A SEARCH FOR LOST PLANETS IN THE KEPLER MULTI-PLANET SYSTEMS AND THE DISCOVERY OF A LONG PERIOD, NEPTUNE-SIZED EXOPLANET KEPLER-150 F

Abstract

The vast majority of the 4700 confirmed planets and planet candidates discovered by the Kepler space telescope were first found by the Kepler pipeline. In the pipeline, after a transit signal is found, all data points associated with those transits are removed, creating a "Swiss cheese"-like light curve full of holes, which is then used for subsequent transit searches. These holes could render an additional planet undetectable (or "lost"). We examine a sample of 114 stars with 3+ confirmed planets to see the effect that this "Swiss cheesing" may have. A simulation determined that the probability that a transiting planet is lost due to the transit masking is low, but non-neglible, reaching a plateau at ~3.3% lost in the period range of P = 400 - 500 days. We then model the transits in all quarters of each star and subtract out the transit signals, restoring the in-transit data points, and use the Kepler pipeline to search the transit-subtracted (i.e., transit-cleaned) light curves. However, the pipeline did not discover any credible new transit signals. This demonstrates the validity and robustness of the Kepler pipeline's choice to use transit masking over transit subtraction. However, a follow-up visual search through all the transit-subtracted data, which allows for easier visual identification of new transits, revealed the existence of a new, Neptune-sized exoplanet. Kepler-150 f (P = 637.2 days, R_P = 3.86 R_⊕) is confirmed using a combination of false positive probability analysis, transit duration analysis, and the planet multiplicity argument.

Keywords: planetary systems; planets and satellites: detection.

Figure 1

The probability of a planet that was originally detectable (i.e., had 3+ detectable transits) that became undetectable (i.e., had < 3 detectable transits) after the in-transit data points of the successfully fit KOIs were removed. A transit was ruled “detectable” if at least 50% of it was contained within the data (i.e., < 50% in a data gap). The probability of losing a planet at a certain period averaged over all stars is highlighted in black, while the star-by-star probabilities are shown in a transparent gray, so that darker areas correspond to higher density regions.

Figure 2

Thirty day portion of the light curve for Kepler-18 (KIC 8644288). Top: The PDCSAP flux, which was our starting point for the analysis. Middle: The PDCSAP flux detrended for variability. Bottom: The detrended flux after removing the transit signals from the KOIs in Kepler-18. Note that the y-scaling changes in each panel.

Figure 3

Phase-folded light curve for all three CPs in Kepler-253. Top: After detrending, but before transit subtraction. Bottom: After detrending and transit subtraction. Data points are transparent to emphasize the transit shape.

Figure 4

Phase-folded, flux-normalized light curve for Kepler-150 f. Transits by other planets in this window were modeled and subtracted out. Blue circles show the first transit, while red squares show the second. The black line is the best fit from TAP.

Figure 5

Potential single transit in the light curve of Kepler-208 at 786.7641 BKJD. Due to its likely nature as a background eclipsing binary FP, we did not attempt a transit fit.