The behavior of Q in layers B and G in the SCS data is consistent with the reflectivity seen in the seismic data and the frequency shift seen in the spectrogram. Southwest of Site 994, the reflections between 0.55 and 0.8 s are relatively weaker than they are to the northeast and the frequency content and Q values are significantly higher. Northeast of Site 995 the reflection strengths increase as the frequencies and Q values decrease. Northeast of Site 997 there is a slight increase in frequency and Q. We know from drilling (Paull, Matsumoto, Wallace, et al., 1996) that these highly reflective layers contain significant quantities of free gas (up to 12% by volume at Site 997, [Dickens et al., 1997]), but at least the presence of gas could probably have been determined before drilling. Gas seems to be the only geologically reasonable way to explain Q as low as 10-20 in a fine-grained sediment drift. At Site 995, the results from the VSP are reasonably consistent with the SCS inversion, indicating a Q > 20 for the HSZ (Q = 200-300 in SCS data), and Q = 6 for the GSZ (Q = 10-20 in SCS data). On the basis of the scatter in the curves from the SCS data, we believe the GSZ Q estimate from the VSP is more reliable.
Determining the amount of gas from the curves in Figure 3 is somewhat more difficult. We know from a pressurized core recovery that gas occupies ~12% of the total volume near the BSR at Site 997, where Q in the GSZ is 40-80. Although values of Q are even lower at Site 995, it is uncertain whether this is due to a higher concentration of gas, a more extensive vertical distribution of gas, or both. The latter case seems to be supported by the seismic data, where a thinner zone of stronger reflections exists at Site 997 (0.55-0.64 s) as compared to a thicker zone of only slightly weaker reflectors at Site 995 (0.55-0.8 s). The slightly, but systematically lower Q found in layer G suggests even more gas than in layer B, especially at Site 995, reinforcing the notion of a thick gas zone that elevates the reflectivity of the sediments below the BSR (Holbrook et. al, 1996). Recall also that Q may not uniquely determine the percent saturation. The data of Murphy (1982) in sandstones suggests a peak attenuation at ~90% water saturation after which attenuation decreases with decreasing water saturation. Quantification may only be possible where small quantities of gas are present or may yield a minimum estimate of the amount of gas. In either case, a more consistent means of quantifying gas in situ is required for corroboration.
The results of the Q inversion in the HSZ are more difficult to interpret than the results in the GSZ. There are no apparent lateral changes in the seismic data or spectrogram that seem to correlate to the lateral changes in Q for layers C and H. If higher values of Q were associated with higher concentrations of hydrate, we might expect more laterally consistent values, based on the relatively consistent chloride anomaly proxy (Paull, Matsumoto, Wallace, et al., 1996). Instead the minimum occurs about 2 km southwest of Site 997, directly below the ridge crest. Although the Q in layers C and H may be augmented by scattering, it is difficult to see how the scattering could change so systematically across the section. Also, if this change was some artifact of the modeling, it would likely be more closely correlated to the curves from the GSZ that appear to reach a minimum 5-7 km to the southwest.