Answers are grounded in this paper and cite their sources.
Earth noise below 0.5 cps does not differ significantly from stationary normally distributed random noise for record lengths of the order of hours. Earthquake waves, as well as locally generated ground motion such as that due to wind, produce records which differ significantly from the above. The cross-frequency coherence is a sensitive test for nonstationary effects such as earthquakes, and the more generalized two-dimensional spectrum appears to be the appropriate one for looking at transient events. For stationary processes the bispectrum is a convenient method for testing nonlinear interactions, but microseisms measured so far have produced no cases of significant bispectrums.
No takes yet. Share an insight, caveat, or question.
Richard A. Haubrich (1965) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: