Energy & Fuels 2005, 19, 459-470
459
Global Distribution of Methane Hydrate in Ocean Sediment Jeffery B. Klauda*,† and Stanley I. Sandler Department of Chemical Engineering, Center for Molecular and Engineering Thermodynamics, University of Delaware, Newark, Delaware 19716 Received August 9, 2004. Revised Manuscript Received November 29, 2004
In this paper, we present an equilibrium thermodynamic model to accurately predict the maximum depth of hydrate stability in the seafloor, including the effects of water salinity, hydrate confinement in pores, and the distribution of pore sizes in natural sediments. This model uses sediment type, geothermal gradient, and seafloor depth as input to predict the thickness of the hydrate zone. Using this hydrate model and a mass-transfer description for hydrate formation, we have also developed a predictive method for the occurrence of methane hydrates in the ocean. Based on this information, a prediction for the distribution of methane hydrate in ocean sediment is presented on a 1° latitude by 1° longitude (1° × 1°) global grid. From this detailed prediction, we estimate that there is a total volume of 1.2 × 1017 m3 of methane gas (expanded to atmospheric conditions), or, equivalently, 74 400 Gt of CH4 in ocean hydrates, which is 3 orders of magnitude larger than worldwide conventional natural gas reserves. Of this number, we estimate that 4.4 × 1016 m3 of methane expanded to standard temperature and pressure (STP) exists on the continental margins.
Introduction Gas hydrates are an interesting class of solid compounds that develop from water and one or more hydrophobic solutes. These solutes are trapped in hydrate cages, where the guest-host interactions are weak and there is sufficient free space for the guests to translate and rotate. Although 130 guests are known to form clathrate hydrates,1 our interest here is only in methane, because of its abundance in nature, specifically in the seafloor. The thermodynamic stability of methane hydrates in the seafloor, and the resulting gas hydrate stability zone (GHSZ), is a function of temperature, pressure, and water salinity. In regions where hydrates occur, the hydrostatic pressure at the seafloor exceeds the hydrate stability pressure at the seafloor temperature, so that the upper limit for hydrate formation is the seafloor above which there is insufficient methane to form hydrate. With increasing depth below the seafloor, the hydrostatic pressure increases linearly, and because of the geothermal gradient, so does the temperature. However, the hydrate equilibrium pressure increases approximately exponentially with temperature. Therefore, there is a maximum depth of hydrate stability that occurs where the hydrostatic and hydrate equilibrium pressures are equal. The thickness of the hydrate zone is then from the seafloor to this limit of hydrate stability.
Because methane hydrates in ocean sediment occur in porous media, for sufficiently small pore sizes, capillary effects are also important in determining the limit of hydrate stability and the hydrate thickness.2 Previous equilibrium pressure measurements in artificial porous media3,4 have shown an increase in equilibrium pressure above that in the bulk. The model we present here includes the effect of pore size. The global amount of methane contained in hydrates has been debated for many years. Most estimates have been based on a simple region-by-region extrapolation from the amounts of known hydrates in similar global regions.5,6 From the first estimates of gas hydrates in the early 1970s to recent assessments, the amount of estimated methane hydrate reserves has decreased from 3.0 × 1018 to 2.5 × 1015 m3 of methane at standard temperature and pressure (STP).6 The review of gas hydrate reserve estimates by Kvenvolden5 suggested that the consensus amount of methane trapped in hydrates is 1015 m3, and that the amount of methane trapped in permafrost hydrates is at least 2 orders of magnitude less than that in marine hydrates. In that work, a value of 2.1 × 1016 m3 of methane at STP was their best estimate, based on previously published gas hydrate reserve assessments. One of the studies used by Kvenvolden5 to obtain a consensus value was conducted by Harvey and Huang;7
* Author to whom correspondence should be addressed. E-mail:
[email protected]. Phone: (301) 496-9510. Fax: (301) 402-3404. † Currently with Laboratory of Computational Biology, NHLBI, National Institutes of Health, Bldg. 50, Rm. 3308, 50 South Drive, Bethesda, MD 20892. (1) Sloan, E. D. Clathrate Hydrates of Natural Gases, 2nd Edition; Marcel Dekker: New York, 1998.
(2) Clennell, M. B.; Hovland, M.; Booth, J. S.; Henry, P.; Winters, W. J. J. Geophys. Res.sSolid Earth 1999, 104, 22985. (3) Handa, Y. P.; Stupin, D. J. Phys. Chem. 1992, 96, 8599. (4) Uchida, T.; Ebinuma, T.; Ishizaki, T. J. Phys. Chem. B 1999, 103, 3659. (5) Kvenvolden, K. A. Proc. Natl. Acad. Sci., U.S.A. 1999, 96, 3420. (6) Milkov, A. V. EarthsSci. Rev. 2004, 66, 183.
10.1021/ef049798o CCC: $30.25 © 2005 American Chemical Society Published on Web 02/03/2005
460
Energy & Fuels, Vol. 19, No. 2, 2005
their study estimated the global amount of oceanic methane hydrate between the depths of 250 and 3000 m. In their predictions, the seafloor depth, seafloor temperature, and geothermal gradient were interpolated over a 1° latitude by 1° longitude (1° × 1°) grid and hydrates were assumed to form if the organic carbon content was >0.5 wt %. The available area conducive to hydrate formation then was arbitrarily reduced by 25%. Because the amount of hydrate filling of the pore space was unknown, three scenarios were used with a linear decrease in hydrate filling of the pore space from the base to the top of the stability zone: 10% to 2.5%, 20% to 5%, and 40% to 10%. Based on this data, they estimated the amounts of methane expanded to STP from the hydrate ocean reserves to be 23 × 1015 m3, 46 × 1015 m3, and 91 × 1015 m3, respectively.7 More recently, Dickens8 used an alternate approach to estimate the amount of gas hydrate in the seafloor. The hydrate volume was estimated for the continental margins assuming that the cross-sectional area of the hydrate stability zone from the coastline to the continental margin is globally representative of the average Atlantic margin, 104 km2. An isobath of 1000 m was used to determine the span of the global margins, resulting in an estimate of 200 000 km. Although many choices for the global average porosity, geothermal gradient, and hydrate saturation of the pore space were investigated, the most representative range of the predicted gas hydrate is 4.9 × 1015 m3 to 25 × 1015 m3 of global seafloor methane expanded to STP, based on a range of hydrate saturation of the pore space (1%5%) and assuming that all continental margins contain gas hydrates. The most recent estimate for seafloor methane hydrates, 3.0 × 1015 m3 of methane at STP (average of the range reported by Milkov6), used the same approach as Dickens,8 but an average value of 1.2% hydrate saturation of the pore space was used to determine the amount of hydrates on the continental margins. Moreover, this estimate assumed that only 20% of the GHSZ on the continental margins contain methane hydrate, which is based on an inventory of data from the Ocean Drilling Program (ODP) and Deep Sea Drilling Project (DSDP).9 Although this estimate was based on observed values of hydrates in the ocean, the study does not predict the locations of the ocean methane hydrates and assumes that the observations by the ODP/DSDP can be extrapolated to the entire continental margin. Because archaeabacteria produce methane, Gornitz and Fung10 used satellite imagery data across the planet on a 1° longitude by 1° latitude (1° × 1°) grid to estimate regions of high seafloor organic matter, based on sea level phytoplankton concentrations, and, thus, likely hydrate zones. We also assume that methane hydrates are formed from the in situ and proximate production of methane from archaea,11 referred to here as active seafloors; however, we use estimates of seafloor organic carbon obtained directly rather than surmised from sea(7) Harvey, L. D. D.; Huang, Z. J. Geophys. Res.sAtmos. 1995, 100, 2905. (8) Dickens, G. R. Org. Geochem. 2001, 32, 1179. (9) Borowski, W. S.; Paull, C. K.; Ussler, W., III. Mar. Geol. 1999, 159, 131. (10) Gornitz, V.; Fung, I. Glob. Biogeochem. Cycles 1994, 8, 335. (11) Borowski, W. S.; Paull, C. K.; Ussler, W. Mar. Chem. 1997, 57, 299.
Klauda and Sandler
level phytoplankton concentrations. The dependence of the organic carbon concentration on depth below the seafloor is determined by assuming that the loss of organic carbon is due to archaeabacterial conversion of organic matter to methane. This methane is produced over geologic time scales, as a result of continual sedimentation of organic matter from biota. Consequently, the amount of organic carbon decreases as the depth below the seafloor increases. Similar to our published approach for several regional estimates of methane hydrates,12 our global estimate for the amount of methane hydrate is based on a thermodynamic model to determine the thickness of the hydrate stability zones, and a mass-transfer model13 to estimate the extent of hydrate saturation of the seabed pore space. Passive seafloors, areas with fluxes of methane from deep natural gas seeps, are not considered in our calculations, because of the inability to predict these locations. We believe that our global hydrate predictions are an accurate representation of the locations and amounts of methane hydrates from active seafloors (production of methane within the hydrate stability zone) and possibly passive seafloors with low fluxes of methane supported by proximate archae methane production. Hydrate Thermodynamic Stability The most commonly used description of bulk gas hydrates is the statistical mechanical model by van der Waals and Platteeuw (vdWP)14 and extended by Parrish and Prausnitz15 to account for multiple guests. We have developed a classical thermodynamic model for bulk and confined media hydrates that is more accurate than the most recent parametrization of the vdWP model.12,16,17 In our model, the fugacities of water in the hydrate phase (H) and the liquid water (L) phase are set equal: L fH w (T,P) ) f w(T,P)
(1)
where the fugacity of water is
f Lw(T,P) ) xw(T,P)γw(xw,T)Psat,L (T) × w exp
[
]
V Lw(T,P)(P - Psat,L (T)) w (2) RT
The activity of water is computed using the Pitzer model for a single electrolyte in solution,18 as extended to mixed electrolytes by Patwardhan and Kumar,19 with seawater ion concentrations obtained from the CRC Handbook.20 Additional details on the calculations of the fugacity of bulk liquid water can be found elsewhere.17 Naturally occurring hydrates form as nodules in the sediment that progressively protrude into the pore with (12) Klauda, J. B.; Sandler, S. I. Mar. Pet. Geol. 2003, 20, 459. (13) Davie, M. K.; Buffett, B. A. J. Geophys. Res.sSolid Earth 2001, 106, 497. (14) van der Waals, J. H.; Platteeuw, J. C. Adv. Chem. Phys. 1959, 2, 1. (15) Parrish, W. R.; Prausnitz, J. M. Ind. Eng. Chem. Process Des. Dev. 1972, 11, 26. (16) Klauda, J. B.; Sandler, S. I. Ind. Eng. Chem. Res. 2001, 40, 4197. (17) Klauda, J. B.; Sandler, S. I. Ind. Eng. Chem. Res. 2000, 39, 3377. (18) Pitzer, K. S.; Mayorga, G. J. Phys. Chem. 1973, 77, 2300. (19) Patwardhan, V. S.; Kumar, A. AICHE J. 1986, 32, 1419. (20) Lide, D. R.; Frederiskse, H. P. R. CRC Handbook, 76th Edition; CRC Press: Boca Raton, FL, 1995.
Distribution of Methane Hydrate in Ocean Sediment
Energy & Fuels, Vol. 19, No. 2, 2005 461
a convex spherical cap.2 If the pores are small, these nodules will have a higher pressure than the bulk, P, given by
PH pore(r) ) P -
ζHLσHL cos θHL r
(3)
where PH pore(r) is the pressure in the hydrate phase with a pore radius of r, ζ the shape factor, σ the surface tension, and θ the contact angle of the hydrate/liquid water (HL) interface (which is >90°, because the contact surface is convex). This is different from hydrate growth in artificial porous media, because, in that case, hydrates grow from the outside of the porous material into the pores.16 The liquid water then is discontinuous from the bulk and the pressure in the liquid water phase is reduced below that of the bulk. Because no other information is available, the surface tension of the HL interface is assumed to be equal to the surface tension between ice and liquid water (27 mJ/m2).2,3 We also assume that the hydrate phase forms a convex spherical cap; therefore, ζHL ) 2 and θ ) 180°. Natural porous media consist of a distribution of pore sizes; therefore, an improved prediction for hydrate equilibrium is obtained by including the pore distribution.16 We assume that the increase in pressure of the confined hydrate phase is taken into consideration by the Poynting correction in our fugacity-based model:
Figure 1. Gas hydrate stability zone (shaded gray) for a geothermal gradient (GG) of 0.0345 °C/m.
where rj is the average pore radius and xclay is the fraction of clay-sized particles in the sediment sample. (Pressure P in given in units of MPa, and rj is given in angstroms.) Natural sea sediment pore size distributions can be bimodal, especially for samples with a small percentage of clay sediment and high amounts of sand or silt.21,22 These biomodal distributions typically have a single peak, with an average pore size of >300 Å and another broad peak at a larger pore radius. This resulted in an increase in the equilibrium pressure predictions by less than a few percent, compared to a
unimodal distribution. Therefore, the effect of a bimodal distribution is negligible and a single normal distribution peak will be used here. In our calculations, we have used values for the percentage of clay that were interpolated from 250 global data points reported by the DSDP at methane hydrate stability depths below the seafloor.23 The standard deviation of the pore sizes was assumed to be 20 Å; this choice has the largest impact on the calculations for permafrost hydrates near the freezing point of water.16 However, such hydrates do not make a large contribution to our total estimates. The stability of a methane hydrate within the seafloor is dependent on the temperature, pressure, and salinity at a given depth. As the depth below the ocean floor increases, the hydrostatic pressure increases linearly and the temperature increases as a result of the local geothermal gradient. For a given depth below the seafloor, the temperature is first found from the geothermal gradient, and then the hydrate equilibrium pressure is obtained by solving eq 1 with eqs 2 and 3 simultaneously for pressure, using a constant concentration of the salt ions.20 The hydrate is thermodynamically stable when the equilibrium pressure is less than or equal to the hydrostatic pressure, as shown by the gray shaded region in Figure 1. There exists a depth below which the hydrate is not thermodynamically stable because the increasing equilibrium pressure (resulting from the increasing temperature) exceeds the hydrostatic pressure. This maximum depth of hydrate stability is dependent on the geothermal gradient in the sediment. As shown in Figure 1, when the geothermal gradient is increased slightly by 0.002 °C/m, the maximum depth of hydrate stability is decreased from 500 m below the seafloor (mbsf) to 460 mbsf. We have compared our predictions of methane hydrate stability zones to those measured by the ODP and indirect observations from seismographic studies.12,16 Seismic seafloor data are used to locate a region of strong reflections of sound waves due to free methane gas (which is referenced as a bottom-simulating reflector, or BSR) above which hydrates are stable. From
(21) Dewhurst, D. N.; Aplin, A. C.; Sarda, J. P. J. Geophys. Res.s Solid Earth 1999, 104, 29261. (22) Dewhurst, D. N.; Aplin, A. C.; Sarda, J. P.; Yang, Y. L. J. Geophys. Res.sSolid Earth 1998, 103, 651.
(23) Musich, L.; Bearman, S.; Birtley, T.; Hawkins, D.; Long, B.; Marsee, D.; Pinkston, B.; Wood, T.; Woodbury, P. Core Data from the Deep Sea Drilling Project; World Data Center for Marine Geology and Geophysics: Boulder, CO, 2000; Vol. 2001.
fH w (T, P) )
[
]
sat,β VH w (T,P)(P - Pw (T)) × RT VH ∞ w ζHLσHL exp cos θHL φ(r) dr (4) 0 RT r
f H,bulk (T,P) exp w
∫
[
]
where φ(r) is the normal probability distribution function for pore size, V H w (T,P) is the molar volume of (T) is the fugacity of hydrate in the hydrate, and f H,bulk w bulk (details of this calculation can be found elsewhere17). The pore sizes in natural sediment were determined from Dewhurst and co-workers21,22 for London Clay of varying sediment types. A correlation for the pores sizes as a function of clay composition and pressure was fit to their data:16
ln(rj) ) 15.4215 - 21.9773xclay + 11.5670x2clay + 0.2 exp(-0.0278P) (5)
462
Energy & Fuels, Vol. 19, No. 2, 2005
Klauda and Sandler
Figure 2. Phytoplankton pigment global distribution.
ocean drilling data, the depth of the BSR corresponds closely to the deepest occurrence of hydrate in the seafloor.24 Therefore, the BSR is taken to represent the base of hydrate stability:25 below that depth, the hydrostatic pressure is less than the hydrate equilibrium pressure. Hydrates are found in seafloor regions with no observed BSRs;26 in such cases, little or no free gas is required to have strong reflections of sound waves. Our model is a slight improvement over that of the vdWP model in sandy sediment, where the effect of pore size is negligible,16 but greatly improves the accuracy of the predicted hydrate stability zones in clayey sediment, such as the Blake Ridge Region.24 If the effect of hydrate confinement in pores is ignored, the maximum depth of hydrate thermodynamic stability is predicted to be too large by 100 m for ODP site 995 at the Blake Ridge.24 Therefore, the pore size distribution of the sea sediment should be included for clayey sediment. Overall, our predictions for the hydrate depth of seven ODP sites have an absolute average deviation (AAD) of 5%, compared to measured data, which is less than the 12% AAD using the common approach of assuming the phase behavior in pores is identical to that in the bulk.12,16 Sources of Methane in Ocean Sediment Ocean hydrates develop from in situ and proximate methane production from archaea or percolation through (24) Paull, C. K.; Matsumoto, R.; Wallace, P. J.; Black, N. R.; Borowski, W. S.; Collett, T. S.; Damuth, J. E.; Dickens, G. R.; Egeberg, P. K.; Goodman, K.; Hesse, R. F.; Hiroki, Y.; Holbrook, S. W.; Hoskins, H.; Ladd, J.; Lodolo, E.; Loreson, T. D.; Musgrave, R. J.; Nahr, T.; Okada, H.; Pierre, C.; Ruppel, C.; Satoh, M.; Thiery, R.; Watanabe, Y.; Wehner, H.; Winters, W. J.; Wood, W. T. Proc. Ocean Drill. Program: Initial Rep. 1996, 164, 99. (25) Ruppel, C. Geology 1997, 25, 699. (26) Milkov, A. V.; Sassen, R. Mar. Geol. 2001, 179, 71.
the sediment from deep sources of methane. Our predictions are based only on in situ and proximate methane production, because it is not possible to predict global locations of deep methane sources. The amount of organic carbon at the seafloor and in ocean sediment directly determines the ability of archaea to convert organic carbon to methane. Therefore, methane hydrate obtained from methanogenic archaea will occur only in areas where sufficient organic carbon is available by continual sedimentation of organic material. As a result, our model does not predict hydrate formation in passive seafloors that do not have sufficient organic carbon to be converted to methane. In addition, our model may underpredict the volumes of methane hydrate at locations with high fluxes of methane, e.g., Hydrate Ridge27 and some areas in the Gulf of Mexico.26 Therefore, our model serves as a lower bound to the total amount of marine methane hydrates, because of the possible increase in the amount from these additional sources. Gornitz and Fung10 used the sea-level phytoplankton concentrations to determine locations of sufficient seafloor organic carbon to produce methane hydrates. Figure 2 contains the average of sea-level phytoplankton concentrations over the lifetime of the coastal zone color scanner satellite (1978-1986). They suggested that phytoplankton concentrations of >0.5 mg/m3 correspond to the organic seafloor concentrations of >0.5 wt %, which is believed to be necessary for hydrate formation.28-30 Figure 3 contains estimates for the total organic carbon (TOC) at the seafloor, compiled from (27) Milkov, A. V.; Claypool, G. E.; Lee, Y. J.; Xu, W.; Dickens, G. R.; Borowski, W. S. Geology 2003, 31, 833. (28) Collett, T. S. Gas Hydrate Resources of the United States. In 1995 National Assessment of United States Oil and Gas Resources on CD-ROM; Gautier, D. L., Dolton, G. L., Takahashi, K. I., Varnes, K. L., Eds.; USGS Digital Data Series, 1995; Vol. 30; p 85. (29) MacDonald, G. Clim. Change 1990, 16, 247.
Distribution of Methane Hydrate in Ocean Sediment
Energy & Fuels, Vol. 19, No. 2, 2005 463
Figure 3. Total organic carbon (TOC) distribution on the seafloor (>0.4 wt %). Data compiled from refs 31 and 32.
direct measurements of Romankevich31 and Premuzic et al.32 Comparing Figures 2 and 3, we see that there is only a weak correlation between sea-level phytoplankton and seafloor organic carbon. Many areas south of Central America and in southeast Asia have high seafloor organic carbon but low sea-level phytoplankton concentrations, whereas both the northern Atlantic Ocean and the Pacific Ocean south of Alaska have high sea-level phytoplankton concentrations but low seafloor concentrations of organic carbon. Therefore, we used direct measurements of seafloor organic carbon to determine areas of possible hydrate formation, which agrees with a similar method used by Collett28,33 to predict the amount of seafloor gas hydrate. The amount of organic carbon in the seafloor is essential to determine the amount of hydrate filling of the pore space, as will be discussed in the following section. There is a continual renewal of organic carbon on the ocean floor from the sedimentation of organic matter, which is taken into consideration here, and this renewal over geologic time scales has resulted in the deposit of organic matter deep below the surface sediment.13 The amount of organic carbon throughout the sediment may be dependent on time, because of changes in ocean biological productivity. Current seafloor organic (30) Revelle, R. R. Methane Hydrates in Continental Slope Sediments and Increasing Atmospheric Carbon Dioxide. In Changing Climate, Report of the Carbon Dioxide Assessment Committee; National Academic Press: Washington, DC, 1983; p 252. (31) Romankevich, E. A. Geochemistry of Organic Matter in the Ocean; Springer-Verlag: Berlin, 1984. (32) Premuzic, E. T.; Benkovitz, C. M.; Gaffney, J. S.; Walsh, J. J. Org. Geochem. 1982, 4, 63. (33) Collett, T. S. Geologic Assessment of the Natural Gas Hydrate Resources in the Onshore and Offshore Regions of the United States. In Proceedings of the 2nd International Conferences on Natural Gas Hydrates, Toulouse, France, 1996.
carbon data and sedimentation rates were used in our calculations, because how the sedimentation of organic carbon varies over geologic time scales is not well understood. We have found, using the mass-transfer model of Davie and Buffett,13 discussed in the following section, that, for organic carbon contents of
