Scaling K2 Data Products:
Zink et al. 2021 provides the first full K2 transiting exoplanet sample, using photometry from Campaigns 1-8 and 10-18, derived through a fully automated procedure. To encourage transparency and to benefit the community, I provide the following data products:
Citations:
If you make use of any of these products, please be sure to cite Zink et al. (2020; 2021) accordingly.
Software Pipeline:
Our automated pipeline relies on three major software components:
  • EVEREST - A pixel level decorrelation algorithm used to address systematics introduced by the roll of the space craft
  • TERRA - A box least-squares algorithm which detects transit signals in the processed light curves
  • EDI-Vetter - A suite of vetting metrics, built for K2, with the ability to automate false alarm and planet candidate parsing.
  • ________________________________________________________ Masking: Within our automated routine, we apply campaign specific masking. Some problematic cadences produce a surplus of threshold-crossing events, which are likely systematic in nature. We ignore these data points in the pursuit of a pure planet catalog. For more information on this process see Section 2.4.1 of Zink et al. (2020). We provide our list of masked cadences in CSV format for each campaign below:
  • Campaign 1
  • Campaign 2
  • Campaign 3
  • Campaign 4
  • Campaign 5
  • Campaign 6
  • Campaign 7
  • Campaign 8
  • Campaign 10 (part 2)
  • Campaign 11 (part 1)
  • Campaign 11 (part 2)
  • Campaign 12
  • Campaign 13
  • Campaign 14
  • Campaign 15
  • Campaign 16
  • Campaign 17
  • Campaign 18
  • Note: We only use the second part of Campaign 10. In testing we found the inclusion of the first part introduced more noise than value.
    Planet Sample:
    Using our fully automated pipeline, we achieve a sample of K2 transiting planets suitable for demographics. Within this catalog we find:
  • 808 Transit signals
  • 747 Unique planet candidates
  • 51 Multi-planet systems
  • Of the 808 transiting signals identified, 51 signals are detected in multiple, overlapping, campaigns. In the following ASCII file we provide the our best fit planet and stellar parameters for this catalog.
  • Scaling K2 Homogeneous Planet Sample
  • Note: If you make use of the stellar parameters, be sure to cite Hardegree-Ullman et al. (2020) accordingly. ________________________________________________________ Light Curves: For visual inspection and further analysis we provide visualizations of each candidate light curve and the fully detrended photometry.
  • Light Curve Visualizations
  • The format for each file name is as follows: Epic-ID_The-First-Few-Signficant-Digits-of-the-Candidate-Period.pdf i.e., EPIC 220663602 with a candidate at 6.41 days would be: 220663602_06.41.pdf
  • Light Curve Data
  • The target EPIC was used as an identified for each file. The CSV files contain the BDJ-2454833 time (time [days]), the normalized and detrended flux (flux), the flux median absolution deviation (e_flux), and the cadences which were masked (mask).
    Sample Completeness:
    Any catalog of planets will inherit some selection effects due to the methodology of detection, limitations of the instrument, and stellar noise. These biases will effect the sample completeness and must be accounted for when conducting a demographic analysis. The selection of transiting planets can be addressed using analytic arguments, but the instrument and stellar noise contributions to the sample completeness are dependent on the stellar sample and the specifics of the instrument. With an automated detection pipeline this detection efficiency mapping can be achieved through the implementation of an injection/recovery test. Here, artificial signals are injected into the raw photometry and run through the automated software to test the pipelines recovery capabilities, directly measuring the impact of instrument and stellar noise on the catalog. Here we provide the necessary data products to make such measurements. ________________________________________________________ Light Curve Noise Measure: To assess the noise properties within each detreneded light curve, we measure the combined differential photometric precision (CDPP; Christiansen et al. 2012) for transit duration of 1, 1.5, 2, 2.5, 3, 4, 5, 6, 7, 8, 9, and 10 hours. These values enable users to analytically calculate the injected signal strength. The corresponding values are provide in an ASCII file.
  • CDPP Values
  • ________________________________________________________ Injection Test Summary: The injected signal parameters and our pipeline's designation is provided in the following CSV files. We provide flags indicating if the signals was identified by TERRA (found_unVet) and EDI-Vetter (found_Vet).
  • Campaign 1
  • Campaign 2
  • Campaign 3
  • Campaign 4
  • Campaign 5
  • Campaign 6
  • Campaign 7
  • Campaign 8
  • Campaign 10
  • Campaign 11
  • Campaign 12
  • Campaign 13
  • Campaign 14
  • Campaign 15
  • Campaign 16
  • Campaign 17
  • Campaign 18
  • ________________________________________________________ Injected and Pre-processed Light Curves: To enable others to test their own detection routines without enduring the computational overhead, we provide our set of pre-EVEREST injected light curves after being processed by EVEREST (Flux). Each campaign contains a TAR ball with CSV files for each individual target.
  • Campaign 1
  • Campaign 2
  • Campaign 3
  • Campaign 4
  • Campaign 5
  • Campaign 6
  • Campaign 7
  • Campaign 8
  • Campaign 10
  • Campaign 11
  • Campaign 12
  • Campaign 13
  • Campaign 14
  • Campaign 15
  • Campaign 16
  • Campaign 17
  • Campaign 18
  • ________________________________________________________ Pre-processed Light Curves: To ensure consistency in processing, we also provided the corresponding EVEREST data (Flux) without injections. This experienced the exact same software manipulation as that of the injected data.
  • Campaign 1
  • Campaign 2
  • Campaign 3
  • Campaign 4
  • Campaign 5
  • Campaign 6
  • Campaign 7
  • Campaign 8
  • Campaign 10 (part 2)
  • Campaign 11 (part 1)
  • Campaign 11 (part 2)
  • Campaign 12
  • Campaign 13
  • Campaign 14
  • Campaign 15
  • Campaign 16
  • Campaign 17
  • Campaign 18

  • Sample Reliability:
    Despite efforts to remove problematic cadences, instrument systematics pollute the light curve data, creating artificial dips that can be erroneously characterized as a transit signal. To measure the purity in a homogeneous catalog of planets, the rate of these false alarms (FAs) must be quantified.
    The main goal of EDI-Vetter is to parse through all TCEs and remove FAs without eliminating true planet candidates. However, this process is difficult to automate and requires a method of testing the software's capability to achieve this goal. Accomplishing such a task necessitates an equivalent data set that captures all the unique noise properties, that contribute to FAs, without the existence of any true astrophysical signals. With such data available, the light curves can be processed through the detection pipeline. By inverting the light curves and re-running our analysis we achieve such a simulation and provide the resulting FAs in the following CSV file:
  • Reliability Summary
  • Jon Zink
    email: jzink@caltech.edu
    : @jonKzink

    office: cahill 315
    1216 e california blvd
    pasadena, ca 91125