Improved thermal background subtraction with principal component analysis

The Hunt for Observable Signature of Terrestrial Systems (HOSTS Survey) was conducted with the Large Binocular Telescope Interferometer (LBTI) between September 2016 and May 2018. It surveyed successfully 38 nearby stars in order to detect exozodiacal dust. The measurement obtained by the HOSTS survey allowed to constraint the median exozodiacal dust level (zodi level) around nearby star, demonstrating that direct imaging of rocky planets in the habitable zone of their hosts star by future space-based mission is possible. However, in order to really constrain the feasability of futur missions with currently planned design, an improved sensitivity, by a factor 2 at least, is required. The major sensitivity limitation of the HOSTS survey data comes from the imperfect thermal background removal. We here present a method based on principal component analysis to optimize the background subtraction of these data. With this method we obtain the required factor 2 of improvement for the sensitivity.