Urban Single Precipitation Events: A Key for Characterizing Sources of Air Contaminants and the Dynamics of Atmospheric Chemistry Exchanges

Abstract

The chemistry of atmospheric precipitation serves as an important proxy for discriminating the source(s) of air contaminants in urban environments as well as to discuss the dynamic of atmospheric chemistry exchanges. This approach can be undertaken at time scales varying from single events to seasonal and yearly time frames. Here, we characterized the chemical composition of two single rain episodes (18 July 2018 and 21 February 2019) collected in Wrocław (SW Poland). Our results demonstrated inner variations and seasonality (within the rain event as well as between summer and winter), both in ion concentrations as well as in their potential relations with local air contaminants and scavenging processes. Coupling statistical analysis of chemical parameters with meteorological/synoptic conditions and HYSPLIT back trajectories allowed us to identify three main factors (i.e., principal components; PC) controlling the chemical composition of precipitation, and that these fluctuated during each event: (i) PC1 (40%) was interpreted as reflecting the long-range transport and/or anthropogenic influences of emission sources that included biomass burning, fossil fuel combustion, industrial processes, and inputs of crustal origin; (ii) PC2 (20%) represents the dissolution of atmospheric CO₂ and HF into ionic forms; and (iii) PC3 (20%) originates from agricultural activities and/or biomass burning. Time variations during the rain events showed that each factor was more important at the start of the event. The study of both SO₄²⁻ and Ca²⁺ concentrations showed that while sea spray inputs fluctuated during both rain events, their overall impact was relatively low. Finally, below-cloud particle scavenging processes were only observed for PM₁₀ at the start of the winter rain episode, which was probably explained by the corresponding low rain intensity and an overlap from local aerosol emissions. Our study demonstrates the importance of multi-time scale approaches to explain the chemical variability in rainwater and both its relation to emission sources and the atmosphere operating processes.

1. Introduction

The chemical composition of precipitation is an important issue in many regions worldwide as it causes eutrophication, affects ecosystems when leading to acid rain, and can contribute to global climate change [1,2]. The chemical composition of rainwater depends on the particulate or gaseous atmospheric constituents that are emitted locally or transported from distant sources by natural or anthropogenic means [3]. Precipitation is considered one of the most efficient scavengers for removing particulate matter and gaseous pollutants from the atmosphere [2], and its efficiency reflects into its chemical composition and pH conditions [4,5]. Scavenging mechanisms occur as in-cloud scavenging (ICS) or below-cloud scavenging (BCS) as rainout [6,7,8]. While the ICS process takes place within the condensation nuclei inside the cloud droplets, BCS consists of the washing out of aerosols by falling raindrops and/or ice crystals [7]. The initial rain fractions in the washout process mostly remove local anthropogenic origin pollutants [7], whereas during rainout, the chemical composition of precipitation may be linked to events from more distant regions [9].

One of the still outstanding questions regarding the chemical interpretation of precipitation is the time scale of the event, which can be interpreted as monthly/daily averages, a single episode of rain (regardless of the episode duration), or even the short-time monitoring of chemical changes during a single precipitation event. Each observation scale may bring a different set of answers. The monitoring of precipitation chemistry has been previously reported for as long as 20 years, e.g., [6,10,11,12,13,14] as well as for studies based on single rainfall episodes at short time scales, e.g., [15,16,17,18,19,20]. To our best knowledge, single rainfall episode studies have yet to be undertaken in Poland, most of them having considered yearly or multi-year time scales. Moreover, the available single rainfall studies focused mostly on gas–ion–solid phase interactions with the objective of discussing the chemistry and meteorology of the upper atmosphere.

Many factors can affect the chemical composition of rainwater [2,21,22]: local air contaminants, meteorological parameters, location, and processes controlling the transport of air contaminants. Local anthropogenic (urban/industry) sources, mostly through the emission of gaseous SO₂/NO_x, influence the ion composition of rainwater [3,23,24,25]. Natural inputs can also contribute (e.g., coastal Na⁺/Cl⁻-rich sea spray; [26,27,28,29]). Refs. [30,31] showed that the canopy throughfall effect is another natural factor that can affect the chemistry of precipitation at a short event scale. Biomass or forest burning can also influence the chemical composition of rain [32,33]. Finally, changes in the chemistry of precipitation resulting from its interaction with aerosols from various emission sources, e.g., dust, volcano plumes, and soil dust, have also been reported [7,29,34,35,36]. At wider regional and global scales, recent studies have used the chemistry of precipitation, including acidity, to identify the responsible pollution sources and apportion their respective contributions [37,38,39].

Official environmental surveys monitoring precipitation chemistry (e.g., the European Monitoring and Evaluation Programme [EMEP]) provide large-scale and long-term data, giving insights into historical changes. In Poland, the Institute of Meteorology and Water Management (IMWM) along with the Chief Inspectorate of Environmental Protection (CIEP) publishes annual reports documenting the atmospheric pollutant data from selected monitoring stations. Studies focusing on the chemical composition of precipitation have been conducted in Poland [13,14,40,41], including in the city of Wrocław [2,42,43,44,45,46]. In most cases, multivariate statistics, such as principal component analysis (PCA), can provide additional information about the processes and sources affecting the overall chemical composition of precipitation [47,48] or the temporal variations in each PCA principal component [2,49].

Here, this study focuses on the analysis of precipitation chemistry during two single rain episode events (one in summer: 18 July 2018, and one in winter: 21 February 2019) and, to our knowledge, represents a pioneer attempt at linking it to its potential dynamic exchange with urban air contaminants (both gaseous and particle phases). Using statistical analysis (PCA/Spearman rank correlation), our objectives were (i) to identify the sources of the ions present in precipitation and (ii) to evaluate how air contaminants emitted by local emission sources (industry, road traffic, and municipality) may impact the chemistry of atmospheric precipitation.

2. Study Area

Wrocław is the capital city of the Lower Silesia Province in southwestern Poland (Figure 1). It has an urban population of approximately 674,000 inhabitants, with an average population density of 2302 inhabitants per square kilometer as of 31 December 2022. Wrocław is located within a transitional climate zone: it undergoes a large variability in its weather conditions caused by the clash of maritime and continental masses [50]. The climate is shaped by the constantly flowing Icelandic Low and the Azores High, the summer South Asian Low, and the winter Asian High. The way urbanized areas develop affects the bioclimatic diversity, precipitation, and the formation of “heat islands” [51,52]. These modifications are characteristic of large urban and industrial agglomerations. The variability in weather conditions is associated with the frequent and active movement of baric systems, the advection of humid air masses from the Atlantic Ocean, and the influx of dry air masses from the Asian continent [50]. Wrocław is characterized by low precipitation (~167 days/year), with an annual average of 555 mm/year for the period 2016–2020, similar to the average precipitation recorded in the vicinity of its urban area.

Air quality in Wrocław is mostly affected by contaminant emissions from the industrial and energy sectors, as well as by low emission sources that include local road traffic and the municipal and housing sectors [45]. The latter is of particular importance in areas characterized by a dense residential development. The energy, industry, municipal, and housing sectors are sources of air contamination as most of them are relying on the combustion of solid, liquid, and gaseous fuels that release compounds such as sulfur dioxide (SO₂), nitrogen oxides (NO_x), carbon monoxide (CO), and particulate matter (PM) [53].

Wrocław is situated at the junction of two deep geological units: the Fore-Sudetic block and the Fore-Sudetic monocline [54]. The Fore-Sudetic block, southwest of the city, consists of Protero-Palaeozoic crystalline rocks, partly exposed at the surface. The Fore-Sudetic monocline lies north of the city and consists of sedimentary rocks that are not exposed at the surface, overlaid by thick Cenozoic sediments. Loess as well as aeolian sands extend across the northern area, whereas the youngest Holocene deposits, consisting of sands, silts, and peat bogs, are mainly found on the flood terraces of the Odra River and smaller watercourses [55]. Wrocław’s land use reflects a balanced integration of historic background with residential, commercial, industrial, and recreational functions. Industrial land use is concentrated on the outskirts, with manufacturing, logistics, and tech-related industries dominating. Furthermore, Wrocław’s urban planning integrates green corridors (e.g., parks, waterfront areas). Agricultural land remains outside the city, particularly in less urbanized districts. However, these areas are gradually being converted for residential and industrial purposes due to urban expansion.

3. Materials and Methods

3.1. Air Contaminants in Wrocław Area

Atmospheric concentrations for the following contaminants, SO₂, PM_2.5, PM₁₀, CO, O₃, NO_x, NO₂, and NO (Table S1), were obtained from the official CIEP monitoring station (Wyb. J.Conrada-Korzeniowskiego 18 St Wrocław: 51.129378 N, 17.02925 E) for 18 July 2018 and 21 February 2019 (Figure 1, Figure 2 and Figure 3). All data were acquired using the QA/QC reference methodology detailed in the Annex VI of the CAFE Directive (2008/50/EC), from the Environmental Protection Law (Official Gazette 2556/2022). The air quality monitoring station operated by the Chief Inspectorate of Environmental Protection (GIOŚ) is located approximately 6 km north-west of the site where precipitation was collected (University of Wrocław). Both stations are typical of urban background, far from major emission sources, primarily impacted by point sources and dense road traffic. Data from the GIOŚ station were thus considered as representative of the city’s background air pollution.

3.2. Meteorological Conditions in Wrocław Area

Meteorological data (atmospheric pressure, air temperature, relative humidity, wind direction and speed, atmospheric precipitation) were obtained from the monitoring station of the Department of Climatology and Atmosphere Protection (University of Wrocław) in Wrocław (Table S2). A Vaisala HMP45 thermo-humidity probe was used to measure air temperature and relative humidity. Atmospheric pressure was measured with a Vaisal PTA427 barometer. Wind speed and direction were measured using a GILL Windsonic75 ultrasonic anemometer. Measurements were made with a resolution of up to one minute. Atmospheric precipitation was measured using an OTT Parsivel2 laser disdrometer, with a resolution of up to one minute. Precipitation totals were only quantified for the second episode (21 February 2019) due to a failure of the disdrometer during the summer campaign. However, precipitation totals were also obtained from the Institute of Meteorology and Water Management (IMWM) at the Strachowice station in Wrocław for both episodes. These are data with a resolution of up to one minute. Any difference between the data from DCAP (UWr) and IMWM may be attributed to the 15 km distance between the two monitoring stations (Figure 1).

Synoptic charts were prepared using the Royal Netherlands Meteorological Institute model (KNMI; https://www.knmi.nl; KNMI is not involved in this derivative work and does not necessarily endorse its findings). The air mass back trajectories were modeled using the NOAA HYSPLIT model [56,57] (accessed online 29 March 2023). For both sampling periods, 48 h air mass back trajectories were calculated at 6 h intervals at 500/1000/1500 m above ground-level heights. Wind roses were created using Golden Software Grapher 15.0 (Figure 2 and Figure 3), by processing data collected by the Department of Climatology and Atmosphere Protection (University of Wrocław).

3.3. Precipitation Collection

A total of 15 precipitation samples were collected during consecutive summer (18 July 2018) and winter (21 February 2019) seasons. In summer, 2018, the first precipitation sample was collected at 6:00 a.m. and the last one at 2:45 p.m. (UTC time). In winter, sample collection took place from 8:00 a.m. to 2:50 p.m. (UTC time). The duration of sample collection was dictated by rain intensity: during intense rainfalls, samples were taken once the collector was fully filled, whereas during less intense rainfalls, samples were collected after 30 min unless stated otherwise (Table S1, Supplementary Materials). The rain collector was located on the premises of the Department of Climatology and Atmosphere Protection (University of Wrocław; 51°06′19.0″ N and 17°05′20.0″ E). The rain collector consisted of a stainless-steel funnel with a diameter of 1 m and a height of 1.5 m (Figure 1). Before sampling, the funnel was thoroughly cleaned and washed several times with ultrapure (Milli-Q^®) water to ensure that particles that may have deposited onto the surface of the collector were fully eliminated. The fallout was directly collected into 100 mL PE containers. Samples were stored in a freezer until further chemical analysis to avoid any oxidation, reduction, or precipitation processes.

3.4. Determination of Physicochemical Parameters in the Precipitation Samples

Precipitation samples were analyzed for their physicochemical parameters. pH and electrical conductivity (EC) were measured using an Elmetron CX-505 multifunctional device, with precisions of 0.01 for pH and 0.01 µS·cm⁻¹ for EC. Titration was carried out to determine the presence of bicarbonates in the samples: alkalinity was determined on a 25 mL aliquot using 0.01 M HCl in the presence of methyl orange. Alkalinity, expressed in meq units, was converted to mg·L⁻¹ using HCO₃⁻ gram-equivalent. Precision was better than ±0.1 mg·L⁻¹. Anion and cation concentrations were measured by chromatography on a Dionex DX-120 ion chromatograph. Concentrations of F⁻, Cl⁻, NO₂⁻, NO₃⁻, PO₄³⁻, and SO₄²⁻ were measured using an 8 mM sodium carbonate/1 mM sodium bicarbonate eluent and a Dionex Ion Pac TM AS14A column with respective detection limits of 0.05 mg·L⁻¹ for NO₃⁻, 0.02 mg·L⁻¹ for NO₂⁻, and 0.01 mg·L⁻¹ for the remaining ions. Calibration of the instrument was performed by running the Dionex Seven Anion Standard II Thermo Fisher Scientific (cat. no. 057590) and LGS chloride standard (cat. no. 87799-5). Concentrations of Na⁺, NH₄⁺, K⁺, Mg²⁺, and Ca²⁺ were measured using a 20 mM methanesulfonic acid eluent and a Dionex Ion Pac TM CS12A column with detection limits of 0.01 mg·L⁻¹. The Merck Multi-Element Standard VII (cat. no. 1.10322.0100) was used for calibration. Each rainfall sample was analyzed in duplicate, and the final result was calculated as the average of two measurements. The obtained RSD% for anions ranged from 0.23 to 4.54 and from 0.21 to 2.92 for cations. During the analysis, 5 mg·L⁻¹ standards were used as control samples and ultrapure water as blank samples. For control samples, the average RSD% of 0.33 for anions and 0.38 for cations were obtained, whereas no chromatographic signal was detected for ultrapure water.

3.5. Statistical Analysis

Spearman rank correlations (missing data removed in pairs; p < 0.05) were made separately for the meteorological and chemical parameters of each precipitation sample collected during the summer and winter rain episodes, in order to establish interrelationships between variables (

Table 1. Spearman rank-order correlation obtained for the meteorological and chemical parameters measured in precipitations collected during the summer 18 July 2018 rain episode. Missing data were removed in pairs. Bold coefficients are significant at p < 0.05.

	P avg [hPa]	O3 avg [µg·m−3]	PM10 avg [µg·m−3]	PM2.5 avg [µg·m−3]	T2m avg [°C]	RH avg [%]	V avg [m·s−1]	RIMWM_sum [mm]	pH [−log [H+]]	EC [µS·cm−1]	HCO3− [mg·L−1]	F− [mg·L−1]	Cl− [mg·L−1]	NO2− [mg·L−1]	NO3− [mg·L−1]	PO43− [mg·L−1]	SO42− [mg·L−1]	Na+ [mg·L−1]	NH4+ [mg·L−1]	K+ [mg·L−1]	Mg2+ [mg·L−1]
O3 avg [µg·m−3]	0.90
PM10 avg [µg·m−3]	0.79	0.81
PM2.5 avg [µg·m−3]	0.17	0.16	0.39
T2m avg [°C]	0.56	0.30	0.38	0.24
RH avg [%]	−0.83	−0.86	−0.80	−0.26	−0.54
V avg [m·s−1]	0.81	0.65	0.59	0.25	0.56	−0.65
RIMWM_sum [mm]	−0.06	0.16	0.25	0.19	−0.28	−0.06	−0.06
pH [−log [H+]]	−0.29	−0.26	−0.43	−0.34	−0.02	0.29	−0.42	−0.37
EC [µS·cm−1]	−0.07	−0.05	−0.11	−0.16	0.18	0.11	−0.11	−0.40	0.82
HCO3− [mg·L−1]	−0.31	−0.50	−0.23	−0.38	0.03	0.36	−0.34	−0.48	0.50	0.44
F− [mg·L−1]	0.36	0.40	0.27	−0.09	0.12	−0.14	0.41	−0.11	0.21	0.62	0.02
Cl− [mg·L−1]	0.13	0.22	0.11	0.28	−0.07	−0.17	0.11	−0.24	0.03	0.10	−0.10	0.33
NO2− [mg·L−1]	0.14	0.14	0.31	0.09	0.26	0.14	0.31	−0.09	0.31	0.67	0.41	0.77	0.61
NO3− [mg·L−1]	0.83	0.78	0.68	0.09	0.56	−0.65	0.70	−0.17	0.08	0.44	−0.15	0.70	0.14	0.66
PO43− [mg·L−1]	0.01	−0.07	0.10	0.32	0.28	0.07	0.25	0.41	−0.55	−0.35	−0.51	−0.06	−0.27	−0.10	−0.01
SO42− [mg·L−1]	0.29	0.21	0.17	−0.01	0.25	−0.03	0.54	−0.10	0.11	0.51	0.11	0.89	0.23	0.83	0.56	0.11
Na+ [mg·L−1]	0.00	0.08	−0.14	−0.08	−0.10	0.00	0.31	−0.07	0.10	0.22	−0.11	0.46	0.53	0.83	0.14	−0.16	0.55
NH4+ [mg·L−1]	0.21	0.28	0.40	0.09	0.31	−0.25	0.36	0.21	0.08	0.48	−0.05	0.63	0.02	0.77	0.58	0.34	0.60	0.16
K+ [mg·L−1]	−0.08	0.05	0.00	−0.21	−0.50	0.23	−0.28	−0.03	0.38	0.48	0.25	0.57	0.40	0.70	0.18	−0.44	0.27	0.06	0.30
Mg2+ [mg·L−1]	−0.46	−0.46	−0.46	0.46	0.39	0.56	0.05	−0.29	0.10	0.53	0.05	0.36	0.67	0.32	0.10	−0.20	0.67	0.82	0.41	0.21
Ca2+ [mg·L−1]	−0.83	−0.72	−0.67	0.44	0.02	0.73	−0.37	−0.48	0.48	0.57	0.13	0.06	0.41	0.20	−0.22	−0.31	0.24	0.44	−0.02	0.14	0.97

and

Table 2. Spearman rank-order correlation obtained for the meteorological and chemical parameters measured in precipitations collected during the winter 21 February 2019 rain episode. Missing data were removed in pairs. Bold coefficients are significant at p < 0.05.

	P avg [hPa]	O3 avg [µg·m−3]	PM10 avg [µg·m−3]	PM2.5 avg [µg·m−3]	T2m avg [°C]	RH avg [%]	V avg [m·s−1]	RIMWM_sum [mm]	pH [−log [H+]]	EC [µS·cm−1]	HCO3− [mg·L−1]	F− [mg·L−1]	Cl− [mg·L−1]	NO2− [mg·L−1]	NO3− [mg·L−1]	PO43− [mg·L−1]	SO42− [mg·L−1]	Na+ [mg·L−1]	NH4+ [mg·L−1]	K+ [mg·L−1]	Mg2+ [mg·L−1]
O3 avg [µg·m3]	−0.30
PM10 avg [µg·m−3]	0.79	0.00
PM2.5 avg [µg·m−3]	0.64	−0.55	0.70
T2m avg [°C]	0.06	0.83	0.31	−0.26
RH avg [%]	−0.55	−0.47	−0.74	−0.29	−0.64
V avg [m·s−1]	0.21	0.49	0.36	0.13	0.41	−0.53
RIMWM_sum [mm]	−0.46	−0.49	−0.56	−0.18	−0.49	0.87	−0.63
pH [−log [H+]]	0.51	0.16	0.72	0.50	0.25	−0.69	0.21	−0.49
EC [µS·cm−1]	0.50	0.37	0.48	0.17	0.32	−0.77	0.54	−0.90	0.63
HCO3− [mg·L−1]	0.55	0.42	0.74	0.31	0.51	−0.98	0.45	−0.86	0.89	0.90
F− [mg·L−1]	0.55	0.36	0.56	0.14	0.44	−0.88	0.38	−0.79	0.59	0.80	0.90
Cl− [mg·L−1]	0.49	0.46	0.56	0.19	0.44	−0.88	0.55	−0.89	0.74	0.93	0.96	0.87
NO2− [mg·L−1]	0.46	0.43	0.55	0.21	0.40	−0.84	0.52	−0.84	0.75	0.91	0.95	0.89	0.98
NO3− [mg·L−1]	0.29	0.30	0.22	0.13	0.27	−0.63	0.58	−0.85	0.24	0.82	0.60	0.64	0.71	0.69
PO43− [mg·L−1]	0.41	0.57	0.61	0.14	0.60	−0.88	0.71	−0.90	0.53	0.81	0.87	0.77	0.84	0.79	0.72
SO42− [mg·L−1]	0.53	0.43	0.55	0.18	0.47	−0.90	0.57	−0.94	0.61	0.95	0.93	0.89	0.95	0.92	0.83	0.87
Na+ [mg·L−1]	0.48	0.47	0.52	0.10	0.45	−0.84	0.49	−0.86	0.66	0.92	0.90	0.92	0.96	0.96	0.71	0.83	0.92
NH4+ [mg·L−1]	0.33	0.38	0.27	0.03	0.38	−0.67	0.57	−0.85	0.27	0.85	0.65	0.69	0.74	0.72	0.96	0.77	0.86	0.76
K+ [mg·L−1]	0.88	−0.59	0.90	0.80	−0.07	−0.80	−0.02	−0.74	0.78	0.88	0.90	0.81	0.83	0.87	0.63	0.58	0.83	0.88	0.63
Mg2+ [mg·L−1]	0.80	0.42	0.82	0.49	0.19	−0.77	0.27	−0.84	0.92	0.92	0.92	0.84	0.97	0.94	0.61	0.79	0.89	0.94	0.64	0.81
Ca2+ [mg·L−1]	0.78	0.36	0.88	0.53	0.32	−0.84	0.61	−0.86	0.83	0.95	0.93	0.78	0.88	0.87	0.75	0.83	0.90	0.82	0.79	0.89	0.94

for summer and winter, respectively). To reduce the complex correlation matrices to a limited number of uncorrelated factors for both precipitation episodes, a multivariate principal component analysis (PCA) was conducted. PCA is a proven statistical method to characterize the spatio-temporal variability in large data populations, e.g., [58,59,60]. PCA allows defining the principal components (PC) that describe the variability in the analyzed population, with the objectives here to (i) analyze the variability in the winter and summer precipitation events simultaneously, (ii) discriminate the processes controlling their respective chemical compositions, (iii) determine the respective contributions of these processes. PCA was carried out on 16 standardized parameters: PM₁₀, PM_2.5, pH, EC, HCO₃⁻, F⁻, Cl⁻, NO₂⁻, NO₃⁻, PO₄³⁻, SO₄²⁻, Na⁺, NH₄⁺, K⁺, Mg²⁺, and Ca²⁺. During the PCA analysis, three mutually uncorrelated factors or principal components (i.e., processes) were isolated in the factor loadings that can be read as correlation coefficients. On the other hand, factor scores help estimate the intensity of these processes: extreme negative values (factor scores < −1) indicate that the process(es) that each PC represents do(es) not contribute, whereas positive values (factor scores > +1) indicate they strongly contribute. Values between −1 and +1 define the mixing zone of individual processes, e.g., [60,61]. Here, the first three PCs explained 85.5% of the observed chemical variations. The remaining 14.5% were considered random noise that cannot be interpreted using this technique, e.g., [60,62].

4. Results and Discussion

The chemical composition of atmospheric precipitation showed clear seasonality (winter vs. summer) as well as inner variability at the scale of each single rain episode (Figure 2 and Figure 3).

4.1. Summer Rain Episode

During the summer (18 July 2018) precipitation episode, an increase in precipitation was observed from ~06 a.m. on, to reach a maximum at ~11 a.m., with a subsequent decrease, followed by a second slight peak around 12:30 p.m. The precipitation event ended at around 3 p.m. (Figure 2A). Corresponding HYSPLIT back trajectories showed that air masses were generally originating from the north-east and north directions at an average altitude of ~500–1000 m above ground level, rising in their last phase to an altitude of 500–1500 m above ground level (Figure 4A). At that date, Poland was within the range of a low-pressure system centered over southern Russia, with a zone of cloudiness and precipitation moving from the east (Figure 4B). The cloud system was directly related to the atmospheric front separating two polar maritime and tropical masses. The synoptic situation contributed to the rapid development of convection and, subsequently, of thunderstorms and showers.

The beginning of the precipitation event was accompanied by a decrease in temperature of ~1 °C and an increase in the average wind speed [63] from ~1 to 4 m·s⁻¹ (Figure 2A). This resulted in the washing out of atmospheric PM₁₀ and PM_2.5, which was recorded at the CIEP station during the rain episode, characterized by two- and four-fold decreases in the local PM concentrations, respectively (Figure 2B). Similar to the findings of [64], we observed that the washing-out effect was more effective for the coarse (PM₁₀) than for the fine (PM_2.5) fraction (Figure 2B). The onset of the aerosol scavenging was accompanied by a significant short-term peak in the concentrations of most anions and cations in the precipitation (Figure 2C,D). This agrees with what was previously reported for whole rain episodes [7,36] that the authors described as the below-cloud scavenging of aerosols and in-cloud processes, where sulfates, nitrates, and ammonium ions dominate. The atmospheric CO₂ decrease with the increasing precipitation that we observed (Figure 2A,C) may suggest the formation of bicarbonate ions through the dissolution of this CO₂. Moreover, the absence of clear statistical relations between the bicarbonate and PM₁₀/PM_2.5 concentrations (Table 1) seemed to exclude any inputs of CaCO₃-rich particles coming from resuspended soil dust or building material [3,24]. However, ref. [41], studying the carbon isotope compositions (δ¹³C) of the CO₂/HCO₃⁻ system in the atmosphere of Wroclaw, concluded that local CO₂ was not responsible for the presence of bicarbonates in the rain. We thus hypothesized that the formation of bicarbonate ions in the rain was linked to (i) interaction with remote CO₂ generated outside the city and/or (ii) PM_x aerosol scavenged earlier and without any link to the local ones already present in the atmosphere of the city.

The second, less pronounced peak in anion concentrations occurred when the precipitation intensified again, around 1–2 p.m. (Figure 2A). It is noteworthy that ammonium exhibited random fluctuations that were not related to either the duration of the fallout or the PM₁₀/PM_2.5 concentrations (Figure 4D). Ammonium and bicarbonate are the ions that buffer the pH in rainwater [65]. While we observed a positive correlation between the pH and HCO₃⁻ and NH₄⁺ for the winter episode (Table 1), we did not for the summer episode (Table 2). There was also no relationship between the atmospheric SO₂ and SO₄ concentrations in the precipitation, which may exclude the rapid formation of SO₄ through the oxidation of SO₂ from local emission sources (Figure 2C,D). Ref. [43] observed a similar trend during a longer two-week rain episode. The authors concluded that while SO₄ at the beginning of the rain episode originated from high-temperature combustion taking place outside of Wrocław, the final sulfate S (δ³⁴S) and O (δ¹⁸O) stable isotope compositions (after two weeks of rain) indicated that SO₄ resulted in secondary low-temperature SO₄ derived from the interaction between local SO₂ and meteoric water.

Nitrate ions displayed an interesting trend: in addition to two small concentration peaks that were consistent with those of the other ions (Figure 4C), they presented a sharp increase at the end of the rain episode from 12 a.m. to 3 p.m. This increase may have resulted from the dissolution of NOx that presented concomitant fluctuations. These were coupled with a rapid increase in the atmospheric O₃ concentrations (Figure 2B), which can be related to an increase in solar radiation that enhanced forming reactions with ozone precursors.

Furthermore, the Spearman rank correlation matrix (Table 1) between the atmosphere and precipitation chemistry identified significant positive correlations between chlorides/sulfates and sodium, and nitrites/nitrates/sulfates and ammonium (Table 1). It follows that NaCl, Na₂SO₄, (NH₄)₂SO₄, NH₄NO₃, and NH₄NO₂ salts may have represented ion precursors during the scavenging processes occurring during the summer precipitation.

4.2. Winter Rain Episode

The winter episode (21 February 2019) started with an increase in precipitation at ~8 a.m. and peaking at ~3 p.m. The first phase (8 a.m. to 12 a.m.) was characterized by light precipitation, followed by intense rain between 1 and 3 p.m. (Figure 3A). Unfortunately, the second evening rain episode (from 8 p.m. on, Figure 3A) was not sampled. The 48 h HYSPLIT back trajectories (Figure 4C) showed that air masses were generally arriving from the west at an average altitude of ~1000–2000 m above ground level, in agreement with the European baric conditions at that time (Figure 4D). Air temperatures were still quite high for the season, at 7 °C, as the synoptic conditions in Poland were influenced by two low-pressure and two high-pressure systems located over Europe (Figure 4D). Such a pressure system was favorable for frontogenesis, forming a vast area of clouds and precipitation. The gradual development of the high-pressure system resulted in a short-term improvement in the weather during that day, but in the evening, atmospheric fronts bringing heavy rainfalls formed again over Poland. When precipitation started, the temperature slightly decreased, but during the rain event, the temperature increased by ~2 °C and the average wind speed increased from ~2 to 3 m·s⁻¹ (Figure 3A). In stark contrast to the summer rain episode, the washing out of PM₁₀ and PM_2.5 was not discernable when the rain started, especially for PM_2.5 (Figure 3B). This was probably linked to both the low rain intensity during that first stage and the overlap created by aerosol emissions from local municipal and domestic sources, though the PM_x concentrations started decreasing when the rain became more intense 3–4 h later (Figure 3B).

Interestingly, all anions in the rainwater, except nitrate, showed declining concentrations when the rain event started (Figure 3C). This may be explained by either (i) the leaching of primary anions brought about by the fallout or in-cloud scavenging processes from long-distance transport and/or (ii) the dilution of the originally enriched ions (resulting in below-cloud scavenging and the dissolution of local PM_x) by incoming successive precipitation masses that were characterized by low ion concentrations. While the second hypothesis may have been supported by the observed decrease in the PM₁₀ concentrations, we did not observe a concomitant PM_2.5 washout (their concentrations even increased; Figure 3B). This most likely eliminated this hypothesis, in agreement with the findings of [59]. However, the coupled variations in the atmospheric PM_2.5, dissolved nitrate in rain, and ammonium, sodium, and potassium indicated that they all reacted to form secondary fine nitrate salts (NH₄NO₂, NaNO₂, KNO₂) in precipitation. The statistically significant negative correlations between most of the dissolved ions in the precipitation and pH (Table 2) confirmed that they controlled the pH in winter, in agreement with [65]. As for summer, we observed an inverse relationship between the atmospheric NO_x and O₃ concentrations (Figure 3B), which may be attributed to rapid NO/NO₂/O₃ interactions during the precipitation events. However, no relationship was found between the local gaseous NO_x, SO₂ and dissolved nitrite, nitrate, and sulfate. Similarly to the summer rain episode and the findings of [43], this mostly excluded a local origin for these ions (i.e., from the dissolution of local anthropogenic gases).

4.3. Discussing Links Between Local Air Contamination and Rainwater Chemistry

The main objectives of this study were to (i) evaluate the variability in dissolved ion concentrations during and between single rain episodes and (ii) to determine their eventual link to local air contaminants (both aerosols and gases) through dissolution processes. This was motivated by the fact that the Wrocław area is highly impacted by intensive industrial and energy activities that release air contaminants that include SO₂, NO_x, CO, and PM_x (mostly stemming from fuel combustion [53]). Previous short rain episode studies [15,16,17,18,19] focused on the control of scavenging processes by meteorological factors and/or simulated possible factors that were determining the final precipitation chemistry. Others, including [20], showed the dependence of the precipitation chemistry variability to meteorological factors. These latest authors concluded that sea salt (ss) ions had been transported to North Carolina (USA) by Hurricane Irene in 2011, and that they had been subsequently washed out as both the precipitation and hurricane weakened.

Here, we calculated the concentrations of non-sea salt (nss) sulfate and calcium ion concentrations in rainwater using the following equations (results reported in Figure 5):

SO₄²⁻ nss = SO₄²⁻_measured − (SO₄²⁻/Na⁺)_ss × Na⁺_measured

(1)

and

Ca²⁺ nss = Ca²⁺_measured − (Ca²⁺/Na⁺)_ss × Na⁺_measured

(2)

where the sulfate and calcium to sodium ratios of seawater are (SO₄²⁻/Na⁺)_ss = 0.252 and (Ca⁺²/Na⁺)_ss = 0.0384, respectively [66].

In summer (Figure 5A,B), a rapid discharge of sea salts was observed at the start of precipitation, when ss SO₄²⁻ were 23.49% and Ca²⁺ 14.06%. In winter, the discharge of ss ions lasted longer (Figure 5C,D), with respective contributions of 5.13–15.35% for SO₄²⁻ and of 1.86–5.88% for Ca²⁺. These showed that dissolved ions in the Wrocław precipitation were mostly of anthropogenic origin. This conclusion is also corroborated by the previous isotope study of dissolved SO₄²⁻ δ³⁴S and δ¹⁸O in Wrocław precipitation [43,45] that showed that sea spray was not a significant source of that ion. Moreover, we did not observe any relationship between the atmospheric SO₂ concentration and dissolved SO₄²⁻ ions in precipitation. This mostly excluded both its local origin and rapid exchange at the gas–ion level in the local urban atmosphere. It followed that dissolved sulfates in Wrocław most likely resulted in a mixing between primary and secondary sulfates that had been formed during the transport of air masses from remote emission sources to the city (in agreement with [43,45]).

The PCA analysis (Figure 6) indicated that the first three principal components (PC) allowed explaining 85% of the dataset total variability. PC1 (45%) was highly positively influenced by the following parameters (e.g., loadings > 0.5, from highest to lowest): Cl⁻, Mg²⁺, Ca²⁺, SO₄²⁻, EC, NO₂⁻, PO₄³⁻, PM₁₀, and pH (0.44). PC2 (20%) was influenced by HCO₃⁻ and F⁻ (positive loadings >0.5) and negatively correlated to PM_2.5 (−0.74). PC3 (20%) was mostly influenced by the positive loadings of K⁺ (0.91), and moderately by NH₄⁺ (0.74), NO₃⁻ (0.72), and Na⁺ (0.69).

The parameters that PC1 was positively correlated to, and to a lesser extent, to the pH may be interpreted as reflecting in situ PM₁₀ dissolution. This hypothesis was supported by the time variations in the PC1 factor scores (Figure 6) that showed that significant scores (i.e., >1) were only observed at the start of the winter rain episode, characterized by high aerosol concentrations in the city’s atmosphere. This was also confirmed by the presence of short-life nitrite ions that were rapidly oxidized into nitrates at the start of the winter precipitation (Figure 3C). Coupled variations in the atmospheric O₃ and NO₂ concentrations and in NO₂⁻ and NO₃⁻ in rainwater (Figure 3) indicate an increase in NO₂⁻ concentrations in rainwater that may result in the oxidation of atmospheric NO₂. During precipitation, the NO₂⁻ concentrations in rainwater, whereas the NO₃⁻ concentrations rapidly increase during the initial phase of the event. The primary oxidation mechanism in the atmosphere involves ozone, when O₃ reacts with NO₂ to form nitrate radicals (NO₃), which are further oxidized into nitric acid (HNO₃) or hydrolyzed into NO₃⁻ in the aqueous phase. These reactions typically occur at night in the absence of photolytic activity, which stabilizes NO₃ radicals and facilitates their conversion [67]. It can be assumed that cloudy winter weather promotes similar conditions, enhancing this process. Moreover, the photostationary state between NO and NO₂ is disrupted by increasing O₃ levels, reducing the NO₂ concentrations and shifting the equilibrium toward oxidized nitrogen species, such as NO₃⁻. This explains the observed decrease in atmospheric NO₂ and the concurrent rise in NO₃⁻ concentrations in precipitation.

In summary, PC1 was likely related to the long-range transport and/or anthropogenic influence of emission sources that included biomass burning, fossil fuel combustion, and industrial activities [47,68], as well as atmospheric inputs of crustal origin [48].

The positive loadings of PC2 to bicarbonate and fluorine ions, coupled with a negative one to PM_2.5 concentrations and lack of correlation with PM_2.5 (Table 1 and Table 2), suggested that this principal component represented the dissolution of atmospheric CO₂ and HF. The presence of fluorine ions in precipitation is often positively correlated with increased HF concentrations in the atmosphere impacted by industrial activities [41]. PC2 corresponding factor scores were significant (i.e., >1) for both precipitation episodes (Figure 6), both at the beginning of the precipitation and during it. However, as local atmospheric CO₂ and HF concentrations are not monitored in Wrocław, this hypothesis needs to be verified by further research that should include these parameters.

The positive loadings of nitrate, ammonium, sodium, and potassium in the PC3 principal component were likely related to agriculture practices (e.g., volatilization of animal manure, sewage, natural loss by plants, fertilizers, and agricultural activities) or biomass burning [47,49]. The results showed no correlations with the local aerosol concentrations (scores of 0.17 for PM₁₀ and of 0.33 for PM_2.5). Still, the scavenging and dissolution of remote fertilizer-bearing particles during the path of precipitation could not be ruled out [2]. The highest score of 0.91 for K⁺ may have been linked to inputs from biomass burning, but this would need to be further tested by the additional monitoring of levoglucosane, another marker of this type of emission, e.g., [69,70]. Still, this conclusion was supported by the significant correlation we observed between K⁺ and PM_2.5 during winter (Table 2). The time variations in PC3 showed scores > 1 at the start of the winter rain episode (Figure 6B) and a smaller but noticeable peak in the summer episode (Figure 6A). Fertilizers are generally spread in early spring (end of February).

While the study here only represented two distinct snapshots of rain events, our results can still be compared to the one-year study we previously carried out in 2010 in Wrocław [2]. The PCA analysis then revealed two main principal components that were first related to fossil fuel combustion and agriculture, and a second one corresponding to sea spray aerosols [2]. Here, while the PCA analysis also identified fossil fuel combustion and agriculture as contributing factors to the chemistry of rainwater, it did not identify sea spray aerosols as a significant one. That last conclusion was confirmed by the relatively small contributions of ss SO₄²⁻ and Ca⁺ we determined for the two rain episodes (Figure 5). It suggested that sea spray aerosols mostly penetrate the atmosphere of Wrocław at other periods of the season.

5. Conclusions

The analysis of the chemical composition of single-event precipitation in Wrocław (Poland) revealed clear variability both during each of the two episodes and also between them. The ion concentrations generally were higher at the start of the rain episode both in summer and winter, probably reflecting the capture and subsequent removal of air contaminants by fallout and scavenging processes. The results of the combined chemical and statistical analysis showed that interactions between precipitation and air contaminants can be discriminated into three independent groups. (1) The long-range transport and/or anthropogenic influences of emission sources that included biomass burning, fossil fuel combustion, and industrial processes coupled to a lesser extent with material of crustal origin. (2) The dissolution of atmospheric CO₂ and HF into their ionic forms. (3) Finally, ions originating from agriculture activities and/or biomass burning.

In winter, point heating sources were the main source of PM₁₀ from which higher concentrations for Cl⁻, NO₃⁻, SO₄²⁻, Na⁺, NH₄⁺, K⁺, and Ca²⁺ ions were leached. Our results also revealed seasonality in the rainwater chemistry. In summer, the main sources of NO₃⁻ and NH₄⁺ were most likely agricultural practices (i.e., fertilizers), whereas in winter, fossil fuel combustion dominated. The presence of SO₄²⁻ in precipitation was attributed to primary SO₄²⁻ emissions from industrial and energy activities as well as from the household sector. Contrary to our previous year-long monitoring [2], on a shorter time period, sea salt aerosols only slightly contributed to the chemical composition of precipitation. Finally, K⁺, Ca²⁺, and Mg²⁺ ions, both in winter and summer, originated from agricultural practices (i.e., spread of potassium fertilizers such as KCl or K₂SO₄) or natural/anthropogenic dust.

While the study focused on the analysis of two single rain episodes collected at distinct seasons, our results proved that chemical variability exists at such a short scale. They highlighted different emission sources for some of the rainwater ions when compared to similar studies that covered larger time periods, proving that these sources and their respective contributions vary with time. It underlined the need for further studying the chemistry of atmospheric precipitation at such distinct periodic intervals, in order to more accurately identify sources of air contaminants, as well as to characterize the interaction dynamics existing between atmospheric compounds. This should ultimately help design better remediation strategies to preserve rainwater quality.