View previous topic | View next topic

Why is the sky blue? (Cojones)

Page 2 of 2
Goto page Previous  1, 2

11528.  Tue Nov 30, 2004 8:53 am Reply with quote

So yes, the light from our sun - which is intimately related to the star's surface temperature - is yellow.

What is interesting to me about this is that the very early work on light that was done by Newton, using prisms to split light into its different colour constituents and to recombine it, was done using sunlight as his source of light, and he certainly assumed that it was white. I wonder if anybody has replicated Newton's experiments to take account of the peak in the yellow portion, and whether in fact it makes any difference to his results.

11552.  Tue Nov 30, 2004 1:07 pm Reply with quote

I think that "What colour is the sky?" might well be the right way to put the question, especially if we make the correct answer "blue", but then add the interesting stuff as supplementary. This for two reasons: Alan might say "bronze" and get a forfeit that way (because this time we're asking what colour it is, not was), and because we ought to get a laugh just for having the gall to ask such an obvious question. Plus, like playing poker, this might encourage the panel to try calling our bluff a bit more. People like eg Clive tend to be a bit wary because they assume everything's a trap. Give them one that isn't, and that might encourage them to take more chances.

11753.  Thu Dec 02, 2004 2:24 pm Reply with quote

It's only a slight yellow peak, but Newton's theory was still sound - that light mixes together and its component frequencies can be separated by a prism.

In a related note, the universe appears to be beige, although I prefer 'misty buff'.

646784.  Wed Dec 16, 2009 10:11 am Reply with quote

JumpingJack wrote:
These strike me as remarkably feeble excuses.

They are generally mumbled off in a cursory fashion right at the end of an otherwise enormously technical explanation as to the reason why the sky is blue.

Does anyone know the correct answer?


Because there are too many ways to use silly technicalities to wiggle out of the loosely worded question Why is the sky blue? and klaxon poor Alan, kindly allow me to reformulate the question so it can be asked properly:
  • The correct question is: Why does the Earth’s daylight sky appear light blue to a human observer with normal color vision?
  • The correct answer is: Because the daylight sky’s particular spectral mix elicits a human tristimulus response which is a metameric match for light—i.e., desaturated—blue.
To a first approximation, when you see something you think is blue—or amethyst or mauve or chartreuse or golden or grey or brown or black or white—your mind arrives at that color by first measuring the response levels of your S‐cones compared with those of your M‐cones compared with those of your L‐cones.

    For any particular color, infinitely many spectral combinations exist which your mind reports as the same color. These are all metameric aliases.

    For example, we perceive monochromatic light of wavelength 575 nm as yellow. When you superimpose equal parts red light and green light, you produce a color which we also perceive as that same yellow, even though no 575 nm wavelengths were abused in the making of this picture. Similarly, although we perceive monochromatic light of 490 nm as cyan, you can also create cyan by mixing equal parts blue and green light. You cannot tell one from the other; they appear identical. That metamerism exists is not a bad thing; it’s a good thing. (It’s metameric failure, not metamerism itself, which is the bad thing.) Were it otherwise, we could not create all the colors we can on our RGB monitors and TVs or on our CMYK printers.

    Think about it: given that you have only red, blue, and green dots on your computer monitor or television, how can you get violet, whose frequency is below that of your blue dot? You cannot! You can, however, do a fairly good job of emulating it with a particular shade of purple by mixing together just the right amount of red and blue.

    Is it really violet, though? Well, sure! If it quacks like a violet, it is a violet! Or: A violet by any other name…. After all, I bet you didn’t realize my violet was merely a purpled violet, not a violeted one. It’s good enough.

    Under an HSB color model, its hue would still be the same 273° that spectral violet would produce; only the saturation would be less. That’s because spectral violet would be more saturated than its metameric nom d’écran, even if they have the same hue. Your vision system (unconsciously) notices that both red and blue are present, which cannot happen with spectral violet, so it lightens the color a little bit. You’d get no L‐cone response to spectral violet, so your vision system knows your mixed version must be impure and desaturates it for you. Get a good prism and check out the fabulously rich violet end to see what I mean. That’s one color you’ll never see on TV.
By its nature, Raleigh scattering favors frequencies at the violet end of the spectrum; the shorter, the better. Your three cone types react to this mix of a little green plus some cyan plus more blue plus even more violet in a way indistinguishable from the response a desaturated blue produces: the pastel color we call sky‐blue. Note that there exists no single spectral frequency corresponding to sky‐blue; it’s an achromatic color, however oxymoronic that may sound. You can create it by mixing white light with blue light. You can also create it with the sky’s Raleigh scattering, or with any non‐finite number of other means.

For those who’d care to read a more technical explanation of this, physicist Glenn S. Smith has a paper in the American Journal of Physics from 2005 on “Human color vision and the unsaturated blue color of the daytime sky”.

By the way, daylight skies are not all of them sky‐blue. For example, I live at a rarified altitude where the sky is at least azure and often even darker. With less atmosphere, there’s less scattering, so the perceived color is a bit purer.

I fear that I may have so oversimplified matters that folks might think color is just one or another mix of red, blue, and green light—or for daylight sky, of green, cyan, blue, and violet light. This is not so. It is not enough simply to measure the spectra an object reflects, or to describe it in RGB values or SML response levels, to say what color something “is”.

Color is fundamentally an act of perception, an illusion that exists in the human mind alone. Far more sophisticated processing occurs in the human visual system than just measuring RGB levels. This is easily demonstrated by looking at the same‐color illusion. We perceive the squares labelled A and B as two different colors despite their sharing identical RGB values. You can’t even say that the squares are the same color but we don’t see them that way. Because color is perception, the color we think we see is the color we see.

Fancy processing in the human vision system moderates the S‐, M‐, and L‐cone responses using a double‐opponent system of blue‐vs‐yellow, red‐vs‐green, and light‐vs‐dark to achieve, amongst other things, the automatic white‐balancing trick we call chromatic adaptation or color constancy, and it does this far better than any digital camera does, or under the current technology, can. Why, even the lowly goldfish is better than a camera!

By the way, this is also why we can say that sunlight is by definition white: thanks to color constancy, we automatically discard the illuminant. Even more curiously, we can do so for multiple illuminants all of disparate character within the same scene, which your camera cannot do.

Almost every color you ever see is a blend of different spectra. Very, very little that you see is truly spectrally monochromatic. It’s quite rare to see light of a single frequency, or even of a very narrow band of frequencies. Exceptions to this are nearly always light sources produced somewhat exotically, such as lasers, LEDs, and the output of sunlight split by a prism into the colors of the rainbow. Even Iris’s arc, the classical rainbow, appears a bit washed out. The rarely seen circumzenithal arc , an ice‐crystal halo, presents “purer”, more saturated colors than does a rainbow.

I hope you found this interesting. ☻


659458.  Wed Jan 20, 2010 4:26 am Reply with quote

tchrist wrote:
I hope you found this interesting. ☻

Very much so Tom, even if a little late. LEDs only giving out light of a single wavelength is excellent.

659704.  Wed Jan 20, 2010 11:29 am Reply with quote

Tom - don't know how I missed this when you posted it, but it is a superb post and everything that QI should be.

659866.  Wed Jan 20, 2010 5:11 pm Reply with quote

Jenny wrote:
Tom - don't know how I missed this when you posted it, but it is a superb post and everything that QI should be.

Thank you very much, Jenny. I confess that I was a bit disappointed when no one noticed it. I tried to provide better references than just the routine Wikipedia citations that one too often sees. Dr Smith's paper is to my knowledge the first published, quantitative work in a peer-reviewed journal about something many of us have long "just known", and the UK Atmospheric Optics site has always intrigued me.

Add to that appeals to general ignorance via apparent paradoxes like how to show violet in RGB, the conundrum of achromatic colors, and the same-color illusion, then sprinkle lightly with a technogeeky multilingual pun here and a reference to classical mythology there, and it seemed to me the QI sort of article folks hereabouts would particularly enjoy.

As for timeliness, let's just pretend I gave everyone else five years to provide the right answer. :) While I suppose the letter C is a little passé, you could apply this towards H by thinking of hue or the heavens—or heavenly hues. ☺



Page 2 of 2
Goto page Previous  1, 2

All times are GMT - 5 Hours

Display posts from previous:   

Search Search Forums

Powered by phpBB © 2001, 2002 phpBB Group