DISCUSSION

As mentioned previously, P-wave velocities within the GHSZ at all three sites, particularly at Sites 1245 and 1247, are much lower than normal velocities for water-saturated sediments, whereas the measured S-wave velocities are a little higher than those for water-saturated sediments. Low P-wave velocities coupled with slightly higher S-wave velocities are speculated to be caused by the existence of free gas commingled with gas hydrate. Figures F6 and F7 indicate that the amount of free gas within the GHSZ is approximately proportional to the amount of gas hydrate and is almost continuous throughout the zone.

Several studies have shown that it is possible for free gas and gas hydrate to coexist inside the GHSZ, as discussed below:

1. Lee and Collett (unpubl. data) demonstrated that there exist free gas zones within the GHSZ at the Cirque-2 well, Alaska (USA), as evidenced by the lack of pore water needed to form gas hydrate. It is known that to form gas hydrate, not only gas but also water is needed. One characteristic of this phenomenon is that these free gas zones are isolated from gas-hydrate-bearing sediments and appear as isolated occurrences of free gas. Somewhat continuous distribution of free gas in this study area may exclude the lack of water as a dominant cause of free gas within the GHSZ.
2. Guerin et al. (1999) showed that gas hydrate and free gas coexist at Site 995 of ODP Leg 164 below the BSR. Ruppel (1997) and Hovland et al. (1997) reported a difference between the depth of the BSR and the theoretical bottom of the GHSZ. They associated this discrepancy with capillary forces developed in fine-grained sediments, which reduce the temperature where gas hydrates dissociate. Gas hydrate grows preferentially in large pore spaces and, reciprocally, dissociates first in smaller pores with strong capillary effects (Hovland et al., 1997). Guerin et al. (1999) speculated that in the interval between the BSR and the GHSZ, gas hydrate within smaller pores may have dissociated and released free gas, while gas hydrate remains present within the larger pores. Therefore, two phases would then coexist in this interval between the BSR and the bottom of the GHSZ. However, this study differs from the results presented here for Leg 204, where free gas exists throughout the GHSZ.
3. Near Site 1250, a double BSR is observed. Double BSRs are often associated with changes in the GHSZ over time caused by seafloor deposition or erosion, sea level fluctuation, local variation in geothermal state, or differences in gas chemistry. The true nature of a double BSRs at this site is not known at this time, but its presence suggests that gas hydrate and free gas may coexist between the BSRs. Nevertheless, a double BSR cannot explain the coexistence free gas and gas hydrate throughout the GHSZ.
4. Sometimes drilling induces the dissociation of gas hydrate near the borehole and releases free gas (Collett et al., 1999). If this is the case, the amount of free gas recorded during or after drilling may be proportional to the amount of gas hydrate in sediments. Free gas would temporarily coexist with gas hydrate in close proximity to the borehole. This is one possible explanation for the presence of free gas within the GHSZ observed at these sites.
5. Vigorous streams of methane bubbles have been observed emanating from vents or the seafloor in southern summit region (Heeschen et al., 2003). The presence of methane bubbles beneath and at the seafloor suggest rapid transport of methane through the GHSZ along faults or fractures. Gorman et al. (2002) showed that the migration of methane gas through the GHSZ in a low-flux hydrate province is also possible (e.g., at the Blake Ridge). The presence of BSRs coupled with the proximity to the summit region means that free gas may migrate through the GHSZ along faults or fractures at these sites. Therefore, it is likely that free gas coexists with gas hydrate within the GHSZ in dynamic systems.

Based on the above arguments, it is probable that gas hydrate coexists with free gas either released by the dissociation of gas hydrate during drilling and/or transported along the fractures and faults.

As indicated in Table T2, the average gas hydrate saturations from S-wave velocities are 10.2% ± 3.7%, 10.4% ± 5.6%, and 6% ± 3.29% for Sites 1244, 1245, and 1247, respectively. The amounts of gas hydrate estimated from S-wave velocities in three holes are higher than those estimated from resistivities. The average saturation of 10% from the S-wave velocity at Site 1244 is ~50% higher than that from the resistivity log. The gas hydrate saturation estimated at Site 1244 using the chloride anomaly is between 2% and 8% (Shipboard Scientific Party, 2003), which is close to the estimate from resistivity. At Site 1245, low chloride saturation anomalies are interpreted to reflect in situ hydrate saturations below 3%, with one anomaly suggesting a saturation of 15%. It appears that the estimated gas hydrate saturations from S-wave velocities are higher than those from resistivities or chloride anomalies at these sites.

The difference between estimates from S-wave velocity and those from the resistivity comes from many factors, such as

1. The locations of LWD wells and CWL wells are not identical. For example, at Site 1247, the holes are ~75 m apart. If the gas hydrate saturations are heterogeneous, there may be some difference between estimates, but it is unlikely that all three wells have higher saturations from S-wave velocities.
2. The depth of investigation for CWL S-wave velocity and the LWD resistivity may be different. If drilling caused the dissociation of gas hydrate, the effect of dissociation would be more pronounced near the borehole, at the shallow depth of investigation.
3. The assumed parameter, m, for the BGTL is lower than it should be. This implies that the degree of consolidation at each well site is underestimated. Therefore, the gas hydrate saturations could have been overestimated. Even though the estimates from S-wave velocities are higher than those from resistivities, a similar variation of gas hydrate saturations with depth (Figure F6) suggests that the estimates are reasonable.

Figure F7 and Table T2 indicate that free gas saturations estimated below the GHSZ from P-wave velocity using the BGT or from both P-wave and S-wave velocities using the moduli method are similar. The actual amounts of free gas are dependent on the mode of gas distribution or the calibration constant e. Therefore, it is not possible to determine the accuracy of the estimated saturations depicted in Figure F7 and Table T2. However, Figure F7 with Table T2 enables us to examine the accuracy of each method under the same assumptions of gas distribution.

The BGT method with P-wave velocity computes the bulk modulus of sediments using the general Biot coefficient shown in equation 7 by incorporating the effect of differential pressure and consolidation or local conditions of sediments through the BGTL parameter n. On the other hand, the moduli method calculates the bulk modulus of sediment directly from the measured P- and S-wave velocities. Therefore, the moduli method is more accurate and reflects better in situ properties of sediments. A good agreement of saturations estimated from the two methods suggests that (1) the BGTL parameters used in this study are accurate and (2) the theory based on the BGT with parameters derived from the BGTL can be used to predict elastic velocities for partially gas-saturated sediments. Therefore, if both V_P and V_S are available for the estimation of gas saturations, the moduli method is preferable. However, in the case that only V_P is available for an analysis, the BGT based on the BGTL is a viable approach.

The amounts of free gas estimated from the P-wave velocity data depend on saturation or distribution model of the free gas in the pore space as well as the magnitude of velocity reduction. As shown in equation 9, the calibration constant controls the estimated free gas saturations. Lee (2004) indicates that an appropriate gas saturation model for partially saturated unconsolidated sediments is e = 8, based on the data by Domenico (1977). However, it is difficult to assess the accuracy of the calibration constant without other independent estimations of free gas, because e depends on the frequency (Gei and Carcione, 2003) as well as microstructure of the formation (Murphy et al., 1993). Usually, as the frequency of measurements increases, e decreases, but the precise relation depends on the data.

The dotted line in Figure F8 shows the saturations of the free gas estimated from NMR and density porosities using the method of Freedman (1997). Comparing the free gas estimated shown in Figure F7, the free gas estimations from the NMR technique are about four times greater than free gas saturations estimated from velocities. Note that the comparisons are not valid in the gas-hydrate-bearing intervals because the effect of gas hydrate is not included in the estimation of free gas using the moduli method. As indicated in Figure F8, the general trends of free gas with respect to depth are similar to each other but the details are different.

The saturations of free gas estimated from the NMR technique are independent of the mode of gas distribution in the pore space. Therefore, if the NMR and density porosities are accurate, the dotted line in Figure F8 can be used as a ground truth of free gas saturations. The solid line in Figure F8 is the amount of free gas saturations estimated using the moduli method with e = 2 instead of e = 8, as used in Figure F7. The gas saturations using the moduli method with e = 2 is similar to those from NMR and density porosities. If e = 2 instead of e = 8 is appropriate at these borehole sites, the average free gas saturations estimated from the elastic velocities or the moduli are about four times that of those shown in Table T2. However, the amounts of free gas have not been independently confirmed at these sites. Without any additional data to constrain the amounts of free gas at these sites, it is reasonable to say that estimations using e = 2 and e = 8 correspond to upper and lower limits, respectively.