Assessing the residual biases in high-resolution transit absorption spectra correction

In recent years it has become common practice to divide observed transit absorption spectra by synthetic absorption spectra computed for the transit of an atmosphere-less planet. This action supposedly corrects the observed absorption spectrum, leaving the sole atmospheric absorption signature free from the biases induced by stellar rotation and centre-to-limb variations. We aim to show that while this practice is beneficial, it does not completely correct the absorption spectrum from the stellar distortions and that some residual biases remain, leaving a possibly altered atmospheric signature. By reducing the problem to its most basic form, we show that dividing the observed absorption spectrum by a synthetic absorption spectrum of the planet does not isolate the pure atmospheric absorption signature. We also used simulated synthetic transit observations to assess the magnitude of these residual biases in typical transit observations. We show that dividing the observed absorption spectrum by the planetary absorption spectrum results in an atmospheric signature modulated by the ratio of the flux behind the atmosphere and the flux behind the planet. Depending on the non-homogeneity of the stellar spectrum, this leads to distorted atmospheric signatures. Eventually, directly analysing these biased signatures will lead to wrong estimates of planetary atmosphere properties.