The most time consuming, but also most direct, geophysical technique employed on Leg 164 was vertical seismic profiling (both WVSP and ZVSP) at Sites 994, 995, and 997 using air-gun and water-gun sources. Details of the VSP experiments can be found in Paull, Matsumoto, Wallace, et al. (1996). In addition to very accurate estimates of depth to the BSR (440 ± 10 m at Site 995, and 464 ± 8 m at Site 997), the ZVSP results (Fig. 2, Fig. 3) suggest the presence of a thick low-velocity zone (Vp < 1550 m/s) below the BSR. Holbrook et al. (1996) attribute the sharp drop in velocity to gas-charged sediments, the upper limit of which is marked by bright, reversed polarity reflections in the SCS data (Fig. 1, Fig. 3). However, at Site 994 this bright reflector is not continuous with the strong BSR observed at Sites 995 (440 mbsf) and 997 (464 mbsf). Instead, the Site 994 reflector lies at a depth of ~560 mbsf. As it is traced across the ridge, this reflector parallels the BSR at Sites 995 and 997 and even appears to lie above the BSR several kilometers to the northeast (Dillon, Hutchinson, et al., 1996). Although the ~560 mbsf reflector almost certainly corresponds to the top of free gas at Site 994, in a more regional sense the reflector appears to be coincident with the top of a series of buried faults seen in some seismic sections. At Sites 995 and 997, the coincidence of the BSR in the SCS data and the top of the free-gas zone as constrained by the VSP data confirms earlier interpretations of the BSR as a reversed polarity reflector that marks the top of the free-gas zone (Shipley et al. 1979; Paull and Dillon, 1981).
The VSP experiments also provided constraints on Vp of sediments within the gas hydrate zone (GHZ), sediments for which shipboard pore-water analyses yield chloride anomalies consistent with the presence of gas hydrate under in situ conditions (Paull, Matsumoto, Wallace et al, 1996). The measured Vp of 1600-1850 m/s (Fig. 2) in the GHZ from 200 to 450 mbsf is somewhat lower than estimates based on worldwide averages of fine-grained terrigenous sediments presumably free of gas hydrate (Hamilton 1980) and is far lower than that of pure hydrate (3800 m/s; Sloan, 1997) or sediment frozen with water ice (>4000 m/s; Toksoz et al., 1979). Based on the weighted time average equation of Lee et al. (1993), which uses matrix velocity, porosity, and measured velocity to determine gas hydrate concentration, Holbrook et al. (1996) estimate the concentration of gas hydrate in the Blake Ridge to be 5%-7% of pore volume or 2%-3% of total volume, consistent with estimates based on chloride anomalies (Paull, Matsumoto, Wallace, et al. 1996).
The seismic velocities determined from the ZVSP data were confirmed by a WVSP study that revealed ~10% anisotropy in Vp at Site 995 (Pecher et al., 1997). Although this result may have some implications for quantifying the concentration and distribution of gas hydrate (e.g., evenly disseminated vs. as cement in layers), such values of anisotropy are not unusual in fine, layered sediments (Levin, 1979) and may not necessarily reflect the presence of gas hydrate.
The velocities within the GHZ determined from the ZVSPs were also somewhat lower than velocities derived from other measurements in the area (Fig. 2). Using wide angle ocean bottom hydrophone (OBH) data from this location, Katzman et al. (1994) estimate that the velocity of the sediments ~200 m above the BSR is 1900 m/s, similar to the result obtained by Wood et al. (1994) using MCS data acquired with a 6-km-long, surface-towed streamer. The highest velocities reported for GHZ sediments in the area are as high as 2450 m/s and are determined from deep-towed MCS data acquired on the flank of the ridge ~160 km to the south and in 4000 m of water (Rowe and Gettrust, 1993b). The anisotropy discussed earlier may explain the lower velocities determined on the basis of the ZVSPs. The energy recorded by the ZVSPs travels in the vertical, slow direction, whereas the other analyses are based on energy traveling with some horizontal component, resulting in higher estimates of velocity. However, anisotropy of 10% is insufficient to explain the velocity discrepancy between the ridge crest and flank, which are presumably lithologically similar. The difference in velocities might be better explained by higher gas hydrate concentrations on the flank than on the crest of the ridge. As discussed below, higher concentrations of gas hydrate may reflect more vigorous fluid flux. Greater concentrations of gas hydrate on the ridge flank might be expected if the highly disrupted nature of the BSR (Rowe and Gettrust, 1993a) is either a cause or effect of increased fluid flux.
Another assessment of Vp values in the study area has been made by Tinivella and Lodolo (Chap. 28, this volume) based on tomographic inversion of surface-towed MCS data. The results are generally consistent with those obtained from the VSP data but have poorer vertical resolution due to the geometry of the data acquisition.
In marine settings, shear-wave velocity measurements with a surface source may only be obtained from seismic energy that has been converted from P- to S-waves. This requires large incidence angles, or equivalently, large offsets, which were obtained on Leg 164 using walk-away VSP (WVSP) data collected with borehole geophones operated from the JOIDES Resolution at Sites 994 and 995. The data were acquired using a 150-in3 generator-injector air-gun source towed behind a second ship (R/V Cape Hatteras) at offsets up to ~5 km, allowing incidence angles of up to 45° . A better signal to noise ratio was obtained at Site 995 than at Site 994 because of operational and mechanical issues (Paull, Matsumoto, Wallace, et al., 1996). Most analyses have therefore concentrated on the Site 995 data. At Site 995, additional wide-angle data were also obtained through the deployment of a three-component ocean-bottom seismometer.
To date, the analyses of the WVSP data collected on Leg 164 have revealed the first known P-to-S conversions from a BSR (Pecher et al., 1996, 1997), an effect attributed to a sharp change in Vs at the reflector. Other converted shear waves in the data set came from within the GHZ itself, but accurate determinations of the depths of these conversions requires extensive inverse modeling. For the specific geometry of the WVSP experiments, Pecher et al. (1997) show that if the shear modulus changes across a given interface, then a significant Vs impedance contrast can exist without a significant Vp impedance contrast. Thus, if small amounts of gas hydrate cement mostly affect the shear modulus of sediments (Dvorkin and Nur, 1993), then the results of Pecher et al. (1997) suggest that seismically based determinations of the concentration and distribution of gas hydrate may require the use of wide angle data.
The attenuation of P-waves through Blake Ridge sediments has been investigated by Wood et al. (Chap. 27, this volume) using SCS data acquired at the sea surface on a transect coincident with the drill sites (Fig. 1, Line 31; Katzman et al., 1994). The analysis was based on the change in frequency content of the normal incident signal with depth, and the principal results are shown in Wood et al. (Fig. 2B of Chapter 27, this volume). Although detailed investigations of Q in fine-grained sediments are rare, Wood et al. (Chap. 27, this volume) suggest that gas hydrate concentrations on the order of a few percent by volume have little or no effect on P-wave attenuation. Values of Q = 100-400 determined within the gas hydrate-stability zone (GHSZ) are not unexpected for fine-grained sediments lacking gas hydrate (Bowles, 1997). Below the GHSZ, the analysis revealed significantly higher attenuation (Q values as low as ~5), consistent with gas charged sediments (Toksoz et al., 1979). This corroborates the results of other seismic studies that indicate the presence of free gas at this depth (Holbrook et al., 1996).
Leg 164 also provided sample, logging, and paleontological data critical to the interpretation of seismic images that reveal high-angle faulting of Blake Ridge sediments. Figure 4A shows surface-towed, 10- to 240-Hz SCS data collected in the vicinity of Site 997 as part of Line 31 (Katzman et al., 1994). These data are very mildly filtered (cosine ramp from 0 to 1 between 5 and 10 Hz and from 1 to 0 between 200 and 240 Hz) and have not been migrated, because such processing may blur higher frequency waveforms. The image shows faulting (lateral discontinuities) throughout the GHSZ and underlying sediments. However, it is difficult to trace individual faults or determine which, if any, fault or faults may have been intersected during drilling.
Evidence that a fault plane was intersected at Site 997 comes from the composite paleontological record (Fig. 4D) to a depth of 424 mbsf in Hole 997A and from 416 to 750 mbsf in Hole 997B (Paull, Matsumoto, Wallace, et al., 1996). A temporal hiatus between 360 and 375 mbsf (arrow in Fig. 4D) corresponds to ~14 m of missing section. There is no indication in the seismic data of an exceptionally strong reflector or truncated reflectors corresponding to an erosional surface at this depth. Normal faulting, which is seen in seismic data and which can account for missing section, is the more likely explanation.
A higher resolution image of this type of faulting was acquired in 1993 at a location ~1 km to the southeast of the drilling transect using the Naval Research Laboratory's deep-towed acoustics geophysics system (DTAGS, Gettrust et al., 1991). The small sample of unmigrated data shown in Figure 5 is of the same vertical and horizontal dimension as the SCS data in Figure 3. The 24-channel vertical array DTAGS data were moved out, median stacked, and shifted to make the seafloor horizontal; depth converted using velocities obtained at Site 997 by Holbrook et al. (1996); and gained with a linear ramp of slope 0.08 m-1 beginning at 80 mbsf. The DTAGS data exhibit clear images of individual faults and show that the faults are spaced tens to hundreds of meters apart, frequently penetrate the BSR, and are quite prevalent on the ridge crest (Rowe et al., 1995; Wood and Gettrust, 1998). Displacement across the faults, which the DTAGS section clearly shows to be syndepositional, is ~14 m at 340 mbsf but less than vertical resolution of the data at ~30 mbsf. This amount of displacement is consistent with the thickness of missing section estimated from the paleontological results. That the perturbation of the BSR at the location of the faults is significantly smaller than the displacement on the fault might be partially attributed to a small pressure change across the fault (Rowe and Gettrust, 1993a). The faults have normal sense displacement and associated minor antithetic faults that produce small grabens.
The significance of the faulting is clear when viewed in the context of other Leg 164 geophysical data. A massive section of hydrate ~20 cm long (pure hydrate, from which virtually all sediment has been excluded) was recovered at 330 mbsf at Hole 997A but not at Hole 997B, located only 20 m away to the northwest (Fig. 4A). Subsequent logging of Hole 997B revealed a thin (less than a few meters thick) zone of anomalously high acoustic velocity and resistivity at 365 mbsf (Fig. 4B and Fig. 4C, respectively), interpreted as an indicator of localized concentrations of methane hydrate (Paull, Matsumoto, Wallace, et al., 1996). The velocity and resistivity anomalies in Hole 997B occurred 35 m deeper than the massive hydrate in Hole 997A and was inferred to indicate the presence of gas hydrate in ~23% of pore volume (Paull, Matsumoto, Wallace, et al., 1996).
Highly localized concentrations of gas hydrate like those discovered in Hole 997A are consistent with hydrate accumulation controlled by strong fluid advection (Xu and Ruppel, 1999), probably in a high-permeability zone represented by a single high-angle fault crossing both holes. If gas hydrate concentration was instead primarily controlled by sediment stratigraphy, two holes separated by only 20 m might be expected to exhibit sediment anomalously rich in hydrate at nearly identical depths. Although the paleontologically determined intersection of a possible fault and the two holes is inexplicably reported 30-40 m deeper than the level that yielded the massive hydrate, it seems fairly certain that at least one fault was intersected and that this fault likely promotes fluid advection and gas hydrate formation.