Giant airplane-mounted telescope sees first light!

By Phil Plait | May 28, 2010 3:09 pm

Very cool news: the flying infrared observatory, SOFIA (Stratospheric Observatory for Infrared Astronomy) — which has been in the works for many years — has seen first light. What’s remarkable about this observatory is that it’s mounted in a hole in the side of a 747!

Don’t believe me? Check. This. Out:


[Click any of the images here to embiggen.]

See that ginormous square hole in the back of the plane? That’s where the telescope sits, looking out at the sky. Why put it in a plane? Because SOFIA looks in the mid to far infrared, and observations like that are impossible from the ground. Water vapor in the air absorbs the kind of infrared light seen by SOFIA, but by the time you go up to about 10 km (35,000 feet) you’re above 99% of the Earth’s water vapor. That little bit left does still absorb the light, but a telescope at that altitude receives about 80% of the IR an orbiting observatory would.

So, amazingly, they cut a big hole in the side of a modified 747 and stuck a telescope in it. A big one: SOFIA sports a 2.5 meter (8 foot) mirror, which is bigger than Hubble’s!

Here’s a sample of what it saw:


That is, of course, Jupiter. On the left is an image in visible light, and on the right the SOFIA image. It’s a composite of three infrared colors: 5.4 (colored blue in the image), 24 (green), and 37 microns (red). Your eye can see wavelengths only as long as about 0.7 microns, so these represent wavelengths well outside what we can see. See the stripe on the left that’s reddish? In the infrared it’s brighter, because the gas in that belt is transparent to infrared light and we can see deeper into Jupiter’s atmosphere where it’s warmer. Note that the other belt is still missing, in the visible and infrared. Whatever is blocking the light from that belt is opaque to both our eyes and SOFIA’s.

sofia_m82SOFIA also took a look at the nucleus of the nearby weird galaxy M82, which is undergoing a burst of star formation, and is lousy with thick dust that blocks visible light. SOFIA peers through that dust, revealing the star factories hard at work in the center of that galaxy.

According to preliminary reports, SOFIA is performing perfectly, and getting great observations at a far, far cheaper price tag than putting something in orbit. We still need telescopes in orbit to do other work — some IR is still absorbed even at that height — but SOFIA will make terrific achievements. It also has a robust educational arm, which includes taking as many as 200 teachers per year up on observation flights! If you’re an educator and interested in this, contact the team at SOFIA. You may just qualify to get the airplane ride of a lifetime.

Image credits: Anthony Wesley, N. A. Sharp/NOAO/AURA/NSF, NASA/Jim Ross

CATEGORIZED UNDER: Astronomy, Cool stuff, Pretty pictures
MORE ABOUT: infrared, Jupiter, M82, SOFIA

Comments (57)

  1. I’m going to get a teaching job so I can take a ride in that baby!

  2. Tom

    Always amazed when things like this are flown. I have enough trouble just keeping my backyard observatory scope autoguiding without drifting! :-)

  3. Jess Tauber

    Now THAT’S a ‘grab-n’-go’ scope….


  4. Nuke3d

    Not that I have a clue of what telescope diameter is to deliver what kind of resolution for Jupiter, but it seems to me that the resolution is not very high for a mirror that is bigger than hubbles (or where’s the visible light image from?).

    Other than that, awesome stuff :) Go NASA!

  5. Michael Crowe

    Next logical step: plane-mounted death ray! Muahahahaa!!

  6. Johannes

    The resolution achievable by a telescope is determined by the ratio of the mirror diameter and the wavelength at which one observes. Mid infrared is a way larger wavelength than optical wavelengths…

  7. HvP

    As I understand it, in the infra-red range the wavelengths are so long that it requires a larger telescope to gain the same resolution as you used to seeing in the visible range. Keep in mind that the wavelength is up to 37 times longer – that leads to a corresponding reduction in resolution.

  8. @4. Nuke3D
    It’s been a while since my college astronomy class, but if I remember right as wavelength increases, the diameter of the telescope needed to obtain the same resolution increases as well. Plus you have to figure that since it’s in a plane, you’re going to lose some clarity due to turbulence and motion in general. At least, that’s the best I can figure; someone who knows more can fill in the holes.

  9. Nuke3d

    Ok, thanks for the answers. Makes sense :)

  10. Bigfoot

    Odd that telescopes took to space before they learned to fly.

  11. elephant

    Nuke3d: in IR range, images taken with an instrument of the same diameter/aperture have less resolution, this goes linear with wavelength. Assume a vis wavelength of 530 nm (0.53 micrometers(mu m)), then the resolution at 5.4 mu m goes down to 1/10, and at 37 down to 1/70 .

    So a 2.5 meter mirror at this wavelength can only have an angular resolution compareable to a C-11 (11in mirror) (5.4 mu m) or a medium binocular telescope with a diameter of 3.6 cm /ca. 1.5 in (at the long wavelength). So the image of Jupiter probably is taken at the max. resolution of the 2.5m instrument in the plane.

    At those wavelengths, one simply needs so big an aperture, in this case a mirror.

  12. TW

    >>Odd that telescopes took to space before they learned to fly.<<

    They have been flying for decades…

  13. Gary Ansorge

    Looks like a good use for a dirigible; stable platform, lots of load capacity and they’re VTOL. Granted, current designs only “fly” at a few thousand feet but there’s really no reason they can’t be built to fly at 35,ooo and I expect they’d be a whole lot cheaper in the long run than a 747.

    I just have a thing for oldies but goodies,,,

    GAry 7

  14. Does anybody know how long the typical exposure will be and how they stabilize the telescope in all the shaking and wobbling of the plane?

  15. Number 6

    What do they do about vibration? Even a 747, stable as a table by most standards, generates a lot of vibration as it flies along. That’s gotta play hell with keeping the telescope pointed precisely…

  16. MadScientist

    It only took over 20 years. :) Big expensive equipment on a big expensive platform that costs a fortune to maintain – obtaining the funding is quite an achievement.

    I was wondering how they’d mount the telescope. I’m a bit surprised that they’re staring out the side though since they get the maximum atmospheric path when they do that (well, not quite, they don’t have to point to the horizon). I wonder if other considerations prevented them from looking toward the zenith.

    @Number 6: Oh, there’s a whole bag of tricks to use and people have been flying vibration-sensitive instrumentation for decades. You’re right though – it’s quite a challenge. What is typical for large instruments is a combination of several vibration dampeners. If your pointing is very important you may even consider a fast-acting gimbal tied into a 3 axis gyroscopic sensor; fortunately for this application, translation is not really an issue unless you’re looking at a close object like the moon. These newfangled fiber-optic gyroscopes are boring (though incredibly reliable and very tough) – I enjoyed taking the case off the old type (even just an attitude gyro), spinning it up and wagging the gyro around to show people how the top maintained its orientation.

    @Nuke3d: The effective mirror size is important for the resolution in this case since they’re operating channels in the long wavelength side of the mid-infrared. The 37micron channel for example would have a resolution similar to a visible band telescope with a 43mm diameter lens (or mirror) – similar to the “Gallileoscope”. I haven’t looked at Sofia’s design though to see what is limiting the resolution; it could very well be the detector pixel size in this case.

  17. Torbjörn Larsson, OM

    @ #6:

    Vibraton [sic] isolation system: 12 springs/dampers around the hydrostatic bearing in longitudinal and tangential directions
    Primary mirror (PM): 2.7 m diameter, 2.5 m effective aperture, lightweighted ZERODUR structure on 18-point whiffle-tree support, PM ratio f/1.28, aluminum coated
    Image stability (goal): 0.2 arcsec rms for focal plane tracking

    [SOFIA website]

    They also chop the light, which may be used for phase locked loop systems, which can be used to recover signal in noise (say, over time or perhaps over pixels). But I don’t know if they do any of that.

  18. olderwithmoreinsurance

    Looks as though SOFIA is well on it’s way to being a great observatory and a worthy successor to the KAO. It is not, “far, far cheaper than an orbiting observatory” however. I think its price tag is around the $650 M mark, give or take a bit (it’s that price tag and cost over-runs that got it to within a hairsbreadth of being cancelled by NASA). That makes it roughly 25% more expensive than the very successful Spitzer mission and about 1/4 the cost of the combined Herschel and Planck observatories. As the SOFIA E/PO manager Dana Backman (known him for longer than I like to remember) likes to say: “it’s the space observatory that comes home every morning”. (oops, looks as though I mis-remembered it’s budget, a new source says NASA plans on about $3 billion over it’s 20 year lifetime)

  19. Wow that is the second largest stratospheric observatory for infrared astronomy I’ve seen today!

  20. Brian Too

    Simple question: Is the hole in the plane open?

    I mean, there appears to be a door that swings up and over the fuselage. Ok, so they can open and close the viewing door. However if it’s open to the atmosphere, it must cause terrific wind buffeting to the telescope and that’s bad. If you put in some type of glass or plexiglass shield, you rid yourself of the turbulence but maybe you’re cutting down on the sensitivity of the instrument?

  21. #14 >>”Does anybody know how long the typical exposure will be and how they stabilize the telescope in all the shaking and wobbling of the plane?”<<

    Ever since I first heard about SOFIA I've been dying to know. Anyone have an answer to this?

  22. MadScientist

    @Brian #19: Yes, it will be open. You just won’t get a large enough window for the 5-37 micron band – assuming there are any suitable window materials covering that band – my reference books are in my office so I can’t check. At a cruise altitude of 16km, you only have 10% of the atmospheric pressure, so the wind may not be as bad as you imagine – just don’t let anything protrude from the fuselage.

  23. JB of Brisbane

    @20 Brian Too – my guess is there is some sort of transparent window over the opening, which would need to be as clear and free of warping as possible. This then poses the question, is the compartment with the telescope pressurised or not? Inserting a huge window in the side of an airliner would require massive structural strengthening around the opening if normal cabin pressure was used, to stop the window blowing out, but not so if that compartment (or indeed the whole cabin) was not pressurised.

  24. Mike

    The hole is open to the slipstream, and the entire aircraft behind the mounting bulkhead is vented to the outside. There is a cavity door that is closed prior to landing, and a movable aperture that tracks the telescope, to minimize the hole size. Almost all materials are opaque to IR; the structural issues are minor compared to making a hole in the side of a cylinder (in the absence of stiffening, this makes it far weaker to torsional failure — SOFIA is heavily reinforced around the cavity to prevent that). Making a diamond window that big would fund NASA for some time….

    It’s not really a square hole. It’s “D” shaped, but the movable aperture is painted black, so it’s difficult to see. It’s big enough to park my pickup truck in.

    Zenith makes for some significant troubles. There are a myriad of hydraulic lines, structural supports, and cables in the interior of a 747. Low elevation makes more of the sky available (very rapidly), but at the cost of more water vapor; this can be traded off to a point with higher altitude. SOFIA was designed to observe the galactic center from California — at an elevation at transit barely above 20 deg in summer, and getting much more than 40 deg range would require a bigger (and weaker) hole. There may also have been clearance issues with the open cavity door to the primary mirror, were the door to be made bigger.

    For stabilization, the telescope is mounted on a spherical bearing, and is gyroscopically stabilized (much like a satellite) with image tracking to correct for drift. The whole assembly is mounted on a seismic isolation system (springs and air dampers). Much of the “fun” is keeping the dang aircraft out of the way of the telescope. It’s wedged in there really tightly….

  25. Teshi

    “It also has a robust educational arm, which includes taking as many as 200 teachers per year up on observation flights! If you’re an educator and interested in this, contact the team at SOFIA. ”

    Just graduated today! A match made in (almost) heaven?

  26. George Martin

    SOFIA is not the first airborne infrared telescope. Before it, there was the Kuiper Airborne Observatory. That was a 36 inch telescope in a C141. Its lifetime, according to wikipedia, was from 1974 to 1995. I guess I remember the Kuiper because I know someone who flew an observing run on the aircraft. He had to shave his beard so that he could wear an oxygen mask.


  27. Mike

    Before KAO, there was also the Lear Jet Observatory, and the Convair (Galileo). Both had much shorter time-at-altitude and much smaller telescopes.

  28. MadScientist

    @Larsson #17: The chopping I would imagine is for the long wavelength channels (it is not necessary for the 5 micron channel). Bolometers are typical detectors for that band and chopping between the scene and a blackbody with fixed known temperature makes it much easier to amplify the signal. The thermal emissions of the mirrors will also matter so radiometric calibration of the signal is challenging, but even if you didn’t attempt absolute radiometric calibrations you can still get relative signal strengths for your object of interest and that would be good enough for many applications. Without chopping you will need far more stringent tolerances on the performance of your electronics and you would also need a digitizer with a very large dynamic range. So basically, without chopping you probably don’t even get a useful signal.

  29. Messier Tidy Upper

    A jumbo jet with a hole cut in the side of it for a giant telescope?

    No way! No way! .. Wait .. yes way! Wow. [Jaw drop.] 8)

    That SOFIA is a truly remarkable peice of engineering and such a cool idea. :-)

    However, I wonder why not use a *helicopter* as an airborne observatory since it can hover and thus perhaps enable longer exposures on more stable platform – or as (13.) Gary Ansorge suggested a dirigible or zeppelin? Could it be weight issue?

    I’m pretty sure balloons have been used for transporting telescopes too haven’t they?

    @19. Maxwell Smart Says:

    Wow that is the second largest stratospheric observatory for infrared astronomy I’ve seen today!

    Okay, I’ll bite, what would be the *first* largest stratospheric IR observatory you’ve seen today?

  30. Jeffersonian

    Pretty cool, but my mind also goes toward the beauty below.
    That’s the Kaweah Peaks Group of the Great Western Divide Range of the High Sierra Chain in California (reaching 13,800′). Directly under the forward engine is Red Kaweah with Black Kaweah under the trailing engine. Up and left-edge are peaks around Mineral King. All administrated by Sequoia National Park.

  31. @Michael Crowe, #5:

    Next logical step: plane-mounted death ray! Muahahahaa!!

    They already have that. It’s called ALTB, or Airborne Laser Test Bed.

  32. jcm

    “It also has a robust educational arm, which includes taking as many as 200 teachers per year up on observation flights! If you’re an educator and interested in this, contact the team at SOFIA. You may just qualify to get the airplane ride of a lifetime.”

    If only I were an educator. That being said, I think it would also be a great idea to give students a ride and inspire them about science.

  33. MadScientist

    @Messier #29: some reasons why a helicopter is not used instead: (1) the top altitude is *much* lower (2) the vibrations are awful, (3) cargo capacity is much lower, (4) the main rotor severely limits where you can point, (5) the rotor wash may interfere, (6) the generators can’t provide as much power. I’m sure there are other reasons, but a helicopter is simply not comparable. A blimp would have a fairly low ceiling as well. Now if your payload is under ~1 ton, you can use a balloon (and they will go much higher than a 747) however, launching a balloon has its own problems. Getting a 747 into the air is much quicker and more reliable than getting a balloon up, and if it’s fully fuelled and has a relatively small payload it can stay in the air for a very long time and fly a path so that the object of interest can be viewed for much longer than you could with a balloon.

    Ooh, OK, the mid-IR detector (also referred to as ‘far-IR’, which is OK since different folks have a different definition of where ‘far’ starts) is a Si:As detector:

  34. Richard Woods

    For a closer-up in-flight photo, clearly showing the mirror through the open door, see the reprint PDF of “SOFIA’s smooth ride shakes up astronomy” at

    A speed of 250 mph is mentioned, but also, “We will test at all the speeds the plane can fly and all the altitudes planned for the mission’”.

  35. WOW!! I bet something like this would be really effective at peering through my neighbors’ windows.

  36. MadScientist

    @Lugosi #34: Actually it wouldn’t – the currently installed instruments can’t see through glass but just in case they left the windows open, many of the channels can’t even see more than a few meters through the atmosphere at ground level. However, with open windows and the 5 micron channel, then if the aircraft can execute a sharp enough turn it will be able to look in.

  37. @ MadScientist: Boy, you sure know how to spoil a great fantasy.

  38. MadScientist

    @Lugosi: Yeah, my boss tells me the same whenever I tell him how much it would cost to build an instrument or how much time it will take, or how the modifications he wants will ensure that the instrument will not work. So I’ve had over 2 decades of practice as the Official Party Pooper.

  39. #14 Robert,
    I had the same question. I went to the SOFIA web-site and found this answer to the question:

  40. Mike


    250 knots (not MPH) is a bit misleading. The way speeds work on aircraft is a so-called “indicated” airspeed. In a nutshell, speed is measured using a pitot tube — essentially a manometer — that measures the dynamic pressure against a fluid, pointed directly forward. This has an obvious dependence on air density, but that is ignored; since all the aerodynamic behaviors of the aircraft have the same dependence to first order (this means the never-exceed speed lives in indicated, not true, airspeed, for instance). Indicaetd = true airspeed at sea level.

    Virtually all civilian aircraft fly at no more 250 KIAS below 10,000 feet, and jets tend to fly that rather precisely because they can, as that is the ICAO (and FAA) speed limit. In the stratosphere, SOFIA flies around Mach 0.85 (the envelope in the “official” Boeing tables is 0.82-0.88), which is around twice as fast — over 500 MPH true airspeed (485 KTAS, I believe indicated is around 280 KIAS, but SOFIA flies Mach numbers at cruise altitudes).

    The aircraft has been tested at very low airspeed with the cavity open — this is very important for safety; if the door jams, it has to be possible to land safely, and touchdown speed for a 747 with extended flaps is as low as 130 KIAS.

  41. WJM

    Next logical step: plane-mounted death ray! Muahahahaa!!

    = = =

    I just bought one at a garage sale.

  42. bruce

    How about a Ginormous Balloon platform array that stays aloft with helium and solar power? Oh no, wait astronomers don’t know how to fly balloons (see balloon crash in austrailia), i forgot. never mind. Better a bunch of chem trails.

  43. Richard Woods

    @Mike #40:

    Yes, “MPH”.

    Please check the source next time. I understand knots and pitot tubes perfectly well. I wrote that “250 mph is mentioned”, rather than citing it as an accurate speed. I quoted “250 mph” exactly as given in the second sentence of the PDF I linked:

    “Yet NASA’s new SOFIA telescope recently flew in an airplane at 250 mph with doors wide open.”

    That was from an _aerospace_ forum touting itself as “The World’s Forum for Aerospace Leadership”. See (note: it’s a .org, not a .com)

    Your mph/knots quarrel is with them, not me.

  44. JB of Brisbane

    Okay, it’s an open aperture (spelling?) – I stand corrected

    @#29 re #19 – It’s a Get Smart in-joke (notice #19′s name?).

  45. Astrofiend

    42. bruce Says:
    May 29th, 2010 at 2:52 pm

    Have you done the math on just how big a balloon setup you’d need to loft the weight of a 2.5m telescope, plus all additional instrumentation and and auxiliary systems?

    Seriously, the scientists and engineers who come up with this stuff don’t just piss about and decide to ‘just go with something’ when they are in the planning and design stages. Things are done for very bloody good reasons. If it were possible and practical to have this scope sitting on a ginormous balloon platform using naught but Helium and solar power to keep it doing what it needs to do, then I’m pretty sure that it would have been done. There have been plenty of balloon borne scopes before – it’s not like it’s some esoteric option that they missed thinking of.

  46. MadScientist

    @Bruce #42: The bigger the balloon, the more failures. Meteorological balloons carrying very small payloads like pressure/temperature/dewpoint sondes (and slightly larger ones with ozone sondes) have a very high success rate – one failure in thousands of launches. Big balloons like the one whose launch failed in W. Australia are much rarer launches and they have a much poorer success rate by a few orders of magnitude. Aircraft are a far more reliable platform and can do many things a balloon can’t do (though admittedly a balloon can get higher than most aircraft). SOFIA carries multiple instruments (and more are planned for the future) – the cost of teh instrumentation and the time required to build test and calibrate it is just far too much to take chances with a balloon – that is, assuming a balloon large enough even exists for that size payload and its powerhouse.

  47. Matthew Price

    I have the privilege to work with Dr. Luke Keller in the Ithaca College Physics Department. Dr. Keller is one of the leads on the data analysis of SOFIA first light data.
    His blog is at

    There may be information that you would like.

  48. SeminoleAV

    @Madscientist–Reminds me of the ALOTS (Airborne Lightweight Optical Tracking System? ) aircraft (Boeing 707) that was flown out of PAFB in the early 1960s. Had to take the cargo door off and then mount a separate package with a special oval window and all of the tracking optics. To avoid deformation of the “window”, a large number of pistons were used around the perimeter to hold the “window” in place.

  49. Murff

    This is sweet. I got to work on the Kuiper Flying Observatory in the mid 90′s (changing tires!!). I also got to look around in side. The Kuiper had the scope mounted in front of the wings, just aft of the entry door, on a C-141A.

    I’d love to get a look inside this one.

  50. Brian Too

    Correct me if I’m wrong, but aren’t a lot of of the big stratospheric balloons simply abandoned at altitude? They cut the payload loose and parachute it to the ground. You’d not want to try that with a large telescope.

    I also seem to recall that the balloon materials become very brittle way up high due to the cold. It’s commonly -50 to -70° C. Many plastics (as in, balloon envelope materials) will simply shatter if stressed. Again, lots of payload risk there.

  51. MadScientist

    @Brian Too #51: Typically the payload isn’t released on its own – you let it fly until the balloon bursts. I don’t know about those long-endurance balloons though. The large balloons typically get to 30 or 40km altitude. Even the small meteorological balloons, which are latex, frequently get up to about 32 or slightly higher – if you’re not so lucky it might burst as low as 18km or even lower, but 28km is more typical – and that’s the sort of minimum altitude you want if you send up an ozone sonde. As for the weight size of the payload, the USAF and Naval Air Force habitually drop very heavy items on parachute – it’s really not much of an issue. One instrument which has flown many times, for example, is the MIPAS-B (at ~16 launches it’s a very old balloon veteran); the balloons used typically take a payload of 400-800kg (depending what instruments are put onto the platform) and heavier payloads have been put up before.

    @SeminoleAV: Hehehe – I like the strange arrangements people come up with. Image quality is not a big requirement in many of my applications so I can usually get away with warped windows. I’ll be working on an airborne imager soon though and it will have moderately large windows – I’m already getting nightmares about the design work.

  52. #14 >>”Does anybody know how long the typical exposure will be and how they stabilize the telescope in all the shaking and wobbling of the plane?”<<

    Here are some representative numbers: In the 24 micron filter total integration times were 5 seconds and 30 seconds for Jupiter and M82, respectively. Note that in the thermal infrared the detector arrays (Back-Illuminated Blocked Impurity Band hybrid arrays) will typically saturate in a fraction of a second on the sky background so these data were taken at a frame rate of about 70 Hz (i.e. faster than video). Then we add up the individual frames to get the total integration time. To remove the background we "chop"the secondary mirror: the secondary literally wobbles back and forth (at 11 Hz for these observations) allowing the camera to alternately image the sky and the object. Then we subtract the sky.

  53. Markle

    This has been a very long time coming. I remember hearing about this when touring the Kuiper Observatory C-141 back in ’79 or ’80. This was back when it was Moffett Field NAS and the guards carried M16s and serious expressions. Since, the Germans were brought on board, the Berlin Wall was torn down and we’re starting the second decade of a new century. The guards carry radios and they’ll smile and talk to you now.

    The Kuiper required parkas, gloves and an oxygen mask. SOFIA’s got a pressure bulkhead between the mirror and the optical instruments so everybody can work in shirtsleeves.

    For those who want to know 250kts IAS @ 41000ft @ -40C is about 540kts TAS, a plausible cruise speed for SOFIA. 250kts TAS is the stall speed of a clean config 747-400 at ~20,000ft

  54. Jeffrey Steinberg

    Not everyday you see a 747-SP (shorter, longer range). Most are out of service.

  55. Doug

    Has anyone else noticed that at 2.5 m, the mirror is the same size as the 100-in Hooker Telescope at Mount Wilson?! The Hooker was the largest telescope in the world from 1917 to 1948!! Egad!!

  56. c0ld3l3m3nt

    what a waste of money..


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!


See More

Collapse bottom bar

Login to your Account

E-mail address:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »