Estimations of emission factors for fertilizer-induced direct N₂O emissions from agricultural soils in Japan: Summary of available data

Abstract

Agricultural fields are significant sources of anthropogenic atmospheric nitrous oxide (N₂O). We compiled and analyzed data on N₂O emissions from Japanese agricultural fields (246 measurements from 36 sites) reported in peer-reviewed journals and research reports. Agricultural fields were classified into three categories: upland fields, tea fields and rice paddy fields. In this analysis, data measured over a period of more than 90 days for upland fields and 209 days for tea fields were used to estimate annual fertilizer-induced emission factors (EF) because of limitations in the available data. The EF is defined as the emission from fertilized plots minus the background emission (emission from a zero-N control plot), and is expressed as a percentage of the N applied. The mean of N₂O emissions from upland fields with well-drained soils was significantly lower than that from poorly drained soils. Mean (± standard deviation) N₂O emissions measured over a period of more than 90 days from fertilized upland fields were 1.03 ± 1.14 kg N ha⁻¹ and 4.78 ± 5.36 kg N ha⁻¹ for well-drained and poorly drained soils, respectively. Because the ratio of the total areas of well-drained soils and poorly drained soils was different from the ratio of the number of available EF data for each soil category, we used a weighted mean to estimate EF for all upland fields. The EF was estimated to be 0.62 ± 0.48% for all fertilized upland fields. Mean N₂O emissions and the estimated EF for fertilized tea fields measured over a period of more than 209 days were 24.3 ± 16.3 kg N ha⁻¹ and 2.82 ± 1.80%, respectively. The mean N₂O emission and estimated EF from Japanese rice paddy fields were 0.36 kg N ha⁻¹ and 0.31 ± 0.31% for the cropping season, respectively. Significant uncertainties remain in these results because of limitations in the available data.

Key words:

• background emission
• fertilizer-induced emission factor
• rice paddy field
• tea field
• upland field
• soil drainage

INTRODUCTION

Agricultural soil is a major source of nitrous oxide (N₂O), which is a greenhouse gas and also contributes to the destruction of stratospheric ozone.

The Intergovernmental Panel on Climate Change (IPCC) developed guidelines for calculating national inventories of greenhouse gases (Intergovernmental Panel on Climate Change 1997, 2000), which are due to be revised in 2006. The IPCC provided default emission factors for a “Tier 1” method for use when insufficient data are available for a country. The 1996 and 2000 IPCC guidelines (Intergovernmental Panel on Climate Change 1997, 2000) specify a default fertilizer-induced N₂O emission factor (EF) of 1.25% of net N input (based on the unvolatilized portion of the applied N) for direct emission from agricultural soil, and a background N₂O emission of 1 kg N ha⁻¹ year⁻¹ (Intergovernmental Panel on Climate Change 1997), where EF is defined as the emission from fertilized plots minus the background emission (emission from a zero-N control plot), and is expressed as a percentage of the N applied. The IPCC default values in the 1996 and 2000 guidelines for direct N₂O emission and background emission are based on the study by Bouwman (1996) that analyzed field data of at least 1-year duration and found a liner relationship between N input and N₂O emission from upland fields and grasslands in Europe and the USA. The IPCC recommends that country-specific EFs be used when reliable data are available (“Tier 2” method).

The methodology used to estimate direct N₂O emissions from agricultural soil in the USA is based on a combination of Tier 1 and Tier 3 approaches (US Environmental Protection Agency 2006). Specifically, a Tier 3, process-based model (DAYCENT) is used to estimate direct emissions from major crops on mineral (i.e. non-organic) soils; as well as most of the direct emissions from grasslands. The Tier 1 IPCC methodology is used to estimate direct emissions from non-major crops on mineral soils; the portion of the grassland direct emissions from paddocks and forage legume N additions that was not estimated with the Tier 3 DAYCENT model; and direct emissions from drainage and cultivation of organic cropland soils. The methodologies used to estimate direct N₂O emissions from agricultural soil in European countries are Tier 1 (e.g. Austria, Portugal and UK) and Tier 2 (e.g. Sweden and The Netherlands) (European Environment Agency, 2005). The country-specific EF for Sweden is 1–1.5% and the value varies because of changes in the quantities of different types of fertilizers sold (Swedish Environmental Protection Agency 2006). The country-specific EFs for The Netherlands are from 1% for chemical fertilizer and 2% for manure (Netherlands Environmental Assessment Agency 2006). The methodology used to estimate direct N₂O emission from agricultural soil in Australia is Tier 2, and country-specific EFs are estimated for seven different land-use types and vary from 0.3% for non-irrigated crops in dryland areas to 2.1% for horticulture/vegetable crops (Australian Greenhouse Office 2006).

The Greenhouse Gas Inventory Office of Japan (2005) reports country-specific EFs for direct emission from agricultural soils based on a study by Tsuruta (2001). Although those EFs were based on an extensive measurement campaign of N₂O emissions from Japanese agricultural fields conducted from 1992 to 1994 (Japan Soil Association 1996), two issues remain from the report by Tsuruta (2001): (1) because of a lack of data in the measurement campaign, background emission could not be estimated and, thus, was included in the calculated EFs, despite the fact that EF is defined as the emission from fertilized plots minus the background emission and is expressed as a percentage of the N applied, (2) the measurement periods were too short to accurately estimate annual emission (3 months in many cases, and less than 2 months in some cases).

More measurements of N₂O emissions from agricultural fields have since been carried out in Japan, although there are still insufficient field measurements to cover the wide variety of Japanese agriculture. The aim of our study was to assess the currently available field measurements to provide a quantitative basis for estimating Japanese N₂O emission inventories and to address some of the issues in the current estimates.

MATERIALS AND METHODS

Data collection

We compiled measurements of direct N₂O emission from agricultural fields published in peer-reviewed journals and research reports before 2005; the initial dataset comprised 252 field measurements of N₂O emissions from 38 sites. Measurements from atypically managed fields (Taki 1996) or from those collected over a period significantly shorter than the cropping season (e.g. Kim et al. 2002) were excluded. The final dataset used for our assessment comprised 246 measurements from 36 sites.

For each series of measurements, documented information included the N₂O emission, the type and amount of chemical and organic fertilizer applied, crop type, soil type, measurement period, and the location of the field. We calculated the fertilizer-induced N₂O EF, as defined above, only for studies that used a zero-N control plot. If the seasonal flux of emissions was not reported, we estimated it by integrating the average emission over the measurement period, or from figures showing the seasonal change in N₂O fluxes in the reports.

When the amount of organic material input was reported without providing the N content (percent N, fresh weight), we used the following values: swine manure, 1.35%; cattle manure, 0.71%; rice straw compost, 0.417%; rice straw, 0.54%; wheat straw, 0.33%; vetch, 0.59%; and cowpea residue, 0.47% (National Federation of Agricultural Co-operative Associations 1980; Ministry of Agriculture, Forestry and Fishery 1982). When it is stated as “manure” only, we assumed cattle manure.

Measurements with slow-release fertilizer and nitrification inhibitor are not included in this study because those fertilizer treatments are considered to be possible mitigation options for N₂O emissions from agricultural fields (e.g. Intergovernmental Panel on Climate Change 2001).

Data analysis

To provide a quantitative basis for estimating national and regional N₂O emissions from agricultural fields, we calculated the mean, standard deviation and median of N₂O emissions, EFs and background emissions.

Comparisons of mean emissions and EF values were made using SPSS v.12 software (SPSS, Japan inc, Tokyo, Japan). When the data did not fit a normal distribution, but did fit a log-normal distribution, they were log-transformed before the statistical test.

The standard deviations for the area-weighted mean EF for the fertilized upland fields and background emission for upland fields were calculated according to the National Greenhouse Gas Inventory Report of Japan (Greenhouse Gas Inventory Office of Japan 2005). The standard deviation of the uncertainty in the activity data (soil type distribution data from Soil Survey Committee of Japan 1991) was assumed to be 5%.

RESULTS AND DISCUSSION

Categorization of crop type

The 1996 and 2000 IPCC guidelines provide only one default EF to be used for all crop types when there are insufficient data to determine country-specific EF values. The current National Greenhouse Gas Inventory Report of Japan (Greenhouse Gas Inventory Office of Japan 2005) used country-specific EFs for 13 crop types (e.g. vegetables, 0.77%; fruits, 0.69%; tea, 4.74%; potato, 2.01%) based on the study by Tsuruta (2001). Because of the number of categories, some of these EFs were derived from a limited amount of field data. However, excessive differentiation of crop types would produce biased results because an EF based on limited field data would be affected by extreme values. In addition, country-specific EFs used in current estimates of the Japanese greenhouse gas inventory do not consider background emissions. When background emissions are considered in the estimation of EFs, the amount of available data in each crop category is even smaller.

A review by Akiyama et al. (2005) showed that the worldwide mean EF of rice paddy fields is lower than the IPCC default EF, which is based on upland fields. Tsuruta (2001) reported that the EF of tea fields is considerably higher than that of other upland fields. Therefore, the crop types in this study were divided into three categories: rice paddy fields, tea fields and upland fields. Tea fields were excluded from the upland fields category in the present study. Despite the coarse crop categorization used in our study, the lack of zero-N control data limited the amount of data available to calculate EFs for each crop category. Emission factors were calculated for only 24 upland fields, and for none of the tea or rice fields. N₂O emissions from leguminous crops and grasslands were grouped with upland fields in the present study because of a lack of Japanese data for these crop categories, that is, only two EF data from one site were available for grassland and no EF data were available for leguminous crops (Table 1). All N₂O emission and EF data from leguminous crops and grasslands were within the range of upland fields.

N₂O emissions from fertilized upland fields

Soil water content and soil drainage are very important factors affecting N₂O emission (e.g. Akiyama et al. 2004; Bouwman et al. 2002). However, many field studies do not consider soil water content and soil drainage at the measurement site, although they do consider soil type. Soil type can be used as a general indicator of soil drainage, although soil drainage characteristics can vary within a soil type. Most of the studies we compiled classified soil type according to the system developed by the Classification Committee of Cultivated Soils (1996). We used the same classification system in our study, except for one case that was documented only as alluvial soil, which is not a specific soil type. Based on the general characteristics of each soil type (Classification Committee of Cultivated Soils 1996; Soil Survey Committee of Japan 1991), we categorized each soil as either well-drained or poorly drained. Well-drained soils included Andosols (Andosols or Fluvisols), Yellow Soil (Alisols or Cambisols) and Terrestrial Regosols (Regosols). Poorly drained soils included Wet Andosols (Gleysols or Fluvisols), Brown Lowland Soil (Cambisols or Fluvisols), Gley Soil (Fluvisols, Gleysols or Planosols), Gray Lowland Soil (Fluvisols), Gray Upland Soil (Gleysols or Plandosols) and alluvial soil. Soil types in the FAO–UNESCO classification (1990) were shown in parentheses. More than one name was listed because of the inconsistency between the two classification systems.

N₂O emissions from upland fields where chemical or organic fertilizer had been applied varied greatly, and emissions from poorly drained soils in these fields were generally higher than those from well-drained soils (Fig. 1, Table 1). A weak linear relationship between N₂O emission and total N application rate was observed

Figure 1  Relationship between total N input (kg N ha−1) and N2O emission (kg N ha−1) from upland fields; measurement period more than 90 days.

Table 1a N2O-N emissions and fertilizer-induced emission factors from upland fields with the application of chemical or organic fertilizer

Reference	Location	Soil type	Crop	Total N application rate (kg N ha^−1)	Chemical N application rate (kg N ha^−1)	Organic N application rate (kg N ha^−1)	Crop residue (kg N ha^−1)	Type of chemical N	Type of organic N	Measurement period (days)	N2O-N emission (kg N ha^−1)	Fertilizer- induced emission factor (%)
Well-drained soil
Akiyama & Tsuruta 2002	Ibaraki	Andosol	multi	400	400	0	0	CN, AP		365	0.39
Akiyama & Tsuruta 2002	Ibaraki	Andosol	multi	250	250	0	0	CN, AP		365	0.50
Akiyama & Tsuruta 2003a	Ibaraki	Andosol	veg	150	0	150	0		OC	365	0.45
Akiyama & Tsuruta 2003a	Ibaraki	Andosol	veg	150	0	150	0		CM	365	0.12
Akiyama & Tsuruta 2003a	Ibaraki	Andosol	veg	150	150	0	0	U		365	0.19
Akiyama & Tsuruta 2003a	Ibaraki	Andosol	veg	150	0	150	0		FM	127	0.37
Akiyama & Tsuruta 2003a	Ibaraki	Andosol	veg	150	0	150	0		DC	127	0.11
Akiyama & Tsuruta 2003a	Ibaraki	Andosol	veg	150	150	0	0	U		127	0.25
Akiyama & Tsuruta 2003b	Ibaraki	Andosol	veg	150	0	150	0		PM	365	1.84
Akiyama & Tsuruta 2003b	Ibaraki	Andosol	veg	150	0	150	0		SM	365	0.61
Akiyama & Tsuruta 2003b	Ibaraki	Andosol	veg	150	150	0	0	U		365	0.45
Akiyama et al. 2000	Ibaraki	Andosol	veg	200	200	0	0	U, AS		101	0.16
Hou & Tsuruta 2003	Ibaraki	Andosol	veg	250	250	0	0	U		194	0.78	0.16
Hou & Tsuruta 2003	Ibaraki	Andosol	veg	250	250	0	0	U		194	0.78	0.16
Koga et al. 2004	Hokkaido	Andosol	cereal	129.6	110	0	19.6	A(COM)		365	0.56
Koga et al. 2004	Hokkaido	Andosol	cereal	131.2	110	0	21.2	A(COM)		365	0.53
Koga et al. 2004	Hokkaido	Andosol	legume	80	40	0	40	A(COM)		365	0.19
Koga et al. 2004	Hokkaido	Andosol	legume	84.1	40	0	44.1	A(COM)		365	0.31
Koga et al. 2004	Hokkaido	Andosol	veg	237.1	150	0	87.1	AN(COM)		365	0.34
Koga et al. 2004	Hokkaido	Andosol	veg	279.4	150	0	129.4	AN(COM)		365	2.36
Koga et al. 2004	Hokkaido	Andosol	veg	69.3	60	0	9.3	AN(COM)		365	0.09
Koga et al. 2004	Hokkaido	Andosol	veg	289.9	220	0	69.9	AN(COM)		365	0.82
Koshiba et al. 1996	Gunma	Andosol	cereal	240	150	90	0	unknown	CM	102	0.37
Koshiba et al. 1996	Gunma	Andosol	cereal	240	150	90	0	unknown	CM	120	0.37
Koshiba et al. 1996	Gunma	Andosol	cereal	240	150	90	0	unknown	CM	112	0.73
Koshiba et al. 1996	Gunma	Andosol	cereal	450	0	450	0		CM	102	0.66
Koshiba et al. 1996	Gunma	Andosol	cereal	450	0	450	0		CM	120	0.65
Koshiba et al. 1996	Gunma	Andosol	cereal	450	0	450	0		CM	112	1.23
Li et al. 2002	Ibaraki	Andosol	cereal	150.11	150	0	0.11	AS		95	0.60
Li et al. 2002	Chiba	Andosol	veg	150	150	0	0	AS		93	0.22	0.07
Li et al. 2002	Chiba	Andosol	veg	300	300	0	0	AS		93	0.32	0.07
Li et al. 2002	Chiba	Andosol	veg	150	0	150	0		CM	93	0.30	0.13
Li et al. 2002	Chiba	Andosol	veg	300	0	300	0		CM	93	0.45	0.11
Mori et al. 2005	Tochigi	Andosol	grassland	200	200	0	0	unknown		365	0.39
Mori et al. 2005	Tochigi	Andosol	grassland	200	200	0	0	unknown		365	1.59
Mori et al. 2005	Tochigi	Andosol	grassland	200	200	0	0	unknown		365	0.67
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	600	600	0	0	AS		291	2.25	0.34
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	1000	1000	0	0	AS		291	6.28	0.61
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	600	600	0	0	U, A(COM)		325	1.90
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	600	600	unknown	0	U, A(COM)	RSC	325	1.56
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	0	0	unknown	0		RSC	325	0.45
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	600	600	0	0	U, A(COM)		255	2.26
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	600	600	unknown	0	U, A(COM)	RSC	255	3.72
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	0	0	unknown	0		RSC	255	0.82
Noda 2001	Shimane	Yellow Soil	veg	200	200	0	0	AS		201		0.21
Noda 2001	Shimane	Yellow Soil	veg	200	200	0	0	AN		201		0.31
Shoji et al. 2001	Fukushima	Terrestrial Regosols	cereal	150	150	0	0	U		160	3.15	2.02
Suzuki 1996	Tochigi	Andosol	multi	310	310	0	0	AS		167	0.63
Suzuki 1996	Tochigi	Andosol	multi	310	310	0	0	AS		163	0.21
Suzuki 1996	Tochigi	Andosol	multi	465	465	0	0	AS		167	0.98
Suzuki 1996	Tochigi	Andosol	multi	465	465	0	0	AS		163	0.56
Suzuki 1996	Tochigi	Andosol	multi	440.7	310	130.7	0	AS	RSC	167	0.83
Suzuki 1996	Tochigi	Andosol	multi	511	310	201	0	AS	RSC	163	0.26
Wakazawa & Kosugi 2001	Shizuoka	Andosol	veg	510	260	250	0	COM	RSC	127	0.38
Wakazawa & Kosugi 2001	Shizuoka	Andosol	veg	800	260	540	0	COM	SM	127	2.57
Wakazawa & Kosugi 2001	Shizuoka	Andosol	veg	460	260	200	0	COM	BC	127	0.81
Wakazawa & Kosugi 2001	Shizuoka	Yellow Soil	veg	510	260	250	0	COM	RSC	136	0.73
Wakazawa & Kosugi 2001	Shizuoka	Yellow Soil	veg	800	260	540	0	COM	SM	136	1.21
Wakazawa & Kosugi 2001	Shizuoka	Yellow Soil	veg	460	260	200	0	COM	BC	136	0.11
Wakazawa & Kosugi 2001	Shizuoka	Andosol	multi	780	310	470	0	unknown	SM	253	4.68
Wakazawa & Kosugi 2001	Shizuoka	Andosol	multi	850	310	540	0	unknown	SM	256	2.93
Wakazawa & Kosugi 2001	Shizuoka	Andosol	multi	1160	310	850	0	unknown	SM	267	2.18
Wakazawa & Kosugi 2001	Shizuoka	Yellow Soil	multi	780	310	470	0	unknown	SM	256	2.24
Wakazawa & Kosugi 2001	Shizuoka	Yellow Soil	multi	850	310	540	0	unknown	SM	265	1.29
Wakazawa & Kosugi 2001	Shizuoka	Yellow Soil	multi	1160	310	850	0	unknown	SM	267	1.37
Watanabe et al. 1997	Ibaraki	Andosol	grassland	570	0	570	0		CM, UR	112		0.09
Watanabe et al. 1997	Ibaraki	Andosol	grassland	955	0	955	0		SM, UR	112		0.10
Yan et al. 2001	Ibaraki	Andosol	cereal	150	150	0	0	U		138	0.54	0.27
Yan et al. 2001	Ibaraki	Andosol	cereal	250	250	0	0	U		138	0.51	0.15

Table 1b –Poorly drained soil

Reference	Location	Soil type	Crop	Total N application rate (kg N ha^−1)	Chemical N application rate (kg N ha^−1)	Organic N application rate (kg N ha^−1)	Crop residue (kg N ha^−1)	Type of chemical N	Type of organic N	Measurement period (days)	N2O-N emission (kg N ha^−1)	Fertilizer- induced emission factor (%)
Kumagai and Kanno 1997	Yamagata	Gray Lowland Soil	veg	250	250	0	0	unknown		126	3.65
Kusa et al. 2002	Hokkaido	Gray Lowland Soil	veg	270	270	0	0	AN		208	7.80
Kusa et al. 2002	Hokkaido	Gray Lowland Soil	veg	310	310	0	0	AN		183	3.50
Kusa et al. 2002	Hokkaido	Gray Lowland Soil	veg	280	280	0	0	AN		185	5.30
Kusa et al. 2002	Hokkaido	Gray Lowland Soil	veg	320	320	0	0	AN		200	4.70
Kusa et al. 2002	Hokkaido	Gray Lowland Soil	veg	322	322	0	0	AN		184	9.90
Kusa et al. 2002	Hokkaido	Gray Lowland Soil	veg	242	242	0	0	AN		184	15.60
Kusa et al. in press	Hokkaido	Wet Andosol	cereal	128	98	30	0	MIX	CM	198	8.30
Kusa et al. in press	Hokkaido	Wet Andosol	cereal	128	98	30	0	MIX	CM	199	23.30
Kusa et al. in press	Hokkaido	Wet Andosol	cereal	128	98	30	0	MIX	CM	197	15.50
Nishimura et al. 2005	Ibaraki	Gray Lowland Soil	cereal	60	60	0	0	U		365	2.41
Nishimura et al. 2005	Ibaraki	Gray Lowland Soil	multi	120	120	0	0	U		392	3.19
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	200	0	0	AS		94	2.91	1.28
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	0	200	0		CM	94	1.51	0.58
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	0	200	0		FM	94	6.95	3.30
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	188	12	0	AS	CM	94	2.88	1.26
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	170	30	0	AS	CM	94	3.23	1.44
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	140	60	0	AS	CM	94	1.89	0.77
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	109	91	0	AS	CM	94	1.88	0.76
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	188	12	0	AS	FM	94	1.60	0.62
Ohashi 2000	Hokkaido	Brown Lowland Soil	veg	200	109	91	0	AS	FM	94	5.54	2.59
Shiratori et al. 1999	Niigata	Brown Lowland Soil	veg	400	400	0	0	AN		120	15.39
Shiratori et al. 1999	Niigata	Brown Lowland Soil	veg	400	400	0	0	AN		120	7.22
Shiratori et al. 1999	Niigata	Brown Lowland Soil	veg	400	400	0	0	AN		120	1.09
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	cereal	90	90	0	0	COM		198	2.26
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	cereal	90	90	0	0	COM		187	0.77
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	cereal	90	90	0	0	COM		198	2.67
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	1000	0	1000	0		RSC	152	1.05
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	1000	0	1000	0		SM	152	1.42
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	1000	0	1000	0		BC	152	0.40
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	500	0	500	0		RSC	152	0.11
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	500	0	500	0		SM	152	0.80
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	500	0	500	0		BC	152	0.07
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	200	0	200	0		OP	156	2.41
Wakazawa & Kosugi, 2001	Shizuoka	Gray Lowland Soil	veg	200	200	0	0	AS		122	0.24
Measurement period was more than 90 days. Measurements with slow-release fertilizer and nitrification inhibitors are not included. veg, vegetable; multi, multiple cropping in a year; U, urea; AS, ammonia sulfate; CN, calcium nitrate; A, ammonium; AN, ammonium nitrate; MIX, mixed fertilizer; COM, complex fertilizer; AP, ammonium phosphate; CM, cattle manure; SM, swine manure; FM, fish meal; BC, bark compost; PM, poultry manure; DC, dried cattle excreta; UR, urine; OC, oil cake; RSC, rice straw compost; OP, mixed organic fertilizer pellets.

Table 2 Summary of N2O-N emissions and fertilizer-induced emission factors from Japanese upland fields (except tea fields)

Soil drainage^†	n	Mean	Standard deviation	Median	Min.	Max.
N2O-N emission (kg N ha^−1)
Well-drained soil	67	1.03 a**	1.14	0.61	0.09	6.28
Poorly drained soil	35	4.78 b	5.36	2.88	0.07	23.3
Fertilizer-induced N2O-N emission factor (%)
Well-drained soil	15	0.32 a**	0.49	0.16	0.07	2.02
Poorly drained soil	9	1.40 b	0.95	1.26	0.57	3.30
Estimated emission factor for all soil		0.62^‡	0.48^‡‡
Measurement period more than 90 days. ^†Soil drainage was categorized into two classes on the basis of soil type: (1) well-drained soil: Andosol, Terrestrial Regosols, Yellow Soil, (2) poorly drained soil: Wet Andosol, Brown Lowland Soil, Gley Lowland Soil, Gray Lowland Soil, Gray Upland Soil, alluvial soil.
^‡Weighted by distribution of soil type well-drained soil 72%; poorly drained soil 28% (Soil Survey Committee of Japan 1991).
^‡‡Standard deviation was calculated according to the National Greenhouse Gas Inventory Report of Japan (Greenhouse Gas Inventory Office of Japan 2005). Uncertainty of activity data (soil type distribution data from Soil Survey Committee of Japan 1991) was assumed to be standard deviation of 5%. Means followed by “a” or “b” are significantly different (P < 0.01) using a t-test carried out with log-transformed data.

for well-drained soils (adjusted r ²  = 0.38, P < 0.01), which suggests that approximately 38% of the variability was explained by total N input. The total N input was not clearly related to N₂O emissions for poorly drained soils. This result is not surprising because N₂O emissions are affected by many factors other than N input, such as fertilizer type, temperature, soil texture and soil pH (e.g. Bouwman et al. 2002; Granli & Bockman 1994). These factors could not be considered in our study because of limited data. Moreover, many processes in the soil produce N₂O, although it is mainly produced by two microbial processes, nitrification (aerobic) and denitrification (anaerobic) (e.g. Davidson 1991). In addition, N₂O is also produced by nitrifier-denitrification and chemodenitrification, although these processes are less well understood (e.g. Wrage et al. 2001). N₂O emissions also show large annual variation (e.g. Akiyama and Tsuruta 2003a; Kusa et al. 2002) and spatial variation (e.g. Ball et al. 2000; Yanai et al. 2003). Thus, often no relationships between N input and N₂O emissions are observed when N₂O emissions data are collected from many different studies. Worldwide studies of emissions from upland fields (Food and Agriculture Organization/International Fertilizer Industry Association 2001) and rice paddy fields (Akiyama et al. 2005) have also found no clear relationship between N input and N₂O emissions. In contrast, a liner relationship was observed between N input and N₂O emissions when less data were available for N₂O emissions from upland fields in Europe and the USA (Bouwman 1996).

Although N₂O emissions from agricultural fields cannot be described as a simple function of N input, a simple method to estimate Japanese N₂O emission inventories is required. When other contributing conditions are similar, N₂O emissions increase with increases in N input (e.g. Li et al. 2002). Therefore, the mean and median of N₂O emissions and EFs are relevant parameters. In our calculations of these for upland fields (Table 2), the limited availability of zero-N control data reduced the number of values we could determine for EF. Estimates of annual EFs are needed, but measurement periods for most of the studies we considered were considerably shorter than 1 year, and no studies collected data over a period longer than 1 year. Limiting our analysis to data derived over longer periods would raise the possibility of our calculated mean and median values being biased by extreme values in the smaller dataset. If our analysis is not limited in this way, the mean and median EFs would be biased by data measured over too short a period of time. We, therefore, only used data measured over a period longer than 90 days to estimate EF for a whole year, assuming that most of the fertilizer-induced N₂O emission occurs within this period. Although we believe that using data measured over a period longer than 90 days is the best to estimate EF for Japanese upland fields at this point, it should be noted that this estimate should be revised when more data measured over longer periods becomes available in the future. Bouwman et al. (2002) reported that data measured over a period shorter than 120 days were significantly lower than data measured over a period of more than 180–300 days.

Mean N₂O emissions (± standard deviation) for fertilized upland fields with a measurement period longer than 90 days were 1.03 ± 1.14 kg N ha⁻¹ for well-drained soils and 4.78 ± 5.36 kg N ha⁻¹ for poorly drained fields (Table 2). Calculated EFs were 0.32 ± 0.49% for well-drained soils and 1.40 ± 0.95% for poorly drained fields. The large differences between means and medians indicate that the data distributions are skewed.

Mean N₂O emissions and EFs from well-drained fields were significantly lower than those for poorly drained soils. To estimate an EF for both well-drained and poorly drained soils in upland fields we used an area-weighted mean for two reasons: (1) the ratio of the total areas of well-drained soils and poorly drained soils was different from the ratio of the number of datasets available for each soil category, (2) no statistical data were available for the chemical and organic N fertilizer application rate for each soil class of drainage. The areas of well-drained and poorly drained soils used for weighting were based on soil-distribution data published by the Soil Survey Committee of Japan (1991). To calculate the area of each class of soil drainage, soils for which there were no emission data had also to be categorized. Lithosol (Leptosols), Sand-dune Regosol (Arenosols or Regosols), Brown Forest Soil (Cambisols), Red Soil (Alisols, Acrisols or Cambisols) and Dark Red Soil (Luvisols or Cambisols) were categorized as well-drained soils, and Gleyed Andosol (Histosols or Gleysols), Gley Upland Soil (Gleysols, Planosols) and Muck Soil (Histosols) were categorized as poorly drained soils. Upland field areas were estimated to comprise 72% well-drained soils and 28% poorly drained soils. We assumed that this ratio of soil types had not changed since the survey, which was conducted approximately 30 years ago.

The area-weighted mean EF we calculated for Japanese upland fields was 0.62% (Table 2), which was lower than the IPCC default EF of 1.25%. It was also lower than the global EF of 0.8% estimated by the Food and Agriculture Organization/International Fertilizer Industry Association (2001) and the country-specific EFs used in Sweden (1–1.5%; Swedish Environmental Protection Agency 2006) and the Netherlands (1% for chemical fertilizer and 2% for manure; Netherlands Environmental Assessment Agency 2006). This might be because poorly drained soils in Japan are mainly used for rice paddy fields and as a consequence the proportion of well-drained upland field in Japan is relatively high (72%).

Background emission from upland fields

Mean daily background N₂O emissions (from fields without N fertilizer application) for upland fields are shown in Table 3, and the statistical analysis of these data is summarized in Table 4. Mean daily background N₂O emission from well-drained fields was significantly lower than that from poorly drained soils, although only one measurement from poorly drained soils was available. Large variations were observed in background emissions for well-drained soils. Available data for background N₂O emissions were very limited and measurement periods were considerably shorter than 1 year. Therefore, annual background N₂O emission was estimated by multiplying the mean daily N₂O emission by 365. The estimated annual background emission from upland fields was 0.36 kg N ha⁻¹ year⁻¹ for well-drained soils and 1.40 kg N ha⁻¹ year⁻¹ for poorly drained soils. The area-weighted mean background emission for all upland fields (well drained and poorly drained) was 0.65 ± 0.45 kg N ha⁻¹ year⁻¹, which was lower than the IPCC default value (1 kg N ha⁻¹ year⁻¹). This might be because poorly drained soils in Japan are mainly used for rice paddy fields and as a consequence the proportion of well-drained upland fields in Japan is relatively high (72%), similar to the estimated EF for upland fields. It is important to note that there are large uncertainties in these estimates because the difference between N₂O emissions during cropping and those during fallow periods has not been taken into account.

Table 3 Background N2O-N emissions from upland fields

Reference	Location	Soil type	Crop	Measurement period (days)	N2O emission (g N ha^−1 day^−1)
Well-drained soil
Akai et al. 2001	Okayama	Andosol	grassland	60	1.51
Akai et al. 2001	Okayama	Andosol	grassland	60	0.49
Cheng et al. 2002	Ibaraki	Andosol	veg	80	1.53
Hou & Tsuruta 2003	Ibaraki	Andosol	veg	194	1.96
Li et al. 2002	Chiba	Andosol	veg	93	1.15
Nishiwaki & Inoue 1996	Aichi	Yellow Soil	multi	291	0.69
Shoji et al. 2001	Fukushima	Terrestrial Regosols	cereal	160	0.75
Tanizaki 1992	Ibaraki	Andosol	veg	72	0.00
Yan et al. 2001	Ibaraki	Andosol	cereal	138	0.91
Poorly drained soil
Ohashi 2000	Hokkaido	Brown lowland Soil	veg	94	3.83
Measurement period more than 60 days.veg, vegetable; multi, multiple cropping in a year.

Table 4 Summary of daily background N2O-N emission and estimated annual background N2O-N emission from Japanese upland fields

Soil drainage^†	n	Mean	Standard deviation	Median	Min.	Max.
Daily N2O-N emission (g N ha^−1 day^−1)
Well-drained soil	7	1.00 a**	0.63	0.91	0.00	1.96
Poorly drained soil	1	3.83 b		3.83	3.83	3.83
Estimated annual N2O-N emission (kg N ha^−1 year^−1)^§
Well-drained soil		0.36
Poorly drained soil		1.40
Estimated background emission for all soil		0.65^‡	0.45^‡‡
Measurement period more than 60 days. ^†Soil drainage was categorized into two classes on the basis of soil type: (1) well-drained soil: Andosol, Terrestrial Regosols, Yellow Soil, (2) poorly drained soil: Wet Andosol, Brown Lowland Soil, Gley Lowland Soil, Gray Lowland Soil, Gray Upland Soil, alluvial soil.
^§Annual emissions were estimated by integrating daily emission over a year.
^‡Weighted by distribution of soil type well-drained soil 72%; poorly drained soil 28% (Soil Survey Committee of Japan 1991).
^‡‡Standard deviation was calculated according to the National Greenhouse Gas Inventory Report of Japan (Greenhouse Gas Inventory Office of Japan 2005). Uncertainty of activity data (soil type distribution data from Soil Survey Committee of Japan 1991) was assumed to be standard deviation of 5%. Means followed by “a” or “b” are significantly different (P < 0.01) using a t-test.

Figure 2  Relationship between total N input (kg N ha−1) and N2O emission (kg N ha−1) from tea fields; measurement period more than 90 days.

N₂O emission from tea fields

Mean N₂O emission (± standard deviation) for fertilized tea fields was 24.3 ± 16.3 kg N ha⁻¹ (Tables 5,6, Fig. 2). Mean N₂O emission from tea fields was markedly higher than that from upland fields; thus, it is important to estimate EF for tea fields. However, reliable zero-N control data, essential to calculate EF, were not available. Tokuda (2006) reported that N₂O-N emissions from “zero-N control” tea fields were approximately 3.66–4.24 kg N ha⁻¹ for the measurement period of 209 days, but these data were highly likely to be affected by the repeated application of large amounts of N fertilizer in previous years (740 kg N ha⁻¹ year⁻¹, S. Tokuda, 2005, pers. comm.). The distribution of soil types within tea fields was not available, and the estimated background emission used for upland fields in this study was not necessarily appropriate for tea fields. Therefore, we used the Intergovernmental Panel on Climate Change (1997) default value of 1 kg N ha⁻¹ year⁻¹ for background emission, and the estimated fertilizer-induced EF for fertilized tea fields was 2.82 ± 1.80%. Individual measurement periods for emissions from tea fields ranged from 209 to 365 days, which were longer measurement periods than those available for upland fields. In three of the four reports, recorded emissions data covered an entire year. These data showed that emissions during winter were generally very low; thus, emission data measured over a period of more than 209 days, from March to November, would be comparable to data for an entire year. Therefore, we used all available emission measurements to estimate EF for tea fields.

It should be noted that large amounts of N are added to tea field soils from litter fall and trimmed branches. The amount of N from these sources could be as high as 250 kg N ha⁻¹ year⁻¹ (Hoshina 1985). However, the amount of N added through litter fall and trimmed branches was not documented in the published reports used in this study and was, thus, not considered in this study. Further field measurements taking into account the input of N through litter fall and trimmed branches are needed.

N₂O emissions from rice paddy fields

The mean N₂O emission from Japanese rice paddy fields was 0.36 kg N ha⁻¹ for the cropping season (Tables 7,8,

Table 5 N2O-N emissions from tea fields with the application of chemical or organic fertilizer

Reference	Location	Soil type	Total N application rate (kg N ha^−1)	Chemical N application rate (kg N ha^−1)	Organic N application rate (kg N ha^−1)	Crop residue (kg N ha^−1)	Type of chemical N	Type of organic N	Meas- urement period (days)	N2O emission (kg N ha^−1)
Maeda et al. 2001	Aichi	Yellow Soil	727	727	0		unknown		327	61.00
Maeda et al. 2001	Aichi	Brown lowland Soil	826	826	0		unknown		331	25.00
Tokuda & Hayatsu 2004	Shizuoka	Andosol	600	600	0		AS		365	0.60
Tokuda & Hayatsu 2004	Shizuoka	Andosol	1200	1200	0		AS		365	25.22
Tokuda & Hayatsu 2004	Shizuoka	Andosol	600	600	0		AS		365	3.54
Tokuda & Hayatsu 2004	Shizuoka	Andosol	600	600	0		AS		365	3.08
Tokuda et al. 2006	Shizuoka	Yellow Soil	300	300	0		AS		209	4.33
Tokuda et al. 2006	Shizuoka	Yellow Soil	600	600	0		AS		209	12.70
Tokuda et al. 2006	Shizuoka	Yellow Soil	900	900	0		AS		209	29.75
Tokuda et al. 2006	Shizuoka	Yellow Soil	1200	1200	0		AS		209	19.60
Tokuda et al. 2006	Shizuoka	Yellow Soil	300	Unknown	Unknown		AS	OC	209	2.39
Tokuda et al. 2006	Shizuoka	Yellow Soil	500	Unknown	Unknown		AS	OC	209	3.76
Tokuda et al. 2006	Shizuoka	Yellow Soil	720	720	0		AS		209	13.59
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	41.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	39.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	42.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	45.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	32.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	33.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	32.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	33.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	128	0	128			C	365	3.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	128	0	128			C	365	3.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	20.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		U, AS, A(COM)	FM, OC, C	365	29.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		AN	FM, OC, C	365	34.00
Tokunaga et al. 1996	Yamaguchi	Terrestrial Regosols	894	476	418		AN	FM, OC, C	365	41.00
Measurement period more than 209 days. Measurements with slow-release fertilizer and nitrification inhibitor are not included. U, urea; AS, ammonia sulfate; A, ammonium; AN, ammonium nitrate; COM, complex fertilizer; FM, fish meal; OC, oilcake; C, compost.

Table 6 Summary of N2O-N emission (kg N ha−1) and fertilizer-induced emission factor (%) from Japanese tea fields

	n	Mean	Standard deviation	Median	Min.	Max.
N2O-N emission (kg N ha^−1)
	26	24.3	16.3	27.11	2.39	61.0
Estimated fertilizer-induced emission factor (%)^†
	26	2.82	1.80	3.02	0.35	8.25
^†Background emission was assumed to be 1 kg ha^−1 year^−1 (IPCC default value) because no reliable zero-N control data were available. One emission measurement that was smaller than the estimated background emission was excluded.

Figure 3  Relationship between total N input (kg N ha−1) and N2O emission (kg N ha−1) from rice fields; measurement period more than 90 days.

Fig. 3). Rice paddy fields cover 54% of the agricultural land in Japan and are an important land use (Ministry of Agriculture, Forestry and Fishery 2005). It is, therefore, important to estimate an EF for rice paddy fields to allow calculation of the Japanese greenhouse gas inventory. However, none of the published measurements of emissions from Japanese rice fields included zero-N control data. Thus, an EF for this category of land use could not be calculated. Akiyama et al. (2005) reported that the worldwide EF for rice paddy fields (mean ± standard deviation of 0.31 ± 0.31%) is lower than that for upland fields, and we recommend the use of that EF for Japanese rice fields.

Yan et al. (2003) estimated worldwide background emissions from rice paddy fields to be 0.26 kg N ha⁻¹

Table 7 N2O-N emissions from rice paddy fields with the application of chemical fertilizer

Reference	Location	Soil type	Total N application rate (kg N ha^−1)	Chemical N application rate (kg N ha^−1)	Organic N application rate (kg N ha^−1)	Crop residue (kg N ha^−1)	Type of chemical N	Type of organic N	Measurement period (days)	N2O emission (kg N ha^−1)
Minami 1987	Ibaraki	alluvial soil	100	100	0	0	unknown		120	0.33
Minami 1987	Ibaraki	Andosol	100	100	0	0	unknown	–	120	0.55
Minami 1987	Ibaraki	Andosol	90	90	0	0	unknown	–	139	0.27
Nishimura et al. 2004	Ibaraki	Gray Lowland Soil	122.4	90	0	32.4	U	–	365	0.60
Yagi et al. 1996	Ibaraki	Gley Soil	117	90	0	27	A(COM)	–	150	0.07
Yagi et al. 1996	Ibaraki	Gley soil	117	90	0	27	A(COM)	–	150	0.07
Measurement period more than 90 days. Measurements with slow-release fertilizer and nitrification inhibitor are not included. U, urea; A, ammonium; COM, complex fertilizer.

Table 8 Summary of N2O-N emission (kg N ha−1) from Japanese rice paddy fields

Soil drainage^†	n	Mean	Standard deviation	Median	Min.	Max.
Well-drained soil	2	0.41a*	0.20	0.41	0.27	0.55
Poorly drained soil	4	0.33a	0.27	0.33	0.07	0.60
All soil	7	0.36	0.24	0.33	0.07	0.60
Measurement period more than 90 days. ^†Soil drainage was categorized into two classes on the basis of soil type: (1) well-drained soil: Andosol, Terrestrial Regosols, Yellow Soil, (2) poorly drained soil: Wet Andosol, Brown Lowland Soil, Gley Lowland Soil, Gray Lowland Soil, Gray Upland Soil, alluvial soil. Means followed by ‘a’ or ‘b’ are significantly different (P < 0.01) using a t-test.

during the cropping season alone, and annual emissions to be 0.81 kg N ha⁻¹ year⁻¹. Akiyama et al. (2005) estimated worldwide background emissions for rice paddy fields to be 0.33 kg N ha⁻¹ during the cropping season alone and the annual emission to be 1.82 kg N ha⁻¹ year⁻¹. However, as acknowledged by the authors, limited data availability means that there are large uncertainties in both of these estimates.

Other factors affecting N₂O emissions

N₂O emissions from agricultural fields are affected by many factors, including the type of fertilizer used, climate and soil type. However, those factors were not considered in our study because of limitations in the available data. Further, there were insufficient data to allow us to consider the different effects of chemical and organic fertilizers on EFs. We did not consider interannual variation, even though large interannual variation in N₂O emissions has been reported (e.g. Akiyama and Tsuruta 2003a; Kusa et al. 2002).

Our analysis was also affected by the simple crop and soil drainage classification schemes we used. We classified crops into three categories and soil drainage characteristics into two categories: the actual situation is more complicated. Moreover, because soil drainage characteristics at emission measurement sites were not documented in many of the published reports we used, we classified soil drainage on the basis of soil type.

Our analysis of the published data was also affected by variability in both the length of the period of measurement and the frequency of measurement within that period. Some of the data were obtained by automated N₂O flux monitoring, with intensive measurement over a long period (e.g. Akiyama et al. 2000; Akiyama and Tsuruta 2002, 2003a,b; Nishimura et al. 2004). Most of the available data, however, were obtained manually using the static chamber method, with less frequent measurement (typically once or twice per week) and a shorter measurement period. According to Bouwman et al. (2002), experiments with frequent measurements (more than once per day) yield lower N₂O emissions than those with less frequent measurements (less than once per day), and experiments with longer measurement periods yield higher N₂O emission than those with shorter measurement periods. Because of the limitations of the data, we estimated EFs for upland fields using only data measured over a period longer than 90 days, and assumed that the fertilizer-induced N₂O emissions recorded over that period would be representative of annual emissions. These limitations and assumptions will have affected the estimates of our study. More field measurements covering a variety of crop types, in particular frequent measurements covering an entire year, and studies that include zero-N control data are required for further analyses.

CONCLUSIONS

We reviewed published data on N₂O emissions from Japanese agricultural fields to attempt to establish a quantitative basis on which to determine Japanese greenhouse gas emission inventories. We determined fertilizer-induced emission factors of 0.62 ± 0.48% for fertilized upland fields and 2.82 ± 1.80% for tea fields, and we propose that an EF of 0.31 ± 0.31% (Akiyama et al. 2005) be used for rice paddy fields until more comprehensive data are available. We estimated annual background emissions from upland fields to be 0.36 kg N ha⁻¹ year⁻¹ for well-drained soils and 1.40 kg N ha⁻¹ year⁻¹ for poorly drained soils. There are significant uncertainties in our estimates of background emissions because of limitations in the available data, particularly a lack of measurements during the fallow period. More field measurements, especially measurements covering an entire year, and studies that include zero-N control data are required for further analyses.

ACKNOWLEDGMENT

This work was supported by the Global Environment Research Fund “S-2-3a”, Ministry of the Environment, Japan.