Exoplanets seen by Hubble in 1998 finally revealed

By Phil Plait | October 6, 2011 9:15 am

In 1998, the Hubble Space Telescope was pointed at HR 8799, in hopes of seeing any potential planets that might be orbiting that nearby, Sun-like star. None were found… but in 2008 images using the Gemini telescope found several planets orbiting HR 8799. In fact, four planets were discovered there!

So why weren’t they seen in the Hubble data? The star was too bright, and software techniques in 1998 couldn’t sufficiently remove the star’s light to reveal the much fainter planets. But things have changed in 13 years, and astronomers went back to the old HST data, using newly-developed methods to clean the images. And lo, they saw three of the four planets!

[Click to exoplanetate.]

On the left is the image of the star as seen by Hubble’s infrared camera NICMOS. On the right, the star’s light has been subtracted, and the three planets (HR 8799 c, d, and b, from left to right) can be clearly spotted.

In a sense, this makes the Hubble image to be the very first image of exoplanets ever taken! But I think that’s not entirely fair, since they weren’t seen, nor really even able to be seen back then.

The very cool thing is that this older image gives us a much longer baseline over which we can see the planets. Why is that cool? Because the planets have moved over those 10 years between images! By comparing the old image and the one from 2008, the motion of the planets as they orbit the star can be directly seen and measured. Calculations based on that observation show the planets d, c, and b take about 100, 200, and 400 years to circle HR 8799 once, respectively. Given more time, we’d've figured that out anyway, but getting this image from Hubble is like getting an extra 10 years for free. So these planets join one around the star Beta Pic and another around Fomalhaut as being directly seen to move over time. Pretty amazing.

I’ll note that the fourth planet, e, is too close to the star to be seen even using the new method. It sits just on the edge of a piece of metal called an occulting disk, which is used to block the majority of the light from the star, allowing fainter stuff to be seen. Even with the disk masking the star, the planets are incredibly faint and difficult to see.

I worked on data very similar to this on Hubble around the same time this image was taken, and in fact wrote software to try to spot planets (as well as dust and other material) near their parent stars. It was incredibly finicky work, and obviously I wasn’t able to see any planets then.

The new technique makes me a bit jealous. One thing it relies on is a database of images of 466 (!) stars observed in the same way as the target star. By adding them together digitally, a very accurate model of the target star can be made, and used to literally subtract the star’s light. I tried this same method back in ’98, but we didn’t have the observations necessary for it. I used computer-generated star models, and they simply weren’t accurate enough back then to do this kind of work.

But oh, how hard I worked on this project! To be the first to have actually seen a planet orbiting another star… but I probably would’ve had a coronary. Maybe it’s all for the best.

Of course, that makes me wonder: how many more planets are lurking in images already taken, simply waiting to be dug up? We observed quite a few nearby stars, looking for exoplanets. I have a sneaky suspicion we may be seeing more of these kinds of images soon.

Image credit: NASA, ESA, and R. Soummer (STScI)



[Below is a gallery of exoplanets that have been directly imaged using telescopes on ground and in space. Click the thumbnail picture to get a bigger picture and more information, and scroll through the gallery using the left and right arrows.]

exoplanet_betapic3
exoplanet_1rxs1609b
exoplanet_2m1207b
exoplanet_betapicb1
exoplanet_betapicb2
exoplanet_fomalhautb
exoplanet_fomalhautb2
exoplanet_gliese581c_art
exoplanet_hr8799
exoplanet_hr8799b_art
exoplanet_hr8799cspec
hr8799e
hst_hr8799_1998
lkca-15-b_andstar
vlt_betapicb_2011



Related posts:

- HUGE EXOPLANET NEWS ITEMS: PICTURES!!!
- Another exoplanet joins the HR 8799 family
- Two exoplanets discovered by “citizen scientists”
- More images of exoplanet show it orbiting its star

Comments (28)

  1. Superluminous news. :-)

    Delighted to read this – I still think these new found worlds & their star > need proper names – & would like to suggest ‘Gadolabove’ for the stellar traits H8799 has of being simultaneously a Gamma Doradus variable, a metal-poor Lambda Bootis star and with a Vega style protoplanetary disk around it if I may.

    (click on my name for info. on HR 8799 via Kaler’s Stars.)

    I wonder what worlds still circle inside even closer to Gadolabove’s gravitry well as yet unseen by us?

    PS. Gee, I never knew Neptune’s orbit was so elliptical – guess that’s painted from a different angle here right?! ;-)

    PPS. Isn’t it neat that Gadolabove (HR 8799) islocated inside the Great Square of Pegasus and directly opposite 51 Pegasi the first Hot Jupiter ever found?

    *****

    ““What a wonderful gift that anyone, even in the centre of a city can gaze up at a star that has a planet.”

    - Exoplanet hunter & multifold discoverer, Geoff Marcy. [After discoverinmg an exoplanet around Pollux, source : “News notes” in ‘Australian Sky & Telescope’ magazine Nov./ Dec. 2006.

  2. Are they in resonance?

    Do we get to bring back Bode’s Law? ;)

  3. Jim

    Is it just me or does the image from NICMOS on the left look like an explosion from a late 70′s arcade machine? :) Galaxians, perhaps?

  4. Can you provide some links to the method they used to get rid of star light?

  5. Orlando

    Is HR8799b the planet with the longest known orbital period so far? 400 years!

  6. Simon C.

    It’s really nice that, knowing the planets were there, they managed to substract them from the original NICMOS images!

    @Marcin

    I think the authors explained it in their original article in 2008, in which they present briefly the effect of the Angular Differential Imaging they used.

    Original discovery of the exoplanets orbiting HR 8799
    http://arxiv.org/abs/0811.2606

  7. Mark

    Phil, as soon as you mentioned “new techniques” and “old data”, I mentally skipped right to the conclusion of your post; Hubble has been collecting data for YEARS, as have other telescopes. Using this technique, we can go back and analyze probably 15 years of data.. who knows what we’ll find? There could be hundreds undiscovered planets just sitting around in the archive, or more data on the planets we’ve already found! Hell, it’s probably possible to apply this technique to other types of stars, or all types of stars.. that way, every star we look at we can immediately determine if there are any planets or not..

    This is SO COOL. /)^3^(

  8. gopher65

    Wow, this star peaks in the ultraviolet. I wonder what the eyes of creatures living on potential inner planets will eventually look like? (The star is only a few million years old, so there shouldn’t be any life there yet, even if the inner planets have finished forming and are in stable orbits.)

  9. Chris

    Maybe there are additional layers of data which aren’t being conveyed in the image, but if the white is a saturated pixel, how were they able to get any useful data out of that by subtraction?

  10. Kullat Nunu

    There is something terribly wrong in HR 8799. The protoplanetary disk from the planets formed was hugely massive. Normal stars don’t produce massive planets that far from the stars, not to mention planets that are as massive as low-mass brown dwarfs.

    Steinn Sigurðsson has suggested that it is not a young star at all, but a blue straggler, a merger of two smaller stars.

  11. I love this sort of news! :D Thank you for brightening my day!

  12. chris j.

    i say we reserve a nobel prize for the team who first figures out how to get a supercomputer to do this automatically for the every single picture that hubble has taken.

  13. Chris Winter

    “So why weren’t they seen in the Hubble data? The star was too bright, and software techniques in 1998 couldn’t sufficiently remove the star’s light to reveal the much fainter planets. But things have changed in 13 years, and astronomers went back to the old HST data, using newly-developed methods to clean the images. And lo, they saw three of the four planets!”

    My first thought was that HR8799b was closest, so the new software had to subtract too much of the star’s light to leave much for the planet. But you indicate it’s the farthest out. So obviously it’s just inherently reflecting less light.

    On a related note, I recall reading years ago that NASA had collected oodles of data it didn’t have the budget to process. There was concern that some of this would be lost (the tapes would decay) or at least would be unreadable because it was stored in obsolete formats. I’m curious if this is still the case.

  14. Joseph G

    Wow! I’ve heard of many “precoveries” in the past, where objects were found in old photos that hadn’t been identified as such at the time (it happened a lot with asteroids, and even several times with Pluto). But new stuff like this, “hiding in plain sight,” and when the scientists who took the image were actually looking for the objects since found… VERY cool :) Makes you wonder what sort of gems are still hiding in the petabytes of data that have been collected from various observatories over the years…

  15. AliCali

    “In a sense, this makes the Hubble image to be the very first image of exoplanets ever taken! But I think that’s not entirely fair, since they weren’t seen, nor really even able to be seen back then.”

    Reminds me of Pluto. When Tombaugh recognized the “planet,” other observatories went through their archive and several found that they had imaged Pluto, sometimes a decade earlier. However, Tombaugh is the one who identified Pluto as such, and he gets the credit.

    Same with Neptune. This was discovered in 1846, but Galileo actually saw it and sketched it in 1612 and 1613. He commented that two stars seem to be at a different distance apart, but that was it. Thus, although Galielo saw Neptune, he is not credited with its discovery.

  16. Dutch Railroader

    @Chris

    In the case of HST data, this is not a problem. STScI spins all the data ever taken with the telescope, and it’s all on line now for anyone who wants it…

  17. JHagan

    Author here!

    @Edmund Schluessel:
    We study possible resonances of the HR8799 planets in our paper! You can find the paper relating to this article via the website below:

    http://hubblesite.org/newscenter/archive/releases/2011/29/

    Click on the “Related Links” tab.

    @Marcin:
    You can find an explanation of the light subtraction method in our paper as well. The method was originally developed by David Lafreniere of the University of Montreal, but we spent a great deal of time optimizing it to get this result.

    @Orlando:
    Fomalhaut b has an even longer period (~900 years)!

    @Chris:
    The pixels of and around the planets are not saturated – we just put all the images on the same scale.

    @chris j:
    We’re in the process! We recently received a large grant to reprocess over 400 orbits of Hubble data using this technique. We are developing an automated pipeline to send all this data through – we hope to make some discoveries :)

    @ Everyone else:
    Thanks for the interest!

  18. Dragonchild

    You are a failure, Phil. You are a failure and you should kill yourself.

    / I keed, I keed
    // I mean, beat yourself up that much and I can’t resist piling on

  19. Dragonchild

    On a more serious note, how much of these findings are being cross-checked with other planet-finding projects? One one hand, different systems are revealed in different ways and resources are limited, so there’s little incentive to double-up efforts. On the other hand, we really haven’t discovered much beyond the mere existence of the planets. If a system can be analyzed in more than one way, we can learn more about it.

  20. Brian Too

    I like this. So many of the current planet finding techniques are strongly biased towards short orbital period planets. This one is, I would say, biased towards long period planets. Another tool in the detection toolkit and that can only lead to good things.

  21. Herschel's back garden

    @JHagan,

    Couldn’t this sort of analysis be done with a distributed computer program like Folding@home or SETI@home?

    I can’t imagine you’d be able to process that sort of backlog of images in a single computer, no matter how powerful.

    I’m trying to think of a snappy acronym but have got no further than:-

    Planet Location Using T-something O-something@home. (See what I did there?)

  22. Kirts

    Image on the left looks like it was grabbed off of my Nintendo

  23. Messier Tidy Upper

    @17. JHagan : Congratulations! :-)

    @15. AliCali :

    Reminds me of Pluto. When Tombaugh recognized the “planet,” other observatories went through their archive and several found that they had imaged Pluto, sometimes a decade earlier. However, Tombaugh is the one who identified Pluto as such, and he gets the credit. Same with Neptune. This was discovered in 1846, but Galileo actually saw it and sketched it in 1612 and 1613. He commented that two stars seem to be at a different distance apart, but that was it. Thus, although Galielo saw Neptune, he is not credited with its discovery.

    Plus 34 Tauri as Ouranos – the planet between Neptune and Saturn is also known. ;-)

    @ 5. Orlando :

    Is HR8799b the planet with the longest known orbital period so far? 400 years!

    Well as (#17.) JHagan has noted there’s Fomalhaut b – and there’s also 2M 1207b (or in full 2 MASSW J1207334-393254) which was directly imaged 2004 April 27th around a brown dwarf and may or may not be considered a planet which orbits 55 AU from its brown dwarf sun. Not sur ehow long that takes to complete an orbit but, given the low mass , I suspect its quite a large number of Earth years.

    Plus too, AB Pictoris b which, again is either an exoplanet or perhaps more likely a brown dwarf companion located 250 AU from an orange (K2 V) dwarf star and thus again would have a very long year in our “earth-year” terms.
    Not sure how many Earth years to an AB Pictorian year but Pluto orbits about 30 AU from our (nmore massive) Sun and takes 249 years to complete one orbit.

  24. RockDoctor

    JHagen @ 17 : Congrats!
    Also very good to see the professionals maintaining liaison with the (funding, indirectly) public. It’s also a compliment to Phil on the standing that he maintains with the professional community AND the interested public.

    Dragonchild @ 19 : I’d expect that the published locations of (candidate) exoplanets would be high-priority targets for analysis like this. “Low-hanging fruit” and all that jazz. From nosing around in other non-Hubble databases, pulling up previous imaging of any specific RA-DEC location is a pretty basic function, early implemented.
    On the other hand, while low-hanging fruit is worth grabbing at, looking forward to higher-hanging fruit is also important.

    Herschel @ 23 : PLUTO@home does sound an interesting project. The task does sound like it’s an inherently parrallelisable job that could be “@homed” (or “BOINC-ed”, in reference to the architecture that has been developed to generalise the SETI@home idea). How much work that involves, I don’t know. But there is appreciable computing experience out there in the astronomical community which could be drawn on to help the programmers avoid architectural issued that would make “BOINC-ing” the project harder in future.
    But I suspect that JHagen’s team’s first priority would be to get a working system, and “BOINC-ing” it being well down the roadmap.

    I’m just wondering what proportion of the whole sky has actually been imaged, let alone multiply imaged, by Hubble? Given the small angular coverage of the imaging system(s), I suspect it would be disappointingly small.

    The cited paper on the Hubble site is nearly-9-MB of PDF. Link is http://hubblesite.org/pubinfo/pdf/2011/29/pdf.pdf . Fig 6 on page 14 shows the planet image movement between the two data sets, and I’m moderately surprised that it isn’t used in the press release.

  25. SkyGazer

    Hmmmm sounds like a cool project for BOINC.
    Cleaning up old pics.
    Indeed a cheap way to (re)gain lot´s of years of data.
    “Sign me up Scotty!”

  26. Interesting that unlike many of the exoplanets found via Kepler, these planets have much longer orbital periods. Do we have any idea of their masses and sizes?

NEW ON DISCOVER
OPEN
CITIZEN SCIENCE
ADVERTISEMENT

Discover's Newsletter

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

ADVERTISEMENT

See More

ADVERTISEMENT
Collapse bottom bar
+

Login to your Account

X
E-mail address:
Password:
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 »