Analysis of the Trend Characteristics of Air Pollutants in the Fenwei Plain Based on the KZ Filter

Abstract

In order to improve air quality, China has implemented a series of the most stringent control measures ever in recent years. Quantitatively analyzing the contribution of emissions to the trend change in air pollutants is an essential scientific basis for verifying the effectiveness of air pollution control. We based our study on the air quality online monitoring data and meteorological element data of 11 cities in the Fenwei Plain from 2018 to 2022. We quantitatively investigated the changing patterns of NO₂, O₃, PM₁₀, and PM_2.5 and their influencing factors in the major cities of the Fenwei Plain by using the KZ filtering and MLR modeling analysis methods. The results show the following: (1) The long-term fractions of NO₂, PM₁₀, and PM_2.5 in the Fenwei Plain decreased by 10.5, 33.1, and 17.1 μg·m⁻³, with decreases of 25.8%, 29%, and 28.8%, respectively, from 2018 to 2022. The long-term fractions of O₃ showed the characteristics of decreasing and then increasing, with 2020 as the dividing line. (2) The short-term components of NO₂, PM₁₀, and PM_2.5 contributed the most to the total variance, with the proportion of short-term components ranging from 34.7% to 69.8%, 53% to 73%, and 43% to 58%, respectively. The seasonal components of O₃ contributed the most to the total variance, with the proportion of short-term components ranging from 54% to 70.7%. (3) Most cities in the Fenwei Plain had unfavorable meteorological conditions with regard to NO₂, PM₁₀, and PM_2.5 in 2018–2020 and favorable meteorological conditions in terms of NO₂, PM₁₀, and PM_2.5 in 2020–2022. O₃ showed different characteristics from the other three pollutants. Most cities in the Fenwei Plain had meteorological conditions in 2018–2019 that were unfavorable for improving O₃ levels. In 2019–2021, meteorological conditions were favorable for improving O₃ levels, while in 2021–2022, meteorological conditions were unfavorable for improving O₃ levels.

1. Introduction

Air pollution is caused by a combination of enormous emissions of various pollutants and meteorological conditions, with emissions being the leading cause of pollution [1]. Methods for distinguishing the contribution of meteorological conditions and abatement measures to changes in pollutant concentrations can generally be categorized into two approaches. One is based on source-emission inventories and atmospheric chemical transport models, and the other is based on the statistical analysis of observed data. In recent years, many scholars have found that emission reduction is the dominant factor in ameliorating particulate matter pollution in the atmosphere, based on atmospheric chemistry transport modeling and pollution-source emission inventory data [2,3]. One of the advantages of using atmospheric chemistry models to assess the contribution of emissions to atmospheric particulate pollution is the ability to characterize the spatial distribution and quantify the direct response of total air pollutant emissions to changes in air quality [4]. Methods for atmospheric chemical transport modeling are limited by the model performance, emission inventory uncertainties and lags, imperfect chemical mechanisms, and strong nonlinearity of the model [5]. These factors can introduce some uncertainty into the model results.

In recent years, many scholars have begun experimenting with methods based on mathematical statistics, which have the advantage of not requiring emission inventories as inputs or detailed information on the sources and formation mechanisms of pollutants [6]. When utilizing statistical analyses, statistical models of the changes in pollutant concentrations and the emissions from meteorological factors and sources are generally developed. Then, the contributions of meteorological conditions and emission factors to changes in pollutant concentrations are quantified according to these statistical models [7]. Zhai used multiple linear regression (MLR) modeling to quantify the effects of changes in meteorological conditions on PM_2.5 trends in China during 2013–2018 [8]. Some scholars have explored the relationship between pollutant concentration trends and anthropogenic emissions and meteorological conditions in typical urban areas in China by combining Kolmogorov–Zurbenko (KZ) filtering and MLR modeling [9,10]. However, most of these studies have focused on the period before the 2020 epidemic, and there needs to be more understanding of the impact of emission reduction measures on air pollution in recent years. Moreover, previous studies have focused on particulate matter, and studies on other air pollutants such as O₃ and NO₂ are scarce.

The Fenwei Plain has become one of the most polluted areas in China due to its own basin topography and industrial structure [11]. The Ministry of Ecology and Environment proposed a Three-Year Action Plan to win the battle to defend blue skies in 2018. It included the Fenwei Plain as a critical monitoring area, but the exceedance of atmospheric pollutant concentrations still occurred. Since the outbreak of the COVID-19 epidemic, the socioeconomic life of the cities in the Fenwei Plain has been seriously affected, and the characteristics of atmospheric pollutants have shown new changes [12]. The Fenwei Plain is in a critical economic and strategic position in China, and an excellent ecological environment is crucial for the region’s development. In this paper, we based our study on the air quality online monitoring data and meteorological element data of 11 cities in the Fenwei Plain from 2018 to 2022. We quantitatively investigated the changing patterns of NO₂, O₃, PM₁₀, and PM_2.5 in significant cities of the Fenwei Plain, as well as their influencing factors (including anthropogenic discharge and meteorological condition elements), using KZ filtering and multiple linear regression. This will provide scientific evidence for understanding near-surface particulate matter pollution management in this region.

2. Overview of the Study Area

Fenwei Plain (Figure 1) is the general name for the Fen River Plain, Wei River Plain, and their terraces in the Yellow River Basin, including Lvliang (LL), Jinzhong (JZ), Linfen (LF), Yuncheng (YC), Baoji (BJ), Tongchuan (TC), Xianyang (XY), Weinan (WN), Sanmenxia (SMX), Luoyang (LY), and Xi’an (XA); a total of 11 cities. The topographic relief of the Fenwei Plain is generally low in the center and high in the surrounding area, consistent with the altitude distribution [13]. The Fenwei Plain is a vital energy and heavy-industry base in China, accounting for one-fifth of the national raw coal production and about one-tenth of the national coal consumption. It is also one of the critical regions for air pollution prevention and control in China.

3. Data Sources

The air pollutant concentration data are the hour-by-hour observations of NO₂, O₃, PM₁₀, and PM_2.5 concentrations in 11 cities. The data period is 1 January 2018–31 December 2022. There is more than one measurement point for each air pollutant. The measurements were taken as often as once an hour. They were averaged by some algorithm to obtain the hour-by-hour concentration data for each city and posted on the China General Environmental Monitoring Station (http://www.cnemc.cn/, accessed on 4 July 2023). There are 24 pieces of data for each day. For analysis purposes, we performed different operations for different metrics. The daily evaluation indexes for NO₂, PM₁₀, and PM_2.5 are the mean concentration values of 24 h. The daily evaluation index for O₃ is the daily maximum 8 h sliding average (O₃MAD8).

Our research uses the ground meteorological data from the rp5 world historical weather data site (https://rp5.ru/docs/about/cn, accessed on 4 July 2023). Regarding the meteorological data, there is only one measurement point per city, which comes from a monitoring point closest to the city center. The frequency of measurements is once every three hours. These are 2:00, 5:00, 8:00, 11:00, 14:00, 17:00, and 20:00 daily. Many studies have used daily average meteorological data to analyze the effects of pollutants. For a five-year study scale, daily mean concentrations are a good representation of the effects of meteorological conditions on pollutants. Based on previous studies [14], we selected four meteorological indexes: air temperature, air pressure, wind speed, and relative humidity. The meteorological elements used in the study are the daily average temperature, daily average pressure, daily average relative humidity, and daily average wind speed of the city.

4. Research Methods

4.1. Kolmogorov–Zurbenko Filter

KZ filtering is a method based on time-series analysis. The original series can be better decomposed into short-term, seasonal, and long-term component series by setting different parameters. In KZ filtering, the original air pollutant time series can be represented by Equation (1) [15]:

$$\mathrm{X}\left(\mathrm{t}\right)=\mathrm{E}\left(\mathrm{t}\right)+\mathrm{S}\left(\mathrm{t}\right)+\mathrm{W}\left(\mathrm{t}\right)$$

(1)

X(t) is the original time series. E(t), S(t), and W(t) are the long-term component, seasonal component, and short-term component, in that order. The long-term component is mainly influenced by factors such as the long-term trend of pollutant emissions and climate change [16]. The seasonal component arises from seasonal changes in meteorological conditions and energy consumption patterns [17]. The short-time alterations are residuals, mainly affected by the role of small-scale weather regimes and short-time alterations in external emissions [18]. KZ filtering belongs to low-pass filtering, generated by repeated iterations of the moving averages. The kza program package in the R software mainly implements the KZ filtering method:

$$Y_i=\frac{1}{m}{\sum }_{j=-k}^k{X}_{i+k}$$

(2)

$Y_i$ is the filtered time series, and the computed $Y_i$ becomes the input for the second iteration and the iterated p times. $X_i$ is the original data. The sliding window length $m$ = 2$k$ + 1. The parameter i is the interval of the time series, and $j$ is the sliding window variable. kZ(m,p) means that the sliding window is m and has been iterated p times. Adjusting the window length, m, and the number of iterations, p, can control the filtering on different time scales. To filter periods of less than N days, the following criterion is applied to determine the filter’s effective width [19]:

$$\mathrm{m}\times {\mathrm{p}}^{\frac{1}{2}}\le \mathrm{N}$$

(3)

The parameters m and p are the KZ filtering factors, which can realize different filtering function scales. Equation (3) indicates that high-frequency filters with wavelengths less than N can be filtered out. It has been shown that the sum of long-term and seasonal components of meteorological and air quality data can be extracted using KZ(15,5) filtering, defined as ${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)$. ${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)$ is the baseline component that reflects the relatively stable trend of the original data [20,21]. The long-term trend of the data can be obtained by selecting a larger window, and the long-term pollutant component E(t) can be obtained by utilizing KZ(365,3) filtering [22,23]. The seasonal component S(t) is the difference between the baseline component ${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)$ and the long-term component E(t). The short-term component W(t) is the difference between the original series X(t) and the baseline component ${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)$, as shown in Equations (4)–(7), where ${\mathrm{X}}_{\mathrm{b}}\left(\mathrm{t}\right)$ = ${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)$:

$${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)={\mathrm{E}\left(\mathrm{t}\right)+\mathrm{S}\left(\mathrm{t}\right)=\mathrm{K}\mathrm{Z}}_{(15,5)}$$

(4)

$$\mathrm{E}\left(\mathrm{t}\right)={\mathrm{K}\mathrm{Z}}_{(365,3)}$$

(5)

$$\mathrm{S}\left(\mathrm{t}\right)={{\mathrm{X}}_{\mathrm{b}}\left(\mathrm{t}\right)-\mathrm{E}\left(\mathrm{t}\right)={\mathrm{K}\mathrm{Z}}_{(15,5)}-\mathrm{K}\mathrm{Z}}_{(365,3)}$$

(6)

$$\mathrm{W}\left(\mathrm{t}\right)={\mathrm{X}\left(\mathrm{t}\right)-{\mathrm{X}}_{\mathrm{b}}\left(\mathrm{t}\right)=\mathrm{X}\left(\mathrm{t}\right)-\mathrm{K}\mathrm{Z}}_{(15,5)}$$

(7)

4.2. Multiple Linear Regression

Two main factors affect the fluctuation of the long-term pollutant components around the mean value: changes in pollutant emissions and changes in meteorological conditions. The effects of meteorological conditions can be more effectively eliminated by modeling the short-term component of the pollutant time series, W(t), and the baseline component, ${\mathrm{X}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}\left(\mathrm{t}\right)$, with meteorological factors as independent variables [24,25]. The regression model is shown below:

$$\mathrm{W}\left(\mathrm{t}\right)={\mathsf{\alpha }}_0+\sum {\mathsf{\alpha }}_{\mathrm{i}}{\mathsf{\omega }}_{\mathrm{i}}\left(\mathrm{t}\right)+\mathsf{\epsilon }\mathsf{\omega }\left(\mathrm{t}\right)$$

(8)

$${\mathrm{X}}_{\mathrm{b}}\left(\mathrm{t}\right)={\mathrm{b}}_0+\sum {\mathrm{b}}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{j}}\left(\mathrm{t}\right)+\mathsf{\epsilon }\mathrm{b}\left(\mathrm{t}\right)$$

(9)

$$\mathsf{\epsilon }\left(\mathrm{t}\right)=\mathsf{\epsilon }\mathsf{\omega }\left(\mathrm{t}\right)+\mathsf{\epsilon }\mathrm{b}\left(\mathrm{t}\right)$$

(10)

Equations (8) and (9) show the multiple regression models between short-term and baseline air pollutants and meteorological factors. W(t) and X_b(t) are the air pollutants’ short-term and baseline components. ω_i(t) and x_i(t) are the different short-term and baseline components of the i-th meteorological factor. Parameters a₀ and b₀ are the fitting parameters. ε_W(t) and ε_b(t) are the regression residuals of the short-term and baseline components. ε(t) denotes the total residual, which represents the change in the pollutant concentration influenced only by the emission after removing the effect of meteorological conditions. The multiple linear regression was mainly realized by the SPSS27.0 software. The SPSS27.0 software platform provides advanced statistical analysis, many machine-learning algorithms, and text analysis with open-source scalability, which can be integrated with big data and seamlessly deployed into applications.

Next, a KZ(365,3) filter was used on ε(t) to obtain ε_E(t). After removing the effect of meteorological variations, the reconstructed time series of air pollutants E_adj(t) is calculated as the sum of the mean of E(t) and ε_E(t), which can be referred to as meteorological adjustment since it is only affected by emissions and has no meteorological effect:

$${\mathrm{E}}_{\mathrm{a}\mathrm{d}\mathrm{j}}\left(\mathrm{t}\right)=\overline{\mathrm{E}\left(\mathrm{t}\right)}+{\mathsf{\epsilon }}_{\mathrm{E}}\left(\mathrm{t}\right)$$

(11)

The relative contribution of meteorological conditions to pollutants after KZ filtration can be calculated as:

$${\mathrm{P}}_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{b}}=\frac{{\mathrm{K}}_{\mathrm{o}\mathrm{r}\mathrm{g}}-\mathrm{K}}{{\mathrm{K}}_{\mathrm{o}\mathrm{r}\mathrm{g}}}\times 100\mathrm{\%}$$

(12)

where P_contrib is the relative contribution of meteorological conditions to pollutant changes, K_org is the slope of the change in the original pollutant time series, and K is the slope of the change in the meteorologically adjusted pollutant time series.

5. Results and Discussion

5.1. Trend Changes in the Original Series, Short-Term Components, Long-Term Components, and Seasonal Components

Ambient air quality is the standard for implementing the “People’s Republic of China Environmental Protection Law” and “People’s Republic of China Law on Prevention and Control of Air Pollution” to protect and improve the living environment, the ecological environment, and human health standards. Ambient air functional zones are divided into two categories: Type I zones are nature reserves, scenic spots, and other areas requiring special protection; Type II zones are residential areas, mixed commercial and transportation residential areas, cultural areas, industrial areas, and rural areas. The primary quality standard applies to Type I areas, and the secondary quality standard applies to Type II areas. Since the scope of the study is the entire city, only the days exceeding the national secondary quality standard were analyzed.

Figure 2 shows the trends of the original series, short-term components, seasonal components, and long-term components of NO₂, O₃, PM₁₀, and PM_2.5 in the Fenwei Plain from 2018 to 2022. From the original series and the exceeding days, there were only 9 days when the daily average NO₂ concentration in the Fenwei Plain exceeded the standard (80 μg·m⁻³), and all of these were in 2018. It can also be seen that the peak of daily average NO₂ concentration has been significantly reduced. It shows that with the implementation of the Three-Year Action Plan, the NO₂ pollution in the Fenwei Plain has been well managed and shows a continuous downward trend. The days when the O₃MAD8 exceeded the national secondary quality standard (160 μg·m⁻³) in each of the years from 2018 to 2022 were 58, 45, 31, 45, and 37 d, respectively. From 2018 to 2020, most of the exceeding months were concentrated between April and August, and from 2021 to 2022, most of the exceeding months were concentrated between May and September. Thus, it can be considered that the O₃ control and pollution period in the Fenwei Plain is from April to September. O₃ levels rebounded in 2021–2022, presumably due to the declining NO₂. It shows that simply reducing the concentration of NO₂ may not have the effect of reducing the concentration of O₃. As can be seen from Figure 2, the number of exceedance days of particulate matter is still severe, with the annual number of days in which the daily average concentration of PM₁₀ exceeded the national secondary quality standard (150 μg·m⁻³) between 2018 and 2022 being 72, 62, 30, 41, 30 d, respectively. The annual number of days in which PM_2.5 exceeded the national secondary quality standard (75 μg·m⁻³) for the same period was 84, 80, 72, 54, 63 d, respectively. It indicates that the exceedance of particulate matter is still a problem that needs to be emphasized in the Fenwei Plain.

The long-term components of three air pollutants, but not O₃, showed a decreasing trend. The long-term components of NO₂, PM₁₀, and PM_2.5 decreased by 10.5, 33.1, and 17.1 μg·m⁻³, respectively. The decrease rate was 25.8%, 29%, and 28.8%, respectively. In contrast, O₃ levels were shown to decrease and then increase, with 2020 as the dividing line. The long-term components decreased from 112.7 μg·m⁻³ to 98.1 μg·m⁻³ and then increased continuously to level off at 108 μg·m⁻³. PM₁₀ fluctuated slightly in 2021, which might be related to the frequent sand and dust activities in 2021 [26]. In terms of the short-term component, this exhibited significant amplitude, with large fluctuations especially at higher concentrations of the individual pollutants. The drastic changes in the short-term component can be seen as caused by short-term changes in the weather, since the local emission sources do not show significant changes in the short term. Among them, the effect of the dust storm in March 2021 is extremely obvious in the short-term component of PM₁₀. The seasonal components of all four air pollutants show significant cyclical variations. Such a cyclical fluctuation in the seasonal fractions is mainly due to a heating period in winter in the north, with low boundary-layer heights and static weather, as well as frequent dusty weather in spring, which contribute to the accumulation of pollutants. In summer, frequent precipitation, high temperatures, strong air convection, and high boundary-layer thickness contribute to the diffusion and deposition of pollutants [27].

5.2. Relative Contribution of Long-Term, Short-Term, Seasonal, and Unknown Components of the Pollutants

In order to analyze the contribution of each component to the original time series of pollutant concentrations, the relative contributions of the short-term, seasonal, and long-term components to the total variance were calculated. The relationship between the three theoretical short-term, seasonal, and long-term components should be independent, and the sum of the short-term, seasonal, and long-term components is equal to the variance of the original time series [28]. The degree of contribution of the above components to the variance of each city in the Fenwei Plain is shown in Figure 3. It can be seen that the sum of the long-term, seasonal, and short-term components of the four pollutants in different cities in the Fenwei Plain contributes to the total variance in the range of 90.25–95.98%. The larger the total variance is, the stronger the independence among the three components is, indicating that the KZ filter has a better serial decomposition ability with regard to the four pollutants in different cities of the Fenwei Plain.

The short-term component of NO₂ contributed the most to the total variance, reaching 34.7–69.8%, followed by the seasonal component, reaching 19.3–47.3%, and the long-term component contributed the least to the total variance of the original time series, ranging from 2.1% to 12.8%. The short-term component had the most considerable contribution in Lvliang, which was significantly higher than in other cities, indicating that NO₂ in Lvliang was more susceptible to short-term meteorological conditions. The dominant wind direction near the ground in Lvliang City is northeast, accounting for 33.9% of the wind frequency, and the second dominant wind direction is southwest, accounting for 12.9% of the wind frequency. These two wind directions are the same as the direction of the Lvliang Sanchuan River valley, with the northeast wind corresponding to the Beichuan River and the southwest wind corresponding to the main stream of the Sanchuan River [29]. This may be why Lvliang is the city that is the most vulnerable to short-term changes. The O₃ seasonal component contributes the most to the total variance, with 54–70.7%, followed by the short-term component, with 24.8–40%, and the long-term component contributes the least to the total variance of the original time series, with 0.27–1.74%. The seasonal component is larger than the short-term component for all cities in the Fenwei Plain, suggesting that O₃ is mainly related to seasonal variations in meteorological conditions and energy consumption patterns. The most considerable contribution of the seasonal component of O₃ to the total variance is related to its genesis, as solar irradiation can primarily affect the concentration of O₃ itself [30]. At the same time, the stratospheric–tropospheric exchange triggered by tropopause folding is an essential source of the tropospheric O₃ increase in spring [31]. Therefore, the seasonal variation in O₃ is more prominent. The short-term component of PM₁₀ contributes the most to the total variance, with 53–73%, followed by the seasonal component, with 19.1–37.3%, and the long-term component contributes the least to the total variance of the original time series, with 0.75–4.4%. The short-term component is larger than the seasonal component for all cities in the Fenwei Plain, which is the opposite to that of O₃, indicating that PM₁₀ is mainly related to short-term weather changes and short-term fluctuations of local emissions. Among them, Lvliang is the most susceptible city to short-term variations, and Xi’an is the most susceptible city to seasonal variations. In Xi’an, temperature and humidity are the vital meteorological factors causing O₃ exceedance. High temperature and low humidity are favorable to O₃ generation, and it is easy for a high concentration of O₃ to occur when the temperature is higher than 30 °C and the relative humidity is lower than 60% [32]. This may be why Xi’an is the most vulnerable city to seasonal changes. The short-term component of PM_2.5 contributes the most to the total variance, with 43–58%, followed by the seasonal component, with 28.9–49.2%, and the long-term component contributes the least to the total variance of the original time series, with 0.95–13.1%. PM_2.5 is similar to PM₁₀. PM_2.5 is relatively more susceptible to socioeconomic conditions and policies and less susceptible to short-term meteorological factors.

Overall, the seasonal and short-term components strongly influenced the changes in the original time series of the four air pollutants in the Fenwei Plain. It suggests that the changes in air pollutant concentrations are caused by seasonal and short-term fluctuations in pollutant precursor emissions and meteorological conditions [33]. In contrast, the long-term changes in pollutant emissions and meteorological conditions are not the leading causes of the fluctuations in air pollutants in the Fenwei Plain. Therefore, to study the long-term trend changes in pollutants, it is necessary to separate the long-term components from the original time series [34].

5.3. Trend of NO₂ Concentration Change, Meteorological Effect, and Anthropogenic Discharge Contribution

Figure 4 shows the long-term components of the original concentrations in 11 cities in the Fenwei Plain from 2018 to 2022. The 11 graphs on the left show the trend changes in the actual long-term fractions of each pollutant by city, and the 11 graphs on the right show the trend changes in the long-term fractions of each pollutant adjusted for meteorology to exclude the meteorological factor and, thus, affected only by emissions. Combining this with the results of the long-term components in Figure 2, it can be seen that the long-term components of NO₂ in the 11 cities of the Fenwei Plain keep the same trend before and after the adjustment, and all show a decreasing trend. In order to visualize the effect of meteorological conditions on NO₂ concentration during the study period, we examined the difference between the long-term components and the meteorologically adjusted long-term components to obtain the effect of meteorological conditions on the long-term components. A positive difference indicates that the long-term component due to the pollution source is smaller than the long-term component of the combined effect of the pollution source and meteorology, indicating that the meteorological conditions are not conducive to improving pollution. Conversely, a negative difference indicates that the long-term component due to the source of pollution is larger than the long-term component of the combined effect of the source of pollution and meteorology, suggesting that the meteorological conditions are favorable to improving pollution. Most of the cities in the Fenwei Plain had unfavorable meteorological conditions for NO₂ improvement in 2018–2020 and favorable meteorological conditions for NO₂ improvement in 2020–2022, with Xianyang City showing the most prominent performance. Regarding the fluctuating magnitude of meteorological impacts, Linfen, Yuncheng, Xianyang, and Sanmenxia had more significant meteorological impacts. Similar conclusions emerged from Chen in a study of particulate matter in the Yangtze River Delta, i.e., with 2020 as the dividing line, meteorological conditions were favorable and then unfavorable for improving particulate matter. The specific reasons for this pattern need to be further investigated [10].

Table 1. Estimated relative impact of emission reduction and meteorological variations on NO₂ reduction in the Fenwei Plain from 2018 to 2022.

City	NO2 in 2018	NO2 in 2022	NO2 Decrease Rate	Adjusted NO2 in 2018	Adjusted NO2 in 2022	Adjusted NO2 Decrease Rate	Relative Impact of Meteorological Variations (%)	Relative Impact of Emission Reduction (%)	Long-Term Decline Rate of NO2 Concentration
LL	43.43	38.79	4.64	43.46	38.58	4.88	−5.15%	105.15%	10.68%
JZ	41.39	31.78	9.61	42.30	32.19	10.11	−5.23%	105.23%	23.22%
LF	38.26	31.57	6.70	38.53	32.21	6.32	5.65%	94.35%	17.51%
YC	28.44	20.58	7.86	29.47	22.14	7.32	6.83%	93.17%	27.64%
BJ	38.93	27.47	11.46	38.57	28.31	10.25	10.54%	89.46%	29.44%
TC	33.89	25.31	8.58	33.56	27.19	6.37	25.72%	74.28%	25.32%
XY	47.39	35.47	11.92	45.96	38.04	7.92	33.56%	66.44%	25.15%
WN	46.95	35.08	11.88	46.89	36.12	10.77	9.34%	90.66%	25.30%
SMX	36.41	24.07	12.34	36.45	25.87	10.57	14.29%	85.71%	33.89%
LY	40.08	25.24	14.84	40.73	26.40	14.33	3.41%	96.59%	37.03%
XA	52.40	37.27	15.13	52.23	39.22	13.02	13.99%	86.01%	28.87%

shows the ratios of long-term component changes in the NO₂ concentration and the specific contribution ratios of emission reduction and meteorological conditions to the trend changes in concentration for the 11 cities in the Fenwei Plain. All 11 cities in the Fenwei Plain showed a decreasing trend from 2018 to 2022, but each city’s degree of decrease was different. Among them, Luoyang City had the most significant decreasing ratio of 37.0%, and Lvliang City had the smallest decreasing ratio of 10.7%. The overall trend was higher in the eastern Fenwei Plain than in the western Fenwei Plain and higher in the southern Fenwei Plain than in the northern Fenwei Plain. In another study, Zhang pointed out that the Guanzhong region and northern Shaanxi are more polluted, while southern Shaanxi and southern Shaanxi are less polluted, and the cities are more polluted than the suburban counties [35]. This is consistent with our finding that pollution is higher in the north than in the south, and the reason for this trend may be related to the local energy structure and policies. Since the release of the “Ten Atmospheric Rules,” the Fenwei Plain has implemented strict air pollution control measures, such as controlling coal consumption, adjusting the industrial structure, reducing vehicle fuels, and phasing out small coal-fired boilers. These actively implemented clean-air policies have significantly improved the air quality [36]. The effect of meteorological conditions and the implementation of air pollution control measures varied among the cities in the Fenwei Plain, with the contribution of emission sources to the long-term component changes ranging from 66.4% to 105.2%. Among them, the meteorological conditions in Lvliang and Jinzhong cities had an unfavorable effect on NO₂. Among the other nine cities, Xianyang City had the most significant proportion of meteorological conditions with 33.6%, and Luoyang City had the most significant contribution to the proportion of emission reduction measures with 96.59%. The dominant factor in the year-by-year improvement in NO₂ levels in the Fenwei Plain from 2018 to 2022 is anthropogenic emission reduction, in which the role of meteorological conditions has less influence.

5.4. Trend of O₃ Concentration Change, Meteorological Effect, and Anthropogenic Discharge Contribution

The long-term trend of the O₃ concentration before and after meteorological adjustment is shown in Figure 5. It can be seen that the long-term components of all cities in the Fenwei Plain generally decreased after removing the influence of meteorological factors, indicating that meteorological conditions are unfavorable to the improvement of O₃. All cities in the Fenwei Plain showed the same characteristics, i.e., the meteorological conditions were unfavorable to the improvement in O₃ levels in 2018–2019, favorable to the improvement of O₃ levels in 2019–2021, and unfavorable to the improvement of O₃ levels in 2021–2022. Regarding the fluctuation in the magnitude of meteorological impacts, the meteorological impacts were more considerable in nine cities, with the exception of Sanmenxia and Luoyang. The more significant unfavorable effects of meteorological impacts in all cities since 2022 may be related to the frequent and extreme high-temperature weather in the last two years. The O₃ concentrations were related to the precursor emission sources and their location relationships, geographic features, and meteorological conditions. Scholars have investigated the effects of meteorological parameters such as solar radiation, temperature, relative humidity, wind speed, and direction on O₃ concentrations, and the effects of other meteorological parameters varied depending on the location. In contrast, solar radiation and temperature were recognized as the primary influence parameters of O₃ [37].

Table 2. Estimated relative impact of emission reduction and meteorological variations on O₃ reduction in the Fenwei Plain from 2018 to 2022.

City	O3 in 2018	O3 in 2022	O3 Decrease Rate	Adjusted O3 in 2018	Adjusted O3 in 2022	Adjusted O3 Decrease Rate	Relative Impact of Meteorological Variations (%)	Relative Impact of Emission Reduction (%)	Long-Term Decline Rate of O3 Concentration
LL	101.92	92.78	9.14	98.30	90.22	8.08	11.57%	88.43%	8.97%
JZ	115.69	113.72	1.97	110.24	111.61	−1.37	169.49%	−69.49%	1.70%
LF	128.90	113.96	14.94	123.77	109.32	14.45	3.29%	96.71%	11.59%
YC	128.29	117.69	10.61	122.37	113.97	8.40	20.78%	79.22%	8.27%
BJ	99.42	98.11	1.32	94.92	94.29	0.64	51.66%	48.34%	1.33%
TC	114.51	107.46	7.05	109.75	102.28	7.47	−5.93%	105.93%	6.16%
XY	113.31	107.23	6.07	107.99	100.27	7.72	−27.09%	127.09%	5.36%
WN	104.02	107.87	−3.85	102.66	102.30	0.36	109.36%	−9.36%	−3.70%
SMX	113.26	111.30	1.96	113.36	105.22	8.13	−314.22%	414.22%	1.73%
LY	115.49	113.44	2.05	114.86	107.81	7.05	−243.54%	343.54%	1.78%
XA	104.91	104.42	0.49	99.19	99.20	−0.01	102.81%	−2.81%	0.47%

shows the ratios of long-term component changes in O₃ concentrations in 11 cities in the Fenwei Plain and the specific contribution ratios of emission reduction and meteorological conditions to the trend changes in concentrations. It can be seen that, except for Weinan, all other cities showed a decreasing trend in O₃ concentration in 2018–2022. Linfen showed the most significant decrease of 11.6%. Xi’an had the most minor decrease of 0.47%. The effect of meteorological conditions and the implementation of air pollution control measures varied among the cities in the Fenwei Plain. Meteorological conditions in Sanmenxia, Luoyang, Xianyang, and Tongchuan unfavorably affected O₃. The anthropogenic emissions in Jinzhong, Weinan, and Xi’an unfavorably affected O₃. Among the other four cities, Baoji had the most significant percentage of contribution from meteorological conditions with 51.7%, and Linfen had the most significant contribution from emission reduction measures with 96.7%. O₃ is a secondary pollutant. Its concentration changes depending on the precursor emissions, with the system of the atmospheric environment of the integrated impact, spatial and temporal characteristics, and impact factors making things more complex [38]. Therefore, different measures should be taken to reduce O₃ emissions in different cities.

5.5. Trend of PM₁₀ Concentration Change, Meteorological Effect, and Anthropogenic Discharge Contribution

Figure 6 demonstrates the difference between the emission-only and regulated long-term fractions of PM₁₀ for the 11 cities in the Fenwei Plain after adjusting for meteorological conditions from 2018 to 2022. Figure 6 shows that the long-term fractions of each city in the Fenwei Plain generally increased after removing the influence of meteorological factors, indicating that meteorological conditions are favorable to improving PM₁₀. Except for Sanmenxia, Luoyang, and Yuncheng, the effects of meteorological factors on PM₁₀ showed a two-stage decreasing pattern, i.e., the meteorological conditions were unfavorable to PM₁₀ improvement in 2018–2020, the role of meteorological conditions was almost zero in 2020–2021, and the meteorological conditions were favorable to PM₁₀ improvement in 2021–2022.

Table 3. Estimated relative impact of emission reduction and meteorological variations on PM₁₀ reduction in the Fenwei Plain from 2018 to 2022.

City	PM10 in 2018	PM10 in 2022	PM10 Decrease Rate	Adjusted PM10 in 2018	Adjusted PM10 in 2022	Adjusted PM10 Decrease Rate	Relative Impact of Meteorological Variations (%)	Relative Impact of Emission Reduction (%)	Long-Term Decline Rate of PM10 Concentration
LL	102.34	83.96	18.38	99.04	90.19	8.85	51.84%	48.16%	17.96%
JZ	112.94	80.41	32.53	109.46	85.30	24.16	25.75%	74.25%	28.80%
LF	122.10	73.78	48.33	118.09	79.37	38.72	19.89%	80.11%	39.58%
YC	112.18	85.30	26.88	111.47	92.13	19.33	28.08%	71.92%	23.96%
BJ	102.51	69.97	32.55	96.97	77.04	19.93	38.77%	61.23%	31.75%
TC	95.95	70.80	25.15	92.60	75.23	17.37	30.94%	69.06%	26.21%
XY	132.19	96.52	35.67	126.35	104.27	22.08	38.09%	61.91%	26.98%
WN	131.06	90.29	40.77	127.74	97.45	30.29	25.70%	74.30%	31.11%
SMX	111.77	74.93	36.84	113.96	78.25	35.71	3.07%	96.93%	32.96%
LY	115.27	78.74	36.53	115.37	83.73	31.64	13.38%	86.62%	31.69%
XA	117.77	87.80	29.97	112.17	96.58	15.59	47.98%	52.02%	25.45%

shows the ratios of long-term component changes in PM₁₀ concentrations in 11 cities in the Fenwei Plain and the specific contribution ratios of emission reductions and meteorological conditions to the changes in concentration trends. Compared with 2018, PM₁₀ pollution in all 11 cities in the Fenwei Plain significantly improved in 2022. The Linfen PM₁₀ decreased the most, by 39.58%, and Lvliang PM₁₀ decreased the least, by 17.96%. Pollution-source emission reduction measures are still the main factors for improving PM₁₀ pollution in the Fenwei Plain; except for Lvliang and Xi’an, the proportion of local emission reduction is more than 60% in the rest of the cities. Sanmenxia, Luoyang, and Linfen in the eastern part of the Fenwei Plain accounted for 96.93%, 86.62%, and 80.11%, respectively.

5.6. Trend of PM_2.5 Concentration Change, Meteorological Effect, and Anthropogenic Discharge Contribution

Figure 7 demonstrates the difference between the emission-only and moderated long-term fractions of PM_2.5 for the 11 cities in the Fenwei Plain after adjusting for meteorological conditions from 2018 to 2022. From Figure 7, it can be seen that the long-term fractions of each city in the Fenwei Plain generally increased after removing the influence of meteorological factors, indicating that meteorological conditions are favorable to improving PM_2.5. It can be seen that Lvliang and Jinzhong are less affected by meteorological effects. In terms of the time of the influence of meteorological conditions, most cities in the Fenwei Plain, except Yuncheng and Sanmenxia, showed that meteorological conditions were unfavorable to the improvement of PM_2.5 in 2018–2020, and meteorological conditions were favorable to the improvement of PM_2.5 in 2021–2022.

Table 4. Estimated relative impact of emission reduction and meteorological variations on PM_2.5 reduction in the Fenwei Plain from 2018 to 2022.

City	PM2.5 in 2018	PM2.5 in 2022	PM2.5 Decrease Rate	Adjusted PM2.5 in 2018	Adjusted PM2.5 in 2022	Adjusted PM2.5 Decrease Rate	Relative Impact of Meteorological Variations (%)	Relative Impact of Emission Reduction (%)	Long-Term Decline Rate of PM2.5 Concentration
LL	52.58	23.21	29.37	51.43	24.98	26.45	9.92%	90.08%	55.86%
JZ	54.41	42.11	12.30	54.99	42.07	12.92	−5.05%	105.05%	22.61%
LF	70.37	45.31	25.07	69.55	48.27	21.27	15.12%	84.88%	35.63%
YC	61.33	47.24	14.09	64.74	50.28	14.46	−2.66%	102.66%	22.97%
BJ	52.80	41.23	11.56	50.79	46.80	3.98	65.56%	34.44%	21.89%
TC	49.64	34.95	14.69	49.33	39.82	9.51	35.22%	64.78%	29.59%
XY	68.99	50.17	18.82	67.00	59.07	7.93	57.85%	42.15%	27.28%
WN	59.65	48.86	10.79	57.98	54.18	3.80	64.75%	35.25%	18.09%
SMX	59.06	42.18	16.88	61.46	43.95	17.51	−3.77%	103.77%	28.58%
LY	61.35	42.82	18.52	61.19	45.59	15.59	15.83%	84.17%	30.19%
XA	61.62	45.72	15.90	60.20	52.42	7.78	51.08%	48.92%	25.80%

shows the ratios of long-term component changes in PM_2.5 concentrations and the specific contribution ratios of emission reduction and meteorological conditions to concentration trends for the 11 cities in the Fenwei Plain. Compared with 2018, PM_2.5 pollution in all 11 cities in the Fenwei Plain improved significantly in 2022. Among them, the Lvliang PM_2.5 decreased by the most considerable amount, by 55.86%, and Weinan PM_2.5 decreased by the smallest amount, by 18.09%. In Weinan, Baoji, Xianyang, Xi’an, and other cities in the western Fenwei Plain, meteorological factors play a significant role in PM_2.5 emission reduction, accounting for more than half of the overall emission reduction factors. However, emission reduction measures of pollution sources are still the main factors for improving PM_2.5 pollution in the Fenwei Plain, where the meteorological conditions in Jinzhong, Yuncheng, and Sanmenxia negatively affected PM_2.5 emission reduction. Local emission reduction in Lvliang, Yuncheng, and Luoyang played a significant role in PM_2.5 emission reduction, accounting for more than 80% of the total.

6. Conclusions

(1) The long-term fractions of NO₂, PM₁₀, and PM_2.5 in the Fenwei Plain decreased by 10.5, 33.1, and 17.1 μg·m⁻³ from 2018 to 2022, with decreases of 25.8%, 29%, and 28.8%, respectively. Among them, PM₁₀ showed the most apparent decreasing trend. The long-term fraction of O₃ showed a decreasing and then increasing trend, with 2020 as the dividing line, with the long-term fraction decreasing from 112.7 μg·m⁻³ to 98.1 μg·m⁻³ and then increasing to 108 μg·m⁻³ and leveling off.

(2) The short-term components of NO₂, PM₁₀, and PM_2.5 contributed the most to the total variance, with the short-term components accounting for 34.7–69.8%, 53–73%, and 43–58%, respectively. The seasonal components of O₃ contributed the most to the total variance, with 54–70.7%. This indicates that small-scale weather systems and short-term changes in external emissions mainly influence NO₂, PM₁₀, and PM_2.5. Changes in O₃ concentration is mainly related to seasonal changes in meteorological conditions and energy consumption patterns.

(3) Most cities in the Fenwei Plain had unfavorable meteorological conditions for NO₂, PM₁₀, and PM_2.5 in 2018–2020 and favorable meteorological conditions for NO₂, PM₁₀, and PM_2.5 in 2020–2022. Regarding the magnitude of meteorological effects, PM₁₀ > PM_2.5 > NO₂. O₃ showed different characteristics from the other three pollutants, with most cities in the Fenwei Plain experiencing unfavorable meteorological conditions for the improvement of O₃ in 2018–2019, favorable meteorological conditions for the improvement of O₃ in 2019–2021, and unfavorable meteorological conditions for the improvement of O₃ in 2021–2022.

(4) In the long-term component changes in NO₂ concentration, Luoyang city had the most significant decrease ratio of 37.0%, and Lvliang city had the smallest decrease ratio of 10.7%. Among them, the meteorological conditions in Lvliang and Jinzhong cities adversely affected NO₂ concentrations. Among the other nine cities, Xianyang City had the most significant contribution ratio of 33.6% from meteorological conditions, and Luoyang City had the most significant contribution ratio of 96.59% from abatement measures.

(5) Long-term fractional changes in O₃ concentrations. Except for Weinan, all the other cities showed a decreasing trend for O₃ in 2018–2022, with Linfen having the most significant decrease of 11.6%. Xi’an had the most minor decrease of 0.47%. Meteorological conditions in Sanmenxia, Luoyang, Xianyang, and Tongchuan adversely affected O₃ concentrations. Anthropogenic emissions in Jinzhong, Weinan, and Xi’an adversely affected O₃ concentrations. Among the other four cities, Baoji had the most significant percentage of contribution from meteorological conditions with 51.7%, and Linfen had the most significant contribution from emission reduction measures with 96.7%.

(6) The long-term component change in PM₁₀ concentration was 39.58% in Linfen and 17.96% in Lvliang. Pollution-source abatement measures are still the main factor for improving PM₁₀ pollution in the Fenwei Plain, except for Lvliang and Xi’an, where the local abatement accounted for more than 60% of the rest of the cities. Among them, Sanmenxia, Luoyang, Linfen, and other cities in the eastern Fenwei Plain had a high proportion of local emission reduction of 96.93%, 86.62%, and 80.11%, respectively.

(7) The long-term component change in PM_2.5 concentration in Lvliang decreased by the most considerable amount, 55.86%, and the decrease in PM_2.5 in Weinan was the smallest amount, 18.09%. In Weinan, Baoji, Xianyang, Xi’an, and other cities in the western Fenwei Plain, meteorological factors played a prominent role in PM_2.5 emission reduction, accounting for more than half of the overall emission reduction factors. However, emission reduction measures toward pollution sources are still the main factors for improving PM_2.5 pollution in the Fenwei Plain, where meteorological conditions in Jinzhong, Yuncheng, and Sanmenxia negatively affected PM_2.5 emission reduction. Local emission reduction in Lvliang, Yuncheng, and Luoyang played a significant role in PM_2.5 emission reduction, accounting for more than 80% of the total.