PM10 data analysis and temporal distribution
The pollutants exhibit a yearly rise at around the same times every day, followed by a decline at approximately the same time intervals, when the 10-year station data linked to PM10 pollutants is shown (Fig. 8). The main cause of this pattern is that daily activities connected to humans, such heating, traffic, and other activities, happen roughly around the same times. Because the height of the atmospheric boundary layer decreases throughout the night, more fossil fuels are used to heat homes and offices, which is the cause of the increase in PM10 levels after 6:00 PM. According to Givehchi et al. (2013), burning fossil fuels more throughout the night would hasten the release of ash and other comparable, bigger particles into the atmosphere. Autumn and winter are the seasons with the highest PM10 readings. Because there is less traffic on weekends than there is throughout the week, PM10 levels are often lower on weekends than on other weekdays. In a same vein, Al-Bayati, R. M. et al. (2021) discovered that the Van's PM10 readings dropped on weekends. It is imperative that PM10 readings not be interpreted in isolation from other climatic factors (Barlık, 2020). Lower PM10 measurements at air quality stations occur on days and months when wind speeds rise, for example, because increased wind speeds enhance the transfer of air pollutants in the atmosphere.
When examining the amounts (µg/m3) and movement directions of PM10 pollutant values over the years, it can be inferred that the predominant transport of pollutants originates from the southern directions (Fig. 9). Particularly, in 2022, a notable increase in pollutant values is observed compared to other years. This indicates that the region experienced a higher incidence of dust transportation events in 2022.
A significant increase in PM10 concentration is observed with southerly winds. This may indicate that southerly winds carry high particulate matter. The southern regions of the city, bordering countries like Iraq and Syria, are recognized for the substantial annual transportation of dust originating from these areas. There is also a certain concentration increase in the west direction, but this is less intense than in the south. The concentration is lower in the north and east, which may suggest that winds blowing in these directions bring relatively cleaner air. This is because Lake Van is located in the north and west, so there are no settlements and therefore no air pollutants.
As wind speed increases (shown by the outward expanding circles), the concentration of particles generally decreases.
The PM10 pollutant levels show a decreasing trend between 2013 and 2017, followed by a relatively stable pattern from 2017 onwards (Fig. 11). This decline might be associated with the increased use of natural gas in the city center during these years. Regarding the irregular extreme values, as discussed earlier, these deviations could be due to brief but intense dust transportation events impacting the area.
In the city, higher PM10 levels during the winter months of 2013 and 2014 can be associated with increased use of fossil fuels. In his research, Aladağ, E. (2023), also highlighted the rise in PM10 throughout the winter months for the same area between 2015 and 2020. Over the ten-year period, there's been a consistent rise in PM10 levels during the fall and winter seasons (Fig. 12). Hourly analysis also reveals an increase between 18:00 and 03:00.
SO2 data analysis and temporal distribution
Sulfur dioxide (SO2) is a gaseous substance found in the atmosphere, primarily emitted during the combustion of sulfur-containing fossil fuels, notably coal and oil, in energy generation and various industrial processes. SO2 is a byproduct of burning sulfur-rich fuels (Zhao et al., 2009). Over a 10-year period of data from the air quality monitoring station, fluctuations in SO2 levels are observed at similar times of the day (Fig. 13). During fall and winter, especially during the coldest hours, typically at night, SO2 levels tend to increase due to heightened fossil fuel consumption. According to Öztürk and Bayram (2019), the provincial center of Van's fossil fuel use in 2016 increased significantly over the winter months. SO2 levels exceeding 500 µg/m3, according to the national air quality index, fall into the unhealthy air category. Such upper-limit values were typically recorded in January and February of 2013.
The concentration of SO2 in 2013 was rather high and appeared to be localized in one direction (Fig. 14). This might imply that a source of pollution is moving in that direction. The high concentrations appear to drop and exhibit a more uniform distribution in the next years. This might indicate that pollution management efforts are working or that pollution sources are being decreased. The SO2 content is significantly lower and more uniformly distributed by 2021. This year, there has also been a little increase in one direction, which might indicate the emergence of a new source of pollution or the decreasing effectiveness of the present restrictions.
Analysis of the polar chart reveals that the concentration of SO2 is highest in the easterly direction (Fig. 15). This suggests that the pollution source is situated in this direction and that the wind is most likely coming from this direction. The graph indicates that there is a noticeable increase in SO2 concentration in this direction compared to other directions, which is most likely the result of winds blowing in this direction. Pollution levels are often lower in regions where wind speed rises (outward expanding circles). This might indicate that lower amounts are observed at greater wind speeds because the wind disperses the pollution.
It is evident from the graph's overall trend that the concentration of sulfur dioxide has been falling over time. Particularly between 2013 and 2015, there is a notable decrease. While the general trend remains downward, the pace of fall appears to have slowed considerably after 2015.
The graph's red line, which gradually slopes downward over time, depicts the general tendency. (Fig. 16). The peaks in the graph signify specific times of the day, particularly during winter, when there is an hourly increase in SO2 levels, mainly due to heightened consumption of fossil fuels.
The highest annual SO2 values were observed in 2013 and 2014. On an hourly basis, SO2 levels usually increase between 18:00 and 03:00, during the night hours (Fig. 17). These increments were commonly noticed during the fall and winter months.
Multivariate statistical data analysis
In order to enhance the interpretation of meteorological parameters contributing to the phenomenon of air pollution, multivariate data analysis has been conducted using the available data. The aim here is to apply dimensionality reduction techniques to variables that are numerous, collected over long periods, and have a complex structure. This approach will result in dividing all parameters into a smaller number of component groups with higher representational proportions. (Chen vd, 2007). For the selected dataset, the first step was to normalize the data and remove outliers using the R programming software. Subsequently, before proceeding with factor analysis, the Kaiser-Meyer-Olkin (KMO) test and Bartlett's sphericity test were applied to the dataset. The values obtained in the KMO (Kaiser-Meyer-Olkin) test are evaluated within the acceptance range of 0.5 to 1.0 (Table 4). A high test result value is considered to indicate that the elements in that dataset are well represented by other elements. (Kaiser, 1974). If the result of the KMO (Kaiser-Meyer-Olkin) test falls below 0.5, it is generally considered that factor analysis is not appropriate for that dataset. (Field, 2009). Another test, Bartlett's sphericity test, is considered meaningful for factor analysis when the result yields a p-value less than 0.05. The Bartlett test serves as a measure of how significant the correlation matrix is. In other words, when the p-value is less than 0.05, it indicates that the correlation between variables in the dataset is significant enough to make factor analysis meaningful. (Bartlett, 1950).
Table 4
Kaiser-Meyer-Olkin values
KMO Values | Level of Acceptance |
< 0.5 | Unacceptable |
0.5–0.6 | Mediocre |
0.7–0.8 | Good |
0.8–0.9 | Great |
0.9-1.0 | Superb |
According to the results of the KMO and Bartlett tests applied to the air pollution parameters using the SPSS program, the study shows a positive outcome (Table 5), indicating that it is suitable for proceeding to the factor analysis stage. In the case of the air pollution data, the KMO value was obtained as 0.646, and the significance level of the Bartlett test (sig. value) is p < 0.05, which demonstrates that factor analysis is appropriate for this dataset. With the favorable results of the KMO and Bartlett tests, the analysis of the elemental dataset was initiated. The primary objective in this phase is to represent the dataset by gathering the extensive and diverse measured data obtained from stations under fewer subcomponents through their correlations with each other. The behavior of parameters over time based on their correlation with each other is very important. The purpose of factor analysis is to clarify the rational connections between the final component groupings.
In the exploratory factor analysis conducted in the study, the aim was to identify the factors (components) that best represent the air pollution dataset while having the fewest variables (Usman et al., 2017). During the analysis, variance values and eigenvalues for all components were obtained, and the components with eigenvalues greater than 1 were preferred (Table 5). This selection criterion is often used to determine which components are worth retaining, as eigenvalues represent the amount of variance explained by each component (Cliff, 1988). Components with eigenvalues greater than 1 typically capture more variance in the data and are considered to be more meaningful in explaining the underlying structure of the dataset.
Table 5
Eigenvalues and variances
Component | Initial Eigenvalues | Extraction Sums of Squared Loadings |
Total | % of Variance | Cumulative % | Total | % of Variance | Cumulative % |
1 | 2.408 | 30.098 | 30.098 | 2.408 | 30.098 | 30.098 |
2 | 1.318 | 16.469 | 46.568 | 1.318 | 16.469 | 46.568 |
3 | 1.002 | 12.524 | 59.092 | 1.002 | 12.524 | 59.092 |
4 | 0.924 | 11.548 | 70.640 | | | |
5 | 0.863 | 10.781 | 81.421 | | | |
6 | 0.604 | 7.556 | 88.977 | | | |
7 8 | 0.503 0.379 | 6.282 4.741 | 95.259 100.000 | | | |
The rotated component matrix simplifies the interpretation of factors because it attempts to make the loadings more clear and interpretable. After applying the rotation method, the loadings of variables on factors are typically redistributed in such a way that each variable loads more heavily on one or a few factors, making it easier to understand the meaning of those factors.
According to the results in the table above, three factor values with eigenvalues greater than 1 have been obtained. The parameters included in each factor group are as follows: wind speed (ws), wind direction (wd), boundary layer height (Blh) and temperature are part of the first factor group, PM10 and SO2 belong to the second factor group, and precipitation and air pressure data are included in the third factor group (Fig. 19).
A component plot in the context of rotated factor analysis (also known as a rotated space) is a graphical representation that helps visualize the relationships between variables (indicators) and the extracted or rotated factors (Acal et al., 2020). It's a valuable tool for understanding how variables cluster around the factors (Fig. 19), which can aid in the interpretation of the factors and the underlying structure of the data. Understanding the link between air pollution levels and climatic conditions depends on how these data are categorized in PCA. The PCA result's grouping of PM10 and SO2 air pollutants together suggests that they are produced by similar sources and demonstrate similar atmospheric distribution and behavior. There is a significant connection between the variations in the concentrations of these pollutants since they are classified by the same component.
The wind's properties and the vertical structure of the atmosphere are represented by the three parameters Ws (wind speed), Wd (wind direction), and Blh (atmospheric boundary layer height). The direction and speed of the wind play a significant role in how air pollutants spread. An other significant factor influencing the mixing of surface pollutants into the atmosphere is the height of the atmospheric boundary layer (BLH). When these three factors are combined, it may be possible to see how they interact to move and disperse air pollution. Precipitation events are closely linked with air pressure, which acts as an indication of the overall weather. While high-pressure systems are typically linked to dry and steady weather, low-pressure systems are frequently related to precipitation. If these two variables are grouped together, it can mean that specific weather conditions have an impact on both of them in the same way.