Fortifying the characterization of binary mergers in LIGO data

The study of compact binary inspirals and mergers with gravitational wave observatories amounts to optimizing a theoretical description of the data to best reproduce the true detector output. While most of the research effort in gravitational wave data modeling focuses on the gravitational waveforms themselves, here we will begin to improve our model of the instrument noise by introducing parameters that allow us to determine the background instrumental power spectrum while simultaneously characterizing the astrophysical signal. We use data from the fifth LIGO science run and simulated gravitational wave signals to demonstrate how the introduction of noise parameters results in the resilience of the signal characterization to variations in an initial estimation of the noise power spectral density. We find substantial improvement in the consistency of Bayes factor calculations when we are able to marginalize over uncertainty in the instrument noise level.