I don't mean to diminish the cleverness of those involved in this project at all, but the article summary is a little misleading. While the discovery was made with very small-scale telescopes, the confirmation that this was actually a planet came from two large telescopes, the Harlan J. Smith Telescope (2.7 meter aperture) and the Hobby-Eberly Telescope (9.2 meter effective aperture), as the linked article mentions.
Finding extrasolar planets by the transit method, where you moniter large fields of stars and look for brightness variations as a planet passes in front of one of your targets and blocks some light, is pretty straight-forward. You tend to only need somewhere between 0.1% and 1% precision in your photometry, which requires some work to achieve, but is by no means prohibitive. So it's a good technique for amateurs to get involved with, especially when you consider that smaller telescopes tend to have larger fields of view, so you can moniter more stars at once. But the main stumbling block transit-searchers have run into is the false positive rate. The biggest surveys have found a huge false-positive rate (90-95%) among the planet candidates. It turns out there are lots of things that can make a star dim at fixed intervals, from grazing binaries to starspots.
As a result, transit planet candidates are only considered confirmed when there are measurements of a radial-velocity wobble consistent with the orbital period found by the transit. To get the radial velocity precision you need (for the Hot Jupiters transits detect, precision of tens of meters per second is sufficient), it takes a precise, high resolution spectrograph (very expensive), mounted on a large telescope (at least a couple meters).
I should also point out that transit searches are sensitive mainly to close-in planets. The sensitivity function drops very quickly as the planet moves further out (both because you need a longer sustained campaign, and because the chances of the planet's orbit crossing the star decreases). All the transit detections thus far have been from planets with several-day orbits. While this is interesting science, there's a lot of work to be done with planets in other regimes. The straight-up radial velocity technique gets you planets at seperations between 0 and 5 AU or so (over 150 planets found this way so far), the microlensing method can also detect planets at much larger orbital separations (2 or 3 planets up until now), and direct imaging is ideally suited for large-seperation planets (only the 1 good planet at this point). My point is that you can't cover this whole range of parameter space with small telescopes alone. Radial velocity and direct imaging require large investments in hardware, both in the large telescope itself and the instrumentation (disclaimer: I work on direct imaging, that's why I keep bringing it up). It's also important to note that one of the reasons people find transiting planets so interesting is the possibility of getting spectral information out of the planets. NASA's Spitzer space telescope recently detected the secondary eclipse (the loss of light when the planet is hidden behind the star) of two transiting extrasolar planets. This is pretty exciting science, since you can really compare data to models this way, but it requires some extensive telescope set-ups to get it done.
So again, this is certainly a great project for getting amateurs involved in the planet-finding game, and I"m very impressed with this result. But don't close down Keck and the VLT and Hubble just yet; there's a lot of work to be done in extrasolar planet research, and much of it requires large telescopes with new (read: expensive) instruments.
I don't totally agree. Definitely, the extrasolar planets found by the radial velocity planet searches are largely close to their stars, but that's another observational bias. Not only do closer-in planets tug on their stars more (v ~ 1/r^1/2), but it takes longer for a large-separation planet to complete an orbit, and radial velocity teams don't report a planet until they've seen one orbit. Which means the time baseline of the surveys becomes important. The longest target stars have been monitered is 15 years or so. This is, not coincidentally, the orbital period of the largest-separation extrasolar planet known to date, 55 Cnc d (14.7 years, 6 AU). Also, the star 55 Cnc was being carefully monitered all this time in large part because it was known to already harbor a planet (55 Cnc b, has just a 15 day period, and was one of the first half-dozen extrasolar planets discovered).
My point is that while the results of the radial velocity surveys are pretty complete within 3 or 4 AU or so, beyond this the results are heavily driven by observational bias. Not only do you need 15 years of data to close an orbit, you need enough data points to see a much fainter radial velocity signature. For comparison, the next furthest-out extrasolar planet is at 4.5 AU, and of the 136 extrasolar planets found by the radial velocity method, only 5 are beyond 4 AU (see the California and Carnegie planet almanac for details).
Jupiter is in an orbit of 5.2 AU, taking 12 years to go around the sun. So I would submit we've found no planets of Jupiter-like mass at Jupiter-like orbits (closest would be 55 Cnc d, 6 AU, but--at least--4 times the mass of Jupier, or HD 50499, 1.84 Jupiter masses at 4.4 AU). And I'd say further that current observational techniques would really need to stretch to hit such a planet, so I don't think we're not finding them because they aren't there. (Plus, it's widely suspected that radial velocity teams know about a lot of these long-period planets, but are waiting to announce them until the orbits have been confirmed. All the rest of us can do is wait and see).
As to the basic question of this thread, whether you can get other stellar systems similar to our solar system (namely, a rocky planet in a stable orbit in the habitable zone), I think that issue is nowhere near solved. Recent papers have shown that of nearby, sun-like stars, about 10-20% have a planet that can be detected with the radial velocity method (the exact percentage depends on the metallicty of the star). What that means is that we know between a tenth and a fifth of stars have a planet more massive than Jupiter within the inner 3 or 4 AU. That says absolutely nothing about the other 80-90% of stars. What fraction of these have a Jupiter-like planet in a Jupiter-like orbit is very much up for grabs. We know that it's unlikely for an earth to form in most of the planetary systems we've been seeing (migrating giant planets, or planets in eccentric orbits, would almost certainly disrupt the earth-wannabe). But again, that's only 10 or 20% of stars. So, it could very well be that 80-90% of stars have a rocky planet in the habitable zone. We don't know how common our solar system is yet, and it'll likely take future missions (Kepler, TPF, next-generation adaptive optics systems on ground based telescopes) to really find out.
Actually, I assume he did in fact mean the VLTI, or VLT Interferometer. Given the broken English, my first thought was that he was a European working on the project, but I'm too lazy to confirm that. The idea is basically to have a VLBI-like-array, only in the optical and IR, using the four VLT telescopes and a series of outrigger telescopes that can be moved around the main four. As to how it stacks up to Hubble in the visible, I'm not really an expert on that. But with baselines of order a hundred meters, the theory certainly looks good.
I have used the VLT in the infrared, and can tell you it (well, UT-4 with NACO) quite certainly does better than Hubble in resolution. And while my extragalactic friends (er, friends studying things outside the Milky Way, not actual alien friends) may not be happy to hear it, being able to image faint sources isn't the ultimate factor in ranking telescopes. I look for planets, so I care a lot about contrasts, and the VLT can very nicely out-perform Hubble there.
We've recently discovered n close, faint companion to the star AB Dor with the VLT-NACO system, and got some very good science out of it. This is an object Hubble looked for prior to us, and failed to detect. Mind you, we had some special optics that Hubble didn't, so we did have an advantage there.
But ultimately, I agree with your main point, that there are many factors in determining which telescope is best for your science. The VLT excels in areas where Hubble falters, and vice versa. Astronomy will be at its best with every resource it can get; there's a lot less redundancy than people assume. So yeah, boo NASA, and I need to figure out how to fit a "Save Hubble!" bumper sticker on my bike.
Just to throw in some numbers on your first point, yes, Spitzer doesn't have a chance of splitting HD 209458 and its planet (cleverly named HD 209458b) into two objects on the detector. HD 209458 is a star 50 parsecs away, and the semi-major axis of the planet's orbit is 0.05 AU, translating to about 1 milli-arcsecond of separation or so (that's the size of a penny seen from 4000 km away, for reference).
To see that in the mid-infrared (10 microns, say), you'd need a telescope about 2 km across. Best we're up to now is 10 meters, and ground-based telescopes don't do very well in the mid-IR anyway. With Spitzer's relatively puny 0.85 meter mirror, it doesn't stand a chance. The other planet, TReS-1, orbits a star even further away from us, so the situation's worse.
As you said, this is the telescope getting light out of one "pixel" (one resolution element, anyway), and watching how the amount of light changes during a transit. And just to add to your second paragraph, yes, the slashdot headline is incorrect, both planets were discovered previously, then CONFIRMED by Spitzer observations (though getting photons from an extrasolar planet itself is still very cool).
The two planets are both of the transiting variety that pass in front of their parent star (there are about a half-dozen of these we know about), and this is why they were targeting them with Spitzer to begin with, but they were discovered with different techniques. HD 209458b was detected by the radial velocity (wobble) method, then was shown to transit later. TrES-1 was discovered as a star with a variable lightcurve, then confirmed with radial velocity observations.
Personally, I think this means there are THREE different approaches confirming these two planets, not just the two. First, the radial velocity measurements showing the star wobbling during the planet's orbit. Second, the primary transits, where the planets block out visible light from their stars as they pass between us and the stars. And finally the secondary transits, where the stars blocks the planets' infrared emission as the planets pass behind their stars. There aren't many people who still don't believe in extrasolar planet discoveries, but I think this last piece of evidence makes for a pretty bullet-proof case.
As you get to the very lowest masses of the low-mass stars (about.08 solar masses, or 80 Jupiters), I think you do start getting water lines. The issue is that most stellar photospheres (like our sun's, say) are so hot you can't have molecules of any kind, whereas in cooler stars or brown dwarfs you can. Most people are at least vaguely familiar with the idea of atmoic spectra, where you get some number of sharp (more or less) lines throughout the visible, which, in theory, could be mistaken for one another (it certainly happens for some of the low-strength metal lines). Molecules, on the other hand, have large "bands" of absorption, where light over a large wavelength range is heavily absorbed. It's fun quantum mechanics, but basically the central line is some transition to do with the molecule vibrating (the distance between the oxygen and hydrogen atoms in the water molecule oscillating, say), which is then broadened by a number of transitions to do with the whole molecule rotating.
The end result is you get very distinctive molecular features. See the bottom image on this page on another way of looking for planets, where the black line is the spectrum of the first brown dwarf discovered, Gliesse 229B. Everything in that spectrum redward (longer wavelength) of 1.6 microns is heavily suppresed by methane (another common molecule, CH4)...and I think after about 1.8 you start getting into water bands, too, but don't quote me on that. Anyway, that's all a convoluted way of saying if it looks like water, it probably is water. But there's a large range of planets and brown dwarfs expected to have such water bands in the spectrum (indeed, the parent brown dwarf shows similiar spectral features to the companion)
As to the photo, the data have certainly gone through a lot of processing in the reduction process, but that shouldn't be though of as having photoshopped the result. I've used this same instrument, and there are some features I'm curious about (lack of diffraction spikes from spider arms, mainly), but it seems pretty legitimate. It's just important to realize what this image is showing.
For one thing, that isn't reflected light making the "planet" shine, but rather the planet is powering itself. When you form something like a planet or brown dwarf, and it isn't massive enough to start fusion, gravity will make the object contract. The potential energy of that contraction needs to go somewhere, so half of it heats the gas, the other half comes out as radiation (mostly infrared). As the object ages, the contraction slows, and it gets much fainter. As an aside, even after 5 billion years, Jupiter emits more radiation than it receives from the sun. That's part of the trouble with detecting planets in general, in cases like this we know how much energy is coming out of the object, roughly how old it is, and then we have to rely on theoretical models (which aren't well-tested at all) to tell us how massive the object should be.
The second point is to remember that the "size" of the two objects in that image aren't the physical sizes at all. In fact, the telescope is barely able to resolve the separation between the two, let alone physical structure of the surfaces of these bodies (a scale 10,000 times smaller or so). "Size" of the disk on the detector is simply a matter of our imperfect telescope optics (and basic physics of diffraction) taking the points of lights from parent and companion and smearing them out: brighter objects seem to reach further out on the detector. The diffuse structure near the edge of the bright parent is again introduced by the optics, and isn't a real feature.
The image itself is your typical false-color image. All the data are in the infrared, so for us to see it, a quick remapping has to be done. Rather than have red (0.65 microns), green (0.55 microns), and blue (0.45 microns) filters that are then reproduced by red, green, and blue pixels,
The preprint of the paper lists the parent brown dwarf as 2MASS J12073346-3932539 , which is indeed at the above coordinates. The candidate planet (much in the same way Ralph Nader is the candidate president, but there's my bias showing) will be 0.46 arcseconds south and 0.63 arcseconds east.
In case I didn't discourage any amateur astronomers thus far, here's some more: That's a separation of 0.77 arcseconds, when the seeing at most sites is of order 1 arcsecond. The companion is 100 times brighter than the parent brown dwarf in the K band. The parent brown dwarf has a K of about 12, and for an M8 spectral type, that's a V-magnitude of about 19 or 20. For those of you scoring at home, the parent brown dwarf is one million times fainter than anything you can see with the human eye.
The companion is an even redder object, so the colors will be much, much worse at V (there's a reason we try to detect these in the infrared). With a state-of-the-art AO system (look what we did with the same system earlier this year imaging the surface of Titan) on an 8 meter telescope with excellent infrared detectors, the companion lies one magnitude above the detection limit on their sensitivity/separation curves.
Sorry to depress people looking forward to pointing your telescope at this system tonight, but if it makes you feel better, it's probably not a planet.
I just checked that RA, by the way. It's behind the sun right now. You'll have to wait until January to observe it. Or to point your telescope there and not observe it, as the case may be.
Actually, this is an interesting case in that if it is a false alarm, the most likely explanation is that it's a foreground brown dwarf. The ESO team seems to have a noisy spectrum and JHK photometry of the companion. The colors really do suggest an L-dwarf (as opposed to a background K star, say), and the methane feature in the spectrum does seem real, assuming those data were reduced correctly (Difficult under the best conditions, not made easier by a brighter object 0.7 arcseconds away).
But you're absolutely right, until the proper motion study is confirmed, they've got nothing besides staking their claim to the object on the chance they get lucky. The TW Hydra association (to which the "parent" brown dwarf belongs) moves on the sky at about 66 milli-arcseconds a year, so it should be possible in a few years to see if the companion is following or not.
For me, the kicker is trusting the models. All determinations of mass (including of the parent brown dwarf) are done by guessing age, measuring luminosity and/or color, and seeing what the models of brown dwarfs/extrasolar planets say the mass should be. These are cooling tracks calibrated to Jupiter at 5 billion years, predicting how bright something is at 8 million years. Even the people making the models are hesitant to trust their own numbers at the youngest ages (Heck, we don't even know how long it takes to form a large Jupiter out of the disk).
Right now, I'd set the odds at about 80% that it won't be confirmed as a proper motion pair. In which case, my guess would be a foreground brown dwarf of a much larger mass (older and dimmer, but closer to still put it at 18th magnitude in H). If I'm wrong, the open question of these thoroughly untested models remains the cloud of uncertainty hanging over all of this.
It should be noted that while the initial discovery made with a 10 cm telescope set up relatively cheaply, the confirmation that it was in fact a planet came using the multi-million dollar Keck telescope. Transit detections have a notoriously bad record for turning out to be planets (a few percent are actually confirmed, the rest are some other effect), so you need follow-up with high resolution spectra. And for an 11th magnitude star, that requires a large telescope like Keck.
They could have gotten away with a 3-meter like at Lick Observatory, in this case, given the large radial velocity amplitude, but for further-out planets Keck is the only way to go. Trying to do this part of the science with more 10 cm telescopes would be out of the question.
The advantage to small telescopes is the ability to stare at a large section of the sky at once (in the case of the 10-cm telescopes used for this project, each exposure covers 6 degrees of sky). Compare this to the 10-meter Keck telescopes, whose imaging systems have fields of view of about an arcminute (1/60th of a degree). For transit searches, you want to keep staring at a star until you get lucky with a planet passing in front of the star, then confirm that as the transit happens again and again. So your best bet for optimizing your transit search is to look at a lot of stars at once, meaning a big field of view, which you get most easily from small telescopes.
It's not impossible to get large fields of view with large telescopes, but it takes a lot more effort. Check out the plans for building the 8.4 meter LSST for details.
Slashdot Mirror
I don't mean to diminish the cleverness of those involved in this project at all, but the article summary is a little misleading. While the discovery was made with very small-scale telescopes, the confirmation that this was actually a planet came from two large telescopes, the Harlan J. Smith Telescope (2.7 meter aperture) and the Hobby-Eberly Telescope (9.2 meter effective aperture), as the linked article mentions.
Finding extrasolar planets by the transit method, where you moniter large fields of stars and look for brightness variations as a planet passes in front of one of your targets and blocks some light, is pretty straight-forward. You tend to only need somewhere between 0.1% and 1% precision in your photometry, which requires some work to achieve, but is by no means prohibitive. So it's a good technique for amateurs to get involved with, especially when you consider that smaller telescopes tend to have larger fields of view, so you can moniter more stars at once. But the main stumbling block transit-searchers have run into is the false positive rate. The biggest surveys have found a huge false-positive rate (90-95%) among the planet candidates. It turns out there are lots of things that can make a star dim at fixed intervals, from grazing binaries to starspots.
As a result, transit planet candidates are only considered confirmed when there are measurements of a radial-velocity wobble consistent with the orbital period found by the transit. To get the radial velocity precision you need (for the Hot Jupiters transits detect, precision of tens of meters per second is sufficient), it takes a precise, high resolution spectrograph (very expensive), mounted on a large telescope (at least a couple meters).
I should also point out that transit searches are sensitive mainly to close-in planets. The sensitivity function drops very quickly as the planet moves further out (both because you need a longer sustained campaign, and because the chances of the planet's orbit crossing the star decreases). All the transit detections thus far have been from planets with several-day orbits. While this is interesting science, there's a lot of work to be done with planets in other regimes. The straight-up radial velocity technique gets you planets at seperations between 0 and 5 AU or so (over 150 planets found this way so far), the microlensing method can also detect planets at much larger orbital separations (2 or 3 planets up until now), and direct imaging is ideally suited for large-seperation planets (only the 1 good planet at this point). My point is that you can't cover this whole range of parameter space with small telescopes alone. Radial velocity and direct imaging require large investments in hardware, both in the large telescope itself and the instrumentation (disclaimer: I work on direct imaging, that's why I keep bringing it up). It's also important to note that one of the reasons people find transiting planets so interesting is the possibility of getting spectral information out of the planets. NASA's Spitzer space telescope recently detected the secondary eclipse (the loss of light when the planet is hidden behind the star) of two transiting extrasolar planets. This is pretty exciting science, since you can really compare data to models this way, but it requires some extensive telescope set-ups to get it done.
So again, this is certainly a great project for getting amateurs involved in the planet-finding game, and I"m very impressed with this result. But don't close down Keck and the VLT and Hubble just yet; there's a lot of work to be done in extrasolar planet research, and much of it requires large telescopes with new (read: expensive) instruments.
I don't totally agree. Definitely, the extrasolar planets found by the radial velocity planet searches are largely close to their stars, but that's another observational bias. Not only do closer-in planets tug on their stars more (v ~ 1/r^1/2), but it takes longer for a large-separation planet to complete an orbit, and radial velocity teams don't report a planet until they've seen one orbit. Which means the time baseline of the surveys becomes important. The longest target stars have been monitered is 15 years or so. This is, not coincidentally, the orbital period of the largest-separation extrasolar planet known to date, 55 Cnc d (14.7 years, 6 AU). Also, the star 55 Cnc was being carefully monitered all this time in large part because it was known to already harbor a planet (55 Cnc b, has just a 15 day period, and was one of the first half-dozen extrasolar planets discovered).
My point is that while the results of the radial velocity surveys are pretty complete within 3 or 4 AU or so, beyond this the results are heavily driven by observational bias. Not only do you need 15 years of data to close an orbit, you need enough data points to see a much fainter radial velocity signature. For comparison, the next furthest-out extrasolar planet is at 4.5 AU, and of the 136 extrasolar planets found by the radial velocity method, only 5 are beyond 4 AU (see the California and Carnegie planet almanac for details).
Jupiter is in an orbit of 5.2 AU, taking 12 years to go around the sun. So I would submit we've found no planets of Jupiter-like mass at Jupiter-like orbits (closest would be 55 Cnc d, 6 AU, but--at least--4 times the mass of Jupier, or HD 50499, 1.84 Jupiter masses at 4.4 AU). And I'd say further that current observational techniques would really need to stretch to hit such a planet, so I don't think we're not finding them because they aren't there. (Plus, it's widely suspected that radial velocity teams know about a lot of these long-period planets, but are waiting to announce them until the orbits have been confirmed. All the rest of us can do is wait and see).
As to the basic question of this thread, whether you can get other stellar systems similar to our solar system (namely, a rocky planet in a stable orbit in the habitable zone), I think that issue is nowhere near solved. Recent papers have shown that of nearby, sun-like stars, about 10-20% have a planet that can be detected with the radial velocity method (the exact percentage depends on the metallicty of the star). What that means is that we know between a tenth and a fifth of stars have a planet more massive than Jupiter within the inner 3 or 4 AU. That says absolutely nothing about the other 80-90% of stars. What fraction of these have a Jupiter-like planet in a Jupiter-like orbit is very much up for grabs. We know that it's unlikely for an earth to form in most of the planetary systems we've been seeing (migrating giant planets, or planets in eccentric orbits, would almost certainly disrupt the earth-wannabe). But again, that's only 10 or 20% of stars. So, it could very well be that 80-90% of stars have a rocky planet in the habitable zone. We don't know how common our solar system is yet, and it'll likely take future missions (Kepler, TPF, next-generation adaptive optics systems on ground based telescopes) to really find out.
Actually, I assume he did in fact mean the VLTI, or VLT Interferometer. Given the broken English, my first thought was that he was a European working on the project, but I'm too lazy to confirm that. The idea is basically to have a VLBI-like-array, only in the optical and IR, using the four VLT telescopes and a series of outrigger telescopes that can be moved around the main four. As to how it stacks up to Hubble in the visible, I'm not really an expert on that. But with baselines of order a hundred meters, the theory certainly looks good.
I have used the VLT in the infrared, and can tell you it (well, UT-4 with NACO) quite certainly does better than Hubble in resolution. And while my extragalactic friends (er, friends studying things outside the Milky Way, not actual alien friends) may not be happy to hear it, being able to image faint sources isn't the ultimate factor in ranking telescopes. I look for planets, so I care a lot about contrasts, and the VLT can very nicely out-perform Hubble there.
We've recently discovered n close, faint companion to the star AB Dor with the VLT-NACO system, and got some very good science out of it. This is an object Hubble looked for prior to us, and failed to detect. Mind you, we had some special optics that Hubble didn't, so we did have an advantage there.
But ultimately, I agree with your main point, that there are many factors in determining which telescope is best for your science. The VLT excels in areas where Hubble falters, and vice versa. Astronomy will be at its best with every resource it can get; there's a lot less redundancy than people assume. So yeah, boo NASA, and I need to figure out how to fit a "Save Hubble!" bumper sticker on my bike.
Just to throw in some numbers on your first point, yes, Spitzer doesn't have a chance of splitting HD 209458 and its planet (cleverly named HD 209458b) into two objects on the detector. HD 209458 is a star 50 parsecs away, and the semi-major axis of the planet's orbit is 0.05 AU, translating to about 1 milli-arcsecond of separation or so (that's the size of a penny seen from 4000 km away, for reference).
To see that in the mid-infrared (10 microns, say), you'd need a telescope about 2 km across. Best we're up to now is 10 meters, and ground-based telescopes don't do very well in the mid-IR anyway. With Spitzer's relatively puny 0.85 meter mirror, it doesn't stand a chance. The other planet, TReS-1, orbits a star even further away from us, so the situation's worse.
As you said, this is the telescope getting light out of one "pixel" (one resolution element, anyway), and watching how the amount of light changes during a transit. And just to add to your second paragraph, yes, the slashdot headline is incorrect, both planets were discovered previously, then CONFIRMED by Spitzer observations (though getting photons from an extrasolar planet itself is still very cool).
The two planets are both of the transiting variety that pass in front of their parent star (there are about a half-dozen of these we know about), and this is why they were targeting them with Spitzer to begin with, but they were discovered with different techniques. HD 209458b was detected by the radial velocity (wobble) method, then was shown to transit later. TrES-1 was discovered as a star with a variable lightcurve, then confirmed with radial velocity observations.
Personally, I think this means there are THREE different approaches confirming these two planets, not just the two. First, the radial velocity measurements showing the star wobbling during the planet's orbit. Second, the primary transits, where the planets block out visible light from their stars as they pass between us and the stars. And finally the secondary transits, where the stars blocks the planets' infrared emission as the planets pass behind their stars. There aren't many people who still don't believe in extrasolar planet discoveries, but I think this last piece of evidence makes for a pretty bullet-proof case.
As you get to the very lowest masses of the low-mass stars (about .08 solar masses, or 80 Jupiters), I think you do start getting water lines. The issue is that most stellar photospheres (like our sun's, say) are so hot you can't have molecules of any kind, whereas in cooler stars or brown dwarfs you can. Most people are at least vaguely familiar with the idea of atmoic spectra, where you get some number of sharp (more or less) lines throughout the visible, which, in theory, could be mistaken for one another (it certainly happens for some of the low-strength metal lines). Molecules, on the other hand, have large "bands" of absorption, where light over a large wavelength range is heavily absorbed. It's fun quantum mechanics, but basically the central line is some transition to do with the molecule vibrating (the distance between the oxygen and hydrogen atoms in the water molecule oscillating, say), which is then broadened by a number of transitions to do with the whole molecule rotating.
The end result is you get very distinctive molecular features. See the bottom image on this page on another way of looking for planets, where the black line is the spectrum of the first brown dwarf discovered, Gliesse 229B. Everything in that spectrum redward (longer wavelength) of 1.6 microns is heavily suppresed by methane (another common molecule, CH4)...and I think after about 1.8 you start getting into water bands, too, but don't quote me on that. Anyway, that's all a convoluted way of saying if it looks like water, it probably is water. But there's a large range of planets and brown dwarfs expected to have such water bands in the spectrum (indeed, the parent brown dwarf shows similiar spectral features to the companion)
As to the photo, the data have certainly gone through a lot of processing in the reduction process, but that shouldn't be though of as having photoshopped the result. I've used this same instrument, and there are some features I'm curious about (lack of diffraction spikes from spider arms, mainly), but it seems pretty legitimate. It's just important to realize what this image is showing.
For one thing, that isn't reflected light making the "planet" shine, but rather the planet is powering itself. When you form something like a planet or brown dwarf, and it isn't massive enough to start fusion, gravity will make the object contract. The potential energy of that contraction needs to go somewhere, so half of it heats the gas, the other half comes out as radiation (mostly infrared). As the object ages, the contraction slows, and it gets much fainter. As an aside, even after 5 billion years, Jupiter emits more radiation than it receives from the sun. That's part of the trouble with detecting planets in general, in cases like this we know how much energy is coming out of the object, roughly how old it is, and then we have to rely on theoretical models (which aren't well-tested at all) to tell us how massive the object should be.
The second point is to remember that the "size" of the two objects in that image aren't the physical sizes at all. In fact, the telescope is barely able to resolve the separation between the two, let alone physical structure of the surfaces of these bodies (a scale 10,000 times smaller or so). "Size" of the disk on the detector is simply a matter of our imperfect telescope optics (and basic physics of diffraction) taking the points of lights from parent and companion and smearing them out: brighter objects seem to reach further out on the detector. The diffuse structure near the edge of the bright parent is again introduced by the optics, and isn't a real feature.
The image itself is your typical false-color image. All the data are in the infrared, so for us to see it, a quick remapping has to be done. Rather than have red (0.65 microns), green (0.55 microns), and blue (0.45 microns) filters that are then reproduced by red, green, and blue pixels,
The preprint of the paper lists the parent brown dwarf as 2MASS J12073346-3932539 , which is indeed at the above coordinates. The candidate planet (much in the same way Ralph Nader is the candidate president, but there's my bias showing) will be 0.46 arcseconds south and 0.63 arcseconds east.
In case I didn't discourage any amateur astronomers thus far, here's some more: That's a separation of 0.77 arcseconds, when the seeing at most sites is of order 1 arcsecond. The companion is 100 times brighter than the parent brown dwarf in the K band. The parent brown dwarf has a K of about 12, and for an M8 spectral type, that's a V-magnitude of about 19 or 20. For those of you scoring at home, the parent brown dwarf is one million times fainter than anything you can see with the human eye.
The companion is an even redder object, so the colors will be much, much worse at V (there's a reason we try to detect these in the infrared). With a state-of-the-art AO system (look what we did with the same system earlier this year imaging the surface of Titan) on an 8 meter telescope with excellent infrared detectors, the companion lies one magnitude above the detection limit on their sensitivity/separation curves.
Sorry to depress people looking forward to pointing your telescope at this system tonight, but if it makes you feel better, it's probably not a planet.
I just checked that RA, by the way. It's behind the sun right now. You'll have to wait until January to observe it. Or to point your telescope there and not observe it, as the case may be.
Actually, this is an interesting case in that if it is a false alarm, the most likely explanation is that it's a foreground brown dwarf. The ESO team seems to have a noisy spectrum and JHK photometry of the companion. The colors really do suggest an L-dwarf (as opposed to a background K star, say), and the methane feature in the spectrum does seem real, assuming those data were reduced correctly (Difficult under the best conditions, not made easier by a brighter object 0.7 arcseconds away).
But you're absolutely right, until the proper motion study is confirmed, they've got nothing besides staking their claim to the object on the chance they get lucky. The TW Hydra association (to which the "parent" brown dwarf belongs) moves on the sky at about 66 milli-arcseconds a year, so it should be possible in a few years to see if the companion is following or not.
For me, the kicker is trusting the models. All determinations of mass (including of the parent brown dwarf) are done by guessing age, measuring luminosity and/or color, and seeing what the models of brown dwarfs/extrasolar planets say the mass should be. These are cooling tracks calibrated to Jupiter at 5 billion years, predicting how bright something is at 8 million years. Even the people making the models are hesitant to trust their own numbers at the youngest ages (Heck, we don't even know how long it takes to form a large Jupiter out of the disk).
Right now, I'd set the odds at about 80% that it won't be confirmed as a proper motion pair. In which case, my guess would be a foreground brown dwarf of a much larger mass (older and dimmer, but closer to still put it at 18th magnitude in H). If I'm wrong, the open question of these thoroughly untested models remains the cloud of uncertainty hanging over all of this.
It should be noted that while the initial discovery made with a 10 cm telescope set up relatively cheaply, the confirmation that it was in fact a planet came using the multi-million dollar Keck telescope. Transit detections have a notoriously bad record for turning out to be planets (a few percent are actually confirmed, the rest are some other effect), so you need follow-up with high resolution spectra. And for an 11th magnitude star, that requires a large telescope like Keck.
They could have gotten away with a 3-meter like at Lick Observatory, in this case, given the large radial velocity amplitude, but for further-out planets Keck is the only way to go. Trying to do this part of the science with more 10 cm telescopes would be out of the question.
The advantage to small telescopes is the ability to stare at a large section of the sky at once (in the case of the 10-cm telescopes used for this project, each exposure covers 6 degrees of sky). Compare this to the 10-meter Keck telescopes, whose imaging systems have fields of view of about an arcminute (1/60th of a degree). For transit searches, you want to keep staring at a star until you get lucky with a planet passing in front of the star, then confirm that as the transit happens again and again. So your best bet for optimizing your transit search is to look at a lot of stars at once, meaning a big field of view, which you get most easily from small telescopes.
It's not impossible to get large fields of view with large telescopes, but it takes a lot more effort. Check out the plans for building the 8.4 meter LSST for details.