2.1 Measurement site

The measurements were performed from May to September 2013 at Gucheng (39^∘08^′ N, 115^∘40^′ E, 15.2 m a.s.l.), a rural site in the NCP, which is an Integrated Ecological Meteorological Observation and Experiment Station of the Chinese Academy of Meteorological Sciences. In Fig. 1, the location of the site is shown on the NCP map with the NH₃ emission distribution for the year 2012 from the multi-resolution emission inventory of China (http://meicmodel.org/index.html). The measurement site chosen is situated for monitoring regional background concentrations of air pollutants in the North China Plain and has good regional representativeness (Lin et al., 2009). The site is approximately 110 km southwest of Beijing, 130 km west of Tianjin, and 160 km northeast of Shijiazhuang city in Hebei Province. The site is surrounded by farms, dense villages/towns, and the transportation network in the NCP. The main crops in the area surrounding the site are wheat (winter and spring) and corn (summer and fall). The site is influenced by high NH₃ emissions from fertiliser use and animal husbandry in the surrounding area. Being in the warm temperate zone, the site has a typical temperate continental monsoon climate. Precipitation occurs mainly between May and August.

2.2 Sampling and analysis

Ambient NH₃ was measured using an ammonia analyser (DLT-100, Los Gatos Research, USA), which utilises a unique laser absorption technology called off-axis integrated cavity output spectroscopy. The analyser has a precision of 0.2 ppb at a 100 s average and a maximum drift of 0.2 ppb over 24 h. The response time of the analyser is less than 2 s (with an optional external N920 vacuum pump). During the campaign, NH₃ data were recorded at a 100 s average. In principle, the NH₃ analyser does not need external calibration because the measured fractional absorption of light at an ammonia resonant wavelength is an absolute measurement of the ammonia density in the cell (manual of economical ammonia analyser – benchtop model 908-0016, Los Gatos Research). However, we confirmed the good performance of the NH₃ analyser using a reference gas mixture NH₃ ∕ N₂ (Air Liquide, USA) traceable to the US National Institute for Standards and Technology (NIST). The reference gas of NH₃ (25.92 ppm with an accuracy of ±2 %) was diluted to different concentrations using zero air and supplied to the analyser, and a sequence with five points of different NH₃ concentrations (including zero and the concentrations ranging from 45 to 180 ppb) were repeated for several times to check the performance of the analyser.

As shown in Fig. S1 in the Supplement, the analyser rapidly followed changes of the NH₃ concentration, produced stable response under stabilised NH₃ concentrations, and accurately repeated (within the uncertainty) the supplied NH₃ concentrations. During the calibration, it took about 20 min for the instrument to show 90 % of the changes in the NH₃ concentrations supplied through an aerosol filter and a PTFE tubing (4.8 mm ID, about 3 m). However, these time delays also contained the balance time needed for the calibration system. The lag caused by the tubing and analyser should be much smaller. The NH₃ analyser contains an internal inlet aerosol filter, which was cleaned before our campaign. Nevertheless, some very fine particles can deposit on the mirrors of the integrated cavity output spectrometer (ICOS) cell, leading to a gradual decline in reflectivity. However, slight mirror contamination does not cause errors in NH₃ measurements because the mirror reflectivity is continually monitored and the measurement is compensated for using the mirror ring-down time. Interferences to NH₃ measurements can be from the sample inlets, for example, due to water condensation or adsorption–desorption effects (e.g. Schwab, 2008; Norman et al., 2009). Such interferences were not quantified but reduced as much as possible. PTFE tubing (4.8 mm ID), which is one of the well-suited materials for NH₃ measurement (Norman et al., 2009), was used to induce ambient air. The length of the tubing was kept as short as possible (about 5 m) to limit the residence time to less than 3 s. The aerosol filter at the inlet was changed every 2 weeks. Water condensation was avoided. Nevertheless, we cannot exclude the influence from adsorption and desorption, which can also occur on dry surfaces. However, this influence should be small at our site, where the NH₃ concentration is very high, and cause mainly a lag in the recorded NH₃ concentration. As hourly averages are used in the work, this lag may not exert a significant influence on our results.

A set of commercial instruments from Thermo Environmental Instruments Inc. were used to measure O₃ (TE 49C), NO/NO₂/NO_x (TE 42CTL), CO (TE 48C), and SO₂ (TE 43CTL). All instruments were housed in an air-conditioned room in the observation building at the site. Two parallel inlet tubes (Teflon, 4.8 mm ID × 8 m length) were shared by the analysers. The height of the inlets was 1.8 m above the roof of the building and about 8 m above the ground. The inlet residence time was estimated to be less than 5 s (Lin et al., 2009). Zero and span checks were performed weekly on the analysers of these trace gases to identify possible analyser malfunctions and zero drifts. Multipoint calibrations of SO₂, NO_x, CO, and O₃ analysers were performed on the instruments at approximately 1 month intervals. Measurement records were saved as 1 min averages. After the correction of data on the basis of the multipoint calibrations, hourly average data were calculated and used for the analysis.

An ambient ion monitor (AIM; URG 9000D series, USA) was deployed at the site to measure hourly concentrations of water-soluble inorganic components in PM_2.5 during 15 June–11 August 2013. A detailed description of the performance evaluation of the AIM-IC system is reported by Han et al. (2016). Briefly, ambient air was introduced to the AIM with a 2 m Teflon-coated aluminum pipe and particles larger than 2.5 µm were removed by a cyclone at a flow rate of 3 L min⁻¹. A liquid diffusion denuder was used to remove the interfering acidic and basic gases, in combination with a steam-jet aerosol collector followed by an aerosol sample collector, until the particles could be injected into the ion chromatograph (Hu et al., 2014). The detection limits of NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and Cl⁻ were 0.05, 0.04, 0.05, and 0.01 µg m⁻³, respectively. For the AIM, multipoint calibrations were performed weekly using calibration standard solutions. Acceptable linearity of ions was obtained with an R² of ≥ 0.999. The flow rate of the AIM was checked weekly at the sample inlet with a certified flow metre. The flow rate of the AIM was kept at 3 L min⁻¹ with a standard derivation of < 1 %. Hourly data were obtained for the concentrations of water-soluble inorganic ions in summer 2013.

Meteorological parameters were measured at the site. Air temperature and relative humidity (RH) were monitored using a humidity and temperature probe (HMP155, Vaisala, Finland); wind speed and direction were measured using an anemometer (ZQZ-TFD12, Jiangsu Radio Scientific Institute Co. Ltd., China); rainfall was measured using a tilting rain gauge (SL2-1, Tianjin Meteorological Instrument Factory, China). Global radiation observations were made at the site but showed a drift by the end of July 2013. Instead we use the photolysis rate jNO₂ observed using a 2-pi-actinic-flux spectrograph (CCD type, Meteorologie Consult GmbH, Germany) to indicate the radiation condition for photochemistry. Hourly meteorological data were calculated from the in situ measurements and used in this paper. Planetary boundary layer height (PBLH) values at 14:00 BST (all times in Beijing standard time throughout the paper) were derived from the ERA-Interim data using the bulk Richardson number method (Guo et al., 2016; Miao et al., 2017).

2.3 Data analysis

2.3.3 Trajectory calculation

The 72 h backward trajectories were calculated using the HYSPLIT 4.9 model (http://www.arl.noaa.gov/ready/hysplit4.html). The trajectories terminated at the height of 100 m above the ground. The trajectory calculations were performed at four times (00:00, 06:00, 12:00, and 18:00 UTC) per day in summer 2013. Individual back trajectories were grouped into five clusters. The number of clusters is identified according to the changes in total spatial variance (TSV). Five is chosen as the final number of clusters considering optimum separation of trajectories (larger number of clusters) and simplicity of display (lower number of cluster). The corresponding concentrations of trace gases and water-soluble ions were averaged over the period of 3 h ahead and after the arrival time for each backward trajectory for further analysis.

Table 1The comparison of the concentrations of trace gases (ppb) and water-soluble ions in PM_2.5 (µg m⁻³) at Gucheng with other research.

Download Print Version | Download XLSX

3.1 Overview of concentration levels of measured species

During 15 May–25 September 2013, the average concentrations (ranges) of NH₃, SO₂, and NO_x were 36.2 (0.1–862.9), 5.0 (0–86.8), and 15.4 (2.7–67.7) ppb, respectively. As listed in Table 1, the concentration of NH₃ at the NCP rural site was lower than that reported at Asian and African urban sites such as Lahore (Pakistan; Biswas et al., 2008), Colonelganj (India; Behera and Sharma, 2010), and Cairo (Egypt; Hassan et al., 2013) but higher than those from other areas in China, Europe, and North America (Plessow et al., 2005; Yao et al., 2006; Lin et al., 2006; Walker et al., 2006; Hu et al., 2008; Meng et al., 2011, 2014; Shen et al., 2011; Schaap et al., 2011; Makkonen et al., 2012; Behera et al., 2013; Gong et al., 2013; Li et al., 2014). For example, the NH₃ at the NCP rural site was higher than that found at the Shangdianzi regional background station in the NCP (Meng et al., 2011), Lin'an regional background station in the Yangtze River Delta (YRD) in eastern China (Meng et al., 2014), and the rural site in Beijing (Shen et al., 2011). The relatively high concentrations of NH₃ observed in this study were attributed to agricultural activities involving fertiliser use, vegetation, livestock, and human excrement and waste disposal in the surrounding region.

According to an inventory study (Zhang et al., 2010), the total agricultural NH₃–N emission in 2004 in the NCP was 3071 kt yr⁻¹, accounting for 27 % of the total emissions in China with 1620 kt yr⁻¹ of NH₃–N emissions caused by fertiliser applications, which is the largest emission source in China, accounting for more than half of the total agricultural emissions. In recent years, there were a few publications about China's national and regional emission inventories of NH₃ (e.g. Zhou et al., 2015; Xu et al., 2015, 2016; Kang et al., 2016). However, these inventories are based on bottom-up studies, subject to substantial uncertainties in spatial and temporal variations in NH₃ emissions. Ground-based observations of NH₃ have been sparse. Our measurements, together with others, can be used for validating and constraining models that use bottom-up inventories, and hence help to reveal potential bias in NH₃ emission inventories.

The observed concentration of SO₂ at the NCP rural site was markedly lower than that reported for the same period in 2006–2007 (Lin et al., 2009). Because of a series of emission reduction measures implemented in recent years, SO₂ levels have decreased markedly in the NCP (Lin et al., 2011). The average concentration of NO_x was higher than that at the Shangdianzi (Meng et al., 2011) and Lin'an (Meng et al., 2014) regional background stations in the NCP and YRD region of China, which might be due to emissions from agricultural activities and motor vehicle sources (Lei and Wuebbles, 2013; Liu et al., 2013) in the NCP, but it was lower than those at urban sites in India (Behera and Sharma, 2010) and Egypt (Hassan et al., 2013).

The average concentrations (ranges) of NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ in PM_2.5 were 19.8 (1.07–340.6), 20.5 (3.30–116.9), and 11.3 (1.09–109.3) µg m⁻³, respectively, at the NCP rural site during 15 June–11 August 2013. The average concentration of NH${}_{\mathrm{4}}^{+}$ in PM_2.5 was higher than that observed at the rural or urban sites in the NCP (Meng et al., 2011), the YRD (Meng et al., 2014), Beijing (Shen et al., 2011), Guangzhou (Hu et al., 2008), and Hong Kong (Yao et al., 2006) in China, and is comparable to that at the urban site in India (Behera and Sharma, 2010). The average concentration of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in PM_2.5 was higher than that at rural sites in the NCP (Meng et al., 2011) and YRD (Meng et al., 2014) in China, but was lower than that observed at rural sites in Guangzhou (Hu et al., 2008) in China as well as urban sites in India (Behera and Sharma, 2010) and Egypt (Hassan et al., 2013). The average concentration of NO${}_{\mathrm{3}}^{-}$ in PM_2.5 was higher than that observed at the rural sites in the YRD and Guangzhou (Hu et al., 2008) in China and lower than that at urban sites in India (Behera and Sharma, 2010) and Pakistan (Biswas et al., 2008). The elevated NH₃ and NH${}_{\mathrm{4}}^{+}$ in PM_2.5 concentrations at the NCP rural site demonstrate severe ammonia and fine particulate ammonium pollution in this area.

Figure 2Time series of hourly data of NH₃, other trace gases, and meteorological parameters measured during the sampling period (a) and a close-up of the period with extremely high NH₃ values during 27–31 July 2013 (b).

Download

3.2 Ambient ammonia

3.2.2 Diurnal variations in NH₃

The average diurnal variations in NH₃ from June to September 2013 are shown in Fig. 3. As indicated in Fig. 3a, NH₃ concentration maxima and minima were observed during 08:00–13:00 and 19:00–23:00, respectively. As for July, NH₃ concentrations showed a considerably more pronounced diurnal pattern with a maximum of 59.5 ppb at 08:00. The concentration of NH₃ gradually increased during 00:00–03:00, remained relatively constant during 04:00–06:00, and then rapidly increased from 06:00 (beginning just after sunrise). After peaking at approximately 08:00, a decrease was observed until it reached the minimum of 29.8 ppb at 19:00.

Figure 3Diurnal variation in NH₃ (a) and meteorological parameters (b) during the sampling period.

Download

The morning peak of NH₃ was also observed elsewhere and could have resulted from emissions from fertilised soils and plant stomata, evaporation of dew, human sources, and mixing down of ammonia from the residual layer (Trebs et al., 2004; Norman et al., 2009; Bash et al., 2010; Ellis et al., 2011). Figure 3b reveals that the relative humidity (90–89 %) and temperature (21.5–22.1 ^∘C) remained relatively constant before 06:00 but increased later in the morning. The increasing temperature can heat the Earth's surface and vegetation leaves and reduce the RH, potentially leading to evaporation of NH₃ from soil and plants and volatilisation of ammonium aerosol (Trebs et al., 2004; Norman et al., 2009; Ellis et al., 2011), which may increase NH₃ concentrations in the morning. When the emission was occurring into a shallow boundary layer, NH₃ increase would be more prominent. In addition, the morning rise might also be due to the break-up of the nocturnal boundary layer. During the sampling period, the majority of peaks of ammonia over 50 ppb occurred at night, which were attributed to local emissions such as from agricultural activity, into a shallow nocturnal boundary layer. It was supposed by Ellis et al. (2011) that the downward mixing of air containing higher NH₃ from the residual layer could lead to an increase in surface NH₃ after the break-up of the nocturnal boundary layer.

From Fig. 3a, it can be seen that in July the NH₃ level was the highest and peaked earliest. One reason for this might be the increased emissions of local agricultural NH₃ sources in July compared with those in June, August, and September. On average, the level NH₃ in July had a maximum night-time increase (20.0 ppb from 20:00 to 06:00), which is much large than those in June (5.2 ppb), August (9.9 ppb), and September (1.8 ppb). The early morning increase in NH₃ in July started from a much higher level than in other months, resulting in the earliest NH₃ peak in July.

The Gucheng site is an experiment station for agrometeorological studies. Corn is the main crop in the station area and nearly all the agricultural areas surrounding it. In accordance with the climate in the NCP, corn is planted around the middle of June and grows rapidly in July. Therefore, July is the key period for the application of nitrogen fertilisers like urea. As mentioned above, the urea application at the station on 20 July 2013 and a precipitation process afterwards caused huge NH₃ spikes during the end of July (Fig. 2b). In addition, the highest night-time temperature in July (Fig. 3b) could promote the soil emission of NH₃, and the relatively lower wind speed (Fig. 3b) and lower PBLH (Fig. S3) in July were in favour of the accumulation of NH₃ in surface air.

In summary, ambient NH₃ at Gucheng showed interesting diurnal cycles, which look significantly different in different summer months. We believe the interplay of some processes, such as emissions from agricultural sources, meteorological conditions (temperature, relativity humidity, wind speed, PBLH, etc.), and chemical conversion, are important in the determination of levels and diurnal patterns of NH₃ at the site. Are all of these processes important in the morning variation in NH₃ or not? How important are they? What makes the difference in the peaking time and concentration of NH₃ in different months? These are questions to be answered in the future.

3.3 Ambient ammonium aerosol

Secondary inorganic aerosols form from gas-phase precursors, which are mostly from anthropogenic activities such as industrial, agricultural, and motor vehicle emissions. Therefore, the major precursors (NH₃, SO₂, and NO_x) are responsible for the formation of particulate ammonium, sulfate, and nitrate.

The hourly NH${}_{\mathrm{4}}^{+}$ concentrations during 15 June–11 August 2013 ranged from 1.07 to 340.6 µg m⁻³, with an average concentration of 19.8 µg m⁻³. The highest monthly level of NH${}_{\mathrm{4}}^{+}$ appeared in July and the lowest level appeared in June 2013. Similar to NH₃, the concentration of NH${}_{\mathrm{4}}^{+}$ also increased sharply after urea fertilisation, with the highest value (340.6 µg m⁻³) observed at 09:00 on 28 July 2013. The temporal variations in NH${}_{\mathrm{4}}^{+}$ basically coincided with SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ (discussed in Sect. 3.6), reflecting that NH${}_{\mathrm{4}}^{+}$ largely originated from the neutralisation between NH₃ and acidic species.

The highest hourly SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentration (116.9 µg m⁻³) was observed at 10:00 on 9 July and the second highest value was 111.4 µg m⁻³ at 18:00 on 6 August 2013, with an average concentration of 20.5 ± 13.6 µg m⁻³. Despite the lower concentrations of SO₂, higher SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations in summer were attributed to the higher temperature, O₃ concentration, and solar radiation, which increase photochemical activities, atmospheric oxidation, and conversion of SO₂ to SO${}_{\mathrm{4}}^{\mathrm{2}-}$. The average concentration of NO${}_{\mathrm{3}}^{-}$ in PM_2.5 was 11.3 ± 9.1 µg m⁻³. The highest value of 109.3 µg m⁻³ was observed at 14:00 on 22 June 2013 at the highest RH (93 %) and AWC (910 µg m⁻³). The high RH conditions in summer might dissolve a significant fraction of HNO₃ and NH₃ in humid particles, therefore increasing the concentrations of NO${}_{\mathrm{3}}^{-}$ and NH${}_{\mathrm{4}}^{+}$ in the atmosphere (Krupa, 2003; Trebs et al., 2004; Ianniello et al., 2010). The high NO${}_{\mathrm{3}}^{-}$ concentrations were also mostly associated with large AWC, which indicates the importance of heterogeneous hydrolysis in the production of nitrate (Pathak et al., 2009). Conversely, NH₃ more efficiently in reacted with SO₂ in summer to form (NH₄)₂SO₄.

Figure 4 shows correlations of observed NH${}_{\mathrm{4}}^{+}$ with the sum of observed SO${}_{\mathrm{4}}^{\mathrm{2}-}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}+$NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}+$NO${}_{\mathrm{3}}^{-}+$Cl⁻. Although all the correlations are relatively high, the slopes of the regression lines are far from unity. This cannot be due to bias in measurements. The major ion balance shows a ratio of 1.05 : 1.0 for cation : anion. The slope is 0.56 when all three strong acids are considered, suggesting that the neutralisation of the strong acids explains 56 % of the observed NH${}_{\mathrm{4}}^{+}$. In other words, nearly 44 % of the observed NH${}_{\mathrm{4}}^{+}$ was due to the presence of other acids in aerosol particles.

Figure 4Correlation of observed NH${}_{\mathrm{4}}^{+}$ with observed SO${}_{\mathrm{4}}^{\mathrm{2}-}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}+$NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}+$NO${}_{\mathrm{3}}^{-}+$Cl⁻.

Download

3.4 Results from thermodynamic equilibrium simulation

We have used the thermodynamic equilibrium model ISORROPIA-II to investigate gas–aerosol partitioning characteristics. The model outputs include equilibrium SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, NH₃, NH${}_{\mathrm{4}}^{+}$, H${}_{\mathrm{air}}^{+}$, HNO₃, and AWC. As shown in Fig. 5, the modelled SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and NH₃ are highly correlated with the corresponding measurements, with the slopes of regression lines being 0.99, 1.01, and 1.13 for SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and NH₃, respectively. The correlation between modelled and observed NH${}_{\mathrm{4}}^{+}$ is comparably poor but still significant (R²= 0.46, P < 0.01), with a slope of 0.45. The modelled SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ agree nearly perfectly with the measurements, while the modelled NH₃ and NH${}_{\mathrm{4}}^{+}$ show a slight overestimation and a large underestimation of the respective measurements. Considering the unbalance between observed NH${}_{\mathrm{4}}^{+}$ and the sum of observed SO${}_{\mathrm{4}}^{\mathrm{2}-}+$ NO${}_{\mathrm{3}}^{-}+$ Cl⁻ (see Fig. 4), we believe that other acids in aerosol particles are important in the conversion of NH₃ to NH${}_{\mathrm{4}}^{+}$. These other acids may be oxalic acid and other dicarboxylic acids. Although we did not measure organic acids in aerosol, the presence of oxalic acid and other low-molecular-weight dicarboxylic acids in aerosols is often reported (e.g. Hsieh et al., 2007; Kawamura et al., 2010, 2013; Sauerwein and Chan, 2017). There is no doubt that there is a significant amount of dicarboxylic acid over the NCP particularly during summer (Kawamura et al., 2013). Therefore, it is highly possible that neutralising dicarboxylic acids in aerosol particles contributed significantly to the conversion of ammonia to ammonium. The presence of a significant amount of dicarboxylic acid can explain the substantial underestimate of NH${}_{\mathrm{4}}^{+}$ and slight overestimate of NH₃ by the model well as the model simulations did not include organics, which were not observed. The low slope (0.45) of the regression line for modelled and observed NH${}_{\mathrm{4}}^{+}$ implies that organic diacids contributed at least half to the conversion of ammonia to ammonium. Since more NH${}_{\mathrm{4}}^{+}$ existed in aerosol than was required for neutralising inorganic acids, the ISORROPIA model simulated higher equilibrium NH₃ than observed, leading to an overestimate of NH₃.

Figure 5Observed and modelled concentrations of NH₃, NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ in summer 2013.

Download

Comparisons between the measurements and the modelled values give a positive NMB for NH₃ (26.9 %) and a negative NMB for SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (−0.5 %), NO${}_{\mathrm{3}}^{-}$ (−12.5 %), and NH${}_{\mathrm{4}}^{+}$ (−32.6 %). NME gives an estimation of overall discrepancy between measurement and model (Sudheer and Rengarajan, 2015). The NME values for the model–observation comparison of NH₃, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and NH${}_{\mathrm{4}}^{+}$ are 31.6, 0.9, 14.8, and 45.8 %, respectively. These results indicate that the model simulated SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ well but over-predicted NH₃ and under-predicted NH${}_{\mathrm{4}}^{+}$ by around 30 %, mainly because of the neglect of organics. Therefore, organics should be observed and included in the model simulations in future studies to better investigate aerosol components and gas-to-aerosol conversion.

Figure 6Simulated diurnal variation in HNO₃ and H${}_{\mathrm{air}}^{+}$ (a) and calculated diurnal variation in the pH value of aerosol water (b) in summer 2013.

Download

The average concentration of simulated HNO₃ was 0.7 µg m⁻³, showing a maximum value of 7.41 µg m⁻³ at 11:00 on 19 June 2013. The average diurnal variations in HNO₃ and H${}_{\mathrm{air}}^{+}$ are shown in Fig. 6a. Typical high HNO₃ concentrations during daytime and low values at night-time during the observation period are predicted by the model, which is consistent with other studies (Makkonen et al., 2012; Sudheer and Rengarajan, 2015). The diurnal cycle of H${}_{\mathrm{air}}^{+}$ is predicted with the highest level around 17:00. The concentrations of NH₃ were closely associated with H${}_{\mathrm{air}}^{+}$, and higher NH₃ always corresponded to lower H${}_{\mathrm{air}}^{+}$ (Liu et al., 2017). The pH value of aerosol water, estimated based on the simulated results using Eq. (4), are mostly in the range of 2.5–4.5. The fine particles were moderately acidic in summer, with an average pH value of 3.5. On average, pH is over 3.5 during night-time and below 3.5 during daytime (Fig. 6b). Under the medium acidic conditions and high NH₃ concentrations, organic acid like diacids are able to react with ammonia to form ammonium. Because we used ISORROPIA-II for inorganic aerosol composition and no organic acid measurements are available, we cannot analyse the role of organic acids in detail, though the model performed quite well (Fig. S4).

Figure 7Relationship between NH₃ and NH${}_{\mathrm{4}}^{+}$ (a) and diurnal variation in NH_x (b) in summer 2013.

Download

3.5 Relationship between ammonia and ammonium aerosol

The gas-to-particle conversion between NH₃ and NH${}_{\mathrm{4}}^{+}$ has been reported to be strongly affected by temperature, RH, radiation conditions, the concentration of primary acid gas, and other factors. In this study, NH${}_{\mathrm{4}}^{+}$ concentrations correlated significantly and positively with NH₃, with a correlation coefficient of 0.78 and a slope of 1.48 (Fig. 7a, n= 915, P < 0.01), suggesting that NH₃ played an important precursor role in NH${}_{\mathrm{4}}^{+}$ in PM_2.5 formation.

The ratio of NH₃ to NH_x (NH₃+NH${}_{\mathrm{4}}^{+})$ has been used to identify the source of NH_x and the relative contribution of NH₃ to NH_x deposition (Lefer et al., 1999; Walker et al., 2004). A value higher than 0.5 signifies that NH_x is mainly from local NH₃ sources and that the dry deposition of NH₃ dominates the NH_x deposition. Robarge et al. (2002) reported that more than 70 % of NH_x was in the form of NH₃ at an agricultural site in the southeastern United States and concluded that given a larger deposition velocity of ammonia compared with that of ammonium, a considerable fraction of NH_x could be deposited locally rather than be transported out of the region. According to hourly average concentrations, the ratio of NH₃ ∕ NH_x varied from 0.22 to 0.97, with a mean ratio of 0.69 ± 0.14, suggesting that NH₃ remained predominantly in the gas phase rather than the aerosol phase in summer 2013 at Gucheng.

The diurnal changes in gaseous precursors and aerosol species are controlled by emission and deposition processes, horizontal and vertical transport, and gas–particle partitioning. To investigate gas–particle conversion further, diurnal variation in NH_x is shown in Fig. 7b. Between 08:00 and 18:00, a decrease in NH₃ would result in an increase in NH${}_{\mathrm{4}}^{+}$, which coincided with higher sulfate concentrations. The decrease in gas phase ammonia is likely the result of uptake onto aerosols to form (NH₄)₂SO₄. The diurnal variability in NH_x may be controlled by transport and vertical exchange. Between the hours of 08:00 and 18:00, NH₃ decreased by 43 % while NH_x decreased by 49 %, suggesting that NH₃ remained predominantly in the gas phase. Between 19:00 and 07:00, NH₃ increased by 42 % and NH_x increased by 51 %, indicating that gas–particle partitioning contributes significantly to the decrease in gas-phase ammonia during this time.

3.5.2 Diurnal patterns of NHR, SOR, and NOR

Figure 8 presents the diurnal patterns of NHR, SOR, NOR, gaseous precursors, major water-soluble ions, and meteorological factors. As a key species contributing to the oxidisation capacity of the atmosphere, O₃ can promote HNO₃ formation, affecting the conversion ratio of NH₃. O₃ exhibited low levels in the morning and enhanced levels in the late afternoon. The lower morning concentrations may be due to the depositional loss of O₃ under stable atmospheric conditions in early morning hours, and the higher levels in the afternoon could be due to the photochemical production of O₃. The NH${}_{\mathrm{4}}^{+}$ concentration started to increase from morning, reaching the maximum value (16.1 µg m⁻³) at 18:00, with a diurnal difference of 3.7 µg m⁻³. This diurnal pattern may be due to a combination of high NH₃ concentrations, the intense solar radiation at noon, and the high oxidisation capacity of the atmosphere in the afternoon. A clear diurnal cycle of NHR existed, with an amplitude of 0.10 and a peak of 0.35 at 18:00, which is consistent with the higher SOR and RH.

The SO₂ concentration showed a maximum at 09:00, with a secondary peak at 22:00. The concentration of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ showed small peaks at 11:00, 14:00, and 18:00, respectively, but no strong diurnal variation. SOR displayed a diurnal cycle with the highest value of 0.74 observed at 05:00. It is noted that the SOR was lower during daytime when photochemical reaction is intense. A higher SOR during night-time suggests the importance of dark reactions. SO₂ is highly soluble and can easily be absorbed by wet aerosol particles. The high RH at night may have promoted this process.

Figure 8Diurnal variation in NHR, SOR, NOR, gaseous precursors, major water-soluble ions, and meteorological factors in summer 2013.

Download

As for the diurnal cycle of NO_x, a peak was observed at 06:00 when the mixing layer was stable, and a broad valley was observed in the daytime, reflecting the influences of a higher mixing layer and stronger photochemical conversion. During the night, NO_x concentrations increased again, resulting in the second maximum at 23:00. NO${}_{\mathrm{3}}^{-}$ concentrations did not show profound diurnal variations but instead showed slightly higher values during the night time, probably because of the hydrolysis of dinitrogen pentoxide (N₂O₅) and the condensation of HNO₃ under relatively low temperatures. NOR displayed a diurnal pattern with a maximum of 0.28 at 10:00, which was likely related to photochemical reactions under the conditions of high O₃ concentrations and jNO₂ levels.

Night-time formation, aerosol uptake, and hydrolysis of N₂O₅ are highly uncertain as has been pointed out (e.g. Xue et al., 2014). The NO_x concentration during night-time was higher than during daytime, while the NO${}_{\mathrm{3}}^{-}$ level during night-time was only slightly higher than that during daytime. By assuming a high aerosol surface-to-mass ratio (33.7 m² g⁻¹; Okuda, 2013) and a high uptake coefficient (0.1; Seinfeld and Pandis, 2006), we estimate the night-time N₂O₅ under the conditions over our site to be in the range of about 3–10 ppb, corresponding to a HNO₃ production rate of about 1–3 ppb h⁻¹ (or 2.6–7.7 µg m⁻³). This rate of HNO₃ production would cause an obvious night-time production of NH${}_{\mathrm{4}}^{+}$. Indeed, we can see increases in the NH${}_{\mathrm{4}}^{+}$ concentration and NHR at night (Fig. 8). However, a more or less accurate estimate of the relative contribution of the night N₂O₅ chemistry to NH₃ conversion needs to be made in the future.

3.6 A case study of a pollution period

On several days during the study period, very high NH₃ and inorganic PM_2.5 concentrations were observed. Here we make a case study of a pollution period during 7–11 August 2013. Data on gases, major aerosol ions, and some key meteorological parameters are presented in Fig. 9. Some calculated parameters during this period are given in Fig. S5. As shown in Figs. 9 and S5, there was a sharp increase in NO_x during the night and early morning of 10 August, followed by an increase in NH₃ (peak value of 64 ppb) at 03:00. In the meantime, a large peak of AWC occurred and gaseous HNO₃ decreased to nearly zero (Fig. S5), suggesting rapid uptake of wet aerosol. This event caused the first largest peak of SO${}_{\mathrm{4}}^{\mathrm{2}-}+$ NO${}_{\mathrm{3}}^{-}+$ NH${}_{\mathrm{4}}^{+}$. After this event, NH₃ rose again and reached an even higher peak (76.3 ppb) shortly before noon on 10 August. This peak of NH₃ coincided with a valley of NO_x, but the HNO₃ level increase and pH value decrease were observed in parallel. A few hours later SO₂ showed a large peak and the second largest peak of SO${}_{\mathrm{4}}^{\mathrm{2}-}+$ NO${}_{\mathrm{3}}^{-}+$ NH${}_{\mathrm{4}}^{+}$ occurred. These data show that a high NH₃ concentration was accompanied by a large increase in concentrations of SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and NH${}_{\mathrm{4}}^{+}$, which is consistent with the view that NH₃ plays an important role in particulate matter mass formation and that gas–particle conversion occurred when NH₃ was available, though SO${}_{\mathrm{4}}^{\mathrm{2}-}$ partitions to the aerosol phase impact NH₃ level less (Gong et al., 2013).

Figure 9Hourly concentrations of gaseous species, ionic species, and jNO₂ measured in the pollution episode during 7–11 August 2013.

Download

The secondary ion concentrations had similar temporal distributions with slow accumulation and relatively rapid clearing under favourable meteorological conditions. There was a good correlation between NH₃ and NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ (R= 0.33, 0.27, and 0.49, respectively, with P < 0.01). However, there was also a situation in which high NH₃ was not associated with high SO${}_{\mathrm{4}}^{\mathrm{2}-}+$ NO${}_{\mathrm{3}}^{-}+$ NH${}_{\mathrm{4}}^{+}$, as indicated by the data around noon of 8 August (Fig. 9). During this case, AWC was extremely low and RH was around 40 %. These conditions do not favour heterogeneous reactions.

Figure 10Correlations between NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}+$ NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}+$ NO${}_{\mathrm{3}}^{-}+$ Cl⁻ during 7–11 August 2013.

Download

Table 2Occurrence frequency and mean values of NH₃, other trace gases (ppb), and ionic species in PM_2.5 (µg m⁻³) for each type of air mass arriving at Gucheng in summer 2013.

Download Print Version | Download XLSX

During 7–11 August 2013, the relationships of the observed NH${}_{\mathrm{4}}^{+}$ versus those of SO${}_{\mathrm{4}}^{\mathrm{2}-}$, the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$, and the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and Cl⁻ are presented in Fig. 10. It is known that (NH₄)₂SO₄ is preferentially formed and the least volatile, NH₄NO₃ is relatively volatile, and NH₄Cl is the most volatile. NH${}_{\mathrm{4}}^{+}$ is thought to be first associated with SO${}_{\mathrm{4}}^{\mathrm{2}-}$. Afterwards, the excess of NH${}_{\mathrm{4}}^{+}$ is associated with nitrate and chloride (Meng et al., 2015). It is noted that the correlation of NH${}_{\mathrm{4}}^{+}$ with the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ (R= 0.91, slope = 1.23, with P < 0.01) was closer to unity than that of NH${}_{\mathrm{4}}^{+}$ with SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (R= 0.80, slope = 1.65, with P < 0.01), suggesting that both SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ were associated with NH${}_{\mathrm{4}}^{+}$. As shown in Fig. 10, sulfate and nitrate were almost completely neutralised with most of the data above the 1:1 line. A few scattered data below the 1:1 line may be caused by uncertainties in measurements. Little difference was found between the regression slopes of NH${}_{\mathrm{4}}^{+}$ with the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ and the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and Cl⁻ due to the very low amount of NH₄Cl. In this study, the level of NH₃ was high enough to neutralise both SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ and likely to help form (NH₄)₂SO₄ and NH₄NO₃. In addition to these substances, it is likely that NH₃ also reacted with oxalic acid and other dicarboxylic acid to form ammonium oxalate and other organic ammonium aerosols, as discussed above. Data in Fig. 10 seem to be distributed in two groups, with the one (Group 1) covering low to middle levels but exhibiting a larger slope and the other (Group 2) covering middle to high levels but exhibiting a smaller slope. Data in Group 2 were mainly from 10 August 2013 when high NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ values were observed (see Fig. 9). The difference in slope between the two groups might be caused by different fractions of organic diacids that are considered to convert more ammonia to ammonium. Although no measurements of organic diacids are available to prove this, the significant changes in fractions of organic and inorganic aerosol components are often observed during pollution episodes in China (e.g. Wang et al., 2016). Reduced impact from organic diacids may bring the correlation of NH${}_{\mathrm{4}}^{+}$ with the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and Cl⁻ closer to the 1:1 line.

Figure 11The average NH₃, NH${}_{\mathrm{4}}^{+}$, and meteorological data roses in different wind sectors during summer 2013.

Download

3.7 Transport effects on local ammonia and ammonium

The Gucheng site is located in a densely populated rural area in the NCP, and it is influenced by local sources in the surrounding areas and by long-range transport of pollutants from the residential and industrial centres around it. Dependence of the concentrations of NH₃ on wind direction at Gucheng is studied to get insight into the distribution of local emission sources around the monitoring site. As shown in Fig. 11, during the sampling period, the prevailing surface winds at Gucheng were northeasterly and southwesterly. High NH₃ originated from the southwest sector of the measurement site, which may be due to a local unidentified agricultural or industrial source or transport from the Xushui township, which is approximately 15 km away from Gucheng. Lower NH₃ concentrations were observed under winds from other sectors. Since NH₃ is either readily converted to NH${}_{\mathrm{4}}^{+}$ or subjected to dry deposition, high concentrations are found only close to the surface and near the emission sources. Previous studies have reported an inverse relationship between ground-level concentrations of trace gases, such as ammonia, and wind speed (Robarge et al., 2002; Lin et al., 2011; Meng et al., 2011). Thus, NH₃ concentrations might be generally lower at higher wind speeds because of turbulent diffusion.

To identify the impact of long-range air transport on the surface air pollutant levels and secondary ions at Gucheng, the 72 h backward trajectories were calculated using the HYSPLIT 4.9 model.

Figure 12The 72 h backward trajectories for 100 m above ground level at Gucheng during the sampling period in 2013.

Download

As can be seen in Fig. 12, Clusters 1, 2, and 3 represent relatively low and slowly moving air parcels, with Cluster 2 coming from northwest areas at the lowest transport height among the five clusters. The air masses of Clusters 1 and 3 originate from the southeast of Gucheng. Clusters 4 and 5 represent air parcels mainly from the far northwest.

The trajectories in Cluster 2 came from the local areas around Gucheng, and it was the most important cluster to the Gucheng site, contributing 56 % to the air masses. Based on the statistics, the number of trajectories in Clusters 1, 2, and 3 account for 88 % of all the trajectories. As more than 80 % of air masses originating from or passing over the NCP region can influence the surface measurements at Gucheng, the observation results at Gucheng can represent the regional state of atmospheric components in the NCP region well.

Since the emission sources of pollutants are unevenly distributed in the areas surrounding the Gucheng site, air masses come from different directions and contain different levels of pollutants. The corresponding mean concentrations of NH₃, SO₂, NO_x, NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ in PM_2.5 in different clusters of backward trajectories are also included in Table 2 in order to characterise the dependence of the pollutant concentrations on air masses.

Large differences in the concentrations of NH₃, SO₂, NO_x, NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ in PM_2.5 existed among the different clusters, with cluster 2 corresponding to the highest NH₃ concentration (48.9 ppb) and second highest NO_x, NH${}_{\mathrm{4}}^{+}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (14.4 ppb and 17.5 and 22.1 µg m⁻³, respectively).

Cluster 1 corresponds to the highest SO₂, NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ concentrations (7.9 ppb and 22.3, 22.6 and 17.7 µg m⁻³, respectively) and the second highest NH₃ level (32.8 ppb). Cluster 3 had the highest NO_x level (15.1 ppb), the second highest SO₂ and NO${}_{\mathrm{3}}^{-}$ levels (4.8 ppb and 11.8 µg m⁻³, respectively), and had the third highest concentrations of NH_3, NH${}_{\mathrm{4}}^{+}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (28.5 ppb and 14.6 and 20.2 µg m⁻³, respectively).

Based on Table 2, the lowest NH₃, SO₂, NO_x, NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ levels corresponded to Cluster 5, which was expected to bring cleaner air masses to the surface. As demonstrated by backward trajectories, more than half of the air masses during the sampling period from the NCP region contributed to the atmospheric NH₃ variations, and both regional sources and long-distance transport from the southeast played important roles in the observed ammonium aerosol at the rural site in the NCP.