Revisiting the Concentration Observations and Source Apportionment of Atmospheric Ammonia

Ammonia (NH₃) is the most abundant alkaline gas in the atmosphere. While NH₃ has a beneficial role in buffering acid rain (Wang et al., 2012), after deposition it can detrimentally affect Earth’s ecosystems through soil acidification, water eutrophication, and biodiversity loss (Liu et al., 2019). The overabundance of NH₃ in the lower atmosphere is suggested to promote the formation of secondary ammoniated aerosol particles (Wang et al., 2016), with significant impacts on visibility deterioration and human health (An et al., 2019). Recently, NH₃ and ammonium nitrate particles were also found in the upper troposphere during the Asian monsoon and play a hitherto neglected role in ice cloud formation and aerosol indirect radiative forcing (Höpfner et al., 2019). However, the severe lack of NH₃ measurements with sufficient spatial and temporal coverage is currently a barrier to understanding the vital role of NH₃ in air pollution, ecosystem protection, and climate change. It has resulted in unclear regulatory guidelines for mitigating these effects (Pan et al., 2020b).

Anthropogenic emissions of NH₃ in China are more significant than the total emissions of the U.S. and the European Union (Liu et al., 2019). To date, there is still no national NH₃ concentration monitoring network operated by the Chinese government. Following the guidelines of the National Atmospheric Deposition Program in the U.S., the Institute of Atmospheric Physics, Chinese Academy of Sciences, established a Regional Atmospheric Deposition Observation Network in the North China Plain (READ-NCP). This network, including 10 sites covering different land-use types, started monitoring NH₃ concentrations in 2007, and has also obtained significant results with respect to the atmospheric deposition of nitrogen, carbon, sulfate, and metals. Based on the observations of READ-NCP from 2008 to 2010, NH₃ was found to be a significant contributor to nitrogen deposition in this region (Pan et al., 2012). Thus, clarification of NH₃ levels in China can aid policymakers in the protection of ecosystems from excess nitrogen deposition. Due to the lack of data, however, the whole picture of NH₃ distribution in China was poorly understood. In 2015, READ-NCP was extended to a spatially dense and cost-efficient network focusing on NH₃ observations in China (AMoN-China) (Pan et al., 2018). The system currently consists of approximately 100 sites, which is similar to that of the U.S. AMoN (Fig. 1). While the NH₃ concentration was relatively low in the U.S., there is an increasing importance of deposition of reduced nitrogen due to the significant reduction in oxidized nitrogen (Li et al., 2016).

Figure 1.  Surface ammonia concentrations during 2015−16 observed by AMoN in (a) the U.S. (http://nadp.slh.wisc.edu/AMoN/), (b) the UK (https://uk-air.defra.gov.uk/), and (c) East Asia (https://www.eanet.asia) including China (Pan et al., 2018). (d) Long-term surface measurements of ammonia in Africa within the framework of the IDAF (IGAC-DEBITS-AFRICA) program (mean values from 1998 to 2007) (Adon et al., 2010). Global ammonia morning column measurements (2008–16) observed from space by IASI are also shown (https://doi.pangaea.de/10.1594/PANGAEA.894736).

Besides AMoN in China (Pan et al., 2018) and the U.S. (http://nadp.slh.wisc.edu/AMoN), the monitoring of surface NH₃ is also conducted by other networks (Fig. 1), e.g., EANET (The Acid Deposition Monitoring Network in East Asia; https://www.eanet.asia), EMEP (the Co-operative Programme for Monitoring and Evaluation of the Long-Range Transmission of Air Pollutants; http://ebas.nilu.no/Default.aspx) and the IDAF (IGAC-DEBITS-AFRICA) program for African ecosystems (Adon et al., 2010). Most of these networks employed a cost-effective approach by using passive samplers, including ALPHA, Analyst, Radiello, and Ogawa, which have advantages in characterizing the spatial distribution and long-term trends of NH₃. However, the accuracy of these passive NH₃ sampling techniques is not well validated in the field, which represents one of the biggest challenges in NH₃ monitoring (Martin et al., 2019). For example, it is reported that the NH₃ concentrations collected by Radiello passive samplers are approximately 40% lower than the denuder-based reference method (Puchalski et al., 2011). The low NH₃ concentration bias in the passive collection samplers was suggested to be the result of inaccurate effective sampling rates due to incorrect mass transfer correction factors for the environmental conditions (Pan et al., 2020a). Thus, questions remain as to whether the NH₃ concentrations from different networks can be directly compared if they employed different passive samplers. Concurrent measurements of the passive samplers used in various networks are thus further needed, with a collocated reference method, e.g., annular denuders and continuous real-time instruments employing the wet chemistry technique (von Bobrutzki et al., 2010; Martin et al., 2019; Pan et al., 2020a).

The need for source apportionment has increased in recent years as atmospheric NH₃ concentrations and deposition fluxes have shown little change or even increased following more stringent air pollutant controls (Liu et al., 2018). Long-term satellite observation from the Atmospheric Infrared Sounder (AIRS) aboard NASA’s Aqua satellite also implied that NH₃ levels over agricultural regions had experienced significant increasing trends between 2002 and 2013, with an annual increase rate of 2.6%, 1.8% and 2.3% in the U.S., the European Union, and China, respectively (Warner et al., 2017). The increment of atmospheric NH₃ concentrations tended to continue between 2013 and 2017, as observed from space with the Cross-track Infrared Sounder (CrIS) (Shephard et al., 2020). While agricultural activities (fertilization and livestock volatilization) are known to dominate the emissions of NH₃, accounting for over 60% and 80% of the global and Asian inventory (Bouwman et al., 1997; Huang et al., 2012), non-agricultural sources have been suggested as a major NH₃ source at the urban scale (Felix et al., 2014; Pan et al., 2016; Sun et al., 2017; Chang et al., 2019; Walters et al., 2020a).

Ammonia emissions in developing cities are especially important because of their high emissions ratios to CO₂ and rapidly expanding vehicle fleets (Sun et al., 2017). For example, vehicular emissions were found to be a critical NH₃ source in urban Beijing (Ianniello et al., 2010; Meng et al., 2011). Industrial NH₃ emissions, rather than those from vehicles, were also identified in the megacity of Shanghai (Wang et al., 2015). However, in contrast to previous results, Teng et al. (2017) suggested that urban green spaces and evaporation of deposited NH_x (NH₃+NH₄⁺) on wet surfaces, rather than traffic and agricultural emissions, were the primary source for NH₃ in an urban environment during winter in NCP. Thus, there is still no consensus on whether these emissions are among the major sources of urban atmospheric NH₃. Currently, the rapid development of isotope techniques is promising (Liu et al., 2014) and may provide scientists and policymakers with a more robust methodology and reliable evidence to track atmospheric NH₃ sources (Felix et al., 2014; Pan et al., 2016; Chang et al., 2019; Walters et al., 2020a).

The use of nitrogen isotopic composition of NH₃ (δ¹⁵N-NH₃) as a fingerprint identification of NH₃ emissions sources requires distinguishable isotopic signatures (Felix et al., 2013). While this technique has been widely used in Chinese cities, e.g., Beijing (Pan et al., 2016; Zhang et al., 2020) and Shanghai (Chang et al., 2019), considerable uncertainties remain in characterizing the endmembers. In particular, current collection methods are almost exclusively based on passive samplers, which have not been verified for their suitability to characterize δ¹⁵N-NH₃ accurately. Recently, Walters and Hastings (2018) validated an active sampling collection technique using an acid-coated honeycomb denuder to characterize δ¹⁵N-NH₃ under a variety of laboratory-controlled conditions as well as under field conditions. As a reference to this new verified method, Walters et al. (2020a) also found a substantial low bias of 15‰ in the ALPHA passive sampler in characterizing δ¹⁵N-NH₃ from traffic plumes. Such a low bias of passive samplers in characterizing δ¹⁵N-NH₃ was also confirmed in field observations in urban Beijing by Pan et al. (2020a). Thus, previous source apportionment needs to be reevaluated if using an inventory of δ¹⁵N-NH₃ based on passive samplers, especially the ALPHA sampler.

To evaluate the potential influences of the low bias of δ¹⁵N-NH₃ by passive samplers, we revisited the sources of atmospheric NH₃ in urban Beijing using a Bayesian isotope mixing model (SIAR, Stable Isotope Analysis in R) (Kendall et al., 2007). Two scenarios were performed based on an isotopic inventory with and without correction for the passive collection δ¹⁵N-NH₃ bias (Fig. 2). Accordingly, the model was run with δ¹⁵N-NH₃ values of −18.2‰ (corrected) and −33.2‰ (original uncorrected) as input for ambient samples. The latter value represented an annual mean δ¹⁵N-NH₃ value in urban Beijing based on a year-round and weekly collection by the passive ALPHA sampler (Zhang et al., 2020).

Figure 2.  The nitrogen isotopic composition of ammonia characterized at various endmembers. Recent reported isotopic signatures from traffic plumes, fertilizer and livestock (Ti et al., 2018; Kawashima, 2019; Walters et al., 2020a) were updated based on the previous summary by Walters and Hastings (2018). Note that the field sampling was conducted by different collection methods (legend) and is grouped by passive against active samplers (symbols with colors). To correct the low bias of passive data (gray symbols), 15‰ was added to the original values and is shown as corrected (symbols with colors) accordingly. Symbols with the same color and shape represent a series of observations during the same campaign. Data sources: (a) Freyer (1978); (b) Hristov et al. (2009); (c) Heaton (1987); (d) Savard et al. (2017); (e) Smirnoff et al. (2012); (f) Ti et al. (2018); (g) Felix et al. (2013); (h) Walters et al. (2020b); (i) Kawashima (2019); (j) Felix et al. (2014); (k) Chang et al. (2016).

Figure 3a demonstrates that non-agricultural sources contributed only 57% of NH₃ using the inventory without correction (Fig. 2), which is lower than the original estimation of ~72% by Zhang et al. (2020). This difference implied the impacts of different selection of source signatures in these two studies. Also, we have apportioned the source of NH₃ with corrected δ¹⁵N-NH₃ values of both inventories and samples by adding 15‰ to the corresponding passive sampler measurement data. The results showed that 66% of NH₃ was from non-agricultural emissions (Fig. 3b). This attribution may be more reliable due to the updated inventory. The different contributions between Figs. 3a and b for each source, in particular for fertilizers, industry, and vehicles, indicated the uncertainty introduced by the low δ¹⁵N-NH₃ bias of passive samplers.

Figure 3.  Source apportionment of atmospheric ammonia in urban Beijing based on isotopic inventory (a) without and (b) with correction for the passive collection bias in characterizing nitrogen isotopic composition of ammonia, as shown in Fig. 2. The nitrogen isotopic values of −18.2‰ (corrected) and −33.2‰ (original) were selected as input for ambient ammonia samples. The original isotope data of −33.2‰ were the annual mean values observed between March 2016 and March 2017 by Zhang et al. (2020).

It is important to note that tropospheric NH₃ concentrations can be reduced through tight control measures; else they will continue to increase. Constraining NH₃ sources utilizing stable nitrogen isotopes can aid policymakers to draft a mitigation strategy for NH₃ emissions, but this method depends on an accurate characterization of δ¹⁵N-NH₃ from both source and receptor sites. While the isotopic inventory has significant impacts on the source apportionment, a verified collection technique is warranted to improve the source inventory of δ¹⁵N-NH₃. Due to the different lifetime of NH₃ and NH₄⁺ in the atmosphere, the sources of NH₃ and NH₄⁺ at a given site may also be different. Thus, a better knowledge of nitrogen fractionation via atmospheric processes, e.g., gas-to-particle conversion, also helps in source apportionment of atmospheric NH₃ and NH₄⁺. To address this concern, the concurrent determination of different chemical speciation (i.e., δ¹⁵N-NH₃ and δ¹⁵N-NH₄⁺) is highly needed.

Acknowledgements. This study was supported by the National Key Research and Development Program of China (Grant No. 2017YFC0210100), National Research Program for Key Issues in Air Pollution Control (Grant No. DQGG0208) and the National Natural Science Foundation of China (Grant No. 41405144). WWW acknowledges support from the Atmospheric and Geospaces Sciences U.S. National Science Foundation (Grant No. AGS 1351932). We acknowledge the U.K. Department for Environment Food & Rural Affairs (uk-air.defra.gov.uk) as the source of the UK ammonia data (© Crown 2020 copyright Defra via uk-air.defra.gov.uk, licenced under the Open Government Licence). © Crown copyright 2020 and reproduced by permission of the Controller of HMSO and the Queen’s Printer for Scotland.