Centroid vetting of transiting planet candidates from the Next Generation Transit Survey

TL;DR

This paper introduces a fully-automated centroid vetting algorithm for the NGTS, improving false positive identification in ground-based transit surveys by detecting centroid shifts with high precision, thus enhancing candidate validation efficiency.

Contribution

The paper presents the first ground-based implementation of an automated centroid vetting technique for wide-field transit surveys, integrated into NGTS for improved false positive rejection.

Findings

Centroid shifts can be detected with an average precision of 0.75 milli-pixel.

The algorithm is fully automated and integrated into NGTS.

Ground-based surveys can now effectively use centroiding for candidate vetting.

Abstract

The Next Generation Transit Survey (NGTS), operating in Paranal since 2016, is a wide-field survey to detect Neptunes and super-Earths transiting bright stars, which are suitable for precise radial velocity follow-up and characterisation. Thereby, its sub-mmag photometric precision and ability to identify false positives are crucial. Particularly, variable background objects blended in the photometric aperture frequently mimic Neptune-sized transits and are costly in follow-up time. These objects can best be identified with the centroiding technique: if the photometric flux is lost off-centre during an eclipse, the flux centroid shifts towards the centre of the target star. Although this method has successfully been employed by the Kepler mission, it has previously not been implemented from the ground. We present a fully-automated centroid vetting algorithm developed for NGTS, enabled by our high-precision auto-guiding. Our method allows detecting centroid shifts with an average precision of 0.75 milli-pixel, and down to 0.25 milli-pixel for specific targets, for a pixel size of 4.97 arcsec. The algorithm is now part of the NGTS candidate vetting pipeline and automatically employed for all detected signals. Further, we develop a joint Bayesian fitting model for all photometric and centroid data, allowing to disentangle which object (target or background) is causing the signal, and what its astrophysical parameters are. We demonstrate our method on two NGTS objects of interest. These achievements make NGTS the first ground-based wide-field transit survey ever to successfully apply the centroiding technique for automated candidate vetting, enabling the production of a robust candidate list before follow-up.