Spatial Distributions of Atmospheric Ammonia in a Rural Area in South Korea and the Associated Impact on a Nearby Urban Area

Ammonia (NH3) plays an important role in air quality and atmospheric chemistry, yet studies on the characteristics and impacts of NH3 are limited. Herein, we revealed the spatial distribution of atmospheric NH3, as measured by passive samplers, at three different sites (R1, R2, and R3) in the rural area (livestock environment) of Jeongeup, South Korea, from September 2019 to August 2020. At site R1, the boundary of a large-scale pig farm, dramatically high daily mean concentrations of NH3 were observed (118.7 ppb), whereas sites R2 and R3, located ~1 km from R1, exhibited lower concentrations of 18.2 and 30.4 ppb, respectively. In the rural environment, the monthly NH3 variations showed a peak in June (34.2 ppb), which was significantly higher than in the urban and remote areas. To examine the impact of NH3 from the rural area on a nearby urban area in June 2020, simultaneous measurements were performed using a real-time instrument in Jeonju. When high NH3 events occurred in the urban area in June, the results for the NH3 concentrations and observed meteorological conditions in the rural and urban areas showed that the rural area influenced the NH3 levels in the adjacent urban area.


Introduction
Ammonia (NH 3 ) is a base gas that reacts with acidic species in the atmosphere, such as sulfuric acid (H 2 SO 4 ) and nitric acid (HNO 3 ), to produce secondary inorganic aerosols (SIAs). These SIAs contribute to the degradation of visibility and air quality, and have adverse effects on human health [1][2][3].
Global NH 3 emissions have increased from 1.9 to 16.7 Tg between the 1960s and 2010s [4]; southern Asia, including China and India, accounts for more than 50% of the total global NH 3 emissions since the 1980s. More than 90% of the total NH 3 emissions derive from the livestock industry, agricultural activities, and domestic animal fertilizer use [5,6]. Additionally, some studies have shown that industrial and traffic emissions may be a source of NH 3 in urban environments [7,8]. Recently, effective controls have significantly reduced the emissions of some gaseous pollutants, including NO x and SO 2 [9,10], but NH 3 emissions have continued to increase [11,12].
Field measurements have shown that variations in NH 3 concentrations depend on the location and season. Due to elevated emissions from livestock husbandry and agricultural actives, higher concentrations of ambient NH 3 are typically recorded in rural areas [13,14]. For example, the atmospheric NH 3 concentrations measured from April 2009 to August 2011 near the boundary of a pig farm in China ranged between 46.6 and 674.7 ppb [13]. Additionally, Kubota (2020) reported that NH 3 concentrations in high emissions areas (i.e., adjacent to livestock sources) were greater in winter (~71 ppb) than those in summer (~56 ppb), which is not a typical seasonal variation pattern [14]. In contrast, lower concentrations of ambient NH 3 have been reported in urban areas [15,16]. For instance, the NH 3 concentration measured in Seoul, South Korea, was~11.6 ppb from 2010 to 2011; the mean seasonal value of NH 3 was~14.0 and~9.0 ppb in the summer and winter, respectively [15]. Meng (2011) and Zhou (2019) suggested that NH 3 emitted from livestock activities could influence the ambient NH 3 concentrations in nearby urban areas [17,18]. Although atmospheric NH 3 has a significant impact on the air quality in rural areas and adjacent urban areas, data on spatial distributions and characteristics of atmospheric NH 3 are, at present, limited (there are especially limited data available for atmospheric NH 3

measurements in South Korea).
In this study, to explore spatial distributions of atmospheric NH 3 in a rural area, we conducted a one-year measurement (from September 2019 to August 2020) at three different monitoring sites in Jeongeup, a rural site in South Korea. Jeongeup is a typical agricultural area characterized by large-and small-scale livestock farms and facilities. Moreover, to investigate whether the NH 3 emitted from the agricultural source influenced the atmospheric NH 3 in the urban area, simultaneous measurements for atmospheric NH 3 were carried out in both the rural area (Jeongeup) and a nearby urban area (Jeonju) during summer in June 2020.

Monitoring Sites
Atmospheric NH 3 concentrations were measured in the rural area of Ongdong-myeon, Jeongeup (35.655 • N, 126.987 • E) from September 2019 to August 2020. Jeongeup is an agricultural area of approximately 692.7 km 2 , characterized by the largest NH 3 emissions in Jeollabuk-do [19] (Figure 1). Significant livestock populations (such as pigs, cows, and chickens, etc.) are located within the region of Jeongeup (i.e.,~393,000 pigs,~89,000 cows, and~823,000 chickens in 2021 [20]). Based on the Clean Air Policy Support System (CAPSS), the main NH 3 source among the livestock in the rural area is pigs [19]. Three different sites in Ongdong-myeon, Jeongeup, were selected: one site at the boundary (R1) of a large-scale mechanically ventilated pig breeding farm (~12,000 pigs in 2020) and two sites located 1 km from the farm (R2 and R3). Sites R2 and R3 are located to the north and south of the pig farm, respectively ( Figure 1). Additionally, NH 3 concentrations were measured on the second floor of the Natural Science Building at Jeonbuk National University, Jeonju, which is the capital of Jeollabukdo, South Korea (35.847 • N, 127.129 • E). Jeonju is an urban area located approximately 40 km from the monitoring sites in the rural area of Jeongeup. This site is surrounded by business offices, residential buildings, and roads. Major livestock areas emit NH 3 in the western vicinity of this region (Jeongeup, Iksan, and Gimje). Figure 1 shows the locations of the monitoring sites in both the rural and urban areas.

Atmospheric NH 3 Measurements
Passive samplers (RAD 168, Radiello, Italy) were employed to obtain the NH 3 concentrations over the rural sites of R1-R3 at a height of~3 m from the ground from September 2019 to August 2020, with the aim of analyzing their spatial distribution. A passive sampler is consisted of an outer porous cylindrical diffusive body, which controlled the diffusion rate, and a cylindrical inner polyethylene tube coated by phosphorous acid, leading to NH 3 adsorption. This is a widely used instrument to collect atmospheric NH 3 [14,18,[21][22][23]. In this study, atmospheric NH 3 was collected over a one-day period from 09:00 a.m. to 08:00 a.m. local time the following day during the entire sampling period; however, NH 3 was collected over a two-day period in June (because of the rainy season). Sampling could not be conducted in October and November 2019 owing to limited access in the area owing to African swine fever virus (ASF) and in July 2020 during the monsoon season. After NH 3 collection, samples were stored at −18 • C before extraction. The samples were extracted in 6 mL of deionized water (18.2 MΩ·cm, Merck Milli-Q ® , Millipore, Burlington, MA, USA) and sonicated for 45 min. Subsequently, the extracts were analyzed as an NH 4 + concentration using ion chromatography (Aquion, Thermo Scientific, USA). The NH 4 + concentrations obtained were then converted to NH 3 using a previously reported equation [24,25]. Based on the field blanks, the detection limit of NH 3 was calculated to bẽ 0.85 µg/m 3 (~1.2 ppb). A total of 95 samples were analyzed from R1 (32 samples), R2 (32 samples) and R3 (31 samples), as listed in Table 1. After collection, all samples were analyzed within two weeks. To obtain the temporal variations in the atmospheric NH 3 concentrations in Jeonju (the urban area), NH 3 was measured on the second floor of Jeonbuk National University using a cavity ring-down spectroscopy (CRDS) analyzer (Picarro Inc., model G2103, Santa Clara, CA, USA) at 1 s intervals from 1-30 June 2020. The detailed methods and calibration of the instrument describes in Park (2020) [26]. Briefly, the detection limit of the NH 3 analyzer was less than 0.09 ppb and the average precision was 0.3 ppb for 300 s with a response time of less than 1 s [27]. In principle, the NH 3 analyzer does not require additional external calibration; however, in this study, mixtures of a standard NH 3 gas (9.2 ppm, with an accuracy of ±2%; Airkorea, Korea) and N 2 (Airkorea, Korea, 99.999%) were used to confirm the calibration performance of the analyzer. Calibration was conducted using five different NH 3 concentrations (150, 100, 50, 30, and 0 ppb); the resulting R 2 was 0.9997. Hourly averaged data were used for data analysis. Data that exceeded the hourly amount of precipitation of 5 mm were excluded from data analysis to reduce the effect of precipitation.
Hourly averaged meteorological parameters, including air temperature, relative humidity, wind speed, wind direction, and precipitation, were collected at Jeongeup (rural; station id: 47245,~16.5 km from R1) and Jeonju (urban; station id: 47146,~1.5 km from U1) using the automated synoptic observing system (ASOS) from the Weather Data Service of the Korea Meteorological Administration (Available online: https://data.kma.go.kr) (accessed on 13 May 2021).

Modeling of NH 3 Origin
To identify the relative concentrations of the pollutants contributing to the potential source regions at the receptor site, we performed a concentration weighted trajectory (CWT) analysis [28,29]. The study field covering a geographical area from 90 • E to 150 • E and from 20 • N to 60 • N includes 2400 grid cells with a spatial resolution of 1 • × 1 • . The CWT analysis was combined with a 72-h air mass backward trajectory using the National Oceanic and Atmospheric Administration Hybrid Single-Particle Lagrangian Trajectory (HYSPLIT4) model at four times; 00:00, 06:00, 12:00, and 24:00 UTC at 100 m above ground level (AGL). The meteorological data used for backward trajectory calculating were the GDAS (Global Data Assimilation System) with a resolution of 1 • × 1 • data and were downloaded from the web server of NOAA Air Resources Laboratory. Additionally, to access the regional scale transport and local pollution source emissions, a conditional probability function (CPF) analysis was performed using NH 3 concentrations, wind direction, and wind speed data obtained from ASOS. The threshold criterion of the 90th percentile was selected to indicate the directionality of the sources.

Spatial Distributions of Atmospheric NH 3 in the Rural Area
During the observation period from September 2019 to August 2020, the monthly averaged temperature, relative humidity, and wind speed were 13.1 ± 9.8 • C and 71.4 ± 17.8 %, and 1.5 ± 1.8 m/s, respectively; the prevailing wind direction was north in the rural area ( Figure S1). A total of 95 passive samplers were used at the three different sites within the rural area during the entire study period. Table 1 presents a description of each monitoring site and the average seasonal NH 3 concentrations. Figure 2 shows the variation in the daily average NH 3 concentration at R1, R2, and R3 in the rural area. At site R1 (boundary of a large-scale pig farm), dramatically high NH 3 concentrations were recorded, with a daily average of 118.7 ± 51.7 ppb, which varied from a minimum of 40.3 ppb to a maximum of 272.2 ppb (Figure 2a). For the seasonal variation, the average NH 3 concentrations at R1 were 100.5, 128.6, 60.3, and 141.6 ppb in spring, summer, autumn, and winter, respectively (Table 1). Average NH 3 concentrations over 100 ppb were recorded during all seasons, except in autumn owing to frequent heavy rain and typhoons (Figure S1b) in September 2019 ( Figure S2a). As listed in Table 1, in winter, the average seasonal NH 3 concentration at site R1 was approximately two-fold higher than that in autumn. Ammonia is temperature-dependent [30]; high concentrations have been generally reported during summer in various environments [15,21,[30][31][32][33]. However, at the NH 3 point source observed at the boundary of the large-scale pig farm (R1), the NH 3 concentrations were insensitive to the ambient temperature, instead recording a high NH 3 level in winter ( Figure 2 and Table 1). This is possibly due to ventilation differences at a mechanically ventilated pig farm depending on the season. The ventilation system usually operates at a significantly higher frequency in summer and a lower frequency in winter to maintain the internal temperature [34][35][36][37][38][39]. Increased ventilation rates can easily diffuse NH 3 into the atmosphere under high ambient temperatures, producing relatively lower concentrations of atmospheric NH 3 in summer. In contrast, reduced ventilation rates can diffuse concentrated NH 3 under low ambient temperatures, resulting in relatively high concentrations of atmospheric NH 3 in winter [37][38][39]. Additionally, high concentrations of ambient NH 3 at R1 may have also been driven by the livestock industry environments and activities of such a large-scale pig farm. In Asia, large-scale farms are usually equipped with open manure storage facilities; in these facilities, farmers actively store manure and produce fertilizer during winter for use on farmland in spring with the start of agricultural activity [40]. The NH 3 emitted from intensive manure production in open storage facilities in winter increases the atmospheric NH 3 concentration. Previous studies have reported similar results: ambient NH 3 concentrations were higher in winter months than in other months with active fertilization in agricultural areas [14,[41][42][43]. Kubota (2020) conducted atmospheric NH 3 measurements with a passive sampler at the near livestock sources from October 2018 to January 2020, finding that the average NH 3 concentration in winter was higher than that in summer, which is consistent with our results [14]. Additionally, García-Gómez (2016) and Loftus (2016) showed that the highest seasonal NH 3 concentrations occurred in winter, which is related to the presence of livestock in the vicinity [41,42].
At R2 (~1 km north of R1), the atmospheric NH 3 concentration reached 56.1 ppb, with a daily average of 18.2 ± 11.5 ppb during the observation period ( Figure 2b and Table 1). Compared with the NH 3 level observed at R1 and R3, significantly lower atmospheric NH 3 concentrations were recorded at R2 (Figure 2). This is because the monitoring site is windward of the pig farm at R1, with no surrounding farms. The atmospheric seasonal NH 3 concentration at R2 peaked at 28.5 ± 13.6 ppb in summer, and in June at 30.1 ± 15.0 ppb ( Figure S2b), which was approximately two-fold higher than that in the other seasons (15.8 ppb for spring, 11.2 ppb for autumn, and 13.6 ppb for winter), as listed in Table 1. The seasonal variation at R2 was comparable to the variation observed for the peak concentration in winter at the point source of R1, as described above.
At R3 (~1 km south of R1), the daily average NH 3 concentration was 30.4 ± 12.1 ppb ranging from 15.6 to 56.3 ppb (Figure 2c). Seasonally, the average NH 3 concentrations were 27.5, 38.2, 38.7, and 25.0 ppb in spring, summer, autumn, and winter, respectively, yielding negligible seasonal variations compared with R2 (Table 1). There are several small-scale mechanically ventilated pig farms near small households at R3. Continuous sources from mechanically ventilated small farms likely caused such stable variation. In addition, a higher average daily NH 3 concentration was observed at R3 (30.4 ppb) than at R2 (18.2 ppb) during the entire period, as listed in Table 1.
Sites R2 and R3 are located only~1 km from the NH 3 point source at R1, but significantly lower NH 3 levels were observed (Figure 2). Previous studies have also observed this pattern of decreasing NH 3 concentrations with increasing distance from NH 3 emission sources [44][45][46]. For example, López-Aizpún (2018) reported that NH 3 concentrations decreased from 74.7 ppb at 30 m to 2.1 ppb at 1000 m distance from livestock [46].

Comparisons of Atmospheric NH 3 Concentrations in Different Environments
In this study, we defined the representative atmospheric NH 3 concentration for the rural environment as the average NH 3 concentration obtained from sites R2 and R3 in the rural area of Jeongeup. Site R1 was excluded because of its proximity to the NH 3 emission source [47][48][49]. Figure 3 shows the seasonal variations in the ambient NH 3 values obtained from sites R2 and R3 from September 2019 to August 2020. In the rural environment, the daily mean atmospheric NH 3 concentration was 24.2 ± 13.3 ppb, with a seasonal variation of 33.3 ± 14.6 ppb in summer, 22.2 ± 13.6 ppb in autumn, 21.7 ± 11.8 ppb in spring, 19.3 ± 8.8 ppb in winter. Significantly higher NH 3 concentrations were observed in summer compared with winter in the rural area. The highest monthly concentration was recorded in June 2020 at 34.2 ± 15.7 ppb ( Figure S3).
Many studies have reported a strong positive correlation between the ambient NH 3 and temperature; the atmospheric NH 3 concentration increased with an increase in the ambient temperature [21,26,30,50]. In summer, high temperatures favor the volatilization of NH 3 emitted from various sources, such as agricultural activities, leading to a thermodynamically stable phase state as gaseous NH 3 , rather than particulate NH 4 + in the atmosphere [51][52][53]. Our results for the high NH 3 levels recorded in summer is consistent with previous results (Figure 3).  Table 2 summarizes the atmospheric NH 3 concentrations observed in various environments in rural (livestock villages), urban, and remote areas. Although the measurement period was different, significantly greater NH 3 concentrations were observed in rural areas compared with other environments. For example, in Beijing, China, the average recorded NH 3 concentration was 37.0 ppb near the pig facilities [13]. Additionally, in Navarre, Spain, where two high-intensity point-sources of NH 3 are located (pig and cattle farms), the average NH 3 concentration was 33.8 ppb, with a maximum value of 74.7 ppb [46]. In Colorado, USA, the NH 3 had an average concentration of 61.9 ppb, as influenced by emissions from adjacent large concentrated animal feeding operations [54].
The range of NH 3 concentrations reported in urban areas is relatively variable, as listed in Table 2. In China, mean annual NH 3 concentrations are 7.8 ppb in Shanghai [55] and 15.2 ppb in Nanjing [56]. In Korea, the mean NH 3 concentrations in Seoul [15], Mokpo [57], and Jeonju [26] are 11.6, 8.6, and 10.5 ppb, respectively. Compared with other urban areas in Asia, the ambient NH 3 concentrations are relatively lower in New York, USA [18], and Douai, France [58], with values of 3.2 and 4.2 ppb, respectively. In some urban areas, diurnal variations in the NH 3 concentration are dependent on the traffic emissions, which may be an important NH 3 source in urban areas [59][60][61]. Compared with rural and urban areas, significantly lower NH 3 levels, i.e., <~5 ppb, have been observed in remote areas, including coastal areas, mountains, and forests (Table 2) [43,47,56,62,63].

Impact on NH 3 Levels in Nearby Urban Area
As discussed in Sections 3.1 and 3.2, average seasonal atmospheric NH 3 concentrations in the rural area peaked in summer (Figure 3), particularly in June ( Figure S3). To investigate whether the high NH 3 concentrations observed in the rural area influenced the NH 3 levels in nearby urban areas, temporal variations in the atmospheric NH 3 were measured using CRDS in a nearby urban area at site U1, Jeonju. Location of the urban monitoring site (U1), which is~40 km from the rural area, is shown in Figure 1. During 1-30 June 2020, the hourly mean ambient NH 3 concentration at site U1 was 18.6 ± 7.8 ppb, with a minimum at approximately the detection limit of~1 ppb and a maximum of 59.3 ppb ( Figure S4). Figure 4 shows the diurnal variation in the ambient NH 3 measured at U1 in June 2020. A high NH 3 level was maintained in the afternoon with a mean hourly concentration of >20 ppb from 13:00 to 21:00, with a peak value of 23.1 ppb at 18:00. The NH 3 concentration then remained low from night to sunrise. Park (2020) also observed a similar diurnal pattern with a single NH 3 peak in the late afternoon in June 2019 at an adjacent site, i.e., Samcheon-dong in Jeonju [26]. This ambient NH 3 peak was only measured in an urban area characterized by NH 3 transported from an adjacent rural area. Previous studies have hypothesized that a single NH 3 peak, appearing in the late afternoon, was caused by NH 3 emitted from agricultural activities and the evolution of the planetary boundary layer [66][67][68]. In the morning, farmers in rural areas begin to fertilize their land. As the temperature rises in the afternoon, NH 3 volatilizes into the atmosphere, with the occurrence of vertical exchange via extension of the mixing layer. These processes can lead to elevated levels of atmospheric NH 3 in rural areas; the increasing NH 3 concentrations can then be transported to nearby urban areas with increases in the mixing height and wind direction [26,[66][67][68][69]. In this study, based on simultaneous measurements at the rural and urban sites in June, a high NH 3 level was recorded at both the rural (daily average of 49.6 ± 5.3 ppb obtained at sites R2 and R3) and U1 (hourly average of 23.2 ± 7.6 ppb) sites from 2 to 8 June ( Figure 5). On 2 June, a high daily NH 3 concentration of 52.3 ppb in the rural area was recorded with high wind speeds of ≤~5.2 m/s in the afternoon, as shown in Figure 5a. The prevailing wind direction in the rural area then changed from a northeasterly to a southwesterly direction toward the urban site from 2 to 3 June; this was then maintained until 7 June. Throughout the same period, westerly winds were also dominant in the urban area, with low wind speed conditions (average of 1.3 ± 0.9 m/s), and high temperature (average of 23.0 ± 3.5 • C) at U1 (Figure 5b). These stable meteorological conditions in the urban area favored pollutant accumulation, including NH 3 . This resulted in increasing NH 3 concentrations at U1 until 7 June, especially in the late afternoon (Figure 5b), as hypothesized in previous studies [26,[66][67][68]. Therefore, based on our simultaneous measurements, the elevated NH 3 level in the late afternoon in the urban area was most likely transported from the adjacent rural area. Moreover, the CPF analysis revealed that there is a high probability of NH 3 concentrations > 29 ppb (90th percentile) in June ( Figure 6). Additionally, the CWT results showed that the high concentration of atmospheric NH 3 originated domestically, rather than via long-range transport ( Figure S5). These results indicate that the adjacent rural area influenced the high NH 3 concentrations observed in the urban area in June.

Conclusions
To investigate the spatial distributions of atmospheric NH 3 in a rural area, atmospheric NH 3 concentrations were analyzed from 95 samples collected using passive samplers at three different sites (R1, R2, and R3) in Jeongeup, South Korea, from September 2019 to August 2020. During the entire period, the average daily NH 3 concentrations were 118.7 ± 51.7 ppb at site R1 (boundary of a large-scale pig farm), 18.2 ± 11.5 ppb at site R2 (~1 km north of R1), and 30.4 ± 12.1 ppb at site R3 (~1 km south of R1). Significantly high levels of atmospheric NH 3 were recorded at the NH 3 emission source of R1 during winter (average of~141.6 ppb) due to the low ventilation rates and active production of livestock manure. In contrast, there were significant decreases in the atmospheric NH 3 concentrations at R2 and R3, even at a distance of only~1 km from the NH 3 emissions source (R1). In this study, we used the average NH 3 concentration at sites R2 and R3 to determine the representative atmospheric NH 3 concentration in the rural area as R1 is in close proximity to the NH 3 emission source. The average atmospheric NH 3 concentration of the rural areas (average of sites R2 and R3) was 24.2 ± 13.3 ppb, with a seasonal variation of 33.3 ± 14.6 ppb in summer, 22.2 ± 13.6 ppb in autumn, 21.7 ± 11.8 ppb in spring, and 19.3 ± 8.8 ppb in winter. Particularly, the NH 3 concentrations were highest in the summer of June 2020.
To explore the impact of the high NH 3 concentrations monitored in the rural area during June on the atmospheric NH 3 level in a nearby urban area, atmospheric NH 3 concentrations were simultaneously measured in an urban area of Jeonju using a CRDS in June 2020. The hourly mean NH 3 concentration in June was 18.6 ± 7.8 ppb in the urban area, where a high level was maintained in the late afternoon. When high NH 3 episodes in June occurred at the urban site, elevated NH 3 concentrations were also observed in the adjacent rural area. During these episodes from 2 to 8 June, westerly winds were dominant in the urban area with low wind speed conditions and high temperatures, thus leading to stable meteorological conditions. The CPF analysis also showed that there was a high probability of NH 3 concentrations in June. Conclusively, the increasing ambient NH 3 concentrations observed in the urban area in June were influenced by high NH 3 concentrations from the rural area located to the west. These results can provide a more comprehensive understanding of the spatial and temporal distribution of atmospheric NH 3 and its impact, as well as a scientific basis to develop effective control strategies for atmospheric NH 3 levels.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/atmos12111411/s1, Figure S1: Monthly meteorological conditions of (a) wind direction (WD) and wind speed (WS), and (b) cumulative precipitation (Pre.), temperature (T), and relative humidity (RH) at Jeongeup in the rural area from September 2019 to August 2020. Figure S2: Monthly variations in ambient NH 3 concentrations with standard deviations at the monitoring sites of (a) R1, (b) R2, and (c) R3 in a rural area during September 2019-August 2020. Figure S3: Monthly variations in ambient NH 3 concentrations averaged at the rural sites R2 and R3 during September 2019-August 2020. Figure S4: Time series of hourly mean NH 3 concentrations measured at the urban (Jeonju) during 1-30 June 2020. Figure   Data Availability Statement: Publicly available meteorological archived datasets analyzed in this study can be found at https://data.kma.go.kr (accessed on 13 May 2021). The publicly available Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model can be found at https: //www.ready.noaa.gov/HYSPLIT.php (accessed on 13 May 2021) and run either online or offline. The data can be found from the link: ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1/ (accessed on 13 May 2021).