A study of diagnostic techniques for identifying earthquakes

Several diagnostic techniques are examined for identifying earthquakes as events distinct from possible underground nuclear explosions. It is found that the typical or 'mean' earthquake differs in a statistically significant way from the 'mean' explosion for most of the techniques. Because of the variability of earthquake signatures, many earthquake parameters fall within the range observed for explosions. The following numbers refer to the percentage of earthquakes that fall entirely outside the range observed for explosions for the particular techniques: Long-period surface wave energy density, 80 per cent (10 explosions, 10 quakes); L_g/R_g ratios, 50 per cent (50 explosions, 50 quakes); peak amplitude of Love waves, 40 per cent (7 explosions, 24 quakes); peak amplitude of Rayleigh waves, 30 per cent (11 explosions, 21 quakes); excitation of S_g 30 per cent (32 explosions, 13 quakes); P²/P¹ ratio, 20 per cent (45 explosions, 87 quakes); ratio of spectrums of long-period Love and Rayleigh waves, 70 per cent (4 explosions, 11 quakes); prevailing period of Love waves, 100 per cent (4 explosions, 3 quakes). Long-period surface waves are excited by most events of magnitude greater than 4.4. For 3.9 < M_L < 4.4 only 1/4 of the events generated surface waves. It is argued on theoretical grounds that the absolute value and not the shape of the spectrums of surface waves depends on magnitude for earthquakes below a certain magnitude. If this were so, the prevailing periods of Rayleigh waves from small-magnitude explosions and earthquakes would not differ.