Kepler’s Greatest Hits: Revising Occurrence Rate Benchmarks

We present exoplanet occurrence rates for planets between 0.4-22 Re and between 1-500 days. We apply a non-parametric method via a kernel density estimator to measure occurrence and use Monte Carlo methods for uncertainty estimation. We use a full characterization of completeness and reliability measurements from the Kepler DR25 catalog, including detection efficiency, vetting completeness, astrophysical- and false alarm reliability. We also include more accurate and homogeneous stellar radii from Gaia DR2. We revisit benchmark exoplanet occurrence rate measurements from the literature. Because these benchmarks were calculated on intermediate or independent catalogs, our comparison both tests the method and identifies the differences in measurement values due to the different catalogs. We examine occurrence dependencies on planet radius and orbital period, and reproduce benchmarks within one-sigma confidence for all literature values, except for measurements below 1.3 Re. We then explore the shape of the occurrence cliff (2.5–4 Re) in period-radius space and perform preliminary photoevaporation models to explain the structure. Input distributions used to replicate the radius valley can replicate the structure of the occurrence cliff marginalized over orbital period but cannot replicate the full structure in the period-radius plane, exemplifying the need to retain the 2D shape information. Future work will incorporate full Bayesian inference population inference.