Selective principal component extraction and reconstruction: a novel method for ground based exoplanet spectroscopy

Context. Infrared spectroscopy of primary and secondary eclipse events probes the composition of exoplanet atmospheres and, using space telescopes, has detected H2O, CH4 and CO2 in three hot Jupiters. However, the available data from space telescopes has limited spectral resolution and does not cover the 2.4 − 5.2μm spectral region. While large ground based telescopes have the potential to obtain molecular-abundance-grade spectra for many exoplanets, realizing this potential requires retrieving the astrophysical signal in the presence of large Earth-atmospheric and instrument systematic errors. Aims. Here we report a wavelet-assisted, selective principal component extraction method for ground based retrieval of the dayside spectrum of HD 189733b from data containing systematic errors. Methods. The method uses singular value decomposition and extracts those critical points of the Rayleigh quotient which correspond to the planet induced signal. The method does not require prior knowledge of the planet spectrum or the physical mechanisms causing systematic errors. Results. The spectrum obtained with our method is in excellent agreement with space based measurements made with HST and Spitzer and confirms the recent ground based measurements including the strong ~ 3.3μm emission.