Fine particulate matter (PM_2.5) trends in China, 2013–2018: separating contributions from anthropogenic emissions and meteorology

Fine particulate matter (PM_2.5) is a severe air pollution problem in China. Observations of PM_2.5 have been available since 2013 from a large network operated by the China National Environmental Monitoring Center (CNEMC). The data show a general 30 %–50 % decrease in annual mean PM_2.5 across China over the 2013–2018 period, averaging at −5.2 µg m⁻³ a⁻¹. Trends in the five megacity cluster regions targeted by the government for air quality control are $-\mathrm{9.3}\pm \mathrm{1.8}$ µg m⁻³ a⁻¹ (±95 % confidence interval) for Beijing–Tianjin–Hebei, $-\mathrm{6.1}\pm \mathrm{1.1}$ µg m⁻³ a⁻¹ for the Yangtze River Delta, $-\mathrm{2.7}\pm \mathrm{0.8}$ µg m⁻³ a⁻¹ for the Pearl River Delta, $-\mathrm{6.7}\pm \mathrm{1.3}$ µg m⁻³ a⁻¹ for the Sichuan Basin, and $-\mathrm{6.5}\pm \mathrm{2.5}$ µg m⁻³ a⁻¹ for the Fenwei Plain (Xi'an). Concurrent 2013–2018 observations of sulfur dioxide (SO₂) and carbon monoxide (CO) show that the declines in PM_2.5 are qualitatively consistent with drastic controls of emissions from coal combustion. However, there is also a large meteorologically driven interannual variability in PM_2.5 that complicates trend attribution. We used a stepwise multiple linear regression (MLR) model to quantify this meteorological contribution to the PM_2.5 trends across China. The MLR model correlates the 10 d PM_2.5 anomalies to wind speed, precipitation, relative humidity, temperature, and 850 hPa meridional wind velocity (V850). The meteorology-corrected PM_2.5 trends after removal of the MLR meteorological contribution can be viewed as being driven by trends in anthropogenic emissions. The mean PM_2.5 decrease across China is −4.6 µg m⁻³ a⁻¹ in the meteorology-corrected data, 12 % weaker than in the original data, meaning that 12 % of the PM_2.5 decrease in the original data is attributable to meteorology. The trends in the meteorology-corrected data for the five megacity clusters are $-\mathrm{8.0}\pm \mathrm{1.1}$ µg m⁻³ a⁻¹ for Beijing–Tianjin–Hebei (14 % weaker than in the original data), $-\mathrm{6.3}\pm \mathrm{0.9}$ µg m⁻³ a⁻¹ for the Yangtze River Delta (3 % stronger), $-\mathrm{2.2}\pm \mathrm{0.5}$ µg m⁻³ a⁻¹ for the Pearl River Delta (19 % weaker), $-\mathrm{4.9}\pm \mathrm{0.9}$ µg m⁻³ a⁻¹ for the Sichuan Basin (27 % weaker), and $-\mathrm{5.0}\pm \mathrm{1.9}$ µg m⁻³ a⁻¹ for the Fenwei Plain (Xi'an; 23 % weaker); 2015–2017 observations of flattening PM_2.5 in the Pearl River Delta and increases in the Fenwei Plain can be attributed to meteorology rather than to relaxation of emission controls.

1 Introduction

PM_2.5 (particulate matter with aerodynamic diameter less than 2.5 µm) is a severe air pollution problem in China, responsible for 1.1 million excess deaths in 2015 (Cohen et al., 2017). The Chinese government introduced, in 2013, the “Action Plan on the Prevention and Control of Air Pollution” (Chinese State Council, 2013a), called Clean Air Action for short, to aggressively control anthropogenic emissions. Starting that year, PM_2.5 data from a nationwide monitoring network of about 1000 sites became available from the China National Environmental Monitoring Center (CNEMC) of the Ministry of Ecology and Environment of China (MEEC). These data show 30 %–40 % decreases in PM_2.5 across eastern China over the 2013–2017 period (Chinese State Council, 2018a; Zhang et al., 2019). However, interpretation of these trends in terms of emission controls may be biased by interannual variability and trends in meteorology (Zhang et al., 2014; Wang et al., 2014; Zhu et al., 2012; Jia et al., 2015; Li et al., 2018; Yang et al., 2018, 2016; Liang et al., 2016; Cheng et al., 2019; Chen et al., 2019; Silver et al., 2018). Here we use a stepwise multilinear regression (MLR) model to separate the effects of meteorological variability and emission controls on the 2013–2018 trends in PM_2.5 across China.

Meteorology drives large day-to-day, seasonal, and interannual variations in PM_2.5 in China by affecting transport, scavenging, emissions, and chemical production (Wang et al., 2014; Leung et al., 2018; Tai et al., 2012; Zou et al., 2017). The relationships between PM_2.5 and meteorological variables are complex and differ by region and time of year (Shen et al., 2017). For example, wintertime PM_2.5 pollution events in central and eastern China are associated with low wind speed and high relative humidity (RH; Wang et al., 2014; Zhang et al., 2014; Shen et al., 2018; Pendergrass et al., 2019; Moch et al., 2018; Song et al., 2019). On the other hand, high wind speeds in northern China in spring and summer promote dust emission (Lyu et al., 2017; Wang et al., 2004). Precipitation scavenging is a major factor driving PM_2.5 variability in southern and coastal China (Chen et al., 2018; Leung et al., 2018).

Anthropogenic emissions of PM_2.5 and its precursors, including sulfur dioxide (SO₂), nitrogen oxides (NO_x), ammonia (NH₃), and nonmethane volatile organic compounds (NMVOCs), have undergone large changes in China over the past decades. Rapid growth in emissions from 1980 to 2006 led to a general increase in PM_2.5 over China, as demonstrated by visibility data (Che et al., 2007; Han et al., 2016; Wang and Chen, 2016; Fu et al., 2014; Zhang et al., 2012) and, since 1999, by satellite aerosol optical depth (AOD) data (Ma et al., 2016; Lin et al., 2018; Zhao et al., 2017). SO₂ emissions peaked in 2006–2007, NO_x emissions peaked in 2011, and NH₃ emissions peaked around 1996, as estimated from emission inventories (Zhao et al., 2017; Wang et al., 2017; Xia et al., 2016; F. Liu et al., 2016; Lu et al., 2010; Xu et al., 2016; Kang et al., 2016) and observed from satellites (Xia et al., 2016; F. Liu et al., 2016; de Foy et al., 2016; van der A et al., 2017). SO₂ and NO_x emissions have declined since their peaks, whereas emissions of NH₃ have remained relatively stable since its peak (Zhao et al., 2017). The onset of emission controls led to slight decreases in PM_2.5 over the 2006–2012 period, as indicated by satellite AOD data (Ma et al., 2016, 2019; Lin et al., 2018; Zhao et al., 2017) and surface observations (Tao et al., 2017; Wang et al., 2017). The Clean Air Action greatly increased the scope of emission controls. The Multi-resolution Emission Inventory for China (MEIC; http://www.meicmodel.org, last access: 20 March 2019) estimates nationwide emission decreases over the 2013–2017 period of 59 % for SO₂, 33 % for primary PM_2.5, 21 % for NO_x, and 3 % for NH₃, with NMVOCs increasing by 2 % (Zheng et al., 2018). Continued reductions in emissions are required and implemented in 2018 (Chinese State Council, 2018b). Our goal in this work is to quantify the response of PM_2.5 to these rapid emission changes by resolving the effect of meteorological variability, thus allowing improved assessment of the success of the Clean Air Action.

2 Data and methods

2.1 Observations

We use 2013–2018 hourly data for surface air PM_2.5 together with SO₂, nitrogen dioxide (NO₂), and CO concentrations from the CNEMC network (http://106.37.208.233:20035/, last access: 18 March 2019). The network started in January 2013, with 496 sites in 74 major cities across the country (Chinese State Council, 2013b), growing to ∼1500 sites in 454 cities by 2018. PM_2.5 mass concentrations are measured using the micro-oscillating balance method and/or the β-absorption method (MEE, 2012; Zhang and Cao, 2015). SO₂, NO₂, and CO concentrations are measured at the same sites as PM_2.5. NO₂ concentrations are measured by the molybdenum converter method known to have positive interferences from NO₂ oxidation products (Dunlea et al., 2007). SO₂ and CO are respectively measured using ultraviolet fluorescence and infrared absorption (MEE, 2012; Zhang and Cao, 2015). We applied quality control to the hourly CNEMC data following Barrero et al. (2015) to exclude severe outliers (Lu et al., 2018). There are also occasional consecutive repeats of data that may be caused by faulty instruments or reporting (Rohde and Muller, 2015; Silver et al., 2018). Here we removed values from the hourly time series when there are >24 consecutive repeats. These in whole removed 7.4 %, 7.0 %, 6.4 %, and 6.7 % of the PM_2.5, SO₂, NO₂, and CO data respectively.

We correlated these air quality observations with meteorological observations from 839 stations distributed across China (Fig. S1 in the Supplement). The meteorological observations are compiled in the Surface Daily Climate Dataset (V3.0) released by the China National Meteorological Information Center (CNMIC; http://data.cma.cn/, last access: 16 March 2019). These include data for wind speed (WDS), precipitation (PRECIP), relative humidity (RH), and temperature (TEM). We also used the 850 hPa meridional wind velocity (V850) from the MERRA-2 reanalysis produced at 0.5^∘ × 0.625^∘ horizontal resolution by the NASA Global Modeling and Assimilation Office (https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2, last access: 20 March 2019). We choose these meteorological variables for their strong correlations with PM_2.5 identified in previous studies (Wang et al., 2014; Cai et al., 2017; Shen et al., 2017; Leung et al., 2018; Song et al., 2019; Zou et al., 2017). V850 in particular is a strong predictor of PM_2.5 wintertime pollution events in the North China Plain because northerly winds (negative V850) ventilate the region with clean dry air (Cai et al., 2017; Pendergrass et al., 2019).

All data in this work are averaged over 10 d (10 d time resolution). Trend analyses use only those sites with at least 70 % data coverage for each of the 6 years from 2013 to 2018. We did sensitivity tests with data coverage thresholds changing from 70 % to 90 % and obtained similar pollutant trends. To make the most use of available data, 70 % is chosen. For the MLR model, we further average all data on a 2^∘ × 2.5^∘ grid to increase statistical robustness, following Tai et al. (2012) and Shen et al. (2017).

The 2013 Clean Air Action (Chinese State Council, 2013a) identified three megacity clusters as target regions for reducing air pollution: Beijing–Tianjin–Hebei (BTH; 35–41^∘ N, 113.75–118.75^∘ E), the Yangtze River Delta (YRD; 29–33^∘ N, 118.75–123^∘ E), and the Pearl River Delta (PRD; 21–25^∘ N, 111.25–116.25^∘ E). The more recent plan released in July 2018 (Chinese State Council, 2018b) removed PRD from the list of target regions and added the Fenwei Plain (FWP; 33–35^∘ N, 106.25–111.25^∘ E; 35–37^∘ N, 108.75–113.75^∘ E). Previous studies (Zhang et al., 2012) also identified the Sichuan Basin (SCB; 27–33^∘ N, 103.75–108.75^∘ E) as one of the major haze regions in China. We present analyses for these five target regions by averaging the data from all sites with more than 70 % data coverage for each of the 6 years from 2013 to 2018. The only continuous record for 2013–2018 in the FWP region is for Xi'an (13 sites). Additional FWP sites outside Xi'an started operating in early 2015 and are consistent with the Xi'an data, as will be shown below.

2.2 Multiple linear regression model

We construct a stepwise multiple linear regression (MLR) model to quantify the effect of meteorology on PM_2.5 variability. The model fits the deseasonalized and detrended 10 d PM_2.5 mean time series on the 2^∘ × 2.5^∘ grid to the five deseasonalized and detrended 10 d mean meteorological variables (WDS, PRECIP, RH, TEM, and V850). The deseasonalized and detrended time series are obtained by removing the 50 d moving averages from the 10 d mean time series (Tai et al., 2010). This focuses on synoptic scales of variability and avoids aliasing from common seasonal variations and long-term trends between variables (Shen et al., 2017).

Separate fits of PM_2.5 to the meteorological variables are done for each 2^∘× 2.5^∘ grid square and season (DJF, MAM, JJA, and SON). The fit has the form

$$\begin{array}{}\text{(1)}& Y_{\mathrm{d},i}\left(t\right)=\sum _{k=\mathrm{1}}^{\mathrm{5}}{\mathit{\beta }}_{i,k}{X}_{\mathrm{d},i,k}\left(t\right)+b_i,\end{array}$$

where Y_d,i(t) is the deseasonalized and detrended PM_2.5 time series for grid square and season i, and $X_{\mathrm{d},i,k}\left(t\right)$ is the corresponding time series for the deseasonalized and detrended meteorological variable kϵ[1,5]. We fit the regression coefficients β_i,k and the intercept b_i. The regression is done stepwise, adding and deleting terms based on their independent statistical significance to obtain the best model fit (Draper and Smith, 1998). The fits and the selected meteorological variables differ by location and season, but with regional consistency (Table S1 in the Supplement). For meteorological variables not in the final MLR model, the regression coefficients β_i,k in Eq. (1) are zero.

Figure 1 Annual mean PM_2.5 concentrations in China from the CNEMC network. Panels (a) and (b) show values for 2013 and 2018 for sites with more than 70 % data coverage for the corresponding year. Panel (c) shows the ordinary linear regression trends on a 2^∘ × 2.5^∘ grid for sites with more than 70 % data coverage for each of the 6 years from 2013 to 2018. The trends are based on the time series of 10 d mean anomalies, as described in the text. Polygons in (a) define the four target regions of the Clean Air Action (Beijing–Tianjin–Hebei – BTH: 35–41^∘ N, 113.75–118.75^∘ E; Yangtze River Delta – YRD: 29–33^∘ N, 118.75–123^∘ E; Pearl River Delta – PRD: 21–25^∘ N, 111.25–116.25^∘ E; and Fenwei Plain – FWP: 33–35^∘ N, 106.25–111.25^∘ E, and 35–37^∘ N, 108.75–113.75^∘ E), to which we add Sichuan Basin (SCB; 27–33^∘ N, 103.75–108.75^∘ E). Number inset in panel (c) is the trend in mean PM_2.5 over the study region (21–41^∘ N, 103.75–123^∘ E). Dots in panel (c) indicate grid squares with significant trends (p<0.05).

2.3 Application to 2013–2018 PM_2.5 trends

We use the MLR model to remove the effect of meteorological variability from the 2013–2018 PM_2.5 trends, including not only the 10 d synoptic-scale variability but also any interannual variability and 6-year trends. This makes the standard assumption that the same factors that drive synoptic-scale variability also drive interannual variability (Jacob and Winner, 2009; Tai et al., 2012). We thus apply Eq. (1) to the meteorological anomalies $X_{\mathrm{a},i,k}$, obtained by removing the 6-year means of the 50 d moving averages from the 10 d mean time series. The anomalies calculated in this manner are deseasonalized but not detrended. This yields the meteorology-driven PM_2.5 anomalies Y_m,i:

$$\begin{array}{}\text{(2)}& Y_{\mathrm{m},i}\left(t\right)=\sum _{k=\mathrm{1}}^{\mathrm{5}}{\mathit{\beta }}_{i,k}{X}_{\mathrm{a},i,k}\left(t\right)+b_i.\end{array}$$

Consider now the PM_2.5 anomaly Y_a,i for grid square and season i obtained by deseasonalizing but not detrending the PM_2.5 data (by removing the 6-year means of the 50 d moving averages) in the same way as for the meteorological variables. The residual anomaly Y_r,i, after removing meteorological influence from the MLR model, is given by

$$\begin{array}{}\text{(3)}& Y_{\mathrm{r},i}\left(t\right)=Y_{\mathrm{a},i}\left(t\right)-Y_{\mathrm{m},i}\left(t\right).\end{array}$$

The residual is the component of the anomaly that cannot be explained by the MLR meteorological model, and we will refer to it as the meteorology-corrected data. It includes noise due to limitations of the MLR model and other factors but also a long-term trend over the 6-year period that we can attribute to changes in anthropogenic emissions. The same approach was recently applied by Li et al. (2019) to separate anthropogenic and meteorological drivers of ozone trends in China.

3 Results and discussion

3.1 PM_2.5 trends in China, 2013–2018

Figure 1 shows annual mean observed PM_2.5 concentrations from the CNEMC over China for 2013 and 2018 and the linear regression trends on the 2^∘ × 2.5^∘ grid based on the PM_2.5 anomalies Y_a,i(t), including effects of both changing emissions and meteorology. In 2013, PM_2.5 across most of China was well above the Chinese national air quality standard (annual mean of 35 µg m⁻³). BTH and FWP (Xi'an) had the highest PM_2.5 among the five target regions, with annual average concentrations of 108±34 µg m⁻³ (standard deviation describes variability in the annual average across sites in the region) and 108±11 µg m⁻³ respectively, followed by SCB (71±17 µg m⁻³), YRD (67±12 µg m⁻³), and PRD (47±7 µg m⁻³). PM_2.5 decreased dramatically from 2013 to 2018, by 34 %–49 % for the five target regions. Mean 2018 concentrations were 55±13 µg m⁻³ in BTH, 62±4 µg m⁻³ in FWP (Xi'an), 40±6 µg m⁻³ in SCB, 40±7 µg m⁻³ in YRD, and 31±5 µg m⁻³ in PRD.

Figure 2 shows the 2013–2018 relative trends of annual mean PM_2.5 for the five target regions along with the corresponding trends of SO₂, NO₂, and CO concentrations measured at the same sites. Also shown in the bottom panels are the MEIC trends in emissions of primary PM_2.5, SO₂, NO_x, NH₃, and CO for 2013–2017. The PM_2.5 observations show steady decreases for BTH, YRD, and SCB. PRD flattens out in 2015–2017 before decreasing again in 2018. FWP (Xi'an) decreases sharply by 47 % from 2013 to 2015 but rebounds in 2015–2017 before decreasing again in 2018. Trends at other FWP sites that became operational in early 2015 are similar to Xi'an. We argue in Sect. 3.3 that the 2015–2017 flattening at PRD and the anomalous 2013–2015 sharp decrease and 2015–2017 rebound at FWP are driven by meteorology.

Figure 2 Relative trends of 2013–2018 observed concentrations and 2013–2017 MEIC emission estimates for the five target regions of Fig. 1. The observed PM_2.5 trends are shown as thick lines. Values are annual means referenced to 2013. The observed concentrations are averaged over all sites in each region with at least 70 % data coverage for each year. The number of sites for each region is indicated. Fenwei Plain trends are for Xi'an, as other sites did not start operating until early 2015. Post-2015 relative PM_2.5 trends at these other sites are shown as the dashed line.

We see from Fig. 2 that only SO₂ has a decrease steeper than PM_2.5, indicating that SO₂ emission controls have been a major driver of the PM_2.5 trend (Lang et al., 2017; Shao et al., 2018). The overall SO₂ decrease for the five regions is 57 %–76 % from 2013 to 2018. The SO₂ decrease is quantitatively consistent with the decrease in SO₂ emissions estimated by MEIC (Zheng et al., 2018). This drastic cut of China SO₂ emissions is due to installation of scrubbers at coal-fired power plants (Siwen et al., 2015; Karplus et al., 2018; Silver et al., 2018), elimination of small coal boilers, improvement of coal quality (Zheng et al., 2018), and switch from residential coal to cleaner fuels (Zhao et al., 2018). We also see a significant decrease in CO of 18 %–43 % for the five regions from 2013 to 2018, again consistent with the MEIC and suggesting a reduction in organic PM_2.5 emissions. Primary PM_2.5 emissions in the MEIC decreased at a rate comparable to or steeper than CO. Trends in China PM_2.5, SO₂, and NO₂ presented here are consistent with previous studies (Silver et al., 2018; Ma et al., 2019) that cover a shorter time period than 2013–2018.

Figure 3 shows the time series of monthly mean PM_2.5 for the five target regions, illustrating the seasonal and interannual variability. All regions show winter maxima that can be mostly attributed to meteorology, including shallower mixing depth, lower precipitation, and increased stagnation in winter (Wang et al., 2018). Residential heating emissions in winter also contribute to the seasonality in China, north of about 33^∘ N (covering BTH and FWP in this study; J. Liu et al., 2016; Xiao et al., 2015). There is a large interannual variability, particularly in winter, that must be largely driven by meteorology. Studies for BTH have shown that high PM_2.5 in winter months is associated with weak southerly winds, low mixing depths, and high relative humidity (Zhang et al., 2014; Chang et al., 2016; Li et al., 2018; Shao et al., 2018). The relatively clean 2017–2018 winter was due in part to a higher frequency of northerly flow and associated ventilation (CMA, 2018; Yi et al., 2019). In addition, particularly aggressive actions by the government to restrict coal use that winter may have played a role in reducing PM_2.5 levels (Zhang et al., 2019).

Figure 3 Time series in 2013–2018 of monthly mean PM_2.5 concentrations over the five target regions. Values are averages from all sites in the region with over 70 % data coverage for each of the 6 years.

3.2 Meteorological influence on PM_2.5

Figure 4 shows the correlations of 10 d PM_2.5 concentrations with the individual meteorological variables used in the MLR model. Correlation coefficients r as low as 0.3 are statistically significant, more so when consistent across a region. Wind speed is negatively correlated with PM_2.5, as would be expected from ventilation, except in areas of the north where wind promotes dust formation (Lyu et al., 2017; Wang et al., 2004). Precipitation is also generally negatively correlated with PM_2.5, as one would expect from scavenging (Chen et al., 2018). The positive correlation between precipitation and PM_2.5 over northern China in spring is likely a result of high RH associated with precipitation in adjacent days.

Figure 4 Correlation coefficients (r) of PM_2.5 concentration with the individual meteorological variables used in the MLR model: surface wind speed (m s⁻¹), precipitation (mm d⁻¹), relative humidity (RH; %), surface air temperature (^∘C), and 850 hPa meridional wind velocity (m s⁻¹) for different seasons in China. The correlations are based on 10 d average observations on a 2^∘ × 2.5^∘ grid. Dots indicate statistically significant correlations (p<0.05).

Correlation between RH and PM_2.5 is positive over northern China, especially in winter, and negative over southern China, especially in summer. The positive correlation between PM_2.5 and RH over northern China in winter has been reported by previous studies and attributed in part to the role of aqueous-phase aerosol chemistry in driving secondary PM_2.5 formation (Zheng et al., 2015; He et al., 2018; Song et al., 2019; Pendergrass et al., 2019; Tie et al., 2017). The negative correlation of PM_2.5 with RH over southern China likely reflects the association of high RH with precipitation and onshore wind, which facilitate PM_2.5 wet removal and ventilation (Zhu et al., 2012; Leung et al., 2018).

Temperature has a positive correlation with PM_2.5 year-round over most of China (Wang et al., 2014; Leung et al., 2018), even though there is no strong direct dependence of PM_2.5 on temperature (Jacob and Winner, 2009). The correlation likely reflects the covariation of temperature with other meteorological variables, including wind speed, precipitation, and RH (Tai et al., 2012; Zhu et al., 2012). A possible explanation for the negative correlation with temperature in summer over the North China Plain could be the volatilization of ammonium nitrate at high temperatures (Kleeman, 2008). V850 shows strong positive correlations with winter PM_2.5 over most of China, and strong negative correlations with summer PM_2.5 over southern China, especially for the Pearl River Delta.

Figure 5 Resolving meteorological influences on PM_2.5 2013–2018 trends in China. Panel (a) shows the fraction of detrended and deseasonalized variance in 10 d PM_2.5 means explained by the stepwise multilinear regression (MLR) meteorological model. Panel (b) shows the meteorology-corrected trends to be compared to the trends in the original data shown in Fig. 1. Number inset in (b) is the trend in mean PM_2.5 over the study region (same definition as in Fig. 1). Dots indicate significant correlations (p<0.05) in (a) and significant trends (p<0.05) in (b).

Figure 6 Time series of 2013–2018 PM_2.5 10 d mean anomalies for the five target regions of Fig. 1. The anomalies are relative to the 2013–2018 means. The data are averaged over all measurement sites in each region with at least 70 % of data coverage for each year (same as for Fig. 2). The meteorological contribution to the anomalies as determined from the MLR model is shown in red. The long-term trend in the meteorology-corrected residual in blue (Eq. 3) is interpreted as being driven by changes in anthropogenic emissions. Values inset in each panel are the ordinary linear regression trends, with 95 % confidence intervals obtained by the bootstrap method.

Figure 5a describes the ability of the MLR model to account for PM_2.5 variability in relation to wind speed, precipitation, RH, temperature, and V850 as potential predictor variables. Results are presented as the coefficients of determination R² (fraction of variance explained) between observed and model PM_2.5 in the detrended deseasonalized time series. The R² values have been adjusted to account for different numbers of significant explanatory terms (predictor variables). R² values for the five target regions are 0.59 (BTH), 0.46 (YRD), 0.65 (PRD), 0.65 (SCD), and 0.41 (FWP). Figure 5b shows the meteorology-corrected PM_2.5 trends after removal of meteorological variability predicted by the MLR model, i.e., the trends in the residuals Y_r,i(t) in Eq. (3). The meteorology-corrected decreasing trend averaged across China is −4.6 µg m⁻³ a⁻¹, 12 % weaker than in the original data, meaning that 12 % of the PM_2.5 decrease in the original data is attributable to meteorology. We elaborate on this below for the five target regions.

3.3 Meteorology-corrected PM_2.5 trends for the five target regions

Figure 6 shows the 10 d mean PM_2.5 anomalies in the deseasonalized (but not detrended) data for the five target regions (Y_a(t) in Sect. 2.3). Also shown is the meteorological component Y_m(t) derived from the MLR meteorological model and the residual Y_r(t) (meteorology-corrected; Eq. 3) whose long-term trend can be interpreted as being due to changes in anthropogenic emissions. The PM_2.5 anomalies show large features on 10 d basis that can be mostly captured by the MLR model. The residual meteorology-corrected time series is much smoother, as depicted by the narrower 95 % confidence intervals in the anthropogenic residual trends than in the original observed trends. The meteorology-corrected trends differ by 3 % (YRD) to 27 % (SCB) from the observed trends. The YRD trend reflects a significant contribution from the December 2013 outlier, which reflects unfavorable meteorological conditions (Fig. S2) that are not adequately captured by the MLR model. If we exclude this outlier month from the time series, the observed YRD trend becomes $-\mathrm{5.7}\pm \mathrm{0.9}$ µg m⁻³ a⁻¹ and the meteorology-corrected trend becomes $-\mathrm{5.9}\pm \mathrm{0.7}$ µg m⁻³ a⁻¹.

Most remarkably, it appears that the 2015–2017 flattening in the PRD and 2015–2017 increase in the FWP (see Fig. 2) can be mostly attributed to meteorological variability as resolved by the MLR model rather than to emissions. The trend in the residual is more consistent with a steady 2013–2018 anthropogenic decrease in both regions. The MLR model shows that meteorology accelerated the PM_2.5 decline in PRD and FWP from 2013 to 2015 and contributed partly to the 2015–2017 PM_2.5 rebound over FWP. In particular, the high PM_2.5 anomalies in PRD in 2013 and early 2014 are driven by anomalously low V850, and the low PM_2.5 in winter 2015–2016 is associated with anomalously high southerly flow and precipitation (Fig. S4). The low PM_2.5 in FWP in the winter 2014–2015 is associated with anomalously high wind speed, low RH, and low temperature, while the high anomalies in the winter 2016–2017 are associated with anomalously low wind speed, high RH, and high temperature (Fig. S5).

4 Conclusions

Observations of fine particulate matter (PM_2.5) pollution in China from the extensive CNEMC network established in 2013 show large 2013–2018 decreases driven by emission controls with complicating influences from meteorology. Here we used a stepwise multiple linear regression (MLR) meteorological model to investigate and separate contributions from anthropogenic emissions and meteorology to these 6-year trends.

The CNEMC observations show 34 %–49 % decreases in PM_2.5 in the five megacity clusters targeted by the Chinese government's Clean Air Action to reduce anthropogenic emissions. Concurrent observations of SO₂, CO, and NO₂ are qualitatively consistent with these PM_2.5 decreases being driven by drastic cuts in emissions from coal combustion. At the same time, there is large interannual variability driven by meteorology particularly in winter when PM_2.5 is highest.

We used the stepwise MLR meteorological model to relate PM_2.5 anomalies across China to wind speed, precipitation, relative humidity (RH), temperature, and meridional velocity at 850 hPa (V850) as potential predictors. The model accounts for ∼50 % of the variance in the deseasonalized detrended PM_2.5 data, including 41 %–65 % for the five megacity clusters. Application to the PM_2.5 time series shows that meteorological variability contributed significantly to the 6-year trends across China and in the megacity clusters. Removing meteorological variability as given by the MLR model also reduces the uncertainty in the trend that can be attributed to emission controls. We refer to the data series after removal of meteorological variability as the meteorology-corrected data. Thus the 2013–2018 PM_2.5 decrease for Beijing–Tianjin–Hebei is $-\mathrm{9.3}\pm \mathrm{1.8}$ µg m⁻³ a⁻¹ in the original data and is 14 % weaker in the meteorology-corrected data ($-\mathrm{8.0}\pm \mathrm{1.1}$ µg m⁻³ a⁻¹). For the Sichuan Basin where the meteorological correction is particularly large, the PM_2.5 decrease is $-\mathrm{6.7}\pm \mathrm{1.3}$ µg m⁻³ a⁻¹ in the original data and is reduced by 27 % to $-\mathrm{4.9}\pm \mathrm{0.9}$ µg m⁻³ a⁻¹ in the meteorology-corrected data. The average 2013–2018 PM_2.5 decrease over our study domain is −5.2 µg m⁻³ a⁻¹ in the original data (Fig. 1c) and is reduced by 12 % to −4.6 µg m⁻³ a⁻¹ in the meteorology-corrected data (Fig. 5b).

Observations for the 2015–2017 period indicate a flattening of the PM_2.5 trend in the Pearl River Delta and an increase in the Fenwei Plain. We find from the MLR model that these 3-year trends can be explained by meteorological variability (including particularly steep 2013–2015 decreases) rather than by relaxation of emission controls.