Capturing system drift with time series calibration for Global 21-cm cosmology experiments

ABSTRACT To achieve the sensitivity required to detect signals from neutral hydrogen from the Cosmic Dawn and Epoch of Reionization it is critical to have a well-calibrated instrument which has a stable calibration over the course of the observation. Previous Bayesian calibration methods do not explicitly use the time information available and make assumptions on the impedance matching of the reference sources. Here, we present a new calibration method based on noise wave parameters which fits a calibration solution over time and frequency to the data, interpolating the solutions to the times at which the antenna is being measured. To test this method, we simulate a data set using measurements of the REACH receiver, modelling a low-noise amplifier which is drifting over time. Fitting a polynomial surface in frequency and time to the simulated data demonstrates that we can remove the drift in the calibrated solution over time but leaves a chromatic residual. We further show that we can remove assumptions on the reflection coefficients of the reference noise source and the cold load, reducing degeneracies in the parameter fits. Applying this new calibration equation and surface fitting method to the simulated data removes the chromatic residual in the calibrated spectrum and recovers the parameters to within 0.06 per cent of the truth and a 97 per cent reduction in the RMSE of the spectrum of the validation source compared with previous calibration methods. For two parameters, we report up to six times smaller fit error after the degeneracies are removed from the time-based calibration.