Search techniques for gravitational waves from black-hole ringdowns

The first kilometer-scale gravitational wave observatories, LIGO [1] and VIRGO [2], are expected to begin performing searches for astrophysical sources of gravitational waves in the year 2002. By this time, it will be necessary to have data analysis software for on-line searches. Because the anticipated signals are expected to be weak, near-optimal strategies for data analysis will be required in order to extract gravitational wave signals from the interferometer noise. An optimal technique when the waveform of the signal is known in advance is the method of matched filtering [3]. The statistical properties of the matched filter may be characterized when the detector noise is stationary and Gaussian. However, in order to understand the behavior of the matched filter for a real detector, it is necessary to apply the method to a problem of data analysis that resembles, in some way, the expected nonstationary and non-Gaussian LIGO and VIRGO interferometer data. In this paper, I report the results of analyzing data acquired from the Caltech 40-meter prototype interferometer [4] in November 1994. The noise spectrum of the prototype instrument had some of the same characteristics of the spectra expected in LIGO and VIRGO: at low frequency it was dominated by seismic ground motion and at high frequency it became photon shotnoise limited. However the 40-meter prototype experienced many still uncharacterized non-Gaussian transient disturbances. In this respect, the 40-meter prototype is expected to be much worse than the LIGO and VIRGO interferometers. Nevertheless, the prototype forms a useful test-bed for development of robust algorithms that may be used with some degree of confidence in analyzing the data of future instruments.