the country far from artificial light. I've also been camping. I know what you speak of. When camping, you have trees all around blocking essentially the entire sky and, yes, it's very disorienting. In the country, you get a better view of the sky. The stars definitely help, but you can still see the trees against the sky itself $\endgroup$
    – Jim
    Commented Apr 7, 2021 at 12:05

Note well: What we perceive as color is bit of a tricky subject. This is a different question, one that has been asked and answered multiple times at this site. Per the typical human eye response, sunlight at the top of the atmosphere is about as "white" as "white" can be.

Some of that incoming sunlight is reflected back into space, some is absorbed by the atmosphere, and some is scattered by the atmosphere. There are two main kinds of scattering: Rayleigh scattering, which selective scatters high frequency light (violet and blue) much more than low frequency light (red), and Mie scattering, which scatters light more or less evenly across the spectrum. Rayleigh scattering is a clear sky phenomenon. Mie scattering results when small droplets of water are suspended in the air (e.g., clouds).

The sky is blue at noon on a clear day because of Rayleigh scattering. The atmosphere selectively refracts the violet/blue end of the spectrum. After undergoing a large number of such refractions, some of that refracted light comes down to the ground. The result: A blue sky. Note that we would see a white Sun against a black background if this refraction did not occur. This is what astronauts and cosmonauts see in space.

At sunrise and sunset, the violet, blue, and green parts of the incoming sunlight are almost gone. The result: We see a quite reddish sun. We still see a predominantly blue sky even at sunrise and sunset. It's only at the horizon where we see a reddish colored sky. Rayleigh scattering does affect red light as well, just not as strongly. That red colored sky is partly a result of Rayleigh scattering. The high wavelength light has been scattered away. The red light, to a much lesser extent. The reddish sunlight that makes it through hundreds of kilometers of sunrise/sunset sky is eventually scattered as well, so we see that reddish light as a reddish sky. A slightly cloudy makes for even better sunrises and sunsets because now Mie scattering can occur as well. Mie scattering can spread that incoming reddish sunlight across a good portion of the horizon sky.

With regard to green, you can see a greenish sky at sunrise/sunset. There oftentimes is a narrow band between the blue sky away from the horizon and the red sky at the horizon where the sky appears greenish.

The above is a clear sky phenomenon, and it's only a narrow band of the sky that appears green. An infrequent event can occur where the entire sky appears green. This is called a green thunderstorm. This happens when a thunderstorm occurs at just the right time of day, with just the right height of clouds, and just the right lighting. This can make the entire sky rather than just a thin band appear to be grayish green, and sometimes very green.

Update: Green Flash

Here's a picture of a green flash:

Green flashes are a mirage. They are real effects; they are not an optical illusion.

There's a distinction between a mirage and an optical illusion. On driving along a black asphalt road on a clear, hot summer day, you may have seen the road ahead appear to be covered with shimmering water. That's a combination of a mirage (a real effect) and an optical illusion (your brain misinterpreting what you see). The real effect is that atmospheric refraction results in you seeing a reflection of the air on the road. The reflection is anything but even. The road is hot, so hot you might be able to fry an egg on it, and this results in atmospheric disturbances just above the asphalt. The result is that the reflection is rather uneven and changes over time. This makes the reflection appear to shimmer. The reflection and the shimmering: These are real effects. You can photograph them, you can explain them with physics. The optical illusion is that your brain misinterprets these real effects as shimmering water.

Green flashes are best seen where the Sun rises or sets over the ocean, for a number of reasons. One is the flatness of the horizon. You need to be able to "see" the Sun before it has risen / after it has set. Another is the high heat conductivity of water relative to land. This helps create local thermal variations near the surface (similar to the black asphalt described above).

One form of green flash occurs when the ocean surface is warmer than the air above. This results in an inferior-mirage flash that lasts about a second. It occurs just a moment after the top of the Sun appears to have slipped below the horizon at sunset or just a moment before the top of the Sun will appear at sunrise.

Another form of green flash occurs when the ocean surface is cooler than the air above. This results in a mock-mirage flash than lasts about a second. These occur just a moment before the top of the Sun will appear to slip below the horizon at sunset or just a moment after the top of the Sun appears at sunrise.

A third form occurs when there is a strong atmospheric temperature inversion above the observer. This results in a sub-duct flash, where a sometimes sizable portion of the Sun appears to pinch off the setting/rising Sun and turn green. This form lasts significantly longer than the first two forms.

A number of photographs have been taken of all three of the above forms. There is one last form of green flash, the very rare and very short lived green ray that appears to emanate from the top of a green flash. A number of reputable observers have reported this phenomenon, so apparently it is real. However, no reputable observer has yet caught this phenomenon in a photograph.

  • 2
    $\begingroup$ Does pollution (e.g., particulates, CO$_{x}$, NO$_{x}$, SO$_{x}$, etc.) cause Mie scattering as well? I seem to recall that pollution, ironically, aids in making sunsets/sunrises more colorful. $\endgroup$ Commented Sep 25, 2014 at 13:36
  • 1
    $\begingroup$ There are definitely some green pixels in that image, but I think if you're actually looking at the sky it's very difficult to perceive it as green, rather than a sort of superposition of orange and blue. I'll have to pay attention to it next time I see a sunset like that. $\endgroup$
    – N. Virgo
    Commented Sep 25, 2014 at 13:48
  • 1
    $\begingroup$ @honeste_vivere - That's correct. Molecules are much smaller than the wavelength of visible light, so they don't cause Mie scattering. Mie scattering results from particles with about the same size as the wavelength being scattered. With regard to visible light, that's 475 to 650 nanometers. That's the size of a typical water droplet suspended in a cloud -- or a typical chunk of ash from a coal plant. $\endgroup$ Commented Sep 25, 2014 at 14:05
  • $\begingroup$ @DavidHammen Ah, so it's the particulates not the XO$_{x}$'s. Yeah, I suppose those molecules are, at best, a few nm in size. $\endgroup$ Commented Sep 25, 2014 at 18:43
  • $\begingroup$ "At sunrise and sunset, the violet, blue, and green parts of the incoming sunlight are almost gone." But the opposite side of the sky is still blue, so blue is not gone. Then, different colors should "go" no all at once, but one after another (depending on wavelengths). This means that we should see a rainbow-like gradient, but we don't. Your answer does not explain why. $\endgroup$ Commented Jul 2, 2015 at 20:27

The accepted airglow answer might be technically true, but it does not answer the question! The existence of an additional and very faint source of green light in the atmosphere does not explain the absence of the green light in the sunset sky gradient.

I wasn't satisfied with other answers either. The only satisfactory answer I could find is this one. Below I explain it in my own words and add my own details. Treat this answer as a personal opinion and not a source of truth.

So there are two main factors that contribute to the blue color for the sky at day and red at sunset.

The first one is Rayleigh scattering. The sunlight contains all colors, but they scatter in different amounts:

Rayleigh scattering graph

Blue is the first to scatter, so the sky normally is blue.

The second factor is that the thickness of the atmosphere is uneven when measured from a point on the ground to the zenith and from the same point on the ground to the horizon:

zenith vs horizon

The closer the sun gets to the horizon, the longer path does the light take through the thickness of the atmosphere. The longer the path, the more other colors get scattered.

But the farther a color is from blue in the spectrum, the more it tends to follow a straight path through the thickness of the atmosphere. Red is the toughest: even when it scatters, it scatters at small angles. Thus, you can see red only around the sun because that's where the angles are the smallest.

But every color scatters at it's own angle, not just blue and red. This means that we should see a rainbow-like gradient on the sunset sky, but we don't. Why?

The matter is that the green is always there, but it's never there alone. It's always mixed either with red (forming yellow) or blue (forming turquoise).

Thus, the sunset gradient is as follows: R-RG-GB-B, that is red-yellow-turquoise-blue. And that is indeed the gradient we see in the sky. The blue color of the evening sky is not purely blue, it's turquoise.

But our eyes and brain perform color correction and still identify it as blue. We have no color reference when we look at the sky. If you had a pure blue color object in your hand and compared the color to the color of the blue sky close to the yellow part, you'll notice that the sky is not blue but rather turquoise (or even aquamarine, see photos below).

But why is green a dirty (mixed) color and red and blue are pure colors in that gradient? That makes no sense. They should then be all mixed?

Well, the answer is that red and blue are at the ends of the gradient and they have nothing to mix with. Here's the graph of colors vs intensity in the sun spectrum:

Solar spectrum

If the sun's spectrum were wider (with large amounts of ultraviolet on the blue side and infrared on the red side) and our eyes could register those extra colors, the sunset gradient would be like this:

I-IR-RG-GB-BU-U, that is
infrared -- infrared-red -- yellow -- turquoise -- blue-ultraviolet -- ultraviolet.

See? This imaginary sky has no pure red and no pure blue.

Note that even the red color of the sunset sky is not technically pure red. It still contains some blue and green and all other intermediate colors. It's just the matter of which color(s) dominate.

And finally, sometimes you do actually see the green (aquamarine) color in the sunset gradient:

sunset photo sunset photo

  • 2
    $\begingroup$ see, I like this answer, it adds a very nice sense of completion to the entire thread. Every good question needs at least two answers: the accepted one that is technically correct (which the Central Bureaucracy insists is the best kind of correct) and another to fill out all the remaining details. One that people can read for further information and to get a more complete picture. That one is this answer. +1 $\endgroup$
    – Jim
    Commented Jul 3, 2015 at 12:30
  • $\begingroup$ It is still a bit mystery to me though. We can clearly identify most other colours on the sky. Violet? Check. Blue? Check. Yellow? Check. Orange? Check. Blue? check. We can also identify different variations of all these colours... except bright yellow... which happens to be close to green. $\endgroup$
    – Spero
    Commented Nov 21, 2017 at 19:05
  • 1
    $\begingroup$ @Spero Your eyes and brain perform color correction that you don't realize. Try comparing sky colors against a paper color palette, and you'll see that what you call blue is not purely blue and red isn't purely red. $\endgroup$ Commented Nov 23, 2017 at 12:04
  • $\begingroup$ @Spero It also has to do with the peak spectral sensitivity of each of the three types of cones in our retinas. Our eyes are not spectrometers, equally sensitive to all wavelengths. The colors we perceive are not, strictly speaking, a physical attribute of the various wavelengths of light. They are the way our specific eye-brain system perceives various wavelengths of light. There are some colors that we perceive that can not be produced by a single wavelength of light. Magenta, for instance, is not a spectral color. It requires some of both ends of the visible spectrum and not much else. $\endgroup$
    – Michael C
    Commented Dec 11, 2020 at 10:49
  • 1
    $\begingroup$ In fact, the majority of the colors that humans are capable of perceiving are not spectral colors, but require multiple wavelengths of light to stimulate the three types of retinal cones to various response levels. When viewed in CIE color space, spectral colors are along the edge of the curve between violet and near-IR. Everything else is not a spectral color. Equating a specific wavelength to a specific color is a bit of a false assumption. Animals with different cone responses perceive the same wavelengths of light, and combinations of wavelengths of light, differently than humans do. $\endgroup$
    – Michael C
    Commented Dec 11, 2020 at 10:56

The hand waving explanation in your question is called Rayleigh scattering

Rayleigh scattering results from the electric polarizability of the particles. The oscillating electric field of a light wave acts on the charges within a particle, causing them to move at the same frequency. The particle therefore becomes a small radiating dipole whose radiation we see as scattered light.

The question then becomes why not see the sky as green rather than blue

Rayleigh scattering is inversely proportional to the fourth power of wavelength, so that shorter wavelength violet and blue light will scatter more than the longer wavelengths (yellow and especially red light). However, the Sun, like any star, has its own spectrum and so I0 in the scattering formula above is not constant but falls away in the violet. In addition the oxygen in the Earth's atmosphere absorbs wavelengths at the edge of the ultra-violet region of the spectrum. The resulting color, which appears like a pale blue, actually is a mixture of all the scattered colors, mainly blue and green. Conversely, glancing toward the sun, the colors that were not scattered away — the longer wavelengths such as red and yellow light — are directly visible, giving the sun itself a slightly yellowish hue. Viewed from space, however, the sky is black and the sun is white.

Italics mine. Here color perception from the physiology of the retina enters, and it is not a physics question any longer, but a question of how the cones in the retina interpret the impinging spectral frequencies.

The reddening of sunlight is intensified when the sun is near the horizon, because the volume of air through which sunlight must pass is significantly greater than when the sun is high in the sky. The Rayleigh scattering effect is thus increased, removing virtually all blue light from the direct path to the observer. The remaining unscattered light is mostly of a longer wavelength, and therefore appears to be orange.

So I presume we interpret "blue" even when green is mixed up with it ( as we interpret white when all colors are mixed up) and it is a matter of perception.

The green flash can also be explained with scattering and the difference in the wavelengths between blue and green. Here is a link with photos included. I suspect that the physiology of the eye also plays a role in the perception.

It is interesting that the flashes are not always green , here is an interesting analysis of this refraction process.

  • $\begingroup$ The green flash is very real. It's a mirage (by definition), but a very real mirage. It's a result of atmospheric refraction effect rather than human perception / human physiology. $\endgroup$ Commented Sep 25, 2014 at 16:28
  • $\begingroup$ Your answer does not explain why there is no rainbow-like gradient in sunsets. Also, this is not true: "The Rayleigh scattering effect is thus increased, removing virtually all blue light" -- the opposite side of the sky is still blue, though light travels almost twice as longer distance to there. But the opposite side of the sky is darker, of course. If it's darker because the blue light is removed by Rayleigh scattering, why is the green component not revealed? $\endgroup$ Commented Jul 2, 2015 at 20:34

I scrolled through the answers expecting to find at least one with a chromaticity diagram, but there are none, so I'm writing my own.

Short answer: it simply isn't true that colors are arranged in a line and that to get from blue to orange one must go through green. It isn't true of psychological colors, nor is it true of physical colors (electromagnetic waves).

Here's a CIE xy chromaticity diagram with the approximate location of sky blue and "sky orange" circled.

(I made this by finding some photos of the sky online that looked realistic, using an eyedropper tool on them, converting the sRGB colors to CIE xy, and then placing the circles by hand. I have normal color vision and I used an IPS monitor with accurate sRGB color, but still, the circles should be considered very approximate.)

CIE xy coordinates are rational linear functions of the physical spectrum, which means that linearly interpolating between two physical spectra gets you only the perceptual colors lying on the straight line segment between those colors on the xy diagram. So, supposing that the transition from midday to dusk is roughly linear, we'd expect the intermediate colors to be in the pinkish range, on the opposite side of the white point from green. A small deviation from linearity could produce a pale green, but it would take a large nonlinearity to get a highly saturated green.

You can also understand this by thinking about the spectra directly. The spectrum of sky blue has (compared to neutral white) more light at high frequencies, less at low frequencies, and an intermediate amount at intermediate frequencies. The spectrum of sky orange is roughly the reverse of that. There are not many processes in nature that change the frequency of light, and none of them are involved in determining the color of the sky, so when you model the smooth transition from one of these spectra to the other, you should model it as acting independently on each frequency. When going from blue to orange, the high frequencies start out bright, end up dim, and are in between in between. The low frequencies are the other way around. The middle frequencies start out at medium intensity, end at medium intensity, and are probably at medium intensity in between. Thus, the midway point between the peak at violet and the peak at red isn't a peak at green; it is, rather, a more or less flat spectrum. It could have enough of a peak in the middle to appear as pale green, or enough of a trough to appear as pink, but there's no reason to expect a bright green (or bright magenta).

The only way you could plausibly get a bright perceptual green in the atmosphere is by frequency-dependent diffraction, the same phenomenon that gives you rainbows. This is what causes green flashes, as you can see in Andrew Young's lovely simulations, particularly this one.

There is one physical phenomenon in which a smooth transition from bright blue to bright orange does go through bright intermediate rainbow colors, and that's Doppler shift. But a Doppler shift of that magnitude requires relative speeds that are a substantial fraction of the speed of light. That isn't what happens in the atmosphere.

  • $\begingroup$ I don't see why the transition to midday to dusk should be exactly linear, but it make sense it should be close to linear, so something like this must be the right answer. $\endgroup$ Commented Jan 11, 2020 at 2:14
  • 1
    $\begingroup$ +1 I like this answer. It shows how shifting across the spectrum can never get green unless you filter the other frequencies out. The sky shifts from being a low-pass filter to a high-pass filter, but you'd need a band-pass filter to ever see green. At least, this is true of the light coming from the Sun. If the sky simply emitted green light (who knows if it does though), then you could easily see a green sky $\endgroup$
    – Jim
    Commented Feb 18, 2021 at 19:15

The sun is technically green because the peak of its black body spectrum is near green wavelengths. When light scatters parallel to the plane of incidence (i.e., during the day time), it is blue-shifted. When light scatters perpendicular to the plane of incidence (i.e., sunset or sunrise), it is red-shifted. The light that is not scattered but makes it effectively straight through is not shifted in wavelength. One of my astronomy professors explained that the reason the sun appears yellow to us (and not green) is largely due to the response of our eyes. Regardless, the when sunlight is scattered it tends to move away from green since most of it is already green.

(fun side note: plants used to be red, thus absorbing more green light… though I forgot now why they evolved to be green and now reflect most sunlight)

The image 1 shows the solar spectrum with the visible light portion shown in color. The image can be found here.

solar spectrum

  • 6
    $\begingroup$ The Sun is technically white. Sunlight at noon at the equator is the canonical definition of "white". $\endgroup$ Commented Sep 25, 2014 at 12:51
  • 2
    $\begingroup$ Yes, our eyes respond best to green. We are very, very good at seeing green. Yet we never see the Sun as green. With a very high quality filter, you can look at the Sun from sunrise to sunset. At sunrise, it will be rather reddish. The apparent color changes from reddish to lightly tinted yellow as the sun gains altitude, and then to a nearly pure white. There is no time we see the Sun as green. $\endgroup$ Commented Sep 25, 2014 at 13:54
  • 2
    $\begingroup$ @honeste_vivere Blue and red stars look blue and red. Really. Go to a dark location at night and look up, preferably with binoculars. But you're right, broad spectra that peak in green are perceived as white, which is why there is no blackbody temperature that looks green, whether for stars or glowing hot chunks of metal or anything else. $\endgroup$
    – user10851
    Commented Sep 25, 2014 at 15:53
  • 3
    $\begingroup$ @honeste_vivere: because there's no blackbody (or near-blackbody) spectrum that corresponds to what we perceive as and call "green". A green object, like a leaf, is reflecting a lot of quite a narrow range of light, plus a bit of others. We call this green. A blackbody peaking at the same green frequency as the leaf's reflected light spectrum is emitting much more red and blue light relative to the peak, and we perceive the resulting spectrum as "white". Basically, for the sun to appear green you'd have to cut the red out. $\endgroup$ Commented Sep 26, 2014 at 13:19
  • 4
    $\begingroup$ ... meanwhile, a star (or other hot glowing object) that appears red peaks in the IR (or maybe red. It might be that a red-peaked blackbody spectrum would always look orange rather than red, I just don't know). It has little enough green and blue light that it looks red to us. A blue star peaks in the blue or UV and has little enough red light to appear blue (or anyway distinctly blueish-white). Ultimately the reason green is different from red and blue, despite us basically have RGB sensors in our eyes (actually yellowish/green/blue by peak response), is that green is in the middle. $\endgroup$ Commented Sep 26, 2014 at 13:23