Identifying urban emission sources and their contribution to the oxidative potential of fine particulate matter (PM_2.5) in Kuwait

Abstract

In this study, we investigated the seasonal variations, chemical composition, sources, and oxidative potential of ambient PM_2.5 (particles with a diameter of less than 2.5 μm) in Kuwait City. The sampling campaign was conducted within the premises of Kuwait Institute for Scientific Research from June 2022 to May 2023, covering different seasons throughout the year. The personal cascade impactor sampler (PCIS) operated at flow rate of 9 L/min was employed to collect weekly PM_2.5 samples on PTFE and quarts filters. These collected samples were analyzed for carbonaceous species (i.e., elemental and organic carbon), metals and transition elements, inorganic ions, and DTT (dithiothreitol) redox activity. Furthermore, principal component analysis (PCA) and multi-linear regression (MLR) were used to identify the predominant emission sources and their percentage contribution to the redox activity of PM_2.5 in Kuwait. The results of this study highlighted that the annual-averaged ambient PM_2.5 mass concentrations in Kuwait (59.9 μg/m³) substantially exceeded the World Health Organization (WHO) guideline of 10 μg/m³. Additionally, the summer season displayed the highest PM_2.5 mass concentration (75.2 μg/m³) compared to other seasons, primarily due to frequent dust events exacerbated by high-speed winds. The PCA identified four primary PM_2.5 sources: mineral dust, fossil fuel combustion, road traffic, and secondary aerosols. The mineral dust was found to be the predominant source, contributing 36.1% to the PM_2.5 mass, followed by fossil fuel combustion and traffic emissions with contributions of 23.7% and 20.3%, respectively. The findings of MLR revealed that road traffic was the most significant contributor to PM_2.5 oxidative potential, accounting for 47% of the total DTT activity. In conclusion, this comprehensive investigation provides essential insights into the sources and health implications of PM_2.5 in Kuwait, underscoring the critical need for effective air quality management strategies to mitigate the impacts of particulate pollution in the region.

Graphical Abstract

1. Introduction

Airborne particulate matter (PM) has been linked to several severe health effects, including cardiovascular diseases, respiratory illnesses, lung cancer, and increased mortality, even at relatively low concentrations and short-term exposure (Aldekheel et al., 2023a; Balluz et al., 2007; Pope and Dockery, 2006). Additionally, long-term exposure to high concentrations of particulate matter in the atmosphere can significantly decrease the average life expectancy by a year or more, mainly due to cardiopulmonary mortality and lung cancer (WHO, 2004). The PM_2.5 fraction of airborne particulate matter, defined as particles with an aerodynamic diameter of less than 2.5 μm, accounts for approximately 96% of the particles that exist in the pulmonary parenchyma and can penetrate to the alveoli region (Churg and Brauer, 1997; Feng et al., 2016; Pinkerton et al., 2000). PM_2.5 exposure has also been linked to systemic oxidative stress and cellular damage in humans and animals due to environmentally persistent free radicals, especially in combustion-generated particles, and organic compounds that can produce intracellular reactive oxygen species (ROS) (Gehling et al., 2014; Gehling and Dellinger, 2013; Kouassi et al., 2009; Longhin et al., 2013; Torres-Ramos et al., 2011). Given these health concerns, it is necessary to identify primary PM_2.5 sources in urban areas in order to develop effective mitigation strategies to regulate ambient levels and protect public health.

Kuwait, a small country in the southwestern part of Asia and at the northeastern edge of the Arabian Peninsula, is characterized by its desert environment and extremely hot and arid climate. Kuwait is experiencing unprecedentedly high ambient air temperatures, with the Mutriba area reaching 54°C in 2016, which is the highest recorded temperature in Asia and the third highest in the world in the last 76 years (Alahmad et al., 2019; Merlone et al., 2019). In addition to the elevated temperatures, dust is a common phenomenon in the region due to frequent dust episodes caused by strong winds carrying loose sand and dirt from dry surfaces (AL-Harbi, 2015; Yassin et al., 2018). Previous studies have shown that Kuwait experienced high levels of PM originating from local and regional anthropogenic and natural sources, with daily PM_2.5 and PM₁₀ concentrations potentially exceeding 300 μg/m³ and 2000 μg/m³, respectively, in extreme dust episodes (Alahmad et al., 2021; Al-Hemoud et al., 2018; Brown et al., 2008). According to a comprehensive global air quality report published by the World Health Organization (WHO) in 2016, the average annual PM_2.5 mass concentration in Kuwait was 75 μg/m³, exceeding several neighboring countries, including Iraq (50 μg/m³), Oman (48 μg/m³), Bahrain (60 μg/m³), United Arab Emirates (64 μg/m³), and Iran (42 μg/m³) (WHO, 2016). In the last four decades, Kuwait has experienced remarkable socio-economic growth, driven by the construction of extensive urban freeways and bridges as well as the development of the industrial infrastructure (Al-Awadhi, 2014). This expansion necessitates the implementation of effective regulatory strategies focusing on environmental concerns, particularly the high levels of particulate pollution, to prevent environmental and health deterioration in the region.

Multivariate factor analysis models, including Principal Component Analysis (PCA), UNMIX, and Positive Matrix Factorization (PMF) have been widely utilized for identifying emission sources of particulate pollutants (Deng et al., 2018; Shi et al., 2014; Wang et al., 2012). Several previous research studies have combined the results of PCA analysis with multi-linear regression (MLR) in order to identify the main emission sources and their percentage contribution to target PM species in different regions around the globe (Argyropoulos et al., 2016; Badami et al., 2023b; Deka et al., 2016; Guo et al., 2016; Taghvaee et al., 2019; Ul-Saufie et al., 2013). To date, only two source apportionment studies have been conducted in the state of Kuwait, aiming to identify the potential sources of ambient PM_2.5 (Alahmad et al., 2021; Alolayan et al., 2013). These two studies have reported a wide range of PM_2.5 sources in Kuwait City (Table S1), with soil dust accounting for the majority (54%) of PM_2.5 mass during 2004-2005, and regional pollution being the major contributor (44%) during 2017-2019. Given the limited information regarding PM_2.5 sources in Kuwait, this study provides a comprehensive analysis of the potential emission sources and their contributions to the ambient PM_2.5 oxidative potential. Furthermore, to the best of our knowledge, there has been no prior research conducted to investigate the toxicological characteristics of ambient particles in Kuwait.

The objective of this study is to explore the chemical and toxicological properties as well as potential sources of ambient PM_2.5 in an urban site located in Kuwait City, Kuwait. PM_2.5 samples were collected from June 2022 to May 2023, covering all seasons. The collected samples were analyzed after each season to determine the PM chemical composition (i.e., metals, inorganic ions, carbonaceous species) and oxidative potential (i.e., redox activity). We then employed PCA analysis coupled with MLR to study the contribution of pollution sources to the PM_2.5 oxidative potential in Kuwait.

2. Methods

2.1. Description of the sampling location

Our sampling site, illustrated in Figure S1, is located within the premises of Kuwait Institute for Scientific Research (KISR) in Kuwait City, which is the capital of Kuwait. Kuwait City is the center of the country and includes residential and industrial areas, governmental offices, and numerous private-sector corporations. Given the central location and the close proximity to major residential areas where the majority of the population resides, our sampling site is representative of population exposure to a wide range of pollutants originating from nearby emission sources, including Doha east and west power stations (10 km) and Shuwaikh desalination plant (4 km). In addition, the site is located at a distance of 600 m from Highway 85, which connects the far west residential areas to downtown Kuwait on the east side and serves primarily for workers commuting between their homes and workplaces. Previous studies have conducted their sampling campaigns in the heart of Kuwait City due to its representation of diverse urban pollutant sources and its high population density (Al-Awadhi, 2014; Alolayan et al., 2013; Brown et al., 2008).

2.2. Sampling campaign and meteorological conditions

We collected PM_2.5 samples on a weekly basis, specifically on weekdays, throughout different seasons to provide a comprehensive representation of the entire year (i.e., June 2022 to May 2023). Since the fall season is very short in Kuwait, spanning only four weeks from early November to early December, it was challenging to analyze the seasonal trend within this short period. Therefore, the fall period was considered as a part of the winter season in this study, with a sampling duration spanning from early November 2022 to late January 2023. The sampling campaign of the summer season was carried out from June to July 2022, while the spring period samples were collected from March to May 2023. The meteorological parameters, including temperature, relative humidity, and wind roses (i.e., wind speed and direction) were obtained during our sampling campaign and presented in the supplementary material.

2.3. Instrumentation

We used the personal cascade impactor samplers (PCIS) (Model 225-370, SKC Inc., Eighty Four, PA, USA) (Misra et al., 2002; Singh et al., 2003) operated at a flow rate of 9 L/min and equipped with 2.5 cut-point stage to collect PM_2.5 samples. We employed two samplers simultaneously and connected in parallel to collect particles on both 37 mm PTFE (Pall Life Sciences Inc., Ann Arbor, MI, USA) and quartz (Whatman Company, Marlborough, MA, USA) filters. To minimize particle bouncing and re-entrainment, we applied grease to the impaction surface of the 2.5 μm stage to capture coarse particles and prevent them from bouncing to the subsequent stage. This enabled the collection of fine particles (d_p < 2.5 μm) in the after-filter stage, where the PTFE or quartz filters were placed. Before starting the field sampling, both PTFE and quartz filters were maintained under standard laboratory conditions, specifically a temperature range of 22 - 24 °C and relative humidity of 40 - 50%, in order to achieve equilibration and subsequently determine their pre-sampling weights using Mettler 5 microbalance (MT5, Mettler Toledo Inc., USA). After completing the sampling of each season, the filters were reweighed to obtain the collected PM_2.5 mass on each filter by subtracting the pre-sampling from the post-sampling weight.

2.4. Chemical and toxicological analysis

The collected PM_2.5 samples were sent to the Desert Research Institute (DRI) in Reno, Nevada, for comprehensive analysis of the chemical composition, including metals and trace elements, carbonaceous species, and inorganic ions. The samples were analyzed for the content of metals and trace elements using inductively coupled plasma mass spectroscopy (ICP-MS), employing a hot block acid digestion method to extract particles from the PTFE filter into an acidic solution comprising a mixture of nitric (HNO₃) and hydrochloric (HCl), and hydrofluoric (HF) acids. Post-digestion, the samples were diluted with deionized water, aerosolized, and introduced to the ICP-MS instrument, as detailed by Herner et al. (2006). In addition, the inorganic ion content was measured using ion chromatography (IC), which is a process that involves particle extraction into ultrapure deionized water via sonication, filtration of the liquid solution, and subsequent quantification of ion concentrations, as elaborated by Karthikeyan and Balasubramanian (2006). Moreover, the collected quartz filters were analyzed to obtain the concentration of organic carbon (OC) and elemental carbon (EC) using DRI multiwavelength thermal/optical carbon analyzer (Magee Scientific, Berkeley, CA, USA). This instrument employs a dual-optical system and a multi-stage thermal process to separate and measure different forms of carbon. Further details regarding the design, calibration, and operation of the multiwavelength thermal/optical carbon analyzer can be found in Chen et al. (2015). Sections of the PTFE filters were also sent to the Illinois Lab for Aerosol Research at the University of Illinois Urbana-Champaign for performing the toxicological analysis. Dithiothreitol (DTT) assay is one of the most common and well-established techniques employed to quantify the oxidative stress induced by redox-active PM species and has been extensively used in previous aerosol research studies (Aldekheel et al., 2023b; Badami et al., 2023a; Cho et al., 2005; Gao et al., 2017; Patel and Rastogi, 2018; Verma et al., 2009). First, each PM_2.5 sample was extracted in ultrapure water, followed by the addition of the reducing agent (DTT) and the incubation at 37 °C in potassium phosphate buffer. During the incubation, the PM redox-active compounds in the sample reacted with DTT, leading to the oxidation of DTT into its disulfide form. The linear rate of DTT consumption in the filter extract is proportional to the oxidative potential of redox species present in the aerosol sample, which can generate reactive oxygen species and damage the biological systems. Further detailed information related to the methodology of DTT assay is available in (Cho et al., 2005; Kumagai et al., 2002)

2.5. Principle component analysis (PCA) and multi-linear regression (MLR)

PCA is a statistical tool utilized to simplify complex datasets by reducing the number of variables and identifying patterns and correlations in the data to create new variables (principal components). In the current study, we used Statistical Package for Social Science (SPSS) software (version 25) to perform the PCA analysis using the mass concentrations of EC, OC, metals (i.e., Si, Fe, Al, Mg, Ca), and inorganic ions (i.e., sulfate, ammonium). PCA groups highly correlated chemical species into different components, inferring that these species originate from the same specific emission source of PM_2.5. The varimax orthogonal rotation was employed to improve the fit and interpretability of PCA results by adjusting the factor loadings to ensure that each principal component predominantly corresponds to a single factor (Dallarosa et al., 2005). An eigenvalue greater than 1 was utilized as a threshold to determine the inclusion of potential source factors in the analysis (Argyropoulos et al., 2016). In order to ensure the suitability of the data for the PCA analysis and the reliability of source factors, the Kaiser-Meyer-Olkin (KMO) value was set to be above 0.7 (Altuwayjiri et al., 2022; Argyropoulos et al., 2016). Before performing the PCA analysis, the mass concentrations were standardized using equation 1 to ensure that all variables are on a similar scale and can be compared fairly:

$$Z_{ij}=\frac{C_{ij}-C_j}{{\sigma }_j}$$ (1)

where ${\mathrm{Z}}_{\text{ij}}$ represents the standardized dimensionless value of the jth species in the ith sample, $\text{Cij}$ is the mass concentration of the jth species in the ith sample, $\text{Cj}$ refers to the mean mass concentration of species $\mathrm{j}$, and $\sigma \mathrm{j}$ represents standard deviation of species $\mathrm{j}$. After data standardization, the PCA analysis was performed based on the following equation:

$$Z_{ij}=\sum _{k=1}^P{g}_{ik}{f}_{kj}+e_{ij}$$ (2)

Where $\mathrm{P}$ is the number of source factors in the analysis, ${\mathrm{f}}_{\text{kj}}$ is the loading of the jth species on the source, ${\mathrm{g}}_{\text{ik}}$ represents the contribution of the source in the ith sample (factor score), and ${\mathrm{e}}_{\text{ij}}$ is the unexplained residual that is not captured by the principal components. After identifying the source factors, we used the MLR approach to quantify the contribution of each emission source to the oxidative potential of PM_2.5 during the investigated period (Zuo et al., 2007). The obtained factor scores were used as independent variables in a multi-linear regression analysis, with the extrinsic DTT redox activity as the dependent variable. The standardized regression coefficients (Beta) and the derived regression coefficient (R²) were then used to determine the proportional contributions from various sources to the PM_2.5 redox activity.

3. Results and Discussion

3.1. Seasonal variations in PM_2.5 mass concentrations and chemical composition

3.1.1. Mass concentrations and carbonaceous species

Figure S3 illustrates the measured mass concentrations of ambient PM_2.5 during the summer, winter, and spring seasons in Kuwait. The results indicate noticeably higher PM_2.5 mass concentration (75.2 ± 8.5 μg/m³) during the summer compared to the winter period (60.1 ± 10.8 μg/m³), mainly due to the substantially elevated summer concentrations of crustal elements originating from soil dust (Nava et al., 2012; Zhao et al., 2006). During our summer sampling campaign, we encountered frequent dust events, which is a common occurrence in the region during the summer season (Al-Hemoud et al., 2018; Li et al., 2020). The meteorological department of DGCA in Kuwait reported a higher mean wind speed during the summer season (6.1 ± 2.3 m/s) compared to the spring (4.1 ± 1.2 m/s) and winter seasons (3.3 ± 0.9 m/s). The combination of strong summer winds, known as Shamal, and arid surface conditions facilitated the resuspension of dust from the desert areas (Alahmad et al., 2021; AL-Harbi, 2015; Al-Hemoud et al., 2022; Li et al., 2020; Parolari et al., 2016). The ambient PM_2.5 mass concentration in the spring season (44.5 ± 10.8 μg/m³) was lower than both summer and winter seasons. This resulted from a combination of factors: the spring period lacked the same frequency of dust episodes seen in the summer months, and it also experienced higher temperature and atmospheric mixing height compared to the winter months, which led to the reduction of PM_2.5 concentrations in the spring season. Moreover, the annual average PM_2.5 mass concentration (59.9 μg/m³) was substantially higher than the air quality guidelines set by the World Health Organization (WHO), which recommended an annual mean PM_2.5 concentration of 10 μg/m³ (WHO, 2006). Our reported PM_2.5 mass concentration has also exceeded the national ambient air quality standards (NAAQS) established by the US Environmental Protection Agency (EPA), which specified an annual mean concentration of 12 μg/m³.

The mass concentrations of OC and EC during different seasons are demonstrated in Figure 1. It is evident from the graph that the lowest EC concentration was observed during the summer season (1.79 μg/m³), followed by spring (2.26 μg/m³) and winter (3.36 μg/m³). Considering that EC primarily originates from transportation activities, particularly the exhaust emissions of vehicles due to incomplete fuel combustion (Cyrys et al., 2003; Keuken et al., 2012), the decrease in its concentration during the summer season can be mainly attributed to reduced traffic volume. This reduction in traffic was primarily due to the closure of schools and the fact that the majority of the population chose to leave the country during extremely hot summer months (i.e., June – August) (Al-Awadhi, 2014). During winter, the concentrations of EC increased as a result of enhanced atmospheric stability conditions that lead to a decrease in the boundary layer height, restricting the vertical dispersion of pollutants and increasing their concentrations (Patel et al., 2021; Schwartz et al., 2018). Moreover, the OC concentrations have also shown the same seasonal trend, with values of 5.18 μg/m³, 5.51 μg/m³, and 5.77 μg/m³ during summer, spring, and winter periods, respectively. The seasonal variation in OC concentrations is less pronounced compared to the EC, primarily because OC can originate from primary sources (e.g., transportation and oil combustion) as well as secondary sources (i.e., atmospheric photochemical reactions) (Gianini et al., 2013; Yu et al., 2004). Therefore, the reduced traffic in summer was counterbalanced by the enhanced secondary formation of OC, resulting in a reduced seasonal variability between summer and winter.

Figure 1.

Extrinsic mass concentrations of EC and OC during summer, winter, and spring seasons (June 2022 - May 2023).

3.1.2. Metals and transition elements

Figure 2 depicts the measured extrinsic mass concentrations of metals and transition elements throughout the investigated seasons. During the summer season, we observed significantly higher concentrations of silicon (Si), lithium (Li), magnesium (Mg), aluminum (Al), iron (Fe), calcium (Ca), titanium (Ti), and manganese (Mn) compared to the winter period. In particular, the concentration of silicon during summer (11885.9 ng/m³) was approximately one order of magnitude higher than the winter period (957.3 ng/m³). The higher concentrations of soil elements during summer can be attributed to the prevailing high wind speeds from the northwest, passing through flat and arid desert regions, which enhance the long-range transport of soil particles and contribute to dust events (Al-Dousari et al., 2017; Al-Hemoud et al., 2017). In the winter season, we observed increased concentrations of copper (Cu) and lead (Pb), following a similar seasonal trend to the EC mass concentrations illustrated in Figure 1. This indicates that the reduction in non-tailpipe (i.e., brake and tire wear) emissions during summer can be attributed to the decreased traffic volume during the summer months. Furthermore, the mass concentrations of sulfur (S) and vanadium (V) exhibited higher levels during the winter season in comparison to summer and spring, which can be attributed to various meteorological factors. Primarily, the decreased wind velocity prevalent in the colder months exacerbated the impact of nearby local sources, such as Doha oil-based power plants situated approximately 10 km from our sampling site. Additionally, the formation of a temperature inversion layer near the surface, predominantly occurring in winter, led to the accumulation of particulate pollutants in the lower atmosphere (Beard et al., 2012; Malek et al., 2023; Trivedi et al., 2014). Sodium (Na) concentrations remained consistent across all seasons due to the close proximity of our sampling station to the gulf, where sea salt particles are continuously influx to our sampling point, eliminating any significant seasonal variation.

Figure 2.

Extrinsic mass concentrations of metals and transition elements during summer, winter, and spring seasons (June 2022 - May 2023).

3.1.3. Inorganic ions

The mass concentrations of PM_2.5-bound inorganic ions (i.e., sulfate, nitrate, and ammonium) during the investigated period are depicted in Figure 3. During the summer season, the nitrate concentration was 1.03 μg/m³, significantly lower than the 3.53 μg/m³ observed during the winter season. The extremely high summer temperatures contributed to the dissociation of ammonium nitrate, while the lower winter temperatures coupled with higher relative humidity favored its partition into the particulate phase (Argyropoulos et al., 2016; Samara et al., 2016). Sulfate mass concentrations exhibited less seasonal variation compared to nitrate, with higher levels observed during the winter months compared to summer and spring. Despite the enhancement of secondary sulfate formation during warm periods with high solar radiation, the lower summer concentrations can primarily be ascribed to the strong northwestern winds that transport locally generated SO₂ away from Kuwait City (Al-Awadhi, 2014; Brown et al., 2008). Alolayan et al. (2013) and Brown et al. (2008) reported that sulfate concentrations in Kuwait were substantially higher during the fall season (14 – 15 μg/m³) compared to the summer (8 – 9 μg/m³), attributing this trend primarily to the low-speed winds that intensified the influence of local sulfate sources in fall. Furthermore, the reduced ammonium concentration in the summer can be explained by its reaction with sodium chloride (NaCl) under increased temperatures, leading to the formation of volatile ammonium chloride (NH₄Cl), as reported by Aldabe et al. (2011) and Artíñano et al. (2003).

Figure 3.

Extrinsic mass concentrations of inorganic ions during summer, winter, and spring seasons (June 2022 - May 2023).

3.1.4. Comparison with previous studies conducted in Kuwait City

Alolayan et al. (2013) and Brown et al. (2008) analyzed the chemical composition of ambient PM_2.5 during the period of 2004 - 2005 in Kuwait City, given its proximity to most residential areas and the presence of a wide range of urban PM_2.5 emission sources. Figure 4 shows a comparison between our current research and previous studies on the annual average mass concentrations of PM_2.5, carbonaceous species, inorganic ions, and metal elements. The annual average PM_2.5 mass concentration in the current study (59.9 μg/m³) increased compared to previous years (i.e., 2004 - 2005), potentially due to urbanization, industrial expansion, and economic development in the region. This increase in PM_2.5 concentration is likely linked to various health implications, as highlighted by Colonna et al. (2023), who reported a 0.16% increase in respiratory hospitalization risk for every 1 μg/m³ increment in the ambient PM_2.5 concentration in Kuwait. Furthermore, despite the growth in population and subsequent rise in the number of vehicles in Kuwait over the past two decades, the current study revealed a relatively lower concentration of EC (2.3 μg/m³). This can be attributed to the location of our sampling site, which is adjacent to a less congested highway (i.e., Highway 85), while previous studies conducted their sampling at the intersection of two of the most heavily trafficked highways in Kuwait (i.e., Fourth Ring Road and Highway 50). Another contributing factor could be the recent construction of a new bridge above and parallel to Highway 85, which has substantially alleviated traffic congestion in the area. Although we reported a decrease in EC concentrations, the increase in OC concentration in our study supports its origin from a wide range of sources, including industrial activities and secondary formation in the atmosphere (Altuwayjiri et al., 2021; Saarikoski et al., 2008). Moreover, the increased mass concentration of sulfate observed in our current study compared to the level of 2004-2005 can be attributed to the increased electricity generation from burning fossil fuels, driven by economic and population growth over the past 20 years. Over the past two decades, the electricity generation capacity in Kuwait has significantly increased, predominantly (> 98%) relying on fossil fuel, and it is projected to further increase by 70% in 2035 (32 GW) compared to the level recorded in 2018 (18.8 GW), according to the report published by the Energy Building and Research Center at Kuwait Institute for Scientific Research (2019). Furthermore, the mass concentrations of several crustal metal elements (e.g., Si, Ca, K, Mg) were comparable to previous years. Conversely, the concentration of non-tailpipe tracers such as Zn and Cu has decreased, in alignment with the factors contributing to the decrease in EC concentrations.

Figure 4.

Annual average mass concentrations (per volume of air) of (a) PM_2.5, inorganic ions, and carbonaceous species, and (b) metals and transition elements.

3.2. PM_2.5 oxidative potential

Figure S4 presents PM_2.5 oxidative potential in terms of extrinsic DTT activity across different seasons. The per m³ of air volume normalized DTT activity showed a higher value of 1.31 ± 0.21 nmol/min/m³ during the winter season compared to the summer (0.65 ± 0.18 nmol/min/m³) and spring (0.88 ± 0.24 nmol/min/m³). The annual average extrinsic DTT activity in Kuwait (0.95 ± 0.31 nmol/min/m³) exceeded the PM_2.5 DTT levels in Los Angeles (0.35 nmol/min/m³) (Shirmohammadi et al., 2016) and Atlanta (0.31 nmol/min/m³) (Verma et al., 2014), but remained lower than Milan (3.38 nmol/min/m³) (Hakimzadeh et al., 2020), Athens (5.62 nmol/min/m³) and Beirut (3.51 nmol/min/m³) (Farahani et al., 2022). Saffari et al. (2014) have comprehensively analyzed PM_2.5 toxicological properties using DTT assay in ten distinct locations in southern California. The authors observed higher extrinsic DTT activity during the winter and fall seasons (0.5 – 1.3 nmol/min/m³) compared to the summer and spring (0.2 – 0.6 nmol/min/m³), which aligns with the seasonal trend observed in the current study. The increased PM_2.5 oxidative potential during the winter months in Kuwait can be mostly attributed to the increase in EC and OC concentrations, as they demonstrated a similar pattern to the DTT activity throughout the seasons. To confirm this argument, we conducted a Spearman correlation analysis (Table S2), which revealed a significant correlation between DTT activity and EC (R = 0.86), OC (R = 0.65), Pb (R = 0.87), and Cu (R = 0.65), suggesting a strong contribution of vehicular activities to the oxidative potential of ambient PM_2.5 in Kuwait as it will be discussed later in section 3.3.2. Previous research has consistently demonstrated a robust correlation between DTT activity and organic compounds, including OC, water-soluble OC, water-insoluble OC, PAH, and EC, leading to the conclusion that carbon and organic species play a significant role in driving the oxidative potential of particulate matter (Cho et al., 2005; Saffari et al., 2014; Steenhof et al., 2011). Given the importance of these organics in PM oxidative potential, another potential explanation for the lower summer DTT level is the evaporation of semi-volatile organic compounds (SVOCs) caused by extreme heat. However, the winter period enhanced the partitioning of SVOCs to the particulate phase, resulting in increased DTT activity as also argued by Saffari et al. (2014) who made similar observations in Los Angeles.

3.3. PM_2.5 emission sources and their contribution to the oxidative potential.

3.3.1. Source apportionment of PM_2.5 using PCA

Table 1 summarizes the results of the PCA analysis performed using the volume-normalized mass concentrations of carbonaceous compounds, chemical species, and inorganic ions for the entire study period. Our analysis revealed four distinct PM_2.5 emissions sources in Kuwait, covering a total of 93.92% of the variance in the data. The first factor was predominantly associated with crustal elements, including Si, Al, Ca, Fe, and Mg, implying that this source can be identified as "mineral dust emissions". According to previous research studies, these specific metals and transition elements are commonly considered tracers for soil dust in different regions around the world (Altuwayjiri et al., 2022; Sowlat et al., 2012; Tian et al., 2016). AL-Harbi (2015) reported that the annual dust deposits in Kuwait consisted of a substantial amount of silica which accounted for approximately 17% of the falling dust, further highlighting the importance of this element as a marker for soil dust emissions. In our PCA analysis, the contribution of the mineral dust factor was 36.06% of the PM_2.5 in Kuwait (Figure S5), which aligns with our expectations due to the large desert regions covering the majority of the country. AL-Harbi (2015) performed a comprehensive study to investigate the levels of dust deposition in Kuwait, which was found to be approximately 53.7 ton km⁻² month⁻¹ and significantly higher than various worldwide locations, including North India, Yazd (Iran), Texas (USA), Lanzhou (China), and California (USA), with values of 21, 6.5, 8.5, 11.1, and 1.6 ton km⁻² month⁻¹, respectively.

Table 1.

Principal components and loadings of PM chemical species.

	Principal components
	Mineral dust	Fossil fuel combustion	Traffic	Secondary aerosols
EC	−0.156	0.342	0.883	0.119
OC	−0.320	0.474	0.503	0.430
Sulfate	−0.320	0.320	0.334	0.727
Ammonium	−0.442	0.411	−0.034	0.752
Cu	−0.480	0.372	0.716	−0.209
Pb	−0.366	0.088	0.811	0.338
Si	0.920	−0.211	−0.213	−0.192
Mg	0.880	−0.289	−0.275	−0.242
Al	0.893	−0.284	−0.258	−0.231
Ca	0.896	−0.259	−0.256	−0.209
Fe	0.863	−0.295	−0.259	−0.294
S	−0.219	0.864	0.291	0.330
V	−0.335	0.835	0.236	0.280
Ni	−0.365	0.868	0.261	0.149
Variance (%)	36.06	23.69	20.32	13.85
Cumulative (%)	36.06	59.75	80.07	93.92

Loadings > 0.6 are in bold.

The second factor was primarily associated with high loadings of S, V, and Ni, which are the main markers of fossil fuel combustion (Cheng et al., 2018; Maciejczyk et al., 2021). The oil burned in Kuwait power stations contains a high sulfur content as well as measurable levels of Ni and V (Alolayan et al., 2013; Ramadan et al., 2008). As shown in Figure S6, our sampling site was significantly affected by numerous fossil fuel combustion sources in central and northern Kuwait, including Doha east and west power stations, Doha and Shuwaikh seawater desalination plants, and northern oil fields. According to Ramadan (2022), Doha power stations mainly burn crude oil and heavy fuel oil (HFO), while Shuwaikh seawater desalination plant completely relies on natural gas combustion. The abovementioned sources were the major generators of fuel-combustion tracers (i.e., S, V, Ni), which were transported by the dominant northwesterly winds to Kuwait City. Moreover, the oil and gas operations played a crucial role in emitting sulfur dioxide (SO₂) gas, which contributed to the increase in sulfur concentration in the atmosphere (Al-Awadhi, 2014). In our PCA results, the fossil fuel combustion factor has shown a relatively lower contribution (23.69%) to the PM_2.5 mass concentration in Kuwait City compared to the mineral dust factor.

The third factor was identified as "traffic emissions" due to the presence of tailpipe and non-tailpipe markers, including EC, Pb, and Cu. A large number of previous studies in the literature have considered EC as a major marker for tailpipe emissions (Jain et al., 2018; Yin et al., 2010), while Cu and Pb have been identified as tracers for non-tailpipe emissions (Harrison et al., 2012; Jeong et al., 2022; Querol et al., 2008; Shirmohammadi et al., 2016). This factor accounted for 20.32% of the PM_2.5 mass in Kuwait City, indicating a comparable contribution to the fossil fuel combustion factor. It is worth mentioning that OC loadings were comparable in all factors, except mineral dust factor, supporting the fact that OC originates from diverse primary (i.e., traffic and oil combustion) and secondary sources (Taghvaee et al., 2019).

The fourth factor was named "secondary aerosols" due to the elevated loadings of sulfate (SO₄²⁻) and ammonium (NH₄⁺), accounting for 13.85% of the PM_2.5 concentrations in Kuwait. Previous research studies have used these two species (i.e., ammonium and sulfate) as markers of secondary aerosol formation (Altuwayjiri et al., 2022; Jain et al., 2020; Sricharoenvech et al., 2020). Our sampling site in KISR is located adjacent to Sulaibikhat Bay, which is impacted by the release of industrial, domestic, and hospital sewage from several discharge points (Alshraifi et al., 2009; Mydlarczyk et al., 2020). The decomposition of organic compounds in wastewater releases odorous hydrogen sulfide (H₂S) and ammonia gases into the atmosphere (Widiana et al., 2017), which normally undergo a series of atmospheric chemical reactions that contribute to the formation of secondary ammonium sulfate aerosols. Briefly, H₂S can be rapidly oxidized in the atmosphere to sulfur dioxide (SO₂), which can then be further oxidized to sulfur trioxide (SO₃) that reacts with water droplets, forming sulfuric acid (H₂SO₄) (Abdul Raheem et al., 2009). Ammonium sulfate is formed secondarily from the gas-phase reaction of sulfuric acid (H₂SO₄) with ammonia (NH₃) (Seinfeld and Pandis, 2016; Sinanis et al., 2008).

3.3.2. The contribution of PM_2.5 sources to the oxidative potential using MLR approach

Multi-linear regression was performed between the air volume-normalized DTT activity and the resolved factor scores obtained from the PCA analysis. Table 2 presents the MLR results, including the standardized coefficients and R² value, which were used to quantify the relative contribution of emission sources to the oxidative potential of PM_2.5. It should be noted that the mineral dust factor was excluded from the MLR analysis due to the negative correlation observed between soil dust tracers (e.g., Si, Al) and DTT activity as was presented in Table S2. According to the MLR results, the road traffic factor was the most significant source influencing the oxidative potential of ambient PM_2.5, with a contribution of 47% (standardized Beta coefficient = 0.795) to the DTT activity in Kuwait (Figure 5). This aligned with the significant correlation highlighted previously in Table S2 between DTT activity and traffic emissions, including tailpipe and non-tailpipe markers. It was also supported by previous research studies corroborating the significant effect of vehicular exhaust and road dust (i.e., non-tailpipe) emissions on the redox activity of ambient PM (Hu et al., 2008; Shirmohammadi et al., 2016, 2015). Previous research carried out by Taghvaee et al. (2019) reported that vehicular emissions were responsible for 44% of the overall PM-induced redox activity in Athens, Greece. Furthermore, a study conducted by J. Wang et al. (2020) in Bangkok, Thailand, revealed that fossil fuel combustion, including vehicular exhausts, was the major contributor (63%) to the total PM oxidative potential in the region. Moreover, as shown in Figure 5, the contributions of fossil fuel combustion and secondary aerosols were 19% and 13% to the overall DTT activity, respectively. To further support our MLR findings, Fadel et al. (2023) investigated the predominant sources in Beirut (Lebanon) and revealed that tracers of heavy fuel combustion (e.g., V, Ni) demonstrated a notable contribution (>14%) to the ambient PM oxidative potential. Additionally, Romano et al. (2020) and J. Wang et al. (2020) identified strong correlations between tracers of secondary aerosols (i.e., sulfate, ammonium) and PM oxidative potential. It is important to note that ammonium sulfate is not particularly toxic (Cesari et al., 2019; Fang et al., 2016); however, its high correlation with DTT activity imply the coexistence of other toxic secondary aerosols (e.g., OC, soluble Fe) which are known to significantly contribute to the increase in oxidative potential (Cho et al., 2005; Fang et al., 2016; Giannossa et al., 2022).

Table 2.

Output of multi-linear regression (MLR) analysis between PCA factor scores (independent variables) and extrinsic DTT activity (dependent variable).

Source	Unstandardized coefficient (± Std. error)	Standardized coefficient	P-value	R2
Constant	0.914 ± 0.048		0.000	0.79
Fossil fuel combustion	0.147 ± 0.049	0.329	0.007
Traffic	0.356 ± 0.047	0.795	0.000
Secondary aerosols	0.097 ± 0.049	0.216	0.063

Figure 5.

The percentage contribution of emission sources to PM_2.5 oxidative potential in Kuwait.

4. Summary and conclusions

This study provides a comprehensive insight into the seasonal variability, chemical composition, and source apportionment of PM_2.5 concentrations in Kuwait, emphasizing the crucial link between emission sources and PM oxidative potential. The data revealed notable seasonal variations in ambient PM_2.5 levels, with summer season marked by the highest PM_2.5 concentration (75 μg/m³) primarily due to the increase in soil dust elements. The intense summer winds (known as Shamal) combined with dry ground conditions facilitated the resuspension of dust from the widespread desert regions. To identify the predominant PM_2.5 sources in Kuwait City, PCA was employed and effectively disentangled four distinct principal components (i.e., emission sources), covering 93.9% of the variance in the data, which were mineral dust emissions, fossil fuel combustion, road traffic, and secondary aerosols. Mineral dust, largely attributed to the vast desert landscapes of Kuwait, was the dominant contributor (36.1%), highlighting the region's vulnerability to frequent dust episodes. The fossil fuel combustion factor had a contribution of 23.7%, which emphasized the influence of local power plants and the country's heavy dependence on oil-based energy sources. The road traffic factor, underscored by markers such as EC, Pb, and Cu, revealed a contribution of 20.3% from both tailpipe and non-tailpipe emissions. Secondary aerosols, particularly characterized by sulfate and ammonium ions, highlighted the influence of the indirect emissions resulting from complex atmospheric chemical reactions. Using the MLR approach, this research quantified the contribution of the identified emission sources to the oxidative potential of PM_2.5. Road traffic was the most potent factor affecting PM_2.5 oxidative potential (47%), corroborating findings from previous research studies and underscoring the need for vehicular emission controls. Moreover, this work highlighted that the annual average PM_2.5 concentrations in Kuwait (59.9 μg/m³) considerably exceeded both WHO and US EPA guidelines, which underscores the urgent need for strategic interventions to enhance air quality. It is imperative to note that while the natural origins of mineral dust might be challenging to control, the other identified sources can be managed with stringent policy measures, technological advancements, and public awareness campaigns. This study offers a robust foundation for future air quality management strategies in Kuwait, underlining the importance of a harmonized approach that encompasses monitoring, source identification, and targeted mitigation measures.

Highlights.

• PCA and MLR were used to identify the contribution of PM_2.5 sources to DTT activity

• Annual average PM_2.5 level in Kuwait was six times higher than WHO recommendation

• PM_2.5 sources were mineral dust, fuel combustion, traffic, and secondary aerosols

• Road traffic was the predominant contributor (47%) to PM_2.5 oxidative potential