The Moon Kepler-1625bi
The BBC Science website had an article recently about the possible discovery of an exomoon, or a moon orbiting a planet in another solar system. This discovery was only provisional, but it was intriguing. The picture is from their website story (note that it is an imagined picture, not an actual picture of the system in question).
The full paper is (link at the end of the blog):
HEK VI: ON THE DEARTH OF GALILEAN ANALOGS IN KEPLER,
AND THE EXOMOON CANDIDATE KEPLER-1625B I
A. Teachey, D. M. Kipping & A. R. Schmitt
It appears that a Neptune sized moon might be orbiting Jupiter sized planet, which is itself orbiting fairly close to its parent star. The star in question is one that the Kepler satellite (whose mission is to search for extra-solar planets) catalogue refers to as Kepler-1625, with the second planet Kepler-1625b seeming to be home to the moon Kepler-1625-b-I. Assuming the “sighting” (really a complicated data analysis of light curves) holds up, it is quite likely that the moon could be a captured planet, somewhat as Triton appears to have been captured by Neptune, in our solar system.
The planet in question will be transiting its star in October 2017, and time has been scheduled on the Hubble Space Telescope to observe it, so we should know a lot more by the end of the year.
The Method of Discovery
Some explanation of how this study (which was part of a larger scale exomoon search) was done, is in order. Basically, the light from stars was observed and recorded by the orbiting telescope (Kepler). As planets pass in front of their stars (as seen from the vantage point of Earth), the level of light drops slightly, and that is seen in resulting graphs of the light, known as light curves. That method, by the way, has probably been the major one for discovering exoplanets, so far.
If a planet also has a sizable moon, that dip in the light curve will have a secondary dip, caused by the moon blocking some of the star’s light as well. Obviously this will only be seen when the moon is neither behind, nor in front of the planet, as seen from Earth i.e. when the moon is on either side of the planet, but still in front of the star. The actual “fingerprint” of the moon in the light curve will depend on the geometry of the situation – the period of the moon around the planet, the planet around the star, the orientation of the bodies as seen from Earth, and so forth. The graphs of Kepler-1625b, from the paper, give an idea of what this looks like.
There are a couple of snags. One is that the instrumentation is not yet sensitive enough to easily detect these very small reductions in light from stars. There is a certain amount of noise in this data that can be confusing, and than can look like a planetary or planet-moon signal, even though one isn’t really there. That can be caused by the instrument itself (i.e. electronic noise, such as hot pixels or cold pixels), or by some other real astrophysical effect (e.g. a dark spot on the star itself, rather like a sunspot on our local star, can appear to be a planet or a planet-moon if it persists for a long time).
The other snag is that Kepler has only been observing for a few years, and it can take many revolutions of a planet around its star before a signal can be considered confirmed, with a high level of confidence. Even more so, for a possible moon, as the signal is that much weaker. But, unless the planet is very close to its star, collecting data from multiple orbits can take years. Since the data is generally analysed quite a while after the observations were taken, critical data is probably not going to exist – the star may well have moved out of Kepler’s field of view by then, not to mention that Kepler could be shut down before the interesting data is discovered.
To get around this problem, the researchers decided not to look for individual planet-moon signals, but rather to examine a large ensemble of likely planet-moon systems, and create a “grand light curve”, containing the information from all of these systems, by stacking the data. That could then be compared to synthetic light curves of these same systems, created by simulating the light curve of each planet, assuming that they had moon systems similar to those in our solar system (basically the Galilean moons of Jupiter or a single large moon), and combining them all.
Various synthetic grand light curves could be created, with varying proportions of planets being given moons. One run might have most systems having moons, while another might only have moons in a small proportion of systems. Many simulated grand light curves could be compared to the real data, and the correlations between the two could be compared. When the simulated data gave a good fit to the real data, a case could be made that this is good evidence that the real systems were created by the same system as the artificial data – i.e. that proportion of systems that had moons in the simulated data was similar to the proportion in the actual data, that had moons.
Of course this explanation leaves out a lot of complicated steps, such as finding the systems most likely to show a planet-moon signal, cleaning up the data, normalizing the data so that the various systems could be stacked together, and so on. Producing the synthetic data obviously also was complicated. All of these tasks were also time consuming both in terms of human time and computer time.
The actual paper, in pre-print form, at the arXiv site, goes into great detail about the data analysis. This method took years for the researchers to come up with and refine, and it is therefore very complex. So, as a mere practicing “applied” statistician (though one with a physics degree, and numerous courses in statistical analysis including time series analysis) I will just say it looks like a fascinating data analysis and a very intricate one.
The Larger Picture of Exomoons
The paper actually is actually much larger in scope than Kepler-1625bi. Its purpose was to estimate the proportion of planetary systems that had moons, and therefore give us a sense of just how many exomoons are out there. The complex statistical analysis outlined above was the method used to come up with these estimates.
First, some background on just which planetary systems they focused on. The graph below, taken from their paper, shows the planets selected for the study for further analysis (those in the coloured circles). The planets (strictly speaking Kepler Objects of Interest), vary from an orbit of about 20 days to 400 days, and from a bit over 1 Earth radius to about 12 Earth Radii in size. In solar system terms, that would be from around Mercury to a little past Earth, in terms of distance from their star, and from Earth size to gas giant size, in terms of planet size. The researchers had to confine themselves to this category of planet, since only close-in planets will have many (or any) transits of their stars within the relatively short window of time that Kepler can observe them.
The conclusion of the study, was that for the 284 planets that made the cut (had high quality data and were otherwise viable moon hosting systems), the maximum percentage of them that had moons of this sort was about 38% (with a 95% confidence interval, a statistical measure of how reliable the result probably is). That’s an upper limit - their “likely” estimate of the number of these planets that had moons was about 16%.
Given that all of the large planets in our solar system host a family of moons, this relatively low estimate of the proportion of exoplanets with moons came as something of a surprise. Of course, these exoplanets were “warm Jupiter” type planets, near their hosting stars, whereas the gas giants in our solar system are far from the sun. So, this evidence might suggest that large planets near a star are not likely to host large moons (only about one in six).
Perhaps migrating close to their host star is destabilizing, and these planets lose their moons in the process. The data also showed some tentative evidence that Io type moons might be favored (those are moons very close to their host planet, the way Io is close to Jupiter), so perhaps migrating “strips off” those moons that are orbiting further from their planets.
The paper does some interesting (though again very preliminary) comparisons of what kinds of planets are more or less likely to have moons, based on their analysis of the 284 planetary systems (remember that all of these are large planets close to their host stars):
· Larger planets are somewhat more likely to have moons than smaller planets.
· Colder planets are somewhat more likely to have moons than hotter planets.
· There seems to be little difference between cooler stars and hotter stars, though cooler stars (thus older) might be slightly favored.
· Inner planets are more likely to have moons that outer planets.
· There appears to be little difference between single planet systems and multiple planet systems.
· Habitable zone planets are more likely to hold moons than non-habitable zone planets. Note, though, that there aren’t very many habitable-zone planets in the data, so that result might not be very robust.
With some further statistical analysis, Kepler-1625-b-I was identified as an interesting outlier in the dataset. Some comparisons of its three transits were then done with simulated data, and the speculations about the Jupiter sized planet with a Neptune sized moon came from that additional analysis. It should be noted that we have nothing like that situation in our solar system (the largest moon is still much smaller than Earth, let alone Neptune), but evolving observations have surprised us many times in the recent past, so it would be folly to rule this out.
Also, the point about habitable vs non-habitable zone planets is interesting. In this case, the planets being examined are mostly too close to their stars to be in the habitable zone (where water can be a liquid). But some planets are in the “close-in” habitable zone. Could a large, rocky moon of such a planet hold life? One is inclined to say “why not?”, or at any rate “it’s an interesting subject for speculation”. Perhaps the moons Endor in Star Wars or Andoria in Star Trek will turn out to have real-universe counterparts.
And then there’s the SF planet Kordea, featured in some books published by Dodecahedron Books, the plucky little company that brings you these wonderful blogs …
SF series, The Witches’ Stones:
Helena Puumala's SF Romance series features the planet Kordea, home to a race of beautiful and powerful psychic aliens, known as the Witches of Kordea. The planet has seven moons, an extraordinary arrangement for a terrestrial sized planet in its star's habitable zone, as is noted in Book 1, which you can get from the link below: :).
In fact, the moons of Kordea become a central element in Book 2. A terrestrial planet with seven moons would be cool (though it would probably be a very unstable arrangement).
The Witches' Stones, Book 1 - Rescue from the Planet of the Amartos
Young Earth woman and spaceship mechanic, Sarah Mackenzie, has unwittingly triggered a vast source of energy, the Witches' Stones, via her psychic abilities, of which she was unaware. She becomes the focal point of a desperate contest between the authoritarian galactic power, known as The Organization, and the democratic Earth-based galactic power, known as The Terran Confederation. The Organization wants to capture her, and utilize her powers to create a super-weapon; the Terra Confederation wants to prevent that at all costs. The mysterious psychic aliens, the Witches of Kordea also become involved, as they see her as a possible threat, or a possible ally, for the safety of their own world.
A small but fast scout-ship, with its pilot and an agent of the Terra Confederation, Coryn Leigh, are sent to rescue her from a distant planet at the very edge of the galaxy, near space claimed by The Organization. Battles, physical and mental, whirl around the young woman, as the agent and pilot strive at all costs to keep her from the clutches of the Organization.
The Witches' Stones, Book 2 - Love and Intrigue, Under the Seven Moons of Kordea
Sarah has taken refuge on the planet of Kordea, where she is also learning how to control her psychic abilities, through the tutelage of the Witches of Kordea. Coryn Leigh has now taken up the position of Confederation diplomat to the Kordeans, but he is also charged with keeping the Mackenzie girl safe at all costs. During their time on the planet, an attraction between them grows, though they try to deny it, to themselves and each other.
But The Organization has plans of its own, including threatening the destruction of the planet Kordea, via destabalizing the orbit of Lina, one of its many moons. The Organization proves that its threats are in deadly earnest, so, ultimately Sarah, Coryn and the Witches of Kordea must take the fight to the enemy. Thus is borne a dangerous mission, to a planet where their foe has based the weapon that threatens Kordea, and ultimately, the balance of power throughout the galaxy. Sarah and Coryn agree that the machine must be destroyed, even at the possible cost of their own lives and growing love.
The Witches' Stones, Book 3 - Revenge of the Catspaw
Sarah and Coryn have become married, under the traditions of the Witches of Kordea. But the marriage is performed by the Eldest of the most important coven, a rare honour, that comes with a blessing and a curse. The slow working out of this blessing and curse forms the backdrop to the story.
Having come so close to their goal of enhancing their weaponry via Witches' Stone power, The Organization will not give up. In order to lure Sarah into their trap, and thus have her become their Catspaw (someone who is forced into helping another, against their will) they need bait, and Coryn becomes the bait. He also comes under the domination of a particularly nasty Elite of The Organization, one "Evil Evilla" Copoz.
Sarah, and a picked group of companions, must re-enter The Organization space, this time to the very heart of the empire, to rescue her husband, as he has done for her in the past. They do so at great peril, but nothing can stop the terrible Revenge of the Catspaw.