Author(s): , ,
Institution(s): 1. Millennium Institute of Astrophysics, 2. Pontificia Universidad Catolica de Chile
Transmission spectroscopy allows study of the atmospheres of exoplanets without the need of spatially resolving them from their parent stars and is one of the most valuable follow-up possibilities offered by transiting systems. The measurement of a transmission spectrum, i.e. the apparent planetary size in units in the stellar radius as a function of wavelength, is conceptually simple, but the expected features that need to be discerned are on the order of one part in a thousand or less, and need to be extracted against a background of (potentially correlated) noise and systematic effects with amplitudes greatly exceeding that of the sought signal. In this talk I will describe how we have tackled the estimation of transmission spectra in a ground based survey we are carrying out with IMACS at Las Campanas Observatory, the Arizona-CfA-Catolica Exoplanet Spectroscopy Survey. Our treatment assumes an additive model consisting of the signal, common systematics and one of a set of stochastic processes with different memory characteristics for the noise. Common systematics are estimated from comparison stars using principal component analysis and the model parameter posterior distributions are estimated using MCMC. Model comparison is used to let the data select the model with the most appropriate noise component. I will illustrate the performance of our approach, and discuss possible avenues of improvement. I will also illustrate the importance of potential biases arising from our incomplete knowledge of stellar properties. In particular, I will show that limb darkening assumptions can limit the accuracy of our estimates of planetary radii above the achievable precisions in regimes currently being probed.