China is the largest rice producer in the world. Methane (CH4) emission from its rice fields has been widely measured since the late 1980s. This study collected the results of available field research, totaling 204 season-treatment measurements conducted on 23 sites. Analysis of these data shows that input of organic material, such as green manure, animal waste, and crop straw, increases CH4 emission by a factor of 2. Average CH4 flux from intermittently irrigated rice fields is 53% of that from continuously flooded rice fields; and average CH4 emission flux from late rice fields is 1.6 and 2.3 times greater than flux from early and single rice fields, respectively. There are regional differences in emission factors and a trend of decreasing emission from south to north. On the basis of earlier estimates of CH4 emission from Chinese rice fields, and recent reports on the use of crop residue and green manure, it is presumed that half of the rice fields in China receive organic input. From the frequency of various water management events indicated in the surveyed field experiments, as well as from specific statements in individual reports, it is presumed that 2/3 of irrigated rice fields have been intermittently flooded. On the basis of these assumptions, the region-specific emission factors, and 1995 data on rice cultivation area, CH4 emission from growing-season rice fields in Mainland China was estimated to be 7.67 Tg yr−1, ranging from 5.82 to 9.57 Tg yr−1, due to uncertainties in the areas receiving organic inputs, and intermittent irrigation. Generalized seasonal flux patterns were developed for early, late, and single rice. Monthly distributions of emission were estimated from these patterns and rice calendars. The highest emission rate occurred in August. Spatially, emission hot spots included the plains of Dongting Lake in Hunan Province, Boyang Lake in Jiangxi Province, the delta region of Qiantang River in Zhejiang Province, and the Sichuan Basin. Nearly 90% of all Mainland China CH4 emission occurred between 23°N and 33°N.
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.
 Rice fields are an important source of atmospheric methane (CH4). As the largest rice producer in the world, China has been of particular concern in the past two decades as a source of CH4. Estimates of CH4 emission amounts from Chinese rice fields vary greatly, however, depending on the estimation methods and field measurements used. Earlier estimates were generally based on a single field observation. One of the earliest calculations was made by extrapolating a flux of 58 mg CH4 m−2 h−1, the average CH4 emission flux for the rice-growing season of two consecutive years in Tuzu, Sichuan Province, to the whole of China. The resulting estimate was 30 Tg CH4 yr−1 [Khalil et al., 1991]. After measuring consecutively for 7 years, however, an average CH4 emission flux of 30 mg m−2 h−1 was found to occur during the rice-growing season [Khalil et al., 1998]. Similarly, Wassmann et al. [1993a] extrapolated the results of measurements in Hangzhou, Zhejiang Province to the entire country, and estimated an emission of 18–28 Tg CH4 yr−1. As field measurements accumulated, more flux data were included in upscaling methods. Yao et al.  used flux data from 6 sites to represent 10 agroecological zones, and estimated 15.3 Tg CH4 yr−1. Evaluating results from 12 field sites, Cai  concluded emission was 8.05 Tg CH4 yr−1, and considered the effects of water regime and organic fertilizer application.
 Many models have been developed to estimate CH4 emission from rice fields. Cao et al.  developed a simplified process-based methane emission model. Rice primary production and soil organic degradation were included as supplies of carbon substrate for methanogens, and environmental controls of methanogenesis were also included; estimated CH4 emission from China's rice fields was 16.2 Tg yr−1. Bachelet et al.  estimated 10.47 Tg CH4 yr−1 using a regression equation of carbon and nitrogen inputs from organic matter and fertilizer. Huang et al.  estimated the emission to be 9.66 Tg CH4 yr−1; this model considered daily methane emission flux as a function of photosynthetic activity, and incorporated organic matter, soil sand content, temperature, and rice cultivar. The most recent, and probably the most sophisticated, model to simulate methane emission from rice is that of Matthews et al. . This process model is based on a rice crop simulation model; it integrates the effects of climate, soil, agricultural management, and the growing of rice on methane flux. They calculated an emission of 3.35–8.64 Tg CH4 yr−1 for China, and concluded that a more realistic estimate was 7.22–8.64 Tg CH4 yr−1.
 It appears that as more field measurements are included in upscaling estimation methods, or as models become more complex, the lower the estimate of CH4 emission from Chinese rice fields becomes. Many field measurements have emerged in recent years, allowing the determination of more reliable, region-specific, emission factors in China. This study makes use of most field measurements available to date and provides an updated estimate of CH4 emission from Chinese rice fields.
 Most of the estimations based on scaling-up methods are for total annual CH4 emission. The few studies that estimated monthly emission assumed a constant daily flux [Matthews et al., 1991; Yao et al., 1996]. In this study, seasonal patterns of emission flux were derived separately for early, late, and single rice. The monthly distribution of CH4 emission from Chinese rice fields was estimated using these patterns and an updated rice calendar by Matthews et al. . Emissions were distributed spatially with the aid of a land-use map.
2. Estimation of CH4 Emission
2.1. Regionalization of Rice Fields
 Rice is widely cultivated in China, and CH4 emission measurements from different areas in China have been shown to vary regionally. Emission flux, under similar management practices (e.g., organic fertilizer use, water status), generally decreases from south to north, with changes in rainfall and temperature. Local crop scientists have categorized Chinese rice fields into six regions with 10 sub-regions, according to differences in climate and agricultural practice [Chinese Academy of Agricultural Sciences, 1986]. The Food and Agriculture Organization of the United Nations (FAO) has a similar zoning system, based on temperature and rainfall differences, that generally places Chinese rice fields into four agroecological zones (AEZs): AEZ 7, the warm/cool humid subtropics with summer rainfall; AEZ 6, the warm sub-humid subtropics with summer rainfall; AEZ 5, the warm arid and semiarid subtropics with summer rainfall; and AEZ 8, the cool subtropics with summer rainfall [International Rice Research Institute (IRRI), 1997]. As field measurements were not available for all of the sub-regions of the Chinese zoning system, and to compare results from other countries, the FAO zoning system was used in this study. However, measurements have consistently shown that CH4 flux from fields in the western part of AEZ 6 is much higher that that in eastern AEZ 6 [Cao and Hong, 1996; Khalil et al., 1998; Cai et al., 2000]. Therefore for the current study, this region was divided into AEZ 6A and AEZ 6B. As statistical data on land area under rice cultivation were only available for political units, the boundaries of each AEZ region were modified to follow the nearest provincial boundary. Figure 1 shows the zones as used in this study: AEZ 7 was the double rice region in south and central China, including Hunan, Jiangxi, Zhejiang, Fujian, Guangdong, Guangxi, and Hainan Provinces; AEZ 6A was the central China single/double rice region, including Shanghai, Jiangsu, Anhui, Hubei, and Henan Provinces; AEZ 6B was the southwest single rice region, including Sichuan, Yunnan, and Guizhou Provinces; AEZ 5 was the north China sub-humid single rice region, including Beijing, Tianjin, Hebei, Shandong, Shanxi, and Liaoning Provinces; AEZ 8 was the single rice region in the northeast and west China.
2.2. Estimation of Region-Specific Emission Factors
 Most published results of flux measurements conducted on rice fields in the 5 AEZs were collected, but measurements from lysimeters or non-typical rice fields were excluded [e.g., Wassmann et al., 1993a; Xu et al., 1997]. In total, 204 season-treatment measurements conducted on 23 individual sites were included in this study. All measurements were made with the closed chamber method. Most measurements were made under conditions of either continuous flooding or intermittent irrigation, but some were taken under conditions of continuous flooding with middle season aeration (MSA) and end season drainage (ESD). The latter water regime was classified as intermittent irrigation, since it included at least two drainage periods during the rice-growing season. CH4 flux varies greatly with water regime and organic input; in Beijing rates as low as 0.17 mg m−2 h−1 have been measured in a single rice field with MSA and ESD [Wang et al., 2000]. However, rates as high as 56 mg m−2 h−1 have been measured in a late rice field with continuous flooding in Hunan [Wassmann et al., 1993b]. It is therefore necessary to re-compile the flux data by region, water regime, input of organic fertilizers, and rice season (Table 1). Fluxes from the same treatment (the same water regime, organic input, and season) conducted on the same site were averaged. Fluxes from the same treatment in the same zone were again averaged to determine the region-specific emission factors (Table 2). Field measurements have been taken, under both intermittent irrigation and continuous flooding, during the dominant rice season of each region, but complementary flux data were not always available for comparing with- and without-organic input treatments (Table 2).
Table 1. CH4 Flux From Irrigated Rice Field During Rice Growing Season on Different Sites of China (mg CH4 m−2 h−1)a
Values without superscript are averaged factors measured in the same region, with the same organic treatment, water regime and of the same rice season.
With organic input.
Without organic input.
Values are estimated, either from the corresponding emission flux for treatment without organic input, using a correction factor of 2, or from the corresponding emission flux for treatment with organic input, using a correction factor of 0.5.
Values are followed by standard error and number of sites; values not followed by standard error are based on only one site.
Table 1 shows 15 pairs of flux data, comparing CH4 emissions from rice fields with and without organic input, controlling for other conditions (i.e., site, rice season, and water regime). The ratios of CH4 flux with organic input to flux without organic input range from 0.7 to 4.2, with an average of 2.08 and a standard deviation of 1.16. This average ratio is very close to the default correction factor of 2.0 from the Intergovernmental Panel on Climate Change (IPCC) methodology for estimation of emission flux from rice fields [IPCC, 1997]. This correction factor was therefore used to estimate CH4 emission flux from fields with organic input, based on the corresponding emission flux from fields without organic input; a factor of 0.5 was used to calculate the flux from fields without organic input from the flux with organic input. The estimated emission factors are indicated by the superscript e in Table 2.
2.3. Main Factors Influencing CH4 Emission Flux From Irrigated Rice Fields
 As mentioned above, there are regional trends in CH4 fluxes. For example, emission fluxes from intermittently irrigated single rice fields without organic input in AEZ 7, AEZ 6A, AEZ 5, and AEZ 8 were 6.67, 3.57, 2.75, and 0.25 mg CH4 m−2 h−1, respectively (Table 2). CH4 fluxes from rice fields in AEZ 6B [Cao and Hong, 1996; Khalil et al., 1998; Cai et al., 2000] were consistently higher than those in AEZ 6A [Chen et al., 1993; Zhen et al., 1997; Lin et al., 2000; Cai et al., 2000]. This is at least partially because AEZ 6B is a mountainous area, so rice fields are usually located in basins and valleys, and these are areas characterized by poor drainage. The average soil pH value in AEZ 6B was significantly lower than that in AEZ 6A [see Knox et al., 2000], and CH4 emission was negatively related to soil pH [Yan et al., 2003]. Because of these regional differences, it was necessary to determine region-specific factors for accurate estimation of rice field CH4 emission, especially in a country as geographically diverse as China.
 Many researchers have reported higher CH4 flux from late rice fields than from early rice fields [Wassmann et al., 1993b, 1996; Lin et al., 2000; Cai et al., 2000]. Data (see Table 2, not including the estimates) indicate that the average flux ratios of late rice to early rice and to single rice are 1.58 and 2.28, respectively. Cai et al.  demonstrated that CH4 flux from a rice field flooded in the preceding fallow season was higher than that from a rice field drained in the preceding fallow season. They also showed that CH4 flux from a late rice field, preceded by early rice, was significantly higher than that preceded by an upland crop. This suggests that the water status prior to the rice-growing season is very important for CH4 emission during the rice-growing season. Since rice fields are generally drained in the fallow season, and since the fallow period before a single rice season is longer than that before an early rice season, CH4 flux from early rice fields tends to be higher than that from single rice fields; CH4 flux from late rice fields tends to be higher than that from early rice. As a result, it is important to distinguish between these three rice seasons when estimating emissions.
 It has been well documented that water regime and organic input are two of the most influential factors for CH4 emission from rice fields [Yagi and Minami 1990; Denier van der Gon and Neue, 1995; Husin et al., 1995; Wassmann et al., 2000]. As mentioned above, CH4 emission was doubled under an average organic material input rate of 10 ton ha−1. Intermittent irrigation may reduce CH4 emission by as much as 93% [Yan et al., 2000]. Table 1 shows 11 pairs of data that compare the effects of intermittent irrigation and continuous flooding on CH4 emission from rice fields, controlling for other conditions (i.e., organic input, rice season, and site). CH4 flux from intermittently irrigated fields was 7–102% of that from continuously flooded fields. The average, 53%, is close to the default IPCC scaling factor for estimating CH4 emission from rice fields with single aeration: 50% [IPCC, 1997].
 CH4 emission is also influenced by other factors, such as rice cultivar [Duan et al., 1999] and soil properties [Cai et al., 1999], but there are insufficient data to derive a quantitative relationship between CH4 emission and these factors. Their effects were therefore not distinguished in this study.
2.4. Area of Rice Fields With Organic Input and Intermittent Irrigation
 Since CH4 emissions are affected, and accordingly classified, by region, season, water regime and organic input status, the area of rice cultivation should be determined per region, season, water regime and organic input combination. The area under rice cultivation per region and per season is available from statistical yearbooks. There is no statistical information, however, on the area of rice fields receiving organic inputs. Huang et al.  assumed that 30% of rice fields received organic inputs in addition to the 150 g m−2 of organic materials remaining from the previous crop. Cai  assumed that 20% of rice fields received animal and human waste, and another 20% received crop straw and green manure inputs. There is a long history of using organic fertilizer in Chinese agricultural practice. Primary sources of organic fertilizer include green manure, animal and human waste, and crop straw. The use of organic fertilizer has decreased dramatically since the late 1970s, due to the increased availability of mineral fertilizer and an increase in labor costs. There were 7.9 million ha of green manure fields in 1979, and only 4.1 million ha in 1992 [Editorial Board of China Agriculture Yearbook, 1981, 1993]. After 1992, the statistic was no longer reported. In recent years, however, there has been an indication that while the use of green manure is decreasing, the area of rice fields receiving crop straw is increasing [Li et al., 2000]. Actually, the government has begun to encourage farmers to return crop straw to the field, rather than burning it, since burning crop residue in situ has caused serious environmental problems, frequently reported in the media. Many types of agricultural devices that facilitate returning crop straw to the field have also been developed in recent years. Although there are no official statistics, in 1999 nearly 30% of crop straw was reportedly returned to the field in 10 major grain-producing provinces http://www.agrionline.net.cn/news/news.asp?id = 20010705c1). Therefore it is presumed that 30% of rice fields receive organic input in the form of crop straw, green manure, or compost (derived from crop residue, green manure, and other organic materials). Another 20% of rice fields are assumed to receive organic input in the form of animal and human waste. In total, 50% of rice fields are assumed to have organic input of various types.
 Statistics on water management of rice fields are also lacking. However, intermittent irrigation and middle season aeration are common treatments in Chinese rice fields for the purpose of supplying oxygen to plant roots and to inhibit invalid tillering. Wu and Ye  estimate that rice fields in China are not flooded for 1/3 of the growing season. Of the 23 experiments surveyed for this study, 7 experiments had a water regime of continuous flooding. In these instances, continuous flooding was presumed to be the dominant local practice. Similarly, 8 experiments used intermittent irrigation as their water regime; intermittent irrigation was presumed to be the dominant local practice in those locations. Six experiments included both continuous flooding and intermittent irrigation treatments, but stated that intermittent irrigation was the representative local water regime. The other 2 experiments also included both water regimes, but did not state which regime was dominant in local practice. It can be inferred from this result that nearly 30% of the irrigated rice fields in China are continuously flooded, and the others are intermittently irrigated. In this estimate it is assumed that 1/3 of the irrigated rice fields are continuously flooded and the remainder are intermittently irrigated.
2.5. CH4 Emission Factors and the Area of Rainfed Rice Fields
 All the above mentioned data are for irrigated rice fields, which comprise more than 90% of all Chinese rice fields. However, a small area of rainfed rice fields exists in the hilly and mountainous regions in south, central, and southwest China. Rainfed rice fields are usually located in the upper portions of landscapes, where groundwater is generally too deep to supply the crop and irrigation water is not easily available. These fields were therefore classified for this study as drought-prone rainfed fields, and the IPCC scaling factor 0.4 [IPCC, 1997] was used to calculate rainfed field CH4 emission from continuously flooded field emissions. Data on the total area of rice cultivation were taken from the China Agriculture Yearbook [Editorial Board of China Agriculture Yearbook, 1996], but the fractions of irrigated and rainfed rice area were taken from the database of Huke and Huke .
2.6. Length of Rice-Growing Season
 The length of the rice-growing period varies with region and rice season, and directly influences total CH4 emission. The early rice-growing period is generally shorter than the late rice-growing period, which, in turn, is shorter than the middle rice growing period. Also, rice in warmer regions has a shorter growing period than rice in cool regions. Because emission factors are calculated from field measurements, the length of the rice-growing season was estimated directly from field reports. The average length of the rice-growing period in the regions where only single rice can be cultivated (i.e., AEZ 5 and AEZ 8) is 130 days. In AEZ 6 and AEZ 7, the growing period of single, late, and early rice is 110, 93, and 77 days, respectively. These data were used in the estimation.
2.7. Calculation of CH4 Emission
 The following equation was applied to data from each province to calculate CH4 emission from irrigated rice fields:
where i is rice-growing region (the 5 regions classified above), j is rice season (early rice, late rice, and middle rice), k is water regime (intermittent irrigation and continuous flooding), m is organic input (‘with’ or ‘without’), EFijkm, Aijkm, Lijkm are CH4 emission factor, area under rice cultivation, and length of rice-growing period, respectively, for conditions i, j, k, and m.
 Emission factors are listed in Table 2. For AEZ 6B, an emission factor was available only for single rice, since it accounted for 98% of the total rice cultivation area. For the small area of double rice, the emission factor of single rice was used as default.
 Late rice is usually planted on the same field as early rice. However, the cultivation area of late rice was significantly larger than that of early rice in some provinces. Therefore some of the late rice was not preceded by early rice, but by upland crops or seedling beds of late rice. This was especially true in Shanghai, where farmers grow vegetables in the first half of the year, and rice in the second half of the year, on the same plot of land. In this case, the emission factor for single rice was used for the area of late rice exceeding the early rice area, per province, since CH4 flux during the rice-growing season is influenced by the water regime of the preceding season.
 Emissions from the small area of rainfed rice fields were calculated using similar methods.
2.8. Amount of CH4 Emission
 Methane emission from rice fields per province was estimated for the year 1995 (Figure 2). Total emission was 7.67 Tg CH4, of which, 1.7 Tg, 2.68 Tg, and 3.28 Tg were from early rice, single rice, and late rice, respectively. Irrigated rice fields accounted for 97% of the total emission. Hunan Province emitted more CH4 than other provinces (1.17 Tg), because it had the largest area under rice cultivation and very high CH4 fluxes [Wassmann et al., 1993b, 1996; Cai et al., 2000]. Qinghai was the only province with no rice production, and therefore no CH4 emission from rice fields. Emissions per province were not always directly related to total rice cultivation area. For example, Anhui and Jiangsu Provinces had much larger rice cultivation areas than Fujian Province, but their CH4 emissions were less than those from Fujian Province. This is partially due to the regional difference in emission factors. More importantly, however, it is because the composition of single rice and double rice (early and late) is different in these provinces. As mentioned before, emission flux from single rice fields is much less than that from double rice fields.
3. Seasonal and Spatial Distribution of CH4 Emission
3.1. Seasonal Distribution
 CH4 emission estimates were made separately for early rice, single rice, and late rice. Information required to estimate the seasonal distribution of emission was a rice calendar and the seasonal pattern of emission flux. A detailed rice calendar for each province of China, primarily based on the work of Matthews et al. , is shown in Figure 3. Two adjustments were made to the calendar. First, the calendar of Matthews et al.  includes the entire rice-growing period, from seeding to harvest, while all the field experiments only measured the emissions during the period from transplanting to harvest. A seedling period of 30–45 days was therefore subtracted from the calendar of Matthews et al. . Second, even within a province, there is some year-to-year, regional, and especially varietal variation in the rice calendar. Matthews et al.  used the “bulk” dates of planting and harvest in the calendar. To reduce the uncertainty, rice per province was divided into two parts, one growing earlier and another growing two weeks or one month later. Other minor revisions to the starting and harvest date of rice were made according to information given by the Chinese Academy of Agricultural Sciences  and Diao .
 The seasonal distribution of CH4 emission was certainly influenced by the seasonal pattern of emission flux, which, in turn, was influenced by many factors: i.e., water status before and during the rice-growing season [Cai et al., 2000], addition of inorganic and organic fertilizers [Lin et al., 2000], type of organic material [Lu et al., 2000], and rice cultivar [Shao and Li, 1996]. There were also large year-to-year and site-to-site differences [Wang et al., 2000; Yao and Chen, 1994]. Since the details of these influencing factors were not known, it was only possible to derive a generalized flux pattern by averaging the results of a variety of treatments. However, detailed flux data for each measurement day are usually not provided in publications. In this study, a detailed flux database developed within the Asia-Pacific Network for Global Change Research (APN) project, Land Use/Management Change and Trace Gas Emissions in East Asia (APN 2001-16), was used to derive generalized flux patterns for early, late, and single rice (Figure 4). The data used to derive flux patterns for early and late rice were from three sites: Changsha [Cai et al., 2000], Yingtan [Cai et al., 2000], and Hangzhou [Lu et al., 2000]; experiments at each site were conducted for at least three consecutive years. There were 27 data sets for early rice and 19 data sets for late rice, both including various treatments. The 58 data sets for single rice were from 7 sites: Suzhou [Cai et al., 2000], Nanjing [Cai et al., 2000], Hangzhou [Lu et al., 2000], Chongqin [Cai et al., 2000], Lesan [Khalil et al., 1998], and Beijing [Wang et al., 2000]. The seasonal flux pattern was expressed as a relative ratio of every measured flux to the corresponding seasonal average flux, making the seasonal flux patterns of different treatment-seasons comparable. Per rice type (early, late, single), flux patterns of different treatment-seasons and different sites were averaged per day, for a daily average flux pattern. On the basis of this, a 15-day moving average flux pattern was then derived.
 The general trends of flux patterns for early, late, and single rice differed (Figure 4). The average flux for early rice in the first 15 days was approximately 0.6 times the seasonal average; subsequently, the moving average was relatively constant, close to the seasonal average. The late rice flux peaked shortly after transplant; the average flux in the first 15 days was about twice the seasonal average. After this period, the flux decreased rapidly; in the late stage, the flux dropped close to 0. For single rice, the flux increased gradually after transplanting, peaked about 50 days later, and then gradually decreased until the end of the season. These characteristics of general flux patterns are associated with agricultural practices. For example, early rice and single rice are planted after a fallow season or an upland crop, during which time the soils are usually aerated. Accordingly, it takes time for methanogenic activity to recover. Late rice, however, is planted immediately after early rice; the field is usually drained for a short period for harvest and planting, and thus the methanogenic activity recovers very quickly. The high temperature in August may also be responsible for the peak fluxes in single rice and late rice fields.
 The estimated monthly distribution of emission is shown in Figure 5. Emission increased linearly from April to August, then sharply decreased. Approximately 1/3 of all annual emission occurred in August: approximately 0.5 Tg from single rice and almost 2.0 Tg from late rice. Total emission for June and July was about 40% of the annual emission, and single rice was the biggest contributor. There was no emission in the 4 months from December through March.
3.2. Spatial Distribution of CH4 Emission
 CH4 emission was estimated per province, and a seasonal land cover database (U.S. Geology Survey (USGS), Global land cover characteristics data base, available online at http.//edcdaac.usgs.gov/glcc/globdoc2_0.html, 2001) was used to distribute the emission within provinces. The USGS database includes 28 land cover types that may contain rice fields. Pixels with double-crop rice as the only element were given a weight factor of 2, pixels with rice rotated with other crops were assigned a weight factor of 1, and pixels with rice mixed with other crops were given a weight factor between 0 and 1. Emission hot spots were the plains of Dongting Lake in Hunan Province, Boyang Lake in Jiangxi Province, the delta region of Qiantang River in Zhejiang Province, and Sichuan Basin (Figure 6). Nearly 90% of all emission occurred between 23° N and 33° N. Figure 7 shows the spatial emission for selected months; double-crop rice pixels were given a weight of 1, because only 1 rice-season can be present at any given time. Methane emission first started in south China in April, and then moved northward. Emission could be found in June in the most northern province, but most emission occurred in southwest China, especially the Sichuan Basin. August emission primarily occurred in southeast China, the double rice region. Starting in September, emission gradually retreated from the north.
 The estimated result of 7.67 Tg CH4 yr−1 is the lowest scaling-up estimate to date. As reported by Sass et al. , studies that incorporate more regional methane emission factors and more detailed climatic and agronomic factors tend to produce lower countrywide emission estimates. The estimate produced in this study is based on region-specific emission factors and considers the effects of water management and organic inputs. All of these factors and effects were derived from a large number of field studies. Most of the studies incorporated local agricultural practices, such as water management and organic amendment, or specified local practice while comparing treatment effects. Of the 23 field sites, 6 were located in AEZ 7, 7 in AEZ 6A, 3 in AEZ 6B, 6 in AEZ 5, and 1 in AEZ 8. While the allocation of study sites partially reflects the distribution of research, it also indicates the distribution of rice cultivation areas throughout the country. Therefore the region-specific emission factors developed from these data are reasonably representative. A few more field measurements are unlikely to significantly alter the emission factors in regions other than AEZ 8 and AEZ 6B. The area of rice cultivation in AEZ 8, however, accounted for only 5% of the total area cultivated in China. In AEZ 6B, the emission factor for intermittently irrigated fields with organic input was 15.6 mg CH4 m−2 h−1, derived from one site. If another measurement for this category became available, with a value of 0.5–18.8 mg CH4 m−2 h−1 (the lowest and highest value in Table 2 for the category of intermittently irrigated with organic input), then the emission estimate for the entire country would fall between 7.27 and 7.75 Tg CH4 yr−1. Since emission factors are developed incorporating the influential factors of water regime, organic input, and rice season, they can be easily updated when new field results become available for any region.
 For some regions, scaling factors (2 or 0.5) were used to estimate emissions with organic input from that without organic input or visa versa (Table 2). If these scaling factors were adjusted to 1.5–2.7 and 0.37–0.67 (95% confidence intervals), the emissions estimate would become 7.54–7.79 Tg CH4 yr−1.
 The estimation was directly based on the 36 emission factors in Table 2 (excluding the derived ones), ranging from 0.5–32.3 mg CH4 m−2 h−1, with a mean value of 10.86 mg CH4 m−2 h−1 and a standard deviation of 7.18 mg CH4 m−2 h−1. The 95% confidence interval is from 8.45–13.28 mg CH4 m−2 h−1, approximately 22.2% variation of the mean. Therefore the estimated CH4 emission may be 5.96–9.37 Tg yr−1 due to the uncertainty in emission factors.
 The effects of intermittent irrigation and organic input seem to be more influential than emission factors in the estimation of CH4 emission. CH4 emission can vary from as low as 4.93 Tg yr−1, if all rice fields are intermittently irrigated and receive no organic fertilizer, to as high as 13.75 Tg yr−1, if all rice fields are continuously flooded and fertilized with organic materials. However, these extremes are unrealistic. On the basis of work by Cai , Huang et al. , various other reports, and personal observation, it is estimated that half of the rice fields in China receive organic fertilizer, and 2/3 of the irrigated rice fields are intermittently irrigated. These assumptions resulted in a CH4 emission estimate of 7.67 Tg for 1995. Given a total rice cultivation area of 30.7 million hectares, the average seasonal CH4 emission was 24.9 g m−2. This is also, approximately, the average of all 204 of the directly measured seasonal CH4 emissions (24.6 g m−2), reflecting the rationality of region-specific emission factors, and including the effects of water and organic fertilizer management.
 Uncertainties still exist, nevertheless, due to the lack of statistics. A 50% uncertainty is expected in the number of fields receiving organic input; the percentage of rice fields receiving organic input may be between 25% and 75%. Because intermittent irrigation is the dominant water management regime in rice cultivation in China, the percentage of intermittently irrigated rice fields should be over 50%. It probably varies, however, from 50% to 95%. As a result of these variations, CH4 emission may range from 5.82 to 9.57 Tg yr−1. To reduce the uncertainty in estimations of CH4 emission from Chinese rice fields, a countrywide survey, even by random sampling, of organic fertilizer use and water management practices is preferable to the measure of adding a few more field measurements.
 Process-based models are increasingly used to simulate trace gas emission from soils. Direct comparison of our emission estimate with modeled results is not possible due to differences in method. However, different assumptions about water regimes and organic inputs can cause discrepancies. For example, the present estimate is less than half of the modeled estimates of Cao et al. . In their model, all rice fields were assumed to be continuously flooded during the entire rice-growing season, while in the current study it was assumed that only 1/3 of the rice fields were continuously flooded. Similarly, Huang et al.  estimated CH4 emission at 9.66 Tg yr−1 using a simulation model, but cautioned that the emission might be overestimated because the effect of intermittent irrigation was not considered. Our estimate was similar to the modeled result of Matthews et al. , which was 3.35–8.64 Tg CH4 yr−1, and when organic amendment and intermittent irrigation were considered, it was 7.22–8.64 Tg CH4 yr−1. This reinforces the point that reliable information on organic fertilizer use and water management is essential for a more accurate estimation of CH4 emission from rice fields. Information on the water status of rice fields will also be essential for estimating CH4 emission from rice fields during the non-rice-growing season, emission not considered in this estimate. Cai et al.  showed that on a permanently flooded single rice field, CH4 emission during the winter season can be comparable in magnitude to emission during the rice-growing season.
 The monthly distribution of CH4 emission is influenced by the rice calendar and seasonal flux patterns. The rice calendar varies with year, cultivar, region, and even with elevation; as a result, usually only a generalized calendar can be given. The calendar of each province provided by Matthews et al.  agrees reasonably well with those given by local crop scientists [Chinese Academy of Agricultural Sciences, 1986; Diao, 1994]. Therefore the former was adopted, with some amendments from the latter. In this study, the rice season is considered to run from transplanting to harvest. All included data, and therefore the final emission estimate, were for this period. Some emission may occur in the fields before seedlings are transplanted, but this would only occur in a small percentage of the fields. This pre-rice CH4 emission begins in March. In addition, some rice fields are flooded during the winter fallow season, and could emit CH4 throughout the entire year.
 The first generalized seasonal flux patterns for early, late, and single rice has been developed in this study. These flux patterns are expected to be representative for each rice season, because they are derived from large amounts of field measurements, especially the flux pattern for single rice, which is derived from 58 individual season-treatment data sets. The seasonal distribution patterns are different from patterns based on constant daily flux. On the basis of constant daily flux, the emissions in April, September, October and November would be higher that those in the current estimation, while the emissions from May to August would be lower (Figure 7).
 Our estimated monthly distributions of emissions differ from those of Matthews et al.  and Yao et al. , in part, because their estimates assumed constant daily flux for a given rice season. More importantly, the difference is because of the emission factors used in estimating emissions from different rice seasons. In the pioneer study of Matthews et al. , a constant daily flux was assumed for all rice seasons, due to a lack of field measurements. As shown by Yao et al. , emission from single rice fields accounted for 2/3 of the total emission; accordingly, emission was highest in June. This single rice result was estimated from 4 field measurements, one of which had an extremely high flux value later shown to be an overestimate [Khalil et al., 1998]. In the current study, emission from the single rice season was estimated using results from 17 measurement sites. The average flux is much lower than that for early rice or late rice; emission from the single rice season accounts for only 35% total emission. As a result, even disregarding the within-season flux variation, August emission would be the highest (Figure 7).
 The geographical distribution of emission is determined by the geographical distribution of rice fields and by the regional emission factors. There is less uncertainty in the geographical distribution of rice fields, First because data on the area of rice cultivation can be collected at provincial or county levels, and Second because the location of rice fields is fairly well known. Any improvement in the accuracy of the region-specific emission factors would improve the geographical distribution estimates.
 Through the analysis of a large number of field measurements from China, it was shown that, on average, addition of organic fertilizer can double CH4 emission from rice fields; CH4 emission from intermittently irrigated fields is approximately 53% of that from continuously flooded fields. Crop rotation greatly affects CH4 emission from rice fields; the average CH4 flux from late rice fields is 1.58 times greater than flux from early rice, and 2.28 times greater than that from single rice. Regional emission factors decrease from south to north. Estimated total CH4 emission for 1995 is 7.67 Tg; this estimate can range from 5.82 Tg to 9.67 Tg, due to uncertainties in the assumptions on areas receiving organic input and intermittent irrigation. Further improvement of emission estimates is dependent on better availability of this basic information, rather than on availability of more field measurements. Generalized seasonal flux patterns were developed for early, late, and single rice fields, and these flux patterns were used to estimate the monthly distribution of emission. Approximately 1/3 of the annual emission occurs in August, and about 40% in June and July. Geographically, most emission is from the area around 30°N, especially Tingtong Lake Plain, the Sichuan Basin, and the delta region of the Qiantang River.
 We would like to thank those who provided data to the APN project: especially Dr. Z. Wang from the Institute of Crop Breeding and Cultivation, Chinese Academy of Agricultural Sciences; Dr. W. Lu from the China National Rice Research Institute; and Dr. X. Zhen from the Institute of Atmospheric Physics, Chinese Academy of Sciences. The flux database is available on CD-ROM upon request to Prof. Zucong Cai, Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China.