Spatiotemporal Distribution, Meteorological Influence, and Potential Sources of Air Pollution over Hainan Island, China

Abstract

Data on particulate matter, gaseous pollutants, and AQI values from three cities (Haikou, Sanya, and Danzhou) between January 2018 and December 2022 were obtained in order to analyze the spatiotemporal distribution characteristics of air pollution, the correlation between pollutants with meteorological conditions, and the potential sources in Hainan Island. The spatiotemporal distribution’s characteristics demonstrated that the annual mean concentrations of SO₂, NO₂, CO, O₃, PM₁₀ and PM_2.5 were 4.34 ± 1.11 μg m⁻³, 9.87 ± 1.87 μg m⁻³, 0.51 ± 0.06 mg m⁻³, 73.04 ± 6.36 μg m⁻³, 27.31 ± 3.63 μg m⁻³, and 14.01 ± 2.02 μg m⁻³, respectively. The yearly mean concentrations were trending downward in the past few years and were below the National Ambient Air Quality Standard (NAAQS) Grade II. Summer was the season with the lowest concentrations of all pollutants (3.84 μg m⁻³, 7.34 μg m⁻³, 0.42 mg m⁻³, 52.80 μg m⁻³, 18.67 μg m⁻³ and 8.67 μg m⁻³ for SO₂, NO₂, CO, O₃, PM₁₀ and PM_2.5, respectively), and afternoons were the time with the lowest concentrations of pollutants (except for 78.04 μg m⁻³ for O₃). The influence of meteorological conditions on pollutants was examined: there was a prominent positive correlation between temperature and O₃ in summer, and relative humidity largely influenced the concentrations of PM. The pollution in Hainan was affected more by regional transport; according to the backward trajectory results, Hainan is susceptible to air masses from Guangdong and Fujian to the northeast, the Indochina Peninsula to the southwest, and the South China Sea to the southeast. The results of PSCF and CWT analyses indicated that Guangdong, Jiangxi, Hunan, and Fujian were the primary potential sources of PM_2.5 and O₃.

1. Introduction

Air pollution has been a significant environmental issue in China over the past few decades due to industrialization, urbanization, and economic expansion, which have raised the risk of sickness [1]. To tackle severe air quality issues, the Chinese Ministry of Environmental Protection amended the Ambient Air Quality Standard in 2013 and increased the number of assessed pollutant items to six (SO₂, NO₂, CO, O₃, PM_2.5, and PM₁₀). Air quality had been improving in China since this standard was enacted [2], but the worsening particulate matter in northern China in winter and the remaining high O₃ concentrations require attention [3]. Recently, numerous studies on the atmospheric environment have been conducted, which have mainly focused on the characteristics of temporal and spatial changes of air pollutants [4], the discussion of the correlation and meteorological driving mechanisms of pollutants [5,6], and the various source analysis of pollutants [7,8]. Based on public air quality data, many researchers have exhaustively reported air pollution in the economically developed Yangtze River Delta [9], Pearl River Delta [10], Beijing-Tianjin-Hebei region [11], as well as the heavily polluted northern and northwestern regions of China [12,13], which has currently attracted great attention of academic research. Less research has been done on Hainan, where the air is relatively clean.

Hainan Island, located far from the Chinese mainland in a tropical climate, is widely recognized for having exceptional air quality. However, mild air pollution has been detected in Hainan recently, attributed to the transboundary transport of pollutants, various meteorological impacts, and anthropogenic emissions [7,14]. The long duration of sunshine combined with intense solar radiation result in a higher O₃ concentration here, which may be threatening to human health [15]. O₃ pollution has become one of the main air pollutants in Hainan due to its frequent occurrence during autumn and winter pollution events [7]. Hainan has comparatively low local pollutant emissions while relatively high background pollution concentrations that influence its air quality. According to reports, 41.6% of Hainan’s PM_2.5 is caused by regional transport from mainland China and neighboring nations in Southeast Asia [16]. The co-pollution of PM_2.5 and O₃ occurs more in relatively clean regions based according to the results of a study on air pollution in 114 Chinese cities [17]. Therefore, it is important to study the current situation and potential sources of air pollutants in Hainan. It is necessary to analyze the air pollutants in less-developed areas over a long period at a high-resolution scale. There are few studies on the six pollutants in Hainan. Some scholars have carried out studies on the characteristics of the changes in the O₃ concentration in Sanya city [18], the PM_2.5 evolution characteristics in Haikou city and its relationship with meteorological factors [19], the spatiotemporal changes in the aerosols over Hainan Island, and the tracking of atmospheric pollutant sources [20]. However, researchers have only studied one or two pollutants and the respective meteorological factors or potential source areas over a short period without carrying out systematic, in-depth, and long-term studies, which made it vital to conduct insightful research on the air pollutants in Hainan.

The prime objectives of this study were as follows: (1) to analyze the characteristics of the annual, seasonal, and diurnal concentrations of and variations in particulate matter and gaseous pollutants in Hainan from 2018 to 2022; (2) to investigate the air pollutants and meteorological factors in Haikou using Pearson correlation analysis to understand the correlation between them in order to reveal the driving mechanism; (3) combined with the concentrations measured from atmospheric stations, to use the potential source contribution factor (PSCF) and concentration weighted trajectory (CWT) methods as the main methods to identify and quantify the potential pollution sources. The calculation of backward trajectories was finished with the help of the HYSPLIT model, and potential sources of PM_2.5 and O₃ in Hainan were identified and quantified using the PSCF and CWT methods. It would be beneficial to test the effectiveness of the implemented regional pollution prevention and control guidelines to provide an important reference value and theoretical and scientific basis for the continuous improvement in air quality.

2. Data and Methods

2.1. Study Area

Surrounded by the sea, Hainan Island is situated in the southernmost section of China. The island is situated on the northern edge of the tropics and experiences a tropical monsoon climate that is warm and humid with abundant rainfall [19]. Its unique geographical environment and climate characteristics affect the spatial and temporal distribution of air pollution, so the atmosphere in the island area not only has the characteristics of an ocean but is also influenced by a land source. Coastal cities, such as Sanya and Haikou, generally exhibit lower levels of gaseous pollutants and particulate matter due to their stronger turbulence and lower anthropogenic emissions [3]. Hainan had 10.27 million permanent residents with a GDP of CNY 681.82 billion at the end of 2022. With a beautiful environment, high vegetation coverage, and rich biodiversity, Hainan is a well-known tourist region and received a total of 60.4 million tourists in 2022 [21].

2.2. Data Sources and Quality Control

We selected three major cities from the National Air Quality Monitoring Center website in Hainan: Haikou, Sanya, and Danzhou. The hourly mean concentration air quality monitoring data (SO₂, NO₂, CO, O₃, PM_2.5, and PM₁₀) over the period of January 2018 to December 2022 (the data for Danzhou were from January 2021 to December 2022) were gathered from the China National Environmental Monitoring Center (data sources can be found in the Supplementary Materials). There were 6 monitoring sites in Haikou, 3 in Sanya, and 2 in Danzhou, most of which were located in urban areas. The precise locations of eleven automatic ambient air monitoring stations are displayed in Supplementary Materials (Figure 1). The China Meteorological Data Service Center provided the surface meteorological data, which included wind direction, speed, daily temperature, relative humidity, and precipitation. The data quality was confirmed to be in compliance with the ambient air quality standards (GB 3095-2012). A validity check was conducted on the data. Invalid values were indicated as those missing or having ≤0 values. The average daily, monthly, and annual concentrations of each pollutant could only be calculated if reliable data were available for more than 20 h, 26 days, and 324 days, respectively, and the 8 h ozone concentration could only be calculated if there were consistent data for at least 6 h out of every 8 h. The concentration from each site was added and divided by the number of sites in the corresponding city to determine the average concentration for each city. Standard deviation is a measure of the dispersion of data, which measures the deviation between a data point and its mean value. The more widely dispersed the data, the higher the standard deviation. The concentration of pollutants in this paper is represented as the mean and standard deviation. The quality assurance and control of the data were carried out in accordance with the People’s Republic of China’s National Environmental Protection Standard (HJ 630-2011), which sets technical criteria for environmental monitoring and quality management. The aforementioned technique has been utilized in several prior investigations to exhibit the dependability of data quality [22,23].

2.3. HYSPLIT

Data were downloaded from the global data assimilation system (GDAS) at the NOAA (National Oceanic and Atmospheric Administration), with a spatial resolution of 1° × 1°. These data were used to calculate the 72 h backward trajectories at a simulated height of 500 m, which were recorded every six hours at 00:00, 06:00, 12:00, and 18:00 (UTC + 8). Integrated with the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT4) model, Yaqiang’s open-source GIS-based MeteoInfo software1.7.6 was utilized to investigate the effects of air transport on PM_2.5 and O₃ in the study area. The 72 h backward trajectories of air mass were identified and clustered according to speed and direction of transmission for analyzing the transport paths of airflows. Utilizing long-term ambient air pollution measurements along with statistical analysis of backward air mass trajectories, TrajStat, a geographic information system (GIS)-based software, was used as an efficient tool for HYSPLIT simulations and has been extensively used in MeteoInfo software. Trajectories could be clustered using the Euclidean distance method, and the appropriate number of clusters could be determined distinctly by total spatial variance (TSV).

2.4. PSCF and CWT

To ascertain the potential sources of PM_2.5 and O₃, based on the results of backward trajectory simulation, a rectangular grid (i, j) with specific resolution was created to cover the study area, and a threshold for pollutant concentration was assigned according to air quality standards. The potential source contribution function (PSCF), a method of identifying potential source areas of high concentrations of pollutants on the basis of a conditional probability function, was adopted to determine the potential source areas of the pollution [24]. The higher the pollutant concentration value of the grid corresponding to the air flow trajectory, the more likely the grid unit is the source area of the air pollution in the target area. The PSCF measures the quantity of pollution trajectory endpoints m_ij that traverse grid (i, j) in relation to the total number of trajectory endpoints n_ij that are present within grid (i, j) [25]. The calculation formula is as follows:

$${\mathrm{PSCF}}_{\mathit{ij}}={\displaystyle \frac{m_{\mathit{ij}}}{n_{\mathit{ij}}}}$$

(1)

In the formula, (i, j) are the longitude and latitude coordinates corresponding to grid points, n_ij is defined as the total number of trajectories passing through grid ij, and m_ij is the total number of trajectory endpoints (the corresponding trajectories are pollution trajectories) obtained when the corresponding trajectories passing through the grid reach the study area and the corresponding pollutants exceed a concentration threshold. By multiplying the weight function W_ij by the entire sample n, the effect of the conditional probability function error is lessened.

$$\mathrm{WPSCF}=W_{\mathit{ij}}·\mathrm{PSCF}$$

(2)

$$W_{\mathit{ij}}=\left\{\begin{array}{c}1.0080\le n_{\mathit{ij}}\\ 0.7020\le n_{\mathit{ij}}\le 80\\ 0.4210\le n_{\mathit{ij}}\le 20\\ 0.05n_{\mathit{ij}}\le 10\end{array}\right\}$$

(3)

In order to overcome the limitation of PSCF’s difficulty in distinguishing between medium and strong sources, concentration weighted trajectory (CWT) was adopted as a remedy, so that a weighted concentration was allotted to each grid cell for quantitatively reflecting the sample concentration of the grid cell’s related trajectories and for further refining the potential pollution sources in the study area. TrajStat can be used to examine the CWT’s contribution to the concentration of air pollutants. The higher the CWT value, the greater the potential contribution of the pollution source to the high level of pollution at the receptor site [26,27].

$${\mathrm{CWT}}_{\mathit{ij}}={\displaystyle \frac{1}{\sum _{l=1}^M{\mathit{\tau }}_{\mathit{ij}1}}}{\sum }_{l=1}^M{C_1\tau }_{\mathit{ij}1}$$

(4)

$${\mathrm{WCWT}}_{\mathit{ij}}={\mathrm{CWT}}_{\mathit{ij}}\times W_{\mathit{ij}}$$

(5)

where C_ij means the average weight concentration of cell grid (i, j); M represents the total number of trajectories; τ_ijl is the time that trajectory l stays in the grid (i, j); C_ij is the average weighted concentration of cell grid (i, j); l is the trajectory; Cl is the concentration of the pollutant corresponding to trajectory l passing through grid (i, j). W_ij is employed to reduce the nondeterminacy of the values in each grid cell, as with the PSCF.

Based on the NAAQS Grade I standard, the criteria for PM_2.5 and O₃ were set at 35 μg m⁻³ and 100 μg m⁻³ as the thresholds in the PSCF and CWT analyses; every grid cell resolution was set at 1° × 1°. Regions with high grid–pixel PSCF and CWT values indicated that they had a greater impact on the study area, and the corresponding region was the potential source of the pollutants affecting Hainan.

3. Results and Discussion

3.1. Characteristics of Air Pollutants

Spatiotemporal Characteristics of Air Pollutants

The mean concentrations of SO₂, NO₂, CO, O₃, PM₁₀, and PM_2.5 were 4.34 ± 1.11 μg m⁻³, 9.87 ± 1.87 μg m⁻³, 0.51 ± 0.06 mg m⁻³, 73.04 ± 6.36 μg m⁻³, 27.31 ± 3.63 μg m⁻³, and 14.01 ± 2.02 μg m⁻³; the five-year averages of the six pollutants in the three cities are displayed in Supplementary Materials. Figure 2 shows the annual variations in the pollutant concentrations from 2018 to 2022, where the annual mean concentrations range from 2.86 to 7.07 μg m⁻³, 6.26 to 13.18 μg m⁻³, 0.39 to 0.59 mg m⁻³, 62.73 to 84.35 μg m⁻³, 20.96 to 34.35 μg m⁻³, and 10.89 to 17.84 μg m⁻³ for SO₂, NO₂, CO, O₃, PM₁₀, and PM_2.5, respectively. The lowest mean concentrations of SO₂ (3.46 ± 0.31 μg m⁻³), NO₂ (8.62 ± 1.57 μg m⁻³), CO (0.46 ± 0.07 mg m⁻³), PM₁₀ (24.61 ± 2.73 μg m⁻³), and PM_2.5 (12.45 ± 1.49 μg m⁻³) were found in Sanya. The mean O₃ concentration in Danzhou was the lowest at 64.8 ± 2.08 μg m⁻³. Danzhou had the highest mean concentration of SO₂ (6.35 ± 0.71 μg m⁻³). Haikou had the highest mean concentrations of NO₂ (11.13 ± 1.67 μg m⁻³), CO (0.54 ± 0.04 mg m⁻³), O₃ (79.01 ± 4.35 μg m⁻³), PM₁₀ (30.08 ± 2.8 μg m⁻³), and PM_2.5 (15.5 ± 1.58 μg m⁻³). The annual mean concentrations of SO₂, NO₂, CO, O₃, and PM₁₀ found in the study area were below the National Ambient Air Quality Standard (NAAQS) Grade I limit. The PM_2.5 concentrations in Haikou and Danzhou slightly exceeded the NAAQS Grade II level (35 μg m⁻³), but they had fallen below the standard by 2022. Five of the air pollutant concentrations in the study area decreased between 2018 and 2022. Particulate matter and NO₂ concentrations decreased more than the others: PM₁₀ (23%, 27%, 13%), PM_2.5 (23%, 25%, 12%), and NO₂ (34%, 44%, 2%) for Haikou, Sanya, and Danzhou, respectively. The concentration of O₃ was stably high, but it had a tendency to fluctuate downward from 2019 onward.

Haikou, the capital of Hainan province, has the most inhabitants in the province, and it produces more pollution from human activities including transportation, logistics, and industrial output. In addition, the monsoons bring pollutants from the north during the winter, which raises the concentration of pollutants [28]. The concentrations of SO₂, CO, and NO₂ are also increased by the emissions of sulfur and nitrogen oxides, of which vehicle emissions have become significant sources [29]. The adoption of environmental protection regulations, such as the promotion of new energy vehicles, was primarily responsible for the decreases in the pollutant concentrations in the study area [30]. The high SO₂ concentration in Danzhou was probably due to the discharge of sulfur-containing fuel from ships [31], and both monitoring stations in Danzhou are located in urban areas. Compared with the other two cities in the study area, Sanya had the best air quality, probably because of its lower air pollution emissions and relatively small share of heavy industry, with tourism being its main industry. Moreover, the mountainous terrain in the center of Hainan Island may affect the movement of pollutants from the north [32].

There were clear seasonal distribution characteristics of the six air contaminants in the three cities as shown in Figure 3: the highest concentration of pollutants was highest in winter, followed by autumn and spring, with the lowest in the summer. The seasonal characteristics of the pollutants in Haikou, Sanya, and Danzhou were essentially similar, with the exception of SO₂ and NO₂.

Danzhou had the highest SO₂ concentration in spring (7.19 μg m⁻³) and the lowest in winter (5.06 μg m⁻³), which was quite different from Haikou and Sanya, whose peaks occurred in winter with lows in spring, possibly due to the higher sulfate, nitrate, and ammonium aerosol concentrations along the shipping routes [33]. The SO₂ concentration in Danzhou needs longer observation time to distinguish its seasonal variation, but its high concentration is an undeniable problem. The concentration of CO displayed a U-shaped seasonal distribution, with the lowest in Sanya (0.42 ± 0.08 mg m⁻³) in summer and the highest in Haikou (0.61 ± 0.03 mg m⁻³) in winter. NO₂ in Haikou was maintained at a high concentration (average at 11.48 μg m⁻³), except in autumn. Sanya had the lowest concentration of NO₂ (7.34 μg m⁻³) in the summer. It is noted that O₃ had a unique seasonal variation trend, with the highest concentration in the winter (86.27 μg m⁻³ for Haikou) and the lowest in the summer (52.80 μg m⁻³ for Danzhou). This was mostly because of two factors: first, there is less precipitation in autumn and winter [34], which slows down the wet deposition of pollutants and possibly keeps them in the atmosphere for longer; second, air convection is weaker in winter [35], resulting in a lower rate of pollutant release. In addition, the highest concentrations were recorded in winter except in Haikou, which reached its peak (95.71 μg m⁻³) in autumn. Volatile organic compounds and nitrogen oxides form a secondary pollutant, ground-level O₃, through a variety of complex photochemical reactions. High temperatures and intense sunlight offer favorable meteorological conditions for these reactions, and the significant emissions of volatile organic compounds during photosynthesis may increase the concentration of O₃ [12]. The winter had the highest concentrations of PM₁₀ and PM_2.5, while the summer had the lowest. Specifically, the summertime PM_2.5 concentration was 8.67 ± 1.31 μg m⁻³ in Sanya, while the highest wintertime PM_2.5 concentration appeared in Haikou (21.55 ± 4.17 μg m⁻³). Similarly for PM₁₀, Danzhou had the lowest concentration (18.67 ± 1.31 μg m⁻³) in summer, and Haikou showed the highest in winter (37.24.55 ± 5.53 μg m⁻³). The wet deposition of abundant precipitation in spring and summer reduces the concentrations of pollutants [35]. Autumn and winter are relatively drier, and the accumulation of particles transported from the northeast of Hainan possibly leads to pollution [5].

Despite the fact that each pollutant’s concentration varied, except for SO₂ and O₃, the diurnal trend in the variation in the concentrations of the other pollutants showed the characteristics of a bimodal pattern, with double peaks and double valleys (Figure 4). The concentration of SO₂ slowly increased at night to reach a peak at about 8:00 and then decreased, with the fastest rate of decline occurring in the afternoon and gradually decreasing to the lowest value of the day in the evening. After increasing steadily from a valley point of 8:00, the concentration of O₃ peaked at 78.04 μg m⁻³ at 16:00. This may have been caused by the highest temperatures and the strongest sunlight occurring in the afternoon [36]. The trends in the variations in NO₂ and CO were the opposite to that of O₃. O₃ chemistry is complicated: reducing NO_X can either increase O₃ (under the NO_X limit condition) or decrease O₃ (under the VOC limit condition) [37]. Around 4:00 was the first valley in Sanya of NO₂ and CO (7.78 μg m⁻³ and 0.45 mg m⁻³), which then increased to around 8:00, which is when the peak concentrations of these pollutants occurred in the three cities (13.9 μg m⁻³ for Haikou and 0.59 mg m⁻³ for Danzhou). The concentrations decreased to the lowest values in Sanya (5.58 μg m⁻³ and 0.50 mg m⁻³) at around 14:00, then increased again in the middle of the night to concentrations slightly below the peak in the morning. The two valleys of particle matter, 21.95 μg m⁻³ and 21.71 μg m⁻³, appeared in Sanya at 7:00 and 15:00; and the peak values of 33.33 μg m⁻³ and 16.67 μg m⁻³ appeared at about 20:00 in Danzhou and Haikou, respectively. The pattern of urban air pollution was a bimodal distribution of particle matter, CO, and NO₂ in front and behind the O₃ peak [38].

3.2. Effects of Meteorological Conditions on Pollutants

The ambient air quality of a given area is profoundly affected by its meteorological conditions, and the diffusion and transport of pollutants in the atmospheric environment are primarily dependent on these conditions. Thus, the relationship between meteorological factors of different orders of magnitude, including temperature (T), relative humidity (RH), precipitation (P), wind direction (WD), and wind speed (WS), and pollutant concentration, was investigated in order to further explore their mechanisms (Figure 5 and Figure 6). The provincial capital of Hainan, Haikou, was selected as the representative city. Daily scale data were utilized, with a sample size of 1826 days. The analysis of the correlations between air pollutants and meteorological parameters is displayed in

Table 1. Correlations of the pollutants and the meteorological factors based on the daily data in four seasons (** p < 0.01; * p < 0.05).

Meteorological Factor	Season	SO2	NO2	CO	O3	PM10	PM2.5
T	Spring	−0.15 **	0.19 **	−0.40 **	−0.22 **	−0.27 **	−0.24 **
	Summer	0.06	−0.08	−0.34 **	0.39 **	0.39 **	0.33 **
	Autumn	−0.34 **	0.13 **	−0.08	−0.30 **	−0.25 **	−0.20 **
	Winter	−0.43 **	−0.30 **	−0.37 **	−0.03	−0.17 **	−0.32 **
P	Spring	−0.04	0.12 *	0.03	−0.08	−0.22 **	−0.21 **
	Summer	−0.01	−0.05	0.17 **	−0.14 **	−0.23 **	−0.13 **
	Autumn	−0.24 **	−0.24 **	−0.13 **	−0.26 **	−0.31 **	−0.28 **
	Winter	−0.14 **	−0.07	0.06	−0.20 **	−0.28 **	−0.20 **
RH	Spring	−0.23 **	−0.18 **	0.28 **	−0.13 **	−0.20 **	−0.14 **
	Summer	−0.16 **	−0.06	0.30 **	−0.46 **	−0.44 **	−0.34 **
	Autumn	−0.72 **	−0.31 **	−0.28 **	−0.62 **	−0.64 **	−0.57 **
	Winter	−0.58 **	−0.10 *	0.09	−0.49 **	−0.39 **	−0.30 **
WS	Spring	−0.18 **	−0.40 **	−0.10 *	−0.21 **	−0.08	−0.16 **
	Summer	−0.17 **	−0.52 **	−0.14 **	−0.21 **	−0.21 **	−0.22 **
	Autumn	0.05	−0.47 **	−0.03	0.06	0.06	−0.04
	Winter	0.04	−0.22 **	0.11*	−0.21 **	−0.13 **	−0.19 **

.

The photochemical reaction rate accelerated in the summer and fall due to the more intense solar radiation, longer daylight hours, and higher temperatures. These may have increased the amount of sulfate and nitrate particles in the atmosphere, which become secondary sources of PM_2.5 and PM₁₀. Simultaneously, an increased amount of particulate matter is carried into the atmosphere from the surface, serving as a primary source of PM_2.5 and PM₁₀ [39]. Under high-temperature conditions, diffusion is accelerated by atmospheric cross-ventilation and turbulence [40,41]. As shown in Table 1, there was a strong negative correlation between T and SO₂, NO₂, CO, and PM_2.5, suggesting that the higher the temperature, the lower the concentrations of these pollutants, while a strong positive correlation between O₃ and T was found in the summer, indicating that T influenced the O₃ concentrations through changes in mass fluxed through the upper boundary layer and gas phase chemistry, as well as vertical diffusivity and RH [42]. Below 26 °C, the concentrations of SO₂, NO₂, CO and PM_2.5 were significantly higher than those at 26 °C. In winter, when the T was between 11 °C and 15 °C and the humidity was about 65%, the PM₁₀ concentration exceeded 100 μg m⁻³. Because the ground was cooler in winter, the increase in temperature led to a thermal inversion layer, which prevented atmospheric movement and could have led to the accumulation of pollutants [43].

There was a substantial positive correlation in the summer between T and O₃. At T values over 20 °C, the concentration of O₃ rose to 140 μg m⁻³ or higher. O₃ was negatively correlated with RH, which means that the low RH in the fall and winter may have boosted O₃ concentrations. Combined with the observations of O₃ and T, we concluded that a high T and low RH are favorable meteorological conditions for the generation of O₃. However, too high a WS value may dilute O₃ precursor compounds, thus limiting the formation of O₃ [44]. Particles suspended in the air can be wet-deposited by precipitation to reduce the concentration of air pollutants, and the solubility of SO₂ and NO₂ enables them to be washed and dissolved by precipitation [45], resulting in a negative correlation between precipitation and particulate matter throughout the year. Non-negligible negative correlations among RH and SO₂, O₃, and particulate matter were discovered, particularly in autumn. The formation of secondary aerosols can be enhanced under high RH, which may lead to an increase in PM concentrations, especially in autumn and winter, thus reducing O₃ concentrations [13,46]. In addition, the decrease in visibility at high RH and the decreases in photochemical activities suppress the formation of O₃ [47]. It was shown that the aerosol content of the El Niño–Southern Oscillation (ENSO) is higher over China in winter, proving the existence of a positive correlation between air temperature and aerosol loading, which may lead to an increase in the related air pollutants [48]. For reinforcing the processes of diffusion and transport, conductive near-surface wind is needed [49]. A substantial negative correlation exists between WS and NO₂, suggesting the ventilation and turbulence of pollutants can be enhanced by a high WS, thereby improving air quality [50].

3.3. Backward Trajectory Analysis

Figure 7 shows the statistical results of the daily average AQI values of the three cities in Hainan province from 2018 to 2022. On the basis of the daily average AQI, three grades of air quality could be distinguished: excellent, good, and mild pollution. As can be seen in Figure 7, during the study period, the proportion of the air quality conditions in the three cities was excellent > good > mild pollution. Danzhou had the best air quality, with only 0.278% of the study period having mild pollution, 11.8% of the period having with good air quality, and 87.9% of having excellent air quality. The air quality in Sanya was second, with 86.2% and 13.2% of the study period having excellent and good air quality, respectively. Among the 1824 days sampled, only 12 days were mildly polluted. The number of days with excellent air quality in Haikou was 76.2%, while the number of days with mild pollution was 2.75%.

The main air pollutants were O₃, PM_2.5, and PM₁₀. In Haikou and Danzhou, O₃ pollution accounted for approximately 85% of the total pollution days, and the proportion of PM_2.5 pollution days was approximately 7–8%. The O₃ pollution in Sanya accounted for up to 96%, suggesting the seriousness of the O₃ pollution in Sanya. Therefore, it was necessary to study the trajectories and potential sources of the air masses of the major pollutants, which could provide a reference for pollutant abatement measures.

The 72 h backward trajectories from 2018 to 2022 were computed and clustered using MeteoInfo software. It is evident that the outcomes of air flow clustering varied noticeably. Figure 8 displays the distribution of the backward trajectory clusters in Haikou (a), Sanya (b), and Danzhou (c). In Haikou, three different directions were noted from the clusters during the study period. Cluster 1 accounted for 62.23% of the air mass, originating from the southeast coast of Fujian and traveling to eastern Guangdong. Cluster 2 (28.65%) originated in the northern South China Sea. Cluster 3 (9.11%) mainly came from Southeast Asia, originating in southern Thailand and being transported across Laos and Vietnam. Four clusters came from three directions in Sanya. Cluster 1 and cluster 4 were mainly from the north, which accounted for 50.19% of the air mass. Cluster 1 (35.02%) originated in the southeast coast of China and passed through the Taiwan Strait. Cluster 4 (15.17%) came from central Hunan and across Guangdong. As for the south side, cluster 2 (24.27%) originated in the eastern South China Sea. Cluster 3 (25.53%) came from the Gulf of Thailand and traveled through Vietnam and Cambodia. In Danzhou, cluster 1 (53.53%) originated in northern Fujian and traveled to eastern Guangdong. Cluster 2 (22.57%) originated in the center part of the South China Sea. Cluster 3 (23.9%) came from the Gulf of Thailand and traveled along the Indochina Peninsula.

The shorter air clusters suggested that the air masses flowed slowly and pollutants built up [51]. Studies showed that the lower PM_2.5 concentrations during COVID-19 were mainly associated with lower emissions from transport and industry, but long-distance transport and secondary pollution offset the reduction in emissions during the pandemic. O₃ concentrations increased as a result of a weakening of the titration action of NO, despite the fact that VOC emissions decreased during the severe lockdown period [10,52]. The air mass from the southwest is influenced by straw burning and wildfires in Southeast Asia, and biomass burning possibly negatively affected the air quality in Hainan through transport [53,54]. The influences of airflow on the concentration of air pollutants in Hainan was dominated by the northeast direction, followed by the southwest and southeast, which highlights the significance of enhancing regional environmental cooperation and putting cooperative air pollution prevention and control measures into place.

3.4. PSCF Analysis

The northeastern portions of Hainan hosted the majority of the potential sources with high WPSCF values (Figure 9a–c). For Haikou, the northeast (Guangdong, eastern Guangxi, eastern Hunan, northwest Jiangxi, and eastern Hubei) had high WPSCF values (Figure 9a). Since long-distance transport was the primary contributor to air pollutants, these observations agreed adequately with the backward trajectory analysis. Southern Hunan and the north-central part of Guangdong were the primary potential sources for Danzhou, having high WPSCF values. These places were located along the route of the dominant cluster 1 (Figure 8c), which had elevated PM_2.5 levels. According to cluster 1’s dominance (Figure 8b), the likely source regions for Sanya that had high WPSCF values were concentrated in Guangdong and southern Hunan.

The high-value WPSCF areas for O₃ in Haikou were concentrated in Jiangxi and western Fujian; the high-value WPSCF area for ozone in Sanya included southern Hubei, Hunan, and Guangdong; and the O₃ high-value WPSCF areas in Danzhou included small areas in central Hunan and northern Vietnam. The WPSCF area covered by O₃ in Danzhou was larger than that covered by PM_2.5, but the scope of the high-value area was smaller, and the pollution situation was less serious than that in Haikou and Sanya. Haikou not only had the highest WPSCF for PM_2.5, but the coverage area of the high-value WPSCF area was larger, indicating more serious pollution. It can be seen that although the air flow was still dominated by the northeast, the WPSCF coverage area for O₃ extended to the southeast more than that of PM_2.5 in the same region, which indicated that O₃ was produced by a wider range of potential sources. This finding demonstrated that long-range transport is an important channel for pollution sources in Hainan [55]. The O₃ levels were connected to the numerous, widely scattered industrial facilities in these northern regions, particularly the Pearl River Delta, which are primarily responsible for the expansion of production spaces, consume large amounts of energy, and emit air pollution [56]. Furthermore, economic growth and population expansion in the Guangdong–Hong Kong–Macao Greater Bay Area (GBA) will keep increasing the pressure to protect resources and the environment [57]. In view of the significant influence of regional pollution transport on the atmospheric environment in the study area, it is necessary to jointly control regional air pollution.

3.5. CWT Analysis

In general, higher WPSCF value coverage is consistent with corresponding regions of higher WCWT values. As shown in Figure 10, Guangdong, southern Hunan, Jiangxi, and western Fujian (WCWT > 20 μgm⁻³) were identified as potential pollution source areas of Haikou, and the areas were large and joined together. The potential source areas of Sanya were in southern Hunan and Guangdong, and the source area was smaller than that of the others. The high WCWT values for Danzhou were located in Guangdong, southern Hunan, and southern Jiangxi.

High WCWT values were observed in southern Jiangxi, proving that long-distance transportation had an impact on PM_2.5 pollution in Hainan during the study period. Although their coverage was larger than the WPSCF results in the study area, the potential source region with larger WCWT valued better aligned with the trajectory clustering of Hainan and the high-WPSCF area. The WCWT values for O₃ were ranged as Haikou > Sanya > Danzhou, which is consistent with the calculation result of the annual mean O₃ concentration. The WCWT coverage for O₃ was almost coincident with that of PM_2.5. The high-WCWT-value coverage for Haikou was concentrated in Jiangxi, Guangdong, and Fujian; that of Sanya covered southern Fujian, Jiangxi, and Guangdong; while that of Danzhou was affected by eastern Guangxi, southern Hunan, southern Fujian, and the Taiwan Strait.

Discrepancies existed between the WCWT and WPSCF results in Hainan; different from the northeast direction, where the high-value coverage was located, the low-value coverage of the WCWT largely extended in the southwestern direction. One of the potential reasons for this is the dominant wind directions of the monsoon climate in Hainan are the northeast and southwest [58]. Even so, because the coincidence of the high-WPSCF-value and high-WCWT-value coverage was fine and relatively concentrated, it could be inferred that Guangdong, Jiangxi, Hunan, and Fujian were the main potential source regions of PM_2.5 and O₃ in Hainan. Particulate matter, O₃, and its precursors were carried by air currents to the study area under the control of the wind direction [28]. In particular, the photochemical reaction of O₃ and its precursors is promoted under favorable meteorological conditions in Hainan [6], so O₃ pollution is particularly prominent in Hainan.

The results demonstrated that the potential source areas of PM_2.5 and O₃ were mainly in the southeastern provinces of China (mainly in Hunan, Jiangxi, and Guangdong). It was suggested that a significant decrease in NO_X should help Hainan control PM_2.5 and O₃ pollution [28]. Researchers predicted that PM_2.5 emissions are inversely correlated with precursor emissions, including ammonia (NH₃), VOCs, NO_X, and SO₂. The average increase in O₃ was driven by a lower reduction in NO_X emissions; although a higher reduction may have mitigated the increase, caution should be taken while implementing reductions in NO_X emissions [59].

4. Conclusions

The spatiotemporal distribution characteristics of six monitored pollutants (SO₂, NO₂, CO, O₃, PM₁₀, and PM_2.5), the relationship between meteorological conditions and the concentrations of pollutants, in addition to potential source regions of PM_2.5 and O₃, were investigated in Hainan. During 2018 to 2022, the annual mean concentrations of SO₂, NO₂, CO, O₃, and PM₁₀ found in the study area were below the NAAQS Grade I level. The PM_2.5 concentrations in Haikou and Danzhou slightly exceeded the NAAQS Grade II level (35 μg m⁻³). The levels of the six pollutants significantly decreased, with the exception of O₃. From a seasonal perspective, the highest concentration of most of the pollutants peaked in winter, followed by autumn and spring, with the lowest in summer. The diurnal distribution displayed a bimodal pattern, with double peaks and double valleys of concentrations, except for O₃, which was low in the morning and high in the afternoon. RH was closely correlated with SO₂, O₃, and particulate matter, and WS largely influenced the concentrations of NO₂. There was a prominent positive correlation between temperature and O₃ in summer, while negative correlations between temperature and SO₂, NO₂, CO, and PM_2.5 were found. The results provide an example and background for similar studies, which may be beneficial for determining the root causes of air pollution and constructive for developing pollution control policies, such as further-refined pollution control measures for individual industries with prominent emissions. At the same time, it instigated the following thought: on the basis of the original air quality standards, clean-air areas like Hainan should prepare and implement a set of more stringent standards, refine the indicators of each type of pollutant, and add some new air pollutants that have received widespread attention in recent years. Moreover, the complex physicochemical relationships between pollutants demand coordinated emission reduction measures to be implemented in the pursuit of a more environmentally friendly effect, highlighting the importance of the joint control of regional air pollution. For example, in terms of the regional transport of pollutants, friendly consultations with Southeast Asian countries have been conducted to reduce emissions by reducing straw burning and controlling forest fires. We have reached consensus on environmental air protection with the southeast coast of China, especially the Guangdong–Hong Kong–Macao Greater Bay Area, which strictly controls industrial pollution in accordance with emission standards. These measures are not only beneficial to Hainan but also to ensuring clean air in the pollution source areas.