Improved Coherent Integration through Fringe Model Fitting

Maximizing the signal-to-noise ratio (SNR) of interferometric measurements of faint or low-visibility sources requires coherent integration. Coherent integration requires the use of a fringe-tracking method for determining the appropriate quantities to use for cophasing a large number of short low-SNR measurements. We present a method for tracking fringes during postprocessing. The method is based on fitting a time-dependent model of the fringe pattern to a sequence of data frames in order to determine the phase parameters for the central frame. This approach was chosen in accordance with the philosophy that optimal parameter estimation depends on a accurate model of the data. The model contains two parameters which must be chosen: the order of the time variation, and the number of simultaneous data frames to fit to. We show that this procedure results in a better SNR than a group-delay approach, particularly for low-NV2 data. In one case, the fringe-modeling approach improved coherent SNR by a factor of 2.6 over the group-delay approach. The fringe-modeling algorithm is tested on observations of the star ξ Bootis obtained from the Navy Prototype Optical Interferometer. We also offer suggestions for future improvements to the fringe-tracking model.