Exomoon Candidate Profile for KID 6867155

 Earth size Exomoon orbiting a Saturn- like planet - Wikipedia

If an exomoon is going to be observed in the Kepler light curves it will have to be larger than any moons seen in our solar system. KID 6867155 is a known Kepler planet candidate that is showing some exomoon characteristics in its second of three observed transits due to transit timing variations observed and cues of a possible exomoon in the second transit drop. We may be seeing the indicators of an actual exomoon orbiting KID 6867155 or evidence of a second unknown planet orbiting the star.

Forum Talk member planetsam:

" Here's KID 6867155, SPH10024957 , a Kepler Favourite. It's listed as a 7.90 Re planet with a period of 235.9742432. We have 3 transits in the data so far. Detrending the SAP and taking the centre of each I get the period of Transit 1-Transit 2 as 235.9656 and Transit 2-Transit 3 as 235.9819. This yields a TTV of approximately 11.7 minutes. The first transit appears slightly distorted based on the floor so I used an additional light detrend of it to compensate, and also treated Transit 2 differently (see below for the reason). Without the improvement of the first transit you still get TTVs of roughly about 10 minutes so it seems real and not just me imagining it though the exact value isn't certain."

" Interesting enough, but some other things caught my eye. For a start, if you look at Transit 2 there is a clear ramp extending beyond the egress. The next thing you notice is that Transit 2's floor is visibly longer in duration (~20 minutes longer than T3's, ~50 minutes longer than T1's, though again that is a little distorted) and flatter than the other transits. I have tried a variety of different ways of cleaning up the transits and using different measures, but I still get the same picture."

" The presence of possible floor duration change coupled with TTVs made me wonder if the impact parameter was changing. If the floor duration change is visible, it might be possible to see the expected correlated total transit duration variation. The problem is this is much more model-fitting dependent than measuring the floor and I think any variation is at or beyond the limit of accuracy. However from the measurements I could get the total duration may show variations of the same order that are out of phase with the floor duration variation. That's not what you would expect from simple planet-planet induced inclination change, but that is however what you'd expect from the effects of an exomoon. Indeed I have a hunch that part of the long duration and flatness of the second transits's floor is because we are actually seeing a relatively large moon transit, re: the egress ramp. Although this cartoon below is exaggerated you can see that a moon that is small compared to the planet (and thus smoother transitions than in this case) transiting with said planet could have the effect of smoothing out and appearing to lengthen the transit floor."

" Which brings me to why I was looking at this system in the first place. I have been reading all the exomoon related papers and came up with some simple parameter values that improved the chances firstly of a system where the planets could realistically harbour large moons (i.e the planets likely have large hill spheres), and cross referenced those with parameters that maximized the ability to detect them both through pure photometry and TTVs. When I applied these constraints to the Kepler Favourites, this was in the list of systems which are most favourable. No doubt I'm barking up the wrong tree again, but I thought I'd share just in case. In any case we will have some SC lightcurves of this star soon which will be interesting."

Forum Talk member Kian Jek:
  

" Well done Sam and this is very interesting indeed! It looks like there might be something here. It is hard to tell because the determination of the transit profiles is tricky with just 3 of them, and an accurate TTV computation requires good approximations of the start and end of transit and the exact mid-point- tl;dr. I've redone the computations you made and I see the presence of TDVs, transit duration variations, which seem significant, and may indicate the presence of a large exomoon, one as large as 1x RE(!) based on the unusual profile of transit 2. But first, here's what I have of the 3 transits fitted with polynomials and the depths, durations and mid-points determined."

" In the image above, the green lines denote the start and end of transits, I used the zero-crossing to make that determination. The differences in the transit durations seem quite large."

" Transit #2 is intriguing because of the apparent 'lip' at the egress that could be the signature of another co-transiting body. The problem with this transit is the missing cadence at the ingress, which could possibly throw any estimates off (this includes the spline interpolation/polynomial fit). What I have done is to simply interpolate the missing value by taking the average of the two on either side, but since this is at the start of the transit, this is unfortunate.

However the transit depths appear to be decreasing, whether this is due to changing inclination (the impact parameter for these transits is very high) or just measurement error is hard to tell, and this could affect the determination of the 'extra' depth in transit #2 due to the putative third body. The radius of the host star has been adjusted down to 0.48x Sol and if that's the case, then the 450ppm dip could be accounted for by an exomoon that is estimated to be around 1.1x RE. This seems a little large for an exomoon but anything is possible. The magnitude of this dip underscores how difficult it is to really discern exomoons that may be much smaller. Finally, from the transit profiles I tried to replicate the O-C plots that Sam made."

" The problem here is that there are 2 ways to determine the minima, using the Kwee-vanWoerden method and a polynomial fit. I used both and they are plotted above. I used the average period as the basis for the C period. The KvW method gives a TTV for the 2nd transit of -6.47 minutes, while the polynomial method gives -7.97 and -7.87 minutes. Unfortunately the error bars for the polynomial fits are just too large. We definitely need to see a few more transits in order to get a better fix on things. But this is a great start and absolutely wonderful if it is true. Goes to show that the human eye may be better than supercomputers in finding exotic phenomena of this sort."

You can follow this discussion here.