Gravitational waves are time-dependent waves in the curvature of space caused by moving masses. Boughn and Kuhn (1984, 1990) have searched for a stochastic background of this gravitational radiation by examining the earth's resonant frequencies for excitations. The vibrations caused by gravitational waves are briefly discussed by Misner, Thorne, and Wheeler (1973) and in more detail by Wagoner and Paik (1976). Tuman (1982, page 129) found that the earth's even harmonics show more excitation than the odd harmonics, as would be expected if the earth was being excited by gravitational radiation.
While the previous studies examined the frequencies around the earth's resonant frequencies, I examined a broad range of frequencies corresponding to periods of one day to one-half hour. Also, where the previous studies have considered mainly stochastic or impulsive gravitational radiation, I considered only radiation produced from binary sources. Signals from these pairs of orbiting masses allow examinations of attributes other than signal strength and frequency shape, such as directional, phase, and polarity consistency. Since many stars exist as binaries, larger objects might also be expected to be found in binary systems. Long-period records of the earth's oscillations measured by the International Deployment of Accelerometers (IDA) should provide a sensitive detector of the gravitational-wave excitations from these binary sources. Detecting these waves with the earth's vibrations would provide information about the earth's deep interior, since the waves act on the earth uniformly with depth, and might also provide new astrophysical information.
To be detected, a gravitational-wave response must be extracted from the background noise in the IDA data. Orbital periods of most astronomical objects are expected to vary at astronomical time scales, so the frequency of a gravitational wave is expected to be sharply defined within a spectrum. The polarity, phase, and direction of the gravitational wave will also be constant for these sources and allow a further discrimination between the noise and signal at a given frequency. Frequency shape and directional consistency are considered as the criteria for detecting a binary source. Polarity and phase measures are not considered here, because they are likely to be useful only after the other measurement criteria are satisfied. To produce these measures, earthquake events were removed from the IDA data, and were then are gained to attenuate noise and to strengthen the weak signals, and finally, directional spectra are produced by weighting by direction and Fourier transforming.
Patterns similar to a synthetic response to a gravitational-wave source were observed in a directional spectrum, but these patterns showed no consistency when the same frequency was compared from year to year, and little consistency was observed even when the preprocessing was changed within the same year. These inconsistencies between years suggest that the patterns are noise caused by earthquakes and muting patterns for each station.
While no signals were detected from these data, more data, better noise attenuation algorithms, and a more sophisticated model of the earth's response might allow weaker signals to be extracted from the noise.