Disentangling planetary and stellar activity features in the CoRoT-2 light curve

Abstract

[Abridged] Context. Stellar activity is an important source of systematic errors and uncertainties in the characterization of exoplanets. Most of the techniques used to correct for this activity focus on an ad hoc data reduction. Aims. We have developed a software for the combined fit of transits and stellar activity features in high-precision long-duration photometry. Our aim is to take advantage of the modelling to derive correct stellar and planetary parameters, even in the case of strong stellar activity. Methods. We use an analytic approach to model the light curve. The code KSint, modified by adding the evolution of active regions, is implemented into our Bayesian modelling package PASTIS. The code is then applied to the light curve of CoRoT-2. The light curve is divided in segments to reduce the number of free parameters needed by the fit. We perform a Markov chain Monte Carlo analysis in two ways. In the first, we perform a global and independent modelling of each segment of the light curve, transits are not normalized and are fitted together with the activity features, and occulted features are taken into account during the transit fit. In the second, we normalize the transits with a model of the non-occulted activity features, and then we apply a standard transit fit, which does not take the occulted features into account. Results. Our model recovers the activity features coverage of the stellar surface and different rotation periods for different features. We find variations in the transit parameters of different segments and show that they are likely due to the division applied to the light curve. Neglecting stellar activity or even only bright spots while normalizing the transits yields a $\sim 1.2\sigma$ larger and $2.3\sigma$ smaller transit depth, respectively. The stellar density also presents up to $2.5\sigma$ differences depending on the normalization technique...