Abstract :
[en] In the mid-infrared, in particular in the N-band (11 $\mu$m), the thermal background is a major limit to the data's sensitivity and a challenge to the extraction of relevant information. In fact, the thermal background can even outshine the star, let alone circumstellar emissions. Thus, an effective background subtraction is required to use the data. The analysis of the Hunt for Observable Signature of Terrestrial Systems (HOSTS) survey data has shown that despite the background subtraction, the residual thermal background still limits the data sensitivity. The background bias is mainly responsible for this sensitivity limitation. Indeed, the background correction of the HOSTS survey makes use of an annulus, around the region of interest, to estimate the residual background. However, the background in those regions can differ and introduce a spatial background bias, limiting the sensitivity. To avoid using the background annulus, and improve the background correction, we developed a Principal-Component-Analysis(PCA)-based method for the background subtraction. We present this method, and its application to high-contrast imaging and aperture photometry.
In the context of the search for Earth-like planets in direct imaging, many challenges arise, including the small inner working angle and contrast. However, an additional, less known challenge resides in the presence of exozodiacal dust in the system. This dust, lying in and around the habitable zone (HZ), can outshine rocky planets and hide them. Detecting and understanding this dust is thus of primary importance to prepare future direct imaging missions dedicated to the detection and characterization of Earth-like planets. Modeling exozodiacal dust in individual systems thus allows to assess both their suitability for exoplanets search, and better understand the dust itself. The HOSTS survey has detected a significant amount of dust in the 110 Her system. This amount of dust would prevent the detection of terrestrial planets. However, the modeling of this system, presented in this thesis, suggests that the dust is located further away from the star than the HZ and might thus not prevent exoplanet detection in the HZ. We also present the impact on the modeling that PCA background subtraction would have.
In addition to imperfect background subtraction, resulting in residual spatial structures, the sensitivity of the data can also be limited by the temporal variation of the background. In addition, the background correction is usually built on images acquired before and after the scientific frames, and the background might change between those exposures. This situation leads to the introduction of a temporal bias in the correction of the background subtraction, which might limit both the accuracy and the precision of the background subtraction. In order to tackle this problem, we developed a second PCA-based method to correct the temporal background variations in addition to the spatial variation corrected by the first method. In this thesis we present the results obtained with this temporal-PCA-based method, as well as its current limitations.
