Why is there a lack of consistency in portrayals of the visible light spectrum?

[Drakkith]

The issue of the blue sky is different to these two. It appears to be over stimulation of the S cells by UV, because they have sensitivity right through to 400nm*, but is related to the first in that if it is 'only green' then the eye 'allows itself' to be easily fooled because it just doesn't 'do green' very well. It'll take other cues preferentially, if there are any, and allow them to over-ride a perception of green.

*(viz. if the eye is still 50% sensitive to 420nm than at 450nm, so if there is a stack of 400-420nm energy bundled into light that is strong in all wavelengths, then it'll tend to perceive that additional 'violet' as shading the white light towards strong blue. Caveat; again, I will repeat that this is a contribution to the effect of 'blue sky' but there is, agreed, a general shift towards blue due to scattering. It is the ratio of those two effects on the final perception of blue I do not believe is known or well understood.)

Again, UV light is filtered out by other areas of the eye and is not seen by cone cells. People that have had the lens in their eye removed or replaced with a lens that does not block UV light have [STRIKE]reported that it appears white, not blue.[/STRIKE]

EDIT: Correction UV light does not appear to be white per this article: http://www.guardian.co.uk/science/2002/may/30/medicalscience.research However it is filtered out by the lens of the eye normally.

 

[cmb]

This is incorrect. A rod cell is desensitized after it has been exposed to a large number of photons. It does not contribute to color vision at all...White light looks white because it is stimulating all our cone cells at the same time. The combined input is filtered and interpreted by the brain and is perceived as white.

I'll happily take the correction on your advisement that this is so, but whether the green perception is significantly rods or M cones (that are not well-discriminating to the L cones), [edit:] and also however much filtering the eye does actually do on the UV content that hits the retina*. Whatever the nature of these fine details, ultimately it is self-evident (from the perceived spectrum of black-body radiators, and absence of green therein) we see green mainly by inference from the content of the rest of the spectrum, rather than direct perception of the peak of radiated emissions. Thus, we cannot directly perceive the peak of 'blue-green wavelengths' in the spectrum of the sky.

*(clearly, there are levels of UV emission that will make it to the retina, else there would be no need for UV safety glasses!)

 

[Drakkith]

I'll happily take the correction on your advisement that this is so, but whether the green perception is significantly rods or M cones (that are not well-discriminating to the L cones), [edit:] and also however much filtering the eye does actually do on the UV content that hits the retina*. Whatever the nature of these fine details, ultimately it is self-evident (from the perceived spectrum of black-body radiators, and absence of green therein) we see green mainly by inference from the content of the rest of the spectrum, rather than direct perception of the peak of radiated emissions. Thus, we cannot directly perceive the peak of 'blue-green wavelengths' in the spectrum of the sky.

*(clearly, there are levels of UV emission that will make it to the retina, else there would be no need for UV safety glasses!)

Ever heard of snow blindness? It is caused by sunburn to the cornea and lens (and other areas) of the eye by UV radiation.

And yes, the issue of color vision is very complex and isn't 100% understood. We don't see the green-blue peak as green blue because the M cone that corresponds to green is much further away at its peak than the S cone is.

 

[cmb]

Ever heard of snow blindness? It is caused by sunburn to the cornea and lens

Yeah, OK, I realized after I posted I didn't cover that 'purpose'.

But you should take a look at UV retinopathy, used to be a common cause of macular holes amongst welders using electric arcs before they figured out why. And there is also a particular condition that eclipse-observing astronomers tend to suffer, because although the Sun is obscured, the corona isn't.

see;

http://eclipse.gsfc.nasa.gov/SEhelp/safety2.html

The tissues in the eye transmit a substantial part of the radiation between 380 and 1400 nm to the light-sensitive retina at the back of the eye.

 

[cmb]

And yes, the issue of color vision is very complex and isn't 100% understood. We don't see the green-blue peak as green blue because the M cone that corresponds to green is much further away at its peak than the S cone is.

OK, so I think we're agreed that there is scattering of solar UV and blue light but, nonetheless, that the sky's spectrum peaks in the 'blue-green' region of the spectrum, although for complex reasons not entirely understood we perceive that as blue.

Is that the conclusion?

 

[D H]

I'll overlay a colour spectrum over the 'spectrum of blue sky' given in the page you've just linked to.

There are (at least) three things wrong with this:

1. Where in the world did you get this spectrum? Light with a wavelength of 500 nanometers is much closer to blue (435 to 490 nm) than it is to green (520 to 570 nm). Part of your problem is this lousy spectrum. The region between 490 and 520 (which coincides nicely with the peak of this spectrogram) is cyan. That region shows up as green in the spectrum you chose for your overlay.
2. The spectrum of the blue sky itself is not quite right. That spectrogram was made by shooting through four panes of glass. That is going to attenuate the high frequency part of the spectrum, and the author of the spectrogram did note that this is the case.
3. The peak of the spectogram of the blue sky without going through four panes of glass has a saturation of about 210^o. That saturation value is the tertiary color azure, halfway between cyan and blue. Some other names for colors with a saturation near 210^o: Dodger blue, cornflower blue, UN blue, and of course sky blue.

 

[cmb]

There are (at least) three things wrong with this:

1. Where in the world did you get this spectrum?

1. 
   
   The blue sky spectrum from the link the other poster made. The spectrum from wiki, under 'spectrum', for a K4 star. It matches every other spectrum I've seen so I saw no reason to question it. The spectrum of the blue sky is what I expected to see, also.
   
   Can you provide alternatives?

 

[DaveC426913]

Ever heard of snow blindness? It is caused by sunburn to the cornea and lens (and other areas) of the eye by UV radiation.

Reaaaaaally.
I'd always wondered what the actual mechanism was.

I assumed it was an optical overload thing.

 

[Sci-Fi]

Meh -.- it doesn't reflect the color of the sea But the sky appears blue due to REFRACTION of sunlight by water(moisture) molecules in the air.The sun light first goes through the molecules as their is a density difference between the water molecules and the air,it undergoes refraction and then it gets reflected internally(within the molecule) and refracts again and comes out of the water molecule.Its like refraction through prism,that creates a spectrum of 7 lights,but prism is angled to form 7 colored lights and due to preferable wavelength and shape blue light appears the most.
EDIT:Some people might replace the word 'molecule' with 'bubble' but the concept is the same.

 

[D H]

It matches every other spectrum I've seen so I saw no reason to question it.

Seriously? About the only consistent thing I've noticed with portrayals of the spectrum of visible light is an utter lack of consistency, particularly between 450 nm and 600 nm (blue to yellow), and particularly so those available on the internet. Just a smattering:

[PLAIN]http://www.antonine-education.co.uk/physics_gcse/Unit_1/Topic_5/em_spectrum.jpg

[URL]http://www.yorku.ca/eye/spectrum.gif[/URL]

[URL]http://withfriendship.com/images/b/9817/visible-spectrum.gif[/URL]

[URL]http://hosting.soonet.ca/eliris/remotesensing/LectureImages/visiblespectrum.gif[/URL]

[PLAIN]http://i231.photobucket.com/albums/ee127/jcrowmag/visible-a.jpg

[URL]http://3.bp.blogspot.com/_pF0P8-ezLug/S_JpwLquzoI/AAAAAAAAAFc/zOjSIBOwt9M/s1600/Our+perception+of+light.jpg[/URL]

There is no consistency here.

The spectrum of the blue sky is what I expected to see, also.

Compare that spectrum to a published one, one that is not taken through four panes of glass that attenuates the blue, and one that is taken with a calibrated device.

J. J. Michalsky et. al., "Shortwave, Clear-Sky Diffuse Irradiance in the 350 to 1050 nm Range: Comparison of Models with RSS Measurements at the Southern Great Plains ARM Site in September/October 2001". http://www.arm.gov/publications/proceedings/conf13/extended_abs/michalsky-jj.pdf

The graph on page 8 shows the peak is somewhere between 420 and 460 nm: indigo to blue. So why do we see a sky the color of the sky as azure? Simple: Our eyes don't see energy flux. We see photon flux. That shifts the color we see toward the red from the peak in the spectral irradiance plot.

Can you provide alternatives?

I just did with regard to the spectrum of the blue sky. Now for some colors. Note that green has a hue of 120 degrees. The sky is not green. It's blue with a tinge of green. So let's look at some colors, starting with some cyans (490 to 520 nm) and moving toward blue (440 to 490 nm).

MediumAquamarine. Hue = 160 degrees[/color]
DarkCyan. Hue = 180 degrees[/color]
DeepSkyBlue. Hue = 195 degrees[/color]
DodgerBlue. Hue = 210 degrees[/color]
Royal Blue. Hue = 225 degrees[/color]
Blue. Hue = 240 degrees[/color]

 

[cmb]

I really haven't seen such a variation of colours in spectra before, as you have found! But I have no contest that blue is <490nm and green is >520nm, and in between is blue-green. I've only pinned up here the work of others that one might readily find on the 'net.

In regards the paper you have found, this looks more authoritative for sure. But what is 'direct irradiance spectra' and 'diffuse irradiance spectra'? I guess the audience of that paper knows the difference, but 'fraid I don't.

The thing I still don't get, you see, is that if you take a sheet of white paper and go sit at the bottom of a well that only looks up to blue sky, with a spectrum as you point to on p8, does the paper still look white and not blue, if the only light coming down the well shaft to the bottom is light that comes from the sky, peaking around 400~450nm? Either it is blue, or it looks blue. If the former, then why don't white things look blue in that light, when shaded from all other direct reflections from the Sun, if the sky looks blue due to that light?

 

[cmb]

OK, I found a further paper; http://www.patarnott.com/atms749/pdf/blueSkyHumanResponse.pdf .

I'm happy with that and accept I was wrong to describe the sky as 'white'.

...the spectral irradiance of daytime skylight was shown to be a metameric match to a mixture of a monochromatic blue light plus white light (unsaturated blue light).

Still, I'd tend to think my basic point, that the blue observed is a constrution of colour perception rather than an actually observed 'peak wavelength', remains.

 

[D H]

In regards the paper you have found, this looks more authoritative for sure. But what is 'direct irradiance spectra' and 'diffuse irradiance spectra'? I guess the audience of that paper knows the difference, but 'fraid I don't.

Imagine looking through a paper towel roll at the Sun (don't do this at home!). The spectrum of the light you see shortly before you go blind is the direct irradiance spectrum. Specifically, the direct irradiance is that light that comes out from from a cone pointed directly at the Sun and having a radius of about 3 degrees. The diffuse sky irradiance is all the light coming from the sky except for that light coming directly from the Sun. In other words, the diffuse sky irradiance is the light from the clear blue sky (or the light filtering through a bunch of clouds, depending on the condition of the sky).

The thing I still don't get, you see, is that if you take a sheet of white paper and go sit at the bottom of a well that only looks up to blue sky, with a spectrum as you point to on p8, does the paper still look white and not blue, if the only light coming down the well shaft to the bottom is light that comes from the sky, peaking around 400~450nm? Either it is blue, or it looks blue. If the former, then why don't white things look blue in that light, when shaded from all other direct reflections from the Sun, if the sky looks blue due to that light?

First off, I doubt that you have ever conducted this experiment. If you had, you would know that the only color you see is the tiny but brilliant patch of blue you see when you look straight up the well. (This is why you cannot see stars from the bottom of a well during the day.) Everything else is black and white. There just isn't enough light coming down the well for your cones to be active.

Here's an alternate experiment: Walk in and around your house with that white sheet of paper in hand. Look at that paper in a nicely sunlit area outside and it will look white. Look at it indoors in a room with soft white lights and it still looks white. It looks white in a room with bluish industrial florescent lights, and in a bathroom with the heat lamp turned on, and in a dressing room with the makeup lights on.

It looks white because in your mind you know that that piece of paper is white, so your mind automagically compensates the signals coming from your eyes that say that the paper is a bit off-white in some regard.

 

[D H]

Still, I'd tend to think my basic point, that the blue observed is a constrution of colour perception rather than an actually observed 'peak wavelength', remains.

The peak wavelength in a spectrograph is a high frequency blue, maybe even indigo. Our eyes don't see energy flux. They see photon flux. The peak in the photon flux is a lower frequency blue, azure. That is more or less what we see.

 

[shiva999]

thanks

 

[sophiecentaur]

Seriously? About the only consistent thing I've noticed with portrayals of the spectrum of visible light is an utter lack of consistency, particularly between 450 nm and 600 nm (blue to yellow), and particularly so those available on the internet. Just a smattering:Compare that spectrum to a published one, one that is not taken through four panes of glass that attenuates the blue, and one that is taken with a calibrated device.

J. J. Michalsky et. al., "Shortwave, Clear-Sky Diffuse Irradiance in the 350 to 1050 nm Range: Comparison of Models with RSS Measurements at the Southern Great Plains ARM Site in September/October 2001". http://www.arm.gov/publications/proceedings/conf13/extended_abs/michalsky-jj.pdf

The graph on page 8 shows the peak is somewhere between 420 and 460 nm: indigo to blue. So why do we see a sky the color of the sky as azure? Simple: Our eyes don't see energy flux. We see photon flux. That shifts the color we see toward the red from the peak in the spectral irradiance plot.I just did with regard to the spectrum of the blue sky. Now for some colors. Note that green has a hue of 120 degrees. The sky is not green. It's blue with a tinge of green. So let's look at some colors, starting with some cyans (490 to 520 nm) and moving toward blue (440 to 490 nm).

MediumAquamarine. Hue = 160 degrees[/color]
DarkCyan. Hue = 180 degrees[/color]
DeepSkyBlue. Hue = 195 degrees[/color]
DodgerBlue. Hue = 210 degrees[/color]
Royal Blue. Hue = 225 degrees[/color]
Blue. Hue = 240 degrees[/color]

You guys are all doing it again. How can you possibly be demonstrating colours on a TV or printed display and try to kid each other that those are the "colours" of spectral / monochromatic sources? They are all desaturated matches to those spectral lines and made up of a mix of two or three non-spectral and quite broadband primaries.
A monochromatic source of light has a colour, of course, but the only colours you will ever see in real life (excluding those that are produced from a spectrometer or a gas discharge tube in a darkened room) will not be of a single wavelength.
Just do yourselves a favour and look at the RGB values of the colours in a scene that you would describe as 'colourful'. Very few of them are actually very saturated ((255,0,0) or (255,255,0) etc

Also, the 'colour' is more than just the 'angle' on the colour chart - it's the distance from the white point, too. That is needed in order to describe the saturation. Why else do we refer to 'colour space' (implying more than one dimension)?

All those curves with colour bands behind them are just 'hints' as to what the actual monochromatic sources will look like (near matches). Moreover, everyone's appreciation of just how good a match they are will be different.
Although wavelength has (matches to) a colour, you cannot assign a wavelength to a colour.

Three colour printing is even worse for the same sort of reason. when you want one really bright or 'special' colour, you have to produce it with a purpose-made pigment (spot colour).