Kepler Object of Interest (KOI) Catalog - Q1–Q17 DR24

First delivery date: 2014 Apr 1
Last delivery date: 2015 Sept 24
Status: Done


  1. Summary of Q1–Q17 DR24 KOI Activity Table
  2. Q1–Q17 DR24 TCEs
  3. Vetting of Q1–Q17 DR24 TCEs
  4. Planetary Parameters
  5. Stellar Parameters

1. Summary of the Q1–Q17 DR24 KOI Activity Table

The Q1–Q17 Data Release 24 (DR24) Kepler light curves were searched for periodic flux decrements consistent with the signals produced by transiting exoplanets, known as Threshold Crossing Events (TCEs). These TCEs were then evaluated by the Threshold Crossing Event Review Team (TCERT) using the Robovetter, which is an automated vetting program designed to mimic the human decision-making process employed for previous catalogs. Hence, the Robovetter first determines if a TCE is transit-like or not. If it is transit-like, the TCE is assigned a KOI number and shown in the Q1–Q17 DR24 KOI activity table. Each KOI is further searched for evidence of a secondary eclipse, which indicates that the target is an eclipsing binary, as well as evidence of a centroid offset or ephemeris match, which indicate that the transit-like signal does not originate from the target. For each KOI, a disposition of CANDIDATE or FALSE POSITIVE is given, as well as a number of major and minor flags to document the Robovetter's decision-making process. The catalog uses the stellar parameters published by Huber et al. (2014), along with a subset of updated stellar parameters (Huber 2014), for the observed Kepler targets. For each KOI, the Q1–Q17 Kepler flux time series data are fit by a Mandel & Agol (2002) transit model, and a Markov Chain Monte Carlo (MCMC) routine is used to obtain accurate uncertainties on the fitted parameters (see Rowe et al. 2014 and Rowe and Thompson 2015).

Note that only KOIs that were detected as TCEs in the Q1–Q17 DR24 run of the Kepler pipeline are included in the Q1–Q17 DR24 KOI table. If a previously known KOI did not produce a TCE in the Q1–Q17 DR24 Kepler pipeline, it is unable to be dispositioned by the Q1–Q17 DR24 robovetter, and thus it is not included in the Q1–Q17 DR24 KOI table.

More details about this catalog, the vetting procedures, and the candidates it contains are available in Coughlin et al. (2015).

2. Q1–Q17 DR24 TCEs

The Kepler pipeline (release SOC 9.2) was run on the Q1–Q17 DR24 light curves to search for TCEs, i.e., periodic flux decrements consistent with the signals produced by transiting exoplanets (Seader et al. 2015). All TCEs found by this pipeline run are available in the Q1–Q17 DR24 TCE table. For the Q1–Q17 DR24 pipeline run, a total of 20,367 TCEs were generated. As can be seen in the figure below, a majority of these TCEs are not valid planet candidates — there is a large excess of not transit-like TCEs at both short and long periods due to astrophysical variability (e.g., rapid rotators with strong starspots, eclipsing binaries, pulsating stars) as well as uncorrected instrumental artifacts (e.g., rolling-band noise, cosmic rays, flares, edge effects). Narrow spikes in the period distribution of TCEs result from flux contamination by bright variable stars (e.g., RR Lyrae, V2083 Cyg, BR Cyg), as well as systematic events common to multiple detectors.

Plot of Q1-Q17 DR24 TCEs, KOIs, and PCs

In previous catalogs, various sets of eclipsing binaries were excluded from the search for signatures of transiting planets. In Q1–Q17 DR24, only sinusoidal, contact eclipsing binaries were excluded from the search, and thus many new TCEs and KOIs are generated due to detached eclipsing binaries that were excluded from previous searches.

Note that the SOC 9.2 search pipeline did not recover many previously identified long-period TCEs because of a subtle flaw in the newly incorporated "statistical bootstrap test", which attempted to adjust each TCE's detection threshold to account for remaining systematics in the light curve (see appendix A of Seader et al. 2015). While this test eliminated many TCEs due to artifacts, it also eliminated a significant number of valid long-period, transit-like signals, especially at low signal-to-noise, which includes previously designated Earth-size, habitable zone planet candidates from the past two catalogs (Rowe et al. 2015; Mullally et al. 2015). The implementation of this bootstrap test has been corrected for future runs. The implications for occurrence rate calculations are discussed by Christiansen (2015).

3. Vetting of Q1–Q17 DR24 TCEs

In Q1–Q17 DR24 the process of vetting TCEs has been fully automated via a series of simple decision trees, collectively called the Robovetter, which attempts to codify the rules developed by the human vetters involved in TCERT, while imposing strict uniformity in their application. Every Q1–Q17 DR24 TCE is dispositioned as either a planetary CANDIDATE (PC) or FALSE POSITIVE (FP), with specific reasons given for any FP TCE. These reasons are presented as a collection of major and minor flags. The four major flags, available as individual columns in the KOI table, are:
  • Not Transit-Like: a TCE whose light curve is not consistent with that of a transiting planet or eclipsing binary, such as instrumental artifacts and non-eclipsing variable stars.
  • Significant Secondary: a TCE whose light curve is observed to have a significant secondary eclipse. Typically, these objects are eclipsing binaries, but a small subset may be due to planet occultations, so this flag can be set and still permit a disposition of CANDIDATE, particularly for large planets with short periods.
  • Centroid Offset: a TCE whose signal is observed to originate on a nearby star, rather than the target star, based on examination of the pixel-level data.
  • Ephemeris Match Indicates Contamination: a TCE that has the same period and epoch as another object, and is not the true source of the signal given their relative magnitudes, locations, and signal amplitudes (see Coughlin et al. 2014).

Minor flags provide more specific information on decisions and can be found in the KOI table's comment field, with the meaning of each minor flag described on the data column description page.

The TCEs from the Q1–Q17 DR24 search were federated with pre-existing KOIs, and their latest dispositions are given in the Q1–Q17 DR24 KOI table. Transit-like TCEs that did not federate with any previously designated KOI were assigned new KOI numbers and their dispositions are given in the Q1–Q17 DR24 KOI table. Pre-existing KOIs that did not federate with Q1–Q17 DR24 TCEs are not shown in the Q1–Q17 DR24 KOI table. Note that some new KOIs were designated while the Robovetter was actively being developed, and thus some new KOIs are ultimately dispositioned as not transit-like FPs. Also note that any Q1–Q17 DR24 TCE that does not have a corresponding KOI in the Q1–Q17 DR24 KOI table can be assumed to be dispositioned as a not transit-like FP.

4. Planetary Parameters

Planetary parameters for the majority of KOIs are based upon fitting the Mandel & Agol (2002) transit model to the Q1–Q17 Kepler flux time series data. The model was fit to the PDC light curves available at the MAST at the time they were fit. For those KOIs that were designated prior to the start of the Q1–Q17 DR24 vetting activity, these light curves correspond to Data Releases 21–23. For those KOIs designated as a result of the Q1–Q17 DR24 activity, DR24 PDC light curves were utilized. For those KOIs newly designated as a result of the Q1–Q17 DR24 activity, DR24 PDC light curves were utilized.

The fitting procedure is the same as that described in Rowe et al. (2013). To summarize, the procedure assumes a circular orbit and fits for the stellar density, impact parameter, radius ratio, period, and epoch. The stellar parameters (see below) and associated uncertainties are used to calculate values such as planetary radius and insolation flux. To best estimate the posterior distribution on each fitted parameter, a MCMC approach was used to account for strong correlations between variables (especially stellar density, impact parameter, and radius ratio). The error bars reported for these values encompass 68% of the posterior distribution, emulating one-sigma error bars. Note that the parameters given in the table are based on the best-fit model, not the median value of the MCMC chains (see Rowe and Thompson 2015).

For some very low signal-to-noise false positives, the above fitting procedure was unsuccessful. In these cases the catalog contains the period, epoch, and transit duration, with no error bars, as reported by the original DV report for each TCE. These cases are identified by the column "Planetary Fit Type" being set to "none".

5. Stellar Parameters

Understanding the stellar population is required in order to understand the planet population found in this transit survey. By combining the works of many published sources, Huber et al. (2014) characterized the entire sample of stars observed by Kepler. This stellar catalog was utilized by the Q1–Q16 KOI catalog. An update to a subset of the stars was subsequently performed to create the Q1–Q17 DR24 stellar table (Huber 2014).

KOIs in the Q1–Q17 DR24 catalog that were designated prior to the start of the Q1–Q17 DR24 vetting activity were fit using stellar values from the Huber et al. (2014) Q1–Q16 stellar table. KOIs in the Q1–Q17 DR24 catalog that are newly designated use stellar values from the updated Q1–Q17 DR24 stellar catalog (Huber 2014). These stellar parameters can be accessed via the Kepler stellar data search interface. A few tens of KOIs have no published stellar information; for these, solar parameters are reported in the KOI table. The "Stellar Parameter Provenance" column will have a value of "q1_q17_stellar", "q1_q16_stellar", or "solar" if the stellar parameters are from the Q1–Q17 DR24 stellar catalog, the Q1–Q16 stellar catalog, or defaulted to solar values, respectively.