Paper detail

Statistical tests with multi-wavelength Kernel-phase analysis for the detection and characterization of planetary companions

Kernel phase is a method to interpret stellar point source images by considering their formation as the analytical result of an interferometric process. Using Fourier formalism, this method allows for observing planetary companions around nearby stars at separations down to half a telescope resolution element, typically 20\,mas for a 8\,m class telescope in H band. The Kernel-phase analysis has so far been mainly focused on working with a single monochromatic light image, recently providing theoretical contrast detection limits down to $10^{-4}$ at 200\,mas with JWST/NIRISS in the mid-infrared by using hypothesis testing theory. In this communication, we propose to extend this approach to data cubes provided by integral field spectrographs (IFS) on ground-based telescopes with adaptive optics to enhance the detection of planetary companions and explore the spectral characterization of their atmosphere by making use of the Kernel-phase multi-spectral information. Using ground-based IFS data cube with a spectral resolution R=20, we explore different statistical tests based on kernel phases at three wavelengths to estimate the detection limits for planetary companions. Our tests are first conducted with synthetic data before extending their use to real images from ground-based exoplanet imagers such as Subaru/SCExAO and VLT/SPHERE in the near future. Future applications to multi-wavelength data from space telescopes are also discussed for the observation of planetary companions with JWST.