Characteristics and Driving Factors of PM_2.5 Concentration Changes in Central China

Abstract

Despite nationwide control efforts, central China experiences persistently high annual PM_2.5 concentrations (~50 μg/m³), which are particularly severe in January (exceeding 110 μg/m³). This study employs an integrated approach combining a Multiple Linear Regression (MLR) model derived from random forest analysis with the WRF-CMAQ chemical transport modeling system to quantitatively disentangle the driving factors of PM_2.5 concentrations in central China. Key findings reveal significant spatiotemporal heterogeneity in anthropogenic contributions, evidenced by consistently higher north–south gradients in regression residuals (reflecting emission impacts), linked to spatially varying industrial and transportation influences. Critically, the reduction in anthropogenic impacts over six years was substantially smaller in winter (January: 27 to 23 μg/m³) compared to summer (15 to −18 μg/m³, July), highlighting the profound role of emissions in driving severe January pollution events. Furthermore, WRF-CMAQ simulations demonstrated that adverse meteorological conditions in January 2020 counteracted emission controls, causing a net increase in PM_2.5 of +13 μg/m³ relative to 2016, thereby offsetting ~68% of the reductions achieved through emission abatement (−19 μg/m³). Significant regional transport, especially affecting northern and central Henan, further weakened local control efficacy. These quantitative insights into the mechanisms of PM_2.5 pollution, particularly the counteracting effects of meteorology on emission reductions in critical winter periods, provide a vital scientific foundation for designing more effective and targeted air quality management strategies in central China.

1. Introduction

PM_2.5 pollution constitutes a critical global environmental challenge, imposing long-term public health burdens [1,2,3] that necessitate sustained governmental and academic focus. In China, two-phase clean air policies, the Air Pollution Prevention and Control Action Plan (2013–2017) and Three-year Blue-sky Plan (2018–2020), achieved significant improvements, reducing PM_2.5 concentrations by 53% (72 to 34 μg/m³) in 74 key cities during 2013–2023 [4,5]. Notwithstanding this progress, 237 of 338 Chinese cities (70.1%) still exceeded the National Ambient Air Quality Standards (NAAQS) in 2023, with central China exhibiting a 2.3 times higher pollution-episode frequency compared to coastal regions [6], highlighting the imperative for intensive efforts dedicated to air quality interventions.

Air quality variations arise from complex emission–meteorology interactions and are characterized by strong spatiotemporal heterogeneity, requiring clarification of emission versus meteorological contributions to regional PM_2.5 improvements [7,8]. Studies indicate that northern China’s winter PM_2.5 (2013–2015) was predominantly sourced from local combustion (64 ± 5%) [9], while national anthropogenic controls drove ~80% of PM_2.5 reductions (2013–2017) through SO₂ (−21%), NOx (−59%) and PM_2.5 (−33%) emission cuts [10]. Regional analyses further demonstrate these dynamics: Zhang et al. (2024) identified local emission reductions as responsible for 41–55% of PM_2.5 declines in the Pearl River Delta (2006–2020) [11], while Dong et al. (2022) revealed stronger anthropogenic emission impacts in winter/spring than in summer/autumn in China’s “2 + 26” cities [12]. Zheng et al. projected that ammonia controls could reduce PM_2.5 by 5% in Wuhan (2025–2030) [13], and Zhang et al. quantified emissions and meteorology as contributing 67% (40.5 μg/m³) and 33% (19.7 μg/m³), respectively, to PM_2.5 variability in Tianjin (2017–2018) [14]. Meteorological contributions have been estimated at 20–48% for historical PM_2.5 trends in eastern China [15,16,17], specifically ~20% for Beijing (2013–2017) [18], whereas emissions dominated declines (47%) in the Beijing–Tianjin–Hebei region [19], collectively underscoring dual controls’ significance.

However, directly linking emission cuts to concentration reductions remains challenging due to emission spatiotemporal heterogeneity, uneven meteorological resource distribution, and nonlinear pollutant–precursor responses. This complexity is exemplified in central China’s Henan Province (40.42–48.67° N, 118.67–128.00° E), persistently the most polluted region since 2017 despite stringent controls. Although its annual PM_2.5 declined from 79 μg/m³ (2015) to 48 μg/m³ (2022), concentrations substantially exceed Yangtze/Pearl River Delta levels (Figure 1a). This anomaly arises from a unique topography: the Taihang Mountains obstruct southward cold-air incursions, forming a Beijing–Hebei–Henan pollution transport corridor that triggers regional winter PM_2.5 episodes exceeding 110 μg/m³ (Figure 1b) [20], while the Funiu and Songshan Mountains impede atmospheric diffusion, driving heterogeneous PM_2.5 improvements across cities (Figure 2a). Despite component observations and model tracing in this region, quantitative assessment of combined emission–meteorology impacts under complex terrain remains scarce.

To accurately evaluate control efficacy and advance pollution management, this study quantitatively assesses emission reduction and meteorological impacts on Henan’s PM_2.5 (2015–2020), focusing on 2016 and 2020 for interannual meteorological comparisons. High-resolution hourly PM_2.5 and meteorological data for 18 prefecture-level cities (China Ministry of Ecology and Environment) are integrated with Multiple Linear Regression (MLR) and coupled WRF-CMAQ simulations to (1) decouple emission and meteorological contributions to PM_2.5 reduction in each city; (2) elucidate discrepancies in January versus December pollution declines; and (3) provide scientific support for targeted pollution control strategies in topographically complex regions.

2. Methods

2.1. Observation Data

Meteorological data are from the China Meteorological Science Data Sharing Service (http://data.cma.cn/) and include parameters such as wind speed (WS in m/s), temperature (Temp in °C), air pressure (Pre in hpa), relative humidity (RH in %), and visibility (Vis in km).

PM_2.5 concentrations are derived from real-time monitoring data from state-controlled stations published by the Ministry of Ecology and Environment of the People’s Republic of China (https://air.cnemc.cn:18007/, accessed on 21 October 2025) on an hourly scale in micrograms per cubic meter (μg/m³), and urban-scale averages are derived from state-controlled stations in each city, with data completeness >90%. To thoroughly analyze the impact of meteorological conditions on PM_2.5 pollution, this study utilizes data from national meteorological observation stations in Henan Province and its surrounding areas. Henan Province maintains a comprehensive national-level surface meteorological monitoring network, comprising 114 stations that cover all 18 prefecture-level cities. These stations are not arbitrarily placed but are systematically planned and selected by the China Meteorological Administration in accordance with national standards such as the “Technical Regulations for Site Selection of Surface Meteorological Observation Stations”. This ensures the representativeness, accuracy, and comparability of the data.

2.2. Statistical Modeling Framework

2.2.1. Rationale for Model Selection

PM_2.5 concentrations are governed by nonlinear interactions between emissions and meteorological factors, exhibiting spatiotemporal heterogeneity across China [21,22,23,24,25]. To address this complexity, machine learning (random forest, RF) and parametric regression (Multiple Linear Regression, MLR) were employed as complementary approaches. RF excels at capturing nonlinear relationships and feature importance ranking [26,27], while MLR provides interpretable quantification of meteorological contributions [28].

2.2.2. Random Forest Implementation

The RF model [29]—an ensemble of decorrelated decision trees—was trained on randomly selected subsets (bootstrap sampling) of 2015–2020 observational data to establish PM_2.5–meteorology relationships, with validation performed on withheld data [30,31]. Key hyperparameters of variables used to grow the trees were set to three, with a minimum node size or depth of five, and the number of trees in the forest was set to 300 in all models. In this process, meteorological variables (excluding temporal features) were normalized prior to training [29,32,33]. Computations used tidyverse (v1.3.1), openair (v2.9-1), and relaimpo (v2.2-6) packages in R (v4.0.2). The accuracy of the RF model has also been validated, as presented in

Table 1. RF model performance for testing data set.

Month	RMSE	R2	MFB	MFE
January	15.04	0.79	−0.03	0.18
July	5.53	0.72	−0.14	0.25
December	8.74	0.82	−0.23	0.31

Note: RMSE: Root Mean Squared Error; R: Correlation Coefficient; MFB: Mean Fractional Bias; MFE: Mean Fractional Error.

below. It is speculated that the notably higher RMSE in January is related to more frequent extremely high PM_2.5 in the observed data, while more efficient anthropogenic emission controls and stable meteorological conditions are the main reason for the higher correlation in December.

2.2.3. Multiple Linear Regression Protocol

City-specific Multiple Linear Regression (MLR) equations for PM_2.5 and six meteorological factors (air temperature, relative humidity, air pressure, precipitation, wind direction, and wind speed) in 18 prefecture-level cities were constructed as the following equations:

$$Y_c(t)=\sum _{k=1}^6{\beta }_k{X}_{c,k}(t)+b$$

(1)

where $Y_c\left(t\right)$ is the predicted PM_2.5 for city c, $X_{c,k}\left(t\right)$ is the hourly observations of the six meteorological elements, $k\in \left[\mathrm{1,6}\right]$ ${\beta }_k$ is the regression coefficient, and b represents the intercept. The regression is performed stepwise, adding and deleting terms based on their independent statistical significance to obtain the best model fit [34]. The meteorological contribution was isolated via

$${∆PM}_{2.5}^{met}=Y_c\left(t\right)-{∆PM}_{2.5}^{obs}\left(t\right)$$

(2)

Monthly aggregation of ${∆PM}_{2.5}^{met}$ enabled quantification of emission-driven changes during key months (January, July, December).

2.3. Chemical Transport Modeling

2.3.1. WRF-CMAQ Configuration

The WRF (version 4.13) and CMAQ (version 5.3.1) models were configured for a three-level one-way nested domain setup (Figure S1). Domains covered East Asia (36 km resolution) and central China centered on Henan Province (12 km resolution) and encompassed part of Henan Province and its vicinity (4 km resolution). Meteorological inputs for CMAQ were generated by WRF. Key physics and surface process WRF options are listed in Table S1. Initial and boundary meteorological conditions were derived from the 1° × 1° resolution, 6-hourly NCEP FNL global reanalysis data [35], with grid nudging enabled. The CMAQ simulation incorporated updated chemical mechanisms (SAPRC07), the AERO7 aerosol module, and the M3Dry bidirectional deposition model. Initial and boundary conditions for the outermost domain (36 km) were based on default CMAQ profiles. For nested domains, initial and boundary conditions were downscaled from the adjacent coarser domain. All simulations were spun up for 5 days. Model performance was evaluated by comparing simulated meteorological parameters (2 m temperature, 2 m relative humidity, 10 m wind speed, surface pressure) and PM_2.5 concentrations with observations (the detailed results of model validation are presented in Figures S2–S4). The verification of the WRF and CMAQ models can be found in Tables S2 and S3 of the Supplementary Materials.

2.3.2. Emission Inventory Processing

Anthropogenic emissions within the 36 km and 12 km domains over China utilized the 2016-based Multi-resolution Emission Inventory for China (MEIC). Emissions for other regions in the 36 km domain employed the Regional Emission inventory in ASia version 2 (REAS2) [36]. Both high-resolution inventories (0.25° × 0.25°) were regridded to their respective model domains using the US EPA’s Spatial Allocator tool. For the innermost 4 km domain focused on Henan Province, the local emission inventory was applied [37], also regridded from its native 3 × 3 km² resolution using the Spatial Allocator. Biogenic emissions across all domains were generated online using MEGAN v2.0 [38] driven by WRF meteorological data. Windblown dust emissions were also calculated online within the CMAQ framework. In this study, a set of VOC and PM speciation profiles were developed using emission measurements in China and temporal disaggregation by hourly allocation via species-specific profiles [39,40].

2.3.3. Sensitivity Experiment Design

The sensitivity experiment simulation method of “fixing emission inventory and changing meteorological conditions” was adopted in this study. Emission–meteorology interactions were quantified through factorial simulations:

$$D1={\displaystyle \frac{M_{2020}{E}_{2016}-M_{2016}{E}_{2016}}{M_{2016}{E}_{2016}}}$$

(3)

where M₂₀₁₆E₂₀₁₆ is PM_2.5 with 2016 meteorology and emissions, M₂₀₂₀E₂₀₁₆ is PM_2.5 with 2020 meteorology and 2016 emissions, and D1 reflects the meteorological deterioration or improvement.

The relative changes in the observed PM_2.5 concentrations are described as D3; then

$$O_{2020}=O_{2016}\times \left(1+D3\right)$$

(4)

$$O_{2020}=O_{2016}\times \left(1+D1\right)\times \left(1+D2\right)$$

(5)

Emission contributions (D2) were derived from observed changes (D3), accounting for covariance:

$$D3=D1+D2+D1D2$$

(6)

3. Results and Discussion

3.1. Emission Separation Contribution Based on Random Forest Method

Figure 2 presents the 2015–2020 average distributions of observed and predicted PM_2.5 concentrations across Henan Province. Observed PM_2.5 (Figure 2a) exhibits a pronounced north–south gradient, with maximum concentrations in Anyang (77 μg/m³), followed by Jiaozuo, Zhengzhou, and Pingdingshan, consistent with regional emission patterns [41]. Minimum values occur in Xinyang (54 μg/m³), succeeded by Zhumadian, Nanyang, and Sanmenxia. The regression-predicted spatial distribution (Figure 1b) replicates this north-high/south-low structure, confirming that the RF model accurately captures the province’s PM_2.5 spatial heterogeneity. Notably elevated pollution in north-central regions reflects pollutant accumulation within the Taihang Mountains transport corridor under prevailing northerly flows.

The regression residual (PM_2.5obs − PM_2.5pred), representing anthropogenic emission contributions (Figure 2c), demonstrates peripheral > central distribution. Maximum residuals occur in northern Puyang (6 μg/m³), followed by Anyang, Hebi, and Kaifeng, indicating persistent emission dominance that outweighed control measures during the study period. Conversely, Xinxiang exhibits the lowest residual (−1 μg/m³), trailed by Zhengzhou, Jiaozuo, and Luohe, signifying effective emission reductions that offset meteorological disadvantages. Eleven cities (61% of the province) in eastern, western, and southern regions show minimal residuals (<0.3 μg/m³), suggesting limited emission policy impacts. Collectively, these patterns demonstrate effective control measures in Xinxiang and Zhengzhou, while northern cities (particularly Puyang–Anyang–Hebi) and Kaifeng/Xuchang require strengthened mitigation.

Interannual variations in regression residuals (PM_2.5obs − PM_2.5pred) during January, July, and December (Figure 3 and Table S4) reveal significantly stronger emission-driven PM_2.5 reductions in summer than in winter, with progressive improvement observed from 2015 to 2020. Winter months (January) consistently exhibited higher anthropogenic contributions to PM_2.5 concentrations compared to summer, constituting a primary driver of severe winter pollution episodes. The emission impact peaked in January 2016 (+40.7% above baseline), declined through 2018 (−7%), then rebounded in 2019 (+2.6%). Conversely, July demonstrated a persistent six-year decreasing trend (−6.6 μg/m³ yr⁻¹), reflecting enhanced emission control efficacy. Following anomalous increases in 2015, both the number of cities exhibiting negative residuals (indicating effective controls) and their absolute magnitude increased annually, reaching maximum coverage (100% of cities) and intensity (−18.4 μg/m³ average) by 2020. Despite temporal adjacency, January and December exhibited fundamentally divergent spatiotemporal patterns. January 2015 featured maximum residuals centered on Zhengzhou (+59.3 μg/m³, Figure 3a), while January 2016 showed impact intensification in central-northern regions, with Luoyang peaking at +72.4 μg/m³ (Figure 3d). The 2017 northwest–southeast gradient culminated in an Anyang extreme (+79.23 μg/m³ Figure 3g), followed by marked provincial improvement in 2018 (mean +7.7 μg/m³, Figure 3j). January 2019 rebounded with western maxima in Nanyang (+52.9 μg/m³, Figure 3m), while January 2020 maintained universally positive residuals with persistent northern/eastern contributions (+3.4~44.4 μg/m³, Figure 3p).

In contrast, December impacts showed minimal 2016 rebound, followed by progressive annual improvement. Though temporally aligned with July trends, December patterns diverged fundamentally from January in both magnitude and spatial heterogeneity. During 2015–2016, anthropogenic emissions generated higher PM_2.5 concentrations in December than in January. Post-2017, December residuals decreased significantly relative to January (Δ = −30 μg/m³), indicating superior year-end control efficacy. This aligns with the observed 34% PM_2.5 reductions in December versus < 5% in January after 2017. Spatial analysis identified central-northern Henan (Anyang–Zhengzhou–Kaifeng corridor) as the persistent hotspot for January–December divergence, where enhanced transboundary transport during peak winter further diminishes local control effectiveness.

The PM_2.5 regression residuals (PM_2.5obs − PM_2.5pred) during January, July, and December in Henan Province are shown in Figure 4. Over the past six years, July residuals exhibited a significant decreasing trend (−5.5 μg m⁻³ yr⁻¹), transitioning from positive values in 2015 (15 μg/m³) to consistently negative readings, culminating at −18 μg/m³ in 2020 (Figure 4a), demonstrating progressively effective summer emission controls. Winter months displayed contrasting patterns: both January and December peaked in 2016 (37 μg/m³ and 48 μg/m^3, respectively), marking the maximum anthropogenic contribution to PM_2.5 elevations. While December achieved negative residuals by 2019 (−2 μg/m³) reaching −5 μg/m³ in 2020—indicating quantifiable mitigation efficacy—January residuals declined only modestly post-2018 despite transient improvement. Notably, January’s persistent positive residuals (minimum 8 μg m⁻³ in 2018) rebounded to 27 μg/m³ in 2019, establishing anthropogenic emissions as a primary driver of winter PM_2.5 extremes. Collectively, anthropogenic contributions decreased province-wide during 2015–2020, with summer outperforming winter (Figure 4b), while December demonstrated progressive improvement (−7.1 μg m⁻³ yr⁻¹), contrasting with January’s stagnant positive residuals (mean +14.3 μg/m³), underscoring the necessity of enhanced targeted control measures at the beginning of the year.

3.2. Meteorological and Pollution Characteristics of the Reference Year

Atmospheric mobility critically regulates pollutant dispersion, transport, and accumulation. The years 2016 and 2020 were selected as benchmark years for comparative analysis based on the preceding findings (detail in Figure S5). The WRF-CMAQ model was used to quantify meteorological impacts on PM_2.5 concentrations in Henan Province, disentangling emission versus meteorological contributions seasonally. This approach elucidates mechanistic pathways for regional air quality improvement.

Boundary layer height (BLH) modulates pollutant diffusion and exhibits established correlations with air pollution [42,43,44]. As shown in Figure 5, BLH distributions during January/July 2016–2020 reveal topographic modulation: the Funiu and Songshan Mountain ranges consistently exhibited 50–100 m lower BLH than adjacent areas. Summer 2020 displayed pronounced north–south BLH disparity (northern mean: 580 ± 40 m vs. southern: 380 ± 60 m), whereas winter exhibited minimal provincial variation. July BLH systematically exceeded January by 300–500 m province-wide. Notably, July 2020 BLH decreased significantly compared with 2016, with southern cities experiencing 300–500 m reductions versus 200 m in northern regions. Spatial patterns shifted from the 2016 east-high (Zhoukou–Shangqiu: 800–900 m) configuration to a 2020 north-high gradient along the Sanmenxia–Luoyang–Zhengzhou–Kaifeng corridor (north: 500–650 m; south: 200–550 m). Winter reductions were less pronounced (ΔBLHJan = 150–200 m), transitioning from 2016’s 250–500 m range to 2020’s more uniform 150–500 m distribution. Seasonal wind characterization reveals higher winter speeds (mean: 2.8 m/s) versus summer (1.9 m/s) [45]. Dominant northerly flows prevail in central-eastern Henan during winter, with distinct northwesterly currents in western cities (Luoyang/Sanmenxia) and easterlies in Xinyang (Figure 5). Summer features southeasterly dominance, though topographic steering generates (1) localized wind shifts in Anyang/Nanyang, (2) reduced velocities westward, and (3) closed cyclonic circulation near Sanmenxia–Jiaozuo–Luoyang. The 2020 winter wind field retained 2016 patterns but with strengthened easterly components and reduced velocities. Collectively, winter meteorological conditions—characterized by lower BLH and altered wind fields—significantly inhibit pollutant dispersion compared to summer, consistent with established research [46].

Figure 6 illustrates 2016/2020 PM_2.5 distributions simulated using the 2016 emission inventory (Equation (2)). The baseline scenario accurately replicates observed northeast-high/southwest-low concentration gradients [36], reflecting northern Henan’s higher emission density from industrial activity and coal consumption versus southern regions’ enhanced dispersion capacity [47]. Despite identical emissions, 2020 PM_2.5 exceeded 2016 levels by 18.7% province-wide, attributable to deteriorated meteorological conditions: reduced mean wind speed (2016: 2.52 m/s → 2020: 1.71 m/s) and temperature (16.78 °C → 15.67 °C) jointly weakened horizontal/vertical pollutant transport. Maximum concentration increases (>25%) occurred along the Taihang Mountain transport corridors (Anyang–Xinxiang–Jiyuan and Anyang–Xinxiang–Kaifeng)—regions within the “2 + 26” air pollution channel [11]—with northeast-to-southwest attenuation. Basin cities (Jiyuan/Sanmenxia/Pingdingshan) showed amplified sensitivity due to topographic confinement. Conversely, southern cities (Xinyang/Nanyang) experienced minimal increases (<10%) owing to lower emissions and attenuated transboundary transport.

3.3. Impact of Meteorological Conditions

Seasonal airflow variations fundamentally alter atmospheric pollutant transport pathways [48], inducing distinct meteorological impacts on air pollution across seasons. The winter versus summer meteorological influences on 2020 air pollutant concentrations in Henan Province were quantitatively assessed (Figure 7 and Figure 8), revealing critical seasonal divergences.

January 2020’s reduced wind speeds and boundary layer heights (Figure 5) enhanced localized accumulation of PM_2.5 from industrial/combustion sources following north–south transport. Consequently, the January PM_2.5 growth ratio (Δ = +20.56%) significantly exceeded annual averages. Spatial distributions show 2016 maxima (>160 μg/m³) concentrated northeast of Zhengzhou, with provincial concentrations spanning 100–130 μg/m³ along the Zhengzhou–Luoyang axis north of the Songshan Mountains (Figure 7a). By 2020, substantial PM_2.5 increases had emerged in Anyang, Hebi, and Pingdingshan. While >100 μg/m³ zones maintained 2016 spatial patterns, intensity amplified, with pollution epicenters shifting northward to Anyang–Hebi–Xinxiang–Jiyuan–Luoyang and northern Zhengzhou–Xuchang (Figure 7b). Enhanced winter pollution stems from synergistic mechanisms: (1) increased atmospheric stagnation frequency inhibits dispersion [49,50]; (2) heating-driven combustion emissions elevate primary pollutants and secondary nitrate (NO³⁻) formation [51]. These forces collectively drive Henan’s characteristic winter PM_2.5 enhancements. Comparative analysis reveals stark north–south divergence in 2020 versus 2016 (Figure 7c,d): maximum meteorological-driven increases (>70%) occurred in northern cities (Anyang/Hebi/Jiyuan), indicating impeded air quality improvement, while southern regions (south Luoyang/north Nanyang) showed reductions (<−10%) from favorable dispersion conditions.

Boundary layer heights were significantly reduced in July 2020 versus 2016 (Figure 5), suppressing vertical diffusion under prevailing east–southeasterly flow. Provincial PM_2.5 changes ranged from −10.43% to +24.93%. While 2016 maxima (30–50 μg/m³) localized in northwestern Xinxiang–Jiaozuo–Jiyuan and the Songshan northern flank (Figure 8a), 2020 high-value zones (>30 μg/m³) expanded province-wide, with northern Zhengzhou peaking at >60 μg/m³ (Figure 8b). Despite superior summer dispersion (enhanced precipitation scavenging [52] and phytoremediation efficiency), eastern Shangqiu and Zhoukou experienced disproportionate increases (+37.02% and +34.42%, respectively, Figure 8c,d). This anomaly reflects limited industrial emissions in eastern Henan (Zhoukou/Shangqiu) and adjacent regions (Fuyang/Bozhou/Huaibei/Suzhou), constraining pollution accumulation despite unfavorable transport.

Utilizing the variable control methodology (Equations (3)–(5)), Figure 9 and Table S5 quantifies emission-attributable PM_2.5 changes across Henan Province. During January–July 2020, all cities exhibited emission-driven improvements versus 2016 (mean ΔPM_2.5 = −19 μg/m³; range: −2 to −39 μg/m³ or −4% to −44%), displaying a distinct north-high/south-low gradient. Maximum reductions occurred in Xinxiang, Jiyuan, and Luoyang (>30 μg/m³), with emission reductions strongly correlating (R² > 0.82) with observed PM_2.5 decreases in most cities. However, in Jiyuan and Anyang, substantial theoretical reductions (Δ = −34 μg/m³ and −27 μg/m³, respectively) were offset by adverse meteorology, yielding net decreases of only −10 μg/m³ and −2 μg/m³. Hebi and Puyang showed PM_2.5 increases (+2 μg/m³ and +3 μg/m³) due to insufficient emission cuts compounded by meteorological deterioration.

Figure 10 and Table S6 reveal pronounced seasonal divergence in emission–meteorology interactions. Despite superior January emission reductions (mean Δ = −33 μg/m³ versus July: −17 μg/m³), net improvements were compromised by extreme winter meteorological deterioration. This reflects enhanced control measures during winter heating periods—including industrial production restrictions and stringent vehicle emission controls—implemented effectively in 2020. Conversely, summer’s naturally favorable dispersion conditions diminished emission reduction impacts, though province-wide improvements were sustained through consistent control policies. City-level analysis identifies Xinxiang, Jiyuan, Luoyang, and Pingdingshan as top performers in emission control (cumulative PM_2.5 reductions: −82, −75, −68, and −63 μg/m³, respectively). Conversely, Puyang’s marginal January reduction (−14 μg/m³) indicates persistent control challenges. Winter meteorology critically offset gains in several cities: unfavorable conditions increased PM_2.5 by +123 μg/m³ (Jiyuan), +102 μg/m³ (Anyang), and +64 μg/m³ (Pingdingshan), explaining Anyang’s and Puyang’s significant observed rebounds (+48 μg/m³ and +26 μg/m³, respectively).

During July 2020’s post-epidemic recovery, emission-driven improvements were moderate (mean Δ = −16 μg/m³) compared with winter. Jiaozuo, Shangqiu, and Anyang achieved notable summer reductions (−37, −34, and −25 μg/m³, respectively), while Xinyang showed minimal improvement (−3 μg/m³). Favorable summer meteorology (ΔPM_2.5 < +12.5 μg/m³ except Shangqiu) facilitated province-wide decreases despite emission rebound pressures. Exemplary cases include Jiaozuo and Shangqiu, where synergistic controls and meteorology drove >25 μg/m³ decreases, demonstrating achievable regional improvement pathways.

4. Conclusions

This study employs a dual-method approach integrating MLR modeling with WRF-CMAQ simulations to quantitatively resolve anthropogenic emission and meteorological contributions to PM_2.5 dynamics in Henan Province. Key findings reveal spatiotemporal heterogeneity in emission controls in Henan Province.

MLR analysis demonstrates significantly higher emission contributions (2015–2020) in north-central Henan, peaking at 6 μg/m³ in Puyang. While Xinxiang and Zhengzhou achieved effective emission reductions, persistent challenges remain in three northern cities, plus Kaifeng and Xuchang. Seasonally, emission control efficacy substantially outperforms winter measures during summer months. January/December emissions exert disproportionately adverse impacts in north-central regions, exemplified by Anyang’s 93 μg/m³ regression differential in December 2016. Post-2017, December control efficacy surpassed January performance (36 vs. 6 μg/m³), though January emissions remain dominant drivers of peak PM_2.5 pollution.

Meteorology–Transmission Synergy hinders the improvement of air quality. CMAQ simulations reveal that 2020 meteorological conditions amplified PM_2.5 pollution by 46% in Anyang versus 2016, with north-central cities (Hebi, Jiyuan) experiencing 3.2× greater meteorological deterioration than their southern counterparts (Xinyang, Nanyang). Winter demonstrates heightened vulnerability through dual mechanisms: (i) weakened horizontal/vertical dispersion enhancing local accumulation; (ii) intensified transboundary transport from the Beijing–Tianjin–Hebei region. This synergistic amplification particularly impacts pollution-corridor cities and topographically disadvantaged areas, substantially diminishing emission control efficacy during high-pollution episodes.

Improving air quality requires a comprehensive balance between emissions, meteorology, and policy impacts. Emission-driven improvements in 2020 showed 40% (2020 vs. 2016) greater efficacy in northern versus southern cities than 2016. However, adverse meteorology imposed a net +13 μg/m³ provincial PM_2.5 increase, offsetting 68% of emission reductions (−19 μg/m³). Winter emission benefits decreased by 48% versus summer, compounded by intensified transboundary transport weakening local controls. Critical cases include Anyang and Puyang, where modest emission improvements (−30 and +14 μg/m³) coupled with meteorological deterioration (+102 and +11 μg/m³) produced significant PM_2.5 rebounds (+48 and +26 μg/m³). These findings necessitate integrated January control strategies: enhanced local source regulation coupled with regional joint prevention protocols to counter transboundary impacts.

However, this study has certain limitations. The relatively short duration of the research period and the inherent difficulties in conducting a quantitative analysis of meteorological influences mean that the findings are largely context-specific. To enhance the practical utility of this research for government early-warning systems, future studies should integrate real-time meteorological forecasts for a more comprehensive assessment.