Rice cultivation is an important anthropogenic source of atmospheric nitrous oxide (N2O) and methane. We compiled and analyzed data on N2O emissions from rice fields (113 measurements from 17 sites) reported in peer-reviewed journals. Mean N2O emission ± standard deviation and mean fertilizer-induced emission factor during the rice-cropping season were, respectively, 341 ± 474 g N ha−1 season−1 and 0.22 ± 0.24% for fertilized fields continuously flooded, 993 ± 1075 g N ha−1 season−1 and 0.37 ± 0.35% for fertilized fields with midseason drainage, and 667 ± 885 g N ha−1 season−1 and 0.31 ± 0.31% for all water regimes. The estimated whole-year background emission was 1820 g N ha−1 yr−1. A large uncertainty remains, especially for background emission because of limited data availability. Although midseason drainage generally reduces CH4 and increases N2O emissions, it may be an effective option for mitigating the net global warming potential of rice fields.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Agricultural soil is a major source of nitrous oxide (N2O). N2O is a greenhouse gas, and it contributes to the destruction of stratospheric ozone. Early studies found N2O emission from paddy fields to be negligible [e.g., Smith et al., 1982]. However, recent studies suggest that rice cultivation is an important anthropogenic source of not only atmospheric methane (CH4) but also N2O [e.g., Cai et al., 1997].
 The Intergovernmental Panel on Climate Change (IPCC) developed Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories [IPCC, 1997] and Good Practice Guidance and Uncertainty Management in National Greenhouse Gas Inventories [IPCC, 2000] for calculating national inventories of greenhouse gases [IPCC, 1997, 2000]. The current IPCC guidelines use a default fertilizer-induced emission factor (EF) of 1.25% of net N input (based on the unvolatilized portion of the applied N) and a background emission rate for direct emission from agricultural soil of 1 kg N ha−1 yr−1 [IPCC, 1997]. In the guidelines, rice paddy fields were not distinguished from upland fields. However, Bouwman et al.  reported on the basis of data published before 1999 that mean N2O emission from rice paddy fields (0.7 kg N2O-N ha−1 yr−1) was lower than that from upland fields, including grasslands (1.1 to 2.9 kg N2O-N ha−1 yr−1). Yan et al.  reported on the basis of data published before 2000 that the EF for rice paddy fields, at 0.25% of total N input, was also lower than that for upland fields, and a background emission of 1.22 kg N2O-N ha−1 yr−1 for paddy fields. More measurements of N2O emission from rice fields have since been performed, although the number of field measurements is still relatively small compared with those of CH4 emission from rice paddy fields or N2O emission from upland fields. The guidelines are due to be updated in 2006, and a simple method for estimating N2O emission from rice fields is urgently needed. Therefore the aim of this study was to assess field measurements available to date to provide a quantitative basis for estimating national or regional N2O emission inventories. Available data on tested options for mitigating N2O emissions from paddy fields were also reviewed.
2. Materials and Methods
2.1. Data Collection
 We compiled measurement results of direct N2O emission from rice fields published in peer-reviewed journals before summer 2004; the initial data set comprised 147 field measurements of N2O collected during the rice-cropping or fallow season from 29 sites. Measurements from atypically managed fields [e.g., Xu et al., 2004], those collected over a period significantly shorter than the cropping season [e.g., Freney et al., 1981], and those in which the measurement period was not specified [e.g., Xing and Zhu, 1997] were excluded. Two data points that were statistical outliers were also excluded [Khalil et al., 1998; Zheng et al., 2000]. In the end, the data set used for our assessment comprised 113 measurements from 17 sites.
 For each series of measurements, documented information compiled included the N2O emission during the cropping season and/or the fallow period, the water regime used during the cropping season, the type and amount of organic amendment and nitrogen fertilization, and the location of the field. The fertilizer-induced N2O emission factor (EF), defined as the emission from fertilizer plots minus that of a zero-N control plot, was calculated whenever a zero-N control plot was included in a reported experiment. If the seasonal flux was not directly reported, we estimated it by integrating the average emission over the season's duration or from figures showing the seasonal change in N2O fluxes.
 The water regime during the rice-cropping season was classified as continuous flooding, midseason drainage, wet-season rain-fed, or unknown. The water regime of fields drained one or more times during the rice-cropping season was classified as midseason drainage, because among the compiled measurements only one experiment [Bronson et al., 1997a] involved a single midseason drainage, and because the number of times a field was drained was not clearly documented in some papers. The water regime of fields drained only at the end of the season, without midseason drainage, was classified as continuous flooding, because in many experiments the continuous flooding treatment also included end-season drainage and because farmers usually drain the rice fields before harvest. To our knowledge, N2O measurements for deep-water rice fields are not available.
 To provide a quantitative basis for estimating national and regional N2O emissions from rice paddy fields, the mean, standard deviation, and median of N2O emission, EF, and background emission for the cropping season were calculated for each water regime.
 Available data covering the whole fallow period for estimating fallow period N2O emission were very limited. Therefore, hourly emissions were first calculated for each measurement series, and then the mean hourly emission was calculated after excluding data collected during a measurement period shorter than 50 days, because emissions values collected over a short period were much higher than those collected over a longer period. Finally, emission during the fallow period was calculated by integrating the mean hourly data over the entire period.
 Comparisons of mean emissions and EF values among different water regimes were made with SPSS version 12 using one-way analysis of variance followed by a Tukey multiple comparison test (SPSS, Inc.). When the data did not fit a normal distribution but fit a lognormal distribution, they were log-transformed before the statistical test.
3. Results and Discussion
3.1. N2O Emissions From Fertilized Fields During the Rice-Cropping Season
 For each water regime, N2O emissions from paddy fields with chemical or organic fertilizer application during the rice-cropping season showed large variation (Tables 1a, 1b, 1c, and 1d, Figure 1). N2O emission from continuously flooded fields was generally lower than 1000 g N ha−1 season−1, except in one case [Xu et al., 2004], and low emissions were reported even with large N inputs (>300 kg N ha−1) [Chen et al., 1997]. In contrast, large N2O emission (>1000 g N ha−1 season−1) was often reported for fields with midseason drainage. N2O emissions from rain-fed fields were reported only by Abao et al. , and emissions differed among fertilizer treatments and also varied with year. The total N input was not clearly related to N2O emission for all water regimes or for continuous flooding. For fields drained midseason, a weak linear relationship was observed (adjusted r2 = 0.28, P < 0.01); that is, about 28% of the variability was explained by total N input. The amount of chemical N input was also not clearly related to N2O emission (data not shown). This result is not surprising because N2O emission is affected not only by N input and the water regime but also by many other factors such as fertilizer type, temperature, soil texture, and soil pH [e.g., Granli and Bockman, 1994; Bouwman et al., 2002], which could not be considered here because of limited data. Moreover, we classified all water regimes into one of three categories, but actual water management practices varied within each category, especially in the case of midseason drainage. Among fields with midseason drainage, the timing, number, and duration of drainage periods vary, and the timing and amount of rainfall greatly affect water management of even irrigated rice fields; furthermore, details of drainage practices were not documented in many papers. Food and Agriculture Organization/International Fertilizer Industry Association  also found no clear relationship between the amount of chemical N input and N2O emissions in upland fields worldwide.
Table 1a. N2O Emissions From Paddy Fields During the Cropping Seasona
 Although N2O emissions from paddy fields cannot be described as a simple function of N input (Figure 1), a simple method is required for estimating national or regional N2O emission inventories, especially for the Tier 1 methodology of the IPCC guidelines. Also, we expect N2O emission to increase with N input under comparable conditions [e.g., Cai et al., 1997]. Therefore we calculated the mean and median N2O emission and the fertilizer-induced emission factor (EF), which is defined as the emission from fertilized plots minus that of a zero-N control plot (Table 2). The difference between the mean and median shows the skewness of the data distribution. Mean N2O emission ± standard deviation and EF for continuously flooded fields with chemical or organic fertilizer application during the rice-cropping season were 341 ± 474 g N ha−1 season−1 and 0.22 ± 0.24%, respectively. Mean N2O emission and EF for fields with midseason drainage and chemical or organic fertilizer application were 993 ± 1075 g N ha−1 season−1 and 0.37 ± 0.35%, respectively. Mean N2O emission and EF for all water regimes were 667 ± 885 g N ha−1 season−1 and 0.31 ± 0.31%, respectively. Mean N2O emission from continuously flooded fields was significantly lower than that from fields with midseason drainage, although N2O emissions within each water regime showed large variation. However, no significant difference in EF between the continuous flooding and midseason drainage water regimes was observed. Note that fewer EF data were available, because we were able to calculate EF only when zero-N control measurements were available. The mean N2O emission from rain-fed fields, provided only by Abao et al. , was lower than but not significantly different from values for fields with other water regimes.
Table 2. N2O Emissions From Rice Paddy Fields With Chemical or Organic Fertilizer Applied During the Cropping Seasona
 Mean background N2O emission (emission from paddy fields without N fertilizer application) during the cropping season was not significantly different between continuous flooding and midseason drainage regimes (Table 3). Large variations were observed within each water regime even in background emissions, and relatively few experiments included zero-N control plots.
Table 3. N2O Emissions (in gN ha−1) From Rice Paddy Fields With No N Fertilizer Input During the Cropping Season
 Data for emission during the fallow period are listed in Table 4. After excluding measurements obtained over periods shorter than 50 days, which were much higher than those obtained over longer periods, we calculated a mean emission of 24 μg m−2 h−1. Note that only one value (16.3 μg m−2 h−1 [Chen et al., 1997]) for a zero-N control plot during cropping season was available for the fallow period among the compiled data. On the basis of the mean hourly emission, the total N2O emission during the fallow period was estimated to be 1495 g N ha−1 period−1, when the cropping season was assumed to be 110 days long. The estimated emission for the fallow period has large uncertainty because the number of measurements was limited, although fallow rice fields are considered to be an important source of N2O. It is not clear why measured N2O emissions were higher during short measurement periods [Bronson et al., 1997b]. Note that winter cropping and dry season cropping after the rice harvest were not considered here. The background emission for an entire year was estimated as 1820 g N ha−1 yr−1 from the sum of the mean background emission for all water regimes and the estimated fallow period emission.
Table 4. N2O Emissions From Paddy Fields During the Fallow Period
3.3. Options for Mitigation of N2O Emissions From Rice Fields During the Rice-Cropping Season
 Among tested options for mitigation of N2O emissions from rice fields during the growing season (Table 5), nitrification inhibitors (dicyandiamide and thiosulfate) and slow-release urea significantly (P < 0.05) reduced N2O emissions, although the data are few and the effectiveness of each mitigation option showed large variation. Neem-coated urea and nimin-coated urea, which are supposed to have nitrification inhibition properties and to be more locally available in India, did not significantly reduce N2O emissions [Majumdar et al., 2000].
Table 5. Available Data on Possible Mitigation Options
Fertilizer-induced N2O-N emission of the tested mitigation option plot compared with that of the conventional fertilizer plot.
Significantly different from conventional fertilizer plot at P < 0.05 by Duncan's multiple range test. Statistical test results are from the original papers.
Fertilizer-induced N2O emission could not be calculated because no zero-N control plot was available. Thus the percent of N2O-N emission (including background emission) from the tested mitigation option plot is compared with that from conventional fertilizer plot is shown.
3.4. Trade-Off of CH4 and N2O Emissions From Rice Paddy Fields
 Midseason drainage is considered to be an effective option for mitigating methane emissions from rice fields [e.g., Yagi et al., 1997]. A statistical analysis of a large field-measurement data set indicated that compared with continuous flooding, a single midseason aeration can reduce the average seasonal CH4 emission by 40%, and multiple aeration reduces it by 48% (X. Yan et al., Statistical analysis of the major variables controlling methane emission from rice fields, submitted to Global Change Biology, 2004) (hereinafter referred to as Yan et al., submitted manuscript, 2004). However, midseason drainage increases N2O emission by creating saturated or nearly saturated soil conditions, which promote N2O production [e.g., Zheng et al., 2000]. Cai et al.  reported that the global warming potential (GWP) of N2O emissions was even higher than that of CH4 emissions from Chinese paddy fields with midseason drainage when large amounts of chemical fertilizer (364.5 kg N ha−1) or farmyard manure (5 t ha−1) were applied. Bronson et al. [1997a] found that the total GWP of continuously flooded fields was lower than that of fields drained midseason when no straw was applied, but it was higher when straw was applied.
 When CH4 emissions estimated by a statistical model proposed by Yan et al. (submitted manuscript, 2004) and mean N2O emissions calculated here (Table 2) were compared between continuous flooding and midseason drainage water regimes with no organic fertilizer or straw application (Table 6), midseason drainage appears to be generally the more effective option for mitigating net GWP; however, 15% to 20% of the benefit gained by decreasing CH4 emission was offset by the increase in N2O emission. Yan et al. (submitted manuscript, 2004) estimated that application of rice straw at 6 t ha−1 (dry weight) before rice planting, which is roughly all the rice straw harvested, would increase CH4 emission 3.1-fold, compared with that from soils without any organic amendment, for both continuous flooding and midseason drainage water regimes. Therefore, midseason drainage may also be an effective mitigation option when straw is applied.
Table 6. Comparison of Estimated CH4 Emission Determined by the Model of Yan et al. (submitted manuscript, 2004) and the Mean N2O Emission Found by This Study for Continuous Flooding and Midseason Drainage
Reduced GHG by Drainage, CO2 Equivalent, kg ha−1 Season−1
The cropping season assumed to be 110 days.
Calculated by using the global warming potential (GWP) for a time horizon of 100 years: CH4 = 23 and N2O = 296 [IPCC, 2001].
CH4Emission, Estimated From Model of Yan et al. (Without Organic Fertilizer or Straw)
Mean N2O Emission by This Study (Without Organic Fertilizer but Including Plots With Straw Application)
Li et al.  also reported that midseason drainage reduces net GWP compared with continuous flooding; 65% of the benefit gained by decreasing CH4 emissions from rice fields in China was offset by an increase in N2O emissions, as determined by the denitrification-decomposition (DNDC) model. Frolking et al.  used the DNDC model and an atmospheric model to simulate the effect of a change in water management from continuous flooding to midseason drainage on N2O and CH4 emissions from rice fields and the relative radiative impact over 500 years. They found that, initially, a change in radiative forcing was dominantly the result of a decrease in CH4 emissions, but long-term radiative forcing was dominantly the result of the increase of N2O emissions; thus, an initial 36-year cooling effect was followed by a long-term warming effect (e.g., 100 years or longer). They also suggested that the overall complexity of the radiative forcing response to the change in water management could not be captured by conventional GWP calculations.
3.5. Seasonal Pattern of N2O Emissions From Rice Fields
 We tried to find a general seasonal pattern of N2O emissions from paddy fields during the rice-cropping period even though many different patterns were reported. We assessed figures showing seasonal changes in N2O flux for measurement series that include data on chemical or organic fertilizer input and measurements from zero-N control plots, but excluded mitigation option experiments.
 One or more N2O peaks after transplanting or sowing were observed in about 70% of 79 measurements. During the flooded period, N2O peaks were reported in about 35% of measurements. N2O peaks were also reported in about 50% of measurements associated with an end-season drainage period, which is common practice both for middle season drainage and continuous flooding regimes.
 In the case of fields with midseason drainage, N2O peaks were observed in 51 of 52 measurements during the midseason drainage period. N2O peaks have been generally observed after top dressing application [e.g., Cai et al., 1999], but some reports showed N2O peaks during the midseason drainage that were not related to the application of top dressing [e.g., Cai et al., 1997; Xiong et al., 2002]. Zheng et al.  reported that N2O emission peaked during the midseason drainage and suggested that soil moisture is the most sensitive factor regulating N2O emissions. They reported explosive emissions of N2O when soil moisture was near 110% of the soil water holding capacity or the water-filled pore space was 99%. It is notable that N2O peaks during the midseason drainage have been reported for both fertilized and unfertilized fields [e.g., Cai et al., 1997], but N2O peaks from unfertilized fields are generally smaller than those from fertilized fields. In contrast, no N2O peak was reported by Yagi et al.  during the midseason drainage in Japan, even after the application of top dressing.
 After the rice has been harvested, large N2O peaks are commonly observed; in many cases, these peaks are larger than those observed during the cropping season [e.g., Hou et al., 2000; Nishimura et al., 2004]. Most measurements, however, were terminated at harvest and thus did not capture N2O emissions afterward.
 Although water management has a large effect on N2O emission [e.g., Zheng et al., 2000], this pattern analysis can be expected to contain errors caused by the difficulty of determining the water regime, because details of water management (timing, number, and duration of drainage) were not described in many papers. In only seven papers among those reviewed were the seasonal changes in water depth for the entire cropping period reported [Yagi et al., 1996; Cai et al., 1997, 1999; Suratno et al., 1998; Zheng et al., 2000; Ghosh et al., 2003; Nishimura et al., 2004]. In addition, details of fertilizer management (timing and application method for both basal and supplemental fertilizer application, N content of organic fertilizer) or the actual cropping period (date of transplanting and harvest) were not described in some papers.
3.6. Other Factors Affecting N2O Emissions From Rice Fields
 N2O emissions from rice paddy fields are affected by many factors, including type of fertilizer, climate, and soil type. However, those factors were not considered here because of limited data availability. For example, fertilizer type varied little; most measurements were from fields to which urea had been applied as fertilizer. Interannual variation was also not considered here, even though large interannual variation in N2O emissions from rice fields has been reported [e.g., Khalil et al., 1998].
 This analysis is also affected by the water regime classification scheme. We classified the water regime into three categories, but actual water management is more varied, especially in the case of midseason drainage. Local practices regarding the timing, number, and duration of drainage periods vary by region, and the timing and amount of rainfall greatly affect water management even for irrigated rice fields. Moreover, the details of drainage were not documented in many papers.
 Some of the data were obtained by automated N2O flux monitoring [Bronson et al., 1997a, 1997b; Abao et al., 2000; Zheng et al., 2000; Nishimura et al., 2004]. Those results showed sharp peaks in N2O emissions from paddy fields, especially during periods of intermittent drainage. Most of the available data, however, were obtained manually once or twice a week by the static chamber method, so sharp peaks might have been missed, causing seasonal N2O emissions to be underestimated. Also, only two papers reported measurements covering the entire year [Zheng et al., 2000; Nishimura et al., 2004], although many data were available on N2O emissions during the rice-cropping period.
 We reviewed published reports of N2O emissions from rice paddy fields and tried to establish a quantitative basis on which to develop national or regional emission inventories. We propose a value of 0.31 ± 0.31% for the fertilizer-induced emission factor during the rice-cropping period and a value of 1820 g N ha−1 yr−1 for background emission for an entire year. The estimated background emission value has large uncertainty, because available measurements were very limited, especially for the fallow period, even though fallow rice fields are considered an important source of N2O.
 Midseason drainage is an effective option for mitigating methane emissions and net GWP from rice fields, although mean N2O emission is increased by midseason drainage compared with continuous flooding. Available data on tested options for the mitigation of emissions from rice fields during the growing season were also assessed. Although few data are available, we found that application of a nitrification inhibitor or slow-release urea significantly reduced N2O emissions. More field measurements, especially for the entire year, including the fallow period, and experiments that include a zero-N control are required for further analyses.
 This work was supported by the Global Environment Research Fund, Ministry of the Environment, Japan.