Blue Sky on Mars
posted by saurabh in Graphs, Mars, Science!, Starry-eyed |One of my favorite conspiracy theories is that NASA is deliberately false-coloring images so that Mars appears to have a red sky, to cover up the fact that it’s actually blue there. I’m not exactly sure why NASA would be doing this, but I’ll admit I’ve been disappointed by the color of the sky (and the ground) in Mars photos, so I’m on board! Let’s deconvolve:
First, for the uninitiated, our a priori expectation should be a blue sky. The sky everywhere should be blue, because the color of the sky is the product of Rayleigh scattering – which is basically to say, the atmosphere tends to scatter higher-frequency light more than it scatters lower-frequency light. If you look up at any random part of the sky that isn’t the sun, the light you’re seeing is light that has scattered off some part of the atmosphere in that direction; in space, that light would have gone straight to its target, and you would see no light coming from that direction. Since the atmosphere scatters violet light more than red light, this, combined with our human visual system’s spectral bias (weak in the blue end of the spectrum), results in our seeing the sky as blue. This simple phenomenon should apply equally well to all atmospheres.
So it’s a bit glum when we’re told that Mars has a red sky – red because it’s full of thick, choking dust. Well, poop. But maybe it’s not so! Maybe NASA is attempting to pull the dust over our eyes, and Mars is really true-blue. Fortunately, we can verify for ourselves. NASA puts up “raw” JPEGs of all the data the rovers send back. The relevant images are the “PANCAM” ones, which apply a series of fairly narrow bandpass filters before the CCD capture (that is, each filter captures an image of the scene at a specific wavelength of light).
The NASA “true color” images are generated by the PANCAM group at Cornell; their methodology is quite rigorous. The naive method (employed by most conspiracy theorists, notably Keith Laney) is simply to use the three filters that closest approximate the human visual peaks – 600, 530 and 480 nm, for Red, Green and Blue – and slap them together with Ye Olde Photoshop (or in my case, Perl) to make a full-color image. This produces very satisfying images. Check it out!
However, this method has a flaw: the human visual system is additive, meaning that single-wavelengths don’t give the whole picture – each color opsin in your eye is stimulated by the whole spectrum, meaning that what you see as “red” might actually be an amalagam of two individually non-red peaks.*
Unfortunately, we don’t have data from the whole spectrum available to us – we only have eight wavelengths, six of them in the visual range. PANCAM takes this data and fits a third-order polynomial to it to generate an approximation of the true spectrum. This spectral data is then converted into the XYZ color space (a standard color space) by convolving it with the XYZ standard observer functions that (more or less) define the primary colors of that space.† Those XYZ values are then mapped to the familiar sRGB space and slapped together to produce a “true color” image. The result, side by side with the “naive” method:
Egad! There’s a world of difference there. So who’s got it right? Hard to tell… unfortunately PANCAM doesn’t have any pictures of the color calibration target that sits on top of the rovers posted in their collection of true color images, so it’s difficult to be sure. However, the average spectra values in the data for the above image for the sky (blue line) vs. the ground (red line) look like so:
This seems to suggest there’s something off about the PANCAM results – the sky should be white-tending-to-blue, and the ground should be red. Of course, it’s possible our visual system is so heavily red-skewed that we’d still see the spectrum on the left as reddish, but I’m inclined to disbelieve it could see the deep, dusty red shown in the PANCAM image. It seems more likely the reconstruction method is flawed somewhere. Furthermore, the naive method should be pretty good at telling us the color of the sky, since the sky color is composed of all wavelengths in varying intensities (which may not be true of rocks on the ground, meaning those colors are more likely to be wrong in the naive method). I’m staying aboard this conspiracy ship!
* Leading me to wonder if we will, one day, encounter a fully spectrographic visual system – one that sees spectra instead of colors. You’d recognize the chemical composition of everything!
† Sloth prevents me from recapitulating their method – all the necessary data is linked from here, if you feel inclined to do it for yourself.
Loading ...
Blargh. A little late I realized that I used the wrong filter for the blue peak – I should have used the 430 nm one instead of the 480 nm one. The difference is not much – the sky actually gets a little bluer.
that actually is kind of cool about mars.
oddd how NASA is dishing out false crap.
The controversies which have continued to hover around Mars in scientific arena have been paralleled by less constrained speculation, rumor and conspiracy mongering with its own following. A historical background, examples, and speculation of the background sociological influences is given here:
http://www.donaldedavis.com/2008%20new/CLRMARS.html
I’m sorry, but I use simple common sense. When I see pictures of Earth’s sky during the day, under non-cloudy conditions, its blue. When I look at the halo around the planet from space, even in lunar images viewing it, its blue around the edges. When I look at the sky of aircraft high in the atmosphere, space is black, and the atmosphere is blue. The sky matches the color from space.
When I look at images of Venus, and images of the soviet’s surface photos, the skies match the color of Venus from space. The planet is yellowy-white, the skies are yellowy-white.
I’m sorry, but I’ve just checked every other true color image from the Hubble Space Telescope, and the halo around Mars when its closer to the sun scatters light identically to the atmosphere on Earth. I’ve seen the illustrations of the “Red Planet” with it’s red skies, and the whole place being just red red red, and I’m sorry, but the dirt is multiple colors, and during duststorms, the planet can turn red, or orange, and even green sometimes, but on cloudless days its blue, because it projects a blue scatter of light from space. The closer Mars is to the Sun in its orbit, the bluer the skies.
Correct the Hubble imagers if they’re simply so wrong in their color calibrations, because according to some, every time their imaging mars close to the Sun they’re screwing the colors up.
Get real.
correction:
“but the dirt is multiple colors, and during duststorms, the planet can turn red, or orange”
should read:
“but the dirt is multiple colors, and during duststorms, the planet’s atmosphere can turn red, or orange”
Your assumption about the spectral composition of the sky on Mars is wrong. Additionally your claim that there’s no image using the full filter set that shows the color calibration target is also wrong. Take for example the Spirit Pancam photos from sol 50. There’s a full visible filter set of images showing a closeup of the calibration target (http://mars5.jpl.nasa.gov/gallery/all/spirit_p050.html).
Furthermore it is not necessary to use the full filter set to reconstruct the sky spectrum to a high degree.
There’s a paper, “Chromaticity of the Martian sky as observed by the Mars Exploration Rover Pancam instruments”, that describes in detail how the color reconstruction process works and how the spectrum of the Martian sky is composed/reconstructed.
It also explains how it is possible to reconstruct the full sky spectrum to a high degree (with errors as low as 1%) even when using only the three filters L4, L5 and L6.
The Martian sky spectrum has its peak at around 610 nm wavelength, falling off strongly towards both the long and short wavelength ends of the visible spectrum, quite different from the sky spectra on Earth. The Pancam results (from the Horizon and Sky Flat Surveys) confirm the findings of the old Viking mission and not so old Pathfinder mission. Besides, the Luminance of the Martian daytime sky is very low compared to Earth’s sky at daytime.
Calculations including dust and Rayleigh scattering show that the average color of the Martian sky is very close to “dark yellowish brown” according to the NBS-ISCC color naming convention.
If the dust in the Martian atmosphere would settle completely (which has not been observed yet) then the Martian sky would look bluish-black to black at daytime (using NBS-ISCC nomenclature). A complete dust-free atmosphere would have an optical depth (tau) near 0.0 (optical depth is a measure of transparency, and is defined as the negative logarithm of the scattered or absorbed portion of light).
The optical depths observed by Spirit and Opportunity range between 1.7 and 0.2, changing with seasonal dust storms etc.)
Below tau=0.2 the color of the sky starts to change considerably because less light is diffused by dust particles that brighten the sky. So the sky gets very dark and finally, if there’s almost no dust anymore the hue of the sky changes to a very dark blue.
The Luminance of the dust-free sky is as low as 0.005, compared to 1.0 which stands for a Lambertian diffusor, meaning a material reflecting all light perfectly. The average Luminance of the average Martian sky (tau = 1.0) is only 0.17, still very low compared to Earth.
That’s not very surprising because the Martian atmosphere is very! thin compared to Earth, what people seem to forget all the time.
Again, you can calculate this with the same methods that are successfully used to calculate the atmospheric appearance of Earth’s atmosphere (using Rayleigh scattering to calculate the light scattering on molecules).
Sorry for posting twice, but this might be interesting for some of you.
I’ve programmed the colorimetric routines to test the algorithms used to reconstruct the color appearance of materials. I can input any material and illuminant spectra and get the corresponding color.
The rendered object resembles the well-known MacBeth Color Checker, a calibration target widely used in photography.
Here are “real life” photographs:
http://www.ae5d.com/macbeth.html
Here’s a full spectral rendering of the chart. The illuminant is the standard daylight at 6500K color temperature, resembling average noon daylight:
http://s47.photobucket.com/albums/f188/albdo-co/?action=view¤t=MacBeth_D65.jpg
And this is an example of choosing the wrong white point. In this case the illumination is D55 (like afternoon daylight) but with a white point resembling the D65 standard. This is a typical problem all photographers know:
http://s47.photobucket.com/albums/f188/albdo-co/?action=view¤t=MacBeth_D50_non-adapted.jpg
And this is how the color chart would look like through the L7 filter of the Pancam, assuming D65 daylight again. As the glass interference filters are very narrow the resulting “color” can’t be rendered correctly on a computer monitor – the color is outside the gamut of the monitor. Nevertheless, this gives a good impression of the general appearance of the colors:
http://s47.photobucket.com/albums/f188/albdo-co/?action=view¤t=MacBeth_L7.jpg
Hmm, intriguing. The thinness of the Martian atmosphere is something I hadn’t considered fully; makes sense that Rayleigh scattering would play a limited role in that case. I’m still not sure I understand Don Davis’s argument about why the average spectra I show above are wrong. Also, I’m not sure why you say you can successfully reconstruct the whole spectrum from three peaks to within 1% error – this might be true in general of the Martian sky, but it’s certainly not true in general of images. A bandpass filter is going to miss sharp peaks, period, yes? How can that always give you an accurate reconstruction?
Yes, but natural spectra are smooth in most cases.
Sharp peaks are created either by emission processes involving excited electrons and such (fluorescent/luminescent materials, Cerenkov radiation, lasers) or interference/diffraction and similar phenomena (CDs, some car paints, some insects).
The spectral composition of the incident sunlight of mars is known. This spectrum is also smooth.
If the background of a spectrum is smooth then even sharp peaks don’t have much impact if they don’t contain very much energy compared to the rest of the spectrum.
The reconstruction of the colors of the colors of photographs is possible because we know the spectral composition of the Martian sky (the color that this article is dealing with). If you know the illumination by the sky you can use this knowledge to “fill the gaps” of sparse filter sets by reverse engineering because the materials (which are known) have smooth spectra and are lit by a known light source (direct illumination by the sun, indirect illumination by the atmosphere).
Nobody said precise reconstruction is always possible!
But as long as you know the material or the illuminant you can reconstruct a good approximation.
In the case there are enough data and facts known to find the appropriate colors! Finding the colors does not necessarily involve a full spectral calculation. The band-pass filter peaks only need to be at the right places (in case of the Mars photos these are the filters L4, L5, L6). Then the whole works similar to what the human vision system does. Human vision collects photons only with three kinds of color receptors, each of them acts as some kind of wide band-pass filter, collecting photons over a wide range of wavelengths. Color perception is a 3-dimensional phenomenon, so during the “calculation” of color the infinite-dimensionality of spectra is reduced to only three dimensions – If you choose Lab, RGB or HSV as description system is irrelevant (all have three dimensions), as they are mathematically equally valid. One can be converted into every other.
The paper below contains example photographs showing how the reconstruction of three-filter photos leads to much better results if the Martian sky spectrum is included in the calculation:
http://marswatch.astro.cornell.edu/Bell_etal_SkyColor_06.pdf
The Cornell university has several interesting PDFs about the Pancam:
http://pancam.astro.cornell.edu/pancam_instrument/publications.html
i was beginning to come to feel i may well be the only guy whom seriously considered this, at the very least at the moment i recognize i’m not idiotic
let me make sure to learn more about a few various articles immediately after i recieve a tad of coffee in me, it is very hard to read without having my caffeine, adios for the time being 
I spent a year researching ancient valley networks on Mars at university and I have wondered about the sky on Mars too. The atmosphere is thin, but not that thin. The surface pressure in reasonably low-lying areas is 1% of the Earth’s at sea level (and can be as high as 1.5% at the lowest lying places, such as the floor of Hellas); but the density of the atmosphere is nearer twice that (at 1.2-2.5% of the Earth’s at sea level) because of the lower gravity. The atmosphere is also deep (considering the pressure – again due to the lower gravity). So could we say that the Rayleigh scattering would be nearer to, say, 4% as much as on the Earth at sea level?
Just three points
1) Some of the more recent images released by the JPL show a blue sky
2) The Hubble photos show, along with plenty of cloud, white haze over large areas; shouldnt that lead a whitish-hazy sky? The opaquity of the atmosphere should be related to water-ice or CO2 ice haze, not just dust. The Martian atmosphere may be thin but it is never far from saturation.
3) The Hubble photos, and also amateur photos of Mars, show a blue colouring near the limb – wouldnt that also indicate a blue sky?
I really wonder – I’d be surprised if the Martian sky was the same bright blue as the Earth’s (although hazy skies could lead to that effect) but I am not comfortable with the official photos either, as if the surface is under a permanent dust storm (if there was that much smog, how can we see the surface?). The excess pink/ bright orange cant be right (the actual space probes are totally mis-coloured in official photos!), so I am left in doubt.
I dont mean to sound daft but why cant they send a cheap digital camera that would take normal photos without needing to “calibrate” the images from multi-million dollar cameras?
Does anyone know what the real colour of the sky is?
It occurs to me that there’s somewhat of a problem with the idea that the atmosphere is (a) so thin that it does not produce significant Rayleigh scattering and (b) still manages to hold enough dust continuously that it can maintain a red sky. Shouldn’t a thinner atmosphere be less turbid and therefore less likely to kick up dust and keep it from falling back to the ground? Sigh – if only I had taken statistical mechanics in college!