6.4.2 Influence of NO x and NH 3 Emission Reductions
on the PM 2.5 Concentration: The Model Study
Numerical simulations are an effective means to evaluate the response of pollutant
concentrations to their precursor emissions and thus to assess the influence of
emission control strategies on air quality. In this area sensitivity analysis is one of
the most commonly used methods. In the sensitivity test, a group of emissions is
changed based on a standard scenario. The difference in the simulation results
between the sensitivity test and the standard scenario can be seen as the influence
of the changes in emissions. This approach is most widely used to predict the impact
on air quality from emission changes of a certain type of pollutant. The sensitivity
analysis is easy to operate and computationally efficient, but it fails to deal with
uncertainty in emission inventories. Some studies assimilate observation data into
chemical transport models to optimize emissions (Zhang et al. 2015, 2016). However, observation data are not always available, especially for the components of
PM 2.5 . On the other hand, data assimilation is inefficient in computation, so it is
useful to do the post-evaluation for major events, the duration time of which is
usually less than 1 month. Response surface modeling (RSM) is one of the statistical
methods that is used to construct relationships between model inputs and outputs
(Wang et al. 2011; Xing et al. 2011) but is not widely used because of its computational complexity.
Model performance in reproducing SNA concentrations is crucial in sensitivity
analysis, as it determines the reliability of the analysis. The models can adequately
capture the magnitude of the SNA concentration and its variation. In general, NO 3
À
is overestimated by many models, where the simulation of SO 4
2À is much better than
NO 3
À , but there is some underestimation during polluted seasons. The lack of
heterogeneous formation of SO 4
2À and NO 3
À during polluted days and the large
uncertainty associated with the NH 3 emission inventory are the main causes of bias
in SNA simulation, but efforts have been made to improve model performance
(Heald et al. 2012; Wang et al. 2014; Zheng et al. 2015).
There is a thermodynamic equilibrium among H 2 SO 4 , HNO 3 , and NH 3 in the
atmosphere. In brief, NH 3 prefers to react with H 2 SO 4 to form (NH 4 ) 2 SO 4 , after which
excessive NH 3 will react with HNO 3 to form NH 4 NO 3 . NH 4 NO 3 is volatile, and it
tends to be in the aerosol phase when the temperature is low and relative humidity is
high. The thermodynamic equilibrium indicates that changes in both SO 2 and NH 3
emissions will lead to changes in the NO 3
À concentration, and a change in NH 3
emissions has a greater influence on NO 3
À than SO 4
2À (Wang et al. 2013, 2014).
The focus on SO 2 emission control before 2010 was not successfully helped to
reduce the PM 2.5 concentration due to the lack of efforts to control NO x and NH 3
emissions (Zhao et al. 2013). Wang et al. (2011) identified the nonlinearity of the
response of PM 2.5 to NO x and NH 3 emission changes. For the year 2005, NO x and
NH 3 contributed 5–11% and 8–11% to PM 2.5 , respectively. They also found that the
90% increase in NH 3 emissions in China from 1990 to 2005 resulted in a 50–60%
increase in the NO 3
À and SO 4
2À concentrations.
6 Contribution of Atmospheric Reactive Nitrogen to Haze Pollution in China
125
on the PM 2.5 Concentration: The Model Study
Numerical simulations are an effective means to evaluate the response of pollutant
concentrations to their precursor emissions and thus to assess the influence of
emission control strategies on air quality. In this area sensitivity analysis is one of
the most commonly used methods. In the sensitivity test, a group of emissions is
changed based on a standard scenario. The difference in the simulation results
between the sensitivity test and the standard scenario can be seen as the influence
of the changes in emissions. This approach is most widely used to predict the impact
on air quality from emission changes of a certain type of pollutant. The sensitivity
analysis is easy to operate and computationally efficient, but it fails to deal with
uncertainty in emission inventories. Some studies assimilate observation data into
chemical transport models to optimize emissions (Zhang et al. 2015, 2016). However, observation data are not always available, especially for the components of
PM 2.5 . On the other hand, data assimilation is inefficient in computation, so it is
useful to do the post-evaluation for major events, the duration time of which is
usually less than 1 month. Response surface modeling (RSM) is one of the statistical
methods that is used to construct relationships between model inputs and outputs
(Wang et al. 2011; Xing et al. 2011) but is not widely used because of its computational complexity.
Model performance in reproducing SNA concentrations is crucial in sensitivity
analysis, as it determines the reliability of the analysis. The models can adequately
capture the magnitude of the SNA concentration and its variation. In general, NO 3
À
is overestimated by many models, where the simulation of SO 4
2À is much better than
NO 3
À , but there is some underestimation during polluted seasons. The lack of
heterogeneous formation of SO 4
2À and NO 3
À during polluted days and the large
uncertainty associated with the NH 3 emission inventory are the main causes of bias
in SNA simulation, but efforts have been made to improve model performance
(Heald et al. 2012; Wang et al. 2014; Zheng et al. 2015).
There is a thermodynamic equilibrium among H 2 SO 4 , HNO 3 , and NH 3 in the
atmosphere. In brief, NH 3 prefers to react with H 2 SO 4 to form (NH 4 ) 2 SO 4 , after which
excessive NH 3 will react with HNO 3 to form NH 4 NO 3 . NH 4 NO 3 is volatile, and it
tends to be in the aerosol phase when the temperature is low and relative humidity is
high. The thermodynamic equilibrium indicates that changes in both SO 2 and NH 3
emissions will lead to changes in the NO 3
À concentration, and a change in NH 3
emissions has a greater influence on NO 3
À than SO 4
2À (Wang et al. 2013, 2014).
The focus on SO 2 emission control before 2010 was not successfully helped to
reduce the PM 2.5 concentration due to the lack of efforts to control NO x and NH 3
emissions (Zhao et al. 2013). Wang et al. (2011) identified the nonlinearity of the
response of PM 2.5 to NO x and NH 3 emission changes. For the year 2005, NO x and
NH 3 contributed 5–11% and 8–11% to PM 2.5 , respectively. They also found that the
90% increase in NH 3 emissions in China from 1990 to 2005 resulted in a 50–60%
increase in the NO 3
À and SO 4
2À concentrations.
6 Contribution of Atmospheric Reactive Nitrogen to Haze Pollution in China
125