<tt>LEO-Vetter</tt> : Fully Automated Flux- and Pixel-level Vetting of TESS Planet Candidates to Support Occurrence Rates

Abstract The Transiting Exoplanet Survey Satellite (TESS) has identified several thousand planet candidates orbiting a wide variety of stars, and has provided an exciting opportunity for demographic studies. However, current TESS planet searches require significant manual inspection efforts to identify planets among the enormous number of detected transit-like signatures, which limits the scope of such searches. Demographic studies also require a detailed understanding of the relationship between observed and true exoplanet populations, a task for which current TESS planet catalogs are rendered unsuitable by the subjectivity of vetting by eye. We present LEO-Vetter , a publicly available and fully automated exoplanet vetting system designed after the Kepler Robovetter, which is capable of efficiently producing catalogs of promising planet candidates and making statistically robust TESS demographic studies possible. LEO-Vetter implements flux- and pixel-level tests against noise/systematic false positives and astrophysical false positives. The vetter achieves high completeness (91%) and high reliability against noise/systematic false alarms (97%) based on its performance on simulated data. We demonstrate the usefulness of the vetter by searching ∼200,000 M dwarf light curves, and reducing ∼20,000 transit-like detections down to 172 uniformly vetted planet candidates. LEO-Vetter facilitates analyses that would otherwise be impractical to perform on all possible signals due to time constraints or computational limitations. Users will be able to efficiently produce their own TESS planet catalog starting with transit-like detections, as well as have the framework needed to characterize their catalog’s completeness and reliability for occurrence rates.