Cite this Article
ZHANG Wei, GU Jiang-Xin, ZHENG Xun-Hua. Direct Nitrous Oxide Emissions Related to Fertilizer-Nitrogen, Precipitation, and Soil Clay Fraction: Empirical Models. Atmospheric and Oceanic Science Letters, 2015, 8(5): 277-282  
Permissions
Copyright?2015, Editorial office of Atmospheric and Oceanic Science Letters
This is an Open Access article under the terms of CCAL.
Direct Nitrous Oxide Emissions Related to Fertilizer-Nitrogen, Precipitation, and Soil Clay Fraction: Empirical Models
ZHANG Wei1,2, GU Jiang-Xin1,3, ZHENG Xun-Hua1,*
1 State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry (LAPC), Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China
2 University of Chinese Academy of Sciences, Beijing 100049, China
3 College of Natural Resources and Environment, Northwest A & F University, Yangling 712100, China
*Corresponding author: ZHENG Xun-Hua, xunhua.zheng@post.iap.ac.cn
Abstract

Direct nitrous oxide (N2O) emissions (DNEs) from croplands are required in national inventories of greenhouse gases. The Intergovernmental Panel on Climate Change (IPCC) guidelines provide an approach using direct emission factors (EFds) to estimate DNEs, which are constants for large regions. The goal of this paper is to establish empirical models to account for the temporal and spatial variations of EFds, which, apart from the nitrogen addition rate, also vary with a range of environmental factors, so as to enhance the accuracy of regional/national DNE estimates. Therefore, the seasonal/annual DNEs ( n = 71) from upland croplands, which are the differences in N2O emissions between fields with and without fertilizer-nitrogen addition, were used to statistically relate DNEs to regulating factors including the fertilizer-nitrogen addition rate ( FN), and environmental (climate and soil) factors. The multivariate stepwise linear regression results showed positive combined effects of FN and clay fraction on DNEs ( R2 = 0.61, p < 0.001). Furthermore, the nonlinear regression of FN, precipitation, and clay fraction was also adopted for prediction ( R2 = 0.50, p < 0.001). Validation with an independent dataset ( n = 31) suggested that both models were better predictors of DNEs than the IPCC model, which only depends on FN. These empirical models may provide simple but reliable approaches for compiling regional/national, and even global inventories of DNEs from croplands. However, both models were restricted to a limited sample size. Understandably, more field observations are still required to further validate the global applicability of these simple approaches.

Keyword: nitrous oxide; direct emissions; empirical model; nitrogen addition; precipitation; clay fraction
1 Introduction

The Kyoto Protocol requires countries to provide national inventories of greenhouse gases, including nitrous oxide (N2O). During recent decades, great efforts have been made to accurately estimate N2O emissions from croplands, which serve as significant anthropogenic sources of N2O. Bouwman (1996) proposed the concept of direct emission factors (EFds) to relate the fertilizer-nitrogen addition rate to cumulative N2O emissions, which has been adopted by the Intergovernmental Panel on Climate Change (IPCC) to compile national inventories of N2O emissions from croplands (IPCC, 2006). The EFd is defined as the fraction of fertilizer-nitrogen released in the form of N2O within the current season or year. Therefore, the product of EFds and anthropogenic nitrogen addition is the direct N2O emission (DNE) amount, which is the major contributor to N2O emissions from croplands, e.g., 70% in China (Yan et al., 2003; Zheng et al., 2004; Gu et al., 2009). The national inventory in China includes the emissions of N2O from both upland and paddy soils using the default EFds of 1% and 0.3%, respectively (Yan et al., 2003; Lu et al., 2006; Zou et al., 2009). However, these constant EFds that are applied ignore the heterogeneity of soil and climate, which, apart from the nitrogen addition rate, are also key factors regulating N2O emissions. Furthermore, the default EFds possess large uncertainties, ranging from 0.3% to 3% for upland soils and 0% to 0.6% for paddy soils. These deficiencies have a substantial effect on the accuracy of DNE estimates. Many previous studies have focused on predicting DNEs from paddy soils during the rice growing season (e.g., Akiyama et al., 2005; Zou et al., 2009). In this study, we concentrated on DNEs from upland soils.

Recent studies show that nitrogen addition is not the only key factor regulating N2O emissions from upland croplands at regional/national scales. Some other empirical models have considered environmental factors as well as the nitrogen addition rate (Bouwman et al., 2002; Sozanska et al., 2002; Roelandt et al., 2005; Stehfest and Bouwman, 2006), such as soil physical and chemical factors (e.g., soil moisture, temperature, mineral nitrogen, and labile carbon) and climate factors (e.g., precipitation and air temperature). However, most environmental factors are influenced by human activities and climate, which, due to their spatiotemporal variation, are difficult to measure at regional/national scales and therefore have to be obtained from the output of other models (e.g., Sozanska et al., 2002). These indirect inputs also result in uncertainties of N2O emission estimates. Moreover, these environmental factors are further regulated by other variables, which are easier to obtain at regional/national scales. For example, soil moisture and mineral nitrogen content are controlled by precipitation and soil organic matter. Lu et al. (2006) modified the IPCC approach for upland soils with seasonal/annual precipitation, which is commonly available at regional/national scales.

The objective of this study was to establish empirical models to estimate DNEs with commonly available input data accounting for climate and/or soil variations at regional/national scales. Observed seasonal/annual N2O emissions were used to identify statistical relationships between DNEs and potential regulating factors, i.e., fertilizer-nitrogen addition rate, climate, and soil properties.

2 Materials and methods
2.1 N2O emissions database

The DNE amount is defined as the difference in N2O emissions between fields with and without fertilizer-nitrogen addition (e.g., Zheng et al., 2004; Stehfest and Bouwman, 2006).

A database was established with seasonal or annual N2O emissions from upland croplands, which was extracted from Jungkunst et al. (2006) and Stehfest and Bouwman (2006, http://www.mnp.nl/images/stehfest_data_ tcm61-29733.xls), as well as other publications (Table 1). The data were selected with strict criteria: (1) The study area was limited to the Northern Hemisphere; (2) Each site required a treatment without fertilizer-nitrogen addition; (3) Cases in cultivated organic soils were eliminated; (4) Observations in legume or vegetable fields were excluded. The database consisted of 155 cases of observed N2O emissions across 19 sites, of which 102 cases were obtained from nitrogen fertilized fields. Most of the data were from Asia and Europe during 1992-2012, including crops of wheat, barley, maize, cotton, rapeseed, and sugar beet. The topsoil soil organic carbon (SOC) and total nitrogen (TN) content, clay fraction, and soil pH of the involved treatments ranged from 1.5-30.9 g C kg-1, 0.1-3.3 g N kg-1, 0.04-0.50, and 5.2-8.7, respectively. There were three sites in France where the DNE observations covered the rapeseed growing season in spring and summer (five to six months). The other DNE observations lasted for more than one year (Table 1).

Climate (i.e., seasonal/annual precipitation, mean air temperature) and soil (i.e., clay fraction, SOC, TN, and soil pH) factors, which in most studies are usually provided as background information, were used to identify the major driving factors of N2O emissions from croplands. In order to calculate the seasonal/annual precipitation and mean air temperature, the climate data were obtained from the National Oceanic and Atmospheric Administration (NOAA), USA, for each site (ftp://ftp.ncdc. noaa.gov/pub/data/gsod/). Detailed information on the observed data, climate, and soil factors are listed in Table 1.

2.2 Statistical analysis

The SPSS Statistics Client 16.0 (SPSS Inc., Chicago, USA) software packages were used for the statistical data analysis.

A random sampling technique was applied to establish independent calibration and validation datasets. After the rearrangement of 102 DNEs based on the generated random numbers, the top 70% were selected to be the calibration dataset (n = 71) and the others were distributed to the validation dataset (n = 31). The DNE values from both the calibration (p = 0.074) and validation (p = 0.408) datasets passed the Kolmorov-Smirnov normality test.

The factors selected for multivariate stepwise linear regression of the calibration dataset included the fertilizer-nitrogen addition rate, mean air temperature, cumulative precipitation, topsoil SOC and TN content, clay fraction, and soil pH, during the observation period. Considering the interaction among the factors, nonlinear regression was also adopted. The criteria of the determination coefficient (R2), colinearities, and the distribution of residuals, were used to select the model. Colinearity was tested using the variance inflation factor (VIF), which has to be lower than 10 (independent among variables).

The established models were validated with the independent validation dataset. The validation results were quantitatively evaluated using the statistical criteria of the root mean square error (RMSE), Nash-Sutcliffe Index (NSI), the slope and determination coefficient (R2) of zero-intercept linear regression.

3 Results
3.1 DNEs and EFds from upland croplands

The DNEs from upland croplands varied inter-annually and spatially (Table 1). The DNEs ranged from -1 to 5.8 kg N ha-1, with an average value of 1.49 kg N ha-1 (n = 102). Approximately 6% of the DNEs were below 0 kg N ha-1, which indicated the added fertilizer-nitrogen did not stimulate N2O emission. For all cases, the DNEs were positively related with the fertilizer-nitrogen addition (p < 0.001, Fig. 1a), which was in accordance with the IPCC model. The EFds ranged from -1.18% to 3.01%, with a mean value of 0.65% (n = 102), which was lower than the IPCC default value of 1%. The results showed that nitrogen addition accounted for 42% of the temporal and spatial variations in DNEs (Eq. (1) in Table 2), which indicated that the DNE amount might be intensively regulated by other environmental factors.

Table 1 Description of the data used in this study (n, number of fertilizer-nitrogen addition treatments, including no-nitrogen control; P, seasonal/annual precipitation; T, mean air temperature; Clay, topsoil clay content; FN, fertilizer-nitrogen addition rate).

Figure 1 (a) Relationship between the direct N2O emission (DNE) amount and fertilizer-nitrogen addition rate (FN); (b, c) combined effects of FN, clay fraction (Clay), and/or precipitation (P) on DNEs from upland croplands as showed in Eq. (2) and Eq. (4) for both calibration and validation datasets; (d) comparison between observations and predictions by the IPCC model, Eq. (2) and Eq. (4) in the validation dataset. Regression functions presented by the lines are shown in Table 2.

3.2 Calibration and validation of the empirical models

According to the stepwise linear regression results provided by SPSS, nearly 61% of the temporal and spatial variations in DNEs could be explained by the combined effects of the fertilizer-nitrogen addition rate and clay fraction (Fig. 1b, Eq. (2) in Table 2), with significant slope coefficients and intercept (p < 0.001). No colinearity existed between nitrogen addition and clay fraction (VIF = 1.1). For the nonlinear regression of two factors, a significant regression was found when the combined effects of the fertilizer-nitrogen addition rate and precipitation were considered (R2 = 0.45, Eq. (3) in Table 2), whose evaluation criteria were better than the other combinations. If the clay fraction was introduced into the above nonlinear regression, the combined effects of the fertilizer-nitrogen addition rate, precipitation and clay fraction explained more of the variance in the DNEs (R2 = 0.50, Fig. 1c, Eq. (4) in Table 2), with both a significant slope coefficient (p < 0.001) and intercept (p = 0.003). The residues of the three regression models followed the normal distribution.

Table 2 Regression equations (DNE, direct N2O emission (units: kg N ha-1); FN, fertilizer-nitrogen addition rate (units: kg N ha-1); Clay, clay fraction; P, cumulative precipitation (units: mm); n, sample size; R2, determination coefficient). Data in brackets are the estimation uncertainties of the parameter at the 95% confidence level. All of the slopes and intercepts involved in the equations are statistically significant (p < 0.01).

Based on the statistical results of different regressions, two established models (Eqs. (2) and (4)) were preferred. An independent validation dataset (n = 31) other than the calibration one was used to test the performance of the new models and IPCC model (Fig. 1d). Equations (2) and (4) predicted the DNEs well, with RMSE/NSI of 0.45/ 0.54 and 0.44/0.57, respectively. For the zero-intercept linear regressions between observations and predictions, the slope/R2 values were 0.96/0.63 (p < 0.001) and 0.87/0.59 (p < 0.001) for the two models, respectively. In comparison, the two established models performed generally better than the IPCC model (RMSE = 0.98, NSI = -1.15, slope = 1.4, R2 = 0.39).

4 Discussion

The validation of the IPCC model for upland soils showed unsatisfactory prediction, with worse evaluations than the two established models. These models could be further used to predict DNEs from upland croplands at regional/national scales in the Northern Hemisphere.

Using the stepwise linear regression method, the nitrogen addition rate and clay fraction were identified as two key factors controlling the DNEs. The results indicated that only 0.5% (0.3%-0.7%) of added fertilizer-nitrogen were released in the form of N2O, which was much lower than the EFds (1%) of the IPCC model. The positive effect of clay fraction on DNEs was consistent with previous studies (e.g., Bouwman et al., 2002), which affects the oxygen and moisture status and gas diffusion in cropland soils. Fine-texture soils with high clay content have more capillary pores within aggregates. Thus, anaerobic conditions may be more easily reached and maintained for longer periods within aggregates, which favors the high N2O emissions. The statistical evaluation of Eq. (2) indicated that the effects of other potential regulating factors should be analyzed. However, this analysis was hindered by the possibility of obtaining the factors in the field. As both the nitrogen addition rate and clay fraction have maintained at relatively stable values in a specific region for a few years, the empirical model of Eq. (2) could not reflect the inter-annual variations of DNEs.

Lu et al. (2006) proposed that the combined effects of the nitrogen addition rate and precipitation (FN × P) on DNEs enabled better predictions than using the nitrogen addition rate only, which were also available in this study. However, when considering the effect of clay fraction (Clay) simultaneously, the model (FN × P × Clay) performance was improved. Due to the introduction of seasonal/annual precipitation into Eq. (4), which is sensitive to temporal and spatial variabilities, the models could predict the inter-annual and spatial variations of DNEs (Roelandt et al., 2005; Lu et al., 2006). The maximum precipitation in the validation dataset appeared in the maize-winter wheat field of Northeast China in 1995 (Dong et al., 2001) and yielded a higher DNE amount, which could be predicted well by the model (2.84 kg N ha-1 for observation vs. 2.95 kg N ha-1 for Eq. (4)). In addition, the DNEs from some sites were affected by the irrigation practice, such as Yuncheng and Huantai in China, Timmerla in Germany, and Longchamp in France. For upland soils, irrigation is a common management practice that might greatly affect N2O emissions (e.g., Scheer et al., 2008). However, most of the observations involved in this study did not provide detailed information about irrigation and thus was ignored in the database. By considering the combined effects of precipitation and irrigation on DNEs in Eq. (4), the inter-annual and spatial variations would be better predicted.

5 Conclusion

Two empirical models for predicting DNEs from upland croplands were established in this study. The nitrogen addition rate and clay fraction were included in the first model, which showed positive combined effects on DNEs. The other showed satisfactory prediction in consideration of the combined effects of the nitrogen addition rate, precipitation, and clay fraction. Validation results indicated that the new models performed better than the IPCC model. The input data required by the two empirical models are easy to obtain at regional/national scales, which provides a methodology for improved DNE estimates used for compiling inventories. However, due to the limited datasets for calibration and validation, more field observations are still required to further validate the global applicability of these simple approaches.

Reference
1 Akiyama H. , K. Yagi, and X. Yan, 2005: Direct N2O emissions from rice paddy fields: Summary of available data, Glob. Biogeochem. Cycles, 19, doi: DOI:10.1029/2004GB002378.
2 Bouwman A. F. , 1996: Direct emissions of nitrous oxide from agricultural soils, Nutr. Cycl. Agroeco. , 46, 53-70.
3 Bouwman A. F. , L. J. M. Boumans, and N. H. Batjes, 2002: Emissions of N2O and NO from fertilized fields: Summary of available measurement data, Glob. Biogeochem. Cycles, 16, doi: DOI:10.1029/2001GB001811.
4 Cai Y. J. , W. X. Ding, and J. F. Luo, 2013: Nitrous oxide emissions from Chinese maize-wheat rotation systems: A 3-year field measurement, Atmos. Environ. , 65, 112-122.
5 Chen Z. M. , W. X. Ding, Y. Q. Luo, et al. , 2014: Nitrous oxide emissions from cultivated black soil: A case study in Northeast China and global estimates using empirical model, Glob. Biogeochem. Cycles, 27, 1311-1326.
6 Ding W. X. , Y. Cai, Z. C. Cai, et al. , 2007: Nitrous oxide emissions from an intensively cultivated maize-wheat rotation soil in the North China Plain, Sci. Tot. Environ. , 373, 501-511.
7 Dong Y. S. , D. Scharffe, Y. C. Qi, et al. , 2001: Nitrous oxide emissions from cultivated soils in the North China Plain, Tellus, 53B, 1-9.
8 Flessa H. , R. Ruser, P. Dorsch, et al. , 2002: Integrated evaluation of greenhouse gas emissions (CO2, CH4, N2O) from two farming systems in southern Germany, Agric. Ecosyst. Environ. , 91, 175-189.
9 Gu J. X. , X. H. Zheng, and W. Zhang, 2009: Background nitrous oxide emissions from cropland in China in the year 2000, Plant Soil, 320, 307-320.
10 Henault C. , X. Devis, J. L. Lucas, et al. , 1998 a: Influence of different agricultural practices (type of crop, form of N-fertilizer) on soil nitrous oxide emissions, Biol. Fertil. Soil. , 27, 299-306.
11 Henault C. , X. Devis, S. Page, et al. , 1998 b: Nitrous oxide emissions under different soils and land management conditions, Biol. Fertil. Soil. , 26, 199-207.
12 Huang G. H. , G. X. Chen, Z. M. Zhang, et al. , 1998: N2O emission in maize fields and its mitigation, Acta Sci. Circumst. (in Chinese), 18, 344-349.
13 IPCC, 2006: 2006 IPCC Guidelines for National Greenhouse Gas Inventories, H. S. Eggleston et al. (Eds. ), IGES, Hayama, available athttp://www.ipcc-nggip.iges.or.jp, 989pp.
14 Jungkunst H. F. , A. Freibaue, H. Neufeldt, et al. , 2006: Nitrous oxide emissions from agricultural land use in Germany—a synthesis of available annual field data, J. Plant Nutri. Soil Sci. , 169, 341-351.
15 Kaiser E. A. , K. Kohrs, M. Kücke, et al. , 1998: Nitrous oxide release from arable soil: Importance of N-fertilization, crops and temporal variation, Soil Biol. Biochem. , 30, 1553-1563.
16 Liu C. Y. , K. Wang, and X. H. Zheng, 2012: Responses of N2O and CH4 fluxes to fertilizer nitrogen addition rates in an irrigated wheat-maize cropping system in northern China, Biogeosciences, 9, 839-850.
17 Liu C. Y. , Z. S. Yao, K. Wang, et al. , 2014: Three-year measurements of nitrous oxide emissions from cotton and wheat-maize rotational cropping systems, Atmos. Environ. , 96, 201-208.
18 Lu Y. Y. , Y. Huang, J. W. Zou, et al. , 2006: An inventory of N2O emissions from agriculture in China using precipitation-rectified emission factor and background emission, Chemosphere, 65, 1915-1924.
19 Meng L. , W. X. Ding, and Z. C. Cai, 2005: Long-term application of organic manure and nitrogen fertilizer on N2O emissions, soil quality and crop production in a sand y loam soil, Soil Biol. Biochem. , 37, 2037-2045.
20 Misselbrook T. H. , L. M. Cardenas, V. Camp, et al. , 2014: An assessment of nitrification inhibitors to reduce nitrous oxide emissions from UK agriculture, Environ. Res. Lett. , 9, doi: DOI:10.1088/1748-9326/9/11/115006.
21 Nadeem S. , T. Borresen, and P. Dörsch, 2015: Effect of fertilization rate and ploughing time on nitrous oxide emissions in a long-term cereal trail in south east Norway, Biol. Fertil. Soil. , 51, 353-365.
22 Roeland t C. , B. Van Wesemael, and M. Rounsevell, 2005: Estimating annual N2O emissions from agricultural soils in temperate climate, Glob. Change Biol. , 11, 1701-1711.
23 Röver M. , O. Heinemeyer, and E. A. Kaiser, 1998: Microbial induced nitrous oxide emissions from an arable soil during winter, Soil Biol. Biochem. , 30, 1859-1865.
24 Scheer C. , R. Wassmann, K. Kienzler, et al. , 2008: Methane and nitrous oxide fluxes in annual and perennial land -use systems of the irrigated areas in the Aral Sea Basin, Glob. Change Biol. , 14, 2454-2468.
25 Shi Y. F. , W. L. Wen, F. Q. Meng, et al. , 2013: Integrated management practices significantly affect N2O emissions and wheat-maize production at field scale in the North China Plain, Nutr. Cycl. Agroecosyst. , 95, 203-218.
26 Sozanska M. , U. Skiba, and S. Metcalfe, 2002: Developing an inventory of N2O emissions from British soils, Atmos. Environ. , 36, 987-998.
27 Stehfest E. , L. Bouwman, 2006: N2O and NO emission from agricultural fields and soils under natural vegetation: Summarizing available measurement data and modeling of global annual emissions, Nutr. Cycl. Agroecosyst. , 74, 207-228.
28 Teepe R. , R. Brumme, and F. Beese, 2000: Nitrous oxide emissions from frozen soils under agricultural, fallow and forest land , Soil Biol. Biochem. , 32, 1807-1810.
29 Yan G. X. , X. H. Zheng, F. Cui, et al. , 2013: Two-year simultaneous records of N2O and NO fluxes from a farmed cropland in the northern China plain with a reduced nitrogen addition rate by one-third, Agric. Ecosyst. Environ. , 178, 39-50.
30 Yan X. Y. , H. Akimoto, and T. Ohara, 2003: Estimation of nitrous oxide, nitric oxide and ammonia emissions from cropland s in East, Southeast and South Asia, Glob. Change Biol. , 9, 1080-1096.
31 Zeng J. H. , Z. P. Wang, Y. M. Zhang, et al. , 1995: N2O flux from soil during wheat-maize rotation and estimation of total emission, Chin. J. Environ. Sci. (in Chinese), 16, 32-36.
32 Zhang Y. , Y. Mu, Y. Zhou, et al. , 2014: Nitrous oxide emissions from maize-wheat field during 4 successive years in the North China Plain, Biogeosciences, 11, 1717-1726.
33 Zheng X. H. , S. H. Han, Y. Huang, et al. , 2004: Re-quantifying the emission factors based on field measurements and estimating the direct N2O emission from Chinese cropland , Glob. Biogeochem. Cycles, 18, doi: DOI:10.1029/2003GB002167.
34 Zhou M. H. , B. Zhu, B. -B. Klaus, et al. , 2013: Nitrous oxide emissions and nitrate leaching from a rain-fed wheat-maize rotation in the Sichuan Basin, China, Plant Soil, 362, 149-159.
35 Zhou M. H. , B. Zhu, N. Bruggemann, et al. , 2014: N2O and CH4 emissions, and NO3ˉ leaching on a crop-yield basis from a subtropical rain-fed wheat-maize rotation in response to different types of nitrogen fertilizer, Ecosystems, 17, 286-301.
36 Zou J. W. , Y. Huang, Y. M. Qin, et al. , 2009: Changes in fertilizer-induced direct N2O emissions from paddy fields during rice-growing season in China between 1950s and 1990s, Glob. Change Biol. , 15, 229-242.