3.3.1. Principal Component Analysis (PCA)

The sample quantity is crucial for good PCA. The PCA of the study was performed considering the chemical constituents of 90 PM2.5 samples. The outliers (those beyond 2SD) were removed, and the dataset was normalized prior to the operation [69]. When the value of KMO is close to 1, it indicates that there is a strong correlation between these variables (KMO indicates the amount of variance shared among the items designed to measure a latent variable when compared to that shared with the error), and these variables are more suitable for PCA [70]. In this study, the SPSS software package (IBM, version 24.0) was used to conduct PCA research on substances in PM2.5 to obtain the emission characteristics

of its pollution sources. For this study, the species of Li, Na, K, Mn, Cu, Zn, As, Pb, Al, Mg, Ca, Fe, Ba, Si, Ti, NH4 +, NO3 <sup>−</sup>, SO4 <sup>2</sup>−, OC and EC had strong correlations in the two seasons, and the PCA results are listed in Table 2.


**Table 2.** Matrix of loading factors of PCA in spring and autumn.

From Table 2, we know that PCA resolved six components explaining 80.5% of the variance in spring.

**Factor 1**: The first factor contributes 18.15% to the total factor contributions and is characterized by a high concentration of SIAs, which indicates that Wenshan was greatly affected by secondary inorganic aerosol pollution in spring. SIAs are mainly generated by the photochemical reactions of precursor gases (SO2, NH3, and NOx), which are emitted from specific identified sources of human activity (coal combustion, vehicle exhaust emission, and biomass burning). Therefore, the strict control of precursor gases is conducive to reducing PM2.5 levels.

**Factor 2**: The second factor contributed 16.28% of the total PM2.5, and mostly originated from natural sources, such as the lifting of dust or mechanical abrasion processes, which was identified by high concentrations of Al, Fe, Si and Ti, indicating the leading contribution of dust [71]. Si and Ti are the key tracers of soil dust caused by winds. The extra Ba is emitted from brake linings and tire tread wear. These results can be explained as a consequence of dust persisting in the atmosphere longer because of calm and low-speed winds.

**Factor 3**: The third factor resolved 14.15%, and represents the factor contribution from industrial emissions. The elements are related to the industrial metal smelting process and represent anthropogenic emissions from various industries near the sampling site [72].

**Factor 4**: This source provided 11.55% of PM2.5. OC and EC are considered to be tracers of motor vehicle emissions, and EC is an indicator of primary emissions of OC [73]. The presence of K also deserved our attention, directly indicating emissions from biomass burning.

**Factor 5**: Cu and Na were apportioned to this factor, which suggests that the effect of this factor was manifold, such as copper smelting and sea salt [74]. The contribution of this factor towards PM2.5 was 10.83%, as revealed by PCA. In addition, Na might travel long distances from the Indian Ocean, and Cu could have come from the nearby industrial area in Honghe Prefecture.

**Factor 6**: This factor is construction cement dust, which is represented by high concentrations of Ca and Mg [75]. This finding indicates that construction and demolition activities were prevalent in the urban areas in Wenshan during the sampling period, without effective measures for dust control. More precise and effective policies are needed for the local government to improve PM2.5 pollution.

In addition, PCA resolved five components explaining 78.7% of the variance in autumn. Different from spring, Factor 1 represent biomass burning and industry sources, contributing 29.40% of PM2.5. Factor 2 includes secondary inorganic aerosols and motor vehicle exhaust emissions, which resolved 18.73% of the factor contribution. Factor 3 represents metal smelting, with remarkable representative Al, Fe and Mg features, which can be attributed to the smelting production activities around the site. Factor 4 and Factor 5 are soil dust and construction dust, which resolved 9.30% and 8.91% of the factor contributions, respectively.

#### 3.3.2. The Long-Range Transport

To better understand the transport of airborne particles from distant sources, the 72 h backward trajectories starting at a height of 100 m at the sampling site were calculated using the Hybrid Single Particle Lagrangian Integrated Trajectory 4.0 (HYSPLIT4) model with a 12 h period (meteorological data from the Global Data Assimilation System (GDAS)). The back trajectories were classified into three clusters using TrajStat in this study.

In spring, the trajectories were grouped into three clusters (Figure 6). Cluster 1 (blue line), from the southwestern direction, was associated with slower and lower air mass trajectories and accounted for 57%. The other two trajectory clusters (green line and red line) came from the north and southwestern directions, accounting for 24% and 19%, respectively. Cluster 1 came from central Myanmar and passed through during spring sampling in northern Vietnam and the Honghe Prefecture in Yunnan Province, China, which explains the effect of Factor 3. At the same time, Cluster 2 (green line) came from the industrial region in Chongqing, which explained the source of biomass burning in Factor 1 and the industrial impact in Factor 3. Figure 6b shows that wind mainly originated from the south during the sampling period, which prevented the diffusion and great accumulation of NOX and SO2. Then, they formed secondary pollution through photochemical reaction transformation, which conforms to the SIA pollution in Factor 1. The higher wind speeds were also consistent with the contribution of Factor 2.

In autumn, the trajectories were grouped into three clusters from the southeastern direction (Figure 7). Cluster 1 (red line) came from Guangxi Province and passed through the industrial region in Baise, which explains the presence of industrial elements in Factor 1. The other two trajectory clusters (blue line and green line) accounted for 33% and 8%, respectively. Figure 7b also shows that the wind mainly originated from the south during the sampling period and resulted in the impossibility of the diffusion and dilution of pollutants, which was also the reason for the SIA pollution in Factor 2.

**Figure 6.** (**a**) Mean 72 h backward trajectories of each trajectory cluster during spring and the percentage of allocation to each cluster. (**b**) Wind roses of Wenshan during spring sampling.

**Figure 7.** (**a**) The mean 72 h backward trajectories of each trajectory cluster during autumn and the percentage of allocation to each cluster. (**b**) Wind roses of Wenshan during autumn sampling.

### **4. Conclusions**

In this study, PM2.5 samples were collected in Wenshan, and their mass concentration, chemical composition and source apportionment characteristics were analyzed in spring and autumn. The mean concentrations of PM2.5 were 48.00 ± 11.01 μg/m<sup>3</sup> and 41.64 ± 10.10 μg/m3 in spring (sampled on 19 April–3 May) and autumn (sampled on 12 October–26 October). The annual mean concentration of PM2.5 at the three sites was 44.85 ± 10.99 μg/m3, which was lower than that in Standard II (75.00 μg/m3) and higher

than that in Standard II (35.00 μg/m3). This means that the air quality in Wenshan is better than that in most cities in China.

WSIIs and OC were the main components of PM2.5, accounting for 26.91% and 23.80% of PM2.5, respectively. SIAs were the major contributors to WSIIs, due to the incomplete combustion of fossil fuels and the slathering of nitrogen fertilizers in agriculture. Wenshan was greatly affected by secondary inorganic aerosol pollution in the two seasons, which contributed 21.82% and 16.50% to the total factor contributions in spring and autumn, respectively. The ratio of NO3 <sup>−</sup>/SO4 <sup>2</sup><sup>−</sup> implied that the contribution of mobile sources was not significantly different from that of other developed areas. The daily mean value of OC/EC was 2.64–4.17 in spring and 2.74–3.65 in autumn, which indicates that the SOC was generated by the photochemical process during the sampling days in Wenshan. Moreover, the OC and EC concentrations in Wenshan had a better correlation in autumn (R = 0.86) than in spring (R = 0.69), which shows that OC and EC were derived from similar sources during autumn and from complex sources during spring. However, elements from anthropogenic sources (Ti, Si, Ca, Fe, Al, K, Mg, Na, Sb, Zn, P, Pb, Mn, As and Cu) accounted for 99.38% and 99.24% of the total inorganic element concentration in spring and autumn, respectively.

Source apportionment showed that SIAs (18.15%), the lifting of dust or mechanical abrasion processes (16.28%), industrial sources (14.15%), motor vehicle emissions (11.55%), copper smelting and sea salt pathways (10.83%), and construction cement dust emissions (9.58%) were the main pollution sources in PM2.5 in spring. Furthermore, source apportionment showed that biomass burning and industry (29.40%), SIAs and motor vehicle exhaust (18.73%), metal smelting (12.33%), soil dust (9.30%) and construction dust (8.91%) emissions were the main pollution sources of PM2.5 in autumn. Different source contributions were found in spring and autumn. According to the research results, the pollution prevention and control suggestions are as follows: (1) Exert related effective management for artificial sources, such as industry and construction sites, to accelerate industrial transformation and upgrading. (2) Adopt emission control measures, such as motor vehicle restrictions and the promotion of new energy transportation methods.

The results of cluster analysis indicate that the long-range transport of air pollutants has a profound effect on local air quality in Wenshan. Wenshan is mainly affected by long-distance atmospheric transmission from the southwest and the northeast in spring and autumn, respectively.

In this paper, chemical composition and source characteristics of PM2.5 in a plateau slope city were first studied, and the main sources of PM2.5 in Wenshan City are resolved. The results can provide scientific data to support PM2.5 pollution control in local and similar cities.

**Author Contributions:** Conceptualization, J.S. and X.H.; investigation, Y.Z. and L.R.; resources, X.H.; data curation, Y.F.; writing—original draft preparation, X.L.; writing—review and editing, J.S. and X.H.; supervision, P.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Key R&D Projects of China (grant number 2019YFC0214405) and the National Natural Science Foundation of China (grant number 21966016 and 21667014).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data used in this paper can be provided by Jianwu Shi (Shijianwu@kust.edu.cn).

**Acknowledgments:** This work was supported by the National Key R&D Projects of China (No. 2019YFC0214405), and the National Natural Science Foundation of China (No. 21966016, 21667014).

**Conflicts of Interest:** The authors declare that there are no competing financial interests that could inappropriately influence the contents of this manuscript.
