Simultaneous inference of telescope pixel response functions and detector systematics for space-based transit searches
Transit searches in space-based missions like Kepler are typically achieved using aperture photometry, in which particular pixels on the detector are summed in order to infer a star’s brightness. When the stellar density is comparable to the detector point spread function, aperture photometry is unreliable as a unique aperture containing only the light from one star is unachievable, and small motions of the stars on the detector can move stars in and out of the aperture, causing brightness variations that overwhelm transit signals. Therefore, regions of high stellar density, such as the clusters observed by the K2 mission and large swaths of the TESS field of view, require different techniques. Here, I will discuss a novel technique to use K2 and simulated TESS data to build a data-driven model of the pixel response function (PRF), and to simultaneously lean both the PRF and the detector flat field in a model-independent way across the entire detector. I will show the results of transiting planet searches in dense clusters observed during the Kepler and K2 missions. This method can be used to reliably recover both stellar variability and transiting planets in clusters observed with K2 and those in the TESS full frame images, and to help localize transiting planet signals in crowded regions where target confusion can be high.