Source identification and characterization of water-insoluble single particulate matter in rainfall sequences

Precipitation chemistry can reflect the quality and pollution status of the local atmospheric environment, making it a good indicator for tracking the impact of human activities on the atmosphere. The research on rain chemistry improves our comprehension of the emission and transport processes of local and regional pollutants and their possible ecological effects. When atmospheric deposition occurs, it is essential to collect the entire event in pre-determined quantities to determine the dissolved and undissolved contaminants in the samples. This technique is useful for establishing a link between the source of pollution and nearby or distant transport. Compared with samples acquired by direct wet or dry precipitation, there is a higher possibility of acquisition of a single atmospheric particulate matter (PM) from samples collected during precipitation events. Well-characterized and accurately identified pollutant species are essential to obtain reliable results in source identification studies (Berberler et al. 2022).

Our comprehension of Earth's cycles and their impacts can be improved by investigating the wet and dry fractions. PM, nitrogen and sulfur compounds, ozone, and other atmospheric elements can reach different ecosystems through processes of natural deposition. Dry deposition is the process by which atmospheric trace gases and PM fall directly to the Earth. When atmospheric gases mix with atmospheric suspended water, wet precipitation occurs, which is then cleared by rain, snow, or fog. Acid rain is an example of wet precipitation that can cause damage to forests, harm small living creatures, and erode steel structures as well as stone buildings and statues. Acid rain, also known as acid deposition, is the process by which acidity is transferred from the atmosphere to the Earth's surface. It can occur through acidic rain, snow, or sleet, all of which are forms of acidic wet deposition (Corell et al. 2021; Likens et al. 2021). The dry deposition of acidic particles and gases can also affect life, even during dry periods. Rainwater is usually slightly acidic because it absorbs carbon dioxide from the atmosphere and organic acids are formed due to biological activity. This study suggests that Antalya, a location with tourist attractions and agricultural areas, is a suitable sampling point for many studies.

Sulfur (S) and nitrogen (N) are the two major chemical constituents of precipitation. Excessive amounts of S and N can reduce the capacity of water bodies to neutralize acid deposition, thereby accelerating the process of acidification. Global climate change has reduced the availability of clean freshwater in many countries. As a result, alternative water sources such as rain, drizzle, fog, and dew are becoming more attractive in many parts of the world. In the absence of rain and under suitable conditions, fog can be a preferred water source (Meunier & Beysens 2016). Rainwater plays a crucial role in removing harmful gases and air pollutants from the atmosphere. Two main systems remove these pollutants during rain events: rainout (in-cloud scavenging) and washout (below-cloud scavenging). Local emissions, meteorological conditions, and pollutant transport processes can affect the chemical composition of rain (Oduber et al. 2021). Excessive deposition of atmospheric pollutants can negatively affect the surface ecosystem in several ways, such as causing eutrophication, increasing the acidity of soils and water bodies, and reducing biodiversity (Tiwari et al. 2012; Nieberding et al. 2018; Kundu et al. 2023).

Rainwater in the atmosphere has the capacity to dissolve over 90% of all pollutants; however, it can also contain large amounts of metals and metalloids, as well as ions (Imo et al. 2021; Zeng et al. 2022). In addition, the composition of rainwater is primarily influenced by its occasional acidity (Zhang et al. 2024). PM deposition can have direct effects on ecosystems, either negative or positive, such as nutrient enrichment. Atmospheric constituents, including ozone, nitrogen, sulfur compounds, and PM, can penetrate the soil, freshwater, marine environments, and plant–animal tissues through naturally consisting dry and wet expression processes (Berberler et al. 2022). The process by which atmospheric trace gases and PM fall directly to the ground is known as dry deposition. Wet deposition is the process by which atmospheric gases combine with water suspended in the atmosphere and are subsequently removed by rain, snow, or fog. It can damage trees, kill insects, remove paint, corrode steel structures, and wear down stone structures and sculptures. Acid rains were an example of wet precipitation. It is important to note that wet precipitation can have negative effects on the environment and human health. Acid rain is most generally referred to by the more technical phrase ‘acid deposition’, which explains the different methods by which acidity flows from the atmosphere to the Earth's surface (Kilic & Pamukoglu 2023).

After reviewing the literature, one study stands out for its use of SEM‒EDS in conjunction with particle size analysis on sequential rain samples. The study found that when identifying pollution sources, the proximity of the sampling area to the seaside and the influence of sea salt can suppress primary pollutants from nearby or remote transport. As a result, the study found that sea salt had a much greater impact on the samples than previously thought. This study analyzed the soluble and insoluble components, including particle size and SEM‒EDS analysis. Particle size and SEM‒EDS analysis were used together with pollution source investigations to identify the primary pollutants affecting rainfall and those transported from nearby and distant locations in each rainwater sample (Kilic & Pamukoglu 2023). A number of studies employing SEM‒EDS for the characterization of PM in samples of both raw and treated water (Pivokonsky et al. 2018), marine water (Reisser et al. 2014), snow deposits (Miler & Gosar 2013), and fallout PMs in HEPA filtered cleanroom (Mohan et al. 2019). When literature research is examined, it has been observed that rainwater and atmospheric transport issues have been investigated. A number of studies in the literature have employed SEM‒EDS as a direct or complementary technique for the characterization of airborne PM in urban atmospheres (Aikawa et al. 2014; Bai & Wang 2014; Blanco-Alegre et al. 2019; Drinovec et al. 2020; Oduber et al. 2021; Bao et al. 2022; Tohidi et al. 2022). A review of the literature reveals that the combination of scanning electron microscopy (SEM) and energy-dispersive X-ray spectrometry (EDS) represents a highly effective technique for determining the elemental composition, size, and morphology of PM in a diverse range of environmental samples. Sequential sampling of rainfall events can be considered one of the effective atmospheric sampling methods for quantifying both water-soluble and insoluble pollutants, as well as for correlating them with their potential local and distant sources. However, the presence of secondary air pollutants (Akoto et al. 2011), which are formed through atmospheric chemical and photochemical reactions of gaseous pollutants, may complicate source identification studies. Rainfall events are stochastic processes, characterized by occurrence frequencies and depths that are treated as random variables defining their temporal structure (White et al. 2013).

• Ten separate rainfall events were progressively sampled for this investigation, and the samples were examined for pH, conductivity, anions, and cations. SEM‒EDS and a particle size analyzer were used to evaluate the morphology, chemical composition, and particle size distribution of water-insoluble PMs in successive samples. One hundred milliliters of the samples were employed directly for the determination of the particle size distribution, utilizing the Malvern Mastersizer 3000 (Malvern, UK). This instrument comprises a total of 64 channels or detectors, which were employed to measure both the red and the blue components of the measurement, without the use of any additional dispersant.

• The analytical results were used to estimate the contributions of local and distant sources to the measured species concentrations as percentages of precipitation and leaching. We generated and reviewed local wind rose plots and estimated back trajectories in the upper atmosphere to assess the possible sources of pollution influencing the rains. In filter sampling, the probability of finding single particles can be reduced as the particles stick to each other and physically cover the ones below. Furthermore, examining rainwater samples collected in low quantities or diluting rainwater may result in a higher number of individual particles. In contrast, filter sampling does not allow for such dilution, which could be considered a drawback.

• In this study, rain samples are collected sequentially, and insoluble particles in rainwater are characterized separately for each sequence. Source types were determined by using the morphological structures and chemical compositions of the particles.

• Back trajectories were used only to predict which sectors the air masses were transported from. Moreover, the sampling strategy is sequential sampling and there were only two studies in Antalya using sequential sampling (Kilic & Kilic 2023).

• The main and innovative difference of this study compared with previous lies in its fresh perspective on the traditional approach to sampling and evaluating rain samples. Unlike prior research, this study introduces several novel elements: a new sampler (a homemade sequential sampler), a new sampling design (sequential sampling), and a new evaluation method (SEM and trajectory combination particle size analyses). In their study on rainwater collected sequentially in Antalya in 2022, Kilic & Pamukoglu (2023) compared the results obtained with the results detected in rainwater collected in 2023. Since Antalya is an important agricultural and tourism city, suspected pollutants and their sources were sampled in rainwater samples collected consecutively for two consecutive years and it was determined to what extent the results differed.

Located on the Mediterranean coast, Antalya's main industries are agriculture and tourism. The Antalya region is in the Mediterranean climate zone. Winters are generally rainy and summers are hot and humid. Precipitation rates have decreased significantly in recent years. According to the meteorological data, there were 57 days of rain in the study area in 2023 (MGM 2023). The temperature in Antalya province varies between 28 and 40 °C in summer. The processing of enhanced chromium ore in the port region near the sampling area, the natural gas cycle, and thermal power plants can be considered the primary sources of air pollution. The increase in vehicles in traffic due to the recent large-scale migration to Antalya and the fact that the city hosts nearly 15 million tourists in the summer months can be listed among the main sources of pollution. Furthermore, the impact of pollutants from structured industrial regions and intensive agricultural operations can be addressed, as can the sampling area's closeness to the D-400 and D-650 highways. According to the reports of the Provincial Directorate of Environment and Urbanization of the Republic of Turkey, the amount of coke or semi-coke fuel type obtained from hard coal used for ferrochrome production in 2023 is 19.967769 tones, 93.756125 cm³ of natural gas, and 3.314820 kg of fuel oil. Motor vehicle exhaust fume pollution is one of the primary sources of air pollution, according to research (Berberler et al. 2022).

A total of 25 samples were collected from 10 different precipitation events between 1 January 2023 and 31 May 2023. For each rainfall event, fractional (sequential) samples were collected by adjusting the collection volume (300 mL) through fixed volume software from the beginning to the end of the event. The collected 300 mL samples were utilized for particle size analysis (100 mL), pH measurement (15 mL), and ion and SEM analyses (85 mL). The sampling system, designed and developed for this study, was mounted 25 m above ground level (AGL) on the terrace to minimize the influence of surface soils on the collected rain samples. The hoses used were connected to an automated rain collection device maintained in a completely isolated room. A high-precision moisture sensor attached to the funnels on the terrace activates when precipitation begins. Fractional samples of 300 mL are automatically collected sequentially into sample collection containers within the device's chamber, which is housed in a closed and protected environment. The initial rainfall event in the sampling area took place on 12 January 2023, with the first sequential rainwater samples collected from this date designated as the A Series. The two fractions within this series were labeled as Fraction A1 and Fraction A2. Subsequently, two additional samples were collected during a rainfall event on January 27, designated as the B Series, with fractions labeled as B1 and B2. On March 12, four consecutive samples were obtained during another rainfall event and designated as the C Series, with fractions labeled as C1, C2, C3, and C4. On the following day, March 13, three samples were collected and labeled as the D Series, with fractions named D1, D2, and D3. For the rainfall event on April 4, three samples were collected and labeled as the E Series, with sequential fractions designated E1, E2, and E3. Two consecutive samples collected during the April 10 precipitation event constituted the F Series, with fractions labeled as F1 and F2. Similarly, the precipitation event on April 11 was labeled as the G Series, with two samples assigned labels G1 and G2. The rainfall event on May 6 resulted in three samples, designated as the H Series, with fractions labeled as H1, H2, and H3. During the precipitation event on May 11, two consecutive samples were collected and labeled as the I Series, with fractions named I1 and I2. Finally, two samples collected during the rainfall event on May 31 were designated as the K Series, with fractions labeled as K1 and K2.

Information about all rainfall events is presented in Supplementary Material 1 Table S1. The maximum number of fractions, four in total, was collected during the rain event that occurred on 12 March 2023. The final rain event took place on May 31 and was collected as two fractions. The volume of all collected rainwater samples is 300 mL. In the SEM‒EDS studies, it was found that the smallest particles among all analyzed particles in each rainwater sample were the most significant. For the evaluation and identification of PMs, we utilized their relative chemical compositions obtained from SEM‒EDS and compared the EDS results and SEM images with existing literature. The primary objective of this study is to demonstrate the advantages of sequential rain sampling in obtaining a greater number of individual atmospheric PMs per sample compared with ground-level wet-only and other PM sampling methods. Each sample contained numerous particles, the majority of which exhibited similar characteristics. It was impractical to present the SEM‒EDS results for every individual particle in this article. Therefore, particles measuring 5 μm and above, which displayed notable differences at the nanoscale, were given greater emphasis than others. Consequently, an effort was made to evaluate particles with diverse pollutant properties within the samples. The objective was to obtain insights into the transportation of pollution sources. The decision to group the precipitation events according to their transport regions was made based on the results of upper-atmosphere air mass trajectories. The quantity of precipitation events occurring during the study period poses the only disadvantage for PM sampling and quantification. Significant efforts were made to sample all precipitation events, depending on the precipitation pattern, the rarity of the events, or regional climatic conditions.

pH and conductivity measurements were performed immediately after the completion of the rainfall events for the sampled series. These measurements were performed using two different Mettler Toledo Seven Easy model instruments. For each fraction in the series, 15 mL samples were taken from 300 mL of the sample collected in clean beakers that had been previously rinsed with distilled water. pH and conductivity values were determined by calculating the averages of two replicate measurements taken with both instruments. After each sample, the probes used were washed with distilled water and prepared for the next measurement (Kilic & Pamukoglu 2023).

A total of 100 mL of sample from each series was utilized to directly determine particle size distributions using the Malvern Mastersizer 3000 (Malvern, UK). This device measures both red and blue components without the use of additional dispersants and is equipped with a total of 64 detectors. To prevent flocculation and agglomeration, particle size results were obtained immediately following pH measurements of the samples. Particle size distributions ranging from 0.1 μm to 5.0 mm were achieved through meticulous analysis.

As indicated in previous studies on particle size analysis (Bayramoğlu-Karşı et al. 2018; Berberler et al. 2022; Kilic & Pamukoglu 2023), we have demonstrated that particle size distributions in rainwater series can be effectively determined. A critical step in the analysis of particle size in rainwater is to thoroughly clean the sample inlet chamber with ultrapure water before each sample, particularly following the results of the previously analyzed sample, due to the low PM content in the series. Another essential consideration in this measurement is that the sample volume used for determining particle size distributions is adequately separated. Therefore, each fraction was collected in a fixed volume of 300 mL. Given the limited PM in rain series samples, prior studies have established that a minimum of 100 mL of sample is required for the analyzer employed.

To determine the dissolved ions in the rainwater samples (Lekouch et al. 2010), 85 mL of each sample was filtered through a 0.22 μm pore size cellulose acetate filter. The resulting filtrate was used to analyze major anions (F⁻, Br⁻, ⁠, Cl⁻, ⁠, ⁠, and ⁠) and cations (Ca²⁺, Na⁺, Mg²⁺, ⁠, K⁺, and Li⁺) without any pretreatment. An IonPac AS19 (250 × 2 mm) column was employed, along with a Guard IonPac AG19 (50 × 2 mm) column. The column temperature was maintained at 40 °C, and the flow rate was determined as 0.3 mL/min, with an injection volume of 500 μL. The chromatographic method for determining ion concentrations in rainwater samples was validated through recovery studies. The parameters for the ion chromatograms are detailed in Supplementary Material 2 Table S2. Ion concentrations were assessed by injecting the filtrates into a Dionex ICS 5000 ion chromatography system (Thermo Fisher Scientific Inc.), which was equipped with an automatic sampling device, following the ISO 10304 method for ion determination in rainwater samples (Wu et al. 2021).

The filters utilized in the SEM analysis were identical to the 0.22 μm cellulose acetate filters employed in ion analyses, through which the samples were filtered. To prevent contamination from external particles, these filters were cut in a fume hood. The cut filter papers were placed in Petri dishes without any contact and were dried in an oven at 40‒50 °C. SEM analyses were conducted using an FEI Quanta 250 microscope, with the samples remaining uncoated and the chamber pressure set to a low vacuum of 60 Pa (Berberler et al. 2022; Kilic & Pamukoglu 2023). The cut filters were placed directly into the sample placement container of the SEM device without any further processing. These filters were specifically chosen and supplied to match the sample container of the device.

The only purpose of back trajectories was to determine the sectors from where the air masses were moved. The several source types were distinguished by the morphological patterns and chemical makeup of the particles. To account for the maximum fractions of the boundary layer, the isentropic HYSPLIT model was run to compute 72-h backward trajectories for 100, 1,000, and 1,500 m AGL. The meteorological input data was selected from the global forecast system (GFS) with a resolution of 0.25°. The HYSPLIT model was used to map the coordinates of the sampling point in the receiving environment, generate the return trajectory, and determine the paths taken by the air mass responsible for the rain events that occurred at heights of 100, 1,000, and 1,500 m AGL. The model type was selected as a backward, vertical motion, entropic trajectory, with heights of 100, 1,000, and 1,500 m AGL (Stein et al. 2015; Rolph et al. 2017). Back trajectories were used only to predict which sectors the air masses were transported from. Different source types were distinguished by the morphological features and chemical compositions of the particles.

The recovery values and controlling quality (R², limit of detection (LOD), and limit of quantification (LOQ)) of the ion results are presented in Supplementary Material 1 Tables S3 and S4. It was performed from 10 replicate analyzes of the lowest calibration level standard. The LOD was calculated as three times the standard deviation. The amounts of ions in the rainwater sample were determined using the conventional addition method (Matrix spike). The resulting concentrations were then quantified using the calibration curve. Three samples were combined to create the zero matrix for standard addition calibrations. One calibration curve was used for each ion. The calibration procedure involved the addition of standards prepared at the concentrations of the stock standard solutions. The LOD and LOQ values of the analytical method used in this study are comparable to those reported in validation studies on water samples in the literature (Montoya-Mayor et al. 2013; Kılıç et al. 2015; Berberler et al. 2022; Ismail & Afify 2022). Supplementary Material 1 Tables S3 and S4 show the recovery values of the ion analysis. In addition, the results obtained for each ion were determined to be within the expected concentration range.

The primary aim of this study was to assess the feasibility of directly measuring particle size distributions during sequences of rainfall utilizing the laser diffraction technique. These measurements were subsequently correlated with the PM data obtained from SEM‒EDS analyses, with the intention of validating particle sizes and exploring the occurrence of potential ‘ghost peaks’. The elevated concentrations of PM and water-soluble ions noted in the later rainfall sequences can largely be ascribed to a decrease in rainfall intensity, which exhibits an inverse correlation with the washout efficiency of PM removal. Moreover, local atmospheric emissions contribute significantly to the overall particulate load. During periods of reduced rainfall intensity, the inadequate number of raindrops fails to effectively wash away pollutants from the vapor phase or to dislodge dust from the surface. The presence of larger particles is generally observed in the initial two rainfall events, suggesting that raindrops resuspend particles that are in contact with dry ground, which are then lifted into the atmosphere by prevailing winds. These particles are subsequently removed from the lower atmosphere by falling raindrops. This observation is further supported by the high concentration of large particles identified in the first two sequences.

SEM‒EDS and a particle size analyzer were employed to assess the morphology, chemical composition, and particle size distribution of water-insoluble PMs in successive samples. In addition, local wind rose plots were generated, and back trajectories in the upper atmosphere were estimated to evaluate potential pollution sources affecting the rain. In filter sampling, the likelihood of detecting individual particles can be diminished as particles tend to adhere to one another, obscuring those beneath. Furthermore, analyzing rainwater samples collected in smaller quantities or diluting rainwater may lead to a higher number of individual particles being detected. In contrast, filter sampling does not permit such dilution, which may be regarded as a limitation of this method. In this study, rain samples were collected sequentially, and insoluble particles in rainwater were characterized separately for each sequence. The types of sources were determined by examining the morphological structures and chemical compositions of the particles (Berberler et al. 2022).

The A Series Fraction A1 contains particles between 0.5 and 1 μm, with a peak at 41 μm. The A Series Fraction A2 contains particles from 1 to 100 μm, with peaks at 0.5 and 100 μm, possibly due to changes in wind direction. In the series, two peaks appeared in the B Series Fraction B1, 12 and 22 μm. In the B Series Fraction B2, the storm showed stability and low wind mobility. In addition, particles from 0.5 to 1 μm and 10 to 100 μm were detected. Only a single 39 μm peak and particles between 0.4 and 1 μm were detected.

A total of four rainwater fractions were collected at the C Series. The C Series Fraction C1 contained particles ranging from 0.3 to 1 μm and 10 to 100 μm with peaks at 2, 4, 5, and 18 μm. The last fractions of the C Series showed a decrease in particles cleaned by rain. Supplementary Material 1 Table S10 shows the particle distribution contents for the 10th percentile (d10), 50th percentile (d50), and 90th percentile (d90) when examining the distribution width of all fractions. In the D Series, the fractions showed a multimodal distribution. The fifth E Series, on the other hand, showed single peaks. The first two or three series typically exhibit particle size distributions with higher volume densities at the coarse end of the multimodal particle size distribution. It may have a single peak centered around 1.0 μm during calm air precipitation, that is when there is no new front bringing rain from different sectors.

In the F Series, two fractions were collected. Later, the rain was carried by different air masses resulting in a multimodal distribution. Particle size plots for other series and particle distributions for d(0.1), d(0.5), and d(0.9) are given in Supplementary Material 1 Table S10 and Figure S1. The media reported Dv(50) values ranging from 0.582 to 17.736 μm for precipitation events, indicating that half the particle sizes were below and half above this median value. Grain sizes corresponding to Dv(10) showed that 10% of the grain population was below the corresponding grain size. In the same way, Dv(90) means that 90% of the distribution is below the value that corresponds to Dv(90). No significant air bubble peak was observed in this study as shown in Figure 6 and Supplementary Material 1 Figure S2. SEM micrographs were used to verify that peaked specimens were present in the areas of suspected grain size distribution.

The results indicated that approximately 99% of the fluoride ions were present in all rainfall series. Furthermore, the concentration of ions believed to originate from non-marine sources was observed to be 53% in the second sample of the first fraction series, in comparison with the other samples. About 99% of nitrate, ammonium, and calcium ions in all fractions, 56–95% of potassium ions in different series, 15–96% of magnesium ions, 3–73% of sodium ions, and 3–73% of sulfate ions. It has been observed that 46–97% of these come from non-marine sources. Three to 54% of the sulfate ions, 5–44% of the potassium ions, 4–85% of the magnesium ions, 27–97% of the sodium ions, and about 10% of the nitrate, ammonium, and calcium ions in all fractions are of marine origin. Figure 3 shows the calcium ion plot for all fractions as an example of the plots obtained as a result of the calculations. Graphs for other ions are given in Supplementary Material 1 Figure S1.

The observed low ratios may be attributed to desert dust inputs that affect precipitation events in the region. It is known that the deficiency of anions in the anion/cation ratio results from the presence of desert dust, which is rich in bicarbonate and calcium carbonate, transported by upper-atmosphere air masses from North Africa and the Middle East (Bayramoğlu-Karşı et al. 2018). Anion deficiency is a phenomenon frequently observed in rain measurement studies (Chandra Mouli et al. 2005). The absence of measurements for acetate (CH₃COO⁻) and formate (HCOO⁻) ions, which can significantly alter the Σanion/Σcation ratio, likely contributes to the anion deficiency observed in these studies. The primary sources of organic acids, which affect the composition of rainwater, are hydrocarbons emitted by vegetation and organic acids originating from industrial activities and subsequent oxidation. To verify the completeness of the measured ions, we added the calculated bicarbonate equivalents to the total anion equivalents. After adding bicarbonates, the equivalent ratios of total anions to total cations for the A, B, C, D, E, F, G, H, I, and K Series were calculated as 0.77 ± 0.30, 0.58 ± 0.21, 0.77 ± 0.55, 1.02 ± 0.18, 0.88 ± 0.33, 1.05 ± 0.35, 0.49 ± 0.10, 0.62 ± 0.32, 0.65 ± 0.01, and 0.32 ± 0.24, respectively, as presented in Table 2. As discussed below, the samples from the K Series precipitation event contained high levels of calcium ions, and thus the anion deficit in these samples can be attributed to bicarbonate ions. In general, anion deficits were observed across all precipitation events, except for the D, E, and F Series rain events, likely due to unmeasured bicarbonate ions (Bayramoğlu-Karşı et al. 2018). However, the anion deficits observed in the B and G Series precipitation events were significant and could not be explained by the absence of bicarbonates. Therefore, the anion deficits in these two series may be attributed to other unmeasured anions from biological and anthropogenic organic acids. Statistically significant correlations between anions and cations related to agricultural activities indicate that the scavenging mechanism is much more effective than the rain mechanism. This observation supports the dominance of local pollution sources over the long-range transport of pollutants represented by the rain process.

The types of pollutants present in rainwater determine its acidity or alkalinity. The solubility of acid gasses like SO₂ and NO_x in the water is the primary source of rainfall acidification. The precursors of the acids in rainwater, SO₂ and NO_x, change into H₂SO₄ and HNO₃ during the air transmission of rain clouds and precipitation in the receiving environment. Besides the strong acids H₂SO₄, HNO₃, and HCl, organic acids from industrial processes are also important in acidifying rainwater (Kulshrestha et al. 2003). The acidity of rainwater can also be influenced by organic acids, which are one of the factors that can alter rainwater as it passes through forested areas. The acidity of rainwater is influenced by some factors, including the deposition of sea salts and the effects of climate change (Wright & Jenkins 2001).

There was no evidence of acid rain based on the pH values from rainwater array readings. This can be explained as follows: Antalya is situated in a dry environment; thus, before the raindrops reach the receiving area, the amount of dust that has been re-suspended from the ground is balanced. Because of the presence of neutralizers (ammonia and calcium in dust lifted from the ground), raindrops are neutralized in the air before falling in this study, which does not include acid rain. High levels of ⁠, ⁠, and were detected in samples; this points to mainly anthropogenic contributions. High amounts of Na⁺ and Cl⁻ found in the composition of rainfall in certain samples suggest the presence of sea salt. Furthermore, the proximity of Antalya province and the sampling region to the sea may have had an impact on pH measurements. It is thought that fossil fuel combustion and winter traffic emissions may have affected the samples by releasing organic carbon and other inorganic gases (CO, SO₂, NO, NO₂). In addition, a significant volume of soils rich in CaCO₃ is brought to the Mediterranean by the invasion of Saharan dust.

This type of dust transport is thought to be the main cause of the neutralizing acidity of precipitation in the western Mediterranean. In recent times, there has been a surge in interest in these issues. Upon examination of the studies, it becomes evident that dust storms that occur in four major continental regions, namely Libya-Egypt and Sahel (Niger, Chad, Mauritania, and Central Mali), have a profoundly negative impact on the region, including Turkey. There are even warnings about the potential health risks and air pollution associated with desert dust in the national press. Table 3 shows the calculations of neutralization factors obtained from rainwater samples. The results indicate that calcium ion is the most dominant neutralizing agent in the rainfall fields. The average NF values of calcium ions ranged from 0.89 to 20.22 for precipitation events. Ammonium and magnesium ions followed calcium ions and neutralized rainwater to a moderate extent compared with calcium. Ammonium NF values ranged from 0.01 to 1.03. Comparing the average NF values for calcium, ammonium, and magnesium, the values for potassium and sodium were computed at lower concentrations. The study's findings suggest that calcium ions serve as the primary neutralizing agent in rain droplets.

Table 3

Neutralization factors of the cations for the fractions

Fractions .	NF Ca2+ .	NF K+ .	NF .	NF Mg2+ .	NF Na+ .
A1	3.46	0.03	0.20	0.37	0.03
A2	1.86	0.04	0.17	0.17	0.00
B1	3.79	0.11	0.00	0.19	0.15
B2	3.57	0.12	0.56	0.35	0.00
C1	2.40	0.11	0.12	0.15	0.00
C2	3.53	0.06	0.10	0.66	0.00
C3	2.78	0.07	0.88	0.15	0.06
C4	2.22	0.06	0.20	0.46	0.00
D1	1.90	0.09	0.43	0.14	0.00
D2	0.94	0.03	0.39	0.34	0.17
D3	1.50	0.04	1.03	0.18	0.00
E1	2.01	0.04	0.92	0.25	0.00
E2	3.70	0.11	0.34	0.06	0.35
E3	2.34	0.08	0.18	0.07	0.00
F1	1.96	0.07	0.21	0.08	0.00
F2	4.13	0.17	0.66	0.22	0.00
G1	3.79	0.09	0.00	0.35	0.10
G2	4.72	0.10	0.00	0.66	0.18
H1	1.94	0.06	0.55	0.15	0.20
H2	0.89	0.02	0.19	0.34	0.17
H3	6.51	0.06	0.01	0.39	0.00
I1	1.72	0.02	0.19	0.16	0.60
I2	1.89	0.02	0.20	0.19	0.00
K1	3.02	0.05	0.16	0.13	0.00
K2	20.22	0.06	0.44	0.29	0.55

Fractions .	NF Ca2+ .	NF K+ .	NF .	NF Mg2+ .	NF Na+ .
A1	3.46	0.03	0.20	0.37	0.03
A2	1.86	0.04	0.17	0.17	0.00
B1	3.79	0.11	0.00	0.19	0.15
B2	3.57	0.12	0.56	0.35	0.00
C1	2.40	0.11	0.12	0.15	0.00
C2	3.53	0.06	0.10	0.66	0.00
C3	2.78	0.07	0.88	0.15	0.06
C4	2.22	0.06	0.20	0.46	0.00
D1	1.90	0.09	0.43	0.14	0.00
D2	0.94	0.03	0.39	0.34	0.17
D3	1.50	0.04	1.03	0.18	0.00
E1	2.01	0.04	0.92	0.25	0.00
E2	3.70	0.11	0.34	0.06	0.35
E3	2.34	0.08	0.18	0.07	0.00
F1	1.96	0.07	0.21	0.08	0.00
F2	4.13	0.17	0.66	0.22	0.00
G1	3.79	0.09	0.00	0.35	0.10
G2	4.72	0.10	0.00	0.66	0.18
H1	1.94	0.06	0.55	0.15	0.20
H2	0.89	0.02	0.19	0.34	0.17
H3	6.51	0.06	0.01	0.39	0.00
I1	1.72	0.02	0.19	0.16	0.60
I2	1.89	0.02	0.20	0.19	0.00
K1	3.02	0.05	0.16	0.13	0.00
K2	20.22	0.06	0.44	0.29	0.55

View Large

Similarity scores are determined by comparing data objects on an attribute-by-attribute basis, typically by summing the squares of the differences in magnitude for each attribute. This calculation is then used to derive a final outcome, referred to as the correlation score (Berman 2016; Profillidis & Botzoris 2019). To identify probable pollution sources and chemical components in rainwater samples, correlation analysis was performed using Statgraphics Centurion. Table 4 shows the r-values (p < 0.05). Pearson correlation was determined to be the most appropriate method for nominal continuous data measurements (Kiernan 2014). The distributions of ions were tested for normality. The term ‘skewness’ is used to describe a measure of the symmetry or shape of the data. When the standard skewness values are found to be outside the range −2.0 to +2.0, it may be concluded that the data are significantly departing from a normal distribution. The term ‘kurtosis’ is used to describe a measurement of how flat or steep the distribution of the data is with respect to a normal distribution. When standard kurtosis values are found to be outside the range −2.0 to +2.0, it may be concluded that the data are significantly departing from a normal distribution. A strong correlation was observed between and (0.6955) in rainwater. Rainfall episodes have the potential to remove these ions that accumulate in the upper atmosphere. Furthermore, since NO_x and SO₂ precursors influence precipitation processes, they may behave similarly (Shi et al. 2007; Karadeniz & Yenisoy-Karakaş 2020). Correlation coefficients for and (0.4576) and and (0.3713) were also calculated. The values for Ca²⁺ and (0.8042) and Ca²⁺ and (0.6655) show notable correlations. The relationship between and can be associated with pollutants originating from the same source region (Rao et al. 2016; Bisht et al. 2017; Nieberding et al. 2018).

Table 4

Binary correlation between ions analysis (p < 0.05, (r) values > 0.5 is shown in bold) (n = 25)

.	Ca2+ .	K+ .	.	Mg2+ .	Na+ .	Cl− .	.	.
Ca2+
K+	0.7208 (0.0000)
	0.1986 (0.3412)	0.4089 (0.0424)
Mg2+	0.6789 (0.0002)	0.6847 (0.0002)	0.1306 (0.5338)
Na+	0.2397 (0.2485)	0.4482 (0.0246)	0.0395 (0.8513)	0.5654 (0.0032)
Cl−	0.4724 (0.0171)	0.8460 (0.0000)	0.2289 (0.2711)	0.6628 (0.0003)	0.6975 (0.0001)
	0.6655 (0.0003)	0.6799 (0.0002)	0.3713 (0.0677)	0.6271 (0.0008)	0.1728 (0.4143)	0.5149 (0.0084)
	0.8042 (0.0000)	0.8655 (0.0000)	0.4576 (0.0214)	0.7476 (0.0000)	0.2754 (0.1827)	0.6455 (0.0005)	0.6955 (0.0001)

.	Ca2+ .	K+ .	.	Mg2+ .	Na+ .	Cl− .	.	.
Ca2+
K+	0.7208 (0.0000)
	0.1986 (0.3412)	0.4089 (0.0424)
Mg2+	0.6789 (0.0002)	0.6847 (0.0002)	0.1306 (0.5338)
Na+	0.2397 (0.2485)	0.4482 (0.0246)	0.0395 (0.8513)	0.5654 (0.0032)
Cl−	0.4724 (0.0171)	0.8460 (0.0000)	0.2289 (0.2711)	0.6628 (0.0003)	0.6975 (0.0001)
	0.6655 (0.0003)	0.6799 (0.0002)	0.3713 (0.0677)	0.6271 (0.0008)	0.1728 (0.4143)	0.5149 (0.0084)
	0.8042 (0.0000)	0.8655 (0.0000)	0.4576 (0.0214)	0.7476 (0.0000)	0.2754 (0.1827)	0.6455 (0.0005)	0.6955 (0.0001)

Note: The values presented in parentheses represent the p-values.

View Large

Ammonia is commonly present in the atmosphere in aerosol form as (NH₄)₂SO₄, NH₄HSO₄, and NH₄NO₃. The correlation between and (r = 0.465) does not differ significantly from that of and (r = 0.3713). This may suggest the presence of dominant atmospheric compounds such as (NH₄)₂SO₄, NH₄HSO₄, and NH₄NO₃ (Zhang et al. 2007; Cao et al. 2009). In addition, the correlation coefficient between sulfate and ammonium ions (r = 0.4576) indicates a lack of significant amounts of ammonium sulfate in the samples. The findings of this study further support the effectiveness of local sources, as ammonium sulfate in atmospheric samples is often used as an indicator of aged particles.

The strong correlation between Ca²⁺‒Cl⁻ and Mg²⁺‒Cl⁻ is likely due to the influence of sea salt, given the proximity of the sampling area to sea level. Correlation values for Ca²⁺‒Mg²⁺ (0.6789) and Ca²⁺‒K⁺ (0.7208) indicate a strong association with crustal sources (Al-Khashman 2009). On the other hand, the relationship between Ca²⁺‒ and Mg²⁺‒ may be attributed to secondary pollutants, likely originating from the urban center and transported over long distances, merging with local pollutants. Mg²⁺, Na⁺, and K⁺ ions have been linked to common sources influenced by dust and sea salt (Huang et al. 2008). The presence of biomass burning in the samples may be suggested by the association between K⁺ and Mg²⁺. In all samples, phosphate concentrations were below the detection limit. The results indicated that only the ions Na⁺ and Cl⁻ exhibited a lack of normal distribution. No major outliers were identified in the data set. Pearson correlation analysis was subsequently applied, assuming the normality of the data.

When studies in the literature are examined, it has been shown that the probability of the presence of atmospheric PM in sequentially collected precipitation events is higher than in samples obtained directly from PM samples only with wet/dry deposition and filter materials (Berberler et al. 2022). Consequently, the elevated concentration of these particles occasionally constrains the number of targeted micron- and submicron-sized single PMs. However, in contrast to samples of PM on filter media collected directly from the atmosphere, there is the possibility of a reduced population of large particles on the filter samples of rain sequences. In sequential rain samples, the sequence volume can be divided into several fractional volumes, filtered, and prepared for SEM‒EDS analyses in order to obtain a greater number of isolated single particles. Compared with aerosol samples collected at the ground level, the amount of dissolved and undissolved pollutants was less affected by crustal materials in the first- and second-order precipitation fractions. The principal aim of this study was to establish the feasibility of directly measuring particle size distributions during rainfall sequences utilizing the laser diffraction technique. Furthermore, the study sought to correlate these measurements with the PM identified in the SEM‒EDS results, both for the validation of particle sizes and for the exploration of potential ghost peaks. A prevalent challenge in particle size distributions is the occurrence of an artifact peak within the range of approximately 40–500 μm, commonly referred to as the air bubble peak, which may arise in aqueous dispersions. Fortunately, the identification of an air bubble peak can be facilitated through the following criteria: (1) there should exist a distinct separation between the bubble peak and the sample particle peak and (2) a microscopic examination can elucidate the presence of particles within the suspected size ranges (Berberler et al. 2022).

Analysis of the EDS results and SEM images for two fractions from rain samples in the B Series reveals notable findings. Examination of the first fraction, B Series Frac.B1(a), indicates the presence of a local soil particle potentially associated with rain samples, with elemental composition as follows: C 54%, O 33.03%, Al 0.25%, Si 5.10%, S 2.56%, and P 0.006%. In addition, an organic particle contaminated with Ti and road dust was detected in the same sample (road dust particle; B Series Frac.B1(c)). Further analysis of B Series Frac.B1(d) identified a silicate mineral composed of Fe, Al, and Mg, suggesting a mineral contribution from rain. In the second fraction of the B Series, B Series Frac.B2(a) prominently featured an organic, carbon-based particle. B Series Frac.B2(b) contained a soil-derived particle. Due to the influence of northern and southern winds on the B Series rain event, sodium-based particles were also detected in the rainwater, likely reflecting contributions from local atmospheric conditions.

Examination of the selected SEM‒EDS results for PM from the four sequentially collected fractions in the third rainfall event (C Series) yielded the following findings. In two images of the first fraction, C Series Frac.C1(a) and C Series Frac.C1(b), particles with identical EDS results and characteristics consistent with contaminated plant residues in the soil were identified. In the second fraction, C Series Frac.C2(a), contaminated plant residues of plant origin were similarly observed. The analysis of the third fraction, C Series Frac.C3(a), revealed microplastic particles (particles smaller than 5 mm) present in rainwater, as shown in images C Series Frac.C3(a) and C Series Frac.C3(b). Finally, in the last fraction, C Series Frac.C4(a), (b), and (c), an agricultural particle containing notable amounts of phosphorus and soil elements, was detected, indicating potential agricultural pollution sources.

The D Series rain event was divided into three sequential fractions, each revealing distinct particle characteristics upon SEM‒EDS analysis. Examination of the first fraction, represented by images D Series Frac.D1(a) and D Series Frac.D1(b), identified microplastic and rubber particles. In the second fraction, D Series Frac.D2, organic particles including pollen grains were observed, along with trace amounts of phosphorus and soil elements, likely originating from bromine-derived fly ash as seen in images Frac.D2(a) and Frac.D2(b). These elements suggest the presence of soil particles of agricultural origin. Analysis of the third fraction, D Series Frac.D3, yielded multiple particle types. In D Series Frac.D3(a), an organic particle associated with traffic, containing elements such as iron, nickel, cobalt, and chromium, was observed. Additionally, D Series Frac.D3(b) revealed the presence of diatoms from marine organisms, while D Series Frac.D3(c) showed fungal spores contaminated with sulfurous aluminosilicates. This series, collected on 13 March 2023, and illustrated in Supplementary Material 3 Figure S4 (100, 1,000, and 1,500 magnification), highlights the influence of local urban sources on air mass composition in the sampling area.

The SEM‒EDS results from Frac.D3(a) suggest that chromium, nickel, cobalt, and iron detected in organic particles likely originated from a nearby chromium mine, contributing to the resuspension of road dust (Berberler et al. 2022). The proximity of the sampling area to a high-traffic ring road also likely increases lead and chromium concentrations in rainwater samples, as these elements can be transported by air masses from Anatolia, where a chromium mine is situated near the Antalya provincial borders. During this rain event, northerly winds influenced the sampling area, later shifting toward the Mediterranean coast. The sea-originating winds carried diatom particles, which were detected in the D Series Frac.D3(b), while fungal spores contaminated with sulfuric aluminosilicates were found in the D Series Frac.D3(c). These observations underscore the significant impact of emissions from heavy traffic and local urban sources on rainwater composition in the sampling area.

The E Series rain samples, collected in three fractions, revealed various particle types upon SEM‒EDS examination. In the first fraction, images of E Series Frac.E1(a) and Frac.E1(b) identified fly ash and road dust particles. The second fraction, Frac.E2(a), showed spherical organic particles alongside road dust. In the third fraction, E Series Frac.E3, local pollutants from agricultural practices were detected, including plant spores, microplastics, unburned coal particles or biochar, and soot or organic particles. Differentiating PM based on source can be challenging due to the contributions from both the local urban atmosphere and other nearby sources. However, sequential rain sampling, combined with particle size analysis and SEM‒EDS, enhances the ability to characterize and trace the origins of both coarse and fine particles. In two consecutive samples from the F precipitation event, three particle types were identified, influenced by air masses originating from Middle Eastern countries and North Africa. In the first fraction, F Series Frac.F1(a), soil particles likely of agricultural origin were detected. In the second fraction, represented by F Series Frac.F2(a) and Frac.F2(b), microplastic particles and compact agglomerates of soot were observed.

Examination of the two fractions from the G Series revealed that air masses from the Middle East, characterized by local wind dominance, influenced the samples. In the first fraction, G Series Frac.G1(a), SEM images showed evidence of agricultural activity, including phosphorus-containing microplastics and microplastic particles polluted with sodium and titanium. In the second fraction, G Series Frac.G2, road dust particles, spherical soot particles from urban traffic, crustal elements, and soil and pollen particles from urban pollution were detected in images Frac.G2(a-1, a-2) and Frac.G2(b). Examination of SEM images taken from the H Series samples revealed different particle types among the fractions. In the first fraction, H Series Frac.H1(a), organic particles were identified. The second fraction, H Series Frac.H2(a), contained sulfur magnesium aluminum silicate. In the third fraction (H Series Frac.H3), phosphorus-based agricultural residues and traffic-related pollutants were observed in the third fraction Frac.H3(a1, a2) and Frac.H3(b) samples. The air masses influencing the I Series rainwater samples exhibited diverse effects. In the first fraction, I Series Frac.I1(a), nitrogen-containing particles of plant origin were observed. In the second fraction, I Series Frac.I2(a), local soil particles were identified, indicating the significant impact of local pollutants on the samples. Analysis of the SEM‒EDS results for the K Series samples indicates that the first fraction, K Series Frac.K1(a), contains elevated levels of plant residues contaminated with soil. The second fraction, K Series Frac.K2, comprising two separate images (Frac.K2(a) and Frac.K2(b)), reveals the presence of unburned coal-derived particles along with trace amounts of soil elements. This suggests that these particles may also be of agricultural origin.

HYSPLIT modeling was utilized in the approach part to look at the sources and source sectors of contaminants in stormwater. Back-trajectory calculations were performed and the results are shown in Figure 8 for the A and F Series as examples. The local wind rose plots have also been calculated using the local meteorological data. For various rainfall series, the results of wind rise plots and back-trajectory computations are available (Supplementary Material 3 Figure S4). Series A precipitation was mainly caused by the central Mediterranean air masses coming from Europe and Anatolia. In this precipitation event, the air mass at 100 m was influenced by pollutants from industry and the atmosphere of the city of Antalya. Contrary to expectations, none of the fractional samples showed acidic properties. The air mass is the main reason for the absence of acid fraction samples. The high calcium carbonate content of the desert dust-bearing countries in the Middle East neutralizes rainwater in the receiving environment without precipitation. As seen in the F Series, precipitation in the Antalya region is affected by desert dust from the Middle East and Africa. Desert dust neutralizes rainwater in the atmosphere, preventing the formation of acid rain. It also causes high levels of iron and other useful minerals to be transported into the area. Wind vane data provides information about local pollutants or traffic sources. As shown in the wind rose drawings in Figure 8, the wind was under the influence of north and northwest winds during the first rain event, and under the influence of north, northwest, northeast, and south winds during the second rain event on the second day.

When the wind rose from the C Series, rain samples were analyzed. It was found that 12% of the wind came from the northeast and 11% from the south, passing over the D-400 motorway and carrying pollutants from heavy traffic. There are also quarries in the vicinity of the sampling area. The sampling area was affected by northerly winds during the collection of the D Series rainfall samples. For the E Series samples, winds were observed to blow from the northwest and south. It was observed that the air masses causing the E Series fractional samples were influenced by the air masses coming from the Middle Eastern countries. Air masses of 1,000 m were observed to arrive in the sampling area along the coast starting from Hatay. Samples of upper atmospheric air masses causing the F Series were observed to originate from North Africa and the Middle East region. No acidic fractions were found in this sample due to the highly alkaline dust. When the wind roses of the G and H Series were analyzed, it was found that they were mostly under the influence of local winds. In the case of the I Series, a severe thunderstorm was at work. This demonstrated that the fractions were impacted by the wind's various directions. The K Series rainfall samples were observed to be under the influence of south and north winds. One hundred meter air masses were observed to be from the European region. It was also under the influence of local winds.

This study sequentially sampled 10 independent rainfall events and analyzed the rain sequences for pH, conductivity, anions, and cations. In addition, the morphology, chemical composition, and particle size distribution of water-insoluble PMs in sequential samples were assessed using SEM‒EDS and a particle size analyzer. The analytical results were used to estimate the contributions of local and distant sources to the measured species concentrations as percentages of precipitation and leaching. In filter sampling, the probability of finding single particles can be reduced as the particles stick to each other and physically cover the ones below. Furthermore, examining rainwater samples collected in low quantities or diluting rainwater may result in a higher number of individual particles. Local and remote pollution source regions were discovered by assessing the efficiency of in-cloud and sub-cloud washing processes in rainfall events impacting the Antalya (Mediterranean) region of Turkey. Due to the arid nature of the Antalya province in Turkey and the influence of desert dust from the Middle East and Africa, acid rain was not found in any of the precipitation fractions. As the rains persisted, changes in the sampling area dropped and then increased once more, typically as a result of variations in the local wind's direction or speed (PM).

In addition, there is a high prevalence of primary particles of biological origin in rainfall, as well as in particles emitted by vehicles. Other studies in the literature on air pollution transport have been on filter collection and analysis. One effective method for determining the origin of industrial insoluble particles is to characterize the insoluble particles present in rainfall. Rainwater can carry the majority of industrial particulates straight from their source. During the analysis of the precipitation series, the average ∑ anion/∑ cation equivalent ratios were determined for specific fractions. Upon evaluation of the obtained data, the ratio was found to be 0.77 ± 0.30 for the A series, 0.58 ± 0.21 for the B series, and 0.77 ± 0.55 for the C series from precipitation events. The values were calculated as 1.02 ± 0.18 for the D series, 0.88 ± 0.33 for the E series, and 1.05 ± 0.35 for the F series. Additionally, for the G, H, I, and K series, the ratios were determined as 0.49 ± 0.10, 0.62 ± 0.32, 0.65 ± 0.00, and 0.32 ± 0.24, respectively. High concentrations of Ca²⁺ and Mg²⁺ ions in summer are caused by local dust and dust input (dust from North Africa or Middle East areas) in Antalya Province, Turkey in the warm seasons. Analysis of the airmass on the day and time of the precipitation sampling showed that both Anatolian and Mediterranean-North African airmass (with desert dust) influenced the precipitation event. High quantities of ⁠, ⁠, and were found, which was evidence for contributions primarily from anthropogenic sources. High Na⁺ and Cl⁻ values in the rainwater composition during the winter months indicate sea salt inputs. The results of the SEM‒EDS analyses. When combined with other chemical analyses, HYSPLIT return trajectory calculations offer insight into the origins of the air masses during rainfall and the types of contaminants that are carried to the receiving environment.

A multitude of studies can be conducted at a sampling point such as Antalya province, which encompasses a plethora of both tourist attractions and agricultural regions. The objective is to ascertain the sources of rainwater pollution on an annual basis. A further recommendation is that the results obtained from the analysis of rainwater samples should be evaluated on a regular basis, with particular attention paid to the seasonal monitoring of pollution in the rainwater that falls in Antalya. It may be necessary to conduct regular studies on the impact of pollutants transported to agricultural areas on groundwater used as a source of drinking water.

This study was supported by the Scientific and Technological Research Council of Turkey (TUBITAK) under Grant Number 123Y072. The authors thank TUBITAK for their support.

M.K. Corresponding author, Investigation, Methodology, Supervision, Validation, and Writing – original draft; S.K. Validation, Data curation, Resources, Writing – original draft, and Writing – review & editing. All authors read and approved the final manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.