The unidirectional causality influence of factors on PM2.5 in Shenyang city of China

Air quality issue such as particulate matter pollution (PM2.5 and PM10) has become one of the biggest environmental problem in China. As one of the most important industrial base and economic core regions of China, Northeast China is facing serious air pollution problems in recent years, which has a profound impact on the health of local residents and atmospheric environment in some part of East Asia. Therefore, it is urgent to understand temporal-spatial characteristics of particles and analyze the causality factors. The results demonstrated that variation trend of particles was almost similar, the annual, monthly and daily distribution had their own characteristics. Particles decreased gradually from south to north, from west to east. Correlation analysis showed that wind speed was the most important factor affecting particles, and temperature, air pressure and relative humidity were key factors in some seasons. Path analysis showed that there was complex unidirectional causal relationship between particles and individual or combined effects, and NO2 and CO were key factors affecting PM2.5. The hot and cold areas changed little with the seasons. All the above results suggests that planning the industrial layout, adjusting industrial structure, joint prevention and control were necessary measure to reduce particles concentration.

variation 20 . Wang et al. analyzed the spatial and temporal changed of PM 2.5 , PM 10 , SO 2 , NO 2 ,CO and O 3 in 31 provincial capitals city of China from 2013 to 2014 21 . Zhang et al. explored that spatial distribution of the annual PM 2.5 concentration in 190 cities of Chinese, PM 2.5 concentrations were higher in northern than in southern regions, and lower in coastal areas than in inland areas, and it has significant seasonal variations 22 . Zhang et al. suggested that spatial and temporal patterns of PM 2.5 in 190 Chinese cities were positively correlated with its population size and polluting emissions, and negatively correlated with precipitation and wind speed 23 . Yan et al. found that temporal and spatial variation of PM 2.5 in Beijing, and the correlation between PM 2.5 and meteorological factors in different seasons 24 . Huang et al. observed that spatial-temporal change patterns of PM 2.5 in Beijing from August 2013 to July 2014, and the relationship between meteorological factors with PM 2.5 25 . Some studies have analyzed the correlation between air pollutants and meteorological factors. Yang et al. pointed out that it played a crucial role of interaction between PM 2.5 and meteorological factors on analyzing air pollution in 68 major cities and 7 geographical regions in China, and meteorological factors included relative humidity, surface pressure, wind speed and temperature, the relationship between meteorological factors with PM 2.5 has spatial and seasonal variations 26 . Li et al. indicated that the relationships between particulate matters with certain meteorological factors in Chengdu city Sichuan province, there was positive correlation between particulate matter and atmospheric pressure, negative correlation between particulate matter and temperature and wind speed, no significant correlation between particulate matter and relative humidity 27 . Chen et al. found that PM 2.5 mass concentration in the Jing-Jin-Ji region has a certain seasonal variation, and meteorological factors have strong correlation with PM 2.5 mass concentration. Moreover, studies have shown that the higher the PM 2.5 mass concentration, the greater influence of meteorological factors on PM 2.5 28 . Similar studies conducted in other cities, such as Beijing, Shanghai, Guangzhou and Wuhan 25,[29][30][31] .
However, a lot of numerical modeling works have also been carried out to quantify the source contributions to the ambient PM 2.5 in China. Wang et al. used PMF model to deduce the contribution sources of PM 2.5 , then they used backward trajectory model to identify the directions of these contribution sources, and determined that the distribution rates of different directions and sources are different 32 . Dou et al. applied chemical mass balance receptor model (CMB) to analyze the contribution of PM 2.5 . Source analytical results show that there are many main sources of PM 2.5 in Xining during the observation period. It mainly includes contribution rate of dust is 26.4%, contribution rate of coal dust is 14.5%, contribution rate of motor vehicle exhaust is 12.8%, contribution rate of secondary sulfate is 9.0%, contribution rate of biomass combustion is 6.6%, the contribution ratio of secondary nitrate is 5.7%, contribution ratio of steel dust is 4.7%, etc. Suggestions put forward to strictly control the pollution sources such as local coal burning and motor vehicles, and control the open source pollution dominated by dust 33 . Chen et al. Combines with the chemical mass balance model (CMB) to study the pollution level and main pollution sources of PM 2.5 in Xingtai city. The results show main sources of PM 2.5 are soot dust (25%), vehicle exhaust (11%), dust (9%), soil wind dust (3%) and construction dust(2%) during the observation period. The suggestions are that measures should took to control coal burning, dust and industrial production, as well as vehicle emissions 34 . Relevant studies have shown that the distribution of contribution sources of PM 2.5 is significantly different in different regions 35 .
Up to now, there are still great uncertainties about the temporal and spatial distribution of atmospheric particulate matter, possible influencing factors and quantify the source contributions to PM 2.5 , because different cities have different urban human characteristics and natural environmental factors, there is no unified conclusion on the causes of haze. A large number of literatures have only studied the important factors affecting the particle concentration, which considered as the sum of all explanatory variables and the relationship between them is direct. There are few studies on the indirect effects of particulate concentration. Without considering any indirect effects, important factors can hidden. Therefore, this paper uses path analysis to analyze the direct and indirect effects of a series of explanatory variables, to better analyze the potential variables affecting the particle mass concentration and better describe the complex relationship with particles 36 .
The northeast region is a traditional industrial area in China, accounting for 13% of China's land area. At the same time, the northeast region is adjacent to countries in Eastern Europe and Asia, so the air pollution problem is more serious, and it is easy to rise to international disputes. Compared with other places, many cities in the northeast region have to face the serious pollution problem in recent years especially Shenyang city. Shenyang city is the largest city in northeast region of China. However, there is very little research on air quality in the northeast. So air pollutants concentration data (PM 2.5 , PM 10 , SO 2 , NO 2 , CO and O 3 ) and meteorological parameters collected from 11 air quality monitoring stations within Shenyang city in 2017. Correlation analysis, path analysis and spatial autocorrelation used to analyze influencing factors of particulate matter pollution. The conclusions of this research provided reference for environmental management decisions.

Results
Overview of particulate matter data. Firstly daily average PM 2.5 and PM 10 mass concentrations at 11 monitoring stations in Shenyang were summarized. It found that annual mean PM 2.5 and PM 10 mass concentrations were 49 μg/m 3 and 84.9 μg/m 3 , respectively during 2017 at Shenyang. And daily average of PM 2.5 in summer and winter was 25.7 μg/m 3 and 63.9 μg/m 3 . The mass concentration in winter was about 2.5 times that in summer.The daily average PM 2.5 mass concentration ranged from 2.7 μg/m 3 to 247.8 μg/m 3 , and the lowest concentration was appeared in Donglinglu station, the highest concentration was in Liaoshenxilu station, which were located in eastern region and western region, respectively. The daily average PM 10 mass concentration varied widely from 2.0 μg/m 3 to 514.3 μg/m 3 , and the lowest concentration appeared in Donglinglu station and the highest concentration appeared in Jingshenjie station. Jingshenjie station located in the western region. It could be seen that there were great differences in spatial distribution of atmospheric particulate matter in Shenyang. In addition, the median of the atmospheric particulate matter mass concentrations in each monitoring station was far lower than their average value, indicating that the daily average value of the atmospheric particles mass concentrations was a right-skewed distribution. The difference in PM 2.5 and PM 10 mass concentrations of each monitoring station had obviously reached the corresponding standard. PM 2.5 and PM 10 of Senlinlu station met good level standard (PM 2.5 < 35 μg/m 3 , PM 10 < 50 μg/m 3 ) with the percentage of 60.7% and 44%, respectively, and significantly higher than other stations reaching good level standard. At the same time, the stations with the lowest proportion of PM 2.5 and PM 10 reaching the good level standard were Xinxiujie station, with only 35.7% and 12.5% reaching the standard respectively. For fair level of PM 2.5 (75 μg/m 3 ) and PM 10 (150 μg/m 3 ), just the opposite. Therefore, it indicated better air quality that the highest proportion with two levels(good level and fair level) was Senlinlu station, and worse air quality that the lowest proportion with two levels (good level and fair level) was Liaoshenxilu station, Wenhualu station and Xinxiujie station. Figure 1 further showed the spatial variation of PM 2.5 and PM 10 . The station numbers showed in Table 1. It shown that Senlinlu station and Yunonglu station in northern of Shenyang had the best air quality, and Liaoshenxilu station in western of Shenyang and Wenhualu station in central of Shenyang had the worst air quality. That was, PM 2.5 and PM 10 decreased gradually from south to north and from west to east.
The temporal-spatial variation of particulate matter mass concentration. As shown in Fig. 2(a), the annual temporal variation of PM 2.5 and PM 10 presented a saddle-like distribution, with most relatively high value appeared in January, November and December, and most relative low value occurred in July and August. While, the peak value about PM 2.5 and PM 10 observed as the section of overview of particulate matter data. Throughout the year, most PM 2.5 and PM 10 mass concentrations were at a mild or fair pollution level on most days. The high value in winter mainly due to winter heating and traffic emissions, the low value caused by meteorological factors. The binomial fitting formula was = + + ax bx c y 2 , where y was the dependent variable, x was the independent variable, a, b and c were the coefficients. The fitting results showed that: daily concentration of PM 2.5 and PM 10 conformed to the quadratic function, opening up, symmetry axis of the fitting function was concentrated in July and August. Figure 2(b) showed monthly average variation patterns of PM 2.5 and PM 10 . PM 2.5 and PM 10 presented a similar variation trend, and both of them with relatively higher mass concentration appeared in winter than that in summer. The relatively low concentration of PM 2.5 and PM 10 in summer related to the increase of strong convective air mass, precipitation and larger boundary layer height in summer. In the summer of 2017, the monthly precipitation was 117.7 mm, and the precipitation days accounted for about 40.2%.
In addition, four peaks of particulate matters mass concentration observed in the month of January, March, October, and December, respectively. The highest monthly average concentration of PM 2.5 and PM 10 mainly occurred at the western region and southern region in January especially. And the highest concentration station was Liaoshenxilu station, followed by Xinxiujie station. The reasons for the highest concentration appeared www.nature.com/scientificreports www.nature.com/scientificreports/ in these areas may be that the western and southern regions were industrial areas and important traffic areas. Moreover, residential heating emission was also an important factor causing the high concentration of particulate matter in winter. Cheng et al. pointed out that residential heating emission was also an important factor contributing to the high concentration of PM 2.5 in Beijing, in addition to geographical, meteorological factors and regional transportation of air pollutants 37 . Chen et al. observed that air quality in most cities in the Jing-Jin-Ji region deteriorated significantly in winter, mainly due to the additional emissions of air pollutants caused by residential heating 28 . Yang et al. found that high concentrations of particulate matter in many cities in northern China due to emissions from residential heating during the cold winter months 26 . The lowest monthly average mass concentration of particulate matter found at the Senlinlu station located in the north of Shenyang in August, that was, since the station was located in the forest area, it was not easy to be affected by human activities, resulting in the low emissions of local pollution sources. Figure 2 (c,d) were the diurnal variations of PM 2.5 mass concentrations and PM 10 mass concentration in four seasons. The diurnal variation trend PM 2.5 and PM 10 in four seasons was same. Winter and spring showed the most obvious changes, followed by autumn and summer. The change of particulate matter at night (0: 00-6:00) in summer and autumn was in steady state, the trend of particulate matter at night (0: 00-6:00) in spring and winter was decrease. In spring, autumn and winter, particulate matter mass concentrations rose from 7:00 in the morning, and the trend was bimodal state throughout the day. That is, the first peak appeared from 9:00 to 10:00 in a day, and the peak of winter morning delayed 1 to 2 hours than other three seasons. It was mainly due to morning rush hour on travel, anthropogenic emission source and relative humidity was high, atmospheric boundary layer height was low, inversion layer closing to the ground appears, pollutant diffusion conditions was relatively bad. After the peak at morning, temperature increases, relative humidity decreases, atmospheric convection movement gradually strengthens, diffusion conditions improvements, traffic pollutants reduces, particulate matter mass concentrations gradually decreases, reached to the lowest point around 17:00 in the day. Particulate matter mass concentrations gradually increased after 17:00, it was due to evening rush hour, increasing of man-made emissions, occurrence of cooking fume pollution, coupled with the industrial electricity enter the cheaper stage, industrial pollution boost, so that the particulate matter mass concentration rose and remained at a high state. In addition, the diurnal variation of particulate matter was higher at night than the day, reached to the lowest point around 17:00. The four seasons have similar trends, but varying ranges with respect to different seasons. It shown from Fig. 2 (c,d) that the variation curve of the particulate matters mass concentration presented bimodal distribution in each season, in which the absolute value was largest between the peak and the valley in spring.
The unidirectional causality influence of factors on PM 2.5 . Here, correlation analysis and path analysis were used to quantify the unidirectional causality effect of influence factors on PM 2.5 concentration.
The Pearson correlation were analyzed between six air pollution, i.e., PM 2.5 , PM 10 , SO 2 , NO 2 , CO and O 3 , and the common meteorological factors(atmospheric pressure, temperature, wind speed and relative humidity) in different seasons in Shenyang. According to the Pearson correlation coefficient shown in Fig. 3, the meteorological factors affecting different air pollutants varied in different seasons. In spring, such as PM 2.5 , SO 2 , NO 2 , CO and O 3, they were more susceptible to the changes of temperature, wind speed and atmospheric pressure, but the correlation between them with the relative humidity was not significant. Pearson coefficient indicates that O 3 was positive correlated with temperature, indicating that O 3 has strong dependence on temperature. Atmospheric pressure was positive correlated with PM 2.5 and PM 10 , which resulted from the fact that atmospheric pressure obstructs the upward movement of particulate matter, and leaded to the accumulation of particulate matter. In summer, the relative humidity was negative correlated with PM 10 , SO 2 , NO 2 and O 3 . This was mainly due to the low relative humidity and strong wind that easily carry surface dust, and they could combine the dust with water vapor, and easily formed fog and haze, and then made gaseous pollutants not easy to spread. SO 2 had strong negative correlation with wind speed, CO has positive correlation with temperature and relative humidity, and NO 2 has strong correlation with temperature, atmospheric pressure, wind speed and relative humidity. PM 10 and O 3 were positive correlated with wind speed, which was obviously due to strong wind speed in summer, which caused dust suspension and regional transport of O 3 . Particulate matter concentration was positive correlated www.nature.com/scientificreports www.nature.com/scientificreports/ with temperature, indicating that secondary particles transformed by photochemical process at higher temperature. Temperature was the main factor of O 3 , The formation of O 3 depended on the intensity and duration of solar radiation. In autumn, wind speed was an important meteorological factor affecting most pollutants. SO 2 , NO 2 , CO, PM 10 and PM 2.5 were negative correlated with wind speed, indicating that wind speed has important effect on pollutant diffusion in autumn. Temperature was negative correlated with NO 2 , PM 10 and PM 2.5 , mainly due to heating emission caused by temperature decline. In winter, relative humidity and wind speed were significant meteorological factors affecting pollutant. O 3 was negative correlated with relative humidity, which caused by reduced visibility under high humidity conditions. Low visibility can weaken photochemical activity, thus reducing O 3 concentration. It can be found that PM 2.5 was significant positively correlated with relative humidity only in winter, but insignificant in other seasons. SO 2 , NO 2 and CO are positive correlated with relative humidity and significantly negatively correlated with wind speed. PM 10 and PM 2.5 are positive correlated with relative humidity and negative correlated with wind speed. In almost all seasons, particulate matter concentration and four gas pollutants were negative correlated with wind speed, which indicated that strong horizontal diffusion reduce the particulate matter concentration, especially PM 2.5 . Particulate matter concentration correlated with relative humidity, which indicated the importance of aerosol particles hygroscopicity. In addition, these gas pollutants also correlated significantly with the relative humidity in autumn and winter. Whether other meteorological factors significantly affected the gaseous pollutants depended on different seasons. As shown in Fig. 3, temperature and air pressure were the main factors affecting PM 2.5 in spring and summer, while in autumn and winter temperature was no longer the main factor affecting PM 2.5 . Instead, wind speed had significant negative correlation with PM 2.5 . The relative humidity was positive correlated with PM 2.5 only in winter and they had no significant correlation in other seasons. Wind speed was the main factor affecting PM 10 . The relative humidity correlated with PM 10 significantly only in summer and winter. Temperature and air pressure had no significant correlation with PM 10 . In general, wind speed was the most important meteorological factor affecting particulate matter mass concentration, and temperature, air pressure and relative humidity were also the key affecting factors in some seasons.
Considering gaseous pollutants in the atmosphere possibly formed secondary particles through atmospheric chemical reactions to realize the transformation from gas to particles. Pearson method used to discuss the effects of four gaseous pollutants and PM 10 on PM 2.5 (Fig. 3). It found that PM 2.5 had significant positive correlation with PM 10 , SO 2 , NO 2 and CO, and weakly correlates with O 3 .
In addition to the analyzing for unidirectional causality influence of factors of meteorological factors and air pollutants on PM 2.5 , path analysis used to further estimate the direct and indirect effects of meteorological factors on air pollutants in different seasons.
Path analysis results (Fig. 4) showed that: In spring, NO 2 has the largest direct effect on PM 2.5 , while SO 2 and CO had the largest indirect effect. CO and NO 2 played the biggest role in direct and indirect effects, and atmospheric pressure and relative humidity played the greatest positive effect. In summer, O 3 had the largest direct effect on PM 2.5 , followed by relative humidity. While, SO 2 and O 3 , as well as temperature had the largest indirect effect. The pollutants with the largest combined effects of direct and indirect effects were SO 2 and O 3 , and the meteorological factors were temperature and wind speed. In autumn, the most direct effect on particulate matter was NO 2 , followed by relative humidity. The most indirect effect was CO, SO 2 and NO 2 . The pollutants with the largest combined effect of direct and indirect effects were CO, SO 2 and NO 2 . In winter, CO and temperature has the largest direct effect on particulate matter, while CO, SO 2 and NO 2 have the largest indirect effect. The pollutants with the largest combined effect of direct and indirect effects were CO, SO 2 and NO 2 . The meteorological factor with the biggest positive effect was relative humidity, and the biggest inhibiting was wind speed.
The above analysis results indicated that NO 2 and CO were the key factors affecting PM 2.5 in Shenyang. This showed that the incomplete combustion of coal in autumn and winter in Shenyang increased CO concentration in the air, and PM 2.5 concentration increased accordingly, incomplete combustion in the heating process can be considered more. Secondly, NO 2 was another major factor affecting PM 2.5 concentration in multiple seasons, which indicated that vehicle exhaust and industrial exhaust gas contributed to the increase of PM 2.5 mass concentration.
Finally, path analysis was used to consider the heavy pollution weather throughout the year (Fig. 4), that was, PM 2.5 and PM 10 concentration were higher than level 2 respectively (PM 2.5 > 75, PM 10 > 150). The results showed that the positive direct action of NO 2 and the negative direct action of temperature were the largest. The positive indirect effect of CO and SO 2 , the negative indirect effect of O 3 and wind speed were the largest. Therefore, the positive comprehensive effect of NO 2 and CO was the largest, and the negative comprehensive effect of wind speed and temperature was the largest. This was consistent with the results of the four seasons.

Spatial relationship between air pollutants and meteorological factors. Moran's I scatter diagram
obtained for particulate matter in Shenyang to test the spatial autocorrelation of PM 2.5 and PM 10 (Fig. 5). The station numbers showed in Table 1. The results showed that the Moran's I index of PM 2.5 and PM 10 in the four seasons were greater than 0 and their p values were all less than 5%, which meant that both PM 2.5 and PM 10 showed significant spatial autocorrelation among the different monitoring points.
Moran's I scatter diagram were divided into four quadrants, which corresponded to: high-high concentration area (upper right), low-low concentration area (lower left), low-high concentration area (upper left) and high-low concentration area (lower right). The high-high concentration area was also known as hot spot, indicating that these areas were at the high risk level; The low-low correlation concentration area was cold spot, showed low risk level. The low-high concentration area showed that low concentration zone was surrounded by high concentration zone. The high-low correlation area was just the reverse with the low-high correlation area.
As could be seen from Fig. 5, most of the monitoring stations such as Xiaoheyan, Wenhualu, Lindongjie, Hunnandonglu, Xinxiujie, Jinshenjie and Liaoshenxilu were located in the high-high concentrated areas, i.e. hot spot, there PM 2.5 mass concentration was higher than other areas in Shenyang city. These stations were mainly located in the center and the southwest of Shenyang city and showed strong spatial correlation. The high-high concentration area remained unchanged with the change of seasons. The stations of Donglinglu, Senlinlu and Yunonglu that were located in the eastern and northern of Shenyang city belonged to the low-low concentration www.nature.com/scientificreports www.nature.com/scientificreports/ area, namely cold spot area, where PM 2.5 concentration was low. The low-low area varied little with the seasons. In addition, although Taiyuanjie monitoring station was also located in the central of Shenyang, it belonged to the low-high area, which surrounded by high pollution area. The low-high region remained unchanged in spring and summer, but moved northward in autumn.
The mass concentration distributions of PM 10 were slightly different from that of PM 2.5 (Fig. 5). In terms of PM 10 , the stations distribution in high-high concentration area was only one less than that of PM 2.5 , which is Xiaoheyan station. The high-high aggregation area in spring and autumn was more widely distributed. The low-low aggregation area of PM 10 were the same stations with that of PM 2.5 . Both Xiaoheyan and Taiyuan streets were in the low-high area surrounded by high pollution. www.nature.com/scientificreports www.nature.com/scientificreports/ The number monitoring sites distributed in the high-high concentration areas was far more than that distributed in low-low concentration areas for both the PM 10 and PM 2.5 . It indicated that the number of high-value concentration areas was far more than that of low-value concentration areas, and the distribution range was wide. This suggests that the joint prevention and control between regions was a necessary measure to reduce particulate matter concentration. The aggregation areas of PM 10 and PM 2.5 had little changes throughout the year. It meant that the spatial correlation analysis results of urban agglomeration in winter and summer were consistent. It indicated that apart from meteorological factors and heating, industrial structure layout of energy consumption between these areas were also an important factor affecting particulate matter concentration. Therefore, planning the industrial layout and adjusting the industrial structure were one of the important means to reduce particulate matter concentration. www.nature.com/scientificreports www.nature.com/scientificreports/ High particulate matter mass concentration analysis. According to the previous analysis in this research, particulate matters concentration closely related to meteorological conditions. Heavy pollution occurred most often in winter, spring and autumn. As shown in Fig. 6, conditions with low wind speed, low temperature and low relative humidity in spring, with low temperature and high relative humidity in autumn, and with low wind speed, low temperature and high relative humidity in winter, pollution weather with particulate matters reaching level 2(fair level) or above was likely to occur. Regional emergency mitigation measures should be taken earlier to prevent and control heavy pollution weather under adverse meteorological conditions.
The particulate matter mass concentration values with different wind direction shown in Fig. 7. It found that high particulate matter mass concentration mainly accompanied by southwest wind, while low particulate matter mass concentration occurred at the same time as northwest wind. As PM 10 was heavier than PM 2.5 , it was not easy to transport in a long distance, so the problem of local pollution in the southwest was more serious.  (1) The annual temporal variation of PM 2.5 and PM 10 presented a saddle-like distribution pattern along with the observation days. The monthly average change pattern of PM 2.5 and PM 10 showed similar trends and the mass concentration in winter was higher than that in summer. The diurnal variation trend was bimodal state throughout the day and in winter and spring the diurnal variation was obvious, followed by autumn and summer. (2) Particulate matter concentration closely related to meteorological conditions. Meteorological factors affecting air pollution most were different in each season in Shenyang. Wind speed was the most important meteorological factor affecting PM 2.5 and PM 10 , temperature, air pressure and relative humidity also the key affecting factors in some seasons. In addition to the meteorological factors, the possible air pollutants affecting particulate matter considered, and the analysis revealed that PM 10 , NO 2 and CO were also the key factors affecting PM 2.5 in Shenyang. The adverse meteorological conditions tended to form severe pollution weather with high particulate matters concentration in some seasons. High particulate matter mass concentration mainly accompanied by the southwest wind, while the low particulate matter mass concentration occurred with the northwest wind. (3) PM 2.5 and PM 10 showed significant spatial autocorrelation. The changes in the aggregation area of PM 2.5 and PM 10 in each season was small. The number of high-value aggregation areas was much higher than that of low-value aggregation areas and the variation range was wide, which indicated that regional joint prevention and control measures should strengthened.

Methods
There were 11 air quality monitoring stations in Shenyang, covering eastern area, southern area, western area, northern area and central area (  Where R j 2 is direct determination coefficient of x j to y, and R jk is the indirect determination coefficient of x j about y through x k . (2020) 10:8403 | https://doi.org/10.1038/s41598-020-65391-5 www.nature.com/scientificreports www.nature.com/scientificreports/ Where n is the number of monitoring points. w ij is the value of the space weight matrix w, which is equal to 1 when the area of monitoring point i is adjacent to the area of monitoring point j, otherwise it is equal to 0. x i , x j respectively represent the pollutant concentration value of monitoring point i and j. The value of Moran's I index is generally between [ 1,1] − . When I 0 > , it means that it is a positive spatial correlation, and the values clustered high in space. When < I 0, it means that it is a negative spatial correlation, and the values clustered high-low in space; When = I 0, it means that there is no spatial correlation. P value is the significance of Moran's I index. When P is lower than 0.05, the spatial correlation is considered significant; otherwise, it is not significant.