Yup. Makes sense. And I of course won't pretend to know whether it's detectable, but part of me thinks that if they see the temporal shifts between planet transits, they'll be on the lookout for amplitude wobbles as well.

Also, though, as the planet/moon gets farther from the star, you might just get two or three separate transits.... what would Jupiter+Callisto or Saturn+Titan look like 200 light years away? Do they resolve as separate transits?

I'm not sure what you mean by "resolvable," Steve. It's not like we're able to resolve a star disk image with a transit visible as a blotted-out circle within it. But I would imagine that the "resolution" in this regard would be the same as Kepler's overall ability to detect a planetary transit. A moon would have to be the minimum size necessary for Kepler to detect it, all by itself, as a transit event. So if Kepler can't detect the dimming of a star's light caused by the transit of a Callisto-sized planet, it ought not be able to detect the additional dimming that would occur with a Callisto-sized moon as it would appear in the first frame of 'squid's excellent illustration. And also, therefore, ought not be able to tell the difference between the first and second frames.

As I understand it, Kepler can detect down to about an Earth-sized planet, correct? Then I would have to think that the smallest gas giant moon it might detect would have to be at least as large as the Earth.

Another very interesting thing, though -- any planetary body with a ring system will block more or less of a star's light depending on the angle the ring plane presents to the viewer. I can well imagine that some percentage of the planets Kepler will discover may indeed have ring systems, and that these ring systems may not always present the same angle to us here on Earth during every single transit. It will be very, very interesting to see how fast the investigators suspect they're seeing ring systems in some of their results...

-the other Doug

I think "resolultion" in this case refers to time. According to the mission website, Kepler samples a star's brightness every fifteen minutes. I don't know how long a typical Kepler transit will be but in the preliminary list of COROT candidates the transits ranged from an hour up to sixteen hours, with perhaps three hours being the norm. So if we assume a transit takes three hours, we have only about a dozen samples per transit. That is a pretty small sample size from which to try and weed out moon data, and it is a reasonable question whether it will be possible. It seems to me that, if we have a particularly long, slow transit (say 8 hours), an unusually large moon (approaching Earth sized or at least larger than Mars), and a bit of luck, it will be possible, but won't be obvious in the raw data. However, over time Kepler will obviously accumulate observations of multiple transits for each planet detected by Kepler, so that with rigorous analysis it might be possible even with less extreme examples. This is more believable after seeing how clean and noise-free the data were at the Aug. 6 news conference.

As for a ring: Doug, I hadn't thought of it, but it will obviously wreak havoc with density assumptions at first, as first contact of the rings will be difficult to differentiate from the planet itself, giving a grossly inflated diameter estimate. I imagine the difference will become apparent over time, as more samples are added to the data set, but in the meantime someone will publish a paper they'll have to retract.

Hmmh - it seems to me then that a number of the extremely low density exoplanets already discovered might actually have ring systems, it's just that we have (as yet) no mechanism (other than a very low apparent density) for making that case. I'm specifically thinking about such oddities as TrES-4 (1.67 Jupiter radii but 1/6th the density) although from the discussions I've seen so far the smart money on all of these appears to be on tidal heating.

Although this does also get me thinking that surely it would be nearly impossible for any large moon ( certainly anything earth sized ) or significant ring system to survive around any of the hot Jupiter class of planets? Tidal forces would be immense and since the numbers indicate these are probably enough to inflate the Hot Jupiters by 10-20% in any case then surely they would be more than enough to rip any significant moon apart and lead to environments far to chaotic for rings to form? Or is there a possibility for a class of "Trojan\Greek" style co-orbital dust clouds rather than rings that might be more stable in such an aggressive environment?

Yes, they would be able to detect the transit of a sufficiently large moon separate from the planet, regardless of if the two are transiting simultaneously, just so long as the planet doesn't eclipse or occult the moon at the time of transit (in which case, surely, it won't in at least some other transits).



Normally, yes, but with the transit light curve of a planet being scrutinsed, that photometric data gets much more attention. Photometric data containing evidence for a planet (one of the later OGLE planets) went un-noticed for quite some time. Kepler and others like it gather a lot of data, which isn't too easy to sift through easily and detect very minute transits. I hope they intensely scrutinize Kepler photometry around transits of medium and long-period planets in search of moons. Though I don't know what you could really say about them (other than their radius, with a significant error) without extensive transit-timing observations (Hey, Kepler might give those too).

Did you see the depth of the HAT-P-7b secondary transit in the raw data during the press release? It's apparently the same depth as an Earth-radius planet in transit. It shouldn't be much of a stretch to imagine a transit depth half of that (not half the radius of the planet, of course, but a planet whose disk has half the 'area').



This paper,

Transit Detectability of Ring Systems Around Extrasolar Giant Planets
http://arxiv.org/abs/astro-ph/0409506

discusses the transits of ringed planets, shows example light curves, and describes how scattering may allow for one to determine the size of particles in the rings.

Earths in both size and orbital period, meaning only three transits over the Kepler mission. With more transits, you can pull smaller planets out of the noise. The Kepler site has some information http://kepler.nasa.gov/sci/basis/sizes.html

So that I understand, are you asking about detecting various Trojan objects sharing the orbit of the 'Hot Jupiter' around the star ? Does anyone recall the maximum mass a Trojan object can have (seems like I recall mass ratios affecting Trojan stability) and does that overlap with what Kepler might be able to detect ?


Fascinating contemplating what might (and might not) turn up in this missions results.

In the usual formulation, the medium body must be at least 25x smaller than the large one, and the smaller body must have "negligible" mass. I'd be surprised if even a moon-sized object were stable even with a 10x Jupiter, but someone would probably have to simulate it numerically to know for sure.

--Greg

Well, I've read that for a distant observer, a Jupiter transit would have a duration of 30 hours, a Saturn transit about 40 hours...
Exo-moons could be detected with more accurate detectors, , a Planet-Moon system would have a characteristic transit timing variation, for instance a Jupiter-Europa system would have a variation in the order of 10 seconds, a Saturn-Titan system in the order of 30 seconds...
Ring system might be easier to detect

By "transit timing variation," are you referring to delays or advances in the transit time due to the planet's motion around a planet-moon barycenter? This is the method described in the discussion earlier in the thread. It would seem to me that, in order for such a system to be detectable with Kepler, the moon would need to be more massive, relative to the planet. We've found larger planets than Jupiter; it's only reasonable to assume that larger moons than Ganymede also exist. And such larger moons--if they indeed exist--could be detectable by their own transits across the star as well, I should think, especially if their presence was already suspected from the timing data.

Helvick's concern about the stability of moon orbits for hot Jupiters due to tides is notable, but again, if we tweak our hypothetical we might avoid it. Wouldn't there be more available stable orbits for a large moon of a planet in a 24-day orbit than of one in a 3-day orbit? As a bonus, the 24-day planet will have lower orbital speed and thus a longer transit, giving a longer sample of planet-moon interaction.

I haven't heard anyone talk about work on direct detection of moons transiting their planets. Any graduate students looking for a thesis idea?

One other possible source of noise occurs to me; it's also an opportunity for stellar studies, I suppose. If a planet crosses a large starspot, the variation in total light will be similar to its eclipsing a large moon. There are of course differences; the region surrounding a starspot is usually brighter than the average star surface, no? Has such an event been modeled? What would the resulting light curve look like?

Another possibility is stellar flares during a transit, which in fact have been detected:

A Stellar Flare during the Transit of the Extrasolar Planet OGLE-TR-10b



Bill

From reading the Kepler site, I think several of these questions have easy answers. The short summary is that it'll probably find Earth-sized moons in Europa-sized (or bigger) orbits but only if the primary planet has a one-year period AND if the moon orbits in the same plane as the planet (so that they both transit the star).

If the satellite is Earth-sized, then when it transits the star, it should show the same light-curve effect as any Earth-sized planet. Depending on how large the orbit is, that transit may or may not overlap the transit of its primary. If the big planet has a 1-year period, then Kepler should detect the satellite in much the same way as it detects any Earth-sized planet. If it has a longer period, then Kepler can't do it. If the satellite is only Ganymede-sized, then the primary would need to be much closer to the star, raising questions about whether satellites have stable orbits -- other than orbits SO close that Kepler can't see them.

Another point no one has mentioned is that this only works if the orbital plane of the satellite is close to the plane of the primary's orbit. Otherwise we'll miss the transit in all probability. That makes Jupiter a better bet than Saturn (other than the need for a 36-year mission, of course.)

Observing moons transiting planets (but not transiting stars) is clearly beyond Kepler's capabilities, since Kepler is barely able to measure the effect of reflected light from a giant planet in a one-week orbit. The light-change from full to crescent is a LOT bigger than the smaller change when a moon transits.

--Greg

Just pondering other possible 'yon wee beasties' Kepler may or may not detect, has anyone considered a light curve for a binary 'hot Jupiter' orbiting a star?

I'm not sure how stable a double planet might be in this regard, but if the twins orbited their barycenter in a multiple of their period around the host star maybe we might have a stable situation ?

No one can say for 100% certain that its impossible, but none have been identified yet. It's quite possible that some of the non-transiting Jovian planets known today are actually double planets, we wouldn't know with just radial velocity alone.

If a double planet transited, it would be pretty much the same dynamics as a planet + moon transit, but the transits would be more similar to each other.

This is the three-body problem again, of course. Someone would probably need to do some heavy-duty numerical simulations to find out whether any such combination is "stable" in the sense of "neither body gets thrown into a different orbit nor do they collide." My guess is that they'll eventually collide unless they are very far from their star.

If they orbit far enough apart, Kepler certainly out to see a two-step phase curve -- even if the pair really were far from the star and Kepler only got to witness a single transit. Of course, a single transit wouldn't eliminate the possibility that two planets in different orbits just happened to transit at the same time. Still, in a few years, we'll have a much better idea whether such things do exist in any numbers.

--Greg