Response of global soil consumption of atmospheric methane to changes in atmospheric climate and nitrogen deposition


  • Qianlai Zhuang,

    Corresponding author
    1. Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana, USA
    2. Department of Agronomy, Purdue University, West Lafayette, Indiana, USA
    3. Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
    • Corresponding author: Q. Zhuang, Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, 550 Stadium Mall Dr., West Lafayette, IN 47907-2051, USA. (

    Search for more papers by this author
  • Min Chen,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana, USA
    Search for more papers by this author
  • Kai Xu,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana, USA
    Search for more papers by this author
  • Jinyun Tang,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana, USA
    Search for more papers by this author
  • Eri Saikawa,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
    Search for more papers by this author
  • Yanyu Lu,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana, USA
    Search for more papers by this author
  • Jerry M. Melillo,

    1. Ecosystems Center, Marine Biological Laboratory, Woods Hole, Massachusetts, USA
    Search for more papers by this author
  • Ronald G. Prinn,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
    Search for more papers by this author
  • A. David McGuire

    1. Department of Biology and Wildlife, University of Alaska Fairbanks, Fairbanks, Alaska, USA
    Search for more papers by this author


[1] Soil consumption of atmospheric methane plays an important secondary role in regulating the atmospheric CH4 budget, next to the dominant loss mechanism involving reaction with the hydroxyl radical (OH). Here we used a process-based biogeochemistry model to quantify soil consumption during the 20th and 21st centuries. We estimated that global soils consumed 32–36 Tg CH4 yr−1 during the 1990s. Natural ecosystems accounted for 84% of the total consumption, and agricultural ecosystems only consumed 5 Tg CH4 yr−1 in our estimations. During the twentieth century, the consumption rates increased at 0.03–0.20 Tg CH4 yr−2 with seasonal amplitudes increasing from 1.44 to 3.13 Tg CH4 month−1. Deserts, shrublands, and xeric woodlands were the largest sinks. Atmospheric CH4 concentrations and soil moisture exerted significant effects on the soil consumption while nitrogen deposition had a moderate effect. During the 21st century, the consumption is predicted to increase at 0.05-1.0 Tg CH4 yr−2, and total consumption will reach 45–140 Tg CH4 yr−1 at the end of the 2090s, varying under different future climate scenarios. Dry areas will persist as sinks, boreal ecosystems will become stronger sinks, mainly due to increasing soil temperatures. Nitrogen deposition will modestly reduce the future sink strength at the global scale. When we incorporated the estimated global soil consumption into our chemical transport model simulations, we found that nitrogen deposition suppressed the total methane sink by 26 Tg during the period 1998–2004, resulting in 6.6 ppb higher atmospheric CH4 mixing ratios compared to without considering nitrogen deposition effects. On average, a cumulative increase of every 1 Tg soil CH4 consumption decreased atmospheric CH4 mixing ratios by 0.26 ppb during the period 1998–2004.

1 Introduction

[2] Methane (CH4), a potent greenhouse gas and a reactive chemical compound in the atmosphere, affects the global climate system and atmospheric chemistry. It is 25 times more effective, on a per-unit-mass basis, than carbon dioxide (CO2) in absorbing long-wave radiation on a 100 year time horizon [Forster et al., 2007]. It contributes 0.48 W m−2 to the total anthropogenic radiative forcing of 2.43 W m−2 by well-mixed greenhouse gases [Forster et al., 2007] and has an indirect effect of 0.13 W m−2 through the formation of other greenhouse gases, mainly tropospheric ozone and stratospheric water vapor [Lelieveld et al., 1998]. Its atmospheric concentrations have increased about 1.5 times since 1750 to the year 2005 level of 1774.62 ± 1.22 ppb observed by U.S. National Oceanic and Atmospheric Administration (NOAA)/Global Monitoring Division [Dlugokencky et al., 2005] and 1774.03 ± 1.68 ppb observed by the Advanced Global Atmospheric Gases Experiment (AGAGE) network with a greater interannual variability in recent decades [Forster et al., 2007]. More recent observation estimates that the global mean atmospheric CH4 concentrations in 2009 reached 1794 ppb [Dlugokencky et al., 2009]. The significant interannual variations of atmospheric CH4 concentrations include (1) abruptly declining in 1992; (2) rising in 1993 and 1994 and peaking in 1998 [Dlugokencky et al., 1996; Cunnold et al., 2003]; (3) approaching a steady state during 1999 and 2002 [Dlugokencky et al., 2003]; (4) reaching an apparent steady state from 1999 to 2005/2006 [Dlugokencky et al., 2009]; and (5) rising again since 2006, specifically the increases are 8.3 ± 0.6 and 4.4 ± 0.6 ppb during 2007 and 2008, respectively [Rigby et al., 2008; Dlugokencky et al., 2009]. To date, these dynamics have not been well understood as to their sources and sinks [e.g., Zhuang et al., 2009; Prinn et al., 2005; Curry, 2007].

[3] Biogeochemistry models have been extensively used to quantify the regional and global land surface emissions, or sources, of atmospheric CH4 [e.g., Walter et al., 2001; Cao et al., 1995; Potter et al., 1996; Zhuang et al., 2004]. The global chemical transport models in combination with satellite and in situ observational data have also been used to constrain surface emissions [e.g., Hein et al., 1997; Houweling et al., 1999; Bergamaschi et al., 2005, 2009; Mikaloff Fletcher et al., 2004; Dentener et al., 2003; Butler et al., 2004; Bousquet et al., 2006; Chen and Prinn, 2006; Frankenberg et al., 2008; Rigby et al., 2008]. Based on these studies, it has been estimated that the current emissions from the Earth's surface range from 503 to 610 Tg CH4 yr−1 [Forster et al., 2007]. A more recent synthesis study identified the challenges to more accurately quantifying biogenic emissions from major source regions and sectors on the Earth's surface, including the Amazon, rice paddies, and the Arctic [Zhuang et al., 2009].

[4] In terms of CH4 sinks, the atmospheric hydroxyl radicals (OH) are found to be a major sink based on model and observational studies of methyl chloroform [e.g., Prinn et al., 2001, 2005; Krol and Lelieveld, 2003; Bousquet et al., 2005; Montzka et al., 2011]. After the OH sink, soil uptake due to methanotrophy is the second largest sink at 15–45 Tg CH4 yr−1 [Forster et al., 2007]. Since soil consumption rates are of a similar magnitude to the rate of CH4 accumulation in the atmosphere, any significant changes in this sink will affect the net accumulation of this potent greenhouse gas in the atmosphere [Dutaur and Verchot, 2007]. To date, the amount of global soil consumption has been quantified with numerous biogeochemistry models [e.g., Potter et al., 1996; Ridgwell et al., 1999; Curry, 2007, 2009; Spahni et al., 2011; Melton et al. 2012]. These studies consider the effects of changes in atmospheric CH4 concentrations, land use, and climate. However, nitrogen deposition has not been included in these analyses, despite it having been shown to significantly affect the consumption rate [e.g., King, 1997; Zhang et al., 2008; Menyailo et al., 2008].

[5] Here we used an extant biogeochemistry model, the Terrestrial Ecosystem Model (TEM) [Melillo et al., 1993; Zhuang et al., 2004, 2006, 2007], to provide a more comprehensive quantification of global soil consumption in both natural and agricultural ecosystems in a spatially and temporally explicit manner for the 20th and 21st centuries. In this analysis, we explicitly evaluated the nitrogen deposition effects in addition to multiple stresses including the changes of climate, atmospheric CH4 concentrations, and land use at the global scale. Using the global soil consumption rates calculated from TEM, we also ran a three-dimensional global chemical transport model, the Model for Ozone and Related Tracers version 4 (MOZART v4) [Emmons et al., 2010] to evaluate the role of this sink in affecting atmospheric mixing ratios for recent years.

2 Method

2.1 Overview

[6] We first revised an earlier version of the TEM by incorporating the effects of changes in nitrogen deposition, soil moisture, and atmospheric CH4 concentrations on methanotrophy rates and also CH4 transport between soils and the atmosphere. Second, we used field data for CH4 consumption rates of various ecosystems (ranging from tundra to tropical forests to cropland ecosystems) to parameterize the model. Spatially explicit data for climate, soils, atmospheric nitrogen deposition, land use and land cover change, as well as atmospheric CH4 concentrations, are then used to extrapolate the parameterizations and overall model to global terrestrial ecosystems at a 0.5° by 0.5° (latitude by longitude) spatial resolution over the 20th and 21st centuries. A set of simulations was conducted to examine the various effects on global soil consumption. Third, we conducted a set of simulations using MOZART v4 to examine how different degrees of global soil methane uptake affect atmospheric methane mole fractions between 1998 and 2004.

2.2 Model Modification

[7] The Terrestrial Ecosystem Model is a process-based biogeochemistry model that couples carbon, nitrogen, water, and heat processes in terrestrial ecosystems to simulate ecosystem carbon and nitrogen dynamics (TEM) [Melillo et al., 1993; Zhuang et al., 2003, 2004, 2007]. In TEM, methanogenesis (CH4 production) and methanotrophy (CH4 consumption) are affected by hydrological and soil thermal dynamics, in addition to other biological and chemical controls, in both non-permafrost and permafrost ecosystems [Zhuang et al., 2001, 2002, 2004]. Specifically, methanotrophy is modeled as an aerobic process that occurs in the unsaturated zone of the soil profile.

[8] The hourly methanotrophy rates within each 1 cm layer of the soil profile are modeled as a function of soil CH4 concentrations, soil temperature, soil moisture, and redox potentials [Zhuang et al., 2004]. In this study, we modified the algorithm for methanotrophy by incorporating the effects of water inhibition on atmospheric CH4 diffusivity to soils and the effects of nitrogen deposition and changes in atmospheric CH4 concentrations on the methanotrophy rate, Mo(z,t), at soil depth z and time t as

display math(1)

where OMAX is the ecosystem-specific maximum oxidation rate that typically ranges between 0.3 and 360 µmol L−1 h−1 [Segers, 1998]; f(CM(z,t)) is a multiplier that enhances methanotrophy rate with increasing soil CH4 concentrations using a Michaelis-Menten function with a half-saturation constant (KCH4) that varies across ecosystems; f(TSOIL(z,t)) is a multiplier that enhances methanotrophy rates with increasing soil temperatures using a Q10 function with Q10 coefficients (OQ10) and reference temperatures (TOR) that vary across ecosystems; f(ESM(z,t)) is a multiplier to account for the biological limiting effect that diminishes methanotrophy rates if the soil moisture is not at an optimum level (Mvopt); and f(ROX(z,t)) is a multiplier that enhances methanotrophy rates as redox potentials increase linearly from −200 mV to 200 mV (as redox potentials become greater than 200 mV, f(ROX(z,t)) is set equal to 1.0) [Zhang et al., 2002].

[9] Methanotrophy is also assumed to cease if soil moisture reaches a critical minimum (Mvmin) or maximum (Mvmax) value. These critical soil moisture values, along with the optimum soil moisture (Mvopt), are assumed to vary among ecosystems [see Zhuang et al., 2004, equation (B3)]. The term f(Ndp(z,t)) represents the effects of nitrogen deposition on the soil CH4 consumption rate, and the term f(Dms(z,t)) represents the moisture limiting effect on CH4 consumption through its effects on atmospheric CH4 diffusivity to soils. In comparison to our original formulation of methanotrophy, terms f(Ndp(z,t)) and f(Dms(z,t)) are new [Zhuang et al., 2004]. To account for the inhibitive effects of nitrogen deposition (mainly NHx including NH3 and NH4) on soil CH4 consumption, we first modeled the distribution of CNHx in soil profiles (µmol gdw−1; µmol per gram dry weight) at depth z (cm):

display math(2)

[10] Here DepNHx (mg m−2) is the total NHx deposition in a certain month, with the assumption that the depth interval is 1 cm. Coefficient A (dimensionless) was estimated to be 0.227 based on Schnell and King [1994]. The soil bulk density, Dbulk (g cm−3) is calculated based on soil texture [Saxton et al., 1986]:

display math(3)

[11] At a given level of CH4 concentrations at depth z (cm), the inhibition rate is calculated as

display math(4)

[12] Here math formula= −0.2434 and math formula = 3.9332 based on Figure 7 in Schnell and King [1994]. The maximum inhibition rate at depth z (cm) at a given level of NHx concentrations at depth z (InhibMax (z)) was then calculated as

display math(5)

where dInhib is the maximum rate of inhibition for 1 µmol gdw−1 NHx, which is estimated as 60% based on Schnell and King [1994].

[13] The moisture limiting effect on CH4 consumption f(Dms(z,t)) is modeled via affecting atmospheric CH4 diffusivity:

display math(6)

[14] Here at the minimum soil moisture SMmin, f(Dms(z,t)) is equal to 1.0. As SM approaches the saturation point SMsat, the diffusivity D(z) decreases as f(Dms(z,t)) decreases. We set a minimum value of 0.0001 for f(Dms(z,t)) since in saturated soils, the diffusivity is usually on the order of 4 times smaller than in dry soils [Walter and Heimann, 2000]. In equation (6), γ, a parameter depending on the aggregation structure, is set at 1.0.

[15] To compute the CH4 consumption in soils, we solved the steady state transport equation for the soil profile at a 1 cm depth step:

display math(7)

using the following boundary conditions:

display math(8)

[16] The diffusion coefficient D(z) (with units of cm−2 h−1) is modeled as a function of soil texture. Equation (7) is solved for CH4 concentrations at different depths from the soil surface to a prescribed bottom level zb [see Zhuang et al., 2004]. The updated CH4 concentrations are used to model f(CM(z,t)) in equation (1). Total CH4 consumed in the soil column (Fcons) within an hour is calculated based on the following:

display math(9)

[17] Here Zb is the lower boundary of the soil depth (cm), which is set to 100 cm.

2.3 Model Parameterization and Extrapolation

[18] We parameterized TEM for representative ecosystem types of both natural and agricultural ecosystems. The information on climate, soils, and observed consumption flux data for the calibration sites is documented in Table 1. The calibration was done by running TEM for the observational period driven with the corresponding local meteorological or climatic data at each site. The climate data from Climatic Research Unit (CRU) [Mitchell and Jones, 2005] or re-analysis data from the National Centers for Environmental Prediction (NCEP; are used (Table 1). Agricultural ecosystems were parameterized for both C3 and C4 crops (Table 2). Parameters in Table 2 were adjusted to allow the simulated daily average CH4 consumption rates to match the observed daily average data with a tolerance of 1%. Specifically, based on the prior knowledge about the values of these four parameters for each site, we used a trial-and-error method to alter their values in model simulations while other parameters of TEM were kept as is. Comparison between the simulated and observed average daily consumption rates for the observation period was conducted to see if the tolerance was reached for each simulation. We iterated these steps until the tolerance value was met to determine the parameter values (Table 2). These site-level parameters were extrapolated to each 0.5° × 0.5° (latitude × longitude) pixel of the global natural ecosystem map that consists of 11 plant functional types (Table 3). For agricultural ecosystem pixels, C3 and C4 parameterizations were extrapolated with an assumption that the land area was occupied equally with these two types of crops. BCI and INQ parameterizations were used for C3 and C4 crop ecosystems for the pixels in the Southern Hemisphere. Site parameterizations of BCI and IW were used for C3 and C4 crop ecosystems, respectively, in the region between 0°N and 45°N. Sites GUK and MG parameterizations were used for the pixels with C3 and C4 crop ecosystems, respectively, above 45°N.

Table 1. Parameterization Sites of Natural and Managed Ecosystems in This Studya
Site NameLocation (Longitude/Latitude)Elevation (m)VegetationDriving Climate DataObserved DataSources and Comments
  1. a

    NCEP: National Centers for Environmental Prediction.

  2. b

    NOAA: National Oceanic and Atmospheric Administration.

  3. c

    CRU2: Climatic Research Unit version 2.0.

Puerto Rico (PR)65°45′W/18°19′N103Tropical forestNCEPa 1948–2005CH4 flux of April 1984Keller et al. [1986]
Nunn, Colorado (NC)104°45′W/40°48′N1584GrasslandNOAAb 1981–1996CH4 flux of 1990Mosier et al. [1991]
Petersham, Massachusetts (BFT)72°10′W/42°30′N277Temperate coniferous forestsNOAA 1982–1989CH4 flux of May–December 1988Castro et al. [1995]
Petersham, Massachusetts (BFD)72°10′W/42°30′N277Temperate deciduous forestsNOAA 1982–1989CH4 flux of May–December 1988Castro et al. [1995]
Heidelberg,Germany (HGX)8°36′E/49.4°N217Xeric woodlandsNCEP 1948–2005Annual consumption of 1987–1988Born et al. [1990]
Heidelberg,Germany (HGM)8°36′E/49.4°N217Mediterranean shrublandsNCEP 1948–2005Annual consumption of 1987–1988Born et al. [1990]
Petersham, Massachusetts (BFE)72°10′W/42°30′N277Temperate evergreen broadleafNOAA 1982–1989CH4 flux of May–December 1988Castro et al. [1995]
Shoshone Mountain, Nevada (MD)116°2/36°37′N2477DesertNOAA 1985–1994CH4 flux of January 1989–June 1991Striegl et al. [1992]
Barro Colorado Island, Las Pavas, Panama (BCI)9°10′N/79°50′W149C3 riceCRU2c 1988Daily soil CH4 consumption 1988Keller et al. [1990]
Ingham, north Queensland (INQ)18°42′S/146°12′E108C4 sugarcaneCRU2 1996Daily CH4 consumption of 1996Weier [1999]
Varanasi, Gangetic Plains, India (VA)25°18′N/80°1′W27C3 riceCRU2 1995Seasonal CH4 consumption of 1995Singh et al. [1999]
Roadside Farm, Iowa (IW)41°56′17″N/93°23′23″W329C4 corn-soybeanNOAA 1993Seasonal CH4 fluxes of 1993Chan and Parkin [2001]
Gullane, Edinburgh, UK (GUK)56°2′7.3″N/2°49′50.2″W234C3 wheatCRU2 1994–1995Daily CH4 consumption of 1994–1995Dobbie and Smith [1996]
Munich, Germany (MG)48°8′20.86″N/11°34′48.7″E494C4 sunflowerCRU2 1992–1993Hourly CH4 consumption of 1992–1993Flessa et al. [1995]
Table 2. Model Parameters for Various Natural and Agricultural Ecosystemsa
  1. a

    OMAX is a maximum methane oxidation rate (µmol L−1 h−1). KCH4 is a Michaelis-Menten coefficient to account for atmospheric methane concentration effects (µmol L−1). OQ10 is an ecosystem-specific Q10 coefficient to account for soil temperature effects on methanotrophy. TOR is the reference soil temperature to account for soil temperature effects on methanotrophy (°C). Site names are in Table 1.

  2. b

    Agricultural sites.

Table 3. Regional Soil Methane Consumption (Tg CH4 yr−1) for Both Natural (NE) and Agricultural Ecosystems (AE) in the 1990s, Partitioned to Different Latitude Bandsa
 45°S South0°S–45°S0°N–45°N45°N NorthGlobal
  1. a

    S3 and S4 simulations are with 2 year NHx lag effects on consumption.


2.4 Spatially Explicit Data Organization for Global Simulations

[19] To make spatially and temporally explicit estimates of CH4 consumption at the global scale with TEM, we utilized the data of land cover, soils, climate, and leaf area index (LAI) from a variety of sources at a spatial resolution of 0.5° latitude × 0.5° longitude. The land cover data include potential vegetation distribution [Melillo et al., 1993], soil texture [Zhuang et al., 2003], and water pH in soils [Carter and Scholes, 2000].

[20] Similar to earlier versions of TEM, the vegetation and soil texture data sets were used to assign vegetation- and texture-specific parameters to each grid cell. Agricultural grid cells in tropical, temperate, and northern high-latitude regions were assigned with different parameterizations for C3 and C4 crops, respectively (Table 2). The remaining spatially explicit data sets needed to provide inputs into the calculation of CH4 consumption included those for wet soils and the fractional inundation of wetlands to derive the proportion of uplands within each 0.5° × 0.5° grid cell [Matthews and Fung, 1987]. Daily climate data sets were developed from the historical monthly air temperature, precipitation, vapor pressure, and cloudiness data sets [Mitchell and Jones, 2005] of the Climatic Research Unit (CRU) of the University of East Anglia in the United Kingdom (Figure 1a). Monthly LAI data required to simulate soil moisture in TEM are from Zhuang et al. [2004]. We used the data of 1992 for each year of the period 1993–2000 based on the data of McGuire et al. [2001] to extend the land cover and land use data to cover the twentieth century.

Figure 1.

Historical climate, atmospheric CH4 concentrations, and atmospheric NHx deposition from 1900 to 2000 are based on Mitchell and Jones [2005], Etheridge et al. [1998], Globalview-CH4 [2005], and Galloway et al. [2003], respectively.

[21] We used the data of atmospheric CH4 mixing ratio [units: parts per billion (ppb)] from Etheridge et al. [1998] (ice cores) and Globalview-CH4 [2005] to develop the transient atmospheric CH4 concentration data with units of μM (10−3 mol m−3):

display math(10)

where p is atmospheric pressure (Pa), R is the universal gas constant (8.314472 J mol−1 K−1), and T is air temperature (K). We used the mean annual surface air temperature data from Smith and Reynolds [2005], the observed atmospheric CH4 concentration ((CH4)ppb) record, and the standard atmospheric pressure for the calculation (Figure 1b).

[22] A spatially explicit data set of NHx deposition from 1900 to 2000 is developed at a 0.5° spatial resolution (Figure 1c). The spatial variation of the magnitudes of NHx deposition was based on the data for the years 1860 and 1993 (see The original NHx data have units of mg N m−2 yr−1 at a spatial resolution of 5° (longitude) by 3.75° (latitude). Here we scaled these NHx deposition data with information from Figure 1a in Galloway et al. [2003] for the period from 1860 to 1940. From 1941 to 2000, the NHx deposition rates were estimated based on these NHx data and population information (see The jump of NHx deposition may be due to the industrial development, population rising, and agricultural fertilizer increases since 1940 (Figure 1c). The linear relationship between NHx deposition and population was assumed based on the fact that most of the reactive nitrogen was created for agricultural production of food to support the population [Galloway et al., 2003]. Spatial heterogeneity of NHx distribution was considered in developing these spatial data by grouping the population into 13 different subregions including Western Europe, Eastern Europe, Former USSR, United States of America, other Western offshoots, Mexico, other Latin America, Japan, China, India, other Asia, Africa, and Australia (its population from 1901 onward information is obtained from Australian Bureau of Statistics). Specifically, for the period from 1860 to 1940, scaling factors were derived by assuming that (1) the spatial pattern of NHx deposition was constant and (2) the ratio of the overall NHx deposition in a given year to that in 1860 is equal to the ratio of the total reactive nitrogen creation in the given year to that in 1860 (where the latter is described in Figure 1a in Galloway et al. [2003]). For the period from 1941 to 2000, a scaling factor, defined as the ratio of the NHx deposition in a given year to that in 1993, was assumed to be (1) as constant in any of the 13 subregions for the given year and (2) equal to the ratio of the human population in that year to the human population in year 1993. The scaled data were then interpolated to a 0.5° resolution using the nearest neighborhood method (Figure 2).

Figure 2.

Global atmospheric NHx deposition in the 1990s (mg N m−2 yr−1) based on Galloway et al. [2003].

[23] To conduct simulations for the 21st century, we utilized climate forcing data sets based on the Intergovernmental Panel on Climate Change (IPCC) future climate scenarios, which represent possible future climates under emissions scenarios from extreme to moderate: (1) more integrated world with rapid economic growth with an emphasis on fossil fuels (A1FI); (2) more divided world with regionally oriented economic development (A2); (3) more divided world but more ecologically friendly (B2); and (4) more integrated world and rapid economic growth but ecologically friendly (B1) [International Panel on Climate Change, 2000, 2001]. Under those scenarios, the global climate has been simulated with Hadley Centre Coupled Model, version 3 and processed to a 0.5° × 0.5° spatial resolution [Mitchell et al., 2004]. The transient atmospheric CH4 concentration data were obtained by linearly interpolating the decadal data for these four future scenarios (Figure 3). For N deposition, we assumed the annual deposition holds at the year 2000 level for the period 2000–2049 and the year 2050 level for the period 2050–2100 due to the limited availability of global spatially explicit atmospheric N deposition data provided by Dentener [2006]. Spatial data of vegetation, soil texture, soil pH, and upland and cropland distributions used in the 21st century were the same data as in the 20th century simulations.

Figure 3.

Climate and atmospheric CH4 concentrations during the 21st century [Forster et al., 2007]: (a) cloudiness (%), (b) annual precipitation (mm), (c) mean annual surface air temperature (°C), (d) vapor pressure (hPa), and (e) atmospheric methane concentrations (ppb).

2.5 Simulation Design

[24] For the twentieth century, we conducted four “core” simulations with the set of parameters for both natural and agricultural ecosystems at the global scale: (1) the first simulation was done with an atmospheric CH4 concentration of year 2000 at 0.075 μM (1700 ppb), but driven with transient climate throughout the century (hereafter referred to as S1); (2) the second simulation was driven with transient climate and atmospheric CH4 concentrations (hereafter referred to as S2); (3) the third simulation was driven with transient climate, atmospheric CH4 concentrations, and NHx deposition (hereafter referred to as S3); and (4) the fourth simulation was driven with transient data of climate, CH4 concentrations, NHx deposition, and land use and land cover change (hereafter referred to as S4). To examine the sensitivity of the global consumption to climate, atmospheric CH4 concentrations, and N deposition, we conducted another eight simulations with the set of parameters by altering air temperature uniformly with ±3°C for each grid cell, precipitation by ±15%, atmospheric CH4 concentrations by ±35% changes, and NHx deposition by ±25%, respectively, while maintaining all other variables to be the same as in the original data. In these simulations, NHx deposition was assumed to have a 2 year lag inhibitive effect on methanotrophy according to field studies [e.g., Schnell and King, 1994]. To further constrain the estimates, we conducted another simulation driven with the climate data from the European Centre for Medium-Range Weather Forecasts (ECMWF) from 1998 to 2008 (, but otherwise with the same forcing as in the S4 simulation with 2 year NHx lag effects (hereafter referred to as S4b simulation).

[25] For the 21st century, we conducted two sets of simulations under the above four IPCC climate scenarios. The first set of simulations assumes that the global NHx deposition remains at the level of year 2000 throughout the century. The second set of simulations is driven with the N deposition data as described above.

3 Results

3.1 Global Soil Methane Consumption During the Twentieth Century

[26] During the twentieth century, the soil methane consumption increase rate was 0.14 Tg CH4 yr−2, from 18 Tg CH4 yr−1 in the first decade to 32 Tg CH4 yr−1 in the 1990s under changing conditions of climate, atmospheric CH4 concentrations, land use change, and N deposition (Figure 4). Natural ecosystems dominated the consumption, while agricultural ecosystems were only responsible for 16% of the total global soil consumption (Table 3). The average maximum consumption was 3.24 Tg CH4 month−1 in July, and the average minimum consumption was 1.30 Tg CH4 month−1 in February during the century. Seasonal amplitude increased from 1.44 in the 1900s to 3.13 Tg CH4 month−1 during the 1990s.

Figure 4.

Annual global soil methane consumption (Tg CH4 yr−1) from 1900 to 2000: The estimates are derived from the simulations driven with changes of climate, atmospheric CH4 concentrations, atmospheric NHx deposition, and land use change. The effects of atmospheric NHx deposition are assumed to be effective within a 2 year time horizon. Simulation S1 is driven with fixed atmospheric CH4 concentration at 0.075 μM, but with transient climate throughout the century. Simulation S2 is driven with transient climate and atmospheric CH4 concentrations. Simulation S3 is driven with transient climate, atmospheric CH4 concentrations, and NHx deposition. Simulation S4 is driven with transient data of climate, CH4 concentrations, NHx deposition, and land use and land cover change.

[27] At the global scale, the consumption increased from north to south (Figure 5). This north-south gradient prevails throughout the twentieth century. This simulated north-south gradient may be biased due to the limited number of parameterization sites in the south (Table 1). With time, there was no significant change in the consumption in the Northern Hemisphere, but a significant enhancement of the consumption occurred in the Southern Hemisphere. The region between 45°S and 45°N was responsible for more than 80% of the total global consumption (Table 3). The southern United States, eastern China, South America, and Australia had relatively high consumption rates (Figure 5). As a result, the northern high latitudes (45°N–90°N) consumed about 5 Tg CH4 yr−1. The United States, China, and Australia consumed 2.84, 2.78, and 2.89 Tg CH4 yr−1, respectively, during the 1990s. The Sahara desert also had a large consumption of 1.36 Tg CH4 yr−1.

Figure 5.

Mean annual global methane consumption (g CH4 m−2 yr−1) for the three periods of (a) 1900–1929, (b) 1930–1969, and (c) 1960–1999. The estimates are derived from the simulation (S4) driven with climate, changes of atmospheric CH4 concentrations, land use and land cover change, and atmospheric deposition of NHx with 2 year lag effects.

[28] The global soil consumption varied with ecosystem types with more than 80% of the consumption in natural ecosystem soils (Table 4). The agricultural ecosystems soils only consumed about 5.13 Tg CH4 yr−1. Tundra ecosystems and boreal forests, occupying 8.8 million km2, took up 1.18 and 1.79 Tg CH4 yr−1, respectively. Temperate forests and grasslands were moderate sinks and were responsible for 5.47 and 0.81 Tg CH4 yr−1, respectively. In contrast, xeric woodlands and shrublands accounted for more than 25% of the total global soil consumption. Deserts consumed 1.50 Tg CH4 yr−1 over their 11 million km2 area. Temperate savannas with an area of 4 million km2 also acted as a significant sink of 3.01 Tg CH4 yr−1. However, tropical evergreen forests occupied a large land area of 12 million km2, but only consumed 0.05 Tg CH4 yr−1. Overall, the annual soil methane consumption was higher in areas with sparser vegetation, ranking downward from deserts to shrublands, to grasslands, and then finally to forest ecosystems.

Table 4. Global Soil Methane Consumption in Different Natural and Agricultural Ecosystems in the 1990s
Vegetation TypesUpland Area (Million km2)CH4 Tg CH4 yr−1)
Alpine Tundra4.700.46
Wet Tundra4.240.72
Boreal Forest8.400.93
Forested Boreal Wetlands1.770.02
Boreal Woodlands5.130.79
Non-Forested Boreal Wetlands0.470.05
Mixed Temperate Forests3.082.53
Temperate Coniferous Forests1.981.67
Temperate Deciduous Forests2.051.23
Temperate Forested Wetlands0.030.04
Tall Grasslands1.670.22
Short Grasslands3.420.59
Tropical Savannas9.571.87
Xeric Shrublands12.951.75
Tropical Evergreen Forests12.310.05
Tropical Forested Wetlands0.320.03
Tropical Deciduous Forests2.680.04
Xeric Woodlands4.695.59
Tropical Forested Floodplains0.050.01
Tropical Non-Forested Wetlands0.010.00
Tropical Non-Forested Floodplains0.090.01
Temperate Non-Forested Wetlands0.120.01
Temperate Forested Floodplains0.030.02
Temperate Non-Forested Floodplains0.020.00
Wet Savannas0.070.00
Salt Marsh0.050.00
Tidal Freshwater Marshes0.000.00
Temperate Savannas4.063.01
Agricultural Ecosystems15.025.13
Temperate Evergreen Broadleaf2.362.45
Mediterranean Shrublands1.100.95

[29] The S1, S2, and S3 simulations estimated global total consumption rates ranging from 32 to 36 Tg CH4 yr−1 in the 1990s (Table 3). The S1 simulation, driven with a constant atmospheric CH4 level and transient climate, showed a steady soil methane consumption rate of 36 Tg CH4 yr−1 during the twentieth century, while the S2, S3, and S4 simulations showed an increasing trend (Figure 4). In S4b simulation driven with ECMWF data, we estimated the global consumption is 33 Tg CH4 yr−1 during the period 1989–2008.

3.2 Global Soil Methane Consumption During the 21st Century

[30] Under future climate conditions, atmospheric CH4 concentrations and N deposition scenarios, global soil consumption is predicted to increase with rates ranging from 0.05 to 1.0 Tg CH4 yr−2 during the 21st century (Figure 6). Holding the N deposition at the level of the year 2000, our simulations of global soil methane consumption range from 45 to 140 Tg CH4 yr−1 by the end of the century. The second set of future simulations with different levels of N deposition before 2050 and after 2050 estimate that the global total only decreases 1% in comparison with the first set of future simulations at the end of the 21st century.

Figure 6.

Future global soil methane consumption under different climate scenarios simulated with a 2 year NHx effect.

[31] Spatial patterns of the consumption varied in different future scenarios as a result of changes in climate and atmospheric CH4 concentrations, as well as N deposition rates (Figure 7). Specifically, under the scenarios A1FI, A2, and B2, the sink in the eastern and western U.S. and the eastern China was strengthened. South America and Australia became even stronger sink regions compared to their current sinks. The increasing consumption rate in boreal regions led to a larger global total sink. In contrast, similar sink rates to the current sink levels remained in scenario B1 at the end of the 21st century.

Figure 7.

Global methane consumption under future climate scenarios (A1FI and B1) partitioned to every 30 year period. The values are mean annual consumption of the 30 year average.

[32] By the end of the 21st century, under the scenarios A1FI, A2, B1, and B2, boreal ecosystems consumed up to 5–20 Tg CH4 yr−1. In contrast, tropical ecosystems retained similar sink strengths in 2100 in comparison to the present. Shrubland and woodland ecosystems became even stronger sinks ranging in 2100 from 12 to 38 Tg CH4 yr−1. Throughout the century, tundra and grassland ecosystems remained as moderate sinks, of 2–13 Tg CH4 yr−1, respectively. As a result, the United States and China each consumed in 2100 about 5–20 Tg CH4 yr−1, and Australia consumed 3–11 Tg CH4 yr−1 with its sink area becoming even larger. Similarly, Russia and Canada increased their sinks by 2100 to 3–11 and 2–8 Tg CH4 yr−1, respectively.

4 Discussion

4.1 Comparison With Other Studies

[33] For the 1990s, our estimates of global soil consumption of 32–36 Tg CH4 yr−1 are significantly constrained in comparison with the range of 17–51 Tg CH4 yr−1 obtained from other existing estimates using various process-based and bookkeeping modeling approaches [e.g., Ridgwell et al., 1999; Curry, 2007; King, 1997; Potter et al., 1996; Dutaur and Verchot, 2007; Dorr et al., 1993; Smith et al., 2000]. The differences among models are due to several factors being considered differently in various studies as enumerated by King [1997]. For instance, Curry [2007] estimates that considering the effects of temperature, moisture, and land cultivation, the global sink strength is between 9 and 47 with an average of 28 Tg CH4 yr−1. His analysis suggests that the sensitivity of consumption is more severe to soil moisture than to air temperature. In his study, deserts are not factored into the calculations. We estimated that deserts, however, while accounting for 10% of the global land area, consume only 1.50 Tg CH4 yr−1 (Table 4). This estimate is lower than an estimate based on an observed consumption rate of 0.66 mg CH4 m−2 d−1 of a desert in the U.S. [Striegl et al., 1992]. Spatially, our simulations also estimated that warmer and drier soils in semiarid steppe, tropical savanna, tropical seasonal forest, and chaparral ecosystems consume the most methane [e.g., Potter et al., 1996]. This is primarily due to a higher atmospheric CH4 diffusion rate into soils in these ecosystems, while climatology and soil wetness conditions and NHx deposition also contribute to the spatial variability. Our parameterization sites are more in midlatitudes of the Northern Hemisphere and less in tropical regions and the Southern Hemisphere, which may also contribute to the spatial simulation results (Table 1). Smith et al. [2000] found that in situ data of methane oxidation follow lognormal distributions. Ideally, we could parameterize the model based on the distribution of the observed data to quantify the regional consumption uncertainty. However, due to lack of time series data, we were not able to do such quantification.

[34] For the future, we estimated that the global soil consumption increases 22% to 280% depending on what IPCC Special Report on Emissions Scenarios (SRES) (A1FI, A2, B1, and B2) are used. Due mainly to a relatively larger temperature response, the northern high latitudes had relatively more enhanced consumption rates in comparison to the Southern Hemisphere (Figure 7). Curry [2009] also finds that under the SRES emission scenario A1B, the model projects a 23% increase in the global annual mean CH4 soil consumption by 2100, and the largest relative increases occur in the northern high latitudes.

4.2 Major Controls to Soil Methane Consumption

[35] During the twentieth century, the consumption increase rate was primarily due to increasing soil temperature resulted from increasing air temperature, moisture, nitrogen deposition, and atmospheric methane concentrations (Table 5). This temperature response is consistent with our sensitivity analysis for the 1990s, which indicates that a 3°C air temperature increase will result in a 21–32% change in global soil methane consumption (Table 6). Our simulations under future climate scenarios also indicated that the increasing surface temperature dominates the future sink strength (Figures 3 and 6). In contrast, Ridgwell et al. [1999] assumed less sensitive temperature response of methane oxidation activity in high temperatures (i.e., >20°C), suggesting that global warming will leave the global methane sink unchanged, and concluded that the microbial activity is instead the major control.

Table 5. Effects of Climate, Annual Soil Temperature at Top 20 cm Depth, and Soil Moisture on Consumption for Different Regions and the Globe During the Twentieth Centurya
 Annual Air TemperatureAnnual PrecipitationMethane Mixing RatioNHx DepositionSoil TemperatureSoil Moisture
  1. a

    The simulated consumption is based on transient atmospheric methane concentrations, changes of NHx deposition, and transient climate with 2 year NHx year lag effects. The values are Pearson correlations between annual methane consumption and environmental factors.

  2. b

    p < 0.01.

45°S South0.17−0.120.99b0.92b0.180.98b
45°N North0.63b0.74b0.98b0.83b0.68b0.99b
Table 6. Sensitivity Studies of Global Soil Methane Consumption to Changes in NHx Lag Effects, Atmospheric Methane Concentration, NHx Deposition, Air Temperature (AT), and Precipitationa
 BaselineNHx 10 YearsCH4 +30%CH4 −30%NHx +25%NHx −25%Precipitation +15%Precipitation −15%AT +3°CAT −3°C
  1. a

    The values are for the 1990s. The simulations are conducted using transient climate with baseline as well as the following nine sensitivity runs: (1) 10 year lag in NHx; (2) 30% increase in atmospheric CH4; (3) 30% decrease in atmospheric CH4; (4) 25% increase in NHx; (5) 25% decrease in NHx; (6) 15% increase in precipitation; (7) 15% decrease in precipitation; (8) 3°C increase in air temperature; and (9) 3°C decrease in air temperature. Percentage changes are calculated based on these S4 sensitivity simulations.

Consumption (Tg CH4 yr−1)31.727.640.523.629.032.732.032.041.725.0
Changes (%)0.0−1328−25−930.90.932−21

[36] Soil moisture significantly correlates with consumption, due to moisture effects on physical limitations of atmospheric CH4 diffusion into soils and microbial activities (R > 0.98) (Table 5). However, there is a smaller and even a negative correlation between precipitation and consumption at the global scale and in the south of 45°S. This is consistent with field studies suggesting that higher precipitation reduced CH4 uptake when either temperature or N deposition (but not both) is increased [Blankinship et al., 2010]. Ridgwell et al. [1999] concluded that the role of soil moisture in limiting uptake due to inhibition of diffusion is not a major factor. Indeed, in our study, a 15% change in annual precipitation did not affect consumption significantly. We found that soil moisture has not been changed to a significant level from the 15% precipitation change in our simulations, and consequently, both the amount of atmospheric methane diffusion to soils and the consumption from affected microbial activity have not been impacted significantly.

[37] Decreasing or increasing atmospheric methane concentrations by 30% resulted in a −25 ~ 28% change in consumption (Table 6). During the twentieth century, rising atmospheric CH4 concentrations increased the soil consumption by 46% (Figure 4). In contrast, the increasing nitrogen deposition rate suppressed soil CH4 consumption (Table 6). Since the inhibition from a single application of ammonium can persist for days to months, even after the added ammonium is no longer detectable [Nesbit and Breitenbeck, 1992; Mosier et al., 1991], we conducted a simulation considering a10 year nitrogen lag effect, which indicated that consumption is reduced by 10% in comparison with simulations run under 2 year lag effect. With the model sensitivity, our simulations S3 and S2 suggested that nitrogen deposition change reduces the global consumption by 10% in the 1990s (Table 3).

[38] Our simulations estimated that 15 Mkm2 agricultural soils consume 5.13 Tg CH4 yr−1 in the 1990s (Table 3). We have not modeled how land use change due to forest conversion, agricultural abandonment, and urbanization affects the soil consumption. Menyailo et al. [2008] showed that afforestation on a well-aerated grassland in Siberia reduces soil CH4 uptake by a factor of 3 after 35 years of tree growth. The decline in CH4 oxidation was due to the reduction in biomass in soils and to a lesser extent by reduced cell-specific activity of CH4 oxidizing bacteria when grasslands were converted to forests. These effects of land use and land cover change should be factored into our future consumption analysis.

4.3 Implication of Global Soil Consumption to Atmospheric Methane Burden

[39] To assess the role of soil uptake in estimating atmospheric methane mixing ratios, we ran MOZART v4 [Emmons et al., 2010] using our four simulation results (S1, S2, S3, and S4). The horizontal resolution of the model is 1.9° latitude × 2.5° longitude with 56 vertical levels from the surface up to approximately 2 mb. The chemical and transport processes in the model are driven by the meteorological fields from the Modern Era Retrospective-analysis for Research and Applications (MERRA), which have interannual variability [Rienecker et al., 2011].

[40] For emissions other than methane consumption (for which we used the simulated values from TEM), we used the Goddard Institute for Space Studies (GISS) data set (for details, see Fung et al. [1991]). The GISS data set includes emissions from animals, landfills, venting of natural gas at wells, pipeline leakage of natural gas, coal mining, termites, hydrates/clathrates, rice cultivation, wetland ecosystems, forested and non-forested bogs, forested and non-forested swamps and alluvial formation, tundra, and biomass burning. In our simulations, these sink and source data were updated with more recent annually-varying emissions including the Advanced Global Atmospheric Gases Experiment (AGAGE v4.2) for anthropogenic sources ( [Prinn et al., 2000], Global Fire Emissions Database version 3.1 for biomass burning [Van der Werf et al., 2010], and Community Land Model version 4 for wetlands emissions [Melton et al., 2012].

[41] We assumed that the loss mechanism for atmospheric CH4 only included the reaction with hydroxyl radical (OH) and O1D, and these fields were taken from Patra et al. [2011]. We scaled the spatial and temporal pattern of the annually repeating (i.e., no long-term trend) OH field using measurements of methyl chloroform (CH3CCl3) and a three-dimensional climatological OH distribution [Spivakovsky et al., 2000], applying a methodology as analyzed earlier [Prinn et al., 2005]. The lifetime of CH4 using these chemical fields and the emissions data mentioned above was approximately 10 years, which is similar to other current estimates of its lifetime [Boucher et al., 2009]. We ran the model with an initial condition constructed with a latitudinal as well as a vertical gradient and did a spin-up using constant emissions and interannual meteorology from 1990 to the end of 1998. We then ran our four simulations from 1998 until the end of 2004.

[42] Since other sinks and sources and transport are kept the same, but the soil consumption rates are allowed to change in our simulations, the comparisons between the simulation S4 and the other simulations allowed us to assess the relative role of soil methane consumption in affecting atmospheric CH4 ratios. During 1998–2004, the largest root-mean-square differences (RMSD) of global soil consumption were 0.40 Tg CH4 month−1 between S2 and S4 simulations, resulting in 28 Tg CH4 more methane in the atmosphere by the end of 2004. This consumption difference causes 7.1 ppb higher methane mixing ratios in the atmosphere in December of 2004 (Table 7). Comparing the S3 with S4 simulations, we found that the global soil methane consumption induced by agricultural soils played a minor role in affecting atmospheric CH4 mole fractions. In contrast, comparing the S2 and S3 simulations, N deposition suppressed the total methane sink by 26 Tg during the period; as a result, the atmospheric CH4 mole fraction of the S3 simulation was 6.6 ppb higher than the estimates based on the S2 simulation in December of 2004. During this period, a cumulative increase of every 1 Tg of soil CH4 consumption will decrease atmospheric CH4 mole fractions about 0.26 ppb. The large uncertainty in global soil CH4 consumption due to limited knowledge about the effects of climate, atmospheric nitrogen deposition, and land use will result in diverse atmospheric CH4 mole fraction estimates.

Table 7. Role of Global Soil Methane Consumption in Atmospheric CH4 Mole Fraction During the Period of 1998–2004a
  1. a

    Differences and root-mean-square differences (RMSD) are calculated with S4 simulation as a baseline. Monthly atmospheric CH4 mole fractions are calculated with the atmospheric transport chemistry model MOZART.

RMSD of atmospheric CH4 mole fraction (ppb month−1)
Difference of atmospheric CH4 mole fraction in December of 2004 (ppb)
RMSD of global monthly soil consumption (Tg CH4 month−1)
Difference of cumulative global soil methane consumption in the end of 2004 (Tg CH4 period−1)24.527.61.8

5 Conclusions

[43] The effects of multiple factors on the global atmospheric CH4 soil consumption rate were analyzed for the 20th and 21st centuries using a process-based biogeochemistry model. The processes associated with the factors including nitrogen deposition, rising atmospheric CH4 concentrations, agricultural land use, and changing climate are modeled. We found that agricultural land and N deposition changes play minor and moderate roles in determining soil methane consumption, respectively. We estimated that global soils consumed 32–36 Tg CH4 yr−1 during the 1990s and that natural ecosystems are the major sinks while agricultural ecosystems only consume 5.13 Tg CH4 yr−1. During the twentieth century, the consumption rates and seasonal amplitudes (the differences between the lowest and highest monthly consumption during a year) increased mainly due to a warming climate. Arid areas like deserts, shrublands, and xeric woodlands were sink hot spots for consumption. During the 21st century, the projected global soil consumption persistently increased under different future climate scenarios. While dry areas persisted as sinks in the projections, boreal ecosystems became stronger sinks mainly due to increasing soil temperatures. Nitrogen deposition would modestly reduce the future sink strength at the global scale. Our global atmospheric chemical transport model simulations indicated that the global soil methane consumption reduced by nitrogen deposition increased the atmospheric CH4 ratios by 6.6 ppb during the period 1998–2004. On average, a cumulative increase of every 1 Tg soil CH4 consumption will change atmospheric CH4 mole fractions by about 0.26 ppb during the period 1998–2004. More accurate quantification of soil sink therefore deserves attention for future Earth system modeling.


[44] This research is supported by the National Aeronautics and Space Administration (NASA) Land Use and Land Cover Change program (NASA-NNX09AI26G), Department of Energy (DE-SC0007007, DE-FG02-06ER64320 and DE-FG02-94ER61937 to MIT), National Science Foundation (NSF-1028291 and NSF-0919331), the NSF Division of Information and Intelligent Systems (NSF-1028291), and the NSF Carbon and Water in the Earth program (NSF-0630319). The computing is supported by Rosen Center of high-performance computing at Purdue. This study is a contribution to a Working Group of “Toward an adequate quantification of CH4 emissions from land ecosystems: Integrating field and in-situ observations, satellite data, and modeling,” supported by the National Center for Ecological Analysis and Synthesis, a center funded by NSF (grant DEB-0553768), the University of California, Santa Barbara, and the State of California.