The WRF-CMAQ Simulation of a Complex Pollution Episode with High-Level O₃ and PM_2.5 over the North China Plain: Pollution Characteristics and Causes

Abstract

The problem of atmospheric complex pollution led by PM_2.5 and O₃ has become an important factor restricting the improvement of air quality in China. In drawing on observations and Weather Research and Forecasting-Community Multiscale Air Quality (WRF-CMAQ) model simulations, this study analyzed the characteristics and causes of a regional PM_2.5-O₃ complex pollution episode in North China Plain, in the period from 3 to 5 April 2019. The results showed that in static and stable weather conditions with high temperature and low wind speed, despite photochemical reactions of O₃ near the ground being weakened by high PM_2.5 concentrations, a large amount of O₃ generated through gas-phase chemical reactions at high altitudes was transported downwards and increased the O₃ concentrations at the ground level. The high ground-level O₃ could facilitate both the conversion of SO₂ and NO₂ into secondary inorganic salts and volatile organic compounds into secondary organic aerosols, thereby amplifying PM_2.5 concentrations and exacerbating air pollution. The contributions of transport from outside sources to PM_2.5 (above 60%) and O₃ (above 46%) increased significantly during the episode. This study will play an instrumental role in helping researchers to comprehend the factors that contribute to complex pollution in China, and also offers valuable references for air pollution management.

1. Introduction

Rapid economic development and significant rises in energy consumption have resulted in regional and complex pollution emerging as the foremost air pollution challenge in China [1,2,3]. The prominent feature of this complex pollution is the high mass concentrations of atmospheric oxidizing species, represented by ozone (O₃), and fine particulate matter (PM_2.5) (an aerodynamic diameter ≤ 2.5 μm) [4]. In recent years, with the implementation of various atmospheric control measures, PM_2.5 pollution has been alleviated to a certain extent in China [5,6]. Nonetheless, owing to the relatively inadequate regulation of volatile organic compound (VOC) emissions and the intricate nonlinear connection between O₃ and its precursors, concentrations of O₃ show an upward trend [7,8]. The challenge of PM_2.5 pollution remains unresolved, and O₃ pollution is gradually emerging, contributing to a severe environmental scenario marked by the increasingly conspicuous features of PM_2.5 and O₃ complex pollution in China. Although O₃ in the stratosphere benefits life on Earth by filtering out harmful ultraviolet radiation from the sun, O₃ and PM_2.5 can have adverse effects on human health on the ground. The level of simultaneous exposure to PM_2.5 and O₃ is highly correlated with an increased risk to human health, including cardiovascular and respiratory diseases [9,10,11].

At present, PM_2.5 is the largest factor causing air pollution in China, followed by PM₁₀ and O₃ [12]. The concentrations of PM_2.5 were higher in winter than in summer, and decreased with increasing altitude [13]. In northern cities of China, PM_2.5 concentrations were generally high [14], and total particle number concentrations did not decrease much from urban to rural areas or even remote environments [15]. This is in contrast to Europe, where average total particle number concentrations in urban environments are usually higher than those in rural environments [16,17]. PM_2.5 concentrations are influenced by anthropogenic emissions and meteorological conditions [18]. The high concentrations of PM_2.5 in northern China were related to emissions from fossil fuel combustion and biomass combustion [19]. More coal-fired industries (such as coal-fired power plants, steel manufacturing, etc.) and household heating based on coal and biomass during winter (mid November to mid March) led to higher emissions and PM concentrations in the northern region [20]. The impact mechanism of meteorological factors on PM concentrations is relatively complex, and the impact of a single meteorological factor depends on its combination with other factors. Moreover, the same meteorological element has different or even opposite effects on the inflow and outflow of PM [21]. PM_2.5 is more susceptible to meteorological conditions than PM₁₀ [22]. The studies showed that terrain played a crucial role in the generation and migration of PM [23]. The wind azimuth has a significant impact on the outflow of PM, while wind speed and direction have a significant impact on the inflow of PM [21]. Temperature and relative humidity are also correlated with the concentration of PM_2.5 [22]. These results reflect the important role of meteorological factors in the air pollution processes that occur in different seasons and significant changes in geological regions.

O₃ is a secondary pollutant formed in the atmosphere through the photochemical reactions of nitrogen oxides (NO_x) and VOCs [24,25]. Power plants, industry, and transportation are the main sources of NO_x [26], while VOCs are released from more diverse sources, such as vehicle exhausts, evaporative fuels and solvents, consumer goods, and trees. There is no significant spatial distribution difference between nitrogen dioxide (NO₂) and O_3, and O₃ is more divergent in space than NO₂, reflecting the complex nonlinear relationship between O₃ and NO_x [14]. NO_x can cause a decrease or increase in O₃ concentrations, depending on the relative ratios of NO_x to VOC. Generally speaking, in urban areas with high NO_x/VOC ratios, the production of O₃ is limited by VOC, resulting in a decreased NO_x titration of O₃ and free radicals. The reduction in NO_x emissions will tend to increase O₃ concentrations. In contrast, in rural areas where NO_x/VOCs ratios are typically low, reducing NO_x emissions will lower O₃ concentrations [27]. In many cities and industrialized regions around the world, the increase in O₃ concentrations in the lower atmosphere remains a persistent environmental problem. The surface O₃ concentrations in developed areas or nearby areas of China are increasing, and the growth rates in rural areas are lower than those in urban core areas, which are both the result of reduced NO_x emissions. In terms of space, there are O₃ hotspots all over eastern China, especially in the North China Plain (NCP) and the Yangtze River Delta (YRD), which are mainly caused by human sources producing high-level O₃ precursors. Some western cities also experienced severe O₃ pollution, such as central Gansu Province, which may be due to the unique terrain (mountainous valleys and basins), coupled with high local O₃ precursor emissions from the petrochemical industry and vehicles [28]. The distribution ranges of O₃ in China were comparable to those of Europe and the United States, but the scale and frequency of high O₃ events in China were much higher. Since the 1990s, severe O₃ pollution in many urban areas of Europe and the United States has been widely alleviated by strict emission control measures [29,30]. Meanwhile, South and East Asia have experienced rapid urbanization and industrialization, leading to a significant increase in anthropogenic O₃ precursor emissions and potentially shifting global air pollution hotspots to densely populated areas such as India [31] and Mexico City [32]. However, after 2020 the overall O₃ pollution levels in most typical regions of China decreased [33], which may be explained by the coordinated control measures for PM_2.5 and O₃ implemented by the Chinese government [34] and the continuous impact of COVID-19 in China. Meanwhile, meteorological changes can also have a significant impact on surface O₃ concentrations. Meteorological conditions have a strong impact on the production and distribution of O₃ by changing air transport patterns, the dry and wet deposition of gases and aerosols, and the rates of chemical reactions and natural emissions [35].

Although the peak seasons for PM_2.5 and O₃ pollutions are different, as shown by cold-season PM_2.5 pollution and warm-season O₃ pollution [3], simultaneous occurrences of high concentrations of PM_2.5 and O₃ (PM_2.5-O₃ complex pollution) have been observed from time to time. Mao et al. [36] pointed out that PM_2.5-O₃ complex pollution events accounted for 33.4% of total O₃ pollution events in Shanghai from 2013 to 2017. Xiao et al. [37] found there was a total of 34 days of PM_2.5-O₃ complex pollution in Tianjin from 2017 to 2019, which appeared between March and September and then slightly increased every year. Simultaneous exposure to both PM_2.5 and O₃ results in disproportionately more severe health effects than individual exposure to either pollutant [38,39]. While previous studies have thoroughly explored PM_2.5 and/or O₃ pollution [40,41,42], there is a notable lack of research that specifically addresses air pollution events characterized by simultaneously high concentrations of PM_2.5 and O₃.

Several studies in China have analyzed the causes of PM_2.5-O₃ complex pollution from the perspective of meteorology. For example, by drawing on field observations in Shanghai from 2014 to 2016, Yang et al. [43] found that days with simultaneously high concentrations of PM_2.5 and O₃ constituted 1.0% of the entire study period. These events were predominantly observed in stable weather conditions marked by high temperatures, relatively high humidity, and low wind speeds. Lai et al. [44] found that the formation of PM_2.5-O₃ complex pollution in the Pearl River Delta (PRD) region was favored in specific weather conditions, including continental cold high, tropical cyclone, and denatured high, with the occurrence frequency accounting for 49.4%, 21.9%, and 20.7%, respectively. Zhao et al. [45] found that, compared with single PM_2.5 or O₃ pollution, complex pollution in Handan imposes more stringent requirements on meteorological elements, requiring temperatures from 21 to 29 °C, high humidity, and low wind speed and air pressure. Wang et al. [46] analyzed two PM_2.5-O₃ complex pollution events in Beijing on 2 August and 21 August 2014, and revealed that the synoptic situations in the complex pollution were characterized as westerly gas flow at a height of 500 hPa height, with low pressure at the ground. In addition to observational analyses, the simulation of PM_2.5-O₃ complex pollution by the chemical transport models (CTMs) can also provide significant information, and contribute to the understanding of pollution characteristics and causes. Lai et al. [47] simulated a regional PM_2.5-O₃ complex pollution event in the PRD region on 27 October 2014 by using the Weather Research and Forecasting-Chemistry (WRF-Chem) model, and found that high O₃ concentrations were associated with both physical and gas-phase photochemical processes; and also found high PM_2.5 concentrations were related to high initial concentrations and a large production amount of secondary inorganic aerosols. However, at present, few studies have used CTMs to analyze PM_2.5-O₃ complex pollution.

Corresponding reports abroad have highlighted simultaneous occurrences of PM_2.5 and O₃ pollution. For example, Kalashnikov et al. [38] defined the extreme values of PM_2.5 and O₃ in each region as the daily average of local PM_2.5 and the 90th percentile value of the maximum daily 8-h average concentration of ozone (MDA8_O₃) within each year, respectively. If PM_2.5 and O₃ exceeds the extreme values, it is considered as the co-occurrences of PM_2.5 and O₃ pollution (complex pollution). A fixed threshold is not used to define complex pollutions because this spatiotemporal extreme value can take into account both the overall air quality improvement due to emission reduction and stricter national air quality standards. The results indicate that, from 2001 to 2020, the frequency, spatial range, and temporal persistence of extreme PM_2.5-O₃ events in the western United States significantly increased, resulting in approximately 25 million more people being exposed to various harmful air pollutants annually [38]. And it is speculated that with the continuous warming of the climate, the possibility of complex pollution events occurring simultaneously in the western United States will increase. However, few worldwide studies have conducted in-depth analysis of the characteristics and causes of PM_2.5-O₃ complex pollution, and the findings of our study could therefore fill the gap in the related fields. Currently, many studies have utilized machine learning techniques and big data analysis to identify hotspots, cold spots, and air pollution patterns in air pollution research [48]. Kovacs and Haidu [49] used principal component analysis of multidimensional satellite images to study NO₂ changes, and modeled tropospheric NO₂ concentrations by using developed principal component analysis models for air pollution estimation, fully demonstrating the reliability of principal component analysis in identifying and predicting patterns of air pollution changes.

The NCP region is densely populated and heavily industrialized, and is the area of China that is most seriously affected by PM_2.5 and O₃ pollution [3,14,28]. In this study, the Weather Research and Forecasting-Community Multiscale Air Quality (WRF-CMAQ) model was utilized to investigate a regional PM_2.5-O₃ complex pollution episode in the NCP region from 3 to 5 April 2019. The study conducted an analysis of pollution characteristics and weather situations, and also examined the formation mechanism and source apportionment. The novelty of this study lies in the fact that typical pollution events in the NCP, where PM_2.5 and O₃ pollution are relatively serious, are selected for research. Secondly, a more in-depth analysis of the formation mechanism of regional PM_2.5-O₃ composite pollution is conducted through PM_2.5 chemical composition analysis and O₃ process analysis. Thirdly, the contribution of regional transports to this pollution episode is revealed through ISAM source analyses. The findings of this research can contribute to a better understanding of the causes of complex pollution and offer valuable insights to current air quality management, not only in China, but also around the world.

2. Materials and Methods

2.1. Observations

The measurements of meteorological parameters used in this study, including 2 m temperature (T2), 2 m relative humidity (RH2), 10 m wind speed (WS10), and 10 m wind direction (WD10), were obtained from the National Climate Data Center (NCDC, ftp://ftp.ncdc.noaa.gov/pub/data/noaa/, accessed on 1 November 2023). NCDC is under the jurisdiction of the National Oceanic and Atmospheric Administration (NOAA) in the United States, and its meteorological observation data have a time resolution of 3 h. The surface weather charts were provided by the Hong Kong Observatory (http://envf.ust.hk/dataview/hko_wc/current/, accessed on 1 November 2023). Hourly PM_2.5 and O₃ observation data of 367 Chinese prefecture-level cities were obtained from the China National Environmental Monitoring Center (CNEMC, http://www.cnemc.cn/, accessed on 1 November 2023). The data collections of CNEMC were conducted with the guidance of the grade II national standard for ambient air quality (GB3095-2012) [50], and the technical guideline on environmental monitoring quality management (HJ630-2011) [51]. In accordance with GB3095-2012, we defined the PM_2.5-O₃ complex pollution as the daily average concentration of PM_2.5 exceeding 75 μg/m³ and MDA8_O₃ exceeding 160 μg/m³ (on the same day). We then selected a regional PM_2.5-O₃ complex pollution episode in the NCP region from 3 to 5 April 2019 and chose 4 typical cities (Handan, Jining, Anyang, Kaifeng) as the analytical objects (Figure 1).

2.2. WRF-CMAQ Model Configuration

In this study, the offline WRF (v3.9.1)-CMAQ (v5.3.2) model [52,53,54] was used to simulate the PM_2.5-O₃ complex pollution. The meteorological output from WRF was processed by Meteorology-Chemistry Interface Processor (v5.1) (MCIP) to the format required by CMAQ. We conducted the simulation period from 00:00 Local Time (LT) on 25 March to 00:00 LT on 10 April 2019, and the first 5 days were used as spin-up to minimize the influence of the initial conditions. Figure 1a shows the model domain, with a horizontal resolution of 12 km covering most of China. In the vertical direction, there were 31 sigma layers in the WRF, with the model top fixed at 100 hPa; and for the CMAQ model, there were 12 layers to enhance the modeling proficiency. The detailed physical parameterizations and chemical options in the WRF-CMAQ model were the same as those in Li et al. [55]. The initial and lateral boundary conditions of meteorological fields were provided by the Fifth Generation Atmospheric Reanalysis of the Global Climate (ERA5) dataset of the European Center for Medium-Range Weather Forecasts (ECMWF), which has a spatial resolution of 31 km, a temporal resolution of 1 h and 38 barometric layers in the vertical direction. The default initial and boundary chemical conditions provided in CMAQ were used. The anthropogenic emissions were derived from th Emission Inventory of Air Benefit and Cost and Attainment Assessment System (EI-ABaCAS) established by Tsinghua University [56,57]. The Biogenic Emission Inventory System version 3.14 (BEISv3.14) was used to calculate the natural sources for biogenic emissions.

2.3. The Integrated Process Rate Analysis

In order to analyze the formation mechanism of O₃, the integrated process rate (IPR) analysis method embedded in the CMAQ model was applied [58]. IPR can calculate the quantitative contribution of individual physical and chemical processes in a specific grid, including gas-phase chemical reactions (CHEM), cloud processes with the aqueous chemistry (CLDS), dry deposition (DDEP), horizontal advection (HADV), horizontal diffusion (HDIF), vertical advection (ZADV), and vertical diffusion (VDIF). In this study, we defined HTRA (HADV + HDIF) and VTRA (ZADV + VDIF) to represent the net effects of horizontal and vertical transports on the O₃ formation, respectively [59]. The CHEM contribution was calculated as the sum of O₃ chemical productions and losses in the atmosphere.

2.4. The Integrated Source Apportionment Method

The integrated source apportionment method (ISAM) module, coupled with the CMAQ model, can tag and track the pollutants from different geographic regions and source types, which have been widely applied in the source apportionment of PM_2.5 and O₃ [60,61,62]. In this study, the ISAM was used to quantify the contributions of different regions, as well as the boundary conditions (BCON) for the regional transports of PM_2.5 and O₃ in the receptor cities. As shown in Figure 1c, d, there are 10 tagged regions inside China, which are defined on the basis of administrative division: Beijing (BJ), Tianjin (TJ), Hebei (HB), Henan (HN), Shandong (SD), Shanxi (SX), Hubei (HUB), Anhui (AH), Jiangsu (JS), and other regions (OTH) inside China

2.5. The Technical Workflow

Figure 2 shows the step-by-step technical workflow of this study. On the basis of air quality monitoring data, this study analyzed the temporal and spatial characteristics of a regional PM_2.5-O₃ complex pollution episode in NCP in 2019; the simulation effects of WRF and CMAQ models were also evaluated using observed data. The WRF-CMAQ model was used, for PM_2.5 chemical composition analysis, O₃ process analysis, and ISAM source apportionment, to analyze the formation mechanism and source apportionment of this complex pollution episode.

3. Results

3.1. Characteristics of the High-Level O₃ and PM_2.5 Complex Pollution Episode

Figure 1a,b show the spatial distributions of the PM_2.5 and MDA8_O₃ concentrations over mainland China from 3 to 5 April 2019, when a large-scale complex pollution process of high-level PM_2.5 and O₃ concentrations occurred in the NCP region. Over a span of three consecutive days, concentrations of both PM_2.5 and O₃ in more than 10 cities simultaneously exceeded the grade II national standard. Among them, Handan, Jining, Anyang, and Kaifeng experienced successive complex pollution on three days. Thus, we focused on these four cities to investigate the characteristics and causes of this PM_2.5-O₃ complex pollution episode. As shown in Figure 3, concentrations of PM_2.5 in all four cities exceeded 75 μg/m³ on 2 April, suggesting that the PM_2.5 had accumulated before the occurrence of the complex pollution. Concentrations of both PM_2.5 and O₃ during the complex pollution episode are high in Figure 3, indicating that, despite the high PM_2.5, concentrations reduced the intensity of solar radiation reaching the surface, there were still active photochemical reactions that could produce O₃.

From 30 March to 5 April, the T and RH values of each city exhibited a predominantly ascending pattern (Figure 3), and with the T values notably peaking during the episode of complex pollution. After 6 April, precipitation occurred successively in the various regions, with decreases of T on 7 April being accompanied by increases of RH. Figure 4 shows the surface weather and surface wind field charts at 8:00 from 3 to 5 April. At 8:00 on 3 April, Shandong Province was located in a center of high pressure, the weather was sunny, and the ground wind speed was low. Governed by an anticyclone, the horizontal airflow in Shandong Province exhibited a clockwise divergence. Then, the high-pressure center moved to the southeast and entered the East China Sea at 8:00 on 4 April. At this time, a low-pressure system in the Northeast emerged and underwent subsequent development. The NCP region was situated in proximity to a subdued low-pressure trough, with surface winds shifting from the southeast towards the NCP, and from the southwest towards the northeast. At 8:00 on 5 April, the NCP region was located between the high pressure in the Qinghai Tibet Plateau and the low pressure in the Japan Sea. Under the control of the pressure equalizing field, there was air convergence at the junction of Hebei, Henan, and Shandong, with small pressure gradient, weak wind and sunny weather. In short, throughout the complex pollution episode, the NCP region experienced predominantly sunny conditions due to the influence of static and stable weather. This entailed high temperatures, low wind speeds, and unfavorable atmospheric diffusion conditions, resulting in the accumulation of pollutants and facilitating photochemical reactions, ultimately leading to the formation of pollution. This was consistent with the research findings of Wang et al. [46] and Li et al. [63].

3.2. Model Evaluation

The modeling results were compared with the observations to evaluate the WRF-CMAQ simulation performance. Table S1 presents the statistical metrics, including mean bias (MB), mean error (ME), root mean square error (RMSE), and the index of agreement (IOA) for T2, RH2, WS10, and WD10 in Handan, Jining, Anyang, and Kaifeng. Additionally, normalized mean bias (NMB), normalized mean error (NME), and the Pearson’s correlation coefficient (R) were added to assess the PM_2.5 and O₃ simulations, as shown in Table S2. The definitions of these statistical metrics can be found in the Supplementary Materials [53].

As shown in Table S1 and Figure 2, the WRF model captured the variations of T2 well, with MB values between −0.46 and 0.39 °C, and IOA values higher than 0.95 in the four cities. The T2 values of Jining were overestimated, and the values of the other three cities were slightly underestimated. Similar to T2, the simulated RH2 showed a good fit with the observations, with IOA values between 0.87 and 0.93. The model overestimated the RH2 of the four cities slightly, with MB values varying from 1.72 to 4.19%. For WS10, with the exception of the underestimation in Handan (MB = −0.73 m/s), positive biases were acquired in the other three cities, with the MB values ranging from 0.67 to 1.70 m/s, and IOA values mainly between 0.71 and 0.80 (except Jining, with 0.53). Since the WS10 data provided by NCDC were presented as integer values, their relative inaccuracy could potentially amplify disparities between observations and simulations. Compared to the other three meteorological parameters, the WD10 simulations were relatively weak, with MB and IOA values ranging from −27.26 to −19.09 °C and 0.52 to 0.79, respectively. The significant bias observed in WD10 resulted from various factors, including the deviation in time resolution between observations and simulations [61], and the insufficient resolution of land use and land cover data [64]. In conclusion, the WRF model could reflect the actual meteorological conditions and provide accurate meteorological fields to the CMAQ simulation.

As shown in Table S2 and Figure 3, the model captured the change trends and peaks of PM_2.5 and O₃, with R values between 0.73 and 0.81 (except the PM_2.5 of Kaifeng, at 0.56). The observed PM_2.5 and O₃ concentrations were underestimated slightly, with NMB values ranging from −20.5 to −4.2% and −25.4 to 1.3%, respectively. Despite the presence of underestimations in PM_2.5 and O₃ during the complex pollution, the model effectively captured variabilities in both pollutants, and the underestimations fell within an acceptable range (±30%, US-EPA) [65], meaning it was therefore feasible to use the model results to analyze complex pollution. In addition to the PM_2.5 and O₃ evaluation of the above four cities, the simulation results over the whole domain during the complex pollution were also evaluated. Figure 1a,b show the comparison of the spatial distributions of the simulated and observed PM_2.5 daily average and MDA8_O₃ concentrations over mainland China from 3 to 5 April 2019. It can be seen that the high pollution areas of simulated PM_2.5 and O₃ were basically distributed in the NCP region and its surrounding areas, being consistent with the observations. In general, the CMAQ model performed well in simulating both PM_2.5 and O₃, and could be used in the subsequent study of the pollution formation mechanism and source apportionment.

3.3. Formation Mechanisms of the High-Level O₃ and PM_2.5 Complex Pollution Episode

3.3.1. Analysis of O₃ Formation Mechanisms

Figure 5a shows the hourly contributions of different atmospheric processes to the formation of surface O₃ in Handan, Jining, Anyang, and Kaifeng at the first modeling layer from 3 to 5 April 2019. Similar to previous studies [59,66], the contribution of CLDS was negligible and was not considered in this study because of the sunny weather and few clouds during this pollution episode. Generally, the contribution of CHEM was positive in the daytime (7:00~18:00), due to the active photochemical reactions, and negative at night, due to the chemical loss caused by O₃ consuming species (i.e., NO). As shown in Figure 5a, when compared to the other three processes, the contribution of CHEM was relatively weak but consistent with the change trend of net O₃, indicating the important role of CHEM in the O₃ formation. As an important way of removing gaseous pollutants, DDEP made a significant negative contribution to O_3, with −6.72 ppb/h in Handan, −12.02 ppb/h in Jining, −11.35 ppb/h in Anyang, and −11.80 ppb/h in Kaifeng, and was highly related to the concentrations of surface O₃ and dry-deposition velocity. The HTRA and VTRA processes were closely related to meteorological conditions (i.e., wind field, boundary layer turbulence), which demonstrated varying positive or negative contributions to O_3, both at different times and in different locations. During the complex pollution episode, the HTRA and VTRA contributed most to the formation of O₃. The high concentrations of O₃ in Handan and Anyang mainly came from both HTRA and VTRA processes, while those in Jining and Kaifeng mainly came from the VTRA process.

Figure 5b,c show the average contributions of each process to O₃ formation at different heights during the daytime (7:00~18:00) and nighttime (19:00~6:00), from 3 to 5 April 2019, respectively. During the daytime, CHEM contributed positively to O_3, at both near-surface and aloft levels (from 50 to 3400 m above ground, model layers 2~12), with a lower contribution at the surface layer than at high altitudes. This was similar to the results of Li et al. [67], who found that, in August, photochemical reactions at the height of 300~1500 m were stronger than those on the ground in the Yangtze River Delta (YRD) region. A large amount of O₃ generated by gas-phase chemistry at high altitudes were transported to the ground, which significantly increased the positive contributions of the surface VTRA to O₃, with mean contributions of 10.98 ppb/h in Handan, 18.51 ppb/h in Jining, 21.97 ppb/h in Anyang, and 20.75 ppb/h in Kaifeng. During the nighttime, the negative contribution of CHEM to O₃ was less evident above five layers (500 m), and O₃ was mainly consumed by gas-phase chemistry in the surface layer, with mean contributions ranging from −3.97 to −2.90 ppb/h in the four cities. Above 500 m, the formation of O₃ was mainly affected by HTRA and VTRA, and these two processes had opposite contributions and offset each other.

Generally, in the severe O₃ pollution episodes, the contribution of CHEM to the surface O₃ in the daytime could reach more than 10 ppb/h [66]. However, our results showed that the mean contributions of CHEM in the four cities were 0.15~3.18 ppb/h during the daytime, of which the highest contribution was 11.0 ppb/h in Kaifeng at 12:00. The contributions of CHEM to the surface O₃ in the daytime were significantly reduced during the complex pollution episode, indicating that the high PM_2.5 concentrations in the atmosphere weakened the intensity of incident solar radiation due to extinction, thereby weakening O₃ photochemistry productions. This was consistent with the results of Gao et al. [68], who also pointed out that the reduction of surface O₃ caused by aerosols would lead to the weakening of dry deposition, slowing down the reduction of surface O₃ to a certain extent. More importantly, the significant reduction of surface O₃ gas-phase chemical production formed a larger O₃ vertical gradient, prompting more air masses with high concentrations of O₃ to enter the surface from the top of the boundary layer, partially offsetting the reduction of O₃ gas-phase chemical production. Therefore, the impacts of aerosols on surface O₃ concentration stemmed from the collaborative effects of diverse physical and chemical processes. Previous studies also showed that the complex physicochemical properties of aerosols impacted the generation and loss of near surface O₃. The O₃ concentrations could be directly affected by changing atmospheric dynamics and photodegradation rates, or indirectly affected by cloud optical thickness and heterogeneous reaction processes [69,70,71].

3.3.2. Analysis of PM_2.5 Formation Mechanisms

The chemical composition of PM_2.5 is intricate, encompassing both primary emissions and secondary generations through homogeneous and heterogeneous chemical reactions. Among them, secondary particles, such as sulfate, nitrate, ammonium salt, and secondary organic aerosols (SOA), are important components, accounting for more than 50% of the total mass of PM_2.5 [72,73], and acting as the key factors in the occurrence of high concentration PM_2.5 pollutions.

Table 1. Simulated concentrations of PM_2.5 and its major components (μg/m³) during complex and non-complex pollution periods.

Pollutants	Complex Pollution Periods				Non-Complex Pollution Periods
	Handan	Jining	Anyang	Kaifeng	Handan	Jining	Anyang	Kaifeng
EC	3.8	2.6	3.2	2.3	2.7	1.9	2.4	2.0
SO42−	7.8	8.3	7.5	6.7	4.4	5.2	4.7	6.0
NO3−	16.5	16.3	16.3	11.6	7.4	8.8	7.7	10.1
NH4+	7.0	7.1	6.8	5.2	3.2	3.8	3.4	4.4
SOA	26.9	27.0	26.7	24.8	11.7	16.1	13.7	19.6
PM2.5	76.6	74.7	73.9	62.5	39.0	45.9	42.3	53.2
Secondary proportion 1	75.9%	78.5%	77.4%	77.4%	68.3%	73.8%	69.7%	75.5%

¹ Refers to the ratio of the sum of SO₄²⁻, NO₃⁻, NH₄⁺, and SOA to PM_2.5.

shows the simulated concentrations of elemental carbon (EC), sulfate (SO₄²⁻), nitrate (NO₃⁻), ammonium (NH₄⁺), SOA, and PM_2.5 during complex and non-complex pollution periods in Handan, Jining, Anyang, and Kaifeng (30 March to 9 April). It is worth noting that the simulated and observed PM_2.5 concentrations reached a high value on 2 April (Figure 2), indicating that the PM_2.5 had accumulated before the occurrence of complex pollution. During the complex pollution episode, the proportion of secondary components in PM_2.5 in the four cities increased from 68.3–75.5% on non-complex pollution days, and from 75.9% to 78.5% on complex pollution days. The concentration of SOA increased the most, and it was also the component with the highest concentration of the secondary aerosol. For secondary inorganic aerosol (SIA), NO₃⁻ concentration was the highest, followed by SO₄²⁻, NH₄⁺. EC is the primary particulate matter, mainly from fuel combustion, that has stable chemical properties in the atmosphere and can represent the primary component of PM_2.5. When compared with the non-complex pollution periods, concentrations of PM_2.5 and its main components increased significantly during the complex pollution periods, and the growth multiple of EC concentrations was lower than those of secondary component concentrations. For example, the EC concentration in Handan increased 1.4 times during complex pollution periods, while the growth multiples of SO₄²⁻, NO₃⁻, NH₄⁺, and SOA concentrations were 1.8, 2.2, 2.2, and 2.3, respectively. This indicates that, during the complex pollution period, the formation and accumulation of secondary components through chemical reactions had a significantly stronger influence on the formation of PM_2.5 pollution than the direct emission of primary components. Li [74] also found, in a study of the atmospheric complex pollution episode in the Yangtze River Delta urban agglomeration, that the concentration of SOA and NO₃⁻ increased most significantly in high concentrations of PM_2.5 pollution events; and also that, during the entire pollution period, the PM_2.5 component with the highest concentration was NO₃⁻, followed by NH₄⁺, SO₄²⁻, organic carbon (OC), and EC.

Figure 6 shows the average daily changes of key meteorological elements and pollutants in Handan, Jining, Anyang and Kaifeng during the complex pollution episode (3 to 5 April). The meteorological elements included planetary boundary layer height (PBLH), RH and T. The pollutants included SO₂, SO₄²⁻, NO₂, NO₃⁻, SOA, VOCs, O₃ and photochemical oxidants (O_x, O_x = NO₂ + O₃), as well as the sulfate oxidation rate (SOR) and nitrate oxidation rate (NOR), which could reflect the secondary conversion rates of gaseous pollutants (SO₂ and NO₂) in the atmosphere. Their specific formulas follow:

$$\mathrm{S}\mathrm{O}\mathrm{R}=\frac{{\mathrm{S}\mathrm{O}}_4^{2-}}{{\mathrm{S}\mathrm{O}}_4^{2-}+{\mathrm{S}\mathrm{O}}_2}$$

(1)

$$\mathrm{N}\mathrm{O}\mathrm{R}=\frac{{\mathrm{N}\mathrm{O}}_3^{-}}{{\mathrm{N}\mathrm{O}}_3^{-}+{\mathrm{N}\mathrm{O}}_2}$$

(2)

Higher values of SOR and NOR indicated a stronger secondary oxidation of SO₂ and NO₂.

Sulfate is generated by the oxidation of SO₂ to sulfuric acid (H₂SO₄) in the gas or liquid phases. In the process of gas phase oxidation, SO₂ is oxidized by OH to form sulfur trioxide (SO₃) and then H₂SO₄. Compared to the liquid phase oxidation, the gas phase oxidation is more important for sulfate formation [75]. In the daytime change curve, the SO₂ concentration began to increase earlier than SO₄²⁻, which indicated there were sufficient precursors for sulfate formation. The SO₂ concentration peaked at about 12:00, later than the early peak time, indicating that SO₂ mainly came from the emissions of overhead power plants and took a certain time to diffuse to the ground [76]. The concentration of SO₄²⁻ increased rapidly between 8:00 and 14:00, as in Handan, where it reached 10.0 from 6.1 μg/m³, which was consistent with the increase of O₃ and O_x during this period. In daytime, sulfate was mainly generated by SO₂ gas phase oxidation, and SO₄²⁻ concentrations were positively correlated with O_x concentrations (r = 0.70), indicating that the photochemical reactions played an important role in the SO₂ gas phase oxidation. Sun et al. [77] revealed that when the SOR value was higher than 0.10, SO₂ could be oxidized to sulfate through photochemical reactions. The average SOR values of Handan, Jining, Anyang and Kaifeng were 0.29, 0.37, 0.32 and 0.37, respectively, indicating the high secondary oxidation rate of SO₂. Therefore, the increase of SO₂ concentrations and the active photochemical reaction during this complex pollution episode emerged as pivotal factors in sulfate formation.

There are two main ways of achieving nitrate formation [78]; one is the gas-phase photochemical reaction between NO₂ and OH radicals, which mainly occurs in the daytime with high photochemical activities. Through these reactions, gaseous nitric acid (HNO₃) is generated, which can be adsorbed on the surface of particles or react with NH₃ to form granular ammonium nitrate. Ammonium nitrate is semi volatile, and there is a dynamic balance between the chemical reactions of HNO₃ and NH₃, which depends on T, RH, HNO₃ and NH₃ concentrations. High temperatures can promote the volatilization of ammonium nitrate.

The second is to generate NO₃ through NO₂ oxidation in the atmosphere, which can further react with NO₂ to form N₂O₅ during the nighttime. This is an important source of nocturnal HNO₃ because of the photolysis of NO₃, and nitrate therefore has different formation mechanisms in different periods of time. Taking Handan as an example (see Figure 6), the concentration of NO₂ increased slightly from 4:00 to 6:00, and the PBLH reached the minimum at the same time. The morning traffic emissions then increased, and the concentration of NO₃⁻ rose rapidly, peaking around 8:00, when the photochemical reaction was more active due to the increase of solar radiation; the O₃ concentration rose rapidly after 6:00, indicating that the formation of high concentration nitrate in the morning (6:00~10:00) was mainly due to gas-phase photochemical oxidation. After 10:00, the NO₃ concentration began to decrease, and decreased rapidly after 14:00 due to the further rapid rise of T and PBLH, which accelerated the volatilization of ammonium nitrate [79] and enhanced the atmospheric diffusion ability. At 16:00, the NO₂ concentration decreased to the minimum, and at 17:00, the NO₃ concentration decreased to a low value, and the T then began to decrease and the RH gradually increased. Although the O₃ concentrations gradually decreased during this period, it was still maintained at a high level (the average concentration between 17:00 and 23:00 was 87 μg/m³). Due to traffic emissions in the evening peak, the NO₂ concentration rose rapidly after falling to the minimum, and peaked at 20:00. At night, the high concentration of NO₂ and the presence of O₃, as well as the external conditions of weakened light and increased RH, promoted the conversion of NO₃ and/or N₂O₅ to HNO₃ through the hydrolysis reaction [80], which increased the NO₃ concentration. The diurnal variation curve of NOR was more consistent with O₃, indicating that the gas-phase photochemical oxidation had an important impact on the secondary conversion of NO₂. The average NOR values of Handan, Jining, Anyang and Kaifeng during the pollution episode were 0.31, 0.44, 0.38 and 0.42, respectively, which were higher than the SOR value, indicating that, compared to SO₂, the secondary conversion rate of NO₂ was higher, and its contributions to the PM_2.5 concentration were more obvious.

As the highest concentration component in secondary aerosols, SOA is mainly formed by VOCs through a series of photochemical oxidation reactions, via a very complex formation mechanism. According to the daily variations in Figure 6, the VOCs concentration reached their peak at 7:00, indicating that a large number of hydrocarbons were emitted from motor vehicles. The massive emission of VOCs and the rapid increase of O₃ concentrations meant SOA started to be generated rapidly around 9:00, and then peaked at 14:00. The correlation coefficient between SOA and O_x was 0.69, indicating that the formation of SOA was mainly promoted by photochemical oxidation during this complex pollution episode. Similar results were observed in the YRD region [81].

The high concentration of PM_2.5 in this complex pollution episode was mainly due to the high initial concentration and large amount of secondary aerosol. High concentrations of O₃ enhanced the atmospheric oxidation ability; and active photochemical reactions promoted the transformation of SO₂ and NO₂ to secondary inorganic salts, and VOCs to secondary organic aerosols, which increased PM_2.5 concentrations and worsened the air quality.

3.4. Source Apportionment of the High-Level O₃ and PM_2.5 Complex Pollution Episode

3.4.1. Analysis of O₃ Source Apportionment

The horizontal and vertical transport processes made important contributions to the high concentrations of O₃ in this complex pollution episode, and ISAM source analysis was able to quantify the regional transport contributions. Figure 7 shows the time series of contributions that different marked areas made to O₃ concentrations in Handan, Jining, Anyang and Kaifeng in the period from 31 March to 5 April 2019, showing significant differences in the impact of regional transports in different cities at different times. During the complex pollution episode, the contribution of the BCON to the O₃ concentrations in the four cities decreased by nearly half (compared to before this pollution episode), but a large proportion still remained, ranging from 28.5% to 39.2%. The contribution of BCON to the O₃ concentrations was predominant when O₃ levels were low (31 March to 1 April), and it conversely remained relatively stable and constituted a substantial proportion (approximately 30%) when the O₃ concentration was high (2 April to 5 April). Li et al. [63] also found that strong ultraviolet radiation and sinking airflow formed by high pressure were the main causes of boundary layer complex pollution, and especially O₃ pollution. Handan and Anyang are adjacent cities at the junction of the Hebei and Henan provinces, and so the time series of regional transport contributions were relatively similar. From 2 to 3 April, Shandong Province was located in the center of high pressure. Under the influence of anticyclone weather, O₃ was transported from Shandong to Hebei and Henan, meaning Shandong made stronger contributions to O₃ in Handan and Anyang higher than in their own provinces (Hebei and Henan). On 4 April, the prevailing south wind blew from Henan to Hebei, meaning Henan made a stronger contribution to O₃ in Handan than Hebei. The O₃ concentrations in Anyang were mostly contributed by Henan, since Anyang is located in the north of Henan Province. In addition, Shandong, Shanxi and Anhui provinces also made certain contributions to O₃ concentrations in Handan and Anyang. By 5 April, Hebei became the largest regional source of O₃ in Handan and Anyang, followed by Henan, Shandong and Tianjin. The O₃ concentrations in Jining on 3 April were mostly contributed by Shandong. Of the foreign sources, Jiangsu and other regions not labeled by ISAM also had significant contributions. In the next two days, the provincial source contributions decreased, having mainly been affected by the transport of foreign sources, such as Jiangsu, Anhui, Henan and Shanxi. On 3 April, the contribution of Henan to O₃ concentrations in Kaifeng was significantly lower than those of other sources, and O₃ was mainly transported from Shandong and Jiangsu provinces. The results indicated that, during the complex pollution episode, the contributions of foreign sources to O₃ concentrations in Handan, Jining, Anyang and Kaifeng reached 49.2%, 57.6%, 46.1% and 50.8%, respectively, much higher than those of provincial sources. From 4 April to 5 April, the contribution of local sources increased, and the formation of O₃ was determined by both local and foreign sources.

Figure 8 shows the spatial distributions of the contributions of different marked areas to mean O₃ concentrations, in the period from 3 to 5 April 2019. BCON contributed the most O₃ concentrations in the whole study area, especially for the northern part of the area, where BCON contributed more than 60 μg/m³ of O₃. In comparison to other provinces, Beijing and Tianjin had smaller areas with low pollution emissions, resulting in a more confined scope of impact. Although O₃ contributed by each area was mainly concentrated in the province and city itself, this also increased the O₃ concentration in the surrounding areas. The adjacent provinces had mutual regional transports. For example, O₃ from Jiangsu contributed 5~20 μg/m³ to the O₃ concentrations in Shandong, eastern Henan and northern Anhui; and O₃ from Shandong contributed 5~20 μg/m³ to O₃ concentrations in Hebei, Shanxi, Henan and Jiangsu, and 20~40 μg/m³ to O₃ concentrations in the Bohai Sea. Other regions contributed a large amount of O₃ to Shanghai and the western region, and also contributed 5~20 μg/m³ to the O₃ concentrations in the NCP region.

3.4.2. Analysis of PM_2.5 Source Apportionment

Figure 9 shows the time series of contributions of different marked areas to the PM_2.5 concentrations in Handan, Jining, Anyang, and Kaifeng, from 31 March to 5 April 2019. In contrast to O₃, PM_2.5 concentrations were jointly affected by the transports of local emission sources and external sources, while the contributions of BCON to PM_2.5 concentrations were found to be negligible. Since the transport path of pollutions was closely related to the evolution of the wind, the variation trends of the PM_2.5 concentrations contributed by the transports of each region were similar to those of O₃. According to the observations, PM_2.5 concentrations in four cities exceeded the standard on 2 April, and had accumulated before the complex pollution episode. During the complex pollution episode, the contributions of foreign sources caused by regional transports increased and the contributions of foreign sources to PM_2.5 concentrations in Handan, Jining, Anyang and Kaifeng ranged from 60.2% to 70.9%, with Shandong Province making an important contribution to the high PM_2.5 concentrations. The regional transports from Shandong contributed a large amount of PM_2.5 in the early stage of the complex pollution, which kept these cities at a high concentration of PM_2.5 in the early morning of 3 April. At 16:00 on 3 April, the PM_2.5 concentrations dropped to the minimum. After Henan’s contributions to the PM_2.5 concentrations in Handan, Anyang and Kaifeng increased, and Jiangsu’s contributions to the PM_2.5 concentrations in Jining also increased, the PM_2.5 concentrations began to rise gradually. On the afternoon of 4 April, after the PM_2.5 concentrations again fell to a low value, the provincial source emerged as the primary contributor to PM_2.5 concentrations in Handan, Anyang and Kaifeng. Conversely, in Jining, PM_2.5 concentrations were primarily influenced by foreign sources, from Jiangsu, Anhui, and Henan, in addition to provincial sources. Meanwhile, PM_2.5 concentrations in Jining were mainly contributed by foreign sources from Jiangsu, Anhui and Henan, along with provincial sources.

Figure 10 shows the spatial distributions of contributions made by different marked to mean PM_2.5 concentrations in the period from 3 to 5 April 2019. The PM_2.5 concentration contributed by BCON was negligible, and is therefore not shown in the figure. The spatial distribution ranges and degrees of PM_2.5 concentration contributed by each marker area were similar to those of O₃. Other regions contributed a large amount of PM_2.5 to the Shanghai, Xi’an and Hunan provinces, along with 2~10 μg/m³ to PM_2.5 concentrations in the NCP region. In contrast, the PM_2.5 concentration contributions of Shanxi Province, Beijing and Tianjin were relatively small. The high concentrations of PM_2.5 in the Hebei, Henan, Shandong, Hubei, Anhui and Jiangsu provinces, along with their mutual influence through transports, led to high PM_2.5 pollution in the NCP region.

4. Discussion

On the basis of the observational data, this study screened out a regional PM_2.5-O₃ complex pollution episode in the NCP region in the period from 3 to 5 April 2019, and analyzed its meteorological and pollution characteristics. The WRF-CMAQ model was employed to simulate the complex pollution processes. Subsequently, the formation mechanisms and sources of PM_2.5 and O₃ during this pollution episode were analyzed on the basis of the model results.

From 3 to 5 April 2019, PM_2.5 and O₃ concentrations in more than 10 cities in the NCP region simultaneously exceeded the standard for three consecutive days. The concentrations of PM_2.5 and O₃ on the complex pollution days were significantly higher than on the non-complex pollution day, indicating there would still be active photochemical reactions when PM_2.5 concentrations were high, leading to excessive O₃ concentrations. PM_2.5 concentrations in daytime were slightly higher than in nighttime on some days, indicating that high O₃ concentrations promoted the oxidation of secondary particles in the complex pollution processes, when the NCP region was affected by static and stable weather conditions, and the weather was sunny with high temperature and low wind speed. The atmospheric diffusion conditions were poor, which was conducive to the generation and accumulation of pollutants, resulting in the complex pollution episode. Note that on some complex pollution days, there were relatively small fluctuations in PM_2.5 concentrations for the whole day, which was because of temperature inversions and a stable nighttime atmosphere.

The process analysis results of O₃ showed that high concentrations of PM_2.5 weakened the photochemical effects of O₃ to a certain extent. However, the photochemical reactions at high altitudes were stronger than those on the ground, making a large number of O₃ generated by the gas-phase chemical reaction transport from the high altitude to the ground, significantly increasing the contribution of vertical transports to the formation of O₃ on the ground. Therefore, high O₃ concentrations in the complex pollution episode were mainly produced by the gas phase chemistry and the vertical transport process. The high PM_2.5 concentrations mainly came from the initial high concentrations and the formation of secondary aerosols. High O₃ concentration enhanced the atmospheric oxidation ability, and the active photochemical reactions promoted the transformation of SO₂ and NO₂ to secondary inorganic salts, and VOCs to secondary organic aerosols. This increased the PM_2.5 concentrations and aggravated air pollution.

The source apportionment results of O₃ showed that the contribution of the BCON to O₃ concentrations in the four cities during the complex pollution episode decreased by nearly half, compared to those before this pollution episode. During the complex pollution episode, the contribution of regional transports increased significantly and the contributions of foreign sources to the O₃ concentrations in the four cities (Handan, Jining, Anyang, Kaifeng) reached 49.2%, 57.6%, 46.1% and 50.8%, respectively. The source apportionment results of PM_2.5 also showed that during the complex pollution episode, the contribution of foreign sources caused by regional transport increased; the contribution of foreign sources to the PM_2.5 concentrations in four cities ranged from 60.2% to 70.9%, with Shandong Province making an important contribution to these high PM_2.5 concentrations. This study used the WRF-CMAQ model to analyze a representative regional PM_2.5-O₃ complex pollution process in the NCP region. It elucidated the formation mechanisms and conducted source analyses for O₃ and PM_2.5, providing valuable insights into the prevention and control of complex pollution in China.

The first limitation of this study lies in the poor simulation of wind direction by WRF, which failed to quantify the impacts and contributions of meteorological conditions in the process of pollutant transport. In the future, quantitative analyses can be conducted on the effects of wind speed and direction on pollutant transport. The second, limitation lies in this study only analyzing a case of PM_2.5-O₃ complex pollution in NCP in April 2019, meaning the findings should be further checked by using different methods. In the future, researchers should carry out case studies of other regions and in other periods, with the aim of better understanding the formation mechanism and pollution sources of complex pollution in China. Third, the areas marked in the source analysis process were based on provinces, and were therefore not accurate enough. In the future, the marked areas can be further refined to specific cities or areas, which will enable more detailed source analysis.

5. Conclusions

This study used the WRF-CMAQ model to analyze a representative regional PM_2.5-O₃ complex pollution episode in the NCP region, elucidated formation mechanisms, and conducted source analyses of O₃ and PM_2.5. The results indicated the high concentrations of O₃ mainly came from vertical and horizontal transports, highlighting that although photochemical reactions of O₃ near the ground were weakened by the high PM_2.5 concentrations, gas-phase chemical reactions at high altitudes would generate a large amount of O₃ and increase the ground level O₃ concentrations through the vertical transports. The high concentration of PM_2.5 came from the regional transport and the accumulation of secondary inorganic salts and secondary organic aerosols, which were facilitated by the high ground-level O₃. These findings provided valuable insights that will contribute to the prevention and control of complex pollution in China.