Characterization of an Instrument Model for Exoplanet Transit Spectrum Estimation through Wide Scale Analysis on HST Data
Instrument models (IMs) enable the reduction of systematic error in transit spectroscopy light curve data, but, since the model formulation can influence the estimation of science model parameters, characterization of the instrument model effects is crucial to the interpretation of the reduced data. We analyze a simple instrument model and assess its validity and performance across Hubble WFC3 and STIS instruments. Over a large, n=63, sample of observed targets, an MCMC sampler computes the parent distribution of each instrument model parameter. Possible parent distribution functions are then fit and tested against the empirical IM distribution. Correlation and other analyses are then performed to find IM relationships. The model is shown to perform well across the 2 instruments and 3 filters analyzed and, further, the Student's t-distribution is shown to closely fit the empirical parent distribution of IM parameters and the Gaussian is shown to poorly model the observed distribution. This parent distribution can be used in the MCMC prior fitting and demonstrates IM consistency for wide scale atmospheric analysis using this model. Finally, we propose a simple metric based on light curve residuals to determine model performance, and we demonstrate its ability to determine whether a derived spectrum under this IM is high quality and robust.