Statistical modeling of global soil NOx emissions



[1] On the basis of field measurements of NOx emissions from soils, we developed a statistical model to describe the influences of soil organic carbon (SOC) content, soil pH, land-cover type, climate, and nitrogen input on NOx emission. While also considering the effects of soil temperature, soil moisture change-induced pulse emission, and vegetation fire, we simulated NOx emissions from global soils at resolutions of 0.5° and 6 hours. Canopy reduction was included in both data processing and flux simulation. NOx emissions were positively correlated with SOC content and negatively correlated with soil pH. Soils in dry or temperate regions had higher NOx emission potentials than soils in cold or tropical regions. Needleleaf forest and agricultural soils had high NOx emissions. The annual NOx emission from global soils was calculated to be 7.43 Tg N, decreasing to 4.97 Tg N after canopy reduction. Global averages of nitrogen fertilizer–induced emission ratios were 1.16% above soil and 0.70% above canopy. Soil moisture change–induced pulse emission contributed about 4% to global annual NOx emission, and the effect of vegetation fire on soil NOx emission was negligible.

1. Introduction

[2] Changes in atmospheric trace constituents increasingly affect the radiative balance, dynamics, and chemistry of the atmosphere. Nitrogen oxides (NOx = NO + NO2) do not directly affect the Earth's radiative balance, but they catalyze the formation of tropospheric ozone (O3), an important greenhouse gas [Intergovernmental Panel on Climate Change (IPCC), 2001] that is harmful to both human health and plant growth [Lippmann, 1989; Gregg et al., 2003]. As O3 forms, NO rapidly converts hydroperoxy radicals (HO2) to hydroxyl radicals (OH), thereby indirectly reducing the atmospheric concentrations of carbon monoxide (CO), methane (CH4), and hydrofluorocarbons (HFCs) [IPCC, 2001]. NOx is removed from the troposphere mostly as nitric acid (HNO3), directly contributing to the acidification and eutrophication of regional ecosystems.

[3] Tropospheric NOx originates mostly as NO, which photochemically equilibrates with NO2 within a few minutes. Fossil fuel combustion is the largest contributor to the global NOx budget. Estimates of contributions from other sources, such as lightning, biomass burning, and soil emission, are uncertain. On the basis of field measurements, soils are known to be an important source of atmospheric NOx [e.g., Galbally and Roy, 1978; Johansson, 1984; Slemr and Seiler, 1984; Anderson and Levine, 1987], and estimates of global soil NOx emissions range from 4 to 21 Tg N yr−1 [Holland et al., 1999], although that range can be narrowed through scrutiny of the estimates. The lowest estimate (4 Tg N yr−1) was used by Dentener and Crutzen [1993] in a model and was derived by reducing the original estimate of 10 Tg N yr−1 [Galbally and Roy, 1978] to account for the reduction effect of the vegetation canopy. This value was therefore not an independent estimate of global soil NOx emission. The second lowest estimates were given by Müller [1992], 6.6 Tg N yr−1 for above-soil emission and 4.7 Tg N yr−1for above-canopy emission. The highest estimate (21 Tg N yr−1) drops to 13 Tg N yr−1 when canopy reduction is included, as the authors acknowledged [Davidson and Kingerlee, 1997]. Therefore, when the canopy-reduction effect is included, estimates of global soil NOx emissions range from 4.7 Tg N yr−1 [Müller, 1992] to 13 Tg N yr−1 [Davidson and Kingerlee, 1997].

[4] Three methods have been used to estimate global soil NOx emission: empirical models, process-based models, and simple scaling-up. Yienger and Levy [1995] described an elegant empirical model in which NOx flux is a function of a baseline flux, a pulse effect, and a canopy-reduction effect. The baseline flux is biome specific and is empirically linear, exponential, or not related to temperature, depending on soil temperature itself and soil dryness. A universal value of 2.5% is used as the NOx emission ratio induced by nitrogen fertilizer. Their model yielded estimates of 10.2 Tg N yr−1 for above-soil emission and 5.45 Tg N yr−1 for above-canopy emission.

[5] Potter et al. [1996] used a process-based model based on the conceptual hole-in-pipe model of Firestone and Davidson [1989] to estimate global soil NOx emission. Nitrogen flux through the pipe, the gross nitrogen mineralization, was simulated by an extended version of the CASA-Biosphere model [Potter et al., 1993]. The hole size, that is, the potential production of nitrogen trace gases (NO + N2O + N2), was set at 2% of the gross mineralized nitrogen. The ratio of NO:N2O:N2 production is determined by the soil moisture computed from a soil moisture submodel. The annual NOx emission, not including the canopy effect, was simulated as 9.7 Tg N yr−1. As noted by Hutchinson et al. [1997], the advantages of this process-based model were compromised by the assumptions that a uniform 2% of the mineralized nitrogen was lost to the atmosphere and that the ratio of the three gases can be approximated by the monthly soil moisture index.

[6] Davidson and Kingerlee [1997] compiled field measurements from various biomes and derived a mean annual NOx flux for each biome. The annual flux was multiplied by the global area of each biome, and a global soil NOx emission of 21 Tg N yr−1 was calculated. An above-canopy emission of 13 Tg N yr−1 was estimated when the canopy-reduction scheme of Yienger and Levy [1995] was applied. However, the above-soil emissions in this study were estimated from direct field measurements, and many of them were above-canopy fluxes [e.g., Slemr and Seiler, 1991; Skiba et al., 1992; Parson et al., 1996]. In such cases, the canopy effect is counted twice. Another weakness in this scaling-up method is the extrapolation of annual fluxes from individual measurement periods. Some measurement periods are very short, and extrapolation without considering the strong seasonality of factors such as fertilization can introduce large uncertainties.

[7] As noted above, large-scale estimations of soil NOx emissions are usually empirical. Even if a process-based model is used, an empirical relationship is assumed for some critical processes [e.g., Potter et al., 1996]. Because NOx production in soils occurs through microbial nitrification and denitrification, models simulating such processes must specify microbial growth rates, reaction rate constants, solubility coefficients, diffusivities, and other terms. Although it may be possible to obtain the required parameters for short time periods at individual sites, it is much more difficult for large-scale process modeling [Hutchinson et al., 1997]. As Jørgensen and Bendoricchio [2001] noted, the biological world is a sloppy place, and very precise predictive models will inevitably be wrong. A more fruitful endeavor would be to build a model that predicts general trends and accounts for the probabilistic nature of the environment.

[8] We present a statistical model that links NOx emissions from soils to the most significant variables, such as nitrogen fertilization, vegetation type, climate, and soil properties. Integrating global data sets of such variables into the simulations introduces environmental and management heterogeneity into the model fields. Such heterogeneity has not been sufficiently considered in previous estimation models.

2. Materials and Methods

[9] We collected NOx flux data from publications and linked the average NOx fluxes during measurement periods to land cover, soil pH, soil organic carbon (SOC), climate, and nitrogen fertilization through a statistical model. Factors that have short-term effects on NOx flux, such as sudden changes in soil moisture (rainfall on dry soils) and fire, were also considered by attaching empirical functions derived from field measurements. Temporal variation of NOx flux was approximated by introducing a temperature function in addition to the seasonality of nitrogen fertilization, soil moisture change-induced pulse emission, and vegetation fire. Finally, global soil NOx emissions were simulated using the statistical model, the empirical functions, and global data set of those controlling factors.

2.1. Observation Data

[10] We surveyed the literature for field measurements of NOx emissions from soils and documented flux, soil properties, fertilization, vegetation, location, and other pertinent information for statistical analyses. Information on some influential factors, such as soil moisture content, was rarely reported. Global data sets for other soil properties closely related to NOx emission, such as NH4+ and NO3 content, were not available for use as input parameters. Therefore not all of the field measurement results or factors could be considered in the statistical analysis. We included only measurements accompanied by information on SOC, soil pH, nitrogen fertilization rate, vegetation type, and location. Field NOx fluxes may vary greatly over time owing to short-term events, such as a sudden rain on dry soil or vegetation fire, and fluxes following nitrogen fertilization tend to be higher than average; thus short-term fluxes may bias the result. When considering nitrogen fertilization treatment, we included only results with a measurement period longer than 4 months (one crop season); for other treatments, we required measurements to be longer than 1 week. After data selection, the measured fluxes fit a lognormal distribution rather than a normal distribution except a few data with negative or zero flux. Because the few negative fluxes were based on relatively short-term measurements (18–24 days) [Slemr and Seiler, 1984, 1991] and for the convenience of statistical analysis (considering a lognormal distribution), we excluded these data as well. The resulting data set is presented in the auxiliary material.

[11] The Koeppen system [Food and Agriculture Organization (FAO), 2003a] was used to classify the climate of the individual sites as dry, tropical, temperate, or cold. Land-cover type was classified as one of the 16 land-cover types in the U.S. Geological Survey (USGS) land-use database with the International Geosphere-Biosphere Programme (IGBP) legend [USGS, 2003]. Wetland rice fields are often flooded, and their NOx emission fluxes are usually much lower than those in the surrounding uplands [Galbally et al., 1987]; thus rice fields were classified as wetland rather than cropland. If an estimate was not reported, the nitrogen input from animal droppings was assumed as 100 kg N ha−1 for measurements from grazing pastures. All of the factors listed in the auxiliary material significantly influence NOx flux in the statistical model, as discussed later.

[12] NO emitted from soils is quickly oxidized to NO2 by O3 near the canopy, and the newly formed NO2 may be deposited onto the plant canopy [Bakwin et al., 1990; Jacob and Wofsy, 1990]. This canopy effect is essential in explaining differences in NOx fluxes from soils measured by different methods or over different ecosystems because the canopy can remove up to 75% of the NO2 [Jacob and Wofsy, 1990]. When flux is measured from a tower over a forest or by a chamber over grasslands, the canopy effect is included in the measured fluxes. However, if a chamber is inserted directly into the forest floor, the measured fluxes might be much higher than the actual fluxes over the forest canopy. Therefore, in our model, reported flux data are always classified as soil flux or canopy flux. If the data represent canopy flux, soil flux is derived from the following equation:

equation image

where CRF is a canopy-reduction factor defined by Yienger and Levy [1995] as

equation image

where SAI and LAI are stomatal area index and leaf area index, respectively, and ks and kc are constants.

[13] The model of Masson et al. [2003] was used to find the average LAI of the specific vegetation type in the measurement grid for the measurement time. A corresponding SAI was derived from the SAI:LAI ratio of Yienger and Levy [1995]. Modeled LAI and SAI values were different from those of Yienger and Levy [1995], so ks and kc were calculated using the CRF of Jacob and Wofsy [1990] for rain forest, a LAI of 6 derived from the Masson model, and an estimated SAI of 0.09; the resulting ks and kc were 11.6 and 0.32, respectively. These values were then used to estimate CRF in our model.

2.2. Statistical Analysis of NOx Flux

[14] To account for the heterogeneity of NOx fluxes from soils, Yienger and Levy [1995] divided soils into wet and dry groups and then derived mean flux for each biome (land-cover type) of each group. Nitrogen-induced emission was uniformly set to be 2.5% of the total fertilizer used. More measurements have been made since 1995, however. Available data now allow us to include the heterogeneity of more environmental factors, such as soil properties, and to consider the probabilistic nature of their effects through statistical analysis of field data. Our statistical model allows us to estimate the effects of different factors interdependently.

[15] The NOx emission data sets did not arise from systematically designed experimental results; rather, we used them because they were available. They are unbalanced data; that is, they have unequal numbers of observations in the subclasses. Unbalanced data are statistically analyzed by mixed models that include both fixed and random effects [Searle, 1987]. For example, Bouwman et al. [2002] used a linear mixed model to analyze log-transformed data of nitrogenous gas emissions from agricultural land. Back-transformation yields an exponential relationship between emission and controlling factors as follows:

equation image

Nitrogen fertilization is one of the factors in this equation. According to Bouwman et al. [2002], this equation indicates that the calculated emission increases more than proportional to nitrogen application rate. Often in large-scale estimation, only the average nitrogen fertilization rate is known for a given area and the exact rate varies greatly within that area. Equation (3), combined with uncertainties in the geographical distribution of nitrogen application rates, inevitably results in overestimation of the emission. To ameliorate this problem, we used a nonlinear mixed model to describe the relationship between log-transformed flux data and influencing factors as follows:

equation image

where “flux” is the average NOx emission flux during the measurement period; “factor” includes land-cover type, climate, SOC, and soil pH; n_eff is the estimated effect of nitrogen fertilization; and Nrate is the nitrogen fertilization rate. The equation for flux is therefore

equation image

and the calculated flux is proportional to the nitrogen application rate.

[16] The SAS/STAT procedure NLMIXED was used to fit equation (4) to the observations shown in the Appendix. All the influencing factors, including land-cover type, climate, SOC, soil pH, and nitrogen fertilization rate, were chosen to have fixed effects in the model, and the measurement site was set to have a random effect. This procedure estimates the parameters by maximizing an approximation to the likelihood integrated over the random effect. Because the effects of soil moisture change–induced pulse emission and vegetation fire are short-lived (discussed later), these factors are not considered in our statistical model, which links average NOx flux to environmental factors over a relatively long period. Rather, these effects are simulated separately, as described later.

[17] As shown in the auxiliary material, land cover and climate were categorical variables, and soil organic carbon content, soil pH, and nitrogen fertilization rate were continuous variables. However, databases of global SOC content and soil pH are available only in category format. Therefore SOC and pH were classified following the classification system in the FAO soil database [FAO, 1995]. The five pH levels were: <4.5, ≥4.5–5.5, >5.5–7.2, >7.2–8.5, and >8.5. The five SOC levels were: <0.2%, ≥0.2%–0.6%, >0.6%–1.2%, >1.2%–2%, and > 2%.

2.3. Simulation of Global Soil NOx Emission

2.3.1. Temperature Effect

[18] Equation (5) models the average flux during the measurement period. To obtain a dynamic emission flux, it is necessary to consider the temperature effect because NOx flux generally increases exponentially with temperature [Williams et al., 1992; Martin et al., 1998]. Some researchers have noted no increase in flux when soil temperatures exceeded 30°C [Rolle and Aneja, 2002] or no decrease in flux with temperature over very dry soil [Cardenaset al., 1993]; both of these findings were likely due to changes in soil water content. Under field conditions, flux changes are the integrated effect of all influencing factors. Under laboratory conditions in which soils were held at constant moisture content, fluxes from different soils increased exponentially with temperature up to 48°C [Ormeci et al., 1999].

[19] Temperature influence on NOx flux is usually expressed as

equation image

where A is the flux at reference temperature T0, and B is the sensitivity of NOx flux to temperature changes. Table 1 lists values of B, either directly taken or derived from the literature; and the results showed that the temperature response to NOx flux is independent of vegetation and climate. Some past studies suggested that soil temperature has little or no effect on NOx flux [Kaplan et al., 1988; Cardenas et al., 1993]. In these studies, there was either little change in soil temperature, such as the tropical area of Kaplan et al. [1988], or the temperature effect was overwhelmed by the soil water content effect in the field [Cardenaset al., 1993]. In the present model, therefore, we used a uniform temperature response of flux based on the averaged B values in Table 1; that is,

equation image

where T is the surface soil temperature in Kelvin, and T0 is the average of the surface soil temperature at all measurement sites during the measurement month in each climate zone. T0 values, calculated from European Centre for Medium-Range Weather Forecasts (ECMWF) monthly mean surface analyses [ECMWF, 2003], are respectively 298.87, 290.67, 295.76, and 286.69 K for tropical, temperate, dry, and cold climates. The average B value of 0.11 is close to the value of 0.103 used by Yienger and Levy [1995].

Table 1. Reported Temperature Effects on NOx Emissions From Soils (B Value)
VegetationClimateB ValueReferencea
Grass leytemperate0.1231
Grass ryetemperate0.0644
Grass ryetemperate0.124
Grain croptemperate0.1412
Average 0.11 

2.3.2. Canopy-Reduction Function

[20] Whereas equation (7) gives the flux from the soil, the above-canopy flux is calculated by applying a CRF as follows:

equation image

[21] The CRF calculation is shown in equation (2). Monthly LAI at a 0.5° resolution for each vegetation type was obtained by running the Ecoclimap model of Masson et al. [2003].

2.3.3. Soil Moisture Effect

[22] As a microbial process, NOx emission from soils is also moisture dependent, but the response is not monotonic. At low moisture content, the emission is substrate limited; at high moisture content, it is limited by gas diffusion. Both field and laboratory measurements have confirmed that NOx emission has an optimum soil moisture content, which varies from 20% to 45% water-filled pore space [Yang and Meixner, 1997; Ormeci et al., 1999]. Thus soil moisture may affect the annual emission and the temporal variation of NOx flux. In the current simulation, the annual emission is basically determined by the factors presented in the statistical model, in which, the climate factor partially reflects moisture effect. In upscaling NOx emission from soils in Zimbabwe, Kirkman et al. [2001] approximated soil moisture effect by a first-order empirical function involving several parameters obtained through laboratory analysis. As the parameterization varies with soil, it is difficult to do at global scale. In the current simulation, we therefore ignored this moisture effect on the temporal variation in NOx emission.

[23] Another effect of soil moisture on NOx emission is the burst of emission when very dry soil becomes wet. This pulse-type emission occurs not only for NOx but also for carbon dioxide (CO2) and N2O [Scholes et al., 1997; Franzluebbers et al., 2000], simply because carbon and nitrogen mineralization both increase when dry soils become wet [Cui and Caldwell, 1997; Franzluebbers et al., 2000]. Pulse emission was not considered in the statistical analysis, and some pulse-affected flux data were excluded from the auxiliary material owing to an insufficiently long measurement period, but pulse emission may have significant effect on short-term flux. Thus we attached a pulse emission function in the statistical model.

[24] It is unclear how the magnitude and duration of the NOx pulse are related to the length of the antecedent drying period and the amount of water added. Johansson and Sanhueza [1988] found a relationship between the amount of rainfall and the length of the pulse, and Yienger and Levy [1995] assumed the magnitude and duration of the pulse to be a function of rain. In contrast, Martin et al. [1998] reported that the pulse was independent of the amount of water added. Davidson [1992a] showed that wetting dry soil to above or below field capacity caused no significant difference in pulse magnitude, and a light rain could cause an even larger pulse [Davidson et al., 1991]. Johansson et al. [1988] also reported that pulses caused by 15-mm rains were not significantly higher than those caused by 3-mm rains.

[25] Martin et al. [1998] reported that the pulse was independent of the antecedent dry period. However, other researchers showed that the pulse magnitude was related to the dry period [Davidson, 1992b] or that the pulse decreased in response to successive wetting [Slemr and Seiler, 1984; Johansson et al., 1988; Davidson et al., 1991]. Again, this is not unique to NOx, but also occurs for CO2, as Fay et al. [2000] showed that the time between rainfall events affects soil CO2 flux.

[26] There is some indication that the duration of the pulse depends on the magnitude of the pulse. When the peak flux value was about 10 times the value before wetting, the flux returned to the original value in about 24 hours [Davidson, 1992a; Levine et al., 1996; Martin et al., 1998]; when the peak flux value was about 20 times the flux value before wetting, the pulse dropped to the control value within 2 days [Johansson et al., 1988]; pulses with peak values of 32–38 times the prewetting flux declined to a stable value within 3 days [Scholes et al., 1997]; and fluxes with peak values 100 times greater than the prewetting flux maintained a flux of over 10 times the original value after 24 hours and the pulse lasted for 7 days [Davidson et al., 1991].

[27] For the above reasons, in this work the peak value of the pulse (relative to the flux prewetting) is determined by the antecedent dry period, with a simple logarithmic relationship,

equation image

where ppeak is the magnitude of peak flux relative to prewetting flux, and tdry is the length of the antecedent dry period in hours. The constants are obtained from data in Table 2.

Table 2. Information on Pulse Emissions That Occurred After Dry Soils Were Wetted
Antecedent Dry DaysPeak PulseTime to End of Pulse, hoursCalculated b, hour−1Reference
43.5–15200.110Martin et al. [1998]
7610–20440.062Johansson et al. [1988]
18032–38680.051Scholes et al. [1997]
138100920.050Davidson et al. [1991]

[28] The flux decays after the peak value as follows:

equation image

where p is the magnitude of flux relative to prewetting, t is the time since the peak in hours, and b is a constant. Integrating equation (10) and rearranging yields p = ppeakebt; thus

equation image

where tend is the time from pulse peak to pulse end (i.e., p = 1). The average b value derived from data in Table 2 is 0.068.

[29] Yienger and Levy [1995] used precipitation data to determine what triggers a pulse. Pulses usually occur with soil moisture contents of 0.5–3% (w/w) [Slemr and Seiler, 1984; Johansson and Sanhueza, 1988; Johansson et al., 1988; Davidson et al., 1991; Davidson, 1992a]. Rain had no stimulating effect on soils with moisture contents of 7.7% (w/w) [Davidson et al., 1991]. However, in a large-scale land-surface model, soil moisture content rarely reaches such low values, except in hot desert areas. Model soil moisture content is usually averaged over a grid. Even in tropical savanna regions, where pulses of NOx emission are often reported, soil moisture content in the 6-hour ECMWF reanalysis data at 0.5° resolution for 2001 [ECMWF, 2003] never reached a value low enough to trigger a pulse. Instead, the soil moisture contents during the dry season in these arid and semi-arid savanna areas were usually below 17.5% (v/v). With the first rain, the moisture content increases sharply. In this work, therefore, dry soil was defined as soil with a moisture content below 17.5% (v/v) in the ECMWF data set, and an increase of >0.5% (v/v) in the moisture content of soil that experiences dry conditions for at least 3 days was assumed necessary to trigger a pulse. An increase of 0.5% (v/v) in 7 cm of surface soil is equivalent to about 3.5 mm of rainfall, which is the rainfall amount often reported to cause a pulse [Johansson and Sanhueza, 1988; Johansson et al., 1988; Martin et al., 1998].

2.3.4. Fire Effect

[30] Johansson et al. [1988] reported that NO flux increased 10 times after fire and decreased slowly during the following 4 days. Poth et al. [1995] reported that NO flux from a site burned 2 days before was higher than that from an unburned site, but flux from a site burned 30 days before was not. Data reported by Levine et al. [1996] showed that burning increased NO fluxes 3–10.8 times, but the fluxes usually returned to preburning level within 4–10 days. Thus it appears that, although burning significantly stimulates gross nitrogen mineralization [Anderson and Poth, 1998] and consequently NO emission from soil, the effect is short-lived. Therefore fire-influenced fluxes were excluded from the statistical analysis. To estimate the effect of global fire on annual soil emission, we assume that a fire increases NO flux from soil to 10 times preburning level, and the stimulating effect disappears exponentially in 7 days. We used the Global Burned Area data set on a 0.5° grid with monthly time resolution for 2000, (, and assumed the fire was evenly distributed with a monthly and a 0.5° grid.

2.3.5. Land-Cover Data Set

[31] A global land-cover data set is needed to model global soil NOx emission. In this study, the 1-km resolution land-cover map from the USGS with the IGBP legend [USGS, 2003] was used. However, the map groups hot desert and semi-desert (sparsely vegetated) into one class, “barren or sparsely vegetated.” Small fluxes occur over sparsely vegetated deserts [Hartley and Schlesinger, 2000], but NOx flux should not be expected over hot deserts such as the Sahara. Therefore “barren” and “sparsely vegetated” land were separated by replacing some of the “barren and sparsely vegetated” pixels in the USGS land-cover map with “barren” pixels of another 1-km resolution land-cover data set from the University of Maryland [Hansen et al., 2000]. Cropland and grassland areas of a political unit in the land-cover map may not match actual data. Thus “croplands,” “croplands/natural vegetation mosaic,” and “grasslands” in the land-cover maps were adjusted so that the calculated areas equaled the data in statistics. Wetland rice is often flooded and accordingly has a different microclimate and smaller NOx emissions [Yan et al., 2003a] than do other croplands. Therefore rice fields were separated from other croplands by using a rice:total crop ratio calculated from statistical data for each country, each state of the United States and India, or each province of China.

[32] The statistical analyses of data in the auxiliary material defined the effects of land-cover types. However, not all land-cover types are included in the auxiliary material. In this model, the effect of “evergreen needleleaf forest” (USGS data set with IGBP category) was also used for “deciduous needleleaf forest,” the effect of “sparsely vegetated” was used for “open shrublands,” and the effect of “wetlands” was used for “wetland rice field.” NOx emissions from urban and built-up areas, snow and ice, hot desert, and water bodies were assumed to be zero.

2.3.6. Nitrogen Input

[33] Agricultural activity data, including cropland area, wetland rice area, grassland area, chemical nitrogen fertilizer consumption, and animal population, for 2001 for countries except China, India, and the United States were obtained from the FAO online database [FAO, 2003b]. Data for individual provinces in China, states in India, and states in the United States were from Editorial Board of China Agriculture Yearbook [2002], Agriculture Statistics Division [2000], and U.S. Department of Agriculture (USDA) [2003], respectively. Note that the data for India and the United States are from 1999 and 1997, respectively, and were scaled to 2001 using national data in the FAO database [FAO, 2003b]. In some countries, part of the chemical nitrogen fertilizer is used on grassland, and ratios of nitrogen fertilizer used for grassland were derived from International Fertilizer Industry Association, International Fertilizer Development Center/Food and Agriculture Organization (IFA/IFDC/FAO) [1999] and USDA [2003]. Total animal manure nitrogen for each political unit was calculated from animal population data and parameters of Mosier et al. [1998]. Manure nitrogen in waste management systems of “liquid system,” “daily spread,” “solid storage and drylot,” and “pasture range and paddock” were assumed to be available to soils after allowing 20% loss through ammonia volatilization. The percentage of manure nitrogen used on crops and grasslands depends on waste management systems and regions, and it varies remarkably across countries. For example, in Africa, where most animals are grazers, only about 4% of the manure nitrogen is applied to croplands. In Asia, about half of manure nitrogen is applied to croplands. In North America, about 20% is applied to croplands (details not shown).

[34] Because it is difficult to determine exactly where chemical nitrogen and animal manure have been applied, they are assumed to be evenly distributed over the croplands within a political unit, which might not be true in practice. This will not greatly affect the total amount of emissions, however, because NOx flux is proportional to the nitrogen fertilization rate in the model. Because nitrogen is applied to support crop growth, the timing of nitrogen application was assumed to parallel LAI changes with a 1-month phase difference. After application, nitrogen is not assumed to persist year-round but is assumed to be effective for one crop season. Therefore we take the effective time as 122 days. If the annual fertilization rate for a cropland is ANrate, for a given day, the effective nitrogen fertilization rate, ENrate, is

equation image

where LAIave is the annual average LAI of the cropland, and LAIm+1 is the LAI of the following month. For grassland, nitrogen is assumed to be evenly applied in a year; therefore

equation image

Accordingly, the Nrate in equations (7) and (8) was replaced by ENrate.

2.3.7. Other Data Sets

[35] Other input data sets include the FAO Koeppen Climate Classification map [FAO, 2003a], an FAO soil database of dominant surface soil pH and SOC content at 5-min resolution [FAO, 1995], and the ECMWF reanalysis data of surface soil temperature and moisture content in 2001 at 6-hour intervals [ECMWF, 2003]. The model was run with a 6-hour time step. In each 0.5° grid, the emission from each type of land cover was calculated separately and then summarized.

3. Results

3.1. Effects of the Variables

[36] The estimated effects of the variables are listed in Table 3, showing that all five variables have significant effects on NOx flux at the <5% level. The use of nitrogen fertilizer increases NOx emission linearly with application rate, as designed in the model, and the absolute amount of fertilizer-induced emission is dependant on environmental factors; that is, fertilizer-induced emission is proportional to background emission. For example, for an application rate of 100 kg N ha−1, fertilizer-induced emission is 3.5 times background emission.

Table 3. Effects of Controlling Variables on NOx Emission From Soils, Obtained by Fitting Equation (4) to Data in Auxiliary Material Using SAS/STAT PROC NLMIXED
VariablesEffectStandard Errort ValuePr > ∣t
  • a

    IGBP land-cover type: 1, evergreen needleleaf forest; 2, evergreen broadleaf forest; 4, deciduous broadleaf forest; 5, mixed forest; 6, closed shrublands; 9, savannas; 10, Grasslands; 11, permanent wetlands; 12, croplands; 16, barren or sparsely vegetated.

Land covera
Soil pH

[37] Figure 1 illustrates the relative effects of various factors and their subclasses. Soils under needleleaf forests show the highest NOx emission potential among all the land-cover types, followed by croplands. Emissions are low from wetland, closed shrublands, and grasslands. NOx emissions are significantly, and counterintuitively, given the temperatures, lower in tropical than in temperate climates. Soils in dry climates show the highest NOx emission potential. NOx emissions are negatively correlated with soil pH and positively correlated with SOC.

Figure 1.

Estimated effects of (a) land cover, (b) climate, (c) soil organic carbon, and (d) soil pH on NOx emission. The effects were estimated from log-transformed flux data.

[38] Modeled fluxes agree reasonably well with the observations in the auxiliary material, especially for croplands sites (Figure 2), the emission from which is more influenced by nitrogen fertilization. However, the fluxes were log-transformed in the model; when back-transformed, the average of the modeled flux is 13% lower than that of observations. This can be partially attributed to the omission of moisture change–induced pulse emission and fire effect. To ameliorate this problem, additional moisture change–induced pulse emission and fire functions were attached to the statistical model as described earlier.

Figure 2.

Comparison of observed fluxes and fluxes simulated by the statistical model. Both observed and simulated plots are of above-soil fluxes (ng N m−2 s−1) and are log-transformed.

3.2. Simulated Annual Emission

[39] The model predicts global above-canopy NOx emissions of 4.97 Tg N yr−1. Table 4 shows emissions from each land-cover type, with croplands and grasslands being the largest contributors. Large emissions from croplands and grasslands can be attributed in part to the use of chemical nitrogen fertilizer and animal manure, which induce emissions of 0.57 and 0.51 Tg N yr−1, respectively. Globally, fertilizer-induced above-soil and above-canopy emissions are 1.16% and 0.70% of the total chemical nitrogen applied. Average NOx flux from savannas is high, second only to croplands. Above-soil fluxes from forests are similar to grasslands but smaller than grasslands after canopy reduction.

Table 4. Global NOx Emission From Soils Before and After Canopy Reduction
SourcesAreaa 106 km2Emission, Tg N yr−1Flux, kg N ha−1 yr−1
Above SoilAbove CanopyAbove SoilAbove Canopy
  • a

    Areas were computed from the mixed 1-km land-use database.

Rain forest13.20.670.210.510.16
Other forests18.80.870.510.460.27
Sparsely vegetated3.
Permanent wetland0.960.0020.0010.020.01

[40] The application of the canopy-reduction function reduces the global total soil emissions of NOx from 7.43 to 4.97 Tg N yr−1 (Table 4), indicating the average canopy-reduction effect is 33%. Canopy reduction is directly related to the LAI, varying greatly with land cover from 15% over sparsely vegetated land to nearly 70% over rain forest.

[41] Table 5 summarizes above-canopy emissions from the main land-cover types for each continent or country. Emissions over Asia are 1.48 Tg N yr−1, representing nearly 30% of the global total, and resulting from heavy use of nitrogen fertilizer. High emission fluxes from savanna produce a total emission of 1.37 Tg N yr−1 in Africa. Emissions over China, India, and the United States, where croplands are by far the dominant source, account for 25% of the total global emissions.

Table 5. Regional Breakdown of Global Above-Canopy NOx Emissions
RegionTotal EmissionaCroplandsGrasslandsSavanna
EmissionPercent of TotalEmissionPercent of TotalEmissionPercent of Total
  • a

    Emission is in Gg N yr−1.

   Other Asian countries61024940.811318.5183.0
   North America63721633.99915.6284.3
   Unites States40616941.67017.2143.5
   Other North American countries2314720.43012.9135.8
South America5668414.820936.913323.4

[42] Figure 3 illustrates the global distribution of above-canopy NOx emissions. High annual emissions occur in Bangladesh, where the animal population is large; northern China and northern India, two main agricultural regions with heavy fertilizer use; the Florida peninsula, which is mainly cropland with soils of low pH and high organic carbon content; and the savannas of Africa. Emissions from forested areas are low, mainly owing to greater canopy reduction. About 65% of the global above-canopy emissions occurs in the tropics (30°S to 30°N), and one third occurs poleward of 30°N. Emissions poleward of 30°S account for less than 5% of the global total. The Northern and Southern Hemispheres contribute two thirds and one third of the global emissions, respectively.

Figure 3.

Simulated global annual soil NOx emission (after canopy reduction). Grid size is 0.5° by 0.5°.

[43] Seasonal variations in above-canopy emissions are controlled mostly by surface soil temperature and LAI, except for the pulse emissions that follow rainfall on dry soils. Because emissions from the Northern Hemisphere dominate global emissions, the global seasonal variation follows that of the Northern Hemisphere, where the highest (lowest) monthly emissions occur in July (February) (Figure 4). Southern Hemisphere emissions show an opposite pattern. However, maximum monthly emissions from savanna in the Northern Hemisphere are in May, not July, and maximum monthly emissions from savanna in the Southern Hemisphere are in October, not December/January. This is because of pulse emissions at the start of the rainy season.

Figure 4.

Seasonality of NOx emissions from global terrestrial ecosystems (dots), Northern Hemisphere terrestrial ecosystems (solid squares), Southern Hemisphere terrestrial ecosystems (open squares), Northern Hemisphere savanna (solid triangles), and Southern Hemisphere savanna (open triangles).

[44] Pulse emissions occur mainly in dry climate regions such as Australia, tropical Africa, the Middle East, and central Asia. Pulse emissions contribute about 4% of the annual global totals. In some specific grids, however, pulse emissions represent more than 20% of the annual total (Figure 5).

Figure 5.

Simulated daily soil NOx emission at a grid point (1°W, 17°N). Emissions during the three pulse events account for 21% of the annual total.

[45] At a global scale, the stimulating effect of fire on soil NOx emission is negligible, only about 0.1% of the annual emission. This is not surprising, since the effect of fire lasts only several days and only 2.3% of the global land area was burned in 2000 [Tansey et al., 2004].

4. Discussion

4.1. Effects of Modeled Factors

[46] Nonlinear mixed models simultaneously estimate the effects of different factors and express them linearly and nonlinearly. Flux data for NOx emissions from soil fit a lognormal distribution and were analyzed after log-transformation. Thus factors with linear effects had exponential effects when flux data were back-transformed. For factors with spatial distributions that are not known precisely but represented by values averaged over a large area, a nonlinear model is preferable to describe the effects, as was the case for the nitrogen fertilization rate in the current nonlinear mixed model.

[47] The NOx emission potentials of different land-cover type estimated through the statistical model generally agree with the following ranking: fertilized agricultural fields > savannas > forests > other natural systems, which is similar to that of Ludwig et al. [2001]. However, large differences exist among different forests. We estimated that soils under needleleaf forests have the highest NOx emission potential (Figure 1). Higher NOx fluxes from needleleaf than from broadleaf forest sites have been observed widely [Gasche and Papen, 1999; van Dijk and Duyzer, 1999; Venterea et al., 2003]. Aber et al. [1998] proposed that deciduous forest have a greater capacity to retain nitrogen input than coniferous forests. We speculate that the higher NOx emission potential of coniferous forest soils can be partially attributed to their low soil pH. As statistical results have shown (Figure 1) and as will be discussed later, low soil pH favors NO production. Although the effect of soil pH was also considered in the statistical model, the classification of soil pH was coarse and the differences could not be fully reflected in the model; however, soils under coniferous forests generally are more acidic than soils under broadleaf forest with similar land-use history [e.g., Gasche and Papen, 1999; Venterea et al., 2003].

[48] The statistical results indicate that, even without nitrogen input, NOx emissions from croplands are greater than those from most other ecosystems. This might be caused by the residual effect of nitrogen input in previous seasons and/or the acceleration of nitrogen mineralization by agricultural activities.

[49] At process level, soil moisture strongly affects NOx emission by controlling the transport of microbial substrates and products, the ratio of NO:N2O:N2, and the induction of pulse emissions. In the statistical model, owing to difficulties in parameterization, we did not explicitly use soil moisture except introducing a pulse function. The effect of climate, however, should to certain extent reflect the moisture effect on NOx emission. The estimated high NOx emission potential from soils in dry climate and low potential from soils in tropical climate agree well with the conclusion that optimum moisture for NOx emission is relatively low (20–45% water-filled pore space [see Yang and Meixner, 1997; Ormeci et al., 1999]).

[50] The statistical results confirm findings from both the laboratory and field that low soil pH values are correlated with high NO emission rates [Nägele and Conrad, 1990; Ormeci et al., 1999; Venterea et al., 2003]. Several reasons may account for this. Following the pioneering work of Wijler and Delwiche [1954], numerous studies showed that, although the total denitrification gaseous emissions (NO, N2O, and N2) to the atmosphere are less from acidic soils than from neutral or slightly alkaline soils, the ratio of NO production to total denitrification gases increases as pH decreases because low soil pH inhibits NO reduction to N2O [see Šimek and Cooper, 2002]. Low soil pH may also promote chemical decomposition of nitrite, yielding NO as the primary product [McKenney et al., 1990]. The low NOx emission potential of alkaline soils (Figure 1) also supports the experimental result that N2 is the dominant, or perhaps only, denitrification product in soils with pHs of 6.9–8.0 [Weier and Gilliam, 1986].

[51] Given the importance of soil organic carbon as the source of energy and substrate for microorganisms in soil, it is reasonable that high organic carbon content favors NOx emissions (Figure 1). Soil organic carbon also enhances NO production through chemical denitrification [Stevenson, 1994].

[52] Although all categorical variables (i.e., climate, land cover, SOC, and pH) significantly influence NOx emissions, soil pH most strongly accounts for the variations in NOx fluxes, as reflected by the large variability in the effects of different pH classes (Table 3). Differences in the effects between pH classes <4.5 and >7.2–8.5 mean that, with other conditions identical, NOx emissions from very acidic soils are 25 times those from very alkaline soils. The least important variable is SOC content.

4.2. Validation

[53] To validate the statistical model with an independent data set, we used field observations that were excluded from the statistical analysis because of the lack of SOC data, as the statistical results showed that SOC is the least sensitive among all the factors in the statistical model (see Table 3). To avoid bias, the selected observations (Table 6) had a measurement period of at least 1 year and the SOC value could be reasonably estimated. The SOC values in these observations were estimated to be >2% because they are either from tropical rain forest or from peatland. The modeled and observed fluxes show a correlation coefficient of 0.88, and the average of the 10 modeled fluxes is very close to that of the observations.

Table 6. Data Used to Validate the Statistical Modela
LocationLand-Cover TypeClimateSOC, %bSoil pHN Input, kg ha−1Measurement, daysFlux, ng N m−2 s−1Reference
  • a

    Selection criteria are: measurement period at least 1 year, and SOC content unknown, but can be best guessed.

  • b

    SOC was estimated to be >2.0%; see text for reason.

  • c

    Inputs are from nitrogen deposition.

  • d

    Nitrogen input from cow manure was estimated at 100 ha N ha−1.

Germanyevergreen needleleaf foresttemperate>2.02.930c109525.531.44Gasche and Papen [1999]
Germanyevergreen needleleaf foresttemperate>2.05.930c109513.111.95Gasche and Papen [1999]
Germanydeciduous broadleaf foresttemperate>2.0420c7307.98.95Gasche and Papen [1999]
Finlandmixed forestcold>2.04.504261.141.33Lång et al. [1995]
Finlandmixed forestcold>2.04.504261.971.33Lång et al. [1995]
Finlandgrasslandcold>2.05.3180d42617.786.83Lång et al. [1995]
Costa Ricaevergreen broadleaf foresttropical>2.03.603652.864.17Keller and Reiners [1994]
Costa Ricagrasslandtropical>2.04.603650.751.65Keller and Reiners [1994]
Costa Ricagrasslandtropical>2.03.903654.693.60Keller and Reiners [1994]
Costa Ricaevergreen broadleaf foresttropical>2.03.703651.254.17Keller and Reiners [1994]

[54] A simulated result over a grid is an average value and only indicates the general trend. Thus it is inappropriate to compare the simulated result of a single grid with site-specific measurements that can vary greatly within a grid. Results can be compared on large scales, however, if the sources are comparable. Kirkman et al. [2001] described extensive measurements and a detailed model of soil NOx emissions over Zimbabwe. Their estimates, based on field measurements and model analysis, were 22.8 and 32.9 Gg N yr−1, respectively. These values agree well with the estimate for Zimbabwe of 27.8 Gg N yr−1 in this study.

4.3. Comparison With Other Estimates

[55] Compared with the results of Yienger and Levy [1995], this work estimates more emissions over eastern China and northern India (Figure 6), both being important agricultural regions with heavy nitrogen input. The main reasons for this difference are that chemical nitrogen consumption increased in these regions since Yienger and Levy published their results, and we account for nitrogen input from animal manure in the current work. However, chemical nitrogen fertilizer–induced emissions from global croplands in this study are only 0.67 Tg N yr−1, smaller than the 1.04 Tg N yr−1 reported by Yienger and Levy [1995]; because available field measurements were limited, Yienger and Levy set fertilizer-induced emissions at 2.5% of the fertilizer applied for above-soil emission and 1.8% after canopy reduction. The estimated fertilizer-induced ratio in the current work is 1.16% for above-soil emissions and 0.70% for above-canopy emissions, in good agreement with those derived by Bouwman et al. [2002] and Yan et al. [2003b] from a large number of field measurements. In this work, agricultural activity data such as animal population and fertilizer consumption were collected at the province or state level for China, India, and the United States. For these reasons, we think the estimate for agricultural emission and its geographical distribution in this work is more reliable than that of Yienger and Levy [1995].

Figure 6.

Differences between simulated annual NOx emissions (after canopy reduction) of this study and that of Yienger and Levy [1995]. Green indicates estimates that were higher in the study by Yienger and Levy, and red indicates estimates that were higher in this study.

[56] Another difference between this work and that of Yienger and Levy [1995] is the estimates for rain forest. Yienger and Levy [1995] assigned a flux of 8.6 ng m−2 s−1 for rain forest on dry soils and 2.6 ng m−2 s−1 for rain forest on wet soils, resulting in an annual emission of 3.4 Tg N (above soil). In our data set (including those data not listed in the Appendix owing to the lack of soil data or an insufficient measurement period), there are 25 measurements (excluding artificially irrigated sites, as reported by Weitz et al. [1998]) for rain forest from 12 studies. Fluxes range from <0.1 ng m−2 s−1 [Erickson et al., 2001] to 10.2 ng m−2 s−1 [Kaplan et al., 1988], with a mean value of 2.5 ng m−2 s−1 and a median value of 1.3 ng m−2 s−1. Our simulated annual emission for rain forest is 0.68 Tg N (above soil), which is equal to an average flux of 1.7 ng m−2 s−1, falling between the mean and median values of observations. Considering that the fluxes assigned by Yienger and Levy [1995] were averaged from only three studies owing to the lack of data at that time, we feel the current estimation is better justified.

[57] This work predicts more emissions from African savanna than Yienger and Levy [1995] predicted, probably because we distinguished savanna from grassland, while Yienger and Levy [1995] did not. Savanna showed higher emission potential than grassland (Table 3). Some forest soils emit high amounts of NOx owing to nitrogen fertilization or heavy nitrogen deposition. Because of a lack of information, nitrogen input to forests was ignored in this simulation, which may have resulted in an underestimation of emission from forest soils.

[58] By applying a statistical model to a data set of NO emission from fertilized fields, Bouwman et al. [2002] found that only nitrogen fertilization, SOC content and drainage status significantly affected NO emission, and accordingly calculated NO emission from global fertilized cropland and grassland to be 1.37 and 0.23 Tg N yr−1, respectively. Their statistical results that NO emission increases with nitrogen fertilization and SOC content are in agreement with ours. While no significant climatic effect was found, Bouwman et al. [2002] showed that well drained soils had a higher NO emission potential than poorly drained ones. We didn't include soil drainage as a factor in our model, but found that soils in dry climate have the highest NOx emission potential. Both results indicate the preference for dry condition. However, we found a preference for acidic environment while Bouwman et al. [2002] found no significant effect of soil pH. Our estimate for cropland and grassland are 1.56 and 1.06 Tg N yr−1, respectively. These values are not directly comparable to the estimates of Bouwman et al. [2002] which were only for fertilized fields. Globally, fertilized cropland was 69% of the total, and fertilized grassland was 8% of the total [Bouwman et al., 2002].

[59] Our estimated NOx emission from the global terrestrial ecosystem is close to the lower limits reported by previous studies; both the above-soil and above-canopy emissions are about one-third those estimated by Davidson and Kingerlee [1997]. However, if the scaling-up method is applied directly to the data in the Appendix for each land-cover type, a global total above-soil emission estimate of 20.5 Tg N yr−1 results. For example, the average above-soil NOx flux from croplands sites listed in the auxiliary material is 11.58 ng m−2 s−1. Using this value as the default flux and scaling it up to all croplands for 1 year results in an estimate of 5.3 Tg N yr−1, which is very close to the estimate for croplands by Davidson and Kingerlee [1997]. Our estimate for croplands is only 2.4 Tg N yr−1. The discrepancies are caused by global heterogeneities in nitrogen fertilization and environmental factors. This highlights the importance of using representative observed fluxes in the scaling-up method and indicates a source of uncertainty in large-scale modeling, namely the accuracy of input data set of influencing factors. For example, soil pH is critical in controlling NOx flux (Figure 1), and the effect of nitrogen fertilizer is dependent on soil properties. A small deviation from actual values in the input data set may result in great uncertainty in both the total amount of emission and its geographical distribution.

[60] Finally, another source of uncertainty is the robustness of the effect of influencing factor in the model. Because the number of observations for some subclasses is very limited, a few more observation data may cause big change in the final result. More field observations are needed, especially those from land cover types that are not represented in the auxiliary material, and those accompanied by detailed environmental information such as soil properties.

5. Conclusions

[61] We developed a statistical model to describe the influences of major environmental and management variables on soil NOx emissions. The results show that NOx emission is significantly affected by land-cover type, climate, soil pH, SOC content, and nitrogen fertilization. Combining factors included in the statistical model and empirical functions of soil temperature, moisture change-induced pulse emission and vegetation fire, we simulated global NOx emission from soils. The estimated annual emission is near the lower limits reported by previous studies. Estimated global monthly soil NOx emission data at 0.5° × 0.5° resolution for 2001 are available for download at

[62] This work demonstrates that large uncertainties in the estimation of global soil NOx emission can be explained, at least in part, by the variability in environmental variables such as soil pH and organic carbon content. This uncertainty highlights the importance of representative observations and the reliability of input data sets. Our data analysis also indicated the insufficiency of observations, especially those accompanied by detailed environmental information.