Our paper entitled “Single Transit Candidates from K2: Detection and Period Estimation” is out on arXiv today (and under review at MNRAS)! Here is a brief overview:
It’s now more than a year since the Kepler space telescope renewed its search for transiting exoplanets. It has spent that time looking at a series of parts of the sky for 80 days at a time, substantially shorter than the continuous 4-year view of the original mission. However, the number of planets it should be able to spot as they cross their star remains the same. So, with K2’s limited campaigns, it should see many long-period planets that transit only once.
In fact, using a couple of different estimate for how many planets are out there, we expect to see a detectable single transit around every few thousand stars in K2. Given that it looks at nearly 20,000 at a time, there should be a lot out there to find!
But without seeing a second or third transit, there would seem no way to estimate the orbit of the planet. However, with such good in-transit data from Kepler and knowledge of the planet’s parent star from it’s colours, we can begin to estimate such planet’s orbits. To do this, I developed the transit modelling code (Namaste: An Mcmc Analysis of Single Transiting Exoplanets), which estimates the planetary velocity across the star, and uses Kepler’s laws to turn this into an orbital period. To test this code, we tried it out on six known Kepler planets, which gave extremely promising results, with orbital period estimates often within 10%!
We then had to go hunting for them in K2!To do this we ran a transit model along the curve, storing wherever it found a good. We also checked the resulting transits by eye, and found a few more it inexplicably missed along the way. The results of that were the seven dips you see below; our best single transit candidates.
Each of these dips could be caused by a planet crossing in front of it’s star. Alternatively, something else could be cause the signal, such as the eclipse of a star in front of another star. For three of these transits, we think this scenario is more likely, either because the radius given by Namaste is too big to be a planet, or because the period estimated is just too long. For example, EPIC203914123 gave a whopping great period of 200 years! As the probability of seeing a transit scales with distance, we are very unlikely to be seeing such a planet in transit; instead this is much more likely to be a small star crossing a giant star on a shorter orbit
Three more of these single transits are ambiguous – they could come either from binaries or planets. The uppermost signal, however, is exceptionally well-constrained and very likely to be a planet. We estimate that this object, which crosses the 11th magnitude star EPIC203311200, has a circular period of 540 +410/-230 days and a radius 0.51±0.05 times that of Jupiter. If it turns out to be planetary (and I’m speculating here), this would make it a mid-sized gas giant placed at the outer edges of its star’s Habitable Zone. And while gaseous planets are inhospitable to life, the possibility of habitable or even inhabited moons around EPIC203311200b cannot be ruled out.
As with all planet candidates, however, we need to do more work to confirm this hypothesis. This includes weighing the planet with radial velocity observations and searching for contaminating stars with high-resolution images of the star.
But with many similar missions to K2 (for example, TESS and PLATO) on the horizon, finding and analysing the single transits of long-period exoplanets is likely to become extremely important. It could even let us find planets that can be confirmed by the Gaia mission or future direct imaging. Alternatively, such planets could let us use JWST to sample the as-yet unstudied atmospheres of warm- and cold-jupiters for the first time.
I should point out that this paper is under peer review and may change somewhat on publication. We decided to put on arxiv for a few reasons; the first being that, as evidenced by two candidate papers last week, K2 papers just seem to be going that way. The second being that the excellent Planet Hunters team released a similar technique for single transiting Kepler planets last week, and it felt logical to try to release them close together.
I will also be releasing the transit analysis code Namaste on my github this week (it needs a few tweaks and additions before it’s useable by anyone else unfortunately).