An improved method of determining near-surface Q

We performed field measurements using the modified method of spectral ratio s to estimate shallow seismic Q. Three component seismograms from artificia l sources were recorded to determine Q(p) and Q(s) in the unconsolidated se dimentary layer at the experimental site. This modified spectral ratio meth od was assumed to be frequency dependent, and the amplitude ratios then wer e plotted against the arrival-time difference of any two receivers for one particular frequency. The slope of the regression line in the log-amplitude -time space yields a Q for each frequency. Results show that Q is a functio n of frequency in the frequency range (below 300 Hz) we tested. A simple ma thematical derivation with experimental data strongly suggests that the Q o f shallow seismic waves is frequency dependent. Corrections for geometric s preading are used; however, the original and corrected Qs show no significa nt difference in our data, and therefore the geometric factor may be ignore d in this problem. The conventional frequency-independent spectral ratio me thod is easier and faster to apply, but it gives less stable results than t his modified method. The unstable Q is attributed to geometric amplificatio n effects in the conventional frequency-independent spectral ratio method. The source factor can have an effect on the estimates of Q; however, differ ent seismic sources give about the same Q over the dominant frequency band. We established the frequency function by assuming a simple power law regres sion model, where n(p) similar to 1.1 and k << 1 in Q = k(f)(n). This may c onfirm that the weathered unconsolidated layer is saturated partially, and Q(s) > Q(p) stresses that attenuation in our study is physically a local co mpressional mechanism.