Classification of ozone pollution and analysis of meteorological factors in the Yangtze River Delta

ABSTRACT Serious regional ozone (O3) pollution often plagues the Yangtze River Delta (YRD). The formation mechanism of these regional pollution events, including the meteorological and emission factors leading to these pollution events and how to affect the distribution of O3, still needs further research and exploration. In this study, we first define the standard of O3 regional pollution in the YRD, and then select 248 regional pollution cases from 2015 to 2020 according to the defined standard. For the pollution cases in pollution months (May and June), PCT (principal component analysis in T-mode) classification method is used to classify the ozone concentration distribution in YRD area. The regional distribution of the O3 concentrations in the YRD is divided into five types, and the overall type (Type 1) accounts for 15%, which is related to the control of YRD area by high-pressure center. Under the control of high pressure, the weather is sunny with the high temperature, and this weather condition is favorable for ozone generation and intercity transmission, causing extensive pollution. The double center type (Type 2) accounts for 8%. This type of YRD is controlled by the front of the high pressure (the high-pressure center is located in North China), and the weather in the middle and north is conducive to the generation and transmission of O3. Inland type (Type 3) accounts for 24%. The main body of this type of high pressure is located in Mongolia. The easterly wind in YRD area is conducive to the inland transmission of O3 precursors. The northern coastal type (Type 4) accounts for 44%. This type of YRD area is mainly controlled by the weak pressure field. The weather in the northern coastal area is sunny and the solar radiation for a long time is conducive to the formation of O3. The southern coastal type (Type 5) accounts for 10%, the solar radiation is strong in the southern region mainly under the influence of the post-offshore high pressure. This study provides new insights into the relationship between O3 pollution distribution types and atmospheric circulation in YRD area, and reveals the difference of potential meteorological impacts of different O3 pollution distribution types.


Introduction
High concentrations of O 3 have adverse effects on human health (Cohen et al., 2017), forests and crops (Monks et al., 2015;Yue et al., 2017). Therefore, ozone (O 3 ) pollution is considered to be an important air quality problem and has attracted extensive attention in the field of atmospheric environment and atmospheric chemistry.
Since the implementation of the action plan for the prevention and control of air pollution in 2013, the concentration and exceeding standard rate of air pollutants (PM 2.5 , SO 2 , NOx and CO) has decreased significantly. However, the proportion of O 3 as the primary pollutant increases year by year Lu et al., 2018;Tang et al., 2009;Wang et al., 2017). With the accelerating process of urbanization, industrialization and regional economic integration, the temporal and spatial distribution characteristics of air pollution in China have changed significantly, from single air pollution to compound pollution, and the temporal and spatial variation characteristics of urban O 3 pollution have gradually changed from a single city to an urban agglomeration (Gao et al., 2016;Li et al., 2017;Maji et al., 2019;Yang et al., 2014). Ma et al. (2016) pointed out that China's high concentration O 3 pollution is mainly concentrated in four urban agglomerations, namely Beijing-Tianjin-Hebei, Yangtze River Delta (YRD), Pearl River Delta and Chengdu-Chongqing, and the regional characteristics of almost every heavy pollution process are very significant. The main characteristics of regional O 3 pollution events in the YRD are the continuous occurrence of high concentration pollution areas and wide spatial range, which brings difficulties to the emergency early warning of O 3 pollution joint prevention and control.
As the main component of photochemical smog, near surface O 3 is a secondary pollutant, which is mainly produced by a series of photochemical reactions with various precursors (CO, NOx and VOCs, etc.), which is closely related to the type, distribution, intensity and meteorological conditions of emission sources (Lei et al., 2019;Wang et al., 2021). In addition, local O 3 concentration will also be affected by atmospheric transport related to wind speed, wind direction and planetary boundary layer height (Bossioli et al., 2007;Darby, 2005;Souri et al., 2016). Many researchers have done a lot of studies about the relationship between weather patterns and O 3 pollution. For example, Russo et al. (2014) objectively classified and analyzed the atmospheric circulation affecting Portugal from 2002 to 2010. Gao et al. (2020) found that the highest average O 3 concentration is related to the southerly flow introduced by the Western Pacific subtropical high, the saddle area low pressure and the warm flow introduced by the typhoon, while the low average O 3 concentration is related to the northwest flow brought by the Siberian high. Strong warm weather patterns such as the front of continental high pressure and low pressure are the atmospheric circulation situation field that is easy to cause high O 3 values in Beijing-Tianjin-Hebei region (Wang, Ye, etal., 2018;Xue et al., 2014). Liao et al. (2017) studied the relationship between air pollutants (e.g. PM 2.5, O 3 ) and the types of atmospheric circulation in the YRD from 2013 to 2016, and found that O 3 pollution is closely related to the types of westerlies (mainly in summer). However, their research focused on the impact of different atmospheric circulation types on O 3 concentration in a certain region, and did not analyze the differences of the atmospheric circulation of different spatial distribution types of ozone concentration.
The effects of meteorological conditions controlled by specific synoptic patterns on regional surface ozone episodes have been widely discussed. Qi et al. (2017) analyzed the variation characteristics of O 3 concentration near the ground in Hangzhou and its relationship with meteorological elements. The research showed that the variation of O 3 concentration was positively correlated with UV radiation and temperature, and negatively correlated with relative humidity. In the east wind or partial east wind, the O 3 concentration is high, and the pollution source affecting the increase of O 3 concentration in Hangzhou mainly comes from the East.  analyzed the variation characteristics and meteorological causes of O 3 concentration near the ground in summer in eastern China from 2013 to 2017. For clarifing the impacts of synoptic climate systems and their interaction on long-term variation of ozone, Song et al. (2022) use the daily meteorological data and hourly surface ozone concentrations in the summer of 1950-2014 from CMIP6 to investigate the relationship between ozone variations and synoptic climate types (SCTs). At present, most studies focus on analyzing the formation and evolution mechanism of a single specific city or a single O 3 pollution process in the YRD (Jia et al., 2017;Li et al., 2020;Qin et al., 2021;Roxanne, 2004;Xu et al., 2017) and rarely involve the judgment criteria of regional O 3 pollution events. Therefore, there is a lack of regular understanding of the regional O 3 pollution process in the YRD, such as impact scope, pollution degree and duration. Secondly, because the formation of regional O 3 pollution process is closely related to meteorological conditions, the differences of circulation evolution and pollution sources will cause great differences in pollution scope and evolution process. However, previous studies lack the analysis of the differences of meteorological conditions between different O 3 pollution distribution types in the YRD. In this study, we try to reveal the difference of potential meteorological impacts controlling the different O 3 pollution distribution types in YRD, so as to provide support for O 3 pollution forecasting. Therefore, this study attempts to deepen the understanding of O 3 regional pollution in YRD area and explore the main circulation characteristics of different types of regional O 3 pollution, including (1) Using the surface O 3 observation data of the YRD from 2015 to 2020, this paper discusses the discrimination standard of regional O 3 pollution in the YRD; (2) The main distribution types of regional pollution and the probability of occurrence of different types are obtained by using objective methods to classify the O 3 concentration of regional pollution; (3) The atmospheric circulation characteristics of different pollution distribution types are comprehensively analyzed, and the atmospheric circulation and meteorological differences of different distribution types are revealed.

O 3 observation data
The in situ monitoring data for the hourly concentrations of O 3 can be acquired from the National Environmental Monitoring Center (https://air.cnemc.cn:18007/), and O 3 concentration was measured by Ultraviolet photometric method. The hourly pollutant concentration for a city is calculated as the average of the pollutant concentrations from several national monitoring sites in that city, which can better characterize the pollution level of the city. According to the technical regulation for ambient air-quality assessment (HJ 663-2013, http://www.mee.gov.cn/), we explored how the maximum daily 8 h running average O 3 (MDA8 O 3 ) concentration varies. The MDA8 O 3 concentration was calculated for each monitoring site based on the hourly data from the time period 08:00-24:00 for the days with at least 14 h of measurement data. If less than 14 h of valid data is available, the results are still valid if the MDA8 O 3 concentration exceeds the national concentration limit standard. Each city has at least two monitoring sites, and the MDA8 O 3 levels for a city are the corresponding averages over all sites in that city. The MDA8 O 3 values were collected for the period from 2015 to 2020, including 256 state-controlled points in 41 cities, including 73, 114, 10 and 59 stations in Anhui, Jiangsu, Shanghai and Zhejiang, respectively. Hourly averaged O 3 concentration in each city from January 2015 to December 2020 are used in this study. The procedures from Song et al. (2018) are used for the quality control for the measurements.
As shown in Figure 1, from 2015 to 2020, the O 3 concentration in the YRD experienced a first increase year by year, and the lowest value in 2015 was 89.6 μg·m −3 . It reached a high value in 2017 and 2018, respectively, 104.7 μg·m −3 and 103.2 μg·m −3 . From 2019 to 2020, the concentration shows a downward trend, from 102.2 μg·m −3 decreased to 98.9 μg·m −3 . From the perspective of monthly variation, the O 3 concentration is high from April to September in the summer half of each year. From the O 3 pollution exceeding standard rate of 41 cities in the YRD from 2015 to 2020, it is found that the number of exceeding standard days from May to June is the most, 26% and 25%, respectively, accounting for 51% of the total exceeding standard days in the whole year, and more than 90% of the severe and above large-scale pollution days occur in these 2 months. This is consistent with the analysis results of Wang et al. (2018) on the temporal and spatial characteristics of O 3 pollution in the YRD. It can be seen that the seasonal outbreak characteristics of O 3 concentration in the YRD are very obvious. We define the period from May to June as O 3 "pollution months" and study the regional O 3 pollution process of pollution months.

Meteorological and emission data
The atmospheric circulation data adopt the latest ERA-5 hourly reanalysis data (Kalnay et al., 1996) of the European medium range weather forecast center from 2015 to 2020, with a range of 20-50°N and 100-130°E, including Chinese Mainland, Mongolia and Siberia. The horizontal resolution is 0.25°×0.25°, 4 hours per day are 00:00, 06:00, 12:00 and 18:00 (UTC), including sea-level pressure (SLP) and 10 m horizontal wind (U and V), 500 hPa potential height and wind (U and V), and 850 hPa temperature. The surface meteorological observation data adopt the hourly observation data of temperature, relative humidity, horizontal wind speed and direction of 228 regional meteorological stations in the YRD from 2015 to 2020. The quality control is in strict accordance with the People's Republic of China meteorological industry standard QX/T 118-2010.
To investigate the O 3 source in the YRD region, the emission inventory of O 3 precursors over YRD is used for studying the spatial distributions of O 3 . It is obtained from Multiresolution Emission Inventory for China (MEIC, http://meicmodel.org/), with 0.25 degree resolution, including VOC and NOx across the YRD.

Classification of the spatial distribution of ozone
YRD is located in the east of China, as shown in Figure 1. It is composed of four administrative regions of Shanghai, Anhui Province, Jiangsu Province and Zhejiang Province, the location distribution of 41 major cities is shown in Figure 2(b). The northern part of the YRD is mainly plains, while the southern part is mainly low mountains and hills.
In order to reveal the influence range and differences of regional pollution processes in the YRD, this paper objectively classifies the concentration distribution of all regional O 3 pollution processes. The classification method adopts the PCT (principal component analysis in T-mode) method in the objective classification software (Philipp et al., 2016) developed by the European Union cost733 project. The principle is based on the T-mode principal component analysis method improved by Huth (2010). By comparing the applicability of five circulation classification methods to weather and climatology, Huth et al. found that T-mode oblique rotation principal component analysis (PCA) method was superior to traditional clustering methods in terms of the stability of classification results and low dependence on preset parameters (Huth, 2010). It decomposes the original highdimensional data matrix Z into two low-dimensional matrices, F and A. Z = FAT, F = (fim) and A = (aj) being the matrices of principal component (PC) scores and loadings, respectively. Here i indexes grid point and j and m index days. All principal components are sorted by the size of the corresponding eigenvalues, and the larger the eigenvalues, the greater the contribution to the original data. Finally, the 1st-K (K ≤ M) principal component F corresponding to the eigenvalue of the cumulative contribution rate of the original data exceeding a certain percentage (generally 85%), is taken to achieve the purpose of dimensionality reduction. This method can more accurately reflect the characteristics of the original circulation field, without much change due to the adjustment of the circulation classify object, and the resulting space-time field is more stable (Huth, 2010). In recent years, it has been widely used in the circulation classification of O 3 polluted weather in different regions Dong et al., 2020;Han et al., 2019;Savvedra et al., 2012;Trigo & DaCamara, 2000). The objective classification object in this paper is the spatial distribution of O 3 concentration near the ground. Therefore, the classification results mainly reveal the spatial influence range and degree of regional O 3 pollution process in the YRD. In addition, common statistical methods are used in this study, such as synthetic analysis, Pearson correlation analysis, and two tailed Student ttest.

Daily discrimination standard of regional O 3 pollution in the YRD
In this paper, the discrimination of regional O 3 pollution process in the YRD is mainly based on two points: Firstly, for there are few (256) observation stations of ambient air quality in the YRD, and they are mainly in urban. Therefore, we select "city" rather than "station" as the research object (a total of 41 cities). Secondly, the important feature of the regional process is contiguous. Considering Shanghai is relatively small and adjacent to Jiangsu, so Shanghai is merged with Jiangsu as a city. Thus, YRD includes Jiangsu, Anhui and Zhejiang provinces. If the O 3 concentration reaches the pollution level (MDA8 O 3 >160 μg·m −3 ) more than 10% cities in one province is defined as a regional O 3 pollution day. The relevant studies focus on continuous and large range in the discrimination of regional pollution processes, and discrimination depends on the number of stations and spatial uniformity in the study area (Niu et al., 2018;Wang, Ye, et al., 2018). According to the standards defined herein, there are 248 regional O 3 pollution days in the YRD from May to June in the period of 2015-2020. Figure 3a shows the distribution of O 3 daily maximum concentration levels in 41 cities in the YRD (arranged in adjacent order) on regional pollution days (248 days). It shows each regional pollution day covers 10-95% of the cities. Taking the occurrence of pollution days in three adjacent cities as the test threshold, the regional pollution days defined in this study can cover 100% of the cases. Therefore, the standard can reflect the large-scale and contiguous characteristics of the regional O 3 pollution. Comparatively, for non-regional O 3 pollution processes, most pollution days in the YRD are isolated and nonadjacent (Figure 3b). From 2015 to 2020, there are 14 heavy regional O 3 pollution days in the YRD, accounting for 6% of the total pollution days, and the moderate pollution days are 126, accounting for 51% of the total pollution days. The O 3 pollution days in YRD increased from 38 in 2015 to 49 and 48 in 2017 and 2018, respectively. Although the number of pollution days and maximum concentration decreased since 2019, pollution coverage did not decrease significantly. The situation of O 3 pollution in the YRD is still serious.

Regional O 3 pollution type in the YRD
The PCT method was used in this study to classify 248 typical regional O 3 pollution days in the YRD, and five O 3 concentration distribution types were obtained (Figure 4), with a cumulative variance contribution of 85%. The five types of O 3 can reflect different spatial distribution patterns in YRD, which were named integral type (Type 1), double center type (Type 2), inland type (Type 3), northern coastal type (Type 4) and southern coastal type (Type 5). Table 1 shows some statistics of five types. There are significant differences among the five types in the average of MDA8 O 3 , the average duration of pollution process and the number of pollution processes lasting days.
Type 1 occurred for 37 days (accounting for 15%), and the regional average concentration was the highest among the five types (183.3 μg·m −3 ), the longest duration of the average pollution process (2.17 days), which is shown in Table 1. The distribution is shown in Figure 4(a), the high concentration (larger than 160 μg·m −3 ) covers over 38 cities in Anhui, Jiangsu, Shanghai and Northern Zhejiang, exceeding 2/3 of the total area of the  (Table 1). Among them, the Taihu Lake basin around Jiangsu (Yangzhou, Zhenjiang, Changzhou and Wuxi), Maanshan in Anhui and Huzhou in Zhejiang is the most polluted area where MDA8 O 3 exceeds 200 μg·m −3 . In addition, more than 70% of the overall pollution process lasts for more than 2 days, and the longest process reaches 7 days (May 25, 2017-May 31, 2017. Type 2 has 20 days in total (accounting for 8%). The regional average concentration was 160.7 μg·m −3 , and the average pollution process lasting was 1.33 days. It is shown in Figure 4(b) that the spatial distribution of O 3 concentration, which is characterized by double center. The high concentration covered 21 cities, including most of Anhui, Jiangsu and Northern Zhejiang. The two centers were located in the north of Jiangsu and Anhui (213.3 μg·m −3 ) and central Jiangsu (202.4 μg·m −3 ). The average duration of Type 2 was 1.33 days, and 27% of the pollution process lasted for 2 days or more.
Type 3 has 59 days in total, accounting for 24%. The regional average concentration was 138.8 μg·m −3 . As can be seen from Figure 4(c), O 3 concentration was high in the northwest and low in the southeast of YRD. The concentration in inland Anhui Province was the highest (153.2 μg·m −3 ), and the average concentrations in Huaibei, Suzhou, Bozhou, Fuyang, Huainan, Chuzhou, Hefei and Ma'anshan were greater than 160 μg·m −3 . In Xuzhou, Suqian, Huai'an and Nanjing of Jiangsu Province, the average concentration was 145-175 μg·m −3 . The average duration of inland pollution process lasting was 1.63 days, and 36% of the pollution process lasting for 2 days or more. The interannual variation of Type 3 was obvious, increasing from 3 days in 2015-2016 to 16 days in 2017, becoming the main type of O 3 concentration distribution in the YRD, which indicated O 3 pollution in the YRD has tended to expand inland since 2017.
Type 4 has 108 days in total, accounting for 44%. The regional average concentration was about 135.2 μg·m −3 , and was the most frequency type in YRD. Figure 4(d) indicates Type 4 was characterized by high in the East and low in the West. Type 4 average duration was 1.74 days, about 12 pollution processes lasting for 3 days or more, accounting for 20%. The interannual change trend of the Type 4 was opposite to that of the Type 3 ( Figure 5), which decreased from 31 days in 2015 to 10 days in 2017, and then gradually rebounded to 18 days in 2020.
Type 5 occurred for 24 days, accounting for 10%, the regional average concentration was 128.7 μg·m −3 . Figure 4(e) indicates this type has the smallest impact range (9 cities) and the shortest duration (1.33 days) among 5 types. There were four processes lasting for 2 days or more, accounting for 22%. The high concentration covered the northern part of Zhejiang Province (149.2 μg·m −3 ) and Jiangsu Taihu Lake Basin (146.8 μg·m −3 ). The interannual variation of Type 5 was consistent with that of Type 3 ( Figure 5). The main difference between the two was in the location of the high-value center. Type 1 and type 2 have the largest impact range and the heaviest pollution degree from above -mentioned analysis. The changes of five types relative to 2015 are shown in Figure 5. Type 1 has increased significantly since 2017 and then decreased slowly. Type 2 rose slowly at first and then decreased slowly after 2018. Type 4 decreased significantly from 2016 to 2017, and began to recover slowly from 2018 to 2020. Type 5 showed a slow downward trend in recent years. Type 3 showed an upward trend as a whole, with a large increase from 2016 to 2017, which deserved special attention.
Compared with 2015, the five distribution types had the largest change in 2017, with the increase in Types 1−3 and the decrease in Types 4 and 5. To discuss the impact of atmospheric circulation on this phenomenon, we calculated the difference field of MSL and T2 m in the central and eastern China in 2017 and 2015. Figure 6a clearly shows that  the sea-level pressure in central and eastern China in 2017 increased significantly compared with that in 2015. Generally, the weather under high-pressure control is sunny and cloudless, with less wind. Furthermore, the T2 m difference field shows (Figure 6b) that the temperature in the north of the Yangtze River in the Yangtze River Delta region is higher, indicating that the solar radiation conditions in these regions are conducive to the generation of O 3 (Qi et al., 2017). Types 1−3 are related to the enhancement of the Siberian High, while Type 4 and Type 5 are related to the weak land pressure field in East China (detailed analysis in Section 3.4.3), so the numbers of Types 1−3 increase and the frequencies of Types 4 and 5 decrease.

Characteristics of meteorological elements for each type
Coupled with the increases in nitrogen oxide (NOx) and volatile organic compound (VOC) emissions, O 3 distribution in the lower troposphere is significantly influenced by winds, air temperature, cloud cover and downward shortwave radiation, which all affects the regional transport and chemical formation of O 3 Li et al., 2018;Xu et al., 2008).
From the perspective of VOC emissions (shown in Figure 7a), high emissions were mainly located in the region around Taihu Lake, provincial capitals and economically developed cities, such as Hefei, Nanjing, Hangzhou and Ningbo, among which Shanghai has the largest large value area. The distribution of NO x emission can be seen in Figure 7b. Similarly, Shanghai has the highest concentration and the largest coverage area, followed by the urban agglomeration in southern Jiangsu and the Nanjing-Hefei area. It is well known that VOC and NO x are important precursors for O 3 generation. Under appropriate meteorological conditions, their accumulation and transmission have an important impact on the increase or decrease in local O 3 concentration.
High ozone levels suggest that, besides the trans-boundary pollution, the effect of other factors on ozone concentrations in summer should be considered. It is well known that meteorology plays an important role in the formation, transport and dispersion of air pollutants. As a result, changes in local meteorological conditions, such as solar radiation, temperature, relative humidity and wind speed are important factors affecting the concentration of air pollutants. By analyzing the meteorological data, O 3 concentration and their relationship of five O 3 pollution distribution types in the YRD, the influence of weather factors on O 3 pollution distribution in the YRD is discussed. As can be seen from Table 2, for Type 1, the highest MDA8 O 3 was 183.3 μg·m −3 , about 20-50 μg·m −3 higher than the other four types. The average maximum temperature of Type 1 of YRD was 30°C. There were more sunny days, the average precipitation was less than 0.5 mm, the average relative humidity was 63.5%, and the average wind speed was 2.0 m·s −1 . Among the five types, the average wind speed of Type 1 was the lowest, and the dominant wind direction was southeast or northwest. The emission distribution of VOC and NOx in YRD (Figure 7), combined with small wind speed and favorable wind direction, has a significant influence on the increase and distribution of O 3 concentration. For Type 4, the MDA8 O 3 were 135.2 μg·m −3 , and the distribution of meteorological factors was relatively uniform, while the meteorological conditions were lower than that of Type 1. The dominant wind direction was southwest or south, resulting in Jiangsu Province and Shanghai O 3 pollution. The difference between Type 2, Type 3 and Type 5 of meteorological conditions mainly lies in the maximum temperature, precipitation and wind speed. The wind speed of Type 3 and Type 5 was the largest among the five types, and the dominant wind direction was northeast to east and southeast, respectively. The dominant wind direction was the most important factor affecting the diffusion and transmission process of O 3 pollution.
Taken Shanghai as an example to discuss the influence of radiation and mixing layer height on O 3 concentration, as shown in Figure 8. As the most important meteorological influence factor for O 3 generation, daily total radiation directly participates in O 3 generation. The height of mixed layer plays an important role in O 3 accumulation. When the daily radiation is more than 2500 MJ·m −2 , O 3 is prone to pollution, which reflects the important influence of radiation on O 3 generation. As far as Shanghai was concerned, the overall total radiation of Type 1 was relatively strong. The thickness of the maximum mixing layer was directly proportional to the daily total radiation, and the proportion of O 3 pollution was the highest (59%) among five types. However, it was not always that the higher radiation was, the higher the O 3 concentration was, which was also affected by the pollution transmission and diffusion conditions. For Type 4, there were many days when the total daily radiation was more than 2500 MJ·m −2 in Shanghai, and the radiation promoted O 3 generation significantly. When the radiation was more than 2500 MJ·m −2 and the boundary layer thickness was more than 1000 m, the probability of O 3 pollution was significantly increased (about 46%). For Type 2, Type 3 and Type 5, the occurrence probability of O 3 pollution in Shanghai was low, and the local generation of O 3 was significantly weaker than Type 1 and Type 4.

Atmospheric circulation characteristics
Previous studies have shown that O 3 pollution in eastern China is closely related to atmospheric circulation, which has a direct impact on O 3 pollution concentration distribution and regional transmission (Gao et al., 2016;Shu et al., 2019;Zhu et al., 2019). This section firstly counted the occurrence frequency of the atmospheric circulation types corresponding to the five O 3 pollution types in the YRD, took the circulation type with the highest frequency of each type as the atmospheric circulation characteristics of this type, and calculated the average field of each type of atmospheric circulation and the anomaly field relative to the average circulation. The atmospheric circulation included SLP & 10 m wind field, 500 hPa geopotential height & wind field, 850 hPa temperature & wind field, SLP, 500 hPa geopotential height and 850 hPa temperature anomaly field. Figure 9 shows the 500 hPa geopotential height anomaly field of five types, the geopotential height anomaly in the whole of East Asia were negative except for Type 4. The low potential height of Type 1 (Figure 9a) was the largest among the remaining four types, which deepened the trough in the east of Northeast China, and the YRD was located in the behind of the trough. This situation was conducive to YRD ground high-pressure strength. Different from the other four types, Type 2 showed a small negative anomaly in the Ural Mountain Baikal Lake area (Figure 9b). The meridional degree of Type 3 (Figure 9c) was smaller than Type 1 and Type 2, as a result, the ground cold high pressure was located to the north. Type 4 ( Figure 9d) showed a positive anomaly in the East Asia, indicating the weakening of the East Asian trough. There was a large negative anomaly near Ural Mountain Baikal Lake for Type 5 (Figure 9e), which indicated the trough in this region was relatively strong. While the anomaly in eastern China was small, the geopotential height in this region was close to the multi-year average value from May to June.

850 hPa temperature anomaly field
The 850 hPa temperature anomaly field ( Figure 10) indicates that Type 1, Type 2 and Type 3 were all controlled by negative anomaly in Chinese Mainland, while Type 4 was controlled by positive anomaly. Type 5 was controlled by weak positive anomaly in the central Inner Mongolia and Pingyuan in northern China, and the rest of China was controlled by weak negative anomaly. For Type 1, the temperature anomaly in YRD was between −1.2°C and −0.8° C, and it indicated that there is obviously cold air influencing YRD. The temperature anomaly distribution of Type 2 (Figure 10b) was similar to Type 1. The anomaly was −1.0-−0.7°C in the YRD. Type 3 (Figure 10c) anomaly in the YRD was between −0.6°C and −0.3°C. The whole Chinese Mainland was controlled by the positive temperature anomaly for Type 4 (Figure 10d), and the anomaly was between 0.9°C and 1.2°C in the YRD, and it indicated that the YRD area was controlled by the heating mass, conducive to the photochemical reaction in the northern coastal area of YRD. The temperature anomaly of Type 5 (Figure 10e) in the YRD was close to 0° C, therefore the temperature field characteristics of this type were not obvious. Figure 11 shows the eastern China was controlled by the cold high-pressure circulation for Type 1 (Figure 11a). The high-pressure center was located in the inland area of YRD, and the central pressure was about 1047 hPa, and the overall ground wind speed in the inland area of YRD was lower than 2 m·s −1 . The superposition of local generation and transportation formed a large-scale O 3 polluted air mass and maintained in the YRD, which made the pollution process continue and the pollution range expand. The control system of Type 2  was cold high-pressure circulation (Figure 11b). The main body of cold air mainly affected North China and Shandong. The cold high-pressure center entered the sea from Shandong Peninsula, and the YRD was located in the front of high pressure. The dominant wind in the north of the YRD was northwest, which was conducive to the input of O 3 from North China. In the meantime, the dominant wind in the south of YRD was southeast, which was conducive to the chemical generation of O 3 . On the one hand, downdraft prevailed in the middle and lower troposphere (Figure 9b), led to poor vertical diffusion. While the cold high-pressure circulation was conducive to pollutant transportation from North China to the north of the YRD. On the other hand, the solar radiation is strong under the control of the wedge -shaped weak pressure field, which was conducive to the generation of O 3 in central of Jiangsu. YRD was controlled by the East airflow at the bottom of the high pressure ( Figure 11c). The precursors generated locally in Jiangsu and Shanghai were transported downwind under the guidance of the easterly wind, which led to the increase of O 3 in Anhui. Type 4 has the greatest impact on Jiangsu Province and Shanghai among the five types. YRD was mainly controlled by the weak pressure field (Figure 11d). The weather in coastal areas such as Jiangsu Province was sunny and conducive to the generation of O 3 , especially in the afternoon. YRD was under the rear control of high pressure for Type 5 (Figure 11e), and the dominant wind direction was southeast, which led to O 3 pollution in Zhejiang Province. In general, for Type 1, the YRD area was influenced by the surface high system and downdraft in the middle and lower troposphere, forming a large range of O 3 polluted air mass in the YRD. For Type 2, due to the difference in circulation locations in the southern and northern YRD, the formation mechanism of the two O 3 pollution centers is quite different. The north of YRD area was influenced by the input of polluted O 3 air mass from North China, while the high center of O 3 southern area of YRD was related to the local photochemical reaction under the control of weak air pressure system. Type 3 and Type 5 were all controlled by the easterly or southeast airflow. The difference between them was in the location of the precursor emission source along with the direction of the wind field guiding the air transport, thus resulting in a different spatial distribution. Type 4 was a typical O 3 distribution type in the YRD, which was mainly caused by weak pressure field and convergence in Jiangsu and Shanghai. The strong illumination and radiation led to the local O 3 generation in the basin around Taihu Lake in the afternoon, owing to Taihu Lake being a high emission area for O 3 precursors, such as VOC and NO x (Figure 7). Figure 11. Synthesis of O 3 polluted sea level meteorological field (coloring and isoline) and 10m horizontal wind field in the YRD region of category 5: (a) Integral type (high pressure center); (b) Double center type (high pressure front) (c) Inland type (high pressure bottom); (d) Northern coastal type (weak gas pressure field); and (e) Southern coastal type (high pressure rear).

Sea level pressure field
The high-pressure system was the key control system of regional O 3 pollution in the YRD from the above analysis. The large-scale and long maintenance time of the highpressure system made the pollutants in different cities transmit with each other, which made it easy to form a large-scale polluted air mass. Weak pressure field was one of the most easily occurring weather systems in the YRD in spring, which is significantly related to the Type 4 with the highest frequency.

Discussion
Meteorology plays an important role in modulating tropospheric O 3 (Sekiya & Sudo, 2012). Compared with previous studies, this study attempts to determine the standard of O 3 regional pollution in the YRD. Based on the regional ozone standard, the regional O 3 pollution process in the YRD region from May 2015 to June 2020 was extracted. The objective method was used to classify the distribution of O 3 pollution in the YRD. The first five types with large variance contribution were selected as the main pollution distribution types in the YRD, and then explored the meteorological factors affecting these different types. Finally, the relationship between meteorological and O 3 pollution distribution types was analyzed, and the potential impact of the weather system on regional O 3 pollution was partially revealed.
The spatiotemporal variations of tropospheric O 3 are substantial at global and regional scales. In addition to photochemical reactions associated with O 3 precursor emissions and solar radiation, the atmospheric transport of O 3 and its precursors, including horizontal transport. The magnitude of horizontal transmission can be realized through mode simulation and numerical calculation, such as FLEXPART model and point-to-point transmission model calculated by big data analysis.
Tropospheric O 3 can also change atmospheric chemistry because its photolysis in the presence of water vapor is the primary source of the hydroxyl radical (OH), which is responsible for the removal of many important trace gases (Logan et al., 1981;Thompson, 1992). All these need to be evaluated and analyzed through observational data in the future.

Conclusion
The high-value period of O 3 regional pollution in the YRD occurs from May to June, accounting for 51% of the total days exceeding the standard in the whole year, and more than 90% of the severe regional pollution days occur in these 2 months. A standard for regional O 3 pollution days in the YRD was proposed, which could cover 100% of the YRD regional pollution cases and could reasonably reflect the annual change trend of YRD O 3 regional pollution day. According to a divided spatial study of 248 regional O 3 pollution days in the YRD, five types of regional pollution were discovered by objective classification method. The dominant weather system and its diffusion and transport conditions of O 3 pollution in different types were revealed through circulation synthesis analysis. Highpressure system was the main control system for YRD regional O 3 pollution with largescale and long maintenance. O 3 precursors NO x and VOC were mainly distributed in economically developed areas. Under the control of high pressure, it is conducive to the generation of O 3 and the mutual transmission between cities and easy formation of large-scale O 3 polluted air mass.
Under the background of increasingly serious ozone pollution in the YRD region, the results of this study can deepen our understanding of the relationship between different O 3 regional pollution distribution and large-scale weather system in the YRD, and provide support for pollution prevention and control. In addition, O 3 pollution between cities in the YRD has the characteristics of mutual transmission.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This research is supported by the National Natural Science Foundation of China (Grant no. 42005055), the Natural Science Foundation of Shanghai (Grant no. 19ZR1462100), and the Shanghai Science and Technology Commission (Grant no. 19DZ1205003).

Notes on contributors
Yu Cao is a senior engineer working in Shanghai Meteorological Bureau. Her research interests relate to the meteorological influences upon air pollution. Jinghui Ma received the Ph.D. degree from the Institute of Atmospheric Sciences of Fudan University, China, in 2022. She is a professor level senior engineer working in Shanghai Meteorological Bureau. Her research interests are the synoptic impact mechanism in combination pollution of ozone and PM 2.5 pollution, and sub-seasonal prediction technology of air pollution based on machine learning and predictability analysis.
Yuanhao Qu's research interest focuses on technical development of meteorological conditions for air pollution and forecasting of PM 2.5 and O 3 based on deep learning.

Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.