GEOPHYSICAL RESEARCH LETTERS, VOL. 35, L19806, doi:10.1029/2008GL035291, 2008

Estimating the potential for clathrate instability in the

1%-CO2 IPCC AR-4 simulations Jean-Franc¸ois Lamarque1 Received 9 July 2008; accepted 17 August 2008; published 3 October 2008.

[1] The recent work of Reagan and Moridis (2007) has clathrate destabilizes and the methane can be released into shown that even a limited warming of 1 K over 100 years the ocean water. Under a scenario, one can lead to clathrate destabilization, leading to a significant mechanism for the destabilization is a warming of the ocean flux of methane into the ocean water, at least for shallow seafloor, in response to external forcings such as the release deposits. Here we study the potential for of greenhouse gases (and CO2 in particular) in the atmo- destabilization by identifying the 100-year temperature sphere [Revelle, 1983; Harvey and Huang, 1995]. increase in the available IPCC (Intergovernmental Panel [3] In a recent paper, Reagan and Moridis [2007] (herein- on Climate Change) AR-4 1%-CO2 increase per year (up to after referred to as RM) used a comprehensive model of doubling over pre-industrial conditions, which occurs after methane under the seafloor to identify the potential flux from 70 years) simulations. Depending on assumptions made on methane clathrate destabilization that would be associated with the possible locations (in this case, only depth) of methane a warming of the ocean bottom waters. They found that, for clathrates and on temperature dependence, our calculation shallow deposits (depth <320 m), a warming of 1 K (they only leads to an estimated model-mean release of methane at the studied warmings of 1 K, 3K and 5K) over 100 years leads to a bottom of the ocean of approximately 560-2140 Tg(CH4)/ flux of methane into the ocean water such that all the methane at year; as no actual geographical distribution of methane depth up to 40 meters below the seafloor is released into the clathrates is considered here, these flux estimates must be ocean. This translates into a methane flux peaking at 0.86 ST viewed as upper bound estimates. Using an observed 1% ratio m3/yr/m2, according to their model calculations. Deeper depos- to estimate the amount of methane reaching the atmosphere, its (570 m) also lead to a methane flux, albeit much reduced. our analysis leads to a relatively small methane flux of Deposits at depth of 1000 m were shown to be basically stable approximately 5–21 Tg(CH4)/year, with an estimated inter- under most realistic scenarios of future climate change. model standard deviation of approximately 30%. The role of [4] In our paper, we expand this analysis by using an sea-level rise by 2100 will be to further stabilize methane estimate of the ocean bottom warming that could be clathrates, albeit to a small amount as the sea-level rise is expected in the latter part of the 21st century. A wide variety expected to be less than a few meters. Citation: Lamarque, of coupled climate models (each with their specific represen- J.-F. (2008), Estimating the potential for methane clathrate tation of atmosphere, ocean and sea-ice processes and their instability in the 1%-CO2 IPCC AR-4 simulations, Geophys. interaction, see Table 8.1 of Randall et al. [2007]) have Res. Lett., 35, L19806, doi:10.1029/2008GL035291. participated in the recent IPCC (Intergovernmental Panel on Climate Change) AR-4 simulations [Randall et al., 2007]; this 1. Introduction collection can be seen as representing the state-of-the-science climate models. It is also denominated as the World Climate [2] Methane clathrate is an ice-like structure of water and Research Programme’s (WCRP’s) Coupled Model Intercom- methane [Kvenvolden, 1999, and references therein] that parison Project phase 3 (CMIP3) multi-model dataset. Simu- form under conditions of low temperatures, high pressure lation outputs are available from the PCMDI archive (see and sufficient methane concentrations [Kennett et al., 2003, http://www-pcmdi.llnl.gov/ipcc/data_status_tables.htm). and references therein]; at present times, methane clathrates Among the various simulations each group performed, the 1%- have been found along many outer continental margins of CO2 increase (over pre-industrial conditions) per year simu- the world [Kvenvolden, 1999], but also under the deep lations are the most easily inter-compared as no forcing but ocean floor [Archer, 2007]. They represent a very large CO is changed [Meehl et al., 2007]. In our study, we therefore 15 3 2 methane reservoir (possibly larger than 10 m at present focus on this specific set of simulations to estimate if, under times [Kvenvolden, 1999]) and their destabilization is doubled-CO2 conditions (in 1%/year increase of atmospheric hypothesized to be the main reason for the isotopic carbon CO2 concentration simulations, doubling occurs approximate- excursion observed in the benthic foraminifera record at the ly after 70 years, slightly faster than the observed 20th century Paleocene-Eocene Thermal Maximum [Dickens et al., rate), significant areas of the global ocean have become 1995; Zachos et al., 2001]. Methane clathrates are stable unstable with respect to methane clathrate. under a range of pressures and temperatures [Dickens and [5] Based on the various IPCC scenarios [Meehl et al., Quinby-Hunt, 1994, and references therein]. Outside this 2007], a doubling of CO2 over pre-industrial conditions zone (for warmer temperatures and/or lower pressures), the (i.e., CO2 levels at approximately 530 ppmv) is expected to occur by 2050; after that point, scenarios have divergent 1Atmospheric Chemistry Division, National Center for Atmospheric trajectories (see Figure 10.26 of Meehletal.[2007]). Research, Boulder, Colorado, USA. Therefore, looking at the time of CO2-doubling is a reason- able estimate of what could be expected by mid-21st Copyright 2008 by the American Geophysical Union. century under most assumptions. 0094-8276/08/2008GL035291$05.00

L19806 1of4 L19806 LAMARQUE: CLATHRATE INSTABILITY IN 1%-CO2 AR4 L19806

Table 1. Estimated Globally-Integrated Model-Mean and Standard especially relevant to the small temperature changes. A Deviation of the Seafloor Methane Flux for Each Case and linear interpolation in depth is used for intermediate depths a Interpolation Techniques (320 m, 570 m and 1000 m were used by RM) in all cases. Cubic Interpolant Linear Interpolant [11] First, we filter our results by the seafloor depth Mean Standard Deviation Mean Standard Deviation (specific to each model) such that only depths larger than Case 1 892 272 563 200 320 m (the minimum depth of RM) are considered for Case 2 942 302 581 210 potential methane clathrate release. As the presence of Case 3 1646 642 1053 463 methane clathrates under the ocean seafloor is a function Case 4 2140 659 1407 462 of pressure and temperature [e.g., Dickens and Quinby- a Tg(CH4)/yr, methane flux. For each case see Section 2. Hunt, 1994], the shallower (320 m) clathrate calculation from RM is only relevant to the polar regions. In order to [6] In the following analysis, the benefit of using a variety present a simple analysis of the pressure and temperature of models is that, owing to the fact that they are independent impact on the potential methane flux, we perform a set of enough in their representation of the physical and dynamical four calculations: Case 1) polar (poleward of 60°) regions ocean processes, an agreement between models can be only (with any depth more than 320 m is available for the viewed as a limited indicator of robustness in the associated presence of clathrate); Case 2) same as Case 1) in addition result [Meehl et al., 2007]. Indeed, while climate models use to the rest of the world, but where only depths larger than the same underlying equations for heat, momentum and 900 m have clathrates; Case 3) same as Case 2) with the other conserved quantities, they differ in the numerical addition of the depths between 700 m and 900 m; Case 4) methods needed to solve those equations and, more impor- same as Case 3) with the addition of the depths between 570 tantly, in the representation of subgrid-scale processes. and 700 m. In all cases, no explicit information on the actual [7] The paper is organized as follows. In section 2, we distribution of methane clathrates (beyond the aforemen- describe the methodology used to create an ensemble-mean tioned dependence on ocean depth) is used. By focusing distribution of methane fluxes. This is followed in section 3 only on depth as a measure of stability, we eliminate the by the analysis of the estimated ocean bottom temperature potential temperature biases the ocean models would have, changes and associated methane flux. Discussion and con- which could skew the potential distribution of clathrates. clusions follow in section 4. 2. Methodology 3. Results

[8] We use monthly ocean temperature distributions from [12] We present in Figure 1 the mean distribution of 21 models; no model screening (on resolution for example) ocean bottom 100-year temperature increase (doubled-CO2 is performed and each model is given equal weight in the minus pre-industrial), regardless of sea-floor depth. As the following analysis. One model (UKMO HadCM3) was not ocean models used in this study have a range of horizontal included in this analysis as the study of the ocean bottom resolutions (from less than 1° to more than 3°), we display temperature trend indicates that this model was not in our results as averaged on a 5° (latitude) by 5° (longitude) equilibrium by the time the 1%-CO2 simulation was started. fixed grid. We find that the regions where the largest In addition, four models have performed a 1%-CO2 simu- increases can be found are mostly located in the Northern lation; however, those were dismissed as the doubling was Hemisphere, and preferably above 50°N, except for a few performed with respect to present-day CO2 levels, not pre- isolated regions off the Gulf of Mexico, the coast of Japan, industrial. We therefore end up with a subset of 16 models between Thailand and Borneo and south of Papua. (using only one realization per model) from 12 different [13] For each model, using the 100-year temperature institutions (see Table S1 of the auxiliary material).1 perturbation and ocean seafloor depth from each model at [9] For each model, we compute at each ocean grid point their original resolution, we compute the associated potential the temperature difference in the bottom layer of the ocean methane release at the bottom of the ocean for each case model between years 1–10 and years 91–100 (20 years (Table 1). If we only consider the Polar regions (Case 1), then after reaching CO2-doubling; once reached, CO2 is kept constant for the remainder of the integration), except for the MIROC model simulation for which only 70 years were available from the PCMDI archive. [10] As the results from RM are only available for 100-year temperature increases of 1 K, 3 K and 5 K, we find that, with the additional constraint of no flux for a zero- temperature increase, a third-order polynomial fit provides a reasonable interpolating function for intermediate values (see auxiliary material). We use this interpolating function to calculate the methane flux associated with any tempera- ture perturbation, including those smaller than 1 K; in our study, no temperature increase is larger than 5 K (see next section). An alternate method (linear interpolation) is used to assess the sensitivity to the interpolation technique, Figure 1. Geographical distribution (averaged over a 5° 1Auxiliary materials are available in the HTML. doi:10.1029/ latitude-longitude grid) of the ocean bottom temperature 2008GL035291. increase (K) after 100 years. See text for details.

2of4 L19806 LAMARQUE: CLATHRATE INSTABILITY IN 1%-CO2 AR4 L19806

(16 models from 12 modeling groups) is used to identify areas affected by potential methane clathrate instability by the time of CO2-doubling over pre-industrial conditions. Based on available future scenarios, CO2-doubling is expected to occur by mid-21st century. [17] Using results from the recent paper by Reagan and Moridis [2007], we have identified regions with significant potential methane clathrate destabilization as being mostly located above the Arctic Circle, with some additional limited flux from some regions of intermediate depths. [18] The model-mean globally-integrated integrated po- tential methane flux is found to be approximately 560-2140 Tg(CH4)/yr released at the bottom of the ocean, with the more likely estimates at the low end of the range; this range Figure 2. Geographical distribution (averaged over a 5° is associated with a variety of assumptions to reflect the latitude-longitude grid) of the model-mean potential potential location of methane clathrates and the interpola- methane flux (Case 2, in Tg(CH )/year) at the bottom of 4 tion of the results of RM for intermediate temperature the ocean based on the 100-yr temperature increase from increases. Our estimate is larger than the previously pub- each model. See text for details. lished estimates by Revelle [1983] and Harvey and Huang the globally-integrated mean methane flux is estimate to be [1995]. However, our estimate must be seen as an upper bound since it was not filtered by an actual distribution of 563 Tg(CH4)/year and 892 Tg(CH4)/year, for the linear and cubic interpolants respectively. The robustness of this result methane clathrates (see below). The inter-model standard can be evaluated by looking at the inter-model standard deviation was found to be approximately 30%.It is however deviation in the global methane flux estimates. When using clear that a 16-member ensemble can only provide limited the cubic polynomial interpolant, the standard deviation of statistical information. [19] Recent observations from a large seepage zone the globally-integrated inter-model methane fluxes is 270 indicate that less than 1% of the methane leaving the ocean Tg(CH4)/year. For the linear interpolant, it is 200 Tg(CH4)/ year, indicating that in both cases the inter-model standard bottom reaches the atmosphere [Mau et al., 2007]. There- deviation is on the order of 30%. This shows the significant fore, using that estimate, the total amount released into the impact associated with the inter-model disagreement on atmosphere could be on the order of 5–21 Tg(CH4)/year, temperature trends, even in a simple forcing experiment such significantly less than the estimated 580 Tg(CH4)/yr annual flux of methane (natural and anthropogenic) emitted at the as the 1%-CO2 simulated described here. The largest increase is found between Cases 2 and 3, in which the lower limit of Earth’s surface at present times [Denman et al., 2007], but depth is raised from 900 m to 700 m. Because of the ocean clearly not negligible. Based on Forster et al. [2007], an temperature distribution, it is expected that Case 2 is a more annual flux of 5–21 Tg(CH4)/year would lead to an realistic assumption than Case 3 (and therefore Case 4). additional global radiative forcing of approximately 0.007–0.03 W/m2. [14] Additional information on inter-model differences can be viewed in terms of a probability distribution. For [20] The estimates for the potential methane clathrate flux that purpose, we calculate the fraction of models for which presented here are clearly highly uncertain. In the first place, the methane flux is found within a certain range (from 0 to it is unclear if the models are accurately representing the warming of ocean bottom waters; in particular, the 1%-CO2 4000 Tg(CH4)/year, by increments of 500 Tg(CH4)/year). We find (see auxiliary material) that Cases 1 and 2 provide a simulations used here are forced by a faster rate of CO2 fairly narrow estimates of the seafloor methane flux, with the increase than the observed rate of the 20th century. In addition, deeper penetration of heat (under the seafloor) majority of models falling within the 500–1000 Tg(CH4)/ year range. The inter-model spread widens with Cases 3 and could be expected with longer timescales or larger pertur- 4, to reach a fairly flat distribution, especially for Case 4. This bations. Secondly, we have assumed that all areas affected further indicates the large sensitivity of the deep seafloor by significant warming were associated with methane methane flux estimates to the underlying model simulations. clathrate. Based on oil and gas exploration, Kvenvolden [1999] indicates that methane clathrates can be found in [15] Based on the third-order polynomial approximation described above and the 100-year temperature increase for many continental margins of the globe, including many of each model and the limits of Case 2, we construct (Figure 2) the regions identified in our study with significant warming, a geographical map of the model-mean estimate of methane except notably for the Barents Sea. A separate analysis flux from destabilization. Due to the much larger fluxes for [Gornitz and Fung, 1994] has identified additional regions shallow deposits (restricted in our analysis to poleward of of potential clathrates, especially in the tropical regions, and 60°), the largest fluxes are found above the Arctic Circle; along the coast of East Asia and Africa in particular. similarly, the rest of the world is found to exhibit only small Thirdly, one needs a better estimate of how much methane potential emission fluxes over the continental shelves. from the ocean floor can reach the atmosphere; direct observations have shown that methane bubbles from seep- age can clearly be traced traveling all the way into the 4. Discussion and Conclusions atmosphere [Niemann et al., 2005; Leifer et al., 2006]. It is [16] Analysis of ocean temperature distributions from however unclear if these observations can be generalized as there is an array of environmental variables that could the IPCC-AR4 simulations of 1%-CO2 increase per year

3of4 L19806 LAMARQUE: CLATHRATE INSTABILITY IN 1%-CO2 AR4 L19806 influence whether the methane bubble gets out of the ocean References to the atmosphere, including methane oxidation and ex- Archer, D. (2007), Methane hydrate stability and anthropogenic climate change rate across the air-sea interface [Archer,2007]. change, Biogeosciences, 4, 521–544. Buffett, B., and D. Archer (2004), Global inventory of methane clathrate: Nonetheless, even if methane is oxidized into the ocean, Sensitivity to changes in the deep ocean, Earth Planet. Sci. Lett., 227, some fraction could still potentially reach the atmosphere as 185–199. CO2 (the primary methane oxidation product [Dickens, Denman, K. L., et al. (2007), Couplings between changes in the climate 2003; Kennett et al., 2003]), leading to a radiative impact, system and biogeochemistry, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment albeit much less important due to its much lower (about Report of the Intergovernmental Panel on Climate Change, edited by 25 times less) radiative efficiency (see Table 2.14 of Forster S. Solomon et al., pp. 499–587, Cambridge Univ. Press, Cambridge, U. K. et al. [2007]). The added CO2 to the ocean waters could in Dickens, G. R. (2003), Rethinking the global with a large, addition perturb the ocean water pH [Doney et al., 2007]. dynamic and microbially mediated gas hydrate capacitor, Earth Planet. [21] An additional mechanism one needs to discuss is the Sci. Lett., 213, 169–183. role of sea-level rise. Changes in sea-level have been shown Dickens, G. R., and M. S. Quinby-Hunt (1994), Methane hydrate stability to play an important role in allowing for destabilization in in seawater, Geophys. Res. Lett., 21, 2115–2118. Dickens, G. R., J. R. O’Neil, D. K. Rea, and R. M. Owen (1995), Dis- the case of significant sea-level drop during cold paleocli- sociation of oceanic methane hydrate as a cause of the carbon isotope matic conditions [Archer, 2007]. In addition, Buffett and excursion at the end of the Paleocene, Paleoceanography, 10, 965–971. Archer [2004] predict a 3% decrease in the clathrate Doney, S. C., N. Mahowald, I. Lima, R. A. Feely, F. T. Mackenzie, J.-F. Lamarque, and P. J. Rasch (2007), Impact of anthropogenic atmospheric inventory when their clathrate distribution model is run nitrogen and sulfur deposition on ocean acidification and the inorganic with a 100-m decrease in sea level. According to Meehl et carbon system, Proc. Natl. Acad. Sci. U. S. A., 104, 14,580–14,585. al. [2007], sea-level rise are expected to be smaller than 1 m Forster, P., et al. (2007), Changes in atmospheric constituents and in radia- tive forcing, in Climate Change 2007: The Physical Science Basis. Con- by 2100. While this number is highly uncertain, it is tribution of Working Group I to the Fourth Assessment Report of the expected that the rise will be in the range of a few meters Intergovernmental Panel on Climate Change, edited by S. Solomon et at a maximum. Beyond very small (negative) changes in the al., pp. 129–234, Cambridge Univ. Press, Cambridge, U. K. total clathrate inventory, the additional stabilization from an Gornitz, V., and I. Fung (1994), Potential distribution of methane hydrates in the world’s oceans, Global Biogeochem. Cycles, 8, 335–347. additional few meters of sea-level over the 21st century will Harvey, L. D. D., and Z. Huang (1995), Evaluation of the potential impact be of relevance only for the shallowest deposits, which we of methane clathrate destabilization on future global warming, J. Geo- have excluded in our analysis (a minimum depth of 320 m phys. Res., 100, 2905–2926. Kennett, J. P., K. G. Cannariato, I. L. Hendy, and R. J. Behl (2003), was imposed) due to the limited methane available under Methane Hydrates in Quaternary Climate Change: The Clathrate Gun those deposits. In both cases, the effect of sea-level rise by Hypothesis, 216 pp., AGU, Washington, D. C. 2100 will therefore be to further limit the potential release of Kvenvolden, K. A. (1999), Potential effects of gas hydrate on human wel- fare, Proc. Natl. Acad. Sci. U. S. A., 96, 3420–3426. methane, but only by a marginal amount, as already noted Leifer, I., B. P. Luyendyk, J. Boles, and J. F. Clark (2006), Natural marine by Gornitz and Fung [1994]. seepage blowout: Contribution to , Global Biogeo- [22] Overall, our analysis indicates that, based on the chem. Cycles, 20, GB3008, doi:10.1029/2005GB002668. range of models used in the most recent IPCC Assessment Mau, S., D. L. Valentine, J. F. Clark, J. Reed, R. Camilli, and L. Washburn (2007), Dissolved methane distributions and air-sea flux in the plume of a Report (AR-4), specific areas of the globe could be affected massive seep field, Coal Oil Point, California, Geophys. Res. Lett., 34, by methane clathrate instability by the time of CO2-dou- L22603, doi:10.1029/2007GL031344. bling over pre-industrial conditions, i.e., around 2050. Meehl, G. A., et al. (2007), Global climate projections, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to However, our calculation indicates that the potential meth- the Fourth Assessment Report of the Intergovernmental Panel on Climate ane flux would be of limited significance compared to the Change, edited by S. Solomon et al., pp. 747–845, Cambridge Univ. overall natural and anthropogenic methane emissions. Once Press, Cambridge, U. K. Niemann, H., et al. (2005), Methane emission and consumption at a North methane is released from those instability zones (with Sea gas seep (Tommeliten area), Biogeosciences, 2, 335–351. possibly an additional amount associated with the methane Prather, M. J. (1996), Time scales in atmospheric chemistry: Theory, GWPs for clathrates under the [Harvey and Huang, 1995; CH4 and CO, and runaway growth, Geophys. Res. Lett., 23, 2597–2600. Walter et al., 2006]) and reached the atmosphere, it will lead Randall, D. A. (2007), Climate models and their evaluation, in Climate Change 2007: The Physical Science Basis. Contribution of Working to a positive (although quite limited, based on our analysis) Group I to the Fourth Assessment Report of the Intergovernmental Panel radiative feedback that will further the impact of man-made on Climate Change, edited by S. Solomon et al., pp. 589–662, greenhouse-gas emissions and affect atmospheric composi- Cambridge Univ. Press, Cambridge, U. K. Reagan, M. T., and G. J. Moridis (2007), Oceanic gas hydrate instability tion and climate for decades [Prather, 1996]. and dissociation under climate change scenarios, Geophys. Res. Lett., 34, L22709, doi:10.1029/2007GL031671. [23] Acknowledgments. We would like to thank G. Danabasoglu for Revelle, R. R. (1983), Methane hydrate in continental slope sediments and his help with the processing of ocean model output. J. Kiehl and E. Holland increasing atmospheric , in Changing Climates, pp. 252– provided vary valuable comments on the research and on the manuscript. 261, Natl. Acad., Washington, D. C. Two anonymous reviewers greatly improved earlier version of the Walter, K. M., S. A. Zimov, J. P. Chanton, and D. Verbyla (2006), Methane manuscript. JFL was supported by the SciDAC project from the Depart- bubbling from Siberian thaw lakes as a positive feedback to climate ment of Energy. We acknowledge the modeling groups, the Program for warming, Nature, 443, 71–75, doi:10.1038/nature05040. Climate Model Diagnosis and Intercomparison (PCMDI) and the WCRP’s Zachos, J., M. Pagani, L. Sloan, E. Thomas, and K. Billups (2001), Trends, Working Group on Coupled Modelling (WGCM) for their roles in making rhythms, and aberrations in global climate 65 Ma to present, Science, available the WCRP CMIP3 multi-model dataset. Support of this dataset 292, 686–693. is provided by the Office of Science, U.S. Department of Energy. The National Center for Atmospheric Research is operated by the University ÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀ Corporation for Atmospheric Research under sponsorship of the National J.-F. Lamarque, Atmospheric Chemistry Division, National Center for Science Foundation. Atmospheric Research, 1850 Table Mesa Drive, Boulder, CO 80305, USA. ([email protected])

4of4