Filtering and calibration of data from a resonant-mass gravitational wave antenna

Abstract

Algorithms for extracting a burst gravitational wave signal embedded within the noise of resonant-mass gravitational wave antenna have been well characterized theoretically, but their effects on experimental data, which can be contaminated by non-stationary, non-Gaussian noise, are still being studied. In this paper, we study the effects of three such algorithms, the zero-order prediction, adaptive Wiener-Kolmogorov and non-adaptive Wiener-Kolmogorov algorithms, on data from the resonant-mass gravitational wave antenna, Niobe, at the University of Western Australia. By applying these filters to computer-simulated GW signals, we show that the adaptive Wiener-Kolmogorov filter gives the best noise performance and signal-to-noise ratio in the presence of non-Gaussian noise. By searching for coincidences between the simulated signals, we show that a window larger than the sampling time of the data is necessary to observe a coincidence between all events. A method of applying pulse excitations to Niobe by amplitude modulating the pump oscillator driving the parametric transducer is also described. This method has the potential to be a very accurate calibration technique but uncertainties in the input and output gains reduce its accuracy. Finally, the adaptive and non-adaptive Wiener-Kolmogorov filters are applied to pulses generated by the amplitude modulation method to determine the overall timing delays and energy uncertainties of Niobe and its data acquisition system.