10 replies to this topic

[CPellier]

Hi all,

I finally took some spectra of Uranus in early December and this allows a good comparison of how the two planets can be seen in their differences, especially in colors.

The first graph is a comparison of their respective albedo, i.e. their capacity to scatter back through space the amount of sunlight they receive. These curves are the spectra of the planets whith the Sun spectrum being removed.

The second graph is a comparison of their respective color spectrum, i.e. their light flux as they come to us, before being filtered by our atmosphere, instrumentation, camera or eye. These second curves are the solar spectrum multiplied by the albedos of the planets. They have been re-scaled so they cross at 440 nm, where their respective albedo in blue light is equal.

Both graphs shows that while the relative brightness of the two planets can be considered as equal in blue light, despite a different slope (I find the albedo in the Johnson B photometric band to be equal for both at 0,55), the color differs more noticeably in green, red and infrared where Uranus is slightly, but definitely, more reflective than Neptune.

Uranus is going to be perceived at the eyepiece as less blue than Neptune, slightly more green, and more and more colorless to a point when using big telescopes and/or more clear skies, as the eye is turning into its photopic mode of perception (=bright light) because it then becomes more sensitive to the red emission of the planet. Digital images should be processed accordingly!

It will be interesting to see in the coming years or decades, as the bright polar region of Uranus is slowly turning towards us, if the difference increases again. Uranus is going to be white !

Attached Thumbnails

• 
• 

Edited by CPellier, 13 January 2020 - 04:52 AM.

[Tulloch]

Wow, this is really cool, I don't think I've seen the spectra from both planets put on the same set of axes before. While the two spectra are different, I'm a little surprised how similar they are (but I guess they are pretty much made from the same sort of stuff).

 

I have a couple of questions if that's ok:

 

  1. How did you remove the effects of the atmosphere in the first graph? Were these taken when the planets were at a similar elevation in the sky?

 

  2. Is it possible to convert these spectra into CIELAB colour space to give an average "colour" of each planet? Obviously that requires setting a white point and makes a number of other assumptions, but it might be interesting to see these L*a*b* values and the colours associated with them. Just a thought.

 

Thanks for posting,

 

Andrew

[CPellier]

Hi Andrew,

 

Wow, this is really cool, I don't think I've seen the spectra from both planets put on the same set of axes before. While the two spectra are different, I'm a little surprised how similar they are (but I guess they are pretty much made from the same sort of stuff).

 

I have a couple of questions if that's ok:

 

  1. How did you remove the effects of the atmosphere in the first graph? Were these taken when the planets were at a similar elevation in the sky?

 

In the first graph the type of processing is called "reflectance spectrum" ; it is obtained simply by dividing the spectrum of any solar system object (work as well for moons, asteroids etc.) by the spectrum of a solar star found at the same air mass. By this way, it automatically removes the atmospheric absorption as well as the instrumental response since both effects are equal in the two spectra. So the planets were not at same elevation, but their respective reference stars were. The intensity value of the albedo is found by more complexe calculations, but basically this how the profile is obtained :

 

 

 

 

 2. Is it possible to convert these spectra into CIELAB colour space to give an average "colour" of each planet? Obviously that requires setting a white point and makes a number of other assumptions, but it might be interesting to see these L*a*b* values and the colours associated with them. Just a thought.

 

Good question. I just don't know...

Edited by CPellier, 13 January 2020 - 08:09 AM.

[CPellier]

 I don't think I've seen the spectra from both planets put on the same set of axes before. While the two spectra are different, I'm a little surprised how similar they are (but I guess they are pretty much made from the same sort of stuff).

This work had been done originally by Erich Karkoschka of the LPL in the 90's, I'm just finding again the same results (and so this is a proof that the observations are correctly done). Yes Uranus and Neptune are very alike, they belong to the "ice giants planets", in opposition to the "jovian planets" (Saturn and Jupiter), which are different, but again similar to each other through spectroscopy. Here is a graph showing the albedo of the four giants obtained following Karkoschka's work.

Attached Thumbnails

• 

[Tulloch]

Hi Andrew,

 

In the first graph the type of processing is called "reflectance spectrum" ; it is obtained simply by dividing the spectrum of any solar system object (work as well for moons, asteroids etc.) by the spectrum of a solar star found at the same air mass. By this way, it automatically removes the atmospheric absorption as well as the instrumental response since both effects are equal in the two spectra. So the planets were not at same elevation, but their respective reference stars were. The intensity value of the albedo is found by more complexe calculations, but basically this how the profile is obtained :

 

 

Thanks Christophe, I should have known the answer to that one  .

[Tulloch]

  2. Is it possible to convert these spectra into CIELAB colour space to give an average "colour" of each planet? Obviously that requires setting a white point and makes a number of other assumptions, but it might be interesting to see these L*a*b* values and the colours associated with them. Just a thought.

 

Good question. I just don't know...

Well, after a bit of searching on the net, I came up with this paper and subsequent Excel spreadsheets which might be able to do just this. Written for students, it appears that you can enter your spectrum into the spreadsheet, and it automatically calculates the tristimulus values for the spectra, converting them to L*a*b* (and RGB). All you need is the spectrum levels in 5nm increments (which you should be able to get from your data) and voila!

 

The only issue I can foresee is you don't know the absolute reflectance values from the planet, rather I think you are measuring reflective levels. However, you might be able to estimate the absolute reflectance from your data?

 

The paper (which contains a link to the spreadsheets) is here.

https://pubs.acs.org...jchemed.7b00681

 

Actually, wouldn't the raw (uncorrected) spectra give us the average colours of the planets as they appear to us on Earth (ie with atmospheric dispersion included)?

 

Thanks again,

 

Andrew

Edited by Tulloch, 13 January 2020 - 04:25 PM.

[Tulloch]

Hi again, I hope you don't mind but I took the liberty of digitizing the data from your graphs (assuming these are absolute reflectance values) and plugging the data into the spreadsheet referenced above. After crunching the numbers, I ended up with the L*a*b* and RGB values shown in the image below. These numbers corresponded to the colours shown for the average colours of the planets' discs. What do you think?

 

FYI the data digitizing process was performed using this website: https://automeris.io/WebPlotDigitizer/

 

The colour panels were extracted from this website: https://www.nixsenso...olor-converter/

 

Andrew

Attached Thumbnails

• 

Edited by Tulloch, 14 January 2020 - 12:22 AM.

[CPellier]

That's nice. I think it should not be far from the truth. I'm going to read your previous link. Really interesting indeed.

[Tulloch]

That's nice. I think it should not be far from the truth. I'm going to read your previous link. Really interesting indeed.

Thanks, you should be able to do better with the actual data.

 

I suspect the actual colours of the planets might be just outside the gamut of the sensors. The CIE XYZ colour system xy chromaticity diagram is shown below, where x = X/(X+Y+Z) and y = Y/(X+Y+Z). 

 

If I've calculated correctly , I get the following values, which sit right on or maybe a little outside the gamut of colours for RGB. It's pretty close though.

Neptune (colour spectra): x = 0.2591, y = 0.4127

Uranus (colour spectra): x = 0.2715, y = 0.4392.

 

I'm getting well outside my area of expertise with this analysis, so if I've got this a little (or completely) wrong, I invite anyone to correct my working...

 

Andrew

Attached Thumbnails

• 

[CPellier]

Just took a look at the article and the method, I really think again that the results you have obtained are accurate. Just a remark: do you know if there is some way to introduce parameters such as the level of saturation, not only the hue ? I think that while the hue obtained from Uranus is certainly good, it is also quite more saturated than in real life. One of my theory is that the difference between the two planets is not only to be found in the albedo (hence the hue) but also in the degree of saturation of the hue, Uranus being much less saturated than Neptune.

I can send you the albedo spectra if you want to test further..

 

 

Actually, wouldn't the raw (uncorrected) spectra give us the average colours of the planets as they appear to us on Earth (ie with atmospheric dispersion included)?

 

Yes they would certainly be closer to the human vision, since the camera I'm using looks to have a reasonable match to the human vision, despite its UV and IR extensions. I can send them to you as well if you want, however there will be no intensity calibration.
Last solution (probably the best one) is to use the absolute flux color spectra (I got them) and multiply them by a self-made pseudo "human vision" filter. That could be not uneasy to perform.

[Tulloch]

Just took a look at the article and the method, I really think again that the results you have obtained are accurate. Just a remark: do you know if there is some way to introduce parameters such as the level of saturation, not only the hue ? I think that while the hue obtained from Uranus is certainly good, it is also quite more saturated than in real life. One of my theory is that the difference between the two planets is not only to be found in the albedo (hence the hue) but also in the degree of saturation of the hue, Uranus being much less saturated than Neptune.

I can send you the albedo spectra if you want to test further..

 

Yes they would certainly be closer to the human vision, since the camera I'm using looks to have a reasonable match to the human vision, despite its UV and IR extensions. I can send them to you as well if you want, however there will be no intensity calibration.
Last solution (probably the best one) is to use the absolute flux color spectra (I got them) and multiply them by a self-made pseudo "human vision" filter. That could be not uneasy to perform.

L*a*b* values are supposed to represent the range of colours that human vision is able to interpret, based on the sensitivity of the cones in our eyes. This is calculated from the XYZ matrix (there is a good explanation on page 9 of this reference). I think that the degree of saturation is just pushing the individual colour intensities up or down which results in different L*a*b* (and hence RGB) values. I would expect that the measured spectrum of a "saturated" colour should be quite different that a less saturated one, even if the hue was the same. How that is then converted into RGB is a different story, especially if the colour lies outside the gamut of sRGB.

 

I don't know much about the Star Analyzer 100, but I assume it splits the light into its wavelength components and the camera takes an image of the colour spectrum? Since the camera is not equally sensitive across the visible wavelength bands, I assume you then correct the spectrum by dividing by the camera efficiency to give you a flat wavelength response? I wouldn't have thought it wouldn't need any additional corrections to convert that spectrum into a colour that we would observe physically here on earth (which includes the atmospheric dispersion), but I think the tricky part is to scale this spectrum to get the absolute intensity levels (ie 100% = white, 0% = black) from this measurement, since we don't really know the intensity of the light that hits the planet. (However, maybe there is something I'm missing here).

 

I'm happy to try this again with your accurate data (PM me if you want to send it) and I'll put together something more rigorous in terms of colour matching.

 

Thanks again,

 

Andrew.