Effectiveness of emission control in reducing PM2.5 pollution in central China during winter haze episodes under various potential synoptic controls

Currently, mitigating severe particle pollution in autumn and winter is the key to further improving the air quality of China. The source contributions and transboundary transport of fine particles (PM2.5) in pollution episodes are closely related to large-scale or synoptic-scale atmospheric circulation. How to effectively reduce emissions to control haze pollution under different synoptic conditions is rarely reported. In this study, we classify the synoptic conditions over central China from 2013 to 2018 by using Lamb–Jenkinson method and the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Final (FNL) operational global analysis data. The effectiveness of emission control to reduce PM2.5 pollution during winter haze episodes under potential synoptic controls is simulated by GEOS-Chem model. Among the 10 identified synoptic patterns, four types account for 87 % of the total pollution days. Two typical synoptic modes are characterized by low surface wind speed and stable weather conditions or high relative humidity (A or C type) over central China due to a high-pressure system or a southwest trough and low-pressure system, blocking pollutants dispersion. Sensitivity simulations show that these two heavy pollution processes are mainly contributed by local emission sources with ∼ 82 % for A type and ∼ 85 % for C type, respectively. The other two patterns lead to pollution of transport characteristics affected by northerly or southerly winds (NW or SW type), carrying air pollution from northern or southern China to central China. The contribution of pollution transmission from northern and southern China is 36.9 % and 7.6 %, respectively of PM2.5, and local emission sources contribute 41 % and 69 %, respectively. We also estimate the effectiveness of emission reduction in these four typical severe pollution synoptic processes. By only reducing SO2 and NOx emission and not controlling NH3, the enhanced nitrate counteracts the effect of sulfate reduction on PM2.5 mitigation, with a less than 4 % decrease in PM2.5. In addition, to effectively mitigate haze pollution of NWand SW-type synoptic-controlled episodes, local emission control actions should be in coordination with regional collaborative actions. Published by Copernicus Publications on behalf of the European Geosciences Union. 3144 Y. Yan et al.: Effectiveness of emission control in reducing PM2.5 pollution

Abstract. Currently, mitigating severe particle pollution in autumn and winter is the key to further improving the air quality of China. The source contributions and transboundary transport of fine particles (PM 2.5 ) in pollution episodes are closely related to large-scale or synoptic-scale atmospheric circulation. How to effectively reduce emissions to control haze pollution under different synoptic conditions is rarely reported. In this study, we classify the synoptic conditions over central China from 2013 to 2018 by using Lamb-Jenkinson method and the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Final (FNL) operational global analysis data. The effectiveness of emission control to reduce PM 2.5 pollution during winter haze episodes under potential synoptic controls is simulated by GEOS-Chem model. Among the 10 identified synoptic patterns, four types account for 87 % of the total pollution days. Two typical synoptic modes are characterized by low surface wind speed and stable weather conditions or high relative humidity (A or C type) over central China due to a high-pressure system or a southwest trough and low-pressure system, blocking pollutants dispersion. Sensitivity simulations show that these two heavy pollution processes are mainly contributed by local emission sources with ∼ 82 % for A type and ∼ 85 % for C type, respectively. The other two patterns lead to pollution of transport characteristics affected by northerly or southerly winds (NW or SW type), carrying air pollution from northern or southern China to central China. The contribution of pollution transmission from northern and southern China is 36.9 % and 7.6 %, respectively of PM 2.5 , and local emission sources contribute 41 % and 69 %, respectively. We also estimate the effectiveness of emission reduction in these four typical severe pollution synoptic processes. By only reducing SO 2 and NO x emission and not controlling NH 3 , the en-1 Introduction The regional pollution of fine particles (PM 2.5 ) has attracted worldwide attention from the public and in the scientific community Li et al., 2017c;Lin et al., 2018;Bi et al., 2019) due to its detrimental effect on visibility  and public health (Agarwal et al., 2017;. PM 2.5 pollution in China has been continuously reducing since 2013 as a result of the implementation of the Air Pollution Prevention and Control Action Plan Zhang et al., 2019), especially in the Beijing-Tianjin-Hebei region (BTH) (Li et al., 2017b;Cheng et al., 2019), the Yangtze River Delta (YRD) region, and the Pearl River Delta (PRD) region. However, severe particle pollution still occurs frequently in autumn and winter, which is the major reason restricting the PM 2.5 from reaching the national standard. For example, 12 extremely severe and persistent PM 2.5 pollution episodes occurred in Beijing in January 2013, February 2014, December 2015, December 2016, and January 2017 (Zhong et al., 2018;Sun et al., 2016;Wang et al., 2018). Currently, researching how to effectively reduce emissions in autumn and winter is the key to mitigating haze pollution in China.
The contribution of emission sources has been widely recognized as the decisive factor of PM 2.5 pollution over urban agglomerations, including industrial exhaust, urban transportation, residential emission, power plants, agricultural activities, and bio-combustion (Huang et al., 2014;Tian et al., 2016;Wu et al., 2018;An et al., 2019). The outbreak, persistence, and dissipation of particle pollution generally depends on meteorological conditions and regional synoptic patterns controlled by the large-scale or synoptic-scale atmospheric circulation (Chuang et al., 2008;Zhang et al., 2012;Russo et al., 2014;Zheng et al., 2015;Shu et al., 2017;Li et al., 2019).
Many studies have tried to reveal the relationship between synoptic patterns and severe particle pollution and estimate the meteorological contributions to these pollution episodes. The YRD is mainly affected by pollutants transmitted from northern and the southern China when the East Asian major trough is located at its front (Liao et al., 2017;Shu et al., 2017;Li et al., 2019). Liao et al. (2020) have confirmed that the relative position of the PRD to high-pressure systems imposes significant impacts on the diffusion conditions and the PM 2.5 distributions in the PRD region. For the North China Plain (NCP), a high frequency of stagnant weather accompanied by a small pressure gradient and near-surface wind speed and a shallow mixing layer are major reasons of aerosol pollution over this region in winter . The aerosol pollution formation process in the Sichuan Basin (SCB) is often controlled by the large-scale high-pressure circulation at sea level (Sun et al., 2020). In the Guanzhong Basin, pollution events are generally governed by both the large-scale synoptic situation and the small-scale local circulation. The downhill wind not only forms a convergence zone in the basin but also makes pollutants flow back from the mountain region to the basin (Bei et al., 2017). Leung et al. (2018) also find strong correlations of daily PM 2.5 variability with several synoptic patterns, including monsoon flows and cold-front channels in northern China related to the Siberian High, onshore flows in eastern China, and frontal rainstorms in southern China. These previous studies have highlighted that different levels of PM 2.5 pollution are closely related to the dominant synoptic patterns in different regions, and they attribute the large spatial variability of pollution to the regional transport contributions and the different local sources of PM 2.5 . Thus, heavy pollution prevention and control needs to consider the weather situation, otherwise local emission reduction measures will not work well. However, how to effectively reduce local and regional emissions under different synoptic conditions to control haze pollution is rarely reported.
Various key regions have issued an emergency preplan against the winter haze episodes, but these schemes can only be targeted at a certain city (The People's Government of Beijing Municipality, 2018; The People's Government of Shanghai Municipality, 2018) or a certain urban agglomeration (The People's Government of Guangdong Province, 2014). Although there are many studies targeted PM 2.5 mitigation at a regional scale Zhang et al., 2019;Xing et al., 2018Xing et al., , 2019Fu et al., 2017;etc.), their results can not be directly applied to reduce winter PM 2.5 pollution under various synoptic controls. Moreover, current emission reduction policies in China mainly target sulfur dioxide (SO 2 ) and nitrogen dioxide (NO 2 ) reduction, ignoring the effective emission reduction for ammonia (NH 3 ), although some modeling work has discussed the effectiveness of ammonia emission reduction for PM 2.5 mitigation Ye et al., 2019;. Compared to remarkable reduction in SO 2 , NO 2 , and primary PM emissions, NH 3 emissions has remained stable during 2014-2018 in China . In addition, given the important roles of other PM 2.5 precursors (e.g., non-methane volatile organic compounds) in aerosol formation , cutting emissions other than SO 2 , NO 2 , and PM should be proposed as a next-step mitigation strategy. Therefore, for PM 2.5 mitigation in a specific region during winter haze episodes forced by various synoptic conditions, whether the air pollution emergency management and control schemes are effective and how to improve them have become an urgent scientific questions to be answered.
In order to investigate the effectiveness of emission control for reducing PM 2.5 pollution during winter haze episodes under various potential synoptic controls (PSCs), we take the severe particle pollution of winter haze episodes over Jingzhou, in the hinterland of the Yangtze River middle basin in central China, as an example. Central China is geographically surrounded by major haze pollution regions: the SCB to the west, the PRD to the south, the YRD to the east, and the NCP to the north (Fig. 1). As a regional pollutant trans- Figure 1. The location of Jingzhou (red area) and the major haze pollution regions of the NCP, YRD, PRD, and SCB. The areas framed in red are used to investigate the inter-regional impacts using GEOS-Chem sensitivity simulations. The overlaid map shows the surface elevation (m) from 2 min gridded global relief data (ETOPO2v2) available from the NGDC Marine Trackline Geophysical database (http: //www.ngdc.noaa.gov/mgg/global/etopo2.html, last access: 27 February 2021). port hub with sub-basin topography, central China is a region of transmission pollution characteristics affected by two reported transport pathways from the vast flatland in central eastern China (Yu et al., 2020) and from the NCP region (H. . In combination with high anthropogenic emissions  and secondary aerosol formation (Huang et al., 2020), central China often suffers severe pollution episodes in winter caused by PM 2.5 (Gong et al., 2015;Xu et al., 2017). In this study, we conduct a circulation classification to differentiate the synoptic modes during the severe particle pollution episodes in winter over central China from 2013 to 2018 by using the Lamb-Jenkinson method. Following this, we simulate the PM 2.5 chemical components and the contributions of local sources and transboundary transport of PM 2.5 under different synoptic conditions. Finally, the effectiveness of emission reduction in the main potential synoptic patterns is evaluated by GEOS-Chem model simulations. This study combines atmospheric (circulation classification) and environmental (chemical transport modeling) research methods and could provide a reference for emission control of severe winter haze pollution under different weather types and provide a basis for regional air quality policy-making.

Data
Hourly mass concentrations of PM 2.5 at Jingzhou (112.18 • E, 30.33 • N, 33.7 m) from November 2013 to December 2018 are obtained from Hubei Environmental Monitoring Central Station (http://sthjt.hubei.gov.cn/, last access: 27 February 2021). We screen the pollution days with daily mean PM 2.5 concentrations larger than 150 µg/m 3 for circulation classification.
The meteorological data of surface observations at Jingzhou, including ambient temperature, relative humidity, wind speed, wind direction, and atmospheric pressure, are obtained from the Hubei Meteorological Information and Technology Support Center (http://hb.cma.gov.cn/qxfw/ index.html, last access: 27 February 2021). The data from November 2013 to February 2014 are used to analyze the me-3146 Y. Yan et al.: Effectiveness of emission control in reducing PM 2.5 pollution teorological characteristics during the period that four severe particle pollution events occurred in succession over central China (Fig. S1).
In order to better evaluate the GEOS-Chem model performance, we also use the PM 2.5 observations (a total of 633 sites; from November 2013 to February 2014) from the Ministry of Ecology and Environment of China (MEE, http://www.mee.gov.cn/, last access: 27 February 2021) to conduct the model-observation comparison.

Lamb-Jenkinson circulation classification
The atmospheric circulation classification adopts the Lamb-Jenkinson method proposed by Lamb et al. (1950) and developed by Jenkinson et al. (1977). Compared to the objective classification method principal component analysis (PCA) used in some studies (Chang and Zhan, 2017;Dai et al., 2021), this Lamb-Jenkinson method is a combination of subjective and objective methods. After the objective judgment of the circulation, we also make subjective considerations to overcome the weaknesses of the method, leading to better synoptic significance. Many works of circulation classification have used the Lamb-Jenkinson method and reported that the analysis can respond well to the classification results (Philipp et al., 2016;Santurtun et al., 2015;Pope et al., 2015;Russo et al., 2014;Pope et al., 2014;Trigo and DaCamara, 2000).
To calculate the circulation types of Jingzhou, we mark total 16 points (97.5-127.5 • E, 20-40 • N) at every 10 • longitudes and 5 • latitude with a center point located at 112.5 • E and 30 • N (Fig. S2). Using the sea level pressure of 16 points, we calculate six circulation indexes using the following scheme of central difference: where P (n)(n = 1, 2, 3· · ·16) is the sea level pressure at the nth point; α, α 1 , and α 2 are the latitude values of points C, A 1 , and A 2 , respectively; V is the geostrophic wind; u and v are the latitudinal and meridional components of the geostrophic wind; ξ is the geostrophic vorticity; ξ u is the u meridional gradient; and ξ v is the v latitudinal gradient. Taking the latitude of the center point as the reference frame, the unit of the six circulation indexes is hectopascals per 10 • longitude, the direction of geostrophic wind can be determined by u and v, and cyclones and anticyclones can be determined by ξ . According to the geostrophic wind speed, wind direction, and vorticity value, the circulation is divided into 10 types. The classification standard and corresponding types are shown in Table 1.

GEOS-Chem simulations
The GEOS-Chem chemistry transport model is used (http:// acmg.seas.harvard.edu/geos/, last access: 27 February 2021) to simulate the spatiotemporal distribution of PM 2.5 . The nested model, covering China (15 • S-55 • N, 70-140 • E), is run with a horizontal resolution of 0.25 • latitude × 0.3125 • longitude and 72 vertical layers. The boundary condition of the nested model is provided by the GEOS-Chem global model with a horizontal resolution of 2 • latitude × 2.5 • longitude (Fig. S3). Both global and nested simulations, driven by the GEOS-FP assimilated meteorological data, include detailed tropospheric ozone, NO x , volatile organic compounds (VOCs), HO x , and aerosol chemistry. More details are shown in Yan et al. (2019). In the models, anthropogenic and natural sources are fully considered in GEOS-Chem. Tables S1 and S2 show a list of emission inventories in the global model and nested simulation, respectively. In China, the monthly grid data of 0.25 • × 0.25 • from the MEIC inventory (http://meicmodel.org, last access: 27 February 2021) for CO, NO x , SO 2 , and non-methane volatile organic compounds (NMVOCs) in 2013-2014 is used. Over central China, anthropogenic sources of these species are from our group SEEA (Source Emission and Environment Research) inventory with grid data of 0.1 • × 0.1 • (not shown). The SEEA emission inventory was developed based on the year 2017 for the Wuhan city cluster, and it has been successfully adopted for the air quality simulation and forecasting of the 7th CISM Military World Games in 2019. Other emission descriptions are shown in Sect. S1 in the Supplement.
In order to better simulate the spatiotemporal distribution of PM 2.5 over central China, especially during heavy pollution winter periods, the standard v11-01 of GEOS-Chem is optimized according to the actual situation in China (see details in Sect. S2 in the Supplement), including optimizing PM 2.5 sources and increasing the proportion of sulfate primary emission . The PM 2.5 primary anthropogenic emissions enhance the PM 2.5 concentrations over central China by 5-20 µg/m 3 in winter (Fig. S4). Compared with the results before the model optimization (Fig. S5), the sulfate concentration simulated by the optimized model increased from 10-20 to 30-50 µg/m 3 . Further comparisons of PM 2.5 with observations and inorganic salts (sulfate, nitrate, anticyclone-west (AW) 3 Results and discussion

Classification of PSCs
As shown in Fig. 2, among the circulation patterns of pollution days in Jingzhou from 2013 to 2018, the frequency of SW-type circulation is the highest, accounting for 29 % of the total pollution days. The frequencies of NW type, A type, and C type are also high, accounting for 27 %, 19 %, and 12 %, respectively. The other six circulation patterns are less frequent, with frequencies of less than 5 %. Thus, the above four typical circulation types are considered the main potential synoptic controls on severe particle pollution episodes over central China.

Characteristics of the four main PSCs
SW-type circulation is the predominant PSC of severe PM 2.5 pollution episodes. The circulation at 500 hPa is relatively flat, and the whole East Asia region is affected by the westerly flow (Fig. S6a). The westerly belt fluctuates greatly at 700 hPa, and there are two ridges and a southwest trough in the middle latitudes of Asia (Fig. S7a). Jingzhou is located in the front of a trough, prevailing the weak southwest airflow. At 850 hPa, the cold high-pressure center is formed in Xinjiang, China. Warm low pressure appears in the low-latitude area and weak high pressure appears in the East China Sea (Fig. 3a). In combination with the surface field, a high-lowhigh saddle-like field forms from west to east (Fig. 4a). Such a synoptic type is also the dominant weather system of eastern China (Shu et al., 2017;Yang et al., 2018). Jingzhou is located in the back of the Bohai Sea northeast high-pressure area and the front of southwest warm low-pressure area. Thus, it is affected by the southerly airflow, which could be conducive to the transport of air pollutants formed over southern China to central China. Associated with low local surface wind speed (< 3 m/s) at Jingzhou, the dispersion of local and transported pollutants is inhibited. NW-type circulation mainly occurs in the early winter (December and January). This synoptic pattern is also reported as one of the main types to affect the aerosol distributions over eastern China . Circulation at 500 hPa is controlled by one trough and one ridge, with the weak ridge located in the northwest of China and the shallow trough located in the northeast of China (Fig. S6b). All of East Asia is affected by the westerly current. The trough and ridge at 700 hPa are deepened. Jingzhou is located at the bottom of the shallow trough, prevailing the west-northwest airflow, and affected by the flow around the plateau (Fig. S7b). At 850 hPa, the cold high-pressure center is formed in Xinjiang, and Jingzhou is affected by the northerly airflow, due to being in the front of the high-pressure area (Fig. 3b). For the sea level pressure, the cold high-pressure area is located in western of Mongolia and Xinjiang (Fig. 4b). Jingzhou is located in a region with weak fluctuation in the front of the high-pressure area, and the surface wind speed is slower than 2 m/s. The haze episodes induced by the NW-type synoptic pattern are similar to the transmission-accumulation pollution caused by the SW type, but the transmission path is from northern China to central China.
A-type circulation also mainly occurs in the early winter. The high-altitude circulation field is controlled by one trough and one ridge (Figs. S6c and S7c). East Asia is affected by west-northwest air flow, and the SLP is controlled by a huge high-pressure area, with the center located in the southwest of Lake Baikal (Fig. 4c). A surface high-pressure area favors accumulation of air pollutants, especially over the regions with a high-pressure center (Leung et al., 2018). Jingzhou is in a sparse pressure field in front of the high-pressure area (Figs. 3c and 4c), with an average surface wind speed of ∼ 1.3 m/s. The uniform west-northwest air flow at high altitude leads to low water vapor content and fewer clouds, which is conducive to radiation cooling at night. In addition, due to the weak high-pressure ridge in the north, it is not conducive to the eastward and southward movement of cold air, leading to the stable weather situation and thus severe haze pollution at Jingzhou. This type is also responsible for most of the severe particulate pollution days in the BTH and YRD regions .
C-type circulation mainly occurs in late winter and early spring, when the relative humidity is large with an average value of 74 %. East Asia is controlled by the straight westerly flow, and the southwest shallow trough is obvious at 500 hPa (Fig. S6d). Additionally, the West Pacific subtropical high extends to the west, and central China is affected by the southwest flow. The southwest trough is deepened at 700 hPa, and Jingzhou is located in front of the trough and controlled by the southwest airflow (Fig. S7d). High pressure to the south of Xinjiang and the north of the Tibetan Plateau is strengthened at 850 hPa, and the southwest low-pressure center is formed there (Fig. 3d). Jingzhou is located in the low-pressure system on the SLP field (Fig. 4d), with low surface wind speed (0-3 m/s). Together with the large relative humidity, which can promote the hygroscopic growth of particulate matter (Twohy et al., 2009;Zheng et al., 2015), the haze pollution is persistent and serious at Jingzhou. The impact of low-pressure systems on winter heavy air pollution has also been reported in the northwestern Sichuan Basin (Ning et al., 2018).

PM 2.5 and chemical components under the four main PSCs in control simulations
The spatiotemporal distribution of PM 2.5 and its components under the four typical synoptic controls over central China was simulated by an optimized GEOS-Chem model. In order to reduce the simulation cost, the continuous four severe haze episodes that occurred during Novem-  and grade 6 heavy pollution (PM 2.5 > 250 µg/m 3 , as shown in Fig. 5a and S1a ). The simulation time is started at 1 November 2013, with the first 2 weeks used as spin up to eliminate the impact of initial conditions. The daily and hourly mean PM 2.5 concentrations at Jingzhou in the four typical heavy pollution processes simulated by the control (CON) simulation (Table 2) are compared with the observations (Fig. 5a and Fig. S1a, respectively). The model underestimates the observed PM 2.5 concentrations (by 43.3 µg/m 3 on average), especially in the high PM 2.5 periods (by 116.8 µg/m 3 at the maximum, which occurred from 21 to 23 November 2013). The possible causes for underestimation are meteorological field deviations (an overestimate in temperature and wind speed and an underestimate in humidity; Table S3) and emission errors. Anthropogenic emissions for PM 2.5 precursors used here are for the year 2017 over central China from the SEEA inventory (Table S4). From 2013 to 2017, anthropogenic NO x , SO 2 , and primary PM 2.5 emissions in central China have declined substantially (Table S4) due to implementation of stringent emission control measures for the 12th and 13th Five-Year Plan . The anthropogenic emissions biases may affect our simulations and PM 2.5 attribution results to some extent. Additionally, the underestimation is on a national scale when compared with the MEE observations, with a bias of −29.3, −18.7, −39.0, and −21.4 µg/m 3 on average for SW-type, NW-type, A-type, and C-type synopticcontrolled episodes, respectively (Fig. 6). The national negative biases may be also attributed to the insufficient resolution of the model (Yan et al., 2014) and imperfect chemical mechanisms . Nevertheless, the model can reproduce the evolution of each severe particle pollution episode well, including the accumulation of pollutants, the continuing process and the gradual dissipation of pollution ( Fig. 5a/Fig. S1a).
In order to examine the model performances in the PM 2.5 chemical compositions, we have reviewed the reported concentrations of PM 2.5 and the three inorganic salts (sulfate, nitrate, and ammonium) in other cities (Table 4). The contributions of sulfate, nitrate, and ammonium are 9.1 %-31.9 %, 5.7 %-32.1 %, and 5.9 %-13.3 %, respectively. Figures S8  and S9 show the modeled spatial distribution of PM 2.5 , sulfate, nitrate, and ammonium concentrations averaged in the four typical heavy pollution processes over Jingzhou and China. The fractions of each inorganic salt in PM 2.5 for these four heavy pollution episodes are also shown in Fig. S10. Over central China, the main components of PM 2.5 are the three inorganic salts in these pollution episodes, with the averaged contributions of sulfate, nitrate, and ammonium being ∼ 20 %, ∼ 18 %, and ∼ 13 %, respectively (Table 3). Our modeling results are comparable to the previous observed results (Table 4). Huang et al. (2014) have also reported that the three secondary inorganic particles rank the highest fraction among the PM 2.5 species in central eastern China. As shown in Table 3, in addition to inorganic salts, other chemical components include dust (∼ 15 %), black carbon (∼ 7 %), primary organic aerosol (∼ 14 %), and second organic aerosol (∼ 13 %). In these four pollution events, the differences in mass percentages of each chemical component ranged from 0.1 % (dust) to 6.2 % (sulfate) ( Table 3). See details in Sect. 3.4 for further analysis of the causes for the differences.

Local emissions versus transmission contributions to PM 2.5 under the four main PSCs
In order to investigate the effectiveness of emission control to reduce PM 2.5 pollution of central China in the four typical severe particle pollution episodes, firstly we estimate the local sources versus transmission contributions of PM 2.5 by GEOS-Chem sensitivity simulations (Table 2). Results of XJ0 (emissions outside Jingzhou are zero) indicate the contribution of local emission sources to the PM 2.5 pollution over Jingzhou. The difference between CON and XCC0 (emissions outside central China are zero) shows the transmission contribution of PM 2.5 outside central China to Jingzhou. The difference between CON and NCP0, YRD0, PRD0, and SCB0 (emissions over the NCP, YRD, PRD, and SCB are zero, respectively) represents the contribution of pollution transport from the NCP, YRD, PRD, and SCB regions to Jingzhou, respectively. For the SW-type synoptic situation, differences between the simulation results of NCP0, YRD0, SCB0, and CON show that pollution controlled by SW-type circulation over central China is almost unaffected by the emission sources from northern China, eastern China, or the Sichuan Basin. The concentrations of PM 2.5 and three inorganic salts simulated by NCP0, YRD0, and SCB0 are similar to those simulated by CON, with a difference of less than 3.0 % (Fig. 8). However, as they are affected by the southerly airflow at 850 hPa (Fig. 7), air pollutants formed over southern China could be transmitted to central China, with a transport contribution of 7.6 %. In addition, the contributions from transboundary transport from non-Jingzhou central China is simulated at 12.0 % by comparing the results of XJ0 and XCC0. The transport of air pollutants from the south leads to the smallest proportion of the three inorganic salts (45.7 %) in Jingzhou among the four pollution episodes (50.3 %-55.5 % for the other three episodes) because the emissions of SO 2 , NO 2 , and NH 3 in the south (especially in Guangxi and Guizhou province) are smaller than those in central China (Li et al., 2017a). The low surface wind speed of 2.1 m/s on average (Fig. 5) and the weak ascending in the vertical direction (Fig. 7) at Jingzhou mean that it is not conducive to the dispersion of local pollutants . The high PM 2.5 concentrations are mainly accumulated by local emissions. The simulations of XJ0 and CON show that local emission sources over Jingzhou contribute ∼ 70 % to PM 2.5 .
For the NW-type synoptic mode, affected by the northerly airflow (Fig. 9), it is conducive to the southward movement of Table 2. Description of sensitivity simulations by the GEOS-Chem model. The NCP, YRD, PRD, and SCB are the areas framed in red shown in Fig. 1.
Under the A-type circulation, Jingzhou is controlled by a high-pressure system (Fig. 10), which can lead to stable weather conditions caused by radiation inversion (Guo et al., 2015), and subsidence inversion (Kurita et al., 1985), which is favorable to continuous accumulation of local pollutants (Guo et al., 2015). The distribution of PM 2.5 in China is similar to that of the SW-type weather condition, with an averaged PM 2.5 concentration of 128.6 µg/m 3 over central China. Unlike the SW type, the PM 2.5 at Jingzhou in this synoptic pattern is less affected by transboundary transport, with the total transport contribution of the surrounding four major pollution regions being less than 9 %. The contribution of local emission sources is about 82 % (Fig. 8). Under the C-type synoptic pattern, the southwest lowpressure center is formed at 850 hPa, and Jingzhou is located in the low-pressure system of the SLP field (Fig. 11). In combination with the large relative humidity (78 % on average because the occurrence season of the C type is late winter to early spring; see Fig. 5), it can promote the haze pollu-tion due to its impact on hydrophilic aerosols (Twohy et al., 2009;Zheng et al., 2015). Together with the low wind speed (less than 4 m/s; Fig. 5), it is easy to cause the accumulation of pollutants. The average concentration of PM 2.5 over central China is as high as 203.7 µg/m 3 . Air pollution controlled by this weather condition is the most serious of the  Table 2. four typical synoptic controls. However, in this weather situation, pollutants in northern China are easy to diffuse (Miao et al., 2017;Li et al., 2019), and the concentration of PM 2.5 is significantly lower than that in the former three weather situations (Figs. 11 and S9). The contribution of pollution transport from regions other than central China simulated by GEOS-Chem is less than 8 %, and the contribution of local emission sources at Jingzhou is more than 85 % (Fig. 7).

Effectiveness of emission reduction under the four main PSCs
In order to estimate the effectiveness of emission reduction in severe pollution events forced by the four potential synoptic controls, we conduct sensitivity simulations by applying seven emission scenarios (Table 2). All emission scenarios use the reduction ratio of 20 %, which is close to the average In the JSN emission reduction scenario, the sulfate and ammonium concentrations over Jingzhou are significantly reduced by 3.2-5.8 µg/m 3 (12.7 %-14.5 %) and 0.6-1.9 µg/m 3 (3.2 %-5.9 %) in these four pollution events, respectively. However, the concentration of nitrate increases (1.3 %-1.7 %). This is because there is a competition mechanism between nitrate and sulfate. Ammonium ions always react with sulfate ions first to generate ammonium sulfate, which will continue to react with nitrate ions to generate ammonium nitrate when ammonium ions are rich (Mao et al., 2010). Thus, the reduction of SO 2 emission increases the concentration of nitrate, which offset the contribution of sulfate particle reduction to the environment to some extent. Therefore, the application of JSN emission reduction scheme only reduces the PM 2.5 concentrations by 3.1-7.2 µg/m 3 (2.0 %-3.5 %, Fig. 12). This inefficient emission reduction scheme has been most widely used in heavy pollution areas over China over the past decade, ignoring the synergistic effect of various precursors.
By applying the JSNN and JALL emission reduction scenarios, we aim to evaluate the synergistic effect of multiple precursors on emission reduction. These two scenarios reduce the average sulfate concentration in Jingzhou by 2.8-6.7 µg/m 3 (11.3 %-17.3 %) and 2.9-7.2 µg/m 3 (11.7 %- 17.9 %) and the ammonium concentration by 2.0-4.8 µg/m 3 (12.1 %-16.5 %) and 2.2-4.7 µg/m 3 (13.2 %-17.3 %), respectively. Unlike the increments of nitrate in JSN emission reduction scenario, the nitrate decreases (JSNN: 0.3-1.2 µg/m 3 ; JALL: 0.4-1.5 µg/m 3 ). Therefore, through the application of JSNN and JALL emission reduction schemes, PM 2.5 concentrations decrease by 4.9 %-8.3 % and 9.0 %-15.9 %, respectively (Fig. 12), which is much higher than the improvement in the JSN scenario. M.  have also evaluated the sensitiveness of NH 3 control to PM 2.5 reduction based on observations. However, these results indicate that it is unrealistic to substantially reduce local emissions to achieve the national air quality standard in the long term. Additionally, the sensitivity simulations that exclude emission sources over upwind regions are conducted to estimate the potential PM 2.5 mitigation of inter-regional and intraregional joint control. Our results show that after applying Y. Yan et al.: Effectiveness of emission control in reducing PM 2.5 pollution Figure 10. The same as in Fig. 7 but for A-type synoptic control (14-21 January 2014). the TALL emission reduction scenario, PM 2.5 concentrations have been significantly improved, with the improvement rates increased from 9.0 %-15.9 % (by JALL scenario) to 17.4 %-18.8 % (Fig. 12). The NW-type synoptic-controlled air pollution episode shows the best effect of joint prevention, followed by the SW type. For the NW type, PM 2.5 concentrations are reduced by 26.5 µg/m 3 (16.9 %) by reducing emissions over central China and Northern China (CNALL scheme), which is much more effective than the JALL emis-sion reduction scheme (14.1 µg/m 3 , 9.0 %). In an SW-type controlled pollution episode, this should be otherwise to decrease the emissions over Southern China in addition to central China.

Conclusion
The PM 2.5 pollution in autumn and winter haze periods is now the key obstacle for further improving air quality in Figure 11. The same as in Fig. 7 but for C-type synoptic control (26 January-2 February 2014).
China. The extremely severe and persistent PM 2.5 pollution episodes are attributed to adverse synoptic conditions in addition to high precursor emissions. For the PM 2.5 mitigation during winter haze episodes in specific regions forced by various potential synoptic controls, how to effectively reduce emissions has become an urgent scientific question that needs to be answered. Our results over central China could provide reference for regional air quality policy-making.
Using the Lamb-Jenkinson circulation classification, the top four potential synoptic controls of heavy PM 2.5 pollution days (totally 109 d) over central China from 2013 to 2018 are decomposed as follows: SW type, NW type A type, and C type, accounting for 29 %, 27 %, 19 %, and 12 % of the total pollution days, respectively. In these four PSCs, three inorganic salt aerosols (sulfate: ∼ 20 %; nitrate: ∼ 18 %; ammonium: ∼ 13 %) accounted for ∼ 51 % of PM 2.5 concentrations simulated by optimized GEOS-Chem modeling. In the SW-type (NW-type) synoptic situation, affected by the southerly (northerly) airflow, pollutants over southern (northern) China could be transmitted to central China, with the transport contribution of 7.6 %/37 %. Under A-type (Ctype) weather, affected by stable weather conditions (high relative humidity), the pollution processes are less affected by the emission sources from non-local regions, and the local emission sources dominate the contribution of 82 % (85 %) to PM 2.5 .
By only reducing SO 2 and NO x emission and not controlling NH 3 , due to the competition mechanism between nitrate and sulfate, the concentrations of sulfate and ammonium decrease, but the concentration of nitrate increases instead. The enhanced nitrate counteracts the effect of sulfate reduction on PM 2.5 mitigation, with a less than 4 % decrease in PM 2.5 . Even if the NH 3 emission is also reduced, the PM 2.5 concentration reduction is less than 9 %. By applying the TALL emission reduction scenario, PM 2.5 concentrations would decrease significantly, with the improvement rate increased from 9.0 %-15.9 % (by JALL scenario) to 17.4 %-18.8 %.
These results provide an opportunity to effectively mitigate haze pollution via local emission control actions in coordination with regional collaborative actions according to different synoptic patterns. The NW-type synoptic-controlled air pollution episode shows the best effect of joint prevention, followed by the SW type. It is noted that in this study the division of transmission areas is relatively rough, and more accurate source area identification and refined assessment of the emission reduction effect of multiple pollutants from source groups are needed in follow-up studies.
Author contributions. YY and SK conceived and designed the research. YY performed the data processing, model simulations, and analyses. YZ assisted in the circulation classification. JW provided the emission data over central China. SK, TZ, and DL contributed the funding acquisition. YY wrote the paper with input from all authors.
Competing interests. The authors declare that they have no conflict of interest.