Adaptive Optics, Or: Laser Eye Surgery For Telescopes.

If you think astronomers are too mature to make PEW PEW sound effects while they're doing this, you would be sadly misinformed.

Most publicity photo sets of the really big telescopes and observatories will invariably include one like the above, featuring one of the telescopes shooting a bloody great laser beam into the sky. It makes for a very pretty and very striking image, but it’s also something that I know confuses a lot of people as they can’t really see why a telescope would need a laser attached to it unless astronomers are harbouring some hitherto undisclosed universal megalomaniac quality that makes them want to turn their observatory into the Death Star. Lasers are not something that fit into the popular conception of an observatory, and the reason for this is that they’re a fairly recent innovation: sticking lasers onto your telescope only really became a thing back in the mid-nineties with the advent of something called adaptive optics.

I’m slightly too young to remember the kerfuffle over the Hubble Space Telescope when it was launched back in 1990, but by all accounts it was a Pretty Big Deal indeed. Hubble wasn’t the first space telescope but it was the largest, and it was – unusually – designed to observe visible wavelengths of light. The subsequent discovery that the HST’s mirror was flawed and the three-year effort to devise a solution1 significantly dented the HST’s – and NASA’s – prestige, but once they had it fixed it began to produce startlingly sharp and clear images that vastly surpassed anything that ground-based telescopes of the time could achieve. This was because of the telescope’s location; the atmosphere causes interference as incoming light bounces and scatters off of air molecules and makes ground-based images rather noisy, so situating the Hubble outside of the atmosphere enabled it to avoid these unpleasant effects and produce images of unsurpassed clarity.

Hubble before and after the servicing mission. "Whoops" doesn't even begin to sum it up.

Ground-based telescopes go to great lengths to try to minimise atmospheric distortion and light pollution, which is why they are invariably situated in some inconceivably remote spot at a very high altitude (usually on top of a mountain): the higher up they are, the less atmosphere they have to peer through and the less distorted their images will be. But you can only go so far, and one of the reasons the Hubble was launched was because ground-based telescopes had reached the limit of their usefulness given the technology available at the time. In order to advance astronomy further putting a telescope in space seemed like the only way to go.

However, nursing a space mission from initial concept to eventual launch (especially one as high-profile and long-term as the Hubble) takes a very, very long time. The proposal that would eventually produce the HST started life as something called the Large Space Telescope back in the late 1960s; it was slated for launch in 1979 but a series of delays including the Challenger shuttle disaster pushed the launch date all the way forward to 1990, over two decades since the original proposal. In those two decades ground-based telescopes hadn’t quite caught up, but they were getting close.  In particular improved computer technology and the advent of mass-produced charge-coupled devices (CCDs)  were making processing digital images on the fly a hell of a lot easier, and unlike Hubble — which launched with reel-to-reel tape recorders as its storage medium, this being cutting edge technology at the time Hubble was proposed – astronomers could just go out and stick this stuff onto their ground-based telescopes,  giving them an efficient and easy upgrade.

Eventually technology progressed to the point where it become possible for computers to process atmospheric distortions and respond to them in real-time, adapting the shape and inclination of a mirror inside the telescope to compensate for those distortions and produce something closer to what we’d see if the atmosphere wasn’t in the way. In essence adaptive optics act as a pair of glasses for the telescope  situated in the optical path in front of the telescope’s main mirror and cancelling out the noise and blurriness introduced by the atmosphere – except because atmospheric conditions are constantly changing, the optics inside the telescope also have to constantly change and adapt on the fly to continue to compensate for atmospheric noise. This requires very fast processing of the image on the order of microseconds so that the mirror can be adapted to compensate for the atmospheric distortions now, which is why the advances in technology had to happen before adaptive optics became possible.


Of course you can’t correct for atmospheric conditions if you don’t know what those atmospheric conditions are, and this is where the laser comes in. In order for adaptive optics to work you need some form of reference to tell you what stuff should look like if the atmosphere weren’t there. This can be done by finding a particularly bright guide star with a known brightness and comparing it to what the telescope is actually seeing, but this method can be inconvenient and/or impractical since guide stars do not exist in all regions of the sky. If the thing you’re trying to look at is located too far away from a guide star, your only alternative is to create an artificial one by beaming a laser into the sky. The laser is set to a certain wavelength that will excite certain types of atoms in the air, causing them to glow. Since astronomers know exactly what the glowing atoms should look like – they’re the ones who calibrated the laser, after all – they can use the difference between that and what they’re actually seeing to calibrate their telescope.

So ground-based telescopes have clawed their way back to some level of parity with space-based ones, but they do still have some drawbacks. The adaptive optics technique restricts their field of view to the area immediately around the object they’re trying to image, and the adaptive optics system itself isn’t perfect and can’t produce an image quite as good as what the Hubble is capable of. In particular a project like the Hubble Deep Field would be impossible with a ground-based telescope, since the light from distant galaxies is so faint that only the Hubble has a realistic chance of detecting it. On the other hand the Hubble cost, like, $2.5 billion dollars just to construct and launch, whereas conventional observatories are cheaper and easily-accessible thanks to being located on Earth. It’s swings and roundabouts, really.


  1. The mirror was ground incorrectly and produced a consistent distortion in every image taken using the Hubble. Replacing the mirror once the telescope was in orbit was impossible, so instead astronauts replaced some of the other instruments on the Hubble with equipment that introduced a counterbalancing distortion that cancelled out the imperfections of the mirror.
Tagged , , ,

9 thoughts on “Adaptive Optics, Or: Laser Eye Surgery For Telescopes.

  1. Smurf says:

    Fixing the hubble mirror issue is one of the more impressive problem solving projects I’ve ever read about. I love that hubble is basically wearing glasses.

    • Hentzau says:

      They were very lucky that a solution was even possible. If the same thing happens with the James Webb telescope all the way out at L2 NASA is going to be so completely boned.

  2. Gap Gen says:

    I heard the problem with the mirror was that they calibrated it on something attached on the ceiling, and fucked up the measurement between the distance to the mirror and the thing on the ceiling. But yeah, AO is certainly part of the reason why the JWST is focussing on wavelengths outside of visible light – AO is just so damn good at correcting for atmospheric turbulence in visible wavelengths.

    Also, the Imax 3D film about the Hubble repair mission is well worth a look, if it happens to still be on anywhere. Some amazing people who risked their lives to fix an old telescope.

    • Hentzau says:

      The problem with the mirror was that 1) it had to be made to insanely precise specifications, 2) this precluded the use of conventional null correctors to calibrate the mirror so a special one had to be made that would produce a mirror with the required specifications and 3) this led to the company that made the mirror only calibrating it on a single null corrector that happened to have one of the lenses 1.3 mm out of place. So while using the special corrector meant the mirror was incredibly precise, the out-of-place lens meant it was incredibly precisely *wrong*.

      • Smurf says:

        I did like that they eventually checked it again on another device that DID show there was an error, but they dismissed it because the first null corrector didn’t show an error so they assumed the second device was faulty. Rather than there actually being and error.

        I wouldn’t do that in my work, never mind if I was working on a billion dollar space telescope for NASA.

        • Gap Gen says:

          Note: This is why I work in simulations, where if I fuck up I don’t have to send a team of astronauts to fix my mistake.

  3. innokenti says:

    That right there is incredibly cool. Thanks for the revelation!

  4. pertusaria says:

    Wow, great read. I really liked the point about the surprisingly late use of old storage technology.

    • Hentzau says:

      One of the things that’s stuck with me from my uni undergraduate course was being told that satellite/probes being launched at the time (circa 2002-2003) were running on the equivalent of a 486 processor, this being what was current at the time the satellites were proposed.

Leave a Reply