Impacts of Meteorology and Emissions on O3 Pollution during 2013–2018 and Corresponding Control Strategy for a Typical Industrial City of China

The air quality of Handan, a typical industrial city in China, has been significantly improved through atmospheric pollution control, except for ozone (O3) pollution. We found that, in summer, emissions of anthropogenic volatile organic compounds (VOCs) and NOx decreased yearly in Handan, but the O3 concentration significantly declined yearly during 2013–2015, whereas it experienced worsening O3 pollution after 2015. Therefore, we used the Weather Research and Forecasting Community Multiscale Air Quality (WRF–CMAQ) modeling system to simulate the influence of the meteorological conditions and emission changes in Handan during the typical period (June) of O3 pollution in 2013–2018. For benchmarked June 2013, the results showed that the reduction of the O3 concentration in June of 2014–2016 was mainly caused by emission reduction, while in June of 2017–2018, the combined effect of changes in emissions and meteorological conditions led to aggravated O3 pollution. Sensitivity analysis indicated that combined VOCs and NOx emission controls would effectively reduce incremental O3 formation when the abatement ratio of VOCs/NOx should be no less than 0.84, and we found that VOCs reduction would continusouly bring about O3 decreases under various NOx reductions, but its positive sensitivity to O3 would become smaller with NOx reduction. However, the positive influence of NOx reduction on O3 would happen until NOx reduction exceeding 45–60%. The findings will be helpful in formulating emission control strategies for coping with O3 pollution in an industrial city.


Introduction
With the accelerated pace of economic globalization and regional economic integration, urbanization and industrialization in China have substantially progressed, which has led to an increase in anthropogenic emissions and atmospheric contamination from economically developed regions; in particular, regional complex air pollution characterized by fine particulate matter (PM 2.5 ) and ozone (O 3 ) has become increasingly prominent [1]. Since 2013, in order to improve the air quality in Chinese cities and weaken the consequences of air pollution on climate and the environment, the "Action plan for the prevention and control of air pollution" of the State Department has continuously improved the ambient air quality. These measures have achieved significant success; the annual average PM 2.5 declined 30-50% across China during 2013-2018 [2]. In contrast, during the same period, surface O 3 was the only pollutant with a continuously increasing annual average concentration [3,4]. According to the China Environmental Status Bulletin, the 90th percentile daily maximum 8 h average ozone concentration increased from 139 µg·m −3 in 2013 to 151 µg·m −3 in 2018, and approximately 1000 stations in China experienced serious O 3 pollution in the summer of 2013-2017 [5]. The O 3 concentration not only in China, but also in the United States [6], Europe [7], and many parts of the world, has shown consecutive scenarios, to understand the sensitivity of precursors and determine the suitable proportion of VOC and NO x emission reduction in downtown Handan.

Observation Data
In this study, pollutant observation data with a temporal resolution of 1 h were obtained in 2013-2018 from the observed urban air quality data released in real time by the China National Environmental Monitoring Center (http://106. 37.208.233:20035, accessed on 16 February 2021) [26], which is composed of the average values of 4 national controlled air quality monitoring sites in the Handan ( Figure 1) and average values of 14 national controlled air quality monitoring sites in the Beijing. Every 3 h of observation data from the Handan Meteorological Station (meteorological station 53892) were adopted based on the Meteorological Information Comprehensive Analysis and Processing System, including parameters such as temperature and wind speed.
Atmosphere 2021, 12, x FOR PEER REVIEW 3 of 18 scenarios, to understand the sensitivity of precursors and determine the suitable proportion of VOC and NOx emission reduction in downtown Handan.

Observation Data
In this study, pollutant observation data with a temporal resolution of 1 h were obtained in 2013-2018 from the observed urban air quality data released in real time by the China National Environmental Monitoring Center (http://106. 37.208.233:20035, accessed on 16 February 2021) [26], which is composed of the average values of 4 national controlled air quality monitoring sites in the Handan ( Figure 1) and average values of 14 national controlled air quality monitoring sites in the Beijing. Every 3 h of observation data from the Handan Meteorological Station (meteorological station 53892) were adopted based on the Meteorological Information Comprehensive Analysis and Processing System, including parameters such as temperature and wind speed.

Model Configuration
In this study, an integrated WRF-CMAQ modeling system was used to quantify the impact of meteorological conditions and precursor emissions on O3 pollution and assess the O3 formation sensitivity. The air quality model, including North China, was established by CMAQ v5.0.1, which is a widely used O3 simulation tool. The CMAQ simulation was performed over two nested domains ( Figure 1). The outermost domain (D1) covers most areas of the BTH urban agglomeration and surrounding areas (including most areas of Shanxi, Shandong, Henan, Inner Mongolia, and other provinces) at a 9 km × 9 km grid resolution. The innermost domain (D2) covers the entire Handan region and its surrounding areas at a 3 km × 3 km grid resolution. The meteorological input data files of CMAQ

Model Configuration
In this study, an integrated WRF-CMAQ modeling system was used to quantify the impact of meteorological conditions and precursor emissions on O 3 pollution and assess the O 3 formation sensitivity. The air quality model, including North China, was established by CMAQ v5.0.1, which is a widely used O 3 simulation tool. The CMAQ simulation was performed over two nested domains ( Figure 1). The outermost domain (D 1 ) covers most areas of the BTH urban agglomeration and surrounding areas (including most areas of Shanxi, Shandong, Henan, Inner Mongolia, and other provinces) at a 9 km × 9 km grid resolution. The innermost domain (D 2 ) covers the entire Handan region and its surrounding areas at a 3 km × 3 km grid resolution. The meteorological input data files of CMAQ were provided by the WRFv3.5.1. The National Center for Environmental Prediction (NCEP)'s Final Operational Global Analysis (FNL) data with a horizontal resolution of 1 • × 1 • at every 6 h were used to drive the WRF model. The output files of the WRF model were processed for the CMAQ adopted Meteorology Chemistry Interface Processor program. In addition, the domains of the WRF model were slightly greater than those of CMAQ. A detailed parameterization scheme is described in Appendix A (Table A1).
Moreover, downtown Handan was selected as the receptor area for analysis and discussion. The anthropogenic emission data of gaseous pollutants (SO 2 , NO x , CH 4 , CO, NH 3 and VOCs groups) and particulate matter (organic carbon, elemental carbon, PM 2.5 and PM 10 ) of the BTH region for 2013-2018 were based on the research results of this research group, and biogenic emission were calculated from the Model of Emission of Gases and Aerosols from Nature (MEGAN) [27]. The high-resolution emission inventory of the BTH region specific establishment method can refer to the previous research results of the research team [28,29], which was established based on the county level, and the evaluation of Zhou et al. [30] of this emission inventory had relatively satisfactory performance. The emission inventory of the rest of the region was taken from MEIC emission inventory dataset (http://www.meicmodel.org/, accessed on 10 November 2020). This was updated according to relevant policies released by the local government. The WRF-CMAQ model startup time was advanced by 72 h to eliminate the influence of clean initial conditions and the default clean profiles are used as the initial boundary chemical conditions. The model validation results show that WRF-CMAQ model exhibited an acceptable performance for O 3 . More details about model verification can see Appendix A (Tables A2 and A3).

Quantitative Assessment Method of Meteorological and Emission Impacts
The WRF-CMAQ model simulation system and emission inventory of Handan and its surrounding areas from June of 2013-2018 were used as the input file to establish a quantitative assessment method for the impact of emissions and meteorology on the evolution of O 3 pollution. The critical factor leading to the change in O 3 pollution in the typical period of O 3 pollution during June of 2013-2018 in downtown Handan was identified to quantify the contribution of O 3 precursors and meteorological factors to the evolution of O 3 pollution in downtown Handan and finally to provide scientific support for regional O 3 pollution control.

Simulation Scenarios Settings
In this study, the typical period of O 3 pollution during June of 2013-2018 in downtown Handan was selected as the simulation period. Two sets of simulation scenarios were set up to quantitatively evaluate the impact of meteorological conditions and emissions. Simulation scenarios A were used as the base scenario, and the corresponding meteorological conditions and emissions from June 2013 to 2018 were used as the input variables of the model. Scenarios A results were evaluated using observed data to estimate the performance of our model simulation. In scenarios B, June 2013 was selected as a constant meteorological field, and the corresponding emissions from June 2013 to 2018 were adopted as the input variables of the model (in order to ensure that the simulation results better reflect the impact of meteorological conditions on the evolution of O 3 pollution, the wind speed and temperature change from June of 2013-2018 were comprehensively considered). June 2013, which had a weak impact of meteorological conditions on O 3 pollution, was selected as the constant meteorological field of scenarios B. Table 1 shows the specific scheme settings for the simulation scenarios.  (1) and (2), and the MDA8 O 3 concentration was selected as the quantitative evaluation index to quantify the impact of simultaneous and single emission changes in meteorology and emissions on the evolution of O 3 pollution. where

Sensitivity Experimentation
In this study, the brute force method was used to evaluate the O 3 formation sensitivity in downtown Handan [25,31]. Based on June 2018 emissions condition, two scenarios of 30% VOC and 30% NO x emission reduction were designed and additionally simulated for O 3 of June 2018, to determine VOCs-NO x -O 3 sensitivity over downtown Handan. The net increment of MDA8 O 3 concentration of the two simulated scenarios was calculated by making difference with the benchmark scenario, marked them as ∆O VOC and ∆O NO x , respectively. According to the net increment of MDA8 O 3 concentration in the simulated scenarios, the basis for sensitivity analysis of O 3 formation in downtown Handan was determined as follows: ∆O NO x ≤ −4 µg·m −3 and ∆O NO x < 2 × ∆O VOC is an NO xsensitive regime ∆O VOC ≤ −4 µg·m −3 and ∆O VOC < 2 × ∆O NO x is a VOC-sensitive regime; ∆O NO x ≤ −4 µg·m −3 and ∆O VOC ≤ −4 µg·m −3 , while the difference between them is less than 2 times, is an NO x and VOC-sensitive regime; −4 µg·m −3 < ∆O NO x < 4 µg·m −3 or −4 µg·m −3 < ∆O VOC < 4 µg·m −3 is an insensitive regime; and ∆O NO x ≥ 4 µg·m −3 and 0 ≥ ∆O VOC > −4 µg·m −3 is an NO x -titration regime. A positive number indicates an increase in the MDA8 O 3 concentration and a negative number indicates a decrease in the MDA8 O 3 concentration.

Temporal Trends of Ozone Pollution in Downtown Handan
In Figure 2 February) minimum in downtown Handan and Beijing. However, the changes of O3 concentration in downtown Handan from 2013 to 2018 were more obvious than that in Beijing downtown, especially the monthly hourly average O3 concentration from January to December in 2018 were all higher than those in the same months from 2013 to 2016 in downtown Handan. We regarded the characteristics and causes of O3 concentration changes in downtown Handan during 2013-2018 were more worthy of further discussion. Thus, our analysis focused on the vulnerable month (June) of summer as a typical period and selected Handan as the research target city to conduct the following research work. In 2013, in an effort to ease mounting public concern over air quality, the "Action plan for preventing and controlling air pollution" was promulgated by the Chinese government. Because of this policy, China's air pollution improved significantly after 2013.  In 2013, in an effort to ease mounting public concern over air quality, the "Action plan for preventing and controlling air pollution" was promulgated by the Chinese government. Because of this policy, China's air pollution improved significantly after 2013. Figure

Emission Trends of Ozone Precursors during June 2013-2018
NOx and VOCs are important precursors of O3 formation. Handan implemented a strict total pollutant control system, which significantly reduced the anthropogenic emis-

Emission Trends of Ozone Precursors during June 2013-2018
NOx and VOCs are important precursors of O3 formation. Handan implemented a strict total pollutant control system, which significantly reduced the anthropogenic emissions of NOx and VOCs from 2013 to 2018. Figure 4 shows that anthropogenic NOx emissions in Handan reached approximately 21,829.5 t in June 2013, and then the NOx emissions continued decreasing yearly. The NOx emissions were reduced to 11,341.6 t in June 2018. The NOx emissions mainly originated from industrial sources (including the steel industry, building materials industry, fossil fuel combustion in industrial boilers, and petrochemical and chemical industry) and mobile sources (including road vehicles and nonroad engines) with contributions of 43.6-59.6% and 35.2-54.8% in our study period, respectively. From June of 2013-2018, the Handan NOx emissions decreased by 10,487.9 t, which was mainly from industrial sources (by 6,675.5 t) and mobile sources (by 2,772.6 t), as shown in Figure 5.  Handan. Therefore, in order to promote economic growth in the city, the control of industry was relaxed [32], which made the VOC emissions from organic solvent sources and industrial sources present a slightly increasing trend, especially the VOC emissions from organic solvent sources, which increased by 3972.5 t compared with that in the previous year. pollution, the daily average PM2.5 concentration from June 2015 to June 2017 reached the State Ambient Air Quality Secondary Standard for PM2.5 (75 μg·m −3 ) in downtown Handan. Therefore, in order to promote economic growth in the city, the control of industry was relaxed [32], which made the VOC emissions from organic solvent sources and industrial sources present a slightly increasing trend, especially the VOC emissions from organic solvent sources, which increased by 3972.5 t compared with that in the previous year. As VOCs and NOx are primary pollutants, the pollutant emission is linearly related to the concentration level, it is considered that the ratio of precursor concentration to emission is similar. The decrease in the emission of O3 precursors in Handan during June 2013-2018 was notable, but the MDA8 O3 concentration in downtown Handan showed a yearly decreasing from June 2014 to June 2015, and then showed a rapid increasing trend after 2015. It was probably whether precursor reduction brings about O3 increase of downtown Handan, for the change of VOCs-NOx-O3 formation sensitivity. The ratio of VOCs/NOx can be used to qualitatively analyze the relationship between ambient O3 concentration and VOCs and NOx. In the conditions with relatively low ratios of VOCs/NOx, the reaction of OH radicals and NOx was dominant, and O3 production was more sensitive to VOCs concentration; while in the conditions with relatively high ratios of VOCs/NOx, the reaction of OH radicals and VOCs was dominant, and O3 production was more sensitive to NOx concentration [33]. The ratio of VOCs emissions (anthropogenic VOCs and biogenic In the conditions with relatively low ratios of VOCs/NO x , the reaction of OH radicals and NO x was dominant, and O 3 production was more sensitive to VOCs concentration; while in the conditions with relatively high ratios of VOCs/NO x , the reaction of OH radicals and VOCs was dominant, and O 3 production was more sensitive to NO x concentration [33]. The ratio of VOCs emissions (anthropogenic VOCs and biogenic VOCs) to NO x emissions was calculated, as shown in Figure 6. During 2013-2018, the change of VOCs/NO x was obvious in eastern and southern suburban areas, and in June 2013 the high VOCs/NO x ratio of >7.0 occurred in eastern and southern suburban areas, while these suburban areas seemed to move towards a lower VOCs/NO x condition in the next five years as the anthropogenic emissions changed. Low ratios of VOCs/NO x ranging from 2.0 to 3.0, continuously occurred in downtown Handan during 2013-2018, which excluded the possibility that the change of ozone concentration in downtown Handan in recent years was caused by the change of VOCs-NO x -O 3 formation sensitivity.

Contribution of Meteorology and Emissions to Ozone Trends
The evolution of O 3 pollution has a highly non-linear relationship between NO x and VOC emissions. In addition, the concentration of O 3 is closely related to meteorological conditions [34]. Therefore, calculating the correct quantitative contribution of interannual variations in O 3 precursor emissions and meteorology conditions to O 3 pollution evolution is the key to identifying the main influencing factors of O 3 pollution evolution in Handan in recent years. change of VOCs/NOx was obvious in eastern and southern suburban areas, and in June 2013 the high VOCs/NOx ratio of >7.0 occurred in eastern and southern suburban areas, while these suburban areas seemed to move towards a lower VOCs/NOx condition in the next five years as the anthropogenic emissions changed. Low ratios of VOCs/NOx ranging from 2.0 to 3.0, continuously occurred in downtown Handan during 2013-2018, which excluded the possibility that the change of ozone concentration in downtown Handan in recent years was caused by the change of VOCs-NOx-O3 formation sensitivity.

Contribution of Meteorology and Emissions to Ozone Trends
The evolution of O3 pollution has a highly non-linear relationship between NOx and VOC emissions. In addition, the concentration of O3 is closely related to meteorological conditions [34]. Therefore, calculating the correct quantitative contribution of interannual variations in O3 precursor emissions and meteorology conditions to O3 pollution evolution is the key to identifying the main influencing factors of O3 pollution evolution in Handan in recent years.
Downtown Handan was selected as the receptor area, and the typical period of O3 pollution (June) in 2013-2018 was selected as the simulation period for simulation scenario cases (for specific case settings, see Section 2.3.1). Figure 7a Figure 7b shows the quantitative assessment results of the impact of meteorological conditions and emission changes on the evolution of O3 pollution in downtown Handan. Compared with the meteorological conditions in June 2013, the contributions of meteorological conditions to the MDA8 O3 concentration in June 2014-2018 were 12.9 μg·m −3 , 18.9 μg·m −3 , 11.6 μg·m −3 , 11.2 μg·m −3 , 16.9 μg·m −3 had promotional effects; in particular, the contribution of meteorological conditions in June 2015 to the MDA8 O3 concentration had a big effect. The annual variation in the absolute value of O3 concentration caused by emissions was greater than that caused by meteorological conditions. For June

Sensitivity Regime Identification
Changes in VOC and NO x concentrations will affect their ambient concentrations and change the O 3 formation sensitivity, thereby leading to considerable variation in the O 3 concentration [35]. Therefore, it is crucial to clarify the O 3 formation sensitivity before formulating effective O 3 control strategies. Figure 6 shows that the ratios of VOCs/NO x in downtown Handan were similar from June 2013 to 2018, which indicated that the O 3 formation sensitivity had not changed much in downtown Handan. The CMAQ model was adopted to analyze the sensitivity of O 3 formation in downtown Handan during June 2018 for Handan to formulate prevention and control strategies after 2018. As shown in Figure 8, the ∆O VOC concentration was within −43.5 to −1.2 µg·m −3 , with an average of −18.2 µg·m −3 , 30% VOC emission reduction caused a decrease in the MDA8 O 3 concentration. The ∆O NO x concentration was within −6.8 to 24.4 µg·m −3 , 30% NO x emission reduction caused the daily average ∆O NO x concentration to rise or decrease, in terms of the monthly average value of ∆O NO x was 12.2 µg·m −3 , that is, the reduction in 30% NO x emissions will cause the MDA8 O 3 concentration to increase. According to the definition Section 2.4, the O 3 formation sensitivity in downtown Handan was a VOC-sensitive regime during June 2018. The implication is that cutting VOC emissions can effectively control the concentration of O 3 in downtown Handan.
for Handan to formulate prevention and control strategies after 2018. As shown in Figure  8, the ΔOVOC concentration was within −43.5 to −1.2 μg·m −3 , with an average of −18.2 μg·m −3 , 30% VOC emission reduction caused a decrease in the MDA8 O3 concentration. The ΔONOx concentration was within −6.8 to 24.4 μg·m −3 , 30% NOx emission reduction caused the daily average ΔONOx concentration to rise or decrease, in terms of the monthly average value of ΔONOx was 12.2 μg·m −3 , that is, the reduction in 30% NOx emissions will cause the MDA8 O3 concentration to increase. According to the definition Section 2.4, the O3 formation sensitivity in downtown Handan was a VOC-sensitive regime during June 2018. The implication is that cutting VOC emissions can effectively control the concentration of O3 in downtown Handan.

Development of Control Strategies on both VOCs and NOx
The above discussion based on single species (VOCs or NOx) emission reduction demonstrated that O3 formation was quite sensitive to VOCs in downtown Handan, and only reducing VOC emissions can decrease the concentration of O3. However, in reality, the control strategy of just reducing VOC emissions is difficult to achieve, as society tends to control VOCs and NOx simultaneously. Owing to the non-linear response of O3 to changes in emissions of its precursors, an improper abatement ratio of VOCs/NOx was applied, and the O3 concentration could actually increase [35]. It is unclear how much VOC and NOx emissions should be controlled to achieve an efficient O3 decrease. To address this question, based on the atmospheric pollutant emission inventory of Handan and FNL data in June 2018, we used the WRF-CMAQ model simulation to obtain the MDA8 O3 concentration by adjusting the input anthropogenic VOC and NOx ratios, and the difference between the results of the MDA8 O3 concentration and that of the benchmark scenario (VOC and NOx emissions are not reduced) was used to acquire the net

Development of Control Strategies on Both VOCs and NO x
The above discussion based on single species (VOCs or NO x ) emission reduction demonstrated that O 3 formation was quite sensitive to VOCs in downtown Handan, and only reducing VOC emissions can decrease the concentration of O 3 . However, in reality, the control strategy of just reducing VOC emissions is difficult to achieve, as society tends to control VOCs and NO x simultaneously. Owing to the non-linear response of O 3 to changes in emissions of its precursors, an improper abatement ratio of VOCs/NO x was applied, and the O 3 concentration could actually increase [35]. It is unclear how much VOC and NO x emissions should be controlled to achieve an efficient O 3 decrease. To address this question, based on the atmospheric pollutant emission inventory of Handan and FNL data in June 2018, we used the WRF-CMAQ model simulation to obtain the MDA8 O 3 concentration by adjusting the input anthropogenic VOC and NO x ratios, and the difference between the results of the MDA8 O 3 concentration and that of the benchmark scenario (VOC and NO x emissions are not reduced) was used to acquire the net increment in the MDA8 O 3 concentration in downtown Handan. Figure 9 illustrates that the net increment in the MDA8 O 3 concentration increased as the abatement ratio of NO x increased from 0% to 45% regardless of the decrease in the abatement ratio of VOC emissions from 0% to 75%, whereas the net increment in the MDA8 O 3 concentration declined gradually when the abatement ratio of NO x reduction was greater than 60%.
Under the same abatement ratio of NO x emissions, a higher cutdown abatement ratio of VOC emissions always causes a greater decrease in the net increment in the MDA8 O 3 concentration. In particular, with the abatement ratio of VOCs >30%, the net increment in the MDA8 O 3 concentration declined obviously. Under the same VOC emission reduction rate, the net increment of MDA8 O 3 concentration increased with the abatement ratio of NO x emissions from 0% to 45% (or 60%), indicating that VOCs reduction positive sensitivity to O 3 would become smaller with NO x emission reduction. Therefore, the increase in O 3 concentration in 2016 compared to 2015 when the VOC and NO x emissions continue to decrease by 51.1% and 11.2% respectively as suggested in Figure 4, which may be caused by improper NO x emission reduction. In addition, the "inflection point" that occurred in the abatement ratio of NO x reduction was 45% when the reduction range was 0-30% for VOC emissions, while the "inflection point" that occurred in 60% of the abatement ratio of NO x reduction was when the abatement ratio of VOC reduction was ≥45%. He et al. [35] found similar results in the Pearl River Delta region. This might be due to the weaker atmospheric oxidizing capacity under an excessive abatement ratio of VOC (45-75%) conditions than under the reduction range of 0-30% for VOC emissions, which weakened the NO 2 removal through NO 2 + hydroxyl radical (OH)→HNO 3 . Relatively abundant NO 2 generated more nitric oxide (NO) via photolysis reaction that would promote the titration reaction between NO and O 3 . In other words, the excessive abatement ratio of VOC would enlarge the effect range of the NO titration reaction, under this scenario continued to reduce NO x emissions, would reduce the effect of O 3 concentration reduction. It means co-reduction of VOC and NO x is essential, and VOC emission reduction rate ranges from 30% to 45%, resulting in the O 3 improvement.
Atmosphere 2021, 12, x FOR PEER REVIEW 13 of 18 increment in the MDA8 O3 concentration in downtown Handan. Figure 9 illustrates that the net increment in the MDA8 O3 concentration increased as the abatement ratio of NOx increased from 0% to 45% regardless of the decrease in the abatement ratio of VOC emissions from 0% to 75%, whereas the net increment in the MDA8 O3 concentration declined gradually when the abatement ratio of NOx reduction was greater than 60%. NO x reduction percentage (%) 0% VOCs reduction 15% VOCs reduction 30% VOCs reduction 45% VOCs reduction 60% VOCs reduction 75% VOCs reduction Under the same abatement ratio of NOx emissions, a higher cutdown abatement ratio of VOC emissions always causes a greater decrease in the net increment in the MDA8 O3 concentration. In particular, with the abatement ratio of VOCs >30%, the net increment in the MDA8 O3 concentration declined obviously. Under the same VOC emission reduction rate, the net increment of MDA8 O3 concentration increased with the abatement ratio of NOx emissions from 0% to 45% (or 60%), indicating that VOCs reduction positive sensitivity to O3 would become smaller with NOx emission reduction. Therefore, the increase in O3 concentration in 2016 compared to 2015 when the VOC and NOx emissions continue to decrease by 51.1% and 11.2% respectively as suggested in Figure 4, which may be caused by improper NOx emission reduction. In addition, the "inflection point" that occurred in the abatement ratio of NOx reduction was 45% when the reduction range was 0-30% for VOC emissions, while the "inflection point" that occurred in 60% of the abatement ratio of NOx reduction was when the abatement ratio of VOC reduction was ≥45%. He et al. [35] found similar results in the Pearl River Delta region. This might be due to the weaker atmospheric oxidizing capacity under an excessive abatement ratio of VOC (45-75%) conditions than under the reduction range of 0-30% for VOC emissions, which weakened the NO2 removal through NO2 + hydroxyl radical (OH)→HNO3. Relatively abundant NO2 generated more nitric oxide (NO) via photolysis reaction that would promote the titration reaction between NO and O3. In other words, the excessive abatement The development of VOC and NO x emission reduction strategies must ensure continuous reduction or at least no increase in O 3 concentration, and the adoption of an unreasonable abatement ratio of VOCs/NO x would cause an increase in the O 3 concentration. The net increment in the MDA8 O 3 concentration decreased as the abatement ratio of VOCs was reduced by ≥45% regardless of the abatement ratio of NO x or when the abatement ratio of NO x was ≥75%. Considering the policy feasibility, it was still challenging to achieve the emission targets in the current stage. Accordingly, we focused on the reduction range of 0-50% for NO x and 0-45% for VOCs to devise optimal control measures of VOC and NO x emissions. It was determined that when the abatement ratio of NO x was 0% to 60%, the minimum abatement ratio of VOCs/NO x for a zero MDA8 O 3 increment changed from 0.59 to 0.84 (the abatement ratios of VOCs/NO x at the intersections of the curves and the horizontal axis). This indicated that the abatement ratio of VOCs/NO x should be greater than at least 0.84 to prevent an increase in the O 3 concentration in downtown Handan. Similar results showed that the abatement ratio of VOCs/NO x should be no less than 0.72 in Shanghai [18], in the Heshan site of the Pearl River Delta region it should be higher than 1.10 to prevent the increase in the O 3 levels [35], and that in Wuhan it was 1.1 [36], thereby suggesting that different cities should try to control the increase in O 3 and they will need to formulate a localized abatement ratio of VOCs and NO x emissions.

Conclusions
Handan as a typical industrial city in China has been experiencing worsening summer O 3 pollution recently. We found that emissions of anthropogenic VOC and NO We found that downtown Handan is a typical VOC-sensitive, O 3 -polluted region, and the reduction in VOC could lead to a monotonous decrease in the MDA8 O 3 concentration. Therefore, controlling VOC emissions can effectively reduce the formation of O 3 . Afterwards, we designed 36 precursor reduction scenarios based on 2018 emissions and further simulated O 3 over downtown Handan under these reduction scenarios to find that VOCs reduction would continuously bring about O 3 decreases under various NO x reductions, but its positive sensitivity to O 3 would become smaller with NO x reduction. However, the positive influence of NO x reduction on O 3 would happen until NO x reduction exceeding 45-60%. Thus, considering that the control strategy of only reducing VOC emissions is difficult to achieve, so the abatement ratio of VOCs/NO x should be no less than 0.84 to prevent net O 3 increment in downtown Handan. Overall, this study will help local government policymakers to put forward effective control measures for O 3 pollution in Handan or other regions of China.
Author Contributions: Writing-original draft preparation, writing and editing: S.Y. Writingreview and editing, supervision: W.W. and S.C. Data curation: Y.N. and P.G. All the authors contribute to interpretation, discussion, review and editing the manuscript. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest. Based on the observed data of meteorological and pollutants parameters from June 2013-2018 in Handan (the observed data are detailed in Section 2.1), the performance of our model simulation was evaluated. The normalized mean bias (NMB), the normalized mean error (NME) and the correlation coefficient (R) were used to evaluate the performance in simulating horizontal distribution of the WRF and CMAQ simulation results. Table A2 shows the performance statistics for hourly-averaged temperature (T), relative humidity (RH) at 2 m and wind speed (WS) at 10 m during June 2013-2018 in Handan. T, RH and WS were widely applied and extensively discussed for the verification of the WRF model [37]. T with an NMB range of −4.0-3.0% and an R range of 0.74-0.89, the results were better than those of Wang et al. [38], but slightly worse than those of Sun et al. [39]. RH with NMB varying from −2.1-11.1% and an R range of 0.82-0.89. WS simulation results were all higher than the observations, and there was a clear overestimation, while the results basically are consistent with a previous study [40]. In general, the WRF model simulation could be used for the CMAQ model.

Appendix A
The performance of the CMAQ model was assessed by comparing the observations and the simulation results for MDA8 O 3 concentration. Table A3 shows that MDA8 O 3 concentration with an NMB range of −11.7% to −0.8% and an R range of 0.68-0.78, and that the MDA8 O 3 simulation results were underestimated to a certain extent. All NMBs and Rs are within the threshold for satisfactory performance proposed by recommended benchmarks for photochemical model performance statistics [41]. Overall, the WRF-CMAQ model exhibited an acceptable performance for O 3 .