That's How We Roll: The NASA K2 Mission Science Products and Their Performance Metrics

Abstract

NASA's exoplanet Discovery mission Kepler was reconstituted as the K2 mission a year after the failure of the 2nd of Kepler's 4 reaction wheels in May 2013. The new spacecraft pointing method now gives typical roll motion of 1.0 pixels peak-to-peak over 6 hours at the edges of the field, two orders of magnitude greater than for Kepler. Despite these roll errors, the flight system and its modified science data processing pipeline restores much of the photometric precision of the primary mission while viewing a wide variety of targets, thus turning adversity into diversity. We define metrics for data compression and pixel budget available in each campaign; the photometric noise on exoplanet transit and stellar activity time scales; residual correlations in corrected long cadence light curves; and the protection of test sinusoidal signals from overfitting in the systematic error removal process. We find that data compression and noise both increase linearly with radial distance from the center of the field of view, with the data compression proportional to star count as well. At the center, where roll motion is nearly negligible, the limiting 6 hour photometric precision for a quiet 12th magnitude star can be as low as 30 ppm, only 25% higher than that of Kepler. This noise performance is achieved without sacrificing signal fidelity; test sinusoids injected into the data are attenuated by less than 10% for signals with periods up 15 days. At time scales relevant to asteroseismology, light curves derived from K2 archive calibrated pixels have high-frequency noise amplitude within 40% of that achieved by Kepler. The improvements in K2 operations and science data analysis resulting from 1.5 yr of experience with this new mission concept, and quantified by the metrics in this paper, will support continuation of K2's already high level of scientific productivity in an extended K2 mission.