Spatiotemporal changes in fine particulate matter and ozone in the oasis city of Korla, northeastern Tarim Basin of China

Air pollution is a serious environmental health concern for humans and other living organisms. This study analyzes the spatial and temporal characteristics of air pollutant concentrations, changes in the degree of pollution, and the wavelet coherence of the air quality index (AQI) with pollutants in various monitoring stations. The analysis is based on long-term time series data (January 2016 to December 2023) of air pollutants (PM2.5, PM10, and O3) from Korla, an oasis city in the northeastern part of the Tarim Basin, China. The concentrations of PM2.5, PM10, and O3 in Korla showed a cyclical trend from 2016 to 2023; PM10 concentrations exhibited all-season exceedance and PM2.5 exhibited exceedance only in spring. PM2.5 and PM10 showed a seasonal distribution of spring > winter > fall > summer; O3 concentrations showed a seasonal distribution of summer > spring > fall > winter. Strong positive wavelet coherence between PM and Air Quality Index (AQI) data series suggests that the AQI data series can effectively characterize fluctuating trends in PM concentrations. Moreover, PM10 levels IV and VI were maintained at approximately 10%, indicating that sand and dust have a substantial influence on air quality and pose potential threats to the health of urban inhabitants. Based on the results of this study, future efforts must strengthen relative countermeasures for sand prevention and control, select urban greening species with anti-pollution capabilities, rationally expand urban green spaces, and restrict regulations for reducing particulate matter emissions within city areas.

Air pollution is a pressing global environmental concern that has significant impacts on atmospheric visibility, local climate, and the physical and mental health of human beings 1,2 .Air pollution is defined as a condition in which the concentration of specific substances in the air exceeds a certain threshold over a certain period, causing harm to humans or ecosystems 3,4 .Even brief exposure to air pollution raises a noteworthy risk to human health, exacerbating the Global Burden of Disease (GBD) and ultimately contributing to hospitalization and mortality rates 3,5,6 .
Given rapid economic growth, industrialization, and urbanization, the deterioration of urban living environments has become a significant concern.There are up to 3.7 million premature deaths worldwide due to outdoor air pollution 7 .Recent data shows that 7 of the 10 most polluted cities in the world are in China 8 .Air pollution in China exceeds the levels recommended by the WHO 9 and is the leading cause of death after heart attacks, dietary risks, and smoking [10][11][12] .Therefore, air quality has garnered significant attention from both the Chinese government and the general public.
Particulate matter (PM) is of great concern due to its small size and large surface area, which makes it easy to adsorb airborne microorganisms, heavy metals, and other toxic substances.PM can penetrate the human body by respiration and settle in the lungs, posing a threat to human health 13 .The 2018 Environmental Performance Index [EPI] shows that the environmental quality of China ranked fourth, including PM 2.5 14 , China's PM 2.5

Study area
Korla is the capital city of Bayingolin (Bayin'guoleng) Mongolian Autonomous Prefecture (BMAP), China's largest prefecture-level administrative region, is located in Southern Xinjiang.The city sits at the southern foot of the Tianshan Mountains, on the northeastern edge of the Taklamakan Desert, southwest of Lake Bosten.Korla is situated in the transitional zone between the distinct climates of the northern and southern borders; cold and humid air from the north enters the territory of Korla, where it is blocked by the Khora and Kuruk mountains, resulting in a gradual decrease in humidity.It belongs to a warm temperate continental arid climate, characterized by adequate light and heat resources, with an average altitude of 950 m, an average annual rainfall of about 65 mm, and an average annual sunshine of about 2920 hours 23 .The dominant wind direction on the ground throughout the year is northeasterly.The city borders the Taklamakan Desert, experiencing 47.1 days of sand and dust weather per year, leading to severe sand and dust pollution 23,24 .Covering an area of approximately 7,268 km 2 , Korla serves as the gateway to the Middle Road of the ancient Silk Road.

Methodology
Data sources PM 2.5 , PM 10 , and O 3 concentration data for Korla between January 1, 2016, and December 31, 2023, were retrieved from the Air Quality Data Network (www.aqist udy.cn/ histo rydata).We selected three state-controlled monitoring stations in Korla that offer uninterrupted data monitoring, namely, Peacock Park (S1), Cotton Spinning Factory (S2), and Economic Development Zone (S3).These stations represent commercial, industrial, and residential areas within the primary urban area of Korla (Fig. 1 and Table 1).

Air quality index
The AQI was calculated according to the Technical Provisions from Ambient Air Quality Index (for Trial Implementation) [HJ 633-2012] and Air Quality Standards" (GB 3095-2012) 25 , as specified below: where IAQI P stands for the air quality score of pollutant P, C P is the concentration of pollutant P, BP Hi and BP L0 represent the upper and lower limits of the corresponding standard concentration, respectively, IAQI Hi and IAQI L0 are the air quality sub-indexes corresponding to BP Hi and BP L0 , respectively.
where IAQI and n are the air quality subindex and pollutant item, respectively.As known from the formula (2), if the air quality sub-index for multi-pollutants exceeds the standard, then AQI takes the largest air quality sub-index of pollutants.
All three monitoring sites are categorized as the Class II zone of the ambient air functional area classification (AAFC).Based on the associated requirements, the Class II zone applies to National Secondary Concentration Limits [NSCL] 25 .The 24-h average NSCL limits corresponding to PM 2.5 and PM 10 are 75 and 150 µg•m -3 , respectively; the NSCL for the daily maximum 8-h average concentration of O 3 is 160 µg•m -3 .
Based on the AQI levels and corresponding concentration limits of PM 2.5 , PM 10 , and O 3 in the Technical Provisions on Ambient AQI (for Trial Implementation) [HJ 633-2012], PM 2.5 , PM 10, and O 3 were classified into six levels 26 (Table 2).The higher the ambient AQI, the more serious the pollution is. (1)

Data processing
Origin2021 was used to measure the average concentrations of PM 2.5 , and PM 10 over 24 h, and O 3 over 8 h at the three monitoring stations.Additionally, the software was used to visualize the characteristic curves of the spatiotemporal changes of each pollutant, along with the percentage of the pollution level.Additionally, Origin2021 was employed to chart the characteristic curves depicting the spatiotemporal pattern of each pollutant along  characteristics of the object under analysis in both the time and frequency domains.Wavelet coherence was applied to examine the correlation between two sequential data points on several timescales.This coherence is interpreted as a correlation coefficient with a value ranging from 0 to 1, where s represents a smoothing parameter.If smoothing is absent, the wavelet coherence is 1.The squared value of the wavelet coherence coefficient ranges from 0 ≤ R 2 ≤ 1.Values near 0 suggest a poor correlation, whereas values near 1 indicate a robust correlation 27 .Thus, wavelet coherence can be considered a useful approach for assessing the association between specific parameters and time.Using the following equation, the coefficients can be computed from the wavelet energy spectrum 28 : where Y and Z are the data sequences; R 2 YZ (s) represents the wavelet coherence coefficient; W i YZ (s) stands for the wavelet cross-spectrum of the data sequence YZ; W i Y and W i Z are the wavelet coefficients for the data sequences Y and Z, respectively; and " < > " indicates the smooth function of the wavelet energy spectrum.Detailed computations were conducted using the wavelet toolbox in MATLAB.

Characteristics of spatial and temporal variations in air pollutant concentrations
Figure 2 illustrates a nearly identical concentration trend between the air pollutants PM 2.5 and PM 10 in Korla from 2016 to 2023, showing an "L" cycle with an annual peak occurring from January to May and a sustained peak in PM 2.5 concentration changes during this period (winter and spring), particularly evident in 2018.The concentration levels consistently peak, with maximum concentration values reaching up to 500 µg•m -3 ; PM 10 also exhibits persistent peaks during the same period, with longer peak hours compared to previous years.In contrast, variations in PM 2.5 and PM 10 concentrations from June to December (summer and fall) show no significant fluctuations, and the pattern of change remains relatively steady.
The eastern section of Korla is primarily composed of the Taklamakan Desert.During spring and summer months, gusty winds prevail, facilitating the transportation of sand and dust particles into the air, resulting in elevated levels of atmospheric PM 29,30 .Higher atmospheric PM levels during winter months have two main contributing factors.First, the prolonged period of winter heating in the city leads to the accumulation of PM from coal combustion, thereby increasing the concentration 30,31,32 .Second, the winter climate in oasis cities located within arid regions is cold and dry, characterized by stable atmospheric structures that promote inversions.These inversions hinder the dispersion and dilution of atmospheric pollutants 33 .
Additionally, the high humidity and precipitation in summer promote the wet deposition of PM, resulting in low PM 2.5 and PM 10 concentrations during this season.Furthermore, the stronger solar radiation in summer enhances atmospheric turbulence movement, contributing to the diffusion and dilution of atmospheric PM 34 .The concentration of O 3 exhibits an inverted "V" pattern, with apex concentration occurring between April and September (C max ≈ 150 µg m −3 ), and the lowest concentration levels observed in January (C min ≈ 20 µg m −3 ).Solar radiation intensity significantly influences photochemical reactions, with variations in air temperature can affecting radiation intensity.Hence, both high temperatures and radiation promote O 3 production 34,35 .
China's air quality standards are based on the WHO air quality guidelines for PM and transitional objectives that measure changes in 24-h concentrations 25 .These standards require Class II areas of the AAFC to adhere to Class II standards.The particulate pollutant PM 10 exceeded the standard in Korla in almost all seasons, and PM 2.5 exceeded the standard in spring.The frequent sand and dust weather during spring suggests that the impact of sand and dust on PM concentrations is significant, highlighting the concern about atmospheric particulate pollution in Korla.Additionally, the air pollutant O 3 exceeded the national secondary standard only on individual days during 2016-2023, indicating that Korla has taken actions to control total pollutant emissions, strengthen pollution management, control emissions from motorized vehicles, adjust the energy use structures, and achieve effective results.Consequently, air pollution in Korla has gradually shifted from mixed sources (natural sand, dust, and soot) to natural sand and dust 1 .

Seasonal characteristics of PM 2.5 , PM 10 , and O 3
The main traffic corridors of North and South Xinjiang, National Highway 218 and National Highway 314, pass around Korla and provide complex sources of PM.PM 2.5 , and PM 10 at different monitoring stations in Korla from 2016 to 2023 were the highest in spring, lowest in summer, and intermediate and comparable in fall and winter, with mean concentration values changing from spring > winter > fall > summer (Table 3).Based on the mean PM 2.5 concentration at the three monitoring stations, it is evident that S2 had the highest concentration, followed by S1 and S3.The highest concentration of 63.8 ± 2.2µg m −3 was observed in spring at station S2, while the lowest value of 23.4 ± 0.6 µg m −3 occurred in summer at station S3.However, pollutant emissions resulting from energy consumption during industrial production in industrial areas contribute more to PM 2.5 than residential and motorized trips 4 .The pollutant PM 2.5 concentrations at each site indicate an overall decrease from 2016 to 2023.This trend was particularly distinguishable in winter and spring.As shown in Table 3, the analysis of seasonal differences (uppercase letters) and interannual differences (lowercase letters) in PM 2.5 at the three monitoring stations revealed that no significant differences were found between spring and winter within the stations and that the differences between summer and fall were not significant; however, significant differences were found between spring and winter and between summer and fall (P < 0.05).There was no significant difference statistically (P > 0.05) in the annual mean PM 2.5 among the study years at each site.The changes in the mean PM 10 concentrations at the three monitoring sites were opposite to those of PM 2.5 , showing S3 > S1 > S2, but all their peaks occurred in the spring, at 243.0 ± 9.0, 221.6 ± 8.4, and 219.5 ± 8.4 µg m −3 , respectively; the minimum value was 85.6 ± 2.7 µg m −3 (in the summer at site S2).However, PM 10 , a type of particulate pollutant, displayed comparable changes to PM 2.5 , also showing significant variance among seasons (i.e., winter, spring, summer, and fall).However, this variation was less noticeable between the years (P > 0.05).As previously mentioned, the study area demonstrates significant seasonal fluctuations in atmospheric particulate pollutants, specifically PM 2.5 and PM 10 .Potential factors contributing to elevated levels during the winter months include the use of coal-fired heating for residential purposes, while the presence of dusty weather in spring may also play a role.The gradual decrease in PM 2.5 and PM 10 concentrations from 2016 to 2023 indicates the significance of the control measures implemented for air pollutants in Korla.www.nature.com/scientificreports/Changes in O 3 pollutant concentration at all sites from 2016 to 2023 indicate a gradual overall increase with a sharper incline in the spring and summer seasons (approximately a 30 µg m −3 increase), whereas the increase during fall and winter is slightly lower than that during the former seasons.Additionally, the monitoring sites exhibited a seasonal distribution of O 3 concentrations, with the highest concentration documented in the summer at site S3 in 2023 (C max ≈ 127.3 µg m −3 ), and the lowest concentration of 33.2 µg m −3 recorded at site S2 in 2016 during the fall and winter seasons.The O 3 concentrations at each site showed a significant difference between seasons (P < 0.05).Interannual differences in O 3 concentrations were observed only between individual years (e.g., significant differences in concentrations between 2016 and other study years at each station; however, there was no significant variability in O 3 concentrations between each year from 2017 to 2023).A study characterizing the seasonal distribution of O 3 concentrations revealed that high temperatures and intense high-temperature radiation during summer, particularly in Korla, which is close to the desert, significantly affect the variation in O 3 concentrations 36 .Conversely, cooler weather and lower radiation levels during winter did not promote O 3 formation.

Wavelet-based analysis of coherence between AQI with PM 2.5 , PM 10 , and O 3
The relationship between AQI and PM 2.5 , PM 10 , and O 3 was examined using WTC.WTC images display time on the horizontal axis and frequency on the vertical axis 27 .The frequency range covered high frequencies (0-16), intermediate frequencies (16-128), and low frequencies (128-512).The WTC spectral intensity was defined using a color-coded scheme (blue and red indicate low and high intensity, respectively).Our analysis focused on the cone of influence (COI), which represents the colored region where K is not affected by edge effects in the wavelet spectrum; in contrast, the region included in the thick line represents a 95% confidence interval relative to the red noise [i.e., significance based on Monte Carlo simulations] [37][38][39][40] .
The arrows in the wavelet coherence plot demonstrate the relationships and causal processes between AQI and PM 2.5 , PM 10 , and O 3 .The left ( ←) and the right ( →) arrows denote positive and negative correlations, respectively 40 .In addition, the rightward and downward (or leftward and upward) directions depict a causeeffect relationship between the first and second parameters, while the rightward and upward (or leftward and downward) directions indicate a cause-and-effect relationship between the second and first parameters 27 .
As shown in Fig. 3a, from 2016 to 2023, AQI and PM 2.5 pollutant levels at site S1 show a positive correlation at various frequencies, including the medium-high (0-64) and low (256-512) frequency ranges.Most arrows depict a rightward and upward trend, indicating that the PM 2.5 pollutant sequence affects the AQI sequence.There is a short-term positive correlation between AQI and PM 2.5 in the 64-128 frequency range from 2018 to 2022.As shown in Fig. 3b, there is a long-term positive correlation between AQI and PM 10 over the range 0-512 (high-low) from 2016-to 2023, with a weaker correlation observed at intermediate-frequencies (16-64) compared with the high and low-frequency ranges.Figure 3c illustrates uninterrupted wavelet coherence between AQI and O 3 at site S1 at low frequencies (256-512).However, a more pronounced hysteresis effect was observed.Furthermore, the wavelet coherence distributions at intermediate-and high-frequency periods (0-128) are sporadic and lack informative value.
Figure 3d displays the WTC results for PM 2.5 and AQI at site S2.A positive correlation is observed, as evidenced by the majority of arrows pointing to the right.Specifically, the high-frequency (0-16) and low-frequency (256-512) ranges show significant coherence.In addition, in the intermediate-frequency (16-64) range, a positive correlation between AQI and pollutant PM 2.5 exists only for short periods, and the correlation is weak (e.g., 2016-2017 and 2020-2021).Over the 2016-2023 period, a positive correlation between AQI and PM 10 in the 0-512 range was observed with a high-to-low trend.This correlation was short-term, isotropic, and limited to the intermediate-frequency range (16-64; Fig. 3e).Furthermore, the robust correlation between AQI and PM 10 during these periods demonstrated that the AQI data accurately portrayed the varied trends of PM 10 pollutants.A significant downward leftward arrow indicates a continuous wavelet negative correlation between AQI and O 3 at S2, specifically in the low-frequency range (256-512; Fig. 3f).
From 2016 through 2023, the majority of arrows in the medium-high (0-64) and low (256-512) frequency ranges indicate a positive correlation between AQI and PM 2.5 at the S3 site and an in-phase relationship between the two variables (Fig. 3h).However, the correlation between AQI and PM 2.5 is less significant in the lowintermediate-frequency range (64-128).Figure 3i shows the cross-correlations between AQI values of S3 and PM 10 .Over the long term (2016-2023), a positive correlation between AQI and PM 10 was shown at various frequencies and scales.This link is notable for frequencies ranging from 0 to 16 (high frequencies) and 128 to 512 (low frequencies).Furthermore, there are indications that AQI and PM 10 (upper and lower right arrows) exhibit mutual causality during these periods.The arrows on both sides are within the range of 16-64 (intermediatefrequency; Fig. 3g), indicating irregular fluctuations in the wavelet coherence between the two.In addition, there is continuous wavelet coherence between AQI and O 3 within the low-frequency (256-512) range; however, there is a notable hysteresis effect between the two (downward arrows on the left).
In summary, wavelet coherence analysis showed that AQI at every monitoring station shared continuous wavelet coherence with both PM 2.5 and PM 10 during various periods and displayed the same phase, indicating a substantial correlation between AQI and the pollutants.However, long-term wavelet coherence between AQI and O 3 is fragmented, and a hysteresis effect is apparent.Our results show that utilizing AQI values to examine ozone fluctuation traits is not suitable.Instead, AQI data efficiently characterize PM 10 fluctuations and underpin reliable data for air pollution management and health risk assessment in the region.
Air pollution levels of PM 2.5 , PM 10 , and O 3 Table 2 and Fig. 4a indicate that between 2016 and 2023, the percentage of days when the air quality of PM 2.5 at the three air monitoring stations in Korla was class I (excellent), accounted for one-third of the annual number of days, suggesting an improvement in regional air quality at these stations, with this trend increasing annually.However, the percentage of days with Class II (good) and Class III (slightly polluted) air quality at the three monitoring stations exhibited an undulating trend.The number of days with regional air quality classes I to III at the three monitoring stations reached over 90% of the total number of days in the year, indicating lower pollution levels by PM 2.5 for residents of Korla.However, the percentage of IV (moderately polluted) and V (heavily polluted) days at site S2 was slightly higher than that at sites S1 and S3, possibly due to its location in an industrial area.
Analysis of the air quality class share of PM 10 from 2016 to 2023 showed that class II days accounted for 50%-60% of the annual number of days at all three monitoring stations, with the share of air quality class II at site S1 exceeding 60% in 2020 (Fig. 4b).During the period 2016-2023, the proportion of Class III air quality days at the three monitoring stations displayed a trend of increasing, then decreasing, and then increasing; however, the proportion of Class I air quality changed in the opposite direction, indicating an improving air quality trend in the three regions of Korla.In addition, the percentage of days with air quality levels IV to VI at the three monitoring stations in Korla ranged from 0 to 15% of the number of days in the year.The percentage of days with moderate and heavy pollution showed a trend of decreasing, then increasing, and then decreasing, while the percentage of days with severe pollution showed a slight decrease from year to year.Moreover, the trend of air quality class percentage of particulate pollutant PM 10 at the three stations is almost the same, indicating stable change in PM 10 in Korla during 2016-2023.Natural sand and dust contribute to an increase in PM 10 , reflecting that natural sand and dust are the main factors causing particulate pollution in the city.Figure 4c indicates that compared with PM 2.5 and PM 10 , the air quality class of O 3 in Korla is good, with air quality maintained at excellent (I) and good (II) levels throughout the year.The maximum percentages of air quality class I for O 3 at the three monitoring stations all appeared in 2016 and were in the order of S1 > S2 ≈ S3.This may be because the S1 site is located in the center of Korla, near the Peacock River, where several O 3 precursors (e.g., some oxygenated organic compounds) can be purified by moisture, indirectly reducing the potential for O 3 production 41 .Although the overall O 3 level in Korla is excellent, the proportion of O 3 class I showed a decreasing trend every year from 2016 to 2023, which may be due to an increase in atmospheric O 3 caused by the volatile organic compounds (VOCs) emitted from industrial activities and motorized vehicles accompanying the rapid development of the city 42 .
When air pollutant levels of PM 2.5 and PM 10 reach levels IV-V, they potentially affect the heart and respiratory system of healthy individuals, significantly exacerbating symptoms in cardiorespiratory patients, reducing exercise tolerance, and inducing symptoms in the general population.At severe pollution levels (class VI), exercise tolerance in healthy individuals is markedly reduced, with noticeable and severe symptoms, leading to the early onset of certain diseases 43,44 .Prolonged exposure to high O 3 concentrations damages the human respiratory system and contributes to the development of eye diseases and other related illnesses 45 .Therefore, relevant city departments should actively implement measures for sand prevention and control, such as adopting the straw checkboard engineering model, implementing policies to convert farmland to forests and grasslands, promoting tree planting initiatives, and establishing a monitoring and early-warning system for sand and dust intensity using modern technologies such as satellite remote sensing, radar, and airborne sounding, etc 46,47 .Simultaneously, these measures should aim to reduce sources of particulate pollutant emissions, decrease PM concentrations, and strengthen management and control measures for O 3 to safeguard residents' health, preserve ecological safety, and optimize the environmental ecosystem.

Conclusion
From 2016 to 2023, the concentrations of PM 2.5 and PM 10 in Korla City exhibit a seasonal distribution of spring > winter > autumn > summer, characterized by low concentrations in summer and autumn and high concentrations in winter and spring.In contrast, O 3 demonstrates a seasonal distribution of summer > spring > autumn > winter, displaying an opposite trend to that of particulate matter concentration.Korla City's proximity to the Taklamakan Desert results in dusty weather during spring, contributing to high atmospheric concentrations of particulate matter.The AQI, PM 2.5 , and PM 10 at each monitoring site in Korla City exhibit continuous wavelet coherence at different periods, indicating significant connections between AQI and pollutants.However, the long-term wavelet coherence between AQI and O 3 is fragmented, with a visible lag effect, making it inappropriate to utilize AQI values to examine the properties of ozone oscillations in Korla.Conversely, AQI effectively defines PM 2.5 and PM 10 fluctuations, providing credible data support for regional air pollution control and health risk assessments.Furthermore, the air quality of O 3 in Korla has consistently been outstanding, with air pollution level ratings of PM 2.5 and PM 10 in the range of I-III accounting for more than 90% of the total number of days in the year, indicating overall satisfactory air quality.However, from 2016 to 2023, the number of days with air pollution level ratings of PM 10 , a particulate matter, at the three monitoring stations reaching Class VI (severe pollution) accounted for 5-10% of the total number of days in the year, highlighting the prominent impact of sand and dust and its potential threat to residents' health.Consequently, relevant city authorities should prioritize strengthening sand prevention and control policies and minimizing particulate emissions.
This study enriches research on urban air pollution in arid land oasis cities in Northwest China.Recent prevention and control efforts of urban pollutants have led to significant improvements in air quality in Korla.Urban planners, ecologists, meteorologists, decision-makers, and other stakeholders in the region must collaborate, and continue to implement proactive measures to manage sand and dust effectively, strengthen sand prevention and control policies, select greenery species with strong anti-pollution capabilities, increase urban green coverage, and reduce PM emissions and concentrations.By doing so, they can effectively reduce and mitigate urban air pollution, thereby enhancing air quality for city residents.

Figure 1 .
Figure 1.Map of air quality monitoring stations in the study area.The figure is created using ArcMap 10.5 (ESRI Co., Ltd.; www.esri.com).

Figure 2 .
Figure 2. Variation trends in air pollutant concentrations in Korla from 2016 to 2021.

Figure 4 .
Figure 4. Percentages of days for each PM 2.5 (a), PM 10 (b), and O 3 (c) pollution level at three monitoring sites in Korla.

Table 1 .
Overview of air quality monitoring stations in Korla.

Table 2 .
Ambient air quality level standards and pollutant concentration limits.

Table 3 .
Seasonal changes in PM 2.5 , PM 10, and O 3 concentrations at different monitoring sites in Korla city during 2016-2021.Mean ± standard error, different capital letters indicate statistically significant differences between seasons, and lowercase letters indicate statistically significant differences between study years (P < 0.05).