Introduction
Ammonia (NH3) is an important reactive nitrogen (N) compound, and has
wide impacts on both atmospheric chemistry and ecosystems. As an alkaline
gas in the atmosphere, it can readily neutralize both sulfate and nitric
acid to form ammonium sulfate and ammonium nitrate, which are the major
constituents of secondary inorganic aerosols (Behera and Sharma,
2012). Kirkby et al. (2011) found that atmospheric NH3 could
substantially accelerate the nucleation of sulfuric acid particles, thereby
contributing to the formation of cloud condensation nuclei. The total mass
of secondary ammonium salts accounts for 25–60 % of particulate matter
less than or equal to 2.5 µm in aerodynamic diameter (PM2.5)
(Ianniello et al., 2011; He et al., 2001; Fang et al., 2009), and large
amounts of this fine PM not only cause air pollution but also have a
significant effect on radiative forcing (Charlson et al., 1992; Martin et
al., 2004). In addition, the sinking of NH3 into terrestrial and
aquatic ecosystems can directly or indirectly cause severe environmental
issues, such as soil acidification, eutrophication of water bodies, and even
a decrease in biological diversity (Matson et al., 2002; Pearson and
Stewart, 1993). When deposited into soils, NH3 compounds can be
converted into nitrate (NO3-) through nitrification,
simultaneously releasing protons into the soil, resulting in soil
acidification (Krupa, 2003).
Livestock manure and synthetic fertilizer represent the most important
sources of NH3 emissions, jointly accounting for more than 57 % of
global emissions and more than 80 % of total emissions in Asia (Bouwman
et al., 1997; Streets et al., 2003; Zhao and Wang, 1994). Previous studies
have verified that China emits a considerable proportion of the total global
NH3 emissions budget due to its intensive agricultural activities
(Streets et al., 2003). A major agricultural country, China has undergone
rapid industrialization and urbanization since the Chinese government
implemented its economic reform in 1978. The rapid economic development and
rise in living standards over the last 30 years has resulted in a sharp
increase in grain output and meat production. The use of synthetic
fertilizers, which are applied by Chinese farmers to promote the growth of
crops, has also undergone a considerable, sustained increase. According to
figures from the International Fertilizer Industry Association
(Zhang et al., 2012), synthetic fertilizer production has
increased 3-fold during the past 3 decades, from 10 million tons in 1980
to 43 million tons in 2012. Several factors have contributed to the dramatic
rise in the use of synthetic fertilizers. First, their use grew dramatically
in the latter half of the 20th century in most parts of the world, as
farmers increasingly expected to achieve higher crop yields. Second, N
over-fertilization has been common, resulting in higher NH3
volatilization loss, especially in the North China Plain and Taihu region
(Xiong et al., 2008; Ju et al., 2009). In addition, due to farmers'
increasing labor costs and income from off-farm activities, traditional
farmyard manure has been gradually eliminated in much of China and replaced
by synthetic fertilizers (Ma et al., 2009; Zhang et al., 2011). As a
consequence, the surge in NH3 emissions from synthetic fertilizer
application during this period has been inevitable. Meanwhile, since 1980,
when China began developing a series of policies to support livestock
production, the industry has undergone rapid growth driven by the increasing
demand for beef, pork, mutton, milk, and wool (Zhou et al.,
2007). For example, in 2006, there were 56 million slaughtered
cattle in China, showing a 16-fold increase from the number in 1980; the
number of poultry in production increased 12-fold during this same period
(EOCAIY, 2007). The flourishing livestock industry has produced large
volumes of manure that releases gaseous NH3 through N hydrolyzation and
volatilization. In conclusion, a marked increase in NH3 emissions from
livestock manure and synthetic fertilizer are expected from 1980 to the
present, but specific data on annual emissions and variation in emissions
are lacking.
Changes induced by anthropogenic activities can significantly influence the
global N cycle (Vitousek et al., 1997). Therefore, to better understand
the evolution of the global N budget and the impacts on living systems, it
is essential to quantify NH3 emissions during recent decades in China.
Moreover, the compilation of multi-year regional and national NH3
emissions inventories would also help elucidate the causes of severe air
pollution in China.
In a previous study, we developed a comprehensive NH3 inventory for
2006 to show the monthly variation and spatial distribution of NH3
emissions in China based on a bottom-up method (Huang et al., 2012b). Our
method had several advantages over previous inventories. First, emissions
factors (EFs) characterized by ambient temperature, soil acidity, and other
crucial influences based on typical local agricultural practices were used
to parameterize NH3 volatilization from synthetic fertilizer and animal
manure. In addition, we included as many different types of emission sources
as possible, such as vehicle exhaust and waste disposal. Our NH3
emissions inventory was compared with some recent studies to show its
reliability. Paulot et al. (2014) used the adjoint of a
global chemical transport model (GEOS-Chem) to optimize NH3 emissions
estimation in China; the results were similar to our previous study
(Huang et al., 2012b). In addition, the distribution of the total
NH3 column in eastern Asia retrieved from measurements of the Infrared
Atmospheric Sounding Interferometer (IASI) aboard the European
METeorological OPerational (MetOp) polar orbiting satellites (Van Damme
et al., 2014) was also in agreement with the spatial pattern of NH3
emissions calculated in our previous study. However, there were still some
problems in this method; for example, Huang et al. (2012b) generally
adopted EFs reported in early years and an up-to-date in situ measurement was
needed; moreover, wind speed that could be of importance in emission
estimation was not considered in that study. Though Huang et al. (2012b)
have involved as many NH3 emitters as possible in the inventory, some
minor sources may be neglected like fertilization in orchard, NH3
escape from thermal power plants. Nevertheless, this bottom-up emission
inventory appears to be reliable, and the method can be used to estimate
NH3 emissions in China.
In this study, we mainly focused on compiling long-term emission
inventories based on the method by Huang et al. (2012b). Some improvements to this method
have been made and sources of NH3 in our inventories were listed as
follow: (1) farmland ecosystems (synthetic fertilizer application, soil and N
fixing, and crop residue compost); (2) livestock waste; (3) biomass burning
(forest and grassland fires, crop residue burning, and fuelwood combustion);
and (4) other sources (excrement waste from rural populations, the chemical
industry, waste disposal, NH3 escape from thermal power plants, and
traffic sources). The interannual variation and spatial patterns of NH3
emissions from 1980 to 2012 on the Chinese mainland (excluding Hong Kong,
Macao, and Taiwan) are discussed in this paper.
Methods and data
NH3 emissions were calculated as a product of the activity data and
corresponding condition-specific EFs, according to the following equation:
E(NH3)=∑i∑p∑m(Ai,p,m×EFi,p,m),
where E(NH3) is the total NH3 emissions; i, p, and m represent the
source type, the province in China, and the month, respectively;
Ai,p,m is the activity data of a specific condition;
and EFi,p,m is the corresponding EF. The emissions
were allocated to each 1 km × 1 km spatial resolution on the basis
of land cover, rural population, and other proxies. Further details on the
estimation methods and gridded allocation of the various sources are
presented in Huang et al. (2012b).
Synthetic fertilizer application
NH3 volatilization from synthetic fertilizers represents an important
pathway of N release from the soil, resulting in large losses of soil and
plant N (Harrison and Webb, 2001). We classified the synthetic fertilizers
used in Chinese agriculture as urea, ammonium bicarbonate (ABC), ammonium
nitrate (AN), ammonium sulfate (AS), and others (including calcium ammonium
nitrate, ammonium chloride, and ammonium phosphates). NH3 emissions
were estimated by multiplying gridded (1 km × 1 km) EFs for five
types of fertilizer and consumption, which was calculated as the product of
cultivated area and the application rate to crops (EOCAY,
1981–2013; Zhang et al., 2012; NBSC, 2003–2013a). A crop calendar, which
involves the type of crop cultivated at a specific region and corresponding
fertilization timing was used to identify the monthly fertilizer
application. We considered 16 kinds of crops that are wildly cultivated in
different seasons in China, including early rice, semi-late rice, late rice,
non-glutinous rice, wheat, maize, bean, potato, peanut, oil crop, cotton,
beet, sugarcane, tobacco, vegetables and fruits. We derived monthly
condition-specific EFs for synthetic fertilizer volatilization by
introducing several influencing factors like the type of fertilizer, soil
pH, ambient temperature, fertilization method, and application rate (see
Table 2 in Huang et al., 2012b). Briefly, EFs were characterized by
fertilizer types with ABC and urea more volatile than the other fertilizers.
Liner relationships between the volatilization of mineral fertilizers and
soil pH were developed to correct EFs (Fan et al., 2005; Bouwman et al.,
2002). A threshold of 200 kg N ha-1 was defined as the high
fertilization rate and when the local fertilization rate exceeded this
value, we multiplied EFs by 1.18 (Fan et al., 2006). We
derived the relationship between the emission rate and temperature for
various fertilizers from EEA (2009) and Lv et al. (1980).
Compared to Huang et al. (2012b) the effects of wind speed and in situ
measurements of NH3 flux conducted by our research group in a typical
cropland were involved to further refine the EFs for synthetic fertilizer
emissions in this study.
In situ measurement
For acquiring up-to-date EFs that could reflect NH3 volatilization from
synthetic fertilizer application in present Chinese agricultural practice,
we measured NH3 EF by using micrometeorological method for a whole year
in a typical farmland in the North China Plain and an inverse dispersion
model was also used to derive the ammonia EFs (Huo et al., 2014, 2015).
The in situ results could represent better than those used in Huang et al. (2012)
which were derived from studies in early years. The soil pH and mean air
temperature in this farmland was 8.2 and 15 ∘C, respectively. The
measurement yielded an NH3 EF for urea of 12 % ± 3 % in this
case. Huang et al. (2012b) develop a linear relationship between NH3
volatilization and soil pH to involve the impact of soil acidity on EFs
according to Cai et al. (1986) and Zhu et
al. (1989). We applied the condition-specific EF we measured recently to
refine this relationship with linear regression analysis.
Wind speed
In addition to temperature, wind speed is a meteorological parameter that
affects the partial pressure of NH3 by regulating the exchange of
NH3 between the soil and/or floodwater and the air, thereby influencing
NH3 volatilization (Bouwman et al., 2002). Several previous studies
have shown that high winds significantly influence NH3 volatilization
(Denmead et al., 1982; Fillery et al., 1984; Freney et al., 1985). We
followed the approach of Gyldenkaerne et al. (2005) to introduce
the effects of wind speed on NH3 volatilization from synthetic
fertilizer application. The original EFs were multiplied by a factor that
was an exponential function of wind speed. Both the monthly average wind
speed and ambient temperature mentioned above for a 1 km × 1 km
grid were based on the final analysis data set of the National Centers for
Environmental Prediction (NCEP). It should be noted that in this study, we
used mean monthly weather values in the adjustment of EFs rather than the
daily maximum since the daily activity data were not available (we could not
quantify the synthetic fertilizer use each day) or we could not identify the
exact date of fertilizer application and the timing varied annually. On the
other hand, we adopted the parameterization of temperature adjustment
provided by EEA (2009), which is also based on mean temperature. Despite the
uncertainties, we still used mean monthly temperature and wind speed to
produce monthly inventories.
Livestock waste
A mass-flow approach has been widely used to estimate NH3 emissions
from livestock waste (Beusen et al., 2008; Velthof et al., 2012).
Ammoniacal N (TAN) produced from livestock waste can be converted into
gaseous NH3 or lost through other pathways during different process of
manure management (Webb and Misselbrook, 2004; Webb et al., 2006). In
this study, TAN inputted into manure management was the product of the daily
amount of urine and faeces produced (kg (day capita)-1), N content (%), and
TAN content (%) (see Table 3 in Huang et al., 2012b). We assumed
that these parameters have not changed during the 30-year period and some
uncertainties from this assumption would be discussed in Sect. 3.5. We
estimated livestock emissions by multiplying TAN at four different stages of
manure management: outdoor, housing, manure storage, and manure spreading
onto farmland (Pain et al., 1998) with the corresponding
EFs. In the outdoor stage, the excreta were directly deposited in the open
air without any treatment after that while animals' excreta inside buildings
would release emissions during housing, storage and spreading stages. The
periods spent in buildings in a year for different livestock classes were
used to determine the portion of excrement indoors or outdoors. After a
proportion of TAN was depleted through some processes like immobilization,
discharge of NH3, N2O and N2, and the leaching loss of
nitrogen, the rest TAN would flow into next stage (EEA, 2013). In
addition, we also considered three main animal-rearing systems in China:
free-range, intensive, and grazing. The first two systems are extensively
implemented in most rural areas of the country. The free-range system is
characterized by small-scale rearing belonging to individual families and
has been rapidly developed over recent decades (http://www.caaa.cn/). Based
on animal husbandry yearbooks, we defined an intensive rearing system as
that where the number of a single livestock class on a single farm (except
grazing) was larger than a certain value (Table S1 in the Supplement). Under this definition,
an interannual ratio between the free-range system and the intensive one was
introduced to reflect the change of animal rearing types in the inventory
periods. It could represent the changes of Chinese livestock practice in the
inventory period to some degree.
The number of livestock in each class from 1980 to 2012 was provided by
official statistical data and husbandry industry reports (EOCAIY,
1999–2013; EOCAY, 1981–2013). We mainly adopted the EFs in each stage for
different livestock classes that are listed in Table S2 in Huang et al. (2012b). Temperature-dependent volatilization rates were considered by using
specific EFs at different temperature intervals in the manure housing stage
(Koerkamp et al., 1998). We also implemented wind speed and temperature
adjustment in the stages of manure spreading and grazing, based on model
results reported by Gyldenkaerne et al. (2005). Ambient
temperature and wind speed data were extracted from the NCEP final analysis
data set.
Other sources
Activity data set and EFs of other minor NH3 sources used in
our study.
Sources
Activity data set
EFs
Reference
Nitrogen-fixing plants
EOCAY (1981–2013)
0.01 kg NH3 Kg-1 N
EEA (2006)
Compost of crop residues
EOCAY (1981–2013)
0.32 kg NH3 ton-1
Stephen et al. (2004)
Biomass burning
Forest fires
MODIS Burned Area (2000–2012) (Roy et al., 2008), CMF (1990), CMF (1989–1998) and SFA (1999–2000)
1.1 g NH3 kg-1
Andreae and Merlet (2001)
Grassland fires
MODIS Burned Area (2000–2012) (Roy et al., 2008),
0.7 g NH3 kg-1
Seiler and Crutzen (1980)
Crop residues burning
EOCAY (1981–2013) and NBSC (1985–2013)
0.37 (wheat) g NH3 kg-1 0.68 (maize) 0.52 (others)
Li et al. (2007)
Fuelwood combustion
NBSC (1985–2013)
1.3 g NH3 kg-1
Andreae and Merlet (2001)
Human excrement
NBSC (1981–2013b) and NBSC (2003–2013b)
0.787 kg NH3 year-1 cap-1
Buijsman et al. (1987), Moller and Schieferdecker (1989), EPBG (2005)
Chemical industry
Synthetic ammonia
NBSC (1981–2013a)
0.01 kg NH3 ton-1
EEA (2013)
N fertilizers production
NBSC (1981–2013a)
5 kg NH3 ton-1
Stephen et al. (2004)
Waste disposal
Wastewater
NBSC (2003–2013b),
0.003 kg NH3 m-3
EPBG (2005)
Landfill
Du et al. (2006)
0.560 kg NH3 ton-1
Stephen et al. (2004)
Compost
1.275 kg NH3 ton-1
Stephen et al. (2004)
Incineration
0.210 kg NH3 ton-1
Sutton et al. (2000)
Traffic
Light-duty gasoline vehicles
CAAM (1983–2013)
0.023 g NH3 km-1
Liu et al. (2014)
Heavy-duty gasoline vehicles
CAAM (1983–2013)
0.028 g NH3 km-1
Stephen et al. (2004)
Light-duty diesel vehicles
CAAM (1983–2013)
0.04 g NH3 km-1
Stephen et al. (2004)
Heavy-duty diesel vehicles
CAAM (1983–2013)
0.017 g NH3 km-1
Stephen et al. (2004)
Motorcycles
CAAM (1983–2013)
0.007 g NH3 km-1
Stephen et al. (2004)
Ammonia escape
CAEPI (2013)
2.3 mg m-3
NEA (2011)
The other minor NH3 emission sources included agricultural soil,
N-fixing plants, the compost of crop residues, biomass burning, excrement
waste from rural populations, the chemical industry, waste disposal, traffic
sources, and NH3 escape from thermal power plants. The data sources
and EFs for each source type are summarized in Table 1. Further details on
the estimation methods appear in Huang et al. (2012b). NH3 escape,
which was not included in previous inventories, is a new source of NH3
emissions that has emerged during the past 10 years in China, and refers to
the NH3 derived from the incomplete reactions of NH3 additives
used in NOx abatement in thermal power plants (EEA, 2013). We roughly
estimated the amount of NH3 escape by multiplying the total flue gas
released in power plants nationwide by the maximum allowable concentration
of NH3 carried in flue gas (NEA, 2011; CAEPI, 2013).
Monthly emissions
The seasonal NH3 emission estimation for fertilizer application could
be calculated as the product of condition-specific EFs derived from
meteorological factors (average monthly temperature and wind speed) and
monthly fertilizer consumption associated with agricultural timing. For
livestock emissions, we assumed that the number of each livestock category
per month remains constant, because the monthly fluctuation in the
production of meat, eggs and milk is very small (http://www.caaa.cn/). The
monthly EFs were distinguished by average monthly temperature and wind speed
from NCEP. Besides, the emission from biomass burning also shows a temporal
fluctuation. MCD45A1 (MODIS monthly burned area product), MOD14A2 and MYD14A2
products (8-day thermal anomalies/fire products) were utilized to ascertain
the timing of different kinds of biomass. For other minor sources, the
emissions were equally divided into 12 months.
Results and discussion
Annual NH3 emissions
Interannual variation in total NH3 emissions in China from
1980 to 2012; the sources of the emissions were categorized as synthetic
fertilizer application, livestock manure, and other sources.
Over the past 30 years, China has undergone dramatic changes and significant
economic development, and NH3 emissions have changed correspondingly.
Figure 1 illustrates the trends in total NH3 emissions, which are
divided into fertilizer application, livestock waste, and other minor
sources. Total emissions increased from 5.9 to 11.1 Tg between 1980 and
1996, then decreased to 9.7 Tg in 2012. The most important contributor was
livestock manure management, accounting for approximately 50 % of the
total budget. Due to the extremely high consumption and high volatility of
ABC and urea, synthetic fertilizer application was responsible for
30–43 % of the total emissions, second only to livestock manure. However,
in Europe and the United States, where less-volatile synthetic fertilizers
such as AN and AS are more popular (Bouwman and VanderHoek,
1997), livestock manure overwhelmingly dominates the NH3 emissions
inventory (Ferm, 1998). These two primary sources combined accounted
for 80–90 % of the total emissions budget, with other minor sources
accordingly accounting for 10–20 %.
Emissions from livestock waste
Interannual variation in NH3 emissions from livestock manure
for three different rearing systems.
Livestock waste was the largest source of NH3 emissions in China from
1980 to 2012, contributing approximately 50 % of total emissions each
year. Since the 1980s, rapid economic development in China has driven the
large increment of livestock production. The total number of the major
livestock animals, namely, beef cattle, sheep, pigs, and poultry, increased
from approximately 70 to 140 million, 180 to 370 million, 420 to 1400 million, and 0.9 to 10 billion respectively, from the 1980s to mid-2000s
(Fig. S1 in the Supplement). In this period, large quantities of NH3 derived from
livestock waste have been emitted into the atmosphere. As shown in Fig. 2,
emissions increased from 2.9 Tg in 1980 to 6.2 Tg in 2005, more than
doubling during this period, and then decreased to 5.0 Tg in 2012. We
divided livestock NH3 emissions from 1980 to 2012 into four phases. In
the first phase (1980–1990), emissions steadily increased with a mean
growth rate of approximately 3 %. Free-range production contributed most
of the emissions (the population of free-range animals represented more than
90 % of the major livestock animals (EOCAY, 1991). The second phase
(1991–1996) saw the most rapid increase in emissions, and the growth rate
rose to 10 % between the years 1994 and 1995. In 1992, China began to
implement a reform of the socialist market economic system, which had
previously driven livestock production (CAAA, 2009), and accordingly,
more NH3 emissions from livestock waste were emitted. However, in 1997,
there was a 0.5 Tg decrease in livestock emissions, compared to the those in
1996. This observed decline could be attributed to the Asian financial
crisis, which started in 1997 and had a detrimental effect on the
development of the Chinese livestock industry. From 1998 to 2005, as the
third phase, NH3 emissions continually rose in conjunction with an
increase in livestock production due to improvements in cultivation
technique and increases in market demand (Zhang et al., 2003).
Compared to the first two phases, the contribution of intensive rearing
systems to total emissions also increased, with the population of
intensively reared animals representing nearly 20 % of the major livestock
animals (EOCAIY, 2006), because the Chinese government encouraged
large-scale intensive methods for livestock production to gradually replace
traditional free-range systems (CAAA, 2009). After a peak in 2005, there
was a marked decrease, and emissions fluctuated around 5.0 Tg in the fourth
phase, significantly lower than those in the mid-2000s, which can be
explained by a decrease in several major livestock classes, including cattle
and sheep (Fig. S1). Multiple factors inhibited the development of the
livestock industry, and thus reduced NH3 emissions from 2007 to 2012.
These included a rural labor shortage, increased feeding costs for farmers,
and market price fluctuations of meat products (Pu et al., 2008).
Moreover, it should be noted that the class-specific proportions of
intensively reared animals for beef cattle, pigs, and laying hens
significantly increased to approximately 30, 40, and 70 % in 2012,
respectively, which partly accounted for the reduced livestock emissions due
to the lower NH3 EFs of the intensive system compared to the free-range
(EEA, 2013). In contrast to the free-range and intensive systems, in
recent decades NH3 emissions from grazing systems have demonstrated
slight growth, from 0.13 Tg (1980) to 0.20 Tg (2012), without significant
changes.
Table S2 presents the interannual emissions of the typical livestock
categories. Among them, beef cattle were consistently the largest NH3 emitter, contributing to an annual mean of 1.9 Tg NH3; pigs, laying hens,
goats, and sheep were also major contributors to the total emissions in this
period. Poultry had the most rapid growth rate in NH3 emissions.
Nevertheless, a marked downward trend has appeared since 2007 for several
major NH3 sources, including beef cattle, goats, and sheep, which led
to the decrease in total emissions discussed above.
Emissions from synthetic fertilizer application
Interannual variation in NH3 emissions from synthetic
fertilizer in China from 1980 to 2012; types of synthetic fertilizer were
categorized as urea, ABC, and others (AN, AS, and others).
Figure 3 shows the estimations of NH3 emissions from synthetic
fertilizer application for the period 1980–2012. Annual levels consistently
increased from 1980 (2.1 Tg) to 1996 (4.7 Tg), and then declined from 1996
to 2012 (2.8 Tg). ABC and urea were the major sources; NH3 release from
other synthetic fertilizers, such as AS and AN, made a negligible
contribution to emissions (< 0.1 %). In general, the interannual
variation in emissions reflects the changes in farming practices in China.
First, the relative contribution of urea and ABC has gradually changed over
recent decades. During the 1980s, ABC represented a substantial fraction of
synthetic fertilizers used in China, and because of its high volatilization
(Zhu et al., 1989), emissions from this kind of chemical
fertilizer dominated in this period. However, ABC was inefficient for crop
production because of the low N content (17 % N) and high N loss. In the
mid-1990s, China introduced the technology of urea production, which
resulted in widespread application (Zhang et al., 2012). Urea,
characterized by high N concentration (46 % N), has gradually replaced ABC
and become the dominant chemical fertilizer used in cropland over the last
20 years. In 1980, 3.0 million and 5.1 million tons of urea and ABC,
respectively, were produced; by 2012, these values had changed to
approximately 28.8 million and 3.4 million tons, accounting for
approximately 66.7 and 7.9 % of total synthetic fertilizer production
in China, respectively (Fig. S2). Because NH3 volatilization from ABC
is more than 2-fold that from urea (Roelcke et al., 2002; Cai et al.,
1986), the increasing proportion of urea application relative to that of ABC
has caused the decrease in total synthetic fertilizer emissions observed
from the mid-1990s onwards. As shown in Fig. 3, although emissions from
urea application increased by 1.0 Tg from 1996 to 2012, those from ABC fell
by nearly 3.0 Tg. In addition, there have been seasonal disparities in
NH3 emissions from synthetic fertilizers, caused by variation in both
the temperature and timing of fertilizer application for different crops.
Generally, NH3 volatilization began to rise in April with increasing
temperatures, and the highest emissions occurred in summer (June–August),
which can be attributed to the high temperatures and intensive application
of fertilizer. The seasonal distribution of synthetic fertilizer emissions
was almost constant from 1980 to 2012, corresponding to stability in the
seasonal distribution of agricultural activities.
Source apportionments
Contributions to NH3 emissions (Gg) from various sources
from 1980 to 2012.
Synthetic
Agricultural
N-fixing
Compost
Livestock
Biomass
Human
Chemical
Waste
Traffic
Ammonia
total
fertilizer
soil
crop
burning
excrement
industry
disposal
escape
1980
2103
175
20
42
2862
214
362
61
5
7
5851
1981
2077
175
19
43
2888
214
368
60
5
8
5858
1982
2368
175
20
48
3010
220
375
62
6
8
6290
1983
2616
175
18
51
3028
219
383
68
7
9
6574
1984
2812
175
17
54
3111
223
389
74
7
10
6872
1985
2686
175
18
52
3257
218
397
70
8
12
6893
1986
2880
175
18
54
3403
226
405
71
7
14
7252
1987
3015
175
18
57
3509
267
413
82
8
16
7559
1988
3349
174
17
56
3693
231
420
83
9
18
8050
1989
3562
174
17
57
3799
224
430
87
10
20
8381
1990
3474
174
17
63
3872
234
432
89
17
21
8395
1991
3861
174
16
63
3908
234
435
92
28
23
8835
1992
3808
174
16
63
4011
234
438
96
36
27
8902
1993
3803
173
18
66
4259
237
442
93
43
32
9166
1994
4007
173
19
65
4672
236
424
106
45
37
9783
1995
4329
173
17
67
5170
242
404
113
57
40
10 613
1996
4720
174
15
74
5330
255
377
130
60
43
11 177
1997
4528
174
16
72
4844
246
353
126
69
47
10 476
1998
4391
174
16
76
5055
255
327
132
75
51
10 553
1999
4331
174
19
76
5120
257
309
139
80
56
10 562
2000
3797
237
21
76
5349
249
283
146
96
63
0.02
10 317
2001
3835
237
22
69
5391
278
271
154
92
53
0.04
10 403
2002
3957
237
21
69
5527
308
269
171
97
81
0.07
10 738
2003
3692
237
22
65
5783
310
253
173
103
94
0.09
10 733
2004
3683
237
21
70
5970
324
234
203
111
105
0.12
10 958
2005
3492
237
21
76
6159
303
209
232
110
122
0.12
10 962
2006
3319
237
21
77
5867
313
200
238
113
160
0.29
10 545
2007
3258
222
19
79
4992
305
195
258
128
165
0.60
9621
2008
3105
221
20
79
5024
306
185
264
140
191
0.96
9536
2009
3244
221
20
84
5202
315
169
277
152
231
1.77
9917
2010
2967
221
20
85
5104
309
182
271
168
284
3.14
9654
2011
2804
221
19
91
4928
326
131
296
186
335
4.98
9342
2012
2811
221
18
95
5026
332
121
308
268
388
86.63
9674
Source contributions (%) to NH3 emissions in China: (a) 1980; (b) 1996; (c) 2006; (d) 2012.
Table 2 lists NH3 emissions at the national level from 1980 to 2012
from various sources. Other sources (except synthetic fertilizer and
livestock) made no notable contribution to the total budget due to their
relatively low levels; nevertheless, some of these sources exhibited
distinct variation during this period. For example, NH3 emissions
released by biomass burning generally increased, of which crop-residue
burning and housing fuelwood combustion jointly accounted for a large
proportion of the biomass burning emissions. Particularly, in 1987, a large
forest fire occurring in the Greater Khingan Mountains, located in
Heilongjiang Province, released more than 10 Gg NH3. NH3 escape
derived from the denitrification process in thermal power plants increased
substantially, as the implementation of flue gas denitrification has rapidly
increased in the past 10 years, especially in 2012. NH3 emissions from
human excrement decreased from 362 Gg (6.2 % of total) to 121 Gg (1.3 %
of total), due to rural depopulation and improvements in sanitary conditions
over the past 3 decades. The contributions of different sources to total
NH3 emissions in 1980, 1996, 2006, and 2012 are illustrated in Fig. 4.
Emissions from livestock and synthetic fertilizer dominated the total
inventories. Specifically, the proportion of emissions from synthetic
fertilizers of total emissions peaked at 42.3 % in 1996, and then started
to decrease in the following years, which can be attributed to changes in
the types of fertilizer used. Furthermore, as mentioned above, the Asian
financial crisis and the resulting depression in the livestock market led to
a decline in the proportion of emissions from livestock after 1997 and 2006,
respectively. The contributions of the traffic, chemical industry, waste
disposal and NH3 escape from thermal power plants reached the peak
values of 4.0, 3.2, 2.8 and 0.9 %, respectively, in 2012.
Spatial distribution of ammonia emissions
Spatial distribution of NH3 emissions in China in 1980, 1990,
2000, and 2012 (from left to right: total emissions, synthetic fertilizer
emissions, and livestock emissions).
Figure 5 displays the spatial patterns of NH3 emissions in 1980, 1990,
2000, and 2012, respectively. Over recent decades, high emission rates of
greater than 2000 kg km-2 were always concentrated in Hebei, Shandong,
Henan, Jiangsu, Anhui, and East Sichuan provinces, which form the major
areas of intensive agriculture in China. The sum of emissions from these
provinces contributed approximately 40 % of national NH3 emissions
annually from 1980 to 2012. Emissions rates in northeastern China,
consisting of the Liaoning, Jilin, and Heilongjiang provinces, another major
grain-producing area that is also important for cattle breeding, showed an
increasing trend from the 1980s to the 2000s. Again, synthetic fertilizers
and livestock waste dominated the spatial distribution of the total
emissions.
As mentioned above, NH3 volatilization from synthetic fertilizer
application initially increased rapidly, reaching its peak value in the
mid-1990s, and then consistently decreased. This decadal pattern can be
observed in the temporal-spatial distribution of NH3 emissions from
fertilizers (middle column in Fig. 5). High emission rates, with more than
2000 kg km-2 released in both 1980 and 1990, occurred mainly in
Shandong, Henan, and Jiangsu provinces, where farmers consistently
over-applied synthetic fertilizers (Richter and Roelcke, 2000), as
well as in Sichuan, which ranked first in ABC application among all
provinces in the 1980s. In 2000, NH3 emission rates from cultivated
land in Shandong, Henan, Anhui, and Jiangsu provinces (the North China
Plain) generally exceeded 3000 kg km-2, but decreased to less than
2000 kg km-2 in 2012 due to a reduction in the use of ABC. The total
usage of ABC fertilizer in these provinces in 2000 was 5-fold that in 2012.
Hubei, Hunan, Jiangxi, and Guangdong provinces, covering China's major
rice-production areas, displayed significant growth in NH3
volatilization from 1980 to 1990, with emissions approximately doubling;
however, emissions reduced after the mid-1990s possibly because of the
transition of fertilizer usage from ABC to urea in these provinces. In
contrast to the variation observed in the areas mentioned above, NH3
emissions in the Northeast Plain encompassing Jilin, Heilongjiang, and Inner
Mongolia, and in Xinjiang Province, have consistently increased in recent
decades. From the 1990s onwards, grain production in the Northeast Plain
entered a rapid growth period, accompanied by an increasing demand for
synthetic fertilizers.
The spatial distribution of NH3 emissions from livestock waste was
similar to that from synthetic fertilizers, with high emission rates in
eastern China, East Sichuan, and parts of Xinjiang. In the 1980s, Sichuan
was the largest emitter among all of the provinces, accounting for more than
10 % of emissions from livestock manure management, followed by Inner
Mongolia and Henan. More than half of NH3 emissions from livestock in
Sichuan originated from cattle rearing. In Henan, both cattle and goats
played significant roles in the NH3 emissions, whereas in Inner
Mongolia, Qinghai, and Xinjiang, large numbers of sheep were raised and were
responsible for 33, 31, and 42 % of livestock emissions in 1980,
respectively. From the 1980s to 1990s, emissions in the North China Plain
showed more rapid growth than in other areas in China, and almost doubled
during this period in Shandong, Hebei, and Anhui, where the contribution of
beef cattle, pigs, and poultry increased significantly. Until the 2000s, the
North China Plain was the area of highest NH3 emissions, with levels of
3000 kg km-2 throughout most of Hebei, Shandong, and Henan provinces.
The two largest contributors to livestock emissions in these three provinces
were beef cattle and laying hens, which contributed 38 and 19 % in
2000, respectively. Beef cattle and goats were extensively bred in Henan,
Shandong, Sichuan, and Hebei provinces, and in 2012 the decrease in their
population caused a corresponding decrease in NH3 emissions from
livestock manure in these provinces, by approximately 0.14, 0.14, 0.02, and 0.09 Tg, respectively. Emissions from grazing rearing system were
less significant nationally than those from other systems (free-range and
intensive), but they did become important in northern Inner Mongolia,
central and southern Xinjiang, west-central Qinghai, western Sichuan, and
large areas of Tibet.
Monthly variation in ammonia emissions
Monthly distribution of NH3 emissions from different sources
in China: (a) all the sources; (b) synthetic fertilizer; (c) livestock
wastes; (d) crop burning in fields; (e) forest and grass fires.
The monthly variation in NH3 emissions in 1980, 1990, 2000, and 2012
are clearly presented in Fig. 6a. The emissions were primarily concentrated
during April to September due to the intensive agricultural activities and
higher temperatures. Specifically, the different sources showed the diverse
distribution characteristics.
Figure 6b describes the temporal distributions of NH3 emissions from
synthetic fertilizer application in 1980, 1990, 2000, and 2012,
respectively. It is obvious that the monthly emissions from fertilizer
exhibited similar seasonal distribution among different years. Generally,
the largest emissions occurred in summer (June to August), accounting
for 44.8–47.7 % of annual emissions from synthetic fertilizer, which are
attributed to denser fertilization and higher temperatures during this time.
Conversely, because of the less NH3 volatilization related to lower
temperatures and relatively rare cultivation during the winter (December to
February), the NH3 emissions reduced to 7.7–11.1 % of annual
fertilizer emissions. In China, the new spring seeding begins in April and
is accompanied by corresponding fertilizer application. In the following
1–2 months, due to application of top fertilizer and warming temperatures,
particularly in eastern and central provinces such as Jiangsu, Anhui and
Henan, the NH3 emissions continuously increase to August. In the North
China Plain, the winter wheat–summer maize rotation system has been practiced as
a characteristic farming practice. The high emission rates in June and
August could be attributed to the basal dressing and top dressing of summer
plants, such as maize. From autumn on, most of the crops begin to harvest,
which lead to the decline of emissions during this time. In particular,
winter wheat is usually seeded in September with the application of basal
dressing, and the top dressing is applied 2 months later, which could be
responsible for the peak emissions which occurred during September and November.
Besides, owing to more temperature fluctuations and fertilizer application,
the monthly distribution of emissions in the northern regions was more
remarkable than that in the southern regions.
The significant seasonal dependence of NH3 emissions from livestock
wastes in different years can be clearly seen in Fig. 6c. The monthly
distribution of NH3 emissions was highly consistent with the variation
in temperature under the premise of the constant animal population among the
different months we assumed above. The major emissions occurred in warmer
months (May to September), and more than 45 % of the annual livestock
emissions, which could be explained by more NH3 volatilization related
to a substantial increase of temperature. In contrast, the lowest NH3
emissions from livestock wastes were estimated in winter (December to
February), and this is attributed to relatively smaller EFs linked to lower
temperatures.
Apart from the two major sources, the NH3 emissions from biomass
burning also had distinctly temporal disparities in spite of the relatively
small contribution of total emissions.
The temporal variations of emissions from crop burning in fields from 2003
to 2012 (when the annual MODIS thermal anomalies/fire products
(MOD/MYD14A1)) were available) are described in Fig. 6d. The occurrences
of crop burning in fields were concentrated in March to June with another
smaller peak in October, which are consistent with local sowing and harvest
times (Huang et al., 2012a). The highest emissions rates which
occurred in June are mostly attributed to the burning of winter wheat straw
that fertilizes the soil after the harvest (at the end of May) in the North
China Plain. The peak in October can be partly explained by the burning of
maize straw after the harvest (in the end September) in the North China
Plain. In addition, the south China provinces, including Guangdong and
Guangxi, have two or three harvest times every year. The sowing time for
crops here begins in March, when crop residues would be burned to increase
the soil fertility. Simultaneously, in northeast China, there is the local
farming practice of clearing the farmland before sowing in April, which may
emit corresponding NH3 during the spring. In winter, the mature period
of late rice in south China lead to a certain amount of NH3.
Figure 6e displays the seasonal distribution of NH3 emissions from
forest and grass fires from 2001 to 2012 (when the annual MODIS burned area
product (MCD45A1) was available) in China. The weather and vegetation
conditions are regarded as dominating factors that regulate the fire activity
(Perry et al., 2011). The fire emissions were primarily concentrated in
February to April and August to October, because of scarce precipitation,
high wind speed and gradually rising temperature during early spring and
late winter, especially in the southwestern regions. Simultaneously, the lower moisture content of vegetation increases the risk of burning. In addition, the
abundant fallen leaves and crop residues in autumn could make contributions
to the fires dramatically.
Comparison with previous studies
Comparison of total NH3 emissions between this study and
REAS.
Our NH3 emissions inventories provide a detailed description of
interannual variation from 1980 to 2012 in China. A comparison between this
study and the Regional Emission Inventory in Asia (REAS) is presented in
Fig. 7. The figures from REAS for 1980–2000 and 2000–2008 were derived
from version 1.1 (Ohara et al., 2007) and 2.1 (Kurokawa et
al., 2013), respectively. Note that the interannual variability in the
emissions in our study was generally consistent with that in REAS before
1996. However, after that year the annual trend of emissions in our study
differed from those in REAS. In addition, the NH3 emissions in REAS
were generally higher than those in our study. These differences are likely
attributable to differences in the estimations of synthetic fertilizer
emissions, discussed below.
Comparison of NH3 emissions from synthetic fertilizers
between this study and REAS.
Comparison of NH3 emissions (Tg yr-1) from our study with
other published results*.
Base
Total
Synthetic
Husbandry
Biomass
Others
year
Fertilizer
burning
Zhao and Wang (1994)
1990
13.6/8.4
6.4/4.0
4.2/3.9
3.0/0.9
Yan et al. (2003)
1995
4.3/4.3
Streets et al. (2003)
2000
13.6/10.3
6.7/3.8
5.0/5.3
0.8/0.25
1.1/0.95
Yamaji et al. (2004)
1995
5.1/5.2
2000
5.5/5.3
Ohara et al. (2007)
2000
0.5/0.24
Zhang et al. (2011)
2005
4.3/3.5
Zhao et al. (2013)
2010
9.8/3.0
Paulot et al. (2014)
2005–2008
10.4/10.1
Fu et al. (2015)
2011
3.0/2.8
* Before and after the slash represent other studies and this study,
respectively.
In REAS, NH3 emissions from animal manure applied as fertilizer were
included as a category of fertilizer emissions (Yan et al.,
2003). NH3 from the application of animal waste onto croplands was 2.8 Tg
in 2000 in REAS, accounting for approximately 60 % of the total
fertilizer emissions in that year. To render these two inventories
comparable, we excluded the application of animal waste from the fertilizer
emissions in REAS using the value for 2000. A comparison of the emissions
from synthetic fertilizer application is presented in Fig. 8. We found that
the REAS values were 20–50 % higher than ours in 2000–2005, and this
percentage rose to 100 % by 2008, which could be largely responsible for
the differences of total emissions between REAS and our study in the 2000s.
It should be noted that in REAS 2.1, the agricultural emissions were
extrapolated from REAS 1.1 for 2000 using the corresponding activity data of
the target year (Kurokawa et al., 2013), which could have resulted in
considerable inaccuracies due to various missing parameters. On the other
hand, the discrepancy possibly originated from the treatment of the types of
fertilizer and the corresponding EFs considered in the estimation methods.
As mentioned above, the types of fertilizer applied have changed
substantially since 1997. Although the total amount of synthetic fertilizers
increased significantly, the proportion of highly volatile ABC consistently
decreased, which could be responsible for the marked decline in NH3
emissions. However, REAS considered only the total fertilization activities
rather than the change in fertilizer types so the emissions in REAS
continued to increase in recent years. Moreover, we took into account the
local environmental conditions (soil pH, wind speed etc.) and agricultural
practices, and used fields results from Chinese studies to correct the EFs,
whereas REAS employed only uniform EFs based on European studies, and
applied these across the whole of China. Fu et al. (2015) recently
estimated synthetic fertilizer NH3 emissions at approximately 3.0 Tg in
2011 using the bi-directional CMAQ model coupled to an agro-ecosystem model,
which is similar to our value of 2.8 Tg for the same year and supports the
reliability of our inventories.
Table 3 shows the comparison of the emissions from livestock waste in our
study with previous ones. Our results are generally in agreement with those
of Zhao and Wang (1994), Streets et al. (2003), and
Yamaji et al. (2004). The majority of the previous inventories
used European-based EFs, which could introduce significant inaccuracies. Our
study employed a mass-flow approach, and considered three different
livestock rearing systems, as well as four phases of manure management based
on local agricultural practices. The EFs used in our study were also refined
according to environmental conditions. Hence, our estimations employed more
realistic parameters, and the differences between the present study and
previous ones are expected. Paulot et al. (2014) estimated
the annual NH3 emissions of 10.4 Tg using a global 3-D chemical
transport model in 2005-2008 while our result was 10.2 Tg for the
same period and Huang et al. (2012b) estimated 9.8 Tg in 2006. The three
results are quite close. We also found excellent qualitative agreement for
spatial distribution between our estimation and the global NH3 column
retrieved by IASI sensor (Van Damme et al., 2014). Several emission
hotspots shown in this study, including the North China Plain, Sichuan and
Xinjiang provinces (near Ürümqi and in Dzungaria), and the region
around the Tarim Basin were also detected by the IASI sensor.
Uncertainty
Uncertainties in NH3 emissions originated from the values used for
both the activity and EFs. Huang et al. (2012b) summarized the possible
sources of uncertainty in the emissions inventory, including extremely high
activity data for fertilizer use and livestock, the numerous parameters
involved in the EF adjustment, and large variation (≥ 100 %) in the
coefficients of biofuel combustion and chemical industry production. We may
miss some possible sources like NH3 loss from fertilization in orchard
and also overestimated emissions in agricultural soils covered with plastic
shed. In this study, the impacts of wind speed and ambient temperature on
the EFs in agricultural ammonia emissions were isolated but in real
conditions, there might be some interactions between temperature and wind
speed. Ogejo et al. (2010) indicated that parameter interactions may play
a significant role in emission estimation with a process-based model for
ammonia emissions but they also did not consider the interaction between
temperature and wind velocity. Actually, previous studies generally examined
the respective effect of wind speed and temperature on ammonia
volatilization according to controlled experiments (Sommer et
al., 1991) and we expect more experimental evidences for the interaction
effect.
Furthermore, our method mostly used constant parameters for estimating
30-year inventories rather than the time-varying, which may introduce
additional uncertainties. First, the application rate and synthetic
fertilization method may have changed during recent decades because Chinese
farmers have come to expect higher grain production within limited areas of
cropland, which may lead to uncertainties in NH3 loss per unit area.
Second, although we considered interannual changes in the percentage of
intensive rearing systems to livestock emissions, manure management, that
was divided into four phases in our method, could have also changed over
time because it was affected by many factors including the N content of the
feed, housing structure, manure storage system, spreading technique, and
time spent outside or indoors (Zhang et al., 2010). For example, the feed
situation in Chinese agriculture has been changed, e.g. animal horsing
conditions, feedstuff types or feeding periods. Zhou et al. (2003)
conducted rural household surveys on the Chinese household animal raising
practices. They found that in some provinces like Zhejiang, industrial
processed feed had become a major animal feed. The industry processed feed
is easy to digest and absorb, showing more use efficiency than traditional
farm-produced forage. Therefore, the amount of N excreta per animal feed by
industry forage should be less than that by farm forage. But Li et
al. (2009) investigated that compared with 1990s, little has changed in the average N content in
manure from pig, chicken, beef and sheep in recent years
according to a nationwide analysis of 170 samples. On the other hand, rearing
periods for animals like poultry were significantly reduced during recent
years along with the development of breeding technology, that is, manure
excreted per animal per year was supposed to be declining. However, this
change was not considered in this study and it may result in overestimation
of livestock emissions in recent years. In addition, over recent decades,
excessive synthetic fertilizer use has caused significant soil acidification
in China (Guo et al., 2010), but our inventories did not consider the
influence on NH3 volatilization. Monte Carlo is an effective method to
evaluate the uncertainties in various issues including an emission
inventory. In Monte Carlo simulation, random numbers are selected from each
distribution (normal or uniform) of input variables and the output
uncertainty of an emission inventory is based on the input uncertainties
from activity data and emission factors. In this study, we ran 20 000 Monte
Carlo simulations to estimate the range of NH3 emissions with a 95 %
confidence interval for 1980, 1990, 2000, and 2012. The estimated emission
ranges were 4.5–7.4, 6.3–11.1, 8.0–13.4, and 7.5–12.1 Tg yr-1, respectively.
Conclusions
We developed comprehensive NH3 emission inventories from 1980 to 2012
in China. Generally, emissions increased from 1980 to 1996, reaching a peak
value of approximately 11.1 Tg, then fluctuated at around 10.5 Tg from 1997
to 2006, but underwent a sharp decrease after 2006. The interannual
variation in the emissions is attributable to changes in the types of
synthetic fertilizer applied and livestock manure management. These factors
were the two major NH3 sources, accounting for more than 80 % of
total NH3 emissions, while demonstrating different temporal trends.
Emissions from synthetic fertilizers initially rose, from 2.1–4.7 Tg, in
the period 1980–1996, and then decreased to 2.8 Tg by 2012, which was
caused by a change in the relative contributions of urea and ABC consumption
to total emissions. In contrast to synthetic fertilizer emissions, emissions
from livestock, ranging from 2.9–6.1 Tg from 1980 to 2012, rose until 2005,
but significantly decreased after 2006. Other sources were insignificant in
the total budget but they could play a role in specific region or periods
like vehicles on road in big cities, crop residue burning and large wild
fires due to agricultural timing and climate conditions. NH3 emissions
generally peaked in the spring and summer, corresponding to planting
schedules and relatively high temperature that were the two determining
factors for the monthly variation of mineral fertilizer and livestock
emissions, respectively. The emissions from crop residue burning were
generally concentrated from March to June and October when major crops like
winter wheat and corn are harvested. At the regional level, the spatial
patterns of the total emissions have generally been consistent over recent
decades, with high emissions rates of more than 2000 kg km-2
concentrated in Hebei, Shandong, Henan, Jiangsu, Anhui, and East Sichuan
provinces, which represent the major areas of intensive agriculture in
China. Compared to NH3 emissions in REAS, our results are more reliable
because we considered more parameters when calculating specific EFs
according to local conditions and agricultural practices.
It should be noted that gaps still exist in these inventories due to
uncertainties in the activity data, EFs, and related parameters, especially
for earlier years. As many samples as possible should be used in statistical
censuses, and more local field studies should be implemented for better
estimates of the EFs to reduce uncertainties. Such high-resolution
inventories can be used in global and regional modeling to simulate
atmospheric aerosol formation, explore the impacts of NH3 emissions on
air quality, and understand the evolution of the N cycle and atmospheric
chemistry during recent decades. In addition, we expect our results to be
validated by top-down estimates in future studies.