Gas hydrate dissociation prolongs acidification of the Anthropocene oceans

1 Introduction

Gas hydrates, also known as clathrates, are solids that consist of gas molecules surrounded by cages of water molecules [Sloan and Koh, 2007]. Methane (CH₄) hydrates are stable at the relatively low temperatures and relatively high pressures that exist in much of the world's oceans, and they are found in sediments over large areas of the modern continental slopes and rises [Kvenvolden, 1993; Beaudoin et al., 2014]. The total mass of carbon stored in present-day subseafloor gas hydrate systems remains the subject of debate, as obvious from the recent literature. For example, using geochemical modeling, Piñero et al. [2013] and Kretschmer et al. [2015] estimated 550 and 1146 gigatons of carbon (Gt C = 10¹⁵ g C), respectively, in oceanic gas hydrates. However, this work implies <60 kg of methane C m⁻² everywhere across the seafloor, whereas measurements at multiple drill sites indicate >300 kg of methane C m⁻² of the seafloor [Dickens et al., 1997; Milkov et al., 2003; Malinverno et al., 2008], even after excluding underlying free methane gas. Beaudoin et al. [2014] note these issues in their extensive review, emphasizing that the gas hydrate standing stock in modern oceanic sediments is uncertain, but may be ~5000 Gt C. Our study employs a judicious mean value reported by Yamamoto et al. [2014], which was ~2680 Gt C.

Concerns exist that ongoing climate warming will increase the temperature of the oceans, thus causing presently stable hydrates to dissociate and release part of their total subseafloor CH₄ pool to the oceans and atmosphere [Krey et al., 2009; Mascarelli, 2009; Ruppel, 2011; Biastoch et al., 2011; Hunter et al., 2013; Kessler, 2014; Yamamoto et al., 2014]. Archer [2007] has argued not only that such destabilization is likely if current fossil fuel emission scenarios hold but also that CH₄ leakage from the seafloor will be chronic—taking place over hundreds to thousands of years. With chronic release, the CH₄ that escapes from the seafloor will most likely be rapidly and efficiently oxidized to CO₂ within the oceans, or perhaps partially in the atmosphere [Dickens, 2001; McGinnis et al., 2006; Kessler et al., 2011; Zeebe and Ridgwell, 2011; Yamamoto et al., 2014]. Aerobic CH₄ oxidation produces CO₂,

(1)

which would enhance anthropogenic acidification of the oceans. (For those unfamiliar with CO₂-induced acidification of the oceans, we direct the reader to the highly readable expositions by Feely et al. [2009] and Gattuso and Hansson, [2011].) Methane hydrate-related acidification may have occurred during past episodes of global warming, most notably the Paleocene-Eocene Thermal Maximum, circa 56 Ma [Dickens et al., 1995; Zeebe et al., 2009].

The potential effects of the release and oxidation of CH₄ from gas hydrates on the future carbonate chemistry of the global oceans have received only limited consideration [Biastoch et al., 2011; Zeebe and Ridgwell, 2011]. Specifically, it is not known how much the pH and carbonate saturation state of the surface and deep waters will further fall with this additional CO₂. Moreover, the timing and duration of any such effects remains uncertain.

2 Model

This paper examines these questions using the biogeochemical model developed in Boudreau et al., [2010a], which was designed to simulate the oceanic and atmospheric CO₂ system over millennial timescales. To this end, the model includes a simplified and computationally efficient description of the carbonate compensation feedback by seafloor sediments. The current version of this model differs somewhat from that in Boudreau et al. [2010a], as a result of an improved description of the dynamics of the snowline (the water depth where the sedimentary CaCO₃ falls to zero)—see the supporting information for details. Our model is capable of predicting the CaCO₃ content of seafloor sediments with oceanographic depth and the three critical carbonate horizons [Boudreau et al., 2010b]: the saturation horizon (Z_sat) below which oceanic waters are undersaturated with respect to CaCO₃, the carbonate compensation depth (Z_cc), where seafloor dissolution just balances the downward rain of CaCO₃, and the depth at which the CaCO₃ content of the sediment falls to 10% (Z₁₀), as an operational surrogate for the snowline. Past results with this model are entirely consistent with those from more complex models, such as those presented by Archer et al. [1997, 2009b]—see discussion in Boudreau et al. [2010a].

To account for warming feedbacks, our model includes temperature as a variable for the atmosphere and each ocean reservoir—see Figure 1 and Figures S1–S5 in the supporting information. While the atmosphere is treated as compositionally homogeneous, this box was subdivided into high- and low-latitude domains for the purpose of temperature modeling (Figure 1a). The temperature, T (°C), in each of the atmospheric boxes at any given time was related to the partial pressure of CO₂, using the formula T = cln(P/P_o) + T_o [Scheffer et al., 2006], where T_o is the initial (preindustrial) year-averaged temperature in a chosen atmospheric subbox, P_o is the initial (preindustrial) P_CO2, P is the P_CO2 at a subsequent time (atm), and c is a scaling coefficient. We employed values of c = 1.7 and 4.7 in different simulations, as these might represent “short-term climate” and “long-term Earth system” sensitivities. As it turns out, almost identical results are obtained with respect to the carbonate system with either c value (Figures S2–S5).

The anthropogenic component of the added CO₂ in our model follows the IS92a emissions scenario as extrapolated by Boudreau et al. [2010a]; this injection (their Figure 2) introduces ~4025 Gt C as CO₂ over ~600 years and closely resembles other estimates of this forcing [e.g., Caldeira and Wickett, 2003; Feely et al., 2004]. Additional CO₂ release (Figure 1b) from methane hydrate dissociation and oxidation was provided by a smooth fit to the release-versus-time prediction in Yamamoto et al. [2014]. While the location of the oxidation of methane from marine sediments is most likely the oceans, we examined two end-member cases: a deep oceanic scenario, whereby all CH₄ oxidation occurred within the deep water column, and an atmospheric scenario, whereby all the marine-hydrate-generated CH₄ was oxidized within the atmosphere. The total CO₂ released from methane hydrate dissociation and oxidation (Figure 1b) is ~1600 Gt C over 13 kyr, which represents ~60% of the total gas hydrate reserves estimated by Yamamoto et al. [2014].

3 Results

The predicted CO₂ partial pressure in the atmosphere (P_CO2) is displayed for three “end-member” scenarios (Figure 2a): these are the prescribed IS92a CO₂ emissions with (i) no CH₄ release, (ii) seafloor CH₄ release, but immediate transfer to and oxidation in the atmosphere, and (iii) seafloor release of CH₄ and oxidation in the deep ocean. We recognize that scenario (ii) is highly unlikely given the rapid rates of bubble dissolution and CH₄ oxidation witnessed in the ocean water column, but it is included for completeness. All scenarios produce P_CO2 peaks of 1240 ppmv to be reached in 2210 A.D., which is consistent with previous model predictions [e.g., Archer, 2005]. Methane-generated CO₂ does not noticeably affect this P_CO2 maximum, either in timing or in magnitude, because the hydrate dissociation process has a built-in delay of about 1000 years [Archer et al., 2009a; Yamamoto et al., 2014], as a result of the need to transfer heat into the seafloor where the gas hydrates are located. After the maximum atmospheric P_CO2 is reached, all scenarios follow a decline to attain a quasi steady state between years 8000 and 15000 A.D. This decline is slightly more gradual with the atmospheric release scenario, as CH₄-derived CO₂ needs to be adsorbed into surface waters and transferred into the deeper ocean. As expected, the quasi steady state P_CO2 differs between scenarios. The IS92a quasi steady value is ~390 ppmv, while both methane-affected scenarios are ~94 ppmv greater, reflecting the additional carbon input.

The addition of CO₂ to seawater, either directly or via the atmosphere, causes short-term and long-term drops in both the pH and the carbonate saturation state of the oceans. IS92a-driven CO₂ release, with or without a contribution from CH₄ release and oxidation, decreases the pH of the low-latitude surface ocean by ~0.575 by 2210 AD (Figure 2b), which is consistent with predictions by Caldeira and Wickett [2003], Feely et al. [2004], and Zeebe et al. [2008]. However, pH values slowly recover thereafter, and the long-term pH value of the surface ocean, i.e., at 15000 A.D., is only ~0.05 lower under the IS92a scenario (Figure 2b) when compared to preindustrial times, but ~0.12 lower with the additional acidification from CH₄-derived CO₂.

The aragonite saturation index, Ω_a, will drop in the surface ocean in line with the P_CO2 increase (Figure 2c), where Ω_a = [Ca²⁺][CO₃²⁻]/K_a in which square brackets indicate dissolved concentrations, and K_a is the solubility product for the aragonite form of CaCO₃. If Ω_a > 1, the water is supersaturated with respect to aragonite and it is thermodynamically favored to precipitate; if Ω_a < 1, the water is undersaturated with respect to this mineral and it is thermodynamically favored to dissolve. Under the IS92a scenario alone, the surface oceans will approach saturation with respect to aragonite (Ω_a = 1) by year 2200, before recovering steadily toward a value slightly below preacidification values (Figure 2c). Orr et al. [2005] previously predicted that southern ocean surface water could become undersaturated, whereas we find that, on average, ocean waters will remain marginally saturated. This does not mean that (locally) undersaturated waters do not currently exist (e.g., west coast of North America) or cannot be expected in the future, through the effects of upwelling or biogeochemistry; such spatial variation is not captured by the current box model. If gas hydrate dissociation is added to the IS92a emissions, the recovery is delayed and the long-term saturation index is ~0.025 units lower (Figure 2c).

Deep-ocean carbonate chemistry is also affected by both anthropogenic and hydrate-related CO₂ (Figures 2d–2f). With time, the hydrate-based CO₂ release increases ΣCO₂ of the deep oceans by ~0.15 mM over the IS92a source alone; similarly, carbonate alkalinity is some 0.15 meq L⁻¹ higher, due to additional dissolution of seafloor carbonates. These changes in the chemistry of the deep ocean also affect the evolution of the positions of the critical carbonate horizons (Figure 3). The saturation horizon (Z_sat) rapidly rises, reaching a minimum depth of ~722 m by Year 2950 with IS92a forcing alone, deepening thereafter (Figure 3a). In the scenarios with methane release, Z_sat rises to a shallower depth of ~567 m on the same time frame, and deepens more slowly afterward. This Z_sat behavior is very similar whether CH₄ oxidation occurs in the atmosphere or in the oceans. Anthropogenic CO₂ release certainly dominates the upward migration of the saturation horizon; the deep ocean will be almost entirely undersaturated with respect to aragonite and calcite for a period of several millennia. Hydrate-based CO₂ will offset the return to the preindustrial position by ≥300 m at any selected time before year 15,000. Accordingly, the seafloor will become a more corrosive environment for longer periods of time, to the detriment of CaCO₃-secreting benthic organisms, such as deep water corals [Hall-Spencer et al., 2008; Guinotte and Fabry, 2008].

The carbonate compensation depth (Z_cc in Figure 3b) will rise to ~2.2 km by year 2950 in all release scenarios; however, CO₂ added from hydrates will again offset the subsequent return to initial conditions, amounting to ~300 m between the years 8000 and 15000 A.D. The increased undersaturation will lead to further dissolution of previously deposited CaCO₃ in seafloor sediments. As a result, the depth where the CaCO₃ content of the sediments is 10% (Z₁₀ in Figure 3c) will be shallower by an additional ~50 m in year 15,000.

Other possible changes in ocean biogeochemistry could act, in concert with hydrate dissociation, to produce stronger changes, and we investigate two possibilities. First, experimental evidence indicates that the production of CaCO₃ in the surface ocean may decrease in response to ocean acidification [Riebesell et al., 2000; Orr et al., 2005; Hall-Spencer et al., 2008; Riebesell, 2008; Waldbusser et al., 2015]; such a drop in production should lead to diminished CaCO₃ export from the surface to the deep waters. Boudreau et al. [2010a] showed that lower CaCO₃ production/export could be an important factor in enhancing anthropogenic changes in the future oceans. To explore this effect further, we assume that CaCO₃ export from the surface waters to the deep sea, B(t), scales as

(2)

where t is time, B_o is the preindustrial export (Gmol a⁻¹), Ω_a(t) is the aragonite saturation of the surface oceans at time t, Ω_o is the preindustrial value of Ω_a, and n is a constant that determines the degree of dependence of B(t) on the surface saturation. Values of n are unknown; in fact, equation 2 is quite speculative, but it can capture a wide variety of possible responses. For the latter reason, we performed simulations with n = 0, 0.5, 1, and 2.

Figures 4a and 4b illustrates the effects of reduced CaCO₃ export on the evolution of the saturation horizon and the carbonate compensation depth for the three basic scenarios. The curves with n = 0 have B(t) constant and correspond to the simulations in Figures 2 and 3. Changing n to 0.5, 1, or 2 modestly raises the position of the saturation horizon at the atmospheric P_CO2 maximum; conversely, there is a dramatic rise in the carbonate compensation depth, such that it is shallower than 1 km with n = 2. Such a rise would induce dissolution of previously deposited carbonate at much shallower depths. On a longer timescale, the differences between curves with different n values in Figure 4 are generally small and do not enhance the effect of hydrate-derived CO₂ addition to the atmosphere-ocean system. This result indicates that the effects of anthropogenic acidification are sensitive to changes in CaCO₃ export (Figure 1a), as found in Boudreau et al. [2010a], but that this is not true for the effects of chronic CO₂ release from gas hydrates.

There has been speculation that meridional overturning (conveyor-belt circulation) of the oceans may decrease, at least temporarily, in a warming climate [McPhaden and Zhang, 2002; Bryden et al., 2005; Srokosz et al., 2012; Ramstorf et al., 2015]. We have added this effect to our model (Figures 4c and 4d) by dropping the rates of overturning and high-latitude deep-ocean mixing (see U_T and U_M in Figure S1) to 75%, 50%, and 25% of the present day value (via a linear function of sea surface temperature). Such changes lead to marginal shifts in the timing of the minimum depths for the saturation horizon and the carbonate compensation depth, but they do not enhance the effects of slow additional CO₂ from gas hydrate dissociation.

As stated above, our findings are based on the Yamamoto et al. [2014] estimate of the gas hydrate inventory, as well as their release scenario. If a smaller inventory of gas hydrates is instead correct [Piñero et al., 2013], then the effects will be decidedly smaller. Conversely, a doubling of Yamamoto et al. [2014] emissions, equivalent roughly to a doubling of the gas hydrate reservoir [Beaudoin et al., 2014], causes approximately a doubling of the long-term offsets in the deep and surface water pH values and the total CO₂ in the deep water (see Figure S6), such that these effects become an appreciably more important consideration.

4 Conclusions

We conclude that if anthropogenic warming induces dissociation of seafloor gas hydrates and significant release of CH₄ to the ocean or atmosphere [e.g., Archer, 2007; Yamamoto et al., 2014], ocean acidification will occur over a longer time interval; that is, such a release would add a long (time) tail to the acidification. Consequently, the deep ocean will remain more corrosive to calcifying organisms for several millennia. However, the additional effects upon carbonate saturation caused by hydrate-derived CO₂ generally pale in comparison to those predicted from anthropogenic CO₂ emissions. This is principally because of the difference in the duration of CO₂ injection. Notwithstanding these findings, three central uncertainties about future ocean acidification remain: first, the response of CaCO₃-secreting organisms to increasing acidification; second, the true mass of seafloor CH₄ susceptible to thermal perturbation; and third, despite the work of Yamamoto et al. [2014], the exact magnitude and timing of CH₄ escape.

Erratum

In the originally published version of this article, some text in the first paragraph of the Introduction was incorrectly typeset. The following has since been corrected, and this version may be considered the authoritative version of record. In section 1, the first paragraph, the phrase “<60 kg of methane C m⁻² of the seafloor, whereas measurements at multiple drill sites indicate >300 kg of methane C m⁻² everywhere across the seafloor” has been changed to “<60 kg of methane C m⁻² everywhere across the seafloor, whereas measurements at multiple drill sites indicate >300 kg of methane C m⁻² of the seafloor”.