Three downhole tools were used to acquire temperature data at a nominal vertical spacing of 30-50 m to depths as great as ~415 mbsf at Sites 994, 995, and 997. The deepest measurements at both Sites 995 and 997 were to within 55 m above the BSR and provide an unprecedented constraint on temperatures deep within a thick GHZ. Details of the data acquisition and preliminary analysis can be found in Paull, Matsumoto, Wallace, et al. (1996), and Ruppel (1997). The Leg 164 temperature data can be used to constrain the conductive and advective components of heat flux, to evaluate disparities between traditional heat flow data (Ruppel et al., 1995) and in situ measurements, and to determine temperatures at the BSR.
Figure 3 shows the equilibrium temperature data at Sites 994, 995, and 997. The highly linear (uniform) nature of the temperature vs. depth curves is consistent with conduction-dominated heat transfer deeper than 30 mbsf. Calculated thermal gradients are 33.3 ± 1.5° C km-1, 32.7 ± 0.7° C km-1, and 36.9 ± 0.7° C km-1 for Sites 994, 995, and 997, respectively, and the sediments are not in equilibrium with the bottom-water temperatures (BWT) measured at the time of drilling (Ruppel, 1997). Conductive gradients out of thermal equilibrium with bottom water were previously noted in a traditional marine heat-flow survey that acquired temperature and in situ thermal conductivity measurements in the upper 3-5 m of sediment within 1-2 km of each drill site (Ruppel et al., 1995). The lack of thermal equilibrium between BWT and the sediments in both data sets is consistent with large (up to ± 0.4° C), rapid (timescale of weeks) fluctuations in BWT in this area (Ruppel et al., 1995), an inference confirmed by a 6.25-yr temperature record obtained on the Blake Plateau, south of the Leg 164 study area (Broek, 1969).
An important result that emerges from the comparison of the Leg 164 downhole temperature data (C. Ruppel, unpubl. data) and the traditional heat-flow measurements (Ruppel et al., 1995) is a disparity of up to 30% between the thermal gradients determined from the two data sets. Variations in thermal conductivity between the shallow and deep parts of the section cannot explain this disparity. The thermal gradients measured in the upper 3-5 m at the transect site (Ruppel et al., 1995) are remarkably consistent and cover the narrow range from 46.4° to 47.8° C km-1 with uncertainties as large as ±1.5° C. These shallow data reveal no statistically significant increase in thermal gradient between Sites 994/995 and Site 997, in contrast to the ~10% increase in gradient measured between these locations on Leg 164. At Site 997, the vertical and lateral distribution of gas hydrate and seismic images imply that the holes may intersect a buried fault system that could focus upwelling fluids and lead to an increase in the deep thermal gradient at this site. The lateral resolution of the traditional heat flow data set is probably too low to detect such focused flow near Site 997, even if this flow did continue to near-surface depths. It should also be noted that for the traditional marine heat flow data to lie on the same temperature profile as the downhole temperature data requires significant nonuniformity (concave down gradient) in the composite thermal gradient between the seafloor and 30-50 mbsf.
Three independent estimates of temperature at the BSR can be obtained from Leg 164 and peripheral data sets (C. Ruppel, unpubl. data). The simplest method uses the technique described by Yamano et al. (1982) to determine the BSR temperature by combining the gas hydrate phase equilibria and constraints on the depth of the BSR below the seafloor. Assuming that pressure in the sediments is hydrostatic, the BSR temperatures estimated using this method are 21.6° -22.7șC at Sites 995 and 997 for stability boundaries based on the Brown et al. (1996) fit to the Dickens and Quinby-Hunt (1994) results for 3.3% NaCl water and on Sloan (1998) for freshwater. The second estimate can be derived from downward extrapolation of the shallow thermal gradients measured by Ruppel et al. (1995) to the depth of the BSR. This calculation yields BSR temperature estimates of ~24° and ~25.5° C at Sites 995 and 997, respectively. The third and probably most accurate estimate is based on downward projection of the Leg 164 in situ temperature data and yields BSR temperatures of 17.4° -18.8° C and 19.8° -21.1° C at Sites 995 and 997, respectively (Ruppel, 1997).
The disparities between these BSR temperature estimates highlight several problems with using data other than those collected in deep boreholes to constrain the thermal state of sediments within a GHZ. First, if the in situ data acquired on Leg 164 are presumed to provide the most accurate constraint on BSR temperatures, then even very high quality traditional heat flow data may yield an overestimate of actual BSR temperatures (C. Ruppel, unpubl. data). It is striking that this result was obtained in a low advective flux setting (Egeberg and Dickens, 1999) in a sediment drift deposit near a tectonically quiescent continental margin. BSR temperature estimates based on downward extrapolation of surface heat flow may be even more inaccurate in active margin gas hydrate provinces characterized by high fluid-flux rates, variegated sediments, and tectonic deformation. Second, the method of Yamano et al. (1982) only provides a rough measure of the thermal gradient in the sediments above the BSR if the approximate BWT is known and if sediments can be assumed to be in thermal equilibrium with the bottom water. Third, the BSR temperatures at Sites 995 and 997 lie 2.9° C (Ruppel, 1997) below those predicted at the base of the GHSZ using presently accepted stability curves (Dickens and Quinby-Hunt, 1994; Sloan, 1997).
Several physical and chemical processes have been invoked to explain low temperatures at the Blake Ridge BSR. Capillary forces in the fine-grained sediments (e.g., Clennell et al., in press) or controversial third-surface effects associated with the high concentration of clay minerals (Cha et al., 1988; Kotkoskie et al., 1990) may inhibit gas hydrate stability. Alternately, the phase equilibria for systems at these pressures (~32.2 MPa) may still be too poorly known to draw reliable conclusions about the significance of the apparently cold temperatures measured deep within the GHZ, close to the BSR. Indeed, new theoretical curves emerging from statistical thermodynamics calculations (Tohidi et al., 1995) yield dissociation temperatures much closer to those determined for the Blake Ridge BSR although capillary effects may still need to be invoked to explain the full magnitude of the apparent temperature deficit at the BSR (Clennell et al., in press).