What gives more detailed images (higher resolution) – an iPhone or the COBE satellite camera used to propel Big Bang into the leading Cosmology concept?
News Flash: (June 17, 2012) This article was just voted one of the best science blog articles of the past year at 3-Quarks Daily. Thank you for your votes.
Well lets find out by trying this puzzle: See if you can figure out what fairly well-known astronomical phenomenon this is a photograph of :
Here’s a hint. This first photograph of the mystery phenomenon is presented with the same angular resolution as the camera on the COBE satellite: Cosmic Background Explorer – Differential Microwave Radiometer (COBE-DMR).
(“Angular Resolution” simply means how much detail is in an image. A camera with more megapixels has a greater angular resolution; more resolution = more information.)
The COBE satellite’s data / images were described as “echoes of Big Bang” and used to first claim cosmic microwave radiation is purely from “background,” not from stars or galaxies or space dust or gas. COBE’s “background” radiation map was used to eliminate the “ Steady-State” cosmology – Big Bang’s popular competitor at the time.
Can’t see anything? Try this sharper version of the same image with a resolution identical to the best NASA technology (WMAP) for cosmic microwave radiation.
Can you guess what it is yet? or is it still too obscure? Well, when you are ready – lets take a look at it with the resolution of an ordinary camera. You’ve got it – all three are the same image of our moon during the December 2010 Lunar Solstice Eclipse.
The COBE camera’s angular resolution was horrible; only about 7 degrees per pixel, while our Moon appears about a half a degree wide, meaning each gigantic COBE pixel is 14 times wider than our Moon as we see it from Earth.
To represent the resolution of the COBE camera the first eclipse image had to be resized to a single pixel for the entire image (and even that was slightly too generous because the original Eclipse image was not quite 7 degrees wide).
A single COBE pixel gathered all the “light” (microwave radiation) from the whole 49 square degree (7 degrees times 7 degrees) field of view and averaged it to give you only a single data point.
“Statistics are like bikinis. What they reveal is suggestive, but what they conceal is vital.” ~Aaron Levenstein.
We’re going to take a look at how low resolution microwave images conceal a lot of foreground radiation. This is similar to how the low resolution versions of the eclipse images (above) take the bright light of the moon and add it to the black of the surrounding sky beyond (and all the stars and galaxies in the field) and “average” it all to give a result as a single pixel.
That means a single data point, covering the whole 49 square degree view of sky which is many times the size of the moon at 0.2 square degrees. Of course since the image is only one pixel – whatever the brightness is – it can only be one color and one brightness value.
Although the first image appears simply black, it is not. Pure black would be zero (0) out of 100 percent on the brightness scale. Because of the light from our moon and a few stars are mixed in, that single pixel measures brighter than black at about one percent of full brightness: Red=0.8 percent, Green=0.2% and Blue=1.0%. Our whole sky has 41,253 square degrees. This means you can make a whole sky brightness map with only 841 total COBE pixels (29 x 29). That’s less than a kilobyte of data to show the entire night sky.
The second image is our lunar eclipse image resized to show the best resolution of NASA’s newest microwave camera on the WMAP (Wilkinson Microwave Anisotropy Probe) satellite. The WMAP‘s best camera resolution was about 0.22 degrees, meaning each pixel is about half (44 percent) as wide as our Moon as we see it from Earth. (Three of the other four WMAP cameras had angular resolution worse than one pixel per Moon diameter.)
To represent the resolution of the sharpest WMAP camera the image was resized so that the moon is 2.4 pixels wide. While we can now distinguish the moon from the background because the pixels are smaller than the Moon’s diameter – individual stars that we can see in the digital camera version are still invisible to WMAP. That’s because compared to WMAP’s resolution stars are too small (smaller than a pixel) and thus are “averaged” away; essentially mixed together with lots of “black” beyond, losing their individualness.
This means that even though WMAP could distinguish (about 8 pixels worth of) Moon microwaves from the background it simply did not have the angular resolution needed to distinguish microwaves from any individual star or galaxy (except our own Milky Way galaxy, which surrounds us, and a few neighboring galaxies such as Andromeda which is about 2.5 degrees wide).
Now you can tease out a bit better data by cleverly using lots of overlapping images, but that’s the sharpest angular resolution for NASA’s Cosmic Microwave Radiation technology as of May 2012.
Angular resolution will improve later this year or in early 2013 when the European Space Agency’s PLANCK Satellite releases whole sky microwave radiation data with its best angular resolution almost three times as sharp as WMAP (5 arc-minutes vs WMAP’s best at 13.5 arc-minutes). The only cosmic microwave background experiment with resolution better than Planck’s is “Acbar” a ground based telescope that achieves four (4) arcminute resolution.
By comparison the 5-megapixel iPhone-4 camera has a resolution some 226 times sharper than WMAP, NASA’s best microwave radiation camera and and about 87 times more detail than Planck. (Wouldn’t it be amusing to have an iPhone 4 described as a telescope?)
Update: Since I had a couple of very bright folks ask about distinguishing between angular resolution and sensitivity I’ve added this: Sensitivity and resolution are related but independent phenomena. A telescope or a camera’s angular resolution is measured in pixels per degree of sky. It is mostly determined by its sensor and pixel size and its lens focal length. Telephoto lenses give higher angular resolution, as do more pixels. A telescope or a camera’s sensitivity (in this case) is how faint a signal each pixel can detect. It is determined by the quality of the individual pixels/sensels. As a generality a large pixel captures more light / radiation; thus can pick up fainter signals.
COBE and WMAP microwave cameras were designed almost purely for high sensitivity with huge “pixels” – not for angular resolution. (COBE’s DMR camera only had two (2) pixels. That is not a misprint. COBE cameras did not have mega-pixels or even kilo-pixels – just two pixels total. Amazingly even the very first digital camera built in 1975 had thousands of times more pixels (at 100×100) than COBE.) While it is an understandable design choice, it appears that ignoring angular resolution leaves a bit of a problem.
One way to find the angular resolution of a camera is to divide Pixel size in microns by Focal length in millimeters. Then multiply that times 206.3 which gives you “arc seconds per pixel.” At the iPhone4’s 35 mm camera focal-length equivalent of 29.4 millimeters (3.85 mm actual) we get 94 arcseconds per pixel. This means the moon is about 19 pixels wide in an iPhone4 photo. NASA has already done this for COBE (7 degrees = 12.7 moons / COBE pixel or 0.0786 pixels / moon diameter) and WMAP (~2.4 pixels / moon diameter) so we don’t need to calculate much.
Perhaps because of our familiarity with spectacular Hubble images and affordable digital cameras with high resolution we take extremely high resolution images of our Universe for granted. However, for Microwaves, high resolution whole sky images simply do not yet exist. As we’ve seen microwave telescope’s stunningly poor (almost meaningless) angular or detail resolution is dramatically worse than common visible light cameras.
Quantifying this: Cosmic microwave radiation is mapped in degrees and arc minutes as opposed to Hubble’s cameras which are hundreds to thousands of times more detailed – sharper than a twentieth of an arc-second. Now lets take a look at how this poor angular resolution affects the idea of cosmic microwave radiation as “Background.”
Cosmic Microwave Radiation – How much is really Background?
Well, since both foreground and background cosmic microwaves have the same frequency and spectrum you can’t separate the data that way.(1) That leaves separating them by location. That takes quite a bit of work, because you can’t just go out in your backyard, put out a microwave camera and get a map of background radiation. (First we generally need to get the camera into orbit because the microwaves we want, at about 1 millimeter wavelength, faint to begin with, are mostly absorbed above the clouds.)
In short, the primary method is to start with whole sky microwave radiation maps (which were never available until the 1983 Russian spacecraft “Relikt” – six years before COBE launched) then subtract the known foreground microwave radiation. Whatever is left, they call “background” — Cosmic Microwave Background Radiation.
Definitions: Whole sky microwave radiation is easy to discern (it is all non-earth based microwave radiation). Foreground microwave radiation is defined as the microwave radiation coming from anything made of normal matter (solids, liquids, gas and plasma) which includes all the galaxies, stars, gas and dust.
Even microwaves from the most distant galaxy, gas cloud or star are foreground radiation. The biggest contributor of foreground microwave radiation, by far, is our own Milky Way galaxy which you can see as the broad red band in this whole sky map.
Background microwave radiation is not clearly defined, other than — all microwave signals left over when you subtract the foreground. While that doesn’t seem like an fully adequate definition, we’ll use that for this article.
Well, there are a few problems with this foreground subtraction method that are not widely appreciated outside the microwave astrophysics research community.
Microwave Foreground Clearly Exists, No Calculations Needed
To start with some context, we know with certainty foreground microwave radiation exists because it shares virtually identical locations and density (flux) patterns as our Milky Way in visible light and every other wavelength including radio waves, gamma rays and x-rays.
Further, (in theory) you can just put out a microwave camera (in orbit) and get a map of foreground radiation. No calculations are needed at all, just take sky pictures in the microwave band and map them. While foreground microwave radiation clearly exists, an indisputable inventory of all its locations does not – partly due to the low resolution of microwave telescopes and cameras, partly because there is no unambiguous definition of a “background,” and partly because different teams use different “models” and methods to subtract foreground.
Here’s one of the most recent (2010) whole sky microwave radiation images (from Planck) that illuminates our Milky Way’s “calculation-free” Foreground microwave radiation.
Everything Blueish-White is Foreground from our Milky Way.
On the other hand —
Microwave Background Must be Calculated, Does not Clearly Exist
However, by contrast, we do not have anywhere near the same assurance that there is any background microwave radiation because you simply cannot take a picture of it, you have to calculate it. Those calculations are massively complex and based upon often hidden assumptions.
To subtract out the foreground requires choosing one of several elaborate computer models with highly complex calculations (not to mention an armload of assumptions) of what is foreground and what isn’t. Then it takes extraordinary amounts of effort and computer time to arrive at a background map (COBE software took 5 months to process only a year’s data).
In short, Cosmic Microwave Foreground is a photograph of microwave radiation, while Microwave “Background” is really only a statistic derived from a complex computer model with a lot of large assumptions.
Poor Angular Resolution is one of Background CMR’s biggest problems.
COBE and WMAP cameras do some things marvelously (like measure exquisitely faint radiation signals), but that’s for another article. However as described above, Cosmic Microwave Radiation “cameras” have remarkably poor resolution; some 200 times worse resolution than an iPhone4 and thousands of times worse than a good digital camera. If we want to subtract the radiation of foreground stars and galaxies (and gas clouds and dust) to see what remains as “background” microwave radiation — we need to know very precisely where all the foreground sources are, and how much and what kinds of microwave radiation they emit.
However, when we are stuck with poor angular resolution we can’t really identify all the distinct sources of foreground radiation or its edges so we can separate it from the background. Think of the Hubble Ultra-deep field images that are overflowing with tens of thousands of galaxies we’ve never seen before – all unarguably spewing microwaves; just like our home Milky Way (at least 30,000 to as many as several million galaxies in every square degree of sky). The only galaxies we have a fairly good idea of how much, where, and what kind of Microwave radiation is emitted is our own Milky Way and to a much smaller degree, our neighbor Andromeda.
This is precisely because our best cosmic microwave radiation instruments (Planck, WMAP and COBE) have such poor resolution that we simply cannot really measure microwave radiation from even our closest neighbor galaxies such as Andromeda – without having the radiation from at least thousands (if not millions) of other stars and galaxies added in. We only recently found out our home Milky Way’s microwave halo glow extends many times broader than the map area of visible stars. The key point is – when your angular (detail) resolution is so poor that you can’t distinguish stars and galaxies from background, even if your sensor is hyper sensitive — sensitivity is irrelevant.
This is not unlike inviting an attractive stranger for dinner and they refuse to come unless they can bring their parents . . . and friends, and in-laws . . . and outlaws and essentially everyone they’ve every met. Then you can’t see your date, you can only hear them (an ear’s angular resolution at about 15 degrees is only about twice as poor as COBE’s camera). Imagine trying to pick out what your date (microwave background) is saying from out of all the thousands of other conversations (galaxies) of your date’s entourage.
The key idea is: if your angular resolution is poor enough – everything looks like background, even if every single photon comes from specific sources.
Is Microwave Background Evaporating?
In fact, the amount of sky real estate considered “Background” has been diminishing in jumps since the first whole sky microwave map from Russia’s RELIKT-1 surprised researchers by showing huge areas of microwaves coming from our own Milky Way. Newer technology’s better angular resolution cameras show a growing percentage of the microwave radiation sky area that was previously considered “background” is really just from our own galaxy.
This year ESA’s Planck satellite just discovered a huge foreground microwave radiation “haze” from our own Milky Way galaxy that covers at least two-thirds of the sky, and perhaps as much as 75 percent.
That (at least) doubles the area of the sky that is known to have foreground microwave radiation. So it seems a rapidly growing amount of what used to be called “background” microwave radiation – just isn’t. Here are three older whole sky COBE microwave radiation maps where you can see the bands of our Milky Way Galaxy radiation at three different frequencies/wavelengths. (Note how much more “left over” sky area there is in the COBE images than in the newer, higher resolution Planck maps.)
Now compare this new (2010) whole sky microwave radiation image from Planck that includes our Milky Way’s microwave radiation.
Everything blueish white is foreground from our Milky Way. The red areas at the top and bottom are what’s left after subtracting out our Milky Way (but not any distant galaxies, dust or gas clouds). Do you see how our Milky Way’s microwave (foreground) area is so much bigger on the newer and higher resolution map?
Your Thoughts Respectfully Requested
What do you think of this poor angular resolution; and how does it affect your ideas on what is foreground and background?
Do you think that the WMAP (or COBE) cameras can provide undebatable evidence that cosmic microwave radiation is not from our own galaxy or any galaxies in the background? How?
As space-based microwave cameras get higher resolution how do you think the “background” radiation maps will change? I look forward to your thoughts.
PS Just for fun (oops, I meant for comparison) – here’s probably the highest quality (visible light) map of our Milky Way galaxy in a whole sky map. It was painstakingly made by Axel Mellinger I happily urge you visit his website to take a detailed look at our own galaxy.
1. A individual camera “pixel” is termed a “sensel” by camera engineers.
1. The WOMBAT Challenge: A “Hounds and Hares” Exercise for Cosmology, Gawiser et al. 1998 “Most foreground separation techniques rely on assuming that the frequency spectra of the components is constant across the sky . . .”
4. “Removing point sources from CMB maps,” Tegmark, and de Oliveira-Costa (1998) http://xxx.lanl.gov/abs/astro-ph/9802123
5. “Max Tegmark’s CMB data analysis center. http://space.mit.edu/home/tegmark/cmb/pipeline.html
6. “Galactic Foreground Sources (and why I’m ignoring them),” Benjamin Recchie http://www.background.uchicago.edu/~whu/Courses/Ast280/Ben/…/Astro.ppt
7. “COBE-DMR Maps,” http://lambda.gsfc.nasa.gov/product/cobe/dmr_image.cfm
8. “Galactic microwave emission at degree angular scales,” Angelica de Oliveira-Costa, et al (1997) http://iopscience.iop.org/1538-4357/482/1/L17
9. “Computing challenges of the Cosmic Microwave Background,” Bond, Crittenden, Jaffe and Knox. http://www.cita.utoronto.ca/~crittend/CiSE/CiSE_final.pdf
10. “First-Year Wilkinson Microwave Anisotropy Probe (WMAP) observations: Foreground Emission” http://lambda.gsfc.nasa.gov/product/map/dr1/pub_papers/firstyear/foregrounds/57710.web.pdf
11. “CMB Foregrounds” Angelica de Oliveira-Costa’s website https://www.cfa.harvard.edu/~adeolive/foreground.html
12. “Big Bang Leftovers in the Microwave: Cosmology with the Cosmic Microwave Background Radiation,” Dissertation, Eric Joseph Gawiser http://arxiv.org/abs/astro-ph/0005365
13. “Removing point sources from CMB maps,” Max Tegmark & Angelica de Oliveira-Costa, http://space.mit.edu/home/tegmark/cleaning.html
14. “UC Berkeley team challenges cosmologists to analyze simulation of cosmic microwave background” http://www.berkeley.edu/news/media/releases/99legacy/3-16-1999.html
15. “Contribution of Extragalactic Infrared Sources to CMB Foreground Anisotropy,”Gaweiser and Smoot , http://arxiv.org/pdf/astro-ph/9603121.pdf
16. “Microwave Emission at High Galactic Latitudes In The Four-Year DMR Sky Maps,” Kogut et al. 1996, http://arxiv.org/abs/astro-ph/9601060
17. “Non-cosmological Signal Contributions to the COBE-DMR Four-Year Sky Maps,” A.J. Banday, K.M. Gorski, C.L. Bennett, G. Hinshaw, A. Kogut, G.F. Smoot, 1996 http://arxiv.org/pdf/astro-ph/9601064v1.pdf
18. Statistical analysis of undetected point sources in cosmic microwave background maps, F. Argüeso et al. (2006) http://arxiv.org/abs/astro-ph/0609348