A Generalized Method for Characterizing 21 cm Power Spectrum Signal Loss from Temporal Filtering of Drift-scanning Visibilities

Abstract A successful detection of the cosmological 21 cm signal from intensity mapping experiments (e.g., during the Epoch of Reionization or Cosmic Dawn) is contingent on the suppression of subtle systematic effects in the data. Some of these systematic effects, with mutual coupling a major concern in interferometric data, manifest with temporal variability distinct from that of the cosmological signal. Fringe-rate filtering—a time-based Fourier filtering technique—is a powerful tool for mitigating these effects; however, fringe-rate filters also attenuate the cosmological signal. Analyses that employ fringe-rate filters must therefore be supplemented by careful accounting of the signal loss incurred by the filters. In this paper, we present a generalized formalism for characterizing how the cosmological 21 cm signal is attenuated by linear time-based filters applied to interferometric visibilities from drift-scanning telescopes. Our formalism primarily relies on analytic calculations, and therefore has a greatly reduced computational cost relative to traditional Monte Carlo signal loss analyses. We apply our signal loss formalism to a filtering strategy used by the Hydrogen Epoch of Reionization Array (HERA) and compare our analytic predictions against signal loss estimates obtained through a Monte Carlo analysis. We find excellent agreement between the analytic predictions and Monte Carlo estimates and therefore conclude that HERA, as well as any other drift-scanning interferometric experiment, should use our signal loss formalism when applying linear, time-based filters to the visibilities.