Inferring the Spectrum of Accretion onto the Pre-Main Sequence Star LkCa 15

The final stage of pre-main sequence star formation is defined by the magnetospheric accretion of gas from a circumstellar disk onto the stellar surface. The combination of the time-variable emission of the accretion shock and the (possibly) time-variable extinction from orbiting circumstellar dust severely complicates measurements of the accretion rate and the fundamental stellar parameters of the system. The presence of the accretion flux acts to reduce the contrast of the stellar photospheric absorption features in high resolution spectroscopic observations ("veiling'' them). Traditional analysis methods which assume the spectrum of the accretion flux to be continuum-like in nature can yield artificially high accretion rate estimates if in fact the accretion flux is significantly concentrated in wavelength, e.g., in emission lines coincident with photospheric absorption features. To examine the influence of these effects on accretion rate estimates, we obtained 20 epochs of high resolution optical spectra and near-simultaneous ugriz photometry of the K5 T Tauri stars LkCa 14 and LkCa 15, which comprise a non-accreting "template'' star and a moderately accreting "target'' star, respectively. We develop a Gaussian process framework to construct a high-fidelity template of the stellar photosphere, and use this to infer the spectrum of the accretion flux on a per-epoch basis. In combination with the broad-band photometry, we use this framework to disentangle the flux changes due to accretion events and variable dimming from circumstellar dust obscuration. We find that the accretion spectrum is not well-described by a continuum source, and indeed contains low amplitude but numerous emission lines, which would otherwise bias accretion rates obtained via traditional methods.