Evaluation on the effect of regional joint control measures in changing

Li LI, Shuhui ZHU, Jingyu AN, Min ZHOU, Hongli WANG, Rusha YAN, Liping QIAO, Xudong TIAN, 4 Lijuan SHEN, Ling Huang, Yangjun Wang, Cheng Huang, Jeremy C AVISE, Joshua S FU 5 1. School of Environmental and Chemical Engineering, Shanghai University, Shanghai, 200444, China 6 2. State Environmental Protection Key Laboratory of the Cause and Prevention of Urban Air Pollution Complex, Shanghai Academy 7 of Environmental Sciences, Shanghai 200233, China 8 3. Zhejiang Environmental Monitoring Center, Hangzhou, 310014, China 9 4. Jiaxing Environmental Monitoring Station, Jiaxing, 314000, China 10 5. Laboratory for Atmospheric Research, Washington State University, Pullman, Washington, USA. 11 6. Department of Civil & Environmental Engineering, University of Tennessee, Knoxville, TN 37996, USA 12 13 *Correspondence to: C. Huang (huangc@saes.sh.cn), H. L. WANG (wanghl@saes.sh.cn) and L. Li 14 (Lily@shu.edu.cn) 15 #These three people contributed equally to this work. 16 17 Abstract: Heavy haze usually occurs in winter in eastern China. To control the severe air pollution during the season, 18 comprehensive regional joint-control strategies were implemented throughout a campaign. To evaluate the 19 effectiveness of these strategies and to provide some insights into strengthening the regional joint-control 20 mechanism, the influence of control measures on levels of air pollution were estimated with an integrated 21 measurement-emission-modeling method. To determine the influence of meteorological conditions, and the control 22 measures on the air quality, in a comprehensive study, the 2nd World Internet Conference was held during December 23 16~18, 2015 in Jiaxing City, Zhejiang Province in the Yangtze River Delta (YRD) region. We first analyzed the air 24 quality changes during four meteorological regimes; and then compared the air pollutant concentrations before, 25 during and after the regulation under static meteorological conditions. Next, we conducted modeling scenarios to 26 quantify the effects caused due to the air pollution control measures. We found that total emissions of SO2, NOx, 27 PM2.5 and VOCs in Jiaxing were reduced by 56%, 58%, 64% and 80%, respectively; while total emission reductions 28 of SO2, NOx, PM2.5 and VOCs over the YRD region are estimated to be 10%, 9%, 10% and 11%, respectively. 29 Modelling results suggest that during the campaign from December 8 to December 18, PM2.5 daily average 30 concentrations decreased by 10 μg/m with an average decrease of 14.6%. Our implemented optimization analysis 31 compared with previous studies also reveal that local emission reductions play a key role in air quality improvement, 32 although it shall be supplemented by regional linkage. In terms of regional joint control, to implement pollution 33 channel control 48 hours before the event is of most benefit in getting similar results. Therefore, it is recommended 34 that a synergistic emission reduction plan between adjacent areas with local pollution emission reductions as the 35 core part should be established and strengthened, and emission reduction plans for different types of pollution 36 through a stronger regional linkage should be reserved. 37

The 2 nd World Internet Conference was held in Tongxiang,Jiaxing, Zhejiang during 16-18 December, 2015. 52 To reduce air pollution during the conference, Zhejiang Province and the Regional Air-pollution Joint Control 53 Office of the Yangtze River Delta (YRD) region developed an Action Plan for Air Pollution Control during the 54 Conference (henceforth referred to as the Action Plan), which clarified target goals, time periods for implementing 55 controls, regions in which the controls would be applied, and the control measures to be implemented, as described 56 below. Targets: achieve an Air Quality Index (AQI) below 100 in "key areas", an AQI below 150 in "control areas", 57 and to achieve significant improvement of the air quality in the surrounding (or buffer) regions outside the control 58 areas. Time Periods: the time periods of interest for implementing various controls include the early stage (3 months 59 before the conference), the advanced stage (2 weeks to 4 days before the conference) and the central stage (3 days 60 before and 2 days after the conference). Regions: areas within a 50km radius, within a 100km radius and outside of 61 a 100km radius from the centre of Tongxiang were classified as key areas, control areas and buffer areas, 62 respectively. These areas cover 9 cities including Jiaxing, Huzhou, Hangzhou, Ningbo and Shaoxing in Zhejiang 63 province, Suzhou and Wuxi in Jiangsu province and Xuancheng in Anhui province, as shown in Fig.1. 64 the anion system and methane sulfonic acid (MSA) eluent for the cation system. The limit of the detection reported 116 by the manufacturer is 0.1 ug/m 3 for all species. The operation principle of AIM-IC is described in detail by 117

Markovic et al. (2012) 118
Hourly ambient mass concentrations of sixteen elements (K, Ca, V, Mn, Fe, As, Se, Cd, Au, Pb, Cr, Ni, Cu, 119 Zn, Ag, Ba) in PM2.5 were determined by the Xact multi-metals monitor. In brief, the Xact instrument samples the 120 air through a section of filter tape at a flow rate of 16.7 lpm using a PM2.5 sharp cut cyclone. The exposed filter tape 121 spot then advances into an analysis area where the collected PM2.5 is analyzed by energy-dispersive X-ray 122 fluorescence (XRF) to determine metal mass concentrations. The sequence of sampling and analysis were performed 123 continuously and simultaneously on an hourly basis. 124

Potential Source Contribution Analysis 125
TrajStat is a HYSPLIT model developed by Chinese Academy of Meteorological Sciences and NOAA Air 126 Resources Laboratory based on geographic information system (GIS). It uses statistical methods to analyze air mass 127 back trajectories to cluster trajectories and compute potential source contribution function (PSCF) with observation 128 data and meteorological data included . 129 PSCF analysis is a conditional probability function using air mass trajectories to locate pollution sources. It 130 can be calculated for each 1° longitude by 1° latitude cell by dividing the number of trajectory endpoints that 131 correspond to samples with factor scores or pollutant concentrations greater than specified values by the number of 132 total endpoints in the cell (Zeng et al., 1989). Therefore, pollution source areas are indicated by high PSCF values. 133 Since the deviation of PSCF results could increase with the raise of distance between cell and receptor, therefore a 134 weight factor (Wij) was adopted in this study to lower the uncertainty of PSCF results. PSCF and Wij calculations 135 are described in Eq. (1) and Eq. (2), where mij is the number of trajectory endpoints greater than specified values in 136 cell (i, j), nij is the number of total endpoints in this cell (Zeng et al., 1989;Polissar et al., 1999). 137 (2) 139 In this study, the TrajStat modelling system was used to analyze potential source contribution areas of PM2.5 140 in Jiaxing during different pollution episodes with the combination of Global Data Assimilation System (GDAS) 6 trajectories corresponded to those trajectories with PM2.5 hourly concentration higher than 75 μg/m 3 . 143

Model selection and parameter settings 145
In this study, the WRF-CMAQ/CAMx air quality numerical modelling system was used to evaluate the 146 improvement in air quality resulting from the control measures outlined in the Action Plan. It takes into account of 147 modeling variations from different air quality models. For the mesoscale meteorological field, we adopted the WRF  The chemical mechanism utilized in CMAQ was the CB05 gas phase chemical mechanism (Yarwood, et al., 2005) 151 and AERO5 aerosol mechanism, which includes the inorganic aerosol thermodynamic model ISORROPIA (Nenes, 152 et al., 1998) and updated SOA yield parameterizations. The gaseous and aerosol modules used in CAMx are the 153 CB05 chemical mechanism and CF module, respectively. The aqueous-phase chemistry for both models is based 154 on the updated mechanism of the Regional Acid Deposition Model (RADM) (Chang et al., 1987). Particulate Source 155 Apportionment Technology (PSAT) coupled in the CAMx is applied to quantify the regional contributions to PM2.5 156 as well. The WRF meteorological modeling domain consists of three nested Lambert projection grids of 36km-157 12km-4km, with 3 grids larger than the CMAQ/CAMx modeling domain at each boundary. WRF was run 158 simultaneously for the three nested domains with two-way feedback between the parent and the nest grids. Both the 159 three domains utilized 27 vertical sigma layers with the top layer at 100hpa, and the major physics options for each 160 domain listed in Table 1. For the CMAQ/CAMx modelling domain shown in Figure 2, we adopted a 36-12-4km 161 nested domain structure with 14 vertical layers, which were derived from the WRF 27 layers. The two outer domains 162 cover much of eastern Asia and eastern China, respectively, while the innermost domain covers the YRD region. 163 The simulation period was from 1-18 December, 2015, during which 1-7 December was utilized for model spin-up 164 and 8-18 December was the key period for analysis of the modelling results with control measures.  Short-wave Radiation Scheme Goddard Short-wave Radiation Scheme (Chou and Suarez, 1999) Initial and boundary conditions (IC/BCs) for the WRF modeling were based on 1-degree by 1-degree grids 170 FNL Operational Global Analysis data that are archived at the Global Data Assimilation System (GDAS). Boundary 171 conditions to WRF were updated at 6-hour intervals for D01. 172 Anthropogenic source emission inventory in YRD is based on a most recent inventory developed by our group 173 The Sparse Matrix Operator Kernel Emissions (SMOKE, https://www.cmascenter.org/smoke) model is applied to 178 process these emissions for modeling inputs that is more detailed emission processes and not usually used in China. 179

180
Prior to evaluating the effectiveness of the control measures and reactions, the performance of the modelling 181 system was evaluated to ensure it was able to reasonably reproduce the observed meteorological conditions and 182 PM2.5 levels. Statistical indexes used for model evaluation include Normalised Mean Bias (NMB), Normalised 183 Mean Error (NME) and Index of Agreement (IOA). The equations to calculate these statistical indexes are as follows: 184 where Pj and Oj are predicted and observed hourly concentrations, respectively. ̅ is the average value of 185 observations. IOA ranges from 0 to 1, with 1 indicating perfect agreement between model and observation.

Method for quantifying the effectiveness of a control 206
Quantifying the PM2.5 reduction in response to emission reductions was done using the so called Brute Force 9 Method (BFM) (Burr and Zhang, 2011), where a baseline scenario was simulated using unadjusted emissions (i.e., 208 those emissions that would have occurred in absence of the Action Plan) and a campaign scenario was modelled 209 based on the emission controls outlined in the Action Plan. In both cases, the same meteorology and chemical 210 boundary conditions were utilized to drive the photochemical model simulations. Through a comparative analysis 211 of the scenarios, a relative improvement factor (RF) for a given atmospheric pollutant, resulting from emission 212 controls, can be calculated and combined with ground based observations to assess the improvement in air quality 213 associated with those emission controls. and Cd is the concentration improvement caused by the control measures (µg/m 3 ). Utilizing models in a relative 219 sense to assess the efficiency of emission controls on air quality is common practice in regulatory modelling, with 220 the assumption that there may be biases in the absolute concentrations simulated by a modelling system, but that 221 the relative response of that system will reflect the response observed in the atmosphere (US EPA, 2014). 222 3 Results and discussion 223

Photochemical transformation changes of air pollutants during the campaign 224
Ground observational data show that from December 1 to December 23, Jiaxing City experienced four distinct 225 physical and chemical processes that contributed to the observed pollution levels at different periods. For each of 226 these processes, this study utilized the integrated emission-measurement-modeling method to analyze the evolution 227 of air quality from several aspects, including the backward air flow trajectory, potential contribution source areas, 228 meteorological conditions and the variation of PM2.5 concentration. 229

Pollution process before the campaign with local emission accumulation as the main contributor 230
The first time period of interest was from December 6 to December 8. Analysis about the potential source 231 contribution areas resulting from PSCF modelling suggests that the polluted air mass primarily originated from the 232 northwest and northerly airstreams, passing Shandong, the eastern coastal areas of Jiangsu and Shanghai and into 233 northern Zhejiang, as is shown in Fig. 4. Analysis of the large-scale weather patterns showed that the polluted air 234 mass occurred in Beijing, Tianjin, Shandong peninsula and northern Jiangsu as a result of cold air with polluted air 235 mass transported into the region on the morning of December 5. In the southern part of Shandong province, the the plumes in northern Zhejiang expanded. Therefore, during this time period, the pollution was primarily affected 242 by regional transport and worsened by stagnant local conditions in Jiaxing. 243 influenced areas as is shown in Figure 5. Therefore, the pollution process was mainly affected by the transport of 254

Pollution removal process caused by clean cold air during the conference 276
During the conference from December 16 to December 18, weather was affected by the large-scale southward 277 transport of cold dry air in northern Zhejiang, resulting in lower temperature and relative humidity, as well as a 278 significant improvement in the air quality. On the 17th and the 18th, under the control of a high pressure system in 279 northern Zhejiang, the sea level pressure increased, the humidity was lower and the wind speed was reduced. 280 Because of the emission reduction effect of the control measures, the pollutant accumulation rate was likely slowed 281 down and the air quality in northern Zhejiang was good overall. From the analysis of potential sources, PM2.5 282 concentrations in Shandong, Jiangsu and Shanghai were significantly reduced. The PM2.5 concentration during the 283 conference was mainly controlled by local emissions, as is shown in Figure 7.  City before (December 1-7), during (December 8-19) and after the regulation (December 19-31) under stagnant 314 weather conditions. It can be seen that pollutant concentrations during the campaign were less than those before the 315 campaign, in which SO2 had the most significant decline of 40.1%, NOx, CO, PM2.5 and PM10 declined 8.0%, 2.6%, 316 12.5% and 16.3%, respectively, indicating that control measures have significantly improved the air quality in 317 Jiaxing City, especially with respect to SO2 and PM10. 318 After the campaign, all the pollutant concentrations rebounded sharply. SO2, NO, NO2, CO, PM2.5, PM10 319 increased 8.3%, 15.4%, 10.3%, 31.8%, 32.2% and 28.6%, respectively. Concentrations of some pollutants were 320 even higher than those before the campaign, which suggests that the emission intensity of the sources had 321 significantly increased after the campaign. The chemistry also changes if we compare observed data during and after the regulation. As is shown from 339 figure 10, the SO2 concentrations after control is a little bit higher than during control (+5.9%). However, the SO4 2-340 after control is much higher than during control (25.8%). This is probably due to two reasons: firstly, SO2 emissions 341 and primary sulfate emissions increased after the control measures were terminated; secondly, previous studies have 342 reported that increased NOx emissions could accelerate the formation of secondary sulfate (Cheng et al., 2016). 343 This can be clearly seen from the SOR. A different trend is observed for NO2 and NO3 -, with the NO2 concentrations 344 after control being much higher than during control (+9.4%), while the increase of NO3 -(+9.45%) is about the same. 345 Sulfate originates from both primary emissions and secondary formation, but nitrate is mostly secondary. The NOR 346 during and after regulation is about the same and most of the N is in the gas phase as indicated by NOx/(NOx+NO3 -) 347 (0.87). Therefore, the increase of NO3is smaller than SO4 2-. The PM2.5 concentration after control sharply 348 rebounded by 31.8%, indicating that both primary emissions and secondary formation are activated.   There were two regional pollution episodes that occurred during the campaign. The first was on December 10-374 12 caused by the southward motion of northern weak cold air. Polluted air masses from south-eastern Shandong 375 peninsula passed through central eastern Jiangsu and into northern Zhejiang, affecting the air quality in Jiaxing. 376 During this period, the average daily PM2.5 concentration in Jiaxing was 145.7 µg/m 3 , higher than the regional 377 average, and its major chemical components were nitrate (31%), sulphate (18%), ammonium (13%) and organic 378 carbon (13%), with obvious regional secondary pollution characteristics. 379 The second episode occurred from December 14-15, and was caused by the transit of northwestly strong cold 382 air. Polluted air masses came from the northwest direction, moved rapidly to the southeast, passed through Shanxi, 383 Hebei, west Shandong, east Anhui and west Jiangsu and ultimately into Zhejiang province. The air masses left 384 China through south-eastern Zhejiang on the early morning of the 16th. The YRD region was strongly affected by 385 the transport of the polluted air mass, with heavy polluted air masses appearing and lasting for about one day over 386 the YRD region from north to south. PM2.5 peaked in Jiaxing on the 15th with a daily average of 201.6 µg/m 3 . The 387 main chemical components of PM2.5 during the episode were nitrate (25%), sulphate (14%), ammonium (12%) and 388 organic carbon (13%), which is consistent with an aged air mass as well as regional secondary pollution 389 characteristics. 390 The regional linkage was initiated from December 16 to December 18, combined with favourable mixing 391 conditions brought by the cold front. The overall air quality in the YRD region during this time period was good, 392 with an average daily PM2.5 concentration in Jiaxing of 45 µg/m 3 . The major chemical components during this 393 (37%), with some newly formed particles and no obvious regional transport, suggesting that air pollutants were 395 mainly derived from local emissions. 396

Emissions reduction estimation during the campaign 397
The air quality assurance campaign for the 2 nd World Internet Conference was from December 8 to December 398 18. In order to ensure the air quality during the conference, three provinces and Shanghai municipality in the YRD processing sectors also contributed significantly to emission reductions of 13.4%, 8.0% and 6.5%, respectively. 417 In terms of the regional distribution of emission reductions, Jiaxing, Hangzhou, Suzhou and Shaoxing have the 418 largest contribution of around 80%. These four cities contribute 87% to the total emission reduction of PM2.5. 419 Combing all control measures, total emission reductions of SO2, NOx, PM2.5 and VOCs are estimated to be 420 22 and the proportion of their emission reductions would be even larger.

PM2.5 concentration improvement in Jiaxing 430
The WRF-CMAQ air quality model, combined with observations, was used to evaluate the improvment of 431 PM2.5 in Jiaxing due to the emission reductions achieved through the campaign. This analysis utilized two model 432 simulations to assess the impact of the emission reductions: 1) a baseline scenario, which utilized an uncontrolled 433 emission inventory (i.e., the emissions that would have occurred without the campaign), and 2) an emission 434 inventory, which reflects the emission reductions achieved by the campaign. Figure 13 shows the time series of 435 described in section 3.1.2, the prevailing wind direction during this period is NW, and Jiaxing experienced a heavy 441 pollution process with the transit and transport of strong cold air. Therefore, we can not see obvious effect without 442 strong upwind precursor emissions reductions. be seen from the figure, the improvement in PM2.5 before the conference (December 8 and 9) was relatively 448 significant, with a daily average decline of roughly 31% and 35%, respectively, which corresponds to a decrease of 449 around 17 μg/m 3 . The reduction in PM2.5 during December 14-15, two of the days with some of the highest observed 450 PM2.5, was relatively low at around 6%, while daily average PM2.5 concentrations on those days decreased by around December 18 after control measures were implemented in Jiaxing and regionally. The reduction in PM2.5 was the 485 results of both local controls, as well as regional controls which reduced pollution in the air masses transported into 486 Jiaxing. Overall, modelling suggests that the regional controls reduced PM2.5 levels in Jiaxing between 5.5%-16.5% 487 (9.9% average), while local control measures contributed 4.5%-14.4%, with an average of 8.8%. 488 Overall, local control measure in Jiaxing had the largest impact on PM2.5 levels and accounted for 89% of the decline 493 in PM2.5, while regional control measures contributed the remaining 11%. conditions, where local emission accumulation is the main contributor to the pollution process (Sce.1), and the 506 emission reduction scenario where transport of polluted air masses into Jiaxing is a major contributor to the PM2.5 507 levels in Jiaxing. In order to investigate the transport processes further, the latter scenario was further divided into 508 a scenario 24 hours in advance (Sce.2) and a scenario 48 hours in advance (Sce.3).  The WRF-CMAQ modelling system was used to analyse and compare the air quality improvement effect under 520 different pollution process in four scenarios. 521

Analysis of optimization scenario effects 522
In order to evaluate the effect of the different starting time for the same control measures, and the same starting 523 time for local and regional control measures, we investigated four scenarios. Figure 18 shows the percentage 524 reduction in daily average PM2.5 concentrations in Jiaxing City from December 13 to December 18 under the 525 regional emission reduction scenario, the Jiaxing local emission reduction scenario and the transport channel 526 emission reduction scenario. Overall, there are differences in the distribution of PM2.5 under the different scenarios. 527 The air quality improvement due to the regional emission reductions was higher than that of local emission 528 reductions in Jiaxing, and lower than that of channel emission reductions. 529 530 Fig. 18 Decline rates of PM2.5 daily average concentrations in Jiaxing under different scenarios reductions (Base), we can see that PM2.5 daily average concentrations in Jiaxing declined by around 5.5%-16.5% 534 under the regional emission reduction plan (regional emission plan including the local emissions control) from 535 December 13 to December 18 and by around 4.5%-14.4% under the local emission reduction plan. Local emission 536 reductions in Jiaxing contributed 83%-94% to the emission reduction effect. Therefore, local emission reduction in 537 Jiaxing is the key factor in improving the local air quality. 538 Compared with the channel emission reduction scenario 24 hours in advance (11.6%-18.2%), local emission 539 reductions also contributed more than 50% to the improvement effect on December 13, 17 and 18. Therefore, local 540 emission reductions contributed most to the air quality improvement effect in Jiaxing, indicating that local areas are 541 still the most important control areas during the campaign. 542 (2)Effect of emission reductions through transport channels 543 As mentioned above, during the large-scale transport of heavily polluted air masses into the Yangtze River 544 Delta region from December 14 to December 15, the PM2.5 pollution in Jiaxing was significantly affected. Under 545 the local emission reduction scenario (Sce.1) and the regional linkage emission reduction scenario (Base), PM2.5 546 daily average concentrations in Jiaxing decline by only 4.5%-5.9%. If a 30% reduction in emissions from industrial 547 sources in the upwind transport channel is implemented, PM2.5 daily average concentrations in Jiaxing declined by 548 11.6%-13.6%, while local emission reductions contributed less than 40% to the improvement of PM2.5. Therefore, 549 to reduce PM2.5 under these large-scale transport conditions, in addition to intensifying local emission reduction 550 efforts, it is more effective to prevent and control such pollution by adopting emission reductions of industrial 551 sources over key transport channels, especially for elevated sources. 552 In this study, the main transport channel involved is the northwest transport channel in control areas, which 553 basically represents the typical winter transport channel in the region. In this study, the main transport channel 554 involved is the northwest transport channel in control areas, which basically represents the typical winter transport 555 channel in the region. Air quality improvement due to regional emission reductions was slightly larger than that of 556 local emission reductions in Jiaxing, and smaller than that of channel emission reductions. This suggests that 557 emissions reduction in the downwind cities does not have much effect on Jiaxing's air quality. In contrast, emissions 558 reduction based on predicted transport pathway in advance are much more effective than local emissions reduction 559 as well as regional emission reductions. Therefore, a well-designed management plan for the main transport channel 560 is necessary to ensure optimized air quality improvement in autumn and winter, in addition to reducing local According to the comparisons between the emission reduction scenario 24 hours in advance (Sce.2) and the 564 emission reduction scenario 48 hours in advance (Sce.3) during the large-scale PM2.5 transport, we can see that if 565 we take December 13 as the target and adopt channel emission reductions 48 hours in advance, PM2.5 daily average 566 concentrations will decline by 23.1% when compared to the baseline scenario, which is significantly better than the 567 improvement achieved by the emission reduction scenario 24 hours in advance (18.2%). Therefore, early measures 568 to reduce emissions will lead to the improvement of air quality.  industrial enterprises cut down VOCs emission by 66% in total, contributing greatly to the reduction of PM2.5 formed 581 through the conversion of precursor species. The observation data of PM2.5 components suggest that the relative 582 contribution of secondary components dropped significantly during the conference. Production restriction or 583 suspension for industrial enterprises is the main contributor to emission reductions for various pollutants during the 584 campaign, which resulted in the largest improvement in air quality. 585 (2) Motor vehicle pollutant emissions declined significantly. In Jiaxing, motor vehicle restrictions were fully 586 implemented during heavy pollution days, temporary traffic control was implemented during certain periods, and 587 enterprises and institutions had a three-day vacation during the conference. Emission reduction rates for various 588 pollutants from motor vehicle emissions were around 40%-50%. Motor vehicle emission reduction measures 589 contributed to the total emission reductions of nitrogen oxides by 18.2%, fine particles by 3.4% and volatile organic 590 compounds by 10.1%. 591 (3) The effect of dust control measures is remarkable. During the conference, most of the construction sites 592 in Jiaxing were suspended from operation. Increased frequency for road cleaning activities greatly lowered the dust 593 emissions. Speciation of the measured PM2.5 suggest that the mass concentration of crust material, decreased by 14% 594 compared to measurements after the conference. Specially, under static conditions, mineral soluble irons (Ca 2+ and 595 Mg 2+ ) declined 56.8% before and during the campaign. This suggests that the suspension of construction operations 596 and increased frequency of rinsing and cleaning of paved roads significantly reduced dust emissions. 597 (4) Regional linkage between surrounding areas played an important role. PM2.5 is a typical regional air 598 pollutant, with obvious regional transport characteristics. In accordance with the requirements of the campaign 599 scheme, eight cities around Jiaxing have actively implemented emissions reduction measures. During the campaign, 600 PM2.5 concentrations in eight surrounding cities and south-eastern Zhejiang also declined with obvious regional 601

synergies. 602
It is worth noting that the implementation of control measures has also had a negative impact on the economy 603 and the society in the short term while improving the air quality. For example, production restriction or suspension 604 on a large number industrial enterprises were taken at great economic costs, and motor vehicle restriction had a 605 large impact on the society. 606 (5) Suggestions on emission reduction plans: Local emission reductions shall be supplemented by regional 607 linkage. Assessment results show that local emission reductions play a key role in ensuring air quality. Therefore, 608 it is recommended that a synergistic emission reduction plan between adjacent areas with local pollution emission 609 reductions as the core part should be established and strengthened, and emission reduction plans for different types 610 of pollution through a stronger regional linkage should be reserved. Strengthen the pollution reduction in the upper 611 reaches along the transport channel. It is especially crucial to enhance pollution emission reductions in the upper 612 reaches of the channel since long-distance transport of plumes is a problem. This is especially true for key industrial 613 sources and elevated sources. Considering that polluted air mass transport is more frequent in winter, it is necessary 614 to develop emission reduction plans for different plume transport channels, combined with forecasting and warning 615 mechanisms which could be initiated on time.