In this study, eight stations located in the core area of THB were selected as the study area, namely, Xiangyang (Xiangy), Jingmen (Jingm), Jingzhou (Jingz) and Wuhan (Wuh) in Hubei province, and Yueyang (Yuey), Changde (Changd), Yiyang (Yiy) and Changsha (Changs) in Hunan province (Figure 1). Since the PM_2.5 data in Hubei and Hunan areas has been gradually complete since 2015, this study discusses the hourly PM_2.5 concentration observations in November, December, January, February and March that are significantly affected by air pollution from 2015 to 2021. Meanwhile, the hourly PM_2.5 concentration observations of 1572 ambient air quality monitoring stations in China were used to calculate the transport index (TRI). The hourly PM2.5 data are derived from the Air Quality Realtime Publishing Platform of China National Environmental Monitoring Centre (http://106.37.208.233:20035/).

The meteorological data of the corresponding period mainly includes the following three parts: (1) Meteorological elements such as pressure, temperature, relative humidity and wind speed at eight national meteorological stations. (2) The reanalysis data in the horizontal resolutions of 1° × 1° obtained from National Center for Environmental Prediction (NCEP) (https://rda.ucar.edu/datasets/ds083.2/). The reanalysis data were interpolated to eight stations by the principle of proximity. The two parts data are used to analyze the meteorological conditions of the regional pollution and calculate the stable index (STI). (3) The Final Reanalysis Data derived from NCEP’s Global Data Assimilation System (GDAS) with 6-h intervals. This data is used in the Hybrid Single Particle Lagrangian Integrated Trajectory model (HYSPLIT-4) to calculate the TRI in the case of regional pollution.

Through the method proposed in the above section, we found 331 RPDs in the 30 months from 2015 to 2021 (November and December of the year and January, February and March of the following year) in THB. On average, more than one-third of the days in a month would have regional pollution. Therefore, it is imperative to study the temporal and spatial distribution of RPDs. Figure 2 shows the variation of RPDs and the differential distributions of pollution stations during RPDs. Overall, the RPDs show a decreasing trend, but the rate of decrease is slow and small. December and January are the months with the most RPDs, but they do not show a noticeable reduction. In addition, the occurrence time of RPDs is becoming more and more concentrated. RPDs in December and January account for more than 85% of the total RPDs from November 2020 to March 2021. Besides, RPDs show a minimum value in March, with the pronounced reduction trend. Specifically, the RPDs in March 2018–2021 are less than 3 days.

From the perspective of spatial distribution, the distribution of stations with regional pollution over the THB shows interesting characteristics, which can be divided into three stages. Firstly, in the winter halfaear of 2015/2016, the number of polluted stations during RPDs is usually more in Hubei than in Hunan, and only two RPDs show more stations in Hunan. Secondly, in the winter halfaear of 2016–2019, the number of polluted stations in Hubei decreased significantly, and those in Hunan increased gradually. Thirdly, in the winter halfaear of 2019/2021, Hunan has as many polluted stations as Hubei in most RPDs. Especially in 2020/2021, there were only 5 days in which the polluted stations in Hubei were more than those in Hunan, and there was no RPD in which all the polluted stations were in Hubei. It shows that the dispersion of the spatial distribution of RPDs increased significantly, and the polluted stations tend to concentrate in Hunan.

Figure 3 shows the proportion of polluted weather at eight stations and the consistency characteristics among stations during the RPDs. It can be seen that the proportion of polluted weather at four stations in Hubei is significantly higher than that in Hunan. Specifically, the proportion of polluted weather in Xiangy is over 95%. That’s because Xiangy is located in the gap of the northern part of the THB, which provides a natural channel for the regional transport of air pollutants from the North China Plain. As a result, the contribution rate of regional transport of PM_2.5 in this area has reached over 70% (Hu et al., 2021). It is also the area most affected by meteorological conditions (Sun et al., 2022a). In addition, Changd is the city with the highest proportion of polluted weather in Hunan, accounting for 75.8%. Because of its geographical location and meteorological conditions, it is more vulnerable to the impact of pollutant transmission (Dai et al., 2019).

Moreover, when polluted weather occurs, there is a strong correlation among stations, that is, there is a high probability that adjacent areas will have polluted weather at the same time. For instance, when polluted weather occurs in Xiangy, Changd will also have polluted weather 79% of the time. When there is polluted weather in Xiangy and Jingm, the probability of polluted weather in the other six stations is over 90%. On the contrary, when polluted weather occurs at four stations in Hunan, the probability of polluted weather in Xiangy and Jingm is relatively small. Especially in Yuey, the probability of polluted weather in Xiangy and Jingm is only about 63%. It shows that the polluted weather in certain regions of Hunan is weakly affected by the typical transportation pollution. Interestingly, Yuey has a good consistency with other four stations except for Xiangy and Jingm. When polluted weather occurs in these stations, Yuey is most likely to be polluted, indicating that although Yuey has the smallest number of RPDs, the regional polluted weather in the middle and south of the THB often covers Yuey area.

The synoptic patterns of 331 RPDs (the primary pollutant is PM_2.5) in the THB from 2015 to 2021 were divided into four types (Figure 4).

The first type is uniform pressure field (P1). The high-pressure center primarily controls it after the cold high-pressure moves southward. The pressure gradient is small, the ground wind is weak, the wind direction is uncertain (Figure 4A), and the pollution transmission is weak. The atmospheric circulation is often in a static and stable state, with poor diffusion capacity. The long-distance transported pollutants brought by the high-pressure going south are not easy to diffuse and output, and the local accumulation of pollutants is significant. The frequency of P1 type is low (42 days), accounting for 12.7% of the total, and the pollution is mainly caused by local pollution accumulation.

The second type is front of cold high-pressure bottom (P2). In high latitude Mongolia, cold high-pressure often develops strongly and spreads southward (the central intensity reaches 1042 hPa). The high-pressure gradient area in front of the high-pressure bottom is often accompanied by the northerly wind of about 2∼7 m/s (Figure 4B), which spreads from north to south and affects the THB. The strong wind carries pollutants from the north to the south when the cold front goes to the south. Affected by the cold front, the pollution concentration in the THB increases significantly, and the long-distance transportation of pollution is evident. The frequency of P2 type is the highest (180 days), accounting for more than half (54.3%) of the total number. It mainly occurs in autumn and winter. The northerly wind brings strong long-distance input of pollutants when the emission of pollution sources in the north increases.

The third type is the back of high-pressure (P3). Mongolia high-pressure often moves eastward into the sea, and it is easy to form severe pollution in North China and East China in the process of moving eastward and southward. After the high-pressure enters the sea, the THB is at the rear of the high pressure. Affected by the return transportation of weak easterly air flow (2–4 m/s) (Figure 4C), the pollutants in the east are transported westward to the THB, causing pollution growth in this area. However, the frequency of this type is low, accounting for only 24.5% (81 days), and the short distance backflow of the pollutants in the east of the upstream mainly transports it.

The fourth type is low pressure with an inverted trough field (P4). The low level is characterized by the development of warm low-pressure in the southwest region (the central intensity is 1012 hPa), and the THB is at the top of the low-pressure trough. The convergence of easterly and southerly flows is conducive to the horizontal accumulation of pollutants (Figure 4D). Moreover, the southerly air flow (2–6 m/s) guides the water vapor in the southern coastal area to the north, and the humidity in the THB increases. The high humidity and air flow convergence are conducive to the local accumulation of particulate matter and the secondary heterogeneous chemical generation. P4 type has the least occurrence (28 days), accounting for only 8.5%. The upstream pollution of this type is light, and the input is weak. Therefore, it is dominated by the wet growth and accumulation of local pollution sources.

PM_2.5 concentration is a direct reflection of the degree of air pollution, and the height of the boundary layer, relative humidity and wind speed are the meteorological factors that significant impact air pollution (Ding et al., 2016; Dong et al., 2017; Huang et al., 2018). Figure 5 shows the distribution characteristics of PM_2.5 concentration, the boundary layer height, relative humidity and wind speed in four types of regional pollution processes. It can be seen that different from the heavy pollution weather caused by the P1 type in North China and the Yangtze River Delta (He et al., 2018; Dai et al., 2020), the average PM_2.5 concentration caused by the P1 type is the lowest among the four weather types during the regional pollution process in the THB. In addition, the average value of PM_2.5 concentration in Hubei is the highest under the control of the P2 type, while that in Hunan is the highest under the P4 type. It is closely related to the different geographical locations of Hubei and Hunan, which are located in different locations of the weather system, and the stations in Hubei are located in the main channels of pollutant regional transmission (Bai et al., 2022; Wang et al., 2022). The average concentration of PM_2.5 under the influence of P3 type is lower than that of P2 type and P4 type, but higher than that of P1 type, showing a consistent feature at each station.

The atmospheric boundary layer is the atmosphere below the discontinuous interface with turbulent characteristics. The height of the boundary layer affects the vertical diffusion of pollutants (Liu et al., 2018; Li et al., 2019). The average and maximum values of the boundary layer height on the regional pollution day under the influence of P2 type are significantly higher than those under other weather types. For example, the average height of the boundary layer in Xiangy, Jingm and Jingz is about 400 m, while that in the other five stations is about 380 m, and the maximum value basically exceeds 900 m. The boundary layer heights of the other three weather types are similar, with an average height of about 280 m and a maximum value of less than 600 m. The low height of the boundary layer and the weak atmospheric diffusion capacity are conducive to the accumulation of pollutants, which are typical characteristics of stagnant air pollution, while the typical transmission air pollution has a high boundary layer height (Bai et al., 2018; Zhou et al., 2022).

The relative humidity reflects the formation of secondary organic aerosol in air pollution to a certain extent. The higher relative humidity is easy to generate the secondary organic aerosol, and even under the background of significantly reduced emissions, the generation of secondary organic aerosol will dominate the occurrence of polluted weather (Huang et al., 2021). In general, the relative humidity of the RPDs in Xiangy and Jingm is relatively small, while that in other six stations is similar. The highest daily relative humidity of regional pollution in Hubei is under the influence of P4 type, while that in Hunan is under the influence of P3 type. The confrontation between the cold and warm air mass in Hunan makes it easier to have a higher relative humidity (Bai et al., 2022).

The surface wind speed can directly reflect the horizontal diffusion conditions. The maximum average wind speed occurs at all eight stations under the influence of P2 type, while the minimum average wind speed occurs under the influence of P1 type. It can be clearly seen that the average wind speed in Xiangy and Jingm under four weather types is significantly higher than that in other stations. Specifically, the average wind speed in Jingm is close to 5 m/s under the influence of P2 type, while the maximum wind speed can reach about 10 m/s, which is the result of the joint action of special terrain features and atmospheric circulation (Sun et al., 2022b).

The above analysis shows the circulation background, pollution characteristics and meteorological conditions under four weather types. In this section, we will discuss the performance of STI and TPI during these four-types pollution. The pollution indices are built based on the observation data during the winter halfaear of 2015–2021 (November of the year to March of the following year). The index model is then validated using the observation data in November to December 2021 and January to March 2022.