Analysis of the Effect of Optical Properties of Black Carbon on Ozone in an Urban Environment at the Yangtze River Delta, China

Junlin AN, Huan LV, Min XUE, Zefeng ZHANG, Bo HU, Junxiu WANG, and Bin ZHU

1Key Laboratory of Meteorological Disaster, Ministry of Education (KLME)/Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters (CIC-FEMD), Nanjing University of Information Science and Technology, Nanjing 210044, China

2Chinese Academy of Meteorological Sciences, Beijing 100081, China

3Institute of Atmospheric Physics Chinese Academy of Sciences, Beijing 100029, China

4Inner Mongolia Meteorological Bureau, Hohhot 010051, China

ABSTRACT Black carbon (BC) reduces the photolysis coefficient by absorbing solar radiation, thereby affecting the concentration of ozone (O3) near the ground. The influence of BC on O3 has thus received much attention. In this study, Mie scattering and the tropospheric Ultraviolet and Visible radiation model are used to analyze the effect of BC optical properties on radiation. Combined with data of O3 precursors in Nanjing in 2014, an EKMA curve is drawn, and the variations in O3 concentration are further investigated using a zero-dimensional box mechanism model (NCAR MM). When O3 precursors are unchanged, radiation and O3 show a highly similar tendency in response to changing BC optical properties (R=0.997).With the increase of modal radius, the attenuation of fresh BC to radiation and O3 first trends upward before decreasing. In the mixing process, the attenuation of BC to radiation and O3 presents an upward tendency with the increase of relative humidity but decreases rapidly before increasing slowly with increasing thickness of coating. In addition, mass concentration is another major factor. When the BC to PM2.5 ratio increases to 5% in Nanjing, the radiation decreases by approximately 0.13%–3.71% while O3 decreases by approximately 8.13%–13.11%. The radiative effect of BC not only reduces O3 concentration but also changes the EKMA curve. Compared with the NOx control area, radiation has a significant influence on the VOCs control area. When aerosol optical depth (AOD) increases by 17.15%, the NOx to VOCs ratio decreases by 8.27%, and part of the original NOx control area is transferred to the VOCs control area.

Key words: black carbon, ozone, radiation, optical properties, EKMA curve

1. Introduction

Tropospheric ozone (O) is a typical secondary pollutant in the atmosphere and poses a serious threat to human health, vegetation, and the atmospheric environment(Crutzen et al., 1973; Feng et al., 2015; Goodman et al.,2015). Urban air quality environmental monitoring data show that as the Oconcentration near the ground has increased year by year in China, Ohas gradually become the main pollutant affecting air quality in China (Li et al.,2019). Meteorological factors and synoptic patterns have great impacts on the variation of O(Dong et al., 2020).Also, the major contribution to Oconcentrations near the ground is well recognized to come from aerosol particles,which is clear from this study.

Black carbon (BC) accounts for a small fraction of aerosols, but it exerts a large radiative effect on the energy balance. BC is produced by incomplete combustion of carbonaceous fuels (Bond et al., 2004). Strict emissions control and reduction measures implemented in China have made both aerosol (Zhang et al., 2019a) and BC mass concentrations (Zhang et al., 2019b) decrease sharply. By absorbing solar radiation, BC can not only change the global and regional radiative balance but also reduce the photolysis coefficient, thereby reducing the concentration of Onear the ground, which makes the influence of BC on Oalways of concern (Jacobson, 2001; Ramanathan et al., 2008; Yang et al., 2016).

Chen et al. (2019) found a negative correlation between BC and Othrough observations in the Nanjing area. Dickerson et al. (1997) proposed that BC could reduce the surface Oconcentration by reducing the photolysis rate. Jacobson(1998) and Castro et al. (2001) found that aerosols containing BC significantly reduced the photolysis coefficient by absorbing solar radiation, and the near-surface Oconcentration decreased by 5%–8%. Li et al. (2005) simulated a period of heavy air pollution in Houston through a regional chemical transport model and found that BC reduced the photolysis coefficient in the boundary layer by 10%–30% and Oconcentration by 5%–20%. Gao et al. (2018) found that absorption of radiation by BC at the top of the boundary layer inhibits the development of the boundary layer in the daytime, and a large number of Oprecursors are limited to the boundary layer, thus promoting the photochemical generation of O. This would counteract the effect of BC lowering the photolysis coefficient to inhibit Oformation. However,the experimental results from studies of factors such as BC scale distribution and aging process are still highly uncertain (Huang et al., 2015). The optical properties of BC and its radiative effect are not fully understood, and the mechanisms of the influence of BC on Oneeds to be further studied.

The optical properties of BC, which are closely related to its particle size distribution and mixing state, are the basis for simulating its radiative effect (Bond et al., 2013) and directly affect the simulated results for O(Saathoff et al.,2003; Chung et al., 2005; Ma et al., 2011). Based on a number of observations of BC particles, both BC number concentration and mass concentration are characterized by a logarithmic normal distribution (LND) (McMeeking et al., 2010;Huang et al., 2012; Wang et al., 2014). The peak particle size of approximately 120–210 nm (Kondo et al., 2006;Gong et al., 2015) is related to the environment and the emission source (Park and Lee, 2015). After being discharged into the atmosphere, BC tends to absorb other aerosols and mix with them (Oshima et al., 2009). Mixing mode, shell chemical composition, thickness, and environmental humidity are important factors that affect the mixing state of BC.Jacobson (2001) found that BC could significantly enhance radiation forcing when it was mixed with other groups, and the rate of increase was related to the hygroscopic growth effect of the aerosol. Zhang et al. (2008) found that, compared with fresh particles, the absorption efficiency of internal mixed particles was nearly two times higher and the scattering efficiency was approximately ten times higher in a high-humidity environment. Liu et al. (2015) proposed that the enhanced optical absorption effect of BC depends largely on the thickness of the shell layer, which is related to the emission source. Therefore, it is necessary to conduct sensitivity tests on BC optical parameters when studying the influence of BC on O.

In this paper, to analyze the sensitivity of radiation and Oto BC optical properties, the mass concentration ratio of BC in particulate matters (PM) and the spectral distribution and the mixing state of BC were changed. As detailed in the following sections, Mie scattering theory and the program of coated spheres according to Bohren and Huffman(BHCOAT) scattering program were used to calculate optical parameters of BC, the tropospheric Ultraviolet and Visible radiation (TUV) model was used to calculate the value of radiation and photolysis coefficients, and a zerodimensional box mechanism model (NCAR MM) was used to calculate Oconcentrations under different optical parameters of BC. Finally, the variation in the Empirical Kinetics Modeling Approach (EKMA) curves after radiation attenuation was analyzed by using the observed data for nitrogen oxides (NO) and volatile organic compounds (VOCs) in Nanjing.

2. Experimental methods and scheme design

2.1. Experimental methods and data

In this study, a box model with no source emission or advection and diffusion processes and only photochemical processes is used for simulation and calculation. The model is composed of three modules: aerosol optical parameter, radiative transfer, and the chemical mechanism related to phosgene (Fig. 1). The optical properties of BC are mainly calculated by Mie scattering theory. The radiation parameters and Oconcentration are respectively solved by the TUV model and the NCAR MM model. The optical parameters of fresh BC and PMare solved by the Lorenz-Mie Scattering code developed by Michael Mishchenko. Mie scattering assumes that the particles are spherical. Through the input of complex refractive index and spectral distribution, the following optical parameters of the aerosol particle swarm are obtained: single scattering albedo (

ω

), asymmetric factor(g), extinction efficiency (Q), and volume extinction coefficient (k). Refractive index and density of PM(Dubovik et al., 2002; Pitz et al., 2003) and those of BC (Hess et al.,1998; Gong et al., 2015) are shown in Table 1. A logarithmic normal distribution (LND) is selected, and the analytical expression is shown in Eq. (1):

Fig. 1. Structure diagram of box model.

Table 1. Aerosol parameter settings.

where ris the modal radius, σ is the width of the spectral distribution, and N is the particle number concentration.When considering the mixing process and the influence of relative humidity (RH), the k-EC-Mie model is adopted for the optical properties of BC (Zhang et al., 2017). The k-EC-Mie model is based on the core-shell model and assumes that 1)the physical and chemical properties of particles with the same scale are the same; 2) under dry conditions, the particle consists only of BC (core), located in the center of the particle, and an envelope (shell); and 3) regardless of the hygroscopicity of BC, the coating is evenly mixed with water before the hygroscopic growth process. Finally, the layered spherical Mie scattering method (BHCOAT scattering program) proposed by Bohren and Huffman (2008) is used for calculations. The core-shell volume ratio of each particle is set as constant when calculating the optical properties of particle swarm.

The TUV model is a radiation transmission model developed and continuously improved by the National Center for Atmospheric Research (NCAR) and embedded in the NCAR MM model (Madronich and Flocke, 1999). It can calculate various radiation quantities such as solar incident radiation intensity and photolysis rate coefficients of 73 photochemical reactions with wavelengths in the range of 121–1000 nm (Stamnes et al., 1988). The TUV model is widely used in the NCAR MM model, WRF-Chem model,and other regional and global atmospheric chemistry models (Shao et al., 2016; Ryu et al., 2018). Palancar et al.(2013) present a comparison between measured actinic fluxes and those simulated with the TUV model and found that the simulated result is larger and the difference between the two is about 10%–40%. In addition to input parameters such as w and g, aerosol optical depth (AOD) with a wavelength of 0.55 μm near the ground should also be calculated using Eqs. (2) and (3) (Hess et al., 1998):

The NCAR MM model is used to calculate the chemical mechanism related to phosgene. The model includes nearly 5000 gas phase chemical reactions between more than 2000 substances. Assuming no other emission sources,no dilution, and no recombination process, it can calculate the time-dependent chemical evolution of air with known initial components (Madronich and Calvert, 1990). Users define input parameters such as initial gas component and volume fraction and then calculate the Oconcentration by combining with photolysis coefficient output from the TUV model. Lingaswamy et al. (2017) used the TUV model to calculate the photolysis rate coefficients [

J

(O),

J

(NO)], used the NCAR MM model to simulate diurnal variation of CO,and reported their results.

The volume fraction of gas and meteorological data required by the NCAR MM model are provided as hourly means from March 2014 to February 2015. The observation site is located in the Key Laboratory of Aerosol and Cloud Precipitation in Nanjing University of Information Technology (32°12′N, 118°42′E; altitude 62 m). O, NO,NO, and CO were observed using an air pollution environmental monitoring analyzer produced by Thermo Fisher Scientific in the United States, including an ultraviolet luminescence Oanalyzer (Model 49i), infrared absorption CO analyzer (Model 48i), and chemiluminescence NO-NO-NOanalyzer (Model 42i). Detailed instrument parameters and calibration methods can be found in Hansen et al. (1984). VOCs were observed using a GC5000 automatic online gas chromatography-flame ionization detector produced by AMA Company in Germany, and detailed instrument parameters and calibration methods can be found in An et al. (2014).

2.2. Experimental scheme design

The sensitivity experiment consists of a control experiment and four variant schemes. In the experiments, PMand BC are fully independent, and the sum of their mass concentration is always 75 μg m. The mixing mode between the two is external mixing. The external mixing method is actually independent scattering between particles. The aerosol optical parameters of PMand BC need to be calculated independently first, and then the optical properties of the mixed aerosol can be calculated according to the weight of the mass concentration.

In the control test, PMis regarded independently without the influence of BC. Then, the net radiation irradiance (579.85 W m) and the Oconcentration (46.76 ppb)near the ground can be calculated, where the concentration of PMis determined by the Chinese ambient air PMstandard (75 μg m). In sensitivity tests, mass concentration ratio of BC to PM, the spectral distribution, and mixing state are respectively changed (Table 2). Based on the observed data in Nanjing from 2015 to 2018, Tan et al.(2020) found that the annual mean ratio of BC to PMwas about 4.92%, and the monthly mean was about 1.5% to 10.1%. Therefore, in this study, the ratio of BC to PMis set at 3%–9% when discussing the influence of the mass concentration of BC on O, and the BC to PMratio is set at 5% when discussing the influence of the spectrum distribution and mixing process of BC on O. REO (%) is defined as the relative change between a sensitivity test and control test [ Eq.(4)]:

If REO is negative, then BC has an attenuation effect on net radiation irradiance and O. The greater the absolute value of REOand REO, the greater the attenuation of BC to radiation and O. The above Oconcentration refers to the average of a continuous eight hours of an Oconcentration at the largest daily values, and the net irradiance is the average value in the corresponding period.

2.3. Model validation

To verify the applicability of the above PMspectral distribution in the Nanjing area and the accuracy of the NCAR MM model, the observed concentration of Onear the ground is compared with the simulation results. In this study, the measured data during sunny weather with a wind speed less than 4 m sin 2014 are selected as the input to the NCAR MM model, resulting in six days being selected.By using these criteria, the simulation error caused by the transmission effect, wet deposition, and the influence of clouds will be significantly reduced (Shao et al., 2016).According to the above PMspectral distribution and the daily mean value of PMobservation data, the daily ground-level AOD can be calculated. The calculated AOD and other observed data are input into the NCAR MM model for simulation. The 6-hour average value of volume fraction of Oprecursors is used as the limit of each simulation. According to the six selected samples, six simulations are performed. The simulation results show the diurnal variation of Oand are highly correlated with the observations(R = 0.87); all reported correlation coefficients in this paper pass the significance test (P ＜ 0.05). As shown in Fig. 2, compared with the observation data, the simulation results for tendencies in Ovariation and concentration perform well during the daytime (0600–1800 LST, LST = UTC + 8), but simulated Oconcentration is higher at noon, which may be related to real-time cloud changes (Zhang et al., 2015). In general, the observed tendency is consistent with the simulated tendency. The slope of the fitting curve is 1.085, indicating that the simulated values are relatively close to the actual values. This shows that the NCAR MM model can accurately simulate the variation tendencies and concentrations of O,and the spectral distribution setting of PMis reasonable.

Table 2. Schematic setting for the sensitivity analysis.

3. Results

3.1. The influence of BC on radiation

Fig. 2. Comparison of observed ozone and modeled ozone in the northern suburbs of Nanjing for the selected six days in 2014.

Fig. 3. Effects of BC mass concentration on (a) net radiation and on (b) concentration of O3. REONRI and REOO3 represent the relative changes of net radiation and O3 concentration between sensitivity test and control test, respectively. “BC”,“BC-H2SO4”, and “BC-H2SO4” stand for fresh BC particles,old BC that fresh BC particles have coated with sulfate under dry conditions, and old BC that fresh BC particles have coated with sulfate under an environmental RH of 50%, respectively.

Table 3. Optical properties of particles in three states.

To explore the effect of fresh BC spectral distribution on radiation, rand width (

σ

) spectral distribution are changed. The larger the ror

σ

is, the larger the particle size of the particle population. The net radiation attenuation of fresh BC is approximately 1.75%–3.71%. As shown in Fig. 4b, REOfirst decreases and then increases with increasing r. When the value of

σ

is 1.8, 2.0, and 2.2,the ris approximately 0.041 μm, 0.026 μm, and 0.017 μm, respectively (i.e., a shift towards small particles when REOis the smallest). The corresponding effective radius(reff) is 0.097 μm, 0.088 μm, and 0.079 μm, respectively(Table 4). That is, when the reff of particle swarm is approximately 0.1 μm, BC has the strongest attenuation of radiation, which is close to the results of Zhang et al. (2017). It is found that the ratio of σreffπ to wavelength is approximately 1 in the three cases, and the above optical parameter is calculated when wavelength is 0.55 μm. The attenuation of BC is strongest when the ratio of σ reffπ is close to wavelength.

Fig. 4. Effects of spectral distribution on (a) changes in optical parameters when σ is 2 and λ is 0.55 μm and on (b) changes in radiation and concentration of O3.

Table 4. The values of optical parameters with maximum k ′ext .

Two enveloping substances are considered next,namely, sulfuric acid (HSO) and organic carbon (OC).Table 5 shows the relevant parameters of the two enveloping substances, in which k is the moisture absorption parameter. The real and imaginary parts of the complex refraction index of OC are both greater than that of HSO, which makes the

k

of BC enclosing OC larger. Therefore, the attenuation of BC to radiation is stronger when the wrapper is OC. The hysteretic parameter of HSOis 1.13, which is much higher than OC. In other words, when the coating is HSO, RH has a more significant influence on BC (Fig. 5).Taking the wrapper of BC being HSOas an example,Fig. 7 describes the change of particle density and other parameters with changes in RH under different particle sizes.With the increase of RH, the density and refractive index of the shell show a monotonically decreasing tendency (Fig. 7b).However, due to the increase of shell hygroscopicity and particle size, the particle volume and extinction coefficient show a monotonically increasing tendency, which is consistent with other studies (Zhang et al., 2017). When the thickness of coating is small (0.012 μm ＜ r＜ 0.015 μm), the contour lines are dense and nearly straight. In this case, the hygroscopic growth effect of the shell is weak, kand mass are almost unchanged (Fig. 7a), and the effect of RH on

k

is very small. When RH increases from 0 to 95%, the attenuation value of net radiation only changes by 0.36%, and the thickness of coating is the main influencing factor. When the thickness of coating is close to the low center value(0.02 μm ＜ r＜ 0.025 μm), the contour thinning and curvature increase, the influence of RH on BC begins to increase, and the attenuation value of BC to radiation increases with the increase in RH. When the thickness of coating is large (r＞ 0.025 μm), the hygroscopic growth effect of the coating is more obvious, and mass and kincrease significantly (Fig. 7a). In this case, BC reduces the net radiation by 1.25%, with this attenuation effect mainly being caused by the change in RH.

Fig. 5. Effects of mixing state on (a) changes in net radiation when the coating is H2SO4, (b) changes in O3 when the coating is H2SO4, (c) changes in net radiation when the coating is OC, and (d) changes in O3 when the coating is OC.The horizontal coordinate is rmod,N in a dry state, which is proportional to the thickness of the coating, while the colored contours represent the values of REONRI and REOO3.

Fig. 6. Changes in particle mass and kext when RH is 20%(top) and the coating is H2SO4 (bottom).

3.2. The influence of BC on ozone

When using the NCAR MM model to calculate the change of O, the hourly concentration value of Oprecursors is taken as the limiting condition of the model and as input to the model. In the model simulation, the data of Oprecursors from the days with no precipitation and wind speed less than 4 m sin the northern suburbs of Nanjing in 2014 are used. Figure 8 describes the diurnal change process of NO, VOCs, and CO in the northern suburbs of Nanjing for the selected days. The diurnal variations of the three factors are consistent: the concentration reaches the highest value at 0800 LST and the lowest at 1400 LST.

Table 5. Parameter settings for wrappers.

In addition to the influence of Oprecursors, radiation is the most critical factor for O(Sillman, 1995). In the calculations, when conditions for Oprecursors are consistent,the net irradiance is significantly correlated with O(R=0.997), so radiation and Oin all schemes show similar variation tendencies (Figs. 3–5) (the impact of Oconcentration changing with BC radiative effect is described in section 3.1). The higher the mass concentration of BC, the greater the attenuation to Owill be (Fig. 3b). With the increase of r, the decay of fresh BC to Ofirst increases and then decreases, and Oconcentration decreases by approximately 10.13%–13.11%. When BC undergoes a mixing process, Oconcentration is reduced by approximately 8.13%–11.94%. The attenuation value of Ois consistent with that of radiation with respect to changing with the thickness and relative humidity of the coating.From what has been discussed above, when the BC to PMratio is 5% in Nanjing, the change in the attenuation of BC to radiation is approximately 0.98%–1.71%, and the change in attenuation of BC to Ocan reach approximately 1.63%–2.57% due to different spectral distributions of BC;changes in the attenuation of BC to radiation and Oare approximately 0.94%–1.97% and 1.54%–3.06%, respect-ively, due to different mixing processes of BC.

Fig. 7. The change process of optical parameters with RH when the coating is H2SO4.

Fig. 8. The diurnal variation process of NOx, VOCs, and CO in Nanjing in 2014.

After the maximum point,

k

shows a decreasing tendency and the irradiance increases with it. At this time, w and g, which show a slight decreasing tendency for the mixed aerosol, have the effect of reducing the irradiance, and that causes the drastic fluctuations of REOand REO.To further analyze the influence of radiation on the pathway of Ogeneration after the attenuation effect of BC on radiation, the NCAR MM model is used to simulate the concentration of O(

φ

) in the northern suburbs of Nanjing in 2014, and an EKMA curve for the Nanjing area is established (Fig. 9). The volume fraction of Oprecursors, such as NO, and the proportion of alkanes, olefins, and aromatic hydrocarbons in VOCs, etc., are all based on average values of their observational data from 0700 to 0900 LST(Table 6). Other meteorological elements, such as temperature, are based on hourly mean values of the annual observation data in 2014. In NOcontrol areas, OH mainly reacts with VOCs, which have little influence on O. Ois more sensitive to NO, so reducing NOis beneficial for hindering the generation of O. At VOCs ＞ 200 ppbc and NO＜15 ppb (i.e., when the VOCs to NOratio is large), the radiation has little impact on O, and an approximately 0.95 ppb reduction in NOachieves the

φ

in the control experiment. However, when the volume fractions of NOand VOCs are close to the ridge line, the effect of radiation is greater. When

φ

remains the same, the difference in NOconcentration corresponding to the two experiments is as high as 6.00 ppb. In the VOCs control area, the reaction of OH and NOis dominant, and Ois sensitive to VOCs. In this case, reducing NOwill promote the formation of O.In this region, radiation has a significant effect on

φ

. At VOCs ＜ 150 ppbc, NOneeds to increase by approximately 2.14 ppb for

φ

to be consistent with its value when the radiation dose decreases; at VOCs ＞ 250 ppbc, NOneeds to increase by approximately 9.47 ppb. In addition, radiation can also cause the ridge line of the EKMA curve to shift.When the AOD increases from 0.88 to 1.08, the slope of ridge line (the NOto VOCs ratio) decreases from 0.090 to 0.083. The original part of the region located in the NOcontrol area is transferred to the VOCs control area under the influence of radiation.

Fig. 9. Influence of radiation on diurnal maximum volume fraction curve of O3. The control test (AOD=0.88, PM2.5) is shown as the red curve while the sensitivity test (AOD=1.08,9%BC+91%PM2.5) is shown as the black curve. The dotted line is the ridge line of the EKMA curve, which is formed by fitting the turning point of each O3 contour line and the boundary between the VOCs control area and the NOx control area.

Table 6. Initial values of ozone precursors in NCAR MM.

4. Summary

BC can reduce the photolysis coefficient by absorbing solar radiation, thus reducing the near-surface Oconcentration. In this paper, the attenuation degree of BC to radiation is discussed in response to changing the optical properties of BC, including the mass concentration ratio of BC to PM, the spectrum distribution of BC, the thickness of coating, and RH in the internal mixing process. Combined with observed data for Oprecursors in Nanjing in 2014, an EKMA curve was calculated to examine the change in Oconcentration near the ground and pathways for Ogeneration after radiation attenuation caused by BC.

For BC, the attenuation generated by radiation is closely related to its mass, and the greater the mass, the greater the influence on radiation. When the mass concentration of the BC to PMratio is 5% in Nanjing, the attenuation of BC particles to net irradiance is approximately 0.1%–3.70%, and the range of this attenuation is approximately 0.98%–1.71% and 0.94%–1.97%, respectively, due to different spectral distribution and mixing processes of BC.With the increase of r, the radiation attenuation of fresh BC tends to increase first and then decrease. With the increase of

σ

, the radiation attenuation of fresh BC tends to shift to the direction of small particles. The attenuation of BC is strongest when the product of spectral width, effective radius, and π is similar to the wavelength. After BC goes through a mixing process, the attenuation value of radiation presents an upward tendency with an increase in RH,but it first decreases rapidly and then increases slowly with the increasing thickness of coating. When r＜ 0.015μm, the effect of RH is very small and the thickness of coating is the main factor affecting the attenuation capacity of BC. When r＞ 0.025 μm, the moisture absorption of coating increases significantly, with RH as the main influencing factor.It is found that when Oprecursors are unchanged, net irradiance is significantly correlated with O(R=0.997), so in the process of changing with changes in BC optical parameters, radiation and Oshow highly similar change tendencies. When the ratio of BC to PMincreases to 5% in Nanjing, Ocan be reduced by approximately 8.13%–13.11%,and the range of this attenuation is approximately 1.63%–2.57% and 1.54%–3.06%, respectively, due to different spectral distribution and mixing processes of BC. The attenuation effect of BC on radiation not only significantly reduces Oconcentration but also changes the EKMA curve. In the VOCs control area, the effect of radiation on

φ

is relatively significant. When radiation changes while

φ

stays the same, the difference in NOconcentration is approximately 2.14–9.47 ppb. In the NOcontrol area, only when the volume fraction of NOand VOCs is close to the ridge line, the effect of radiation is obvious, and the difference in NOconcentration is approximately 6.00 ppb, at the most, when

φ

stays the same. In addition, when the AOD increases from 0.88 to 1.08, the slope of the ridge line (the NOto VOCs ratio) decreases from 0.090 to 0.083 (i.e.,when the AOD changes by 17.15%, the NOto VOCs ratio will decrease by 8.27%), and part of the original NOcontrol area will be transferred to the VOCs control area.

Acknowledgements

. This work was supported by grants from the National Key Research and Development Program of China (Grant No. 2017YFC0210003), the National Natural Science Foundation of China (Grant No. 42075177), and the Qing Lan Project.