1. Introduction
Deposition to Earth’s surface is the major sink for soluble atmospheric constituents and an important source of nutrients (including N, P, and Fe) and toxic substances (e.g., Hg, Pb) for aquatic, terrestrial, and marine ecosystems. Relative to conditions prior to the industrial revolution, increases in the global human population, per capita resource usage, and associated demands for food and energy have driven significant increases in anthropogenic emissions of S, N, trace metals, base cations, and other materials to the atmosphere and corresponding deposition fluxes to the surface particularly in North America, Europe, and Asia. In addition, regulatory mandates such as the phase out of leaded gasoline (e.g., Shen and Boyle 1987) and restrictions on S and N emissions from fossil-fuel-combustion sources (e.g., Keene et al. 2014) have led to corresponding reductions relative to preregulatory levels in atmospheric concentrations and depositions in downwind regions. Most soluble or nonvolatile compounds emitted to the atmosphere or produced therein from the oxidation of precursors exhibit average atmospheric residence times from tens of hours to a week or more and, consequently, changes in deposition fields in response to changes in emissions extend over broad geographic regions. The biogeochemical perturbations caused by enhanced deposition relative to preindustrial levels include acidification of soils and surface waters, eutrophication, losses of biodiversity, degradation of ecosystem structure and functioning, accelerated rates of corrosion of surfaces, decreased visibility, and negative impacts on human health (e.g., Likens 2013).
Precipitation composition and fluxes have been investigated intensively over the past 35 yr. Many nations in heavily populated regions of the world established networks to quantify temporal and spatial variability in constituent concentrations and wet-deposition fluxes (e.g., Vet et al. 2014). Such programs have been essential in characterizing the influences of humans on the composition of the troposphere and the associated impacts of atmospheric deposition on receiving ecosystems. These data have also been essential in developing and validating regional- and global-scale emissions inventories and chemical-transport models (e.g., Lamarque et al. 2013). In contrast, atmospheric deposition over the more extensive remote regions of the world has received relatively less attention, thereby constraining our understanding of the processes that control tropospheric composition as well as our ability to model atmospheric deposition and associated environmental impacts over the full range of human influence. In particular, available data from remote regions provide useful context for evaluating impacts of anthropogenic processes in populated and industrial regions (e.g., Galloway et al. 1984; Likens et al. 1987). In addition, as the human population continues to grow, many formerly remote regions are undergoing rapid development. Anthropogenic greenhouse warming is also driving large-scale global changes including modification of regional climatic conditions, biogeochemical processes, nutrient inputs, and ecological response in remote areas (Ciais et al. 2014). Measurements of precipitation composition and wet-deposition fluxes in these regions provide important benchmarks against which to gauge such ongoing changes in earth systems.
An earlier analysis of precipitation composition at multiple remote locations was based on 163 samples collected at St. Georges, Bermuda; Amsterdam Island, Indian Ocean; Poker Flat, Alaska; San Carlos de Rio Negro, Venezuela; and Katherine, Australia (Galloway et al. 1982). The longest period of record (for Bermuda) included in that analysis corresponded to less than 1 yr. In this paper, we report results from four research programs that characterized precipitation composition and wet-deposition fluxes at remote marine, coastal, and terrestrial regions of the world. These data correspond to 4958 discrete samples collected on an event or daily basis at 14 sites over a span of 18 yr under the auspices of the Global Precipitation Chemistry Project (GPCP; 1979–95), the Western Atlantic Ocean Experiment (WATOX; 1980–89), the Atmosphere/Ocean Chemistry Experiment (AEROCE; 1988–96), and the Cordillera de Piuchue Ecosystem Study (CPES; 1997–99) (Table 1; Fig. 1).
Precipitation sampling sites.
Locations at which precipitation was sampled for chemical characterization.
Citation: Journal of the Atmospheric Sciences 72, 8; 10.1175/JAS-D-14-0378.1
Various subsets of data reported herein (most of which correspond to relatively shorter periods of record) have been interpreted previously including those for Amsterdam Island (Galloway and Gaudry 1984; Moody et al. 1991; Miller et al. 1993); Katherine, Australia (Keene et al. 1983; Likens et al. 1987); Bermuda (Church et al. 1982; Jickells et al. 1982; Moody and Galloway 1988; Galloway et al. 1989, 1993); Torres del Paine, Chile (Galloway et al. 1996); Lijiang, China (Keene et al. 1989); and Poker Flat, Alaska (Dayan et al. 1985). Subsets of data from multiple sites have also been interpreted to investigate aspects of larger-scale biogeochemical cycles (e.g., Keene and Galloway 1988; Graedel and Keene 1995; Dentener et al. 2006) and to estimate the nature and magnitude of anthropogenic influences in polluted regions (e.g., Galloway et al. 1984; Likens et al. 1987). Previously reported data for all of the above locations together with unpublished results from more recent years have been integrated here into a quality-assured database and interpreted to 1) establish representative ranges in precipitation composition in the absence of significant anthropogenic influences; 2) quantify spatial, seasonal, and interannual variability in precipitation composition and wet-deposition rates in remote areas; and 3) assess corresponding long-term trends in the composition of precipitation.
2. Methods
a. Sampling sites
1) Ireland
Precipitation was sampled on the west coast of Ireland at Adrigole (ADG) from November 1984 to January 1990 and at Mace Head (MHG and MHT) from December 1988 to August 1994 (Table 1). The ADG sampling site is southwest of Adrigole, a small coastal village on the Beara Peninsula (population of ~450), within a private, lightly grazed pasture near one of the sites of the Atmospheric Lifetime Experiment (ALE)/Global Atmospheric Gases Experiment (GAGE)/Advanced GAGE (AGAGE) program (Prinn et al. 2000). The site at Mace Head is located about 90 m inland from the coast, approximately 70 km west-northwest of Galway (population of ~70 000), and was operated in cooperation with University College, Galway. The site has an open westerly exposure to the North Atlantic, with only one uninhabited island several kilometers to the west. The land surrounding the site consists of granite outcroppings, blanket peat bogs, and grassland. During the first 9 months of operation, precipitation was sampled at ground level (MHG). After erection of a 23-m-tall aluminum scaffolding tower, precipitation was sampled from the tower top (MHT). Comparison of results indicates that the non-sea-salt (nss) compositions of precipitation sampled at both elevations are statistically indistinguishable (i.e., slopes and intercepts for reduced major axis regressions of paired data were indistinguishable from 1.0 and 0.0, respectively, at p of 0.05) and, consequently, nss results have been merged into a single dataset (MHGT).
Because of its proximity to the Gulf Stream, the western coast of Ireland experiences a maritime, temperate-latitude climate with relatively cool summers and mild winters that are influenced year round by transient low pressure systems. As a result, wind speed and direction (although predominantly from the west) are variable. The transition from winter to summer circulation is associated with the intensification and northward displacement of the Atlantic anticyclone in conjunction with the weakening of the Icelandic low. Occasionally, coastal land–sea-breeze circulation patterns and easterly winds associated with a secondary low develop in the region. Both conditions may transport pollutants from western Europe over the west coast of Ireland.
2) Bermuda, North Atlantic Ocean
Precipitation was sampled on Bermuda at Harbour Radio Tower (HRT) from April 1980 to August 1989 and at Tudor Hill (BTT) from July 1988 to February 1997. Precipitation was also sampled at BTT from July 2006 to June 2009 as reported and interpreted by Keene et al. (2014). Data from this latter period are used in this analysis only to extend the duration or record for evaluation of long-term trends. All other results for BTT reported herein correspond to the continuous period of record from 1980 to 1997. HRT overlooks the town of St. Georges on the northeast end of the island. The sampling site was located on a grassy hill 59 m above mean sea level with widely scattered trees and an open exposure in all directions. At Tudor Hill on the southwest coast about 20 km from HRT, the sampler was deployed at the top of a 23-m-tall aluminum scaffolding tower, the base of which was about 28 m above mean sea level.
The island of Bermuda consists of a variated carbonate dune field (maximum elevation 150 m), which is deposited on the southeast flanks of a submerged volcanic pedestal about 1000 km east of Cape Hatteras, North Carolina, United States. The island is about 30 km long and 2 km wide and has a permanent population of about 60 000 augmented by a large number of tourists but limited industry. The main town, Hamilton, is situated near the middle of the island; the island’s main power production (by fuel oil turbine) is also located in that area. The second largest town is St. Georges at the east end. During the period of sampling, cruise ships (approximately three per week) docked at both towns during the spring through fall season. A major airport is located west of St. Georges. Bermuda is influenced by a wide range of meteorological conditions associated with transport from North America, the central North Atlantic Ocean, Europe, and Africa. The island’s small size and low topographical relief also result in minimal local (heat island and orographic) influences.
3) Barbados (BAT), North Atlantic Ocean
Precipitation was sampled from November 1988 to February 1995 at the top of a 17-m-tall aluminum scaffolding tower erected on a rocky promontory at Ragged Point, St. Phillip (Table 1); the tower base is 30 m above mean sea level. The island is 32 km long by 22 km wide and is composed of uplifted oceanic sediments. The population of about 250 000 people is concentrated on the southern and western parts of the island, housing density is relatively low in the vicinity of the sampling site, and there is no heavy industry on the island. The airport is 10 km southwest of the site and most ship traffic passes to the west of the island. Barbados lies in the trade winds, which allows sampling of air parcels with long overocean trajectories. Winds are easterly except during the passage of tropical depressions and storms.
4) Amsterdam Island (AMI), Indian Ocean
Precipitation was sampled from May 1980 to November 1996 2 km southeast of the northwest shoreline at an elevation of 200 m above mean sea level (Table 1). Amsterdam Island is part of the French Southern and Antarctic Lands. It is a small (9 km by 5 km), volcanic (extinct) island, located in the prevailing westerlies, approximately 4200 km southwest of Perth, Australia, and 5000 km east of southern Africa. On average, rain falls 241 days yr−1 (Miller et al. 1993). Scientists who permanently staff a small research station are supplied by ship twice per year. The island fauna includes seasonal populations of fur seals and seabirds, as well as a permanent population of cattle. The seals are uniformly distributed along the coast, while large colonies of penguins and albatross are located on the southwest shore of the island (Moody et al. 1991).
5) Chile
Precipitation was sampled in southwestern Chile at Torres del Paine National Park (TDP) from April 1984 to December 1993 and at Cordillera de Piuchue (CDP) from February 1997 to February 1999 (Table 1). TDP is comprised of approximately 240 000 ha located 400 km north of Punta Arenas (population ~120 000) and approximately 50 km downwind from the South Pacific Ocean coastline. TDP and its surroundings are formed by a great diversity of landforms, glaciers, mountains, and plains. The mountains rise to 3050 m and the freshwater bodies are of glacial origin. The Andes Mountains are located between TDP and the Pacific Ocean and diminish the impact of marine influences on precipitation composition at TDP. Average monthly temperatures between 1979 and 1984 ranged from 1.9°C in July to 12.1°C in February (Torres del Paine National Park 1995, unpublished manuscript).
CDP is in the Chiloe National Park (45 000 ha), an old-growth, temperate rain forest ecosystem on the remote western slope of Chiloe Island in southern Chile. The city of Castro (population ~30 000) is located on the opposite coast of Chiloe Island approximately 50 km to the east. The sampling site is at an altitude of 650 m, approximately 15 km downwind from the South Pacific Ocean coastline. The park comprises temperate evergreen forests and open moorlands that have not been significantly impacted by air pollution or other human disturbances and have remained floristically stable throughout the Holocene. Persistent westerlies transport air to the site from the South Pacific Ocean. No roads or structures lie within a broad upwind fetch from the northwest to the southeast and with the exception of light, intermittent ship traffic offshore, there are no significant anthropogenic sources (combustion or agricultural) for thousands of kilometers upwind under typical airflow conditions. Abundant precipitation falls throughout the year peaking during the austral winter; clouds envelop the site about 30% of the time. A small dormitory and field laboratory support onsite operations. Power is provided by wind generators and solar cells (Hedin et al. 1995).
6) Poker Flat (PKF), Alaska, United States
Precipitation was sampled from December 1979 to September 1982 at the Poker Flat Research Range in the Tanana Valley of central Alaska (Table 1). This region lies within the Yukon taiga and supports small, scattered spruce trees. The closest major settlement is Fairbanks (population of ~30 000) approximately 70 km to the southeast. Most precipitation falls between May and September. Estimates of monthly precipitation amount during the period of record were obtained from a National Climate Data Center site located about 17 km from Poker Flat.
7) Lijiang (LIJ), China
Precipitation was sampled at two sites over different time periods (June 1987–December 1989 and January 1990–October 1993, respectively) in Yunnan Province, south-central China, about 30 km northwest of Lijiang (population of ~80 000) in clearings on the east face of Yulong (Jade Dragon) Mountain (Table 1). Temperate alpine coniferous forests cover about 90% of the surrounding region. Most precipitation falls between May and October in association with a large-scale monsoonal flow regime. Precipitation is typically generated in association with westerly flow over the sparsely populated Himalayan Mountains and Plateau. Average monthly temperatures range from 6° to 18°C. From 1987 through 1990, precipitation was sampled at an elevation of 3270 m. In 1990, the collector was relocated to a more accessible site at a lower elevation (2250 m). The two sites were about 10 km apart.
8) San Carlos (SCL), Venezuela
Precipitation was sampled from September 1980 to March 1981 in the tropical Amazonian rainforest adjacent to the Rio Negro River in southern Venezuela near the borders with Columbia and Brazil. Terrain is gently rolling, with hills rising up to 40 m above the lowland (100-m altitude). Rainfall is abundant throughout the year with a maximum from April to September. The collection site was in a clearing surrounded by tierra firme mixed forest on the clay oxisols located about 4 km east the town of San Carlos de Rio Negro (population of ~500) in Amaxonas Province. At the time of sampling, regional food production involved subsistence-level slash-and-burn agriculture on small plots augmented with fishing, hunting, and gathering.
9) Katherine (KAT), Australia
Precipitation was sampled from November 1980 to March 1994 at a research facility operated by the Commonwealth Scientific and Industrial Organization Division of Crops and Pastures near Katherine in north-central Australia about 275 km southwest of Darwin (Table 1). Over the 14-yr sampling period, the population of Katherine increased significantly to about 6000 by the end of the program. The site is surrounded by tropical savanna including scattered eucalyptus trees and numerous ant and termite mounds. Annual shifts in the monsoonal regime result in distinct wet (November–April) and dry (May–October) seasons. Approximately 70% of the area surrounding the site is burned during a typical dry season by fires started naturally (e.g., by lightning) and for agricultural purposes (Likens et al. 1987 and references therein).
10) Mauna Loa (MLO), Hawaii, United States
Precipitation was sampled from April 1983 to February 1994 at the NOAA observatory on Mauna Loa Mountain, about 6 km from the Mauna Loa crater (Table 1). The site consists of exposed rocky terrain with scattered low montane vegetation. The island of Hawaii has a resident population of about 140 000, most of whom reside near the coast. Downslope flow conditions at night typically envelope the site in unmodified free-tropospheric air (Vitousek 2004). Upslope flow conditions during the daytime transport marine boundary layer (MBL) air and associated moisture to the site. Precipitation occasionally falls as snow during winter. Although nominally classified as a “marine” site owing to the upslope transport, the site’s high elevation, distance from the coast, and enhanced dry deposition due to turbulent transport minimized influences of shorter-lived marine-derived constituents including primary marine aerosol.
To facilitate comparisons, data in the tables and figures are classified into four broad subgroups of sites based on proximity to the ocean and altitude: marine (MHG, MHT, ADG, HRT, BTT, BAT, and AMI), near coastal (CDP and TDP), continental (PKF, LIJ1, LIJ2, SCL, and KAT), and high elevation regularly influenced by both MBL and free-tropospheric air (MLO). Within each subgroup, sites are ordered by latitude from north to south.
b. Sample collection and processing
Procedures for sample collection, storage, and analysis are detailed in previous publications (e.g., see Galloway et al. 1982, 1993, 1996; Likens et al. 1987; Moody et al. 1991; Keene et al. 2014). Briefly, precipitation was sampled in polyethylene collectors and transferred to polyethylene bottles, all of which were precleaned at the University of Virginia (UVA) by washing with detergent, soaking overnight in dilute HCl, and copiously rinsing with deionized water. The conductivity of the final rinse from each container was analyzed to verify quality. At sites with reliable power either transmitted from remote power stations or produced locally with wind generators (CDP only), “wet only” precipitation was sampled on an event or daily basis using collectors that automatically opened at the beginning and closed at the end of each period of precipitation. At other sites, “bulk” precipitation was sampled in funnel/bottle assemblies (Table 2) that were manually deployed for a maximum of 24 h before the start of an event and recovered within a maximum of 24 h after the event ended, thereby minimizing contributions from dry deposition (Table 2). Immediately after recovery, 250-mL aliquots (or less for low-volume samples) were transferred to the precleaned bottles and sterilized with addition of 500 μL CHCl3 to prevent compositional changes resulting from microbial growth (e.g., Keene et al. 1983; Herlihy et al. 1987). At most sites, treated samples were temporarily stored under refrigeration (~4°C) and shipped periodically (typically weekly or monthly) to UVA. After receipt, samples were stored refrigerated prior to analysis.
Precipitation amount and numbers of samples represented in quality-assured (QAed) database.
Precipitation depth corresponding to each sample was quantified by sample volume and cross-sectional area of the collector (ADG, MHG), with standard bulk rain gauges to which a drop of mineral oil was added to minimize evaporation (LIJ1, MLO, PKF, SCL, HRT, BAT), with recording rain gauges (MHT), or in parallel with both bulk and recording gauges (AMI, KAT, LIJ2, TDP, CDP, BTT) (Table 2). When both bulk and recording gauges were deployed, precipitation amounts correspond to data for the bulk gauge, which were more reliable and accurate.
c. Sample analysis
With the exception of a subset of samples from HRT (N = 146, collected during the first 2 yr of record) that were analyzed at the Bermuda Biological Station for Research (now the Bermuda Institute for Ocean Sciences), all samples were analyzed at UVA. Over the duration of all programs, H+ was measured with a pH meter and electrode. For the GPCP, WATOX, and AEROCE, NH4+ was measured by automated colorimetry and base cations (Na+, K+, Ca2+, and Mg2+) by atomic absorption spectroscopy. For CPES, NH4+ and base cations were measured by ion chromatography (IC). During the first 1–2 yr of GPCP and WATOX, SO42−, NO3−, and Cl− were measured by automated colorimetry. Thereafter, these analytes were measured by IC. SO42−, NO3−, and Cl− in all AEROCE and CPES samples were analyzed by IC. For the GPCP and WATOX, analysis of total (undissociated plus ionized) formate (HCOOHt = HCOOHaq + HCOO−) and acetate (CH3COOHt = CH3COOHaq + CH3COO−) by ion exclusion chromatography was initiated for some sites in 1982 and expanded to all sites in 1983. For AEROCE and CPES, carboxylic species were measured by IC. Routine IC analysis of CH3SO3− in samples from marine sites (MHGT, BTT, BAT, AMI, TDP) was initiated in 1987. For all programs, analytical performance was routinely quantified via laboratory intercomparisons and analysis of audit solutions from the National Institute of Standards and Technology and World Meteorological Organization, among others. In addition, whenever new analytical techniques were adopted, the new and old methods were intercompared to ensure that results were comparable (slopes and intercepts for reduced major axis regressions of paired data that were statistically indistinguishable from 1.0 and 0.0, respectively, at p of 0.05). However, analytical resolution improved over the course of all programs as more sensitive measurement techniques became available. For example, the analytical precision for SO42− increased and the corresponding detection limit decreased by approximately a factor of 10 from the beginning to the end of the data record reported herein. After analysis, data were reduced, compiled, quality assured, and stored in an interactive data archive. For sites significantly influenced by marine sources (ADG, MHG, MHT, HRT, BTT, AMI, BAT, CDP, and TDP), sea-salt and nss contributions to measured SO42−, Cl−, K+, Mg2+, and Ca2+ were differentiated using Na+ as the reference species (Keene et al. 1986).
d. Data quality
An identical set of quality-control criteria was applied to all data. Only data for samples that had been characterized for all chemical constituents routinely measured at the time of analysis and for which complete descriptive information was available (e.g., precipitation amount, collector deployment, and recovery times, etc.) were retained in the quality-assured database.
Field and laboratory notes were examined for indications of gross contamination (large insects, bird droppings, plant debris, etc.), equipment malfunction, and deviation from specified sampling protocols. Data for samples that contained visible particulates were examined for additional evidence of significant contamination (e.g., unusually high concentrations of NH4+ and nss K+ relative to other samples are indicative of insects, bird droppings, and plant debris). In the absence of obvious contamination, data for these samples were retained in the database. Data for samples associated with collector malfunction (i.e., only a portion of an event was sampled) or obvious contamination due to mishandling during processing were excluded from the quality-assured database. Ion-balance relationships were then employed to characterize unmeasured ionic species and identify significant analytical errors (Table 3). Specifically, samples with anion deficits (∑Cations minus ∑Anions on an equivalent basis) greater than ±5 μeq L−1 of the median deficit for all samples and/or ion imbalances greater than ±15% of the median imbalance were flagged for further examination. Application of this filter is based on the assumption that median deficits and imbalances correspond to contributions from unmeasured ions. Flagged samples were reanalyzed and the data reexamined. When reanalysis yielded results consistent with the original set of analyses, the data were retained in the quality-assured database.
Range and median values for the equivalent ratios for the sums of measured anions vs cations and the corresponding absolute differences between the sums of anions minus cations (μeq L−1).
Although carboxylic acids contributed significantly to ionic composition of precipitation at KAT (Keene et al. 1983), these species were not measured in the first 75 samples collected at that site. Consequently, ion-balance relationships could not be reliably interpreted in terms of data quality. For this subset of samples, concentrations of carboxylic acids were estimated from the loss of H+ in unsterilized aliquots and then incorporated in ion-balance relationships, which were subsequently interpreted as described above [see Likens et al. (1987) for details].
e. Statistical evaluations
The significance of seasonal variability in deposition fluxes of precipitation water and associated dissolved constituents at each site was evaluated with the Kruskal–Wallis nonparametric alternative to the one-way ANOVA. This approach tested the null hypothesis that the distribution of monthly wet-deposition fluxes for a given analyte at a given location represented random samples from an identical population. If no precipitation fell during a given month, the corresponding wet-deposition fluxes for all analytes were set equal to 0.0. If precipitation fell during a month but no quality-assured chemical data were available, all results for that month were excluded from the statistical evaluation.
Interannual variability in deposition rates of precipitation water and associated VWA concentrations of dissolved constituents at each site were evaluated using the coefficient of variation (CV, the ratio of the standard deviation to the corresponding mean expressed as a percent). CV is a dimensionless measure of the normalized dispersion of a probability distribution. The relative importance of precipitation amount as a driver of interannual variability in constituent deposition was inferred based on these statistics.
Long-term trends in analyte concentrations were characterized using two complementary statistics: 1) slopes of standard linear regressions (SLRs) for monthly VWA concentrations versus time and 2) slopes calculated using the bootstrapping method applied to all individual concentrations versus time (Freedman 1981). Relative to SLRs, bootstrapping provides a more robust test of significance in trends. Significance was evaluated based on two-tailed tests of the null hypothesis that the regression slope was statistically indistinguishable from 0.0. For the bootstrap slope, the test was based on 1000 random samples of the corresponding data subset.
3. Results and discussion
Precipitation composition and associated wet-deposition fluxes vary spatially and temporally as functions of the size, duration, and frequency of precipitation events. For example, dilution of scavenged material associated with larger-volume events results in lower concentrations on average relative to lower-volume events (e.g., Galloway et al. 1989). Consequently, precipitation in a semiarid region with infrequent, low-volume precipitation events may exhibit higher aqueous concentrations but lower annual wet-deposition fluxes relative to wetter regions with larger, more frequent events. In addition, variability in the frequency of precipitation events influences the relative importance of contributions from wet versus dry deposition to total atmospheric deposition. Because all these physical factors vary spatially and, at many locations, seasonally, the corresponding composition of precipitation and wet fluxes also vary spatially and temporally.
Precipitation composition and wet-deposition fluxes also vary spatially and temporally in response to variability in and proximity to upwind sources for natural precursor species, rates at which precursors are oxidized to soluble reaction products, the associated phase partitioning of those products with size-resolved aerosols, corresponding lifetimes of precursors and reaction products against dry deposition, and meteorological conditions. For example, variability in temperature influences emissions of NH3 from vegetation and rates of precursor oxidation, wind fields modulate the exchange of particles and gases between Earth’s surface and the atmosphere as well as transport from source regions, cloud cover influences in-cloud chemical processes, and upwind removal via deposition within fetch regions influences distances that emitted species and reaction products are transported from sources. It is evident from the above that the “background” composition of precipitation in the absence of anthropogenic influences corresponds to a range of conditions that vary geographically and, at many locations, seasonally.
In addition, measurements of atmospheric constituents at remote locations often reveal discernible anthropogenic influences associated with emission sources hundreds to thousands of kilometers upwind (e.g., Clarke et al. 2013; Simpson et al. 2014). For example, detailed analyses of aerosol composition over the North Atlantic basin indicate that terrestrial sources account for large fractions of particulate nss SO42−, NO3−, and NH4+ at MGT, BTT, and BAT (e.g., Savoie et al. 1989, 2002; Moody et al. 2014). The stable isotopic compositions of particulate S (e.g., Turekian et al. 2001; Lin et al. 2012) and C (Turekian et al. 2003) over the North Atlantic Ocean also reveal that continental sources typically dominate. Trajectory analyses show that combustion over southern Africa contributes minor but detectable amounts of NO3− in precipitation at AMI (Moody et al. 1991). Obviously, local and regional emissions from anthropogenic sources can also be important. Emissions from widespread biomass burning associated with agriculture (in addition to those from natural wild fires) in the Northern Territory of Australia contribute to the relatively high concentrations of soluble species, particularly HCOOHt, CH3COOHt, and associated H+, in precipitation at KAT at the beginning of the wet season (Likens et al. 1987; Keene et al. 2006; Akagi et al. 2012). Emissions of S and N from ships at sea are also important pollutant sources downwind of major shipping lanes (Corbett et al. 1999; Lawrence and Crutzen 1999; von Glasow et al. 2003). Consequently, relative anthropogenic contributions to precipitation composition in remote regions vary spatially and, thus, aqueous concentrations and deposition fluxes reported herein are considered upper limits for precipitation composition in the absence of anthropogenic influences.
a. Spatial variability
1) Precipitation amount
Average annual precipitation amounts varied from 29 cm yr−1 at PKF to 562 cm yr−1 at CDP (Table 1). Compared to other sampling locations, the high-latitude (PKF and TDP) and high-altitude (MLO) sites receive relatively lower rates of annual precipitation whereas the rainforest sites (SCL and CDP) receive relatively higher rates. Average annual precipitation at the other sites fell within the relatively narrow range from 97 to 152 cm yr−1.
2) pH
Annual volume-weighted-average (VWA) H+ expressed as pH for precipitation at all sites ranged from 4.69 to 5.25 (median of 4.93; Table 4). Interestingly, when data for sites in the relatively polluted North Atlantic region (MHGT, ADG, BTT, HRT, and BAT) and for MLO, which was impacted intermittently by H2SO4 from local volcanic emissions of SO2, are excluded, annual VWA H+ expressed as pH for the other sites fell within an identical range and the median pH increased by only 0.05 units to 4.98. Partitioning this latter subset of sites into continental versus marine-influenced yields ranges from 4.69 to 5.03 (median value of 4.95) for continental sites and from 4.96 to 5.25 (median value of 5.25) for marine-influenced sites. As reported previously, acidities of precipitation in remote continental sites tend to be greater than those at remote marine sites (Keene and Galloway 1988). In comparison, during the period 1965–1995, median pHs in or downwind of industrialized and populated regions in the eastern United States and elsewhere typically ranged from the high threes to mid fours (e.g., Likens 2013). However, as described in more detail below, carboxylic acids are major sources of free acidity in precipitation in remote regions. Because the range in precipitation pH overlaps the pKa values of HCOOH and CH3COOH (3.75 and 4.76, respectively), these organic acids buffer changes in H+ resulting from contributions of mineral acids from anthropogenic sources (e.g., Keene and Galloway 1988).
Overall VWA concentrations based on all data and corresponding annual-average wet-deposition fluxes of chemical species. Mg2+, K+, Ca2+, Cl−, and SO42− concentrations and depositions at marine-influenced sites (italics) correspond to nss fractions.
3) Sulfur
VWA concentrations of nss SO42− at marine-influenced sites ranged from 1.2 to 18.8 μeq L−1 (Table 4) and generally decreased with latitude from north to south (Fig. 2). In the absence of SO2 and particulate SO42− transported from anthropogenic sources upwind, the production of (CH3)2S [dimethyl sulfide (DMS)] by marine phytoplankton in the surface ocean and its subsequent emission to and oxidation by various pathways within the overlying atmosphere is the dominant source of nss S in marine precipitation. The primary terminal products of (CH3)2S oxidation are CH3SO3H [methane sulfonic acid (MSA)] and H2SO4, which are efficiently incorporated into aerosols and cloud droplets and subsequently deposited to the surface via wet and dry deposition (Liss 1999).
(left) Mean annual VWA concentrations and (right) mean annual wet-deposition fluxes of (a),(b) SO42−; (c),(d) CH3SO3−; (e),(f) NO3−; (g),(h) NH4+; (i),(j) total inorganic N (TIN = NO3− + NH4+); (k),(l) HCOOHt; and (m),(n) CH3COOHt. For marine-influenced sites (designed in Table 4), values in (a) and (b) correspond to nss SO42− and, for other sites, they correspond to total SO42−. Vertical bars depict ±1 std dev. When available, corresponding average values for the mid-Atlantic region of the United States during 2002–04 (NADP 2014) are depicted as dashed horizontal lines. Data for SCL correspond to less than one year. HCOOHt and CH3COOHt were measured at PKF during only the final year of operation.
Citation: Journal of the Atmospheric Sciences 72, 8; 10.1175/JAS-D-14-0378.1
VWA concentrations of CH3SO3− at marine-influenced sites ranged from 0.059 to 0.194 μeq L−1 and were relatively higher at higher-latitude sites (Table 4, Fig. 2). On a global basis, about 5% of S emitted in association with (CH3)2S is oxidized to CH3SO3− and the balance to nss SO42− (Rodhe 1999). CH3SO3− accounted for 0.7%, 1.1%, 1.5%, 4.3%, 9.1%, and 13% of the total nss S mass in precipitation at MHGT, BTT, BAT, AMI, CDP, and TDP, respectively. As discussed in more detail below, anthropogenic sources for nss SO42− in the North Atlantic region (MHGT, BTT, and BAT) contributed to the lower percent contributions of CH3SO3− relative to the other more remote sites in the Southern Hemisphere. In addition, production ratios for CH3SO3− versus nss SO42− from (CH3)2S oxidation increase at higher latitudes in part due to a temperature dependence in the branching ratio (Bates et al. 1992; Galloway et al. 1996; Davis et al. 1998; Jourdain and Legrand 2001; Savoie et al. 2002), which contributes to the relatively greater contributions of CH3SO3− to total nss S at the higher-latitude sites in the Southern Hemisphere. With the exception of CDP, average annual deposition fluxes of CH3SO3− at marine-influenced sites fell within the range from 0.7 to 1.7 Eq ha−1 yr−1. The substantially higher deposition rate at CDP (10.1 Eq ha−1 yr−1) was due in part to the relatively higher annual precipitation amount (562 cm yr−1) (Table 1) coupled with relatively greater yields of CH3SO3− from the oxidation of (CH3)2S emitted from the upwind ocean.
During the period 1988–1990, mean mass ratio yields from (CH3)2S oxidation to particulate nss SO42− versus CH3SO3− measured in onshore flow from open-ocean sectors at BAT (19.6), BTT (18.8), and MHGT (3.01) indicate that marine biogenic sources accounted for approximately 50% of total particulate nss SO42− at BAT, 30% at BTT, and 10%–15% at MHGT; the balance originated from anthropogenic sources (Savoie et al. 2002). Assuming that the same yields apply to nss SO42− in precipitation at these sites, marine biogenic sources contributed 1.8 (31%), 1.6 (20%), and 0.4 μeq L−1 (2%) of VWA nss SO42− at BAT, BTT, and MHGT, respectively. The lower percentage contributions of biogenic to total nss SO42− in precipitation relative to aerosols suggest that the air columns scavenged by precipitation at these sites were impacted by anthropogenic sources to a greater degree than near-surface aerosols. We note that aerosol sampling was sector controlled to minimize contributions from local sources whereas precipitation sampling was not. Anthropogenic S scavenged from out-of-sector air parcels may have contributed to these differences, particularly at Mace Head, which exhibited the greatest divergence and is in relatively closer proximity to major combustion sources. However, polluted easterly flow at Mace Head is generally associated with high pressure systems that produce little rain, thereby minimizing potential contributions from out-of-sector precipitation. More efficient transport of pollutant S from continents in association with frontal passages that produce rain may also contribute to relatively greater anthropogenic S in precipitation. Alternatively, more efficient aqueous-phase oxidation of SO2 (both natural and anthropogenic) in association with precipitation events may have contributed to these differences, in which case, the yield ratios for aerosols may underestimate biogenic contributions to nss SO42− in precipitation. Finally, much (CH3)2S emitted from the ocean surface is thought to be oxidized within the MBL (Savoie et al. 2002), although some fraction is detrained to and oxidized within the free troposphere (e.g., Long et al. 2014). In contrast, polluted air may be advected from continents over the ocean above the MBL where it has a longer lifetime against deposition (e.g., Neuman et al. 2006). This differential atmospheric processing of anthropogenic and biogenic S may contribute to relatively higher concentrations of anthropogenic S in the column scavenged by precipitation relative to aerosols in near-surface air (Keene et al. 2014). Causes for these differences cannot be resolved unequivocally. However, assuming that the yield ratios reported by Savoie et al. (2002) are representative of those for precipitation at the North Atlantic sampling sites and that anthropogenic contributions to nss SO42− in precipitation at the other marine-influenced sampling sites are negligible, our results suggest that contributions of marine biogenic sources to annual VWA nss SO42− in marine precipitation range from approximately 0.4 to 3.3 μeq L−1 (corresponding to those at MHGT and AMI, respectively).
VWA concentrations of SO42− at the continental sites (PKF, LIJ1, LIJ2, SCL, and KAT) ranged from 2.9 to 7.7 μeq L−1 and were generally higher and more variable than the VWAs attributed to marine biogenic sources at marine sites. All are probably influenced to some extent by anthropogenic emission products including long-distance transport from Asia at PKF (Dayan et al. 1985; Koch et al. 2007; Yu et al. 2012), transport under infrequent easterly flow from heavily populated regions of China to LIJ1 and LIJ2 (Keene et al. 1989), and biomass burning in the regions surrounding KAT (Likens et al. 1987) and SCL (Holben et al. 1996; Artaxo et al. 2002). As indicated above, the relatively high VWA SO42− at MLO reflects contributions from local volcanic emissions and is not representative of precipitation composition at high-elevation locations remote from such local influences.
In addition to variable influences of upwind emissions, variability in annual precipitation amount and the associated frequency of precipitation events contribute to spatial variability in average annual wet-deposition fluxes of nss SO42− and other species among sites. For example, VWA concentrations of nss SO42− at the three most remote marine-influenced sites (AMI, CDP, and TDP) ranged from 1.2 to 3.3 μeq L−1 (factor of 2.9), whereas the corresponding annual wet-deposition fluxes varied from 8.9 to 101 Eq ha−1 yr−1 (factor of 11; Table 4). Similarly, VWA concentrations at the five continental sites (PKF, SCL, LIJ1 and LIJ2, and KAT) varied by a factor of 3.1, whereas the corresponding annual wet-deposition fluxes varied by a factor of 11 (Table 4). It is evident that the major biological and physical controls on VWA nss SO42− in precipitation and corresponding wet-deposition fluxes in remote regions vary spatially. In addition, as noted above, anthropogenic emissions impact precipitation composition to finite degrees at all of these sites. Consequently, caution is warranted when interpreting the magnitudes of anthropogenic impacts in polluted regions based on direct comparisons with these data. We return to this issue below.
4) Nitrogen
The investigation reported herein focuses on NO3− and NH4+, the major soluble inorganic nitrogen species in the troposphere. However, organic N-containing species including organic nitrates (e.g., hydroxyalkyl nitrates) produced by reactions involving hydrocarbons and NO (Shepson et al. 1996; Munger et al. 1998); reduced species such as urea (e.g., Cornell et al. 1998, 2001), amines, and amino acids (e.g., Gorzelska et al. 1992; Milne and Zika 1993); and primary biological material including bacteria, fungi, and pollen (e.g., Herlihy et al. 1987) are also scavenged by precipitation and contribute to wet-deposition fluxes of N. Although their importance relative to inorganic N compounds is considerably more uncertain (e.g., Keene et al. 2002), available data indicate that measurements of N deposition fluxes based exclusively on inorganic species correspond to lower limits. For example, Neff et al. (2002) estimate that the total deposition of organic N globally is in the range from 10 to 50 Tg N yr−1 and Duce et al. (2008) estimate that organic N accounts for about 20 Tg N yr−1 or about 30% of the total N deposited to the global oceans. Model calculations suggest that approximately 16 Tg N yr−1 of soluble organic N is deposited to the global ocean, of which about 45% originates from anthropogenic sources (Kanakidou et al. 2012). A recent review by Jickells et al. (2013) reports that organic N accounts for about 25% of the total N deposition flux globally. In more polluted regions such as eastern North America, contributions of organic N to total N deposition are relatively lower (e.g., Keene et al. 2002; Likens 2013).
NOx (NO + NO2) emitted from soils, emitted during biomass burning, and produced by lightning is the major precursor for NO3− in precipitation over remote regions. There are no known oceanic sources for NO3− precursors (e.g., Dahl et al. 2005).
Like nss SO42−, VWA concentrations and wet-deposition fluxes of NO3− generally decreased from north to south across the North Atlantic basin (Fig. 2), reflecting variability in relative contributions from surrounding continents (e.g., Galloway et al. 1989; Savoie et al. 1989, 1992, 2002; Neuman et al. 2006; Moody et al. 2014). Mean annual VWA concentrations of NO3− at the three marine-influenced sites in the Southern Hemisphere (AMI, TDP, and CDP) ranged from 0.5 to 1.3 μeq L−1 and were lower by factors of 2–14 relative to those at the North Atlantic sites (Table 4, Fig. 2). VWA NO3− at remote continental sites were somewhat higher than those at marine-influenced sites in the Southern Hemisphere ranging from 1.4 μeq L−1 at LIJ1 to 4.8 μeq L−1 at KAT. The relatively high concentrations at KAT were driven in part by combustion products from biomass burning in association with regional agriculture (Likens et al. 1987).
Variability in annual precipitation amount contributed to corresponding variability in average annual wet-deposition fluxes. The highest deposition fluxes outside the relatively more polluted North Atlantic region and the biomass-burning impacted site at KAT were at the two rain forest sites (CDP and SCL), which experienced the greatest water deposition fluxes, whereas the lowest wet-deposition fluxes were at the two highest-latitude sites (TDP and PKF) and the high-elevation site (MLO), which experienced relatively lower water deposition fluxes (Tables 1 and 4, Fig. 2).
In continental regions, NH3 is produced from the degradation of organic matter in soils and surface waters and subsequently volatilized to the atmosphere. It is also emitted to the atmosphere from supersaturated stomatal linings of plants (e.g., Langford et al. 1992) and from biomass burning (Andreae and Merlet 2001; Keene et al. 2006). Globally, NH3 emissions from natural terrestrial sources are about 6 Tg N yr−1 (Galloway et al. 2004). Model calculations indicate substantial spatial variability in NH3 emissions from the ocean to the atmosphere that total about 8.2 Tg N yr−1 globally (Bouwman et al. 1997; Ciais et al. 2014).
The high mean VWA concentration and annual wet-deposition flux of NH4+ at ADG relative to other sites probably reflects contributions from local agricultural emissions and, consequently, these data are not considered to be regionally representative. Chemical climatologies as a function of transport regime (Keene et al. 2014; Moody et al. 2014) indicate that NH3 emissions over surrounding continents contribute significantly to the higher NH4+ concentrations and wet-deposition fluxes associated with North Atlantic precipitation relative to those at the more remote marine-influenced sites (Table 4, Fig. 2). In contrast, based on isotopic analyses and associated model calculations, Altieri et al. (2014) recently reported that most NH4+ in precipitation at Bermuda originates from marine sources. However, it is difficult to reconcile the dominance of a marine source with key observations reported in the literature, including 1) the substantial emissions of NH3 (as well as SO2 and NOx) over the United States (EPA 2014); 2) the high concentrations of NH3 and particulate NH4+ (as well as SO2, particulate nss SO42−, HNO3, and particulate NO3−) in near-surface air over eastern North America and the association of most NH4+ (and nss SO42−) with submicron aerosol size fractions that have long atmospheric lifetimes against deposition (from several days to a week or more) (e.g., Fischer et al. 2006; Smith et al. 2007); 3) the high frequency of airmass transport from North America to Bermuda (Moody et al. 2014); 4) the significantly higher concentrations of NH4+ in precipitation (factor of 2.5 based on VWA concentrations) and aerosols (factors of 3–4 based on median values) associated with flow from North America to Bermuda relative to open-ocean flow (nss SO42− and NO3− are also significantly higher) (Keene et al. 2014; Moody et al. 2014); and 5) the generally higher wet-deposition fluxes of NH4+ (as well as nss SO42− and NO3−) at Bermuda relative to more remote marine locations (Fig. 2h). The bulk of available evidence supports the hypothesis that much of the NH4+ in precipitation at Bermuda originates from terrestrial sources in North America.
Although AMI supports a large sea-bird rookery and local seal population and exhibits a VWA NH4+ concentration that is 4–5 times higher than that at TDP or CDP, available evidence suggests that NH3 emissions from these local sources did not contribute significantly to NH4+ concentrations in precipitation at that site (Moody et al. 1991). The marked seasonality in per-event depositions supports the hypothesis that most NH4+ in precipitation at AMI originates from regional NH3 emissions from the surrounding ocean (Moody et al. 1991).
VWA NH4+ concentration at the continental sites varied over the relatively narrow range from 2.3 to 4.2 μeq L−1. Although NH3 emissions from anthropogenic sources (particularly from biomass burning and agricultural activities near KAT and agricultural activities upwind of LIJ1 and LIJ2) undoubtedly contributed to NH4+ in precipitation at some sites, the relatively narrow range in VWA concentrations across these sites suggests that natural sources probably dominated. The low VWA concentration at MLO implies that biogenic emissions at cooler, higher-elevation locations are relatively low compared to warmer, lower-elevation locations. Like NO3−, annual wet-deposition fluxes were driven in part by precipitation amount. The rain forest sites (CDP and SCL) that received the highest water deposition exhibited disproportionately greater wet fluxes of NH4+ relative to concentration whereas the high-latitude and high-altitude sites (TDP, PKF, and MLO) that received the lowest water fluxes exhibited relatively lower wet fluxes of NH4+.
5) Carboxylic acids
HCOOH and CH3COOH are emitted directly to the atmosphere by fossil-fuel combustion (Kawamura et al. 1985), biomass burning (Andreae and Merlet 2001; Keene et al. 2006), plants (Andreae et al. 1988; Gabriel et al. 1999), soils (Sanhueza and Andreae 1991), and formacine ants (Graedel and Eisner 1988). These acids are also produced in the atmosphere from the oxidation of precursor hydrocarbons emitted by terrestrial plants and from the ocean surface (e.g., Puxbaum et al. 1988; Granby 1997; Neeb et al. 1997; Paulot et al. 2011). In polluted regions, the oxidation of anthropogenic hydrocarbons produces significant HCOOH and CH3COOH (e.g., de Gouw et al. 2005). The photochemical degradation of particulate organic matter emitted to the atmosphere in association with primary marine aerosol (Zhou et al. 2008) and biomass burning (Paulot et al. 2011) may also produce carboxylic acids. Comparisons between concentrations and deposition fluxes at rural sites within industrialized regions and remote sites suggest that biogenic sources dominate on the global scale (Keene and Galloway 1988). This interpretation is consistent with carbon isotopic analyses, which indicate that greater than 80% of HCOOH and CH3COOH over Europe originates from biogenic sources (Glasius et al. 2000, 2001). The strong correlations between concentrations and wet-deposition fluxes of HCOOH and CH3COOH in precipitation at most locations (Keene and Galloway 1986) imply that similar precursors dominate their production. These species are major sources of acidity in remote regions and thereby regulate important pH-dependent chemical pathways (Keene et al. 1983; Millet 2012).
Compared to nss S and N species in marine-influenced regions, the annual average wet-deposition fluxes of HCOOHt and CH3COOHt fall within relatively narrow ranges (factors of 2.3 and 3.9, respectively; Table 4, Fig. 2), suggesting a fairly uniform global distribution of marine sources. Although the ranges overlap those in marine regions, spatial variability in wet-deposition fluxes of HCOOHt and CH3COOHt among continental sites was substantially greater (factors of 11 and 19, respectively). Biomass burning in the region surrounding KAT contributes to the relatively high VWA concentrations and fluxes of both species compared to other sites (Table 4, Fig. 2). Relatively lower concentrations and fluxes at the high-altitude site (MLO) and the high-latitude continental site (PKF) (Table 4, Fig. 2) suggest relatively lower emissions of precursor biogenic hydrocarbons compared to other regions.
Relative to most other chemical constituents of precipitation, carboxylic acids exhibit proportionately little spatial variability in concentrations and wet-deposition fluxes over the broad geographic regions (Table 4, Fig. 2), suggesting fairly uniform source strengths globally. Given the greater expanse of ocean relative to land and the associated deposition fluxes in marine regions, these data imply that marine sources for these acids are significant and perhaps dominant globally. This interpretation is difficult to reconcile with an analysis based on satellite imagery and inversion model calculations, which concluded that, globally, approximately 90% of HCOOH originates from biogenic sources that are dominated by precursor hydrocarbons emitted from tropical and boreal forests (Stavrakou et al. 2012). However, Stavrakou et al. (2012) did not explicitly consider or evaluate HCOOH production via oxidation of precursor compounds emitted from the ocean surface and, thus, their estimated total production fluxes globally (100–120 Tg HCOOH yr−1 or 26–32 Tg C yr−1) should be considered lower limits and their suggestion that terrestrial sources dominate questionable.
Estimated emission fluxes of reactive VOC precursors from the surface ocean are lower than those from terrestrial vegetation and also insufficient to sustain mixing ratios of short-lived oxidation products such as glyoxal measured in remote marine air uninfluenced by continental emissions (Sinreich et al. 2010). These results suggest the presence of a large as yet uncharacterized source for reactive organic precursors over the remote oceans. Recent estimates suggest that the global production flux of organic matter in association with primary marine aerosol (20–50 Tg C yr−1) (Roelofs 2008; Gantt et al. 2009; Albert et al. 2010; Long et al. 2011) is substantially greater than previously thought and may exceed the total global production flux of particulate organic matter from secondary pathways involving gas-phase (primarily anthropogenic and terrestrial biogenic) precursors [~18 Tg C y−1 (Andreae and Rosenfeld 2008)]. Most (~70%–80%) of this organic matter is associated with submicron-diameter-size fractions (Keene et al. 2007; Facchini et al. 2008) with average atmospheric lifetimes against deposition of a week or more. When exposed to light and reactive trace gases, this freshly emitted particulate organic matter reacts to produce OH, H2O2, and, presumably, other low-molecular-weight products including alcohols, carbonyls, and carboxylic species (Zhou et al. 2008). We hypothesize that the photochemical degradation of primary, marine-derived organic matter is an important and previously unrecognized source for carboxylic acids in marine regions that contributes to sustaining the concentrations and wet-deposition fluxes of HCOOHt and CH3COOHt measured over remote oceanic regions (Table 4, Fig. 2).
6) Chlorine
The production of primary marine aerosol by wind stress at the ocean surface dominates the emission flux of inorganic Cl to the atmosphere on a global scale; other sources (including mineral aerosol, biomass burning, and fossil-fuel and waste combustion) are relatively unimportant globally but may dominate emissions in continental regions (Graedel and Keene 1995; Keene et al. 1999). Acid-displacement, free-radical, and other multiphase chemical transformations in the atmosphere produce inorganic chlorinated products that volatilize to the gas phase including HCl, ClNO2, BrCl, and Cl2 (Keene et al. 2009; Lawler et al. 2009, 2011; Long et al. 2014). Because 1) the dry-deposition fluxes of particulate and volatile Cl species are not identical and 2) some of the volatile species have relatively low solubilities, the total amounts of particulate and volatile Cl scavenged and deposited by precipitation in marine regions may diverge from sea-salt ratios relative to conservative tracers independent of contributions from nss sources. Consequently, calculated values for nss Cl− in marine precipitation cannot be reliably interpreted in terms of sources. Available evidence suggests that VWA concentrations and deposition fluxes of Cl− in remote continental (PKF, LIJ1, LIJ2, SCL, KAT) and high-elevation (MLO) regions (Table 5) originate primarily from the scavenging of crustal dust and biomass-burning products. The inland transport and subsequent scavenging of sea-salt aerosol and chlorine-containing reaction products may contribute significantly to inorganic Cl at continental locations downwind from oceans (e.g., Stallard and Edmond 1981; Young et al. 2013; Jordan et al. 2015). However, the low VWA concentrations of Na+ at continental and high-elevation sites (0.01–5.0 μeq L−1) relative to marine-influenced sites (13–1530 μeq L−1; Table 4) indicate that marine contributions decrease rapidly with distance from coasts, consistent with wet-deposition fields for Na+ over the United States reported by NADP (2014).
p values for Kruskal–Wallis test of seasonal differences in wet-deposition fluxes. Results for SO42−, K+, and Ca2+ at marine-influenced sites (italics) correspond to nss fractions.
7) Base cations
Base cations (Na+, Ca2+, Mg2+, and K+) are injected into the remote atmosphere in association with primary marine aerosols, mineral dust, and biomass-burning emissions. Base cations in wet deposition are important nutrients for terrestrial ecosystems (e.g., Likens et al. 1977; Swap et al. 1992; Likens 2013) and also serve as tracers for distinct source types (e.g., Na+ and Mg2+ for sea salt, nss K+ for biomass burning, nss Ca2+ for aolian dust). With the exception of those downwind of semiarid continental regions (e.g., BAT), concentrations and deposition fluxes of nss base cations at marine sites are generally low (Table 4). For sites such as MHGT, which experiences high wind velocities and associated sea-salt concentrations, calculated nss components are associated with relatively large accumulated uncertainties (e.g., Keene et al. 1986). High winds also increase the potential for contamination by locally generated dust. Consequently, nss concentrations of base cations at such sites should be interpreted with caution. Concentrations and wet-deposition fluxes of base cations at continental sites (KAT, LIJ1, LIJ2, and PKF) are generally greater than those of nss fractions at marine-influenced sites (Table 4). The higher fluxes at continental sites reflect regional sources for base cations associated with mineral dust and biomass burning.
Primary marine aerosol is a major reaction medium over the world’s oceans through which important climate-relevant species, including nss S, NOy, and halogen radicals, are processed (Long et al. 2014). In addition to scattering incident solar radiation, primary marine aerosols are effectively activated into cloud droplets, thereby influencing the microphysical properties of clouds (Bates et al. 2012; Quinn et al. 2014). Although emission fluxes are uncertain, model calculations based on a range in size-resolved production parameterizations indicate that, relative to a base run with no primary marine aerosol, CCN number concentrations increase by 150%–500% when contributions from this source are included (Pierce and Adams 2006). Consequently, primary marine aerosol plays a central role in regulating Earth’s radiative balance (e.g., Quinn and Bates 2011). Sea salt is also an important source of atmospheric alkalinity, which influences pH-dependent chemical transformations in marine air and associated deposition fluxes (e.g., Chameides and Stelson 1992; Keene et al. 1998; Russell et al. 2003; and references therein).
Because sea-salt production over the open ocean is a power function of wind velocity (e.g., Gong et al. 1997; Long et al. 2011), atmospheric concentrations and fluxes of conservative sea-salt tracers such as Na+ reflect global wind fields. Tropical and subtropical marine regions (BAT, BTT, HRT) exhibit lower concentrations relative to windier, higher-latitude, temperate regions (MHGT, ADG, AMI, CDP) (Table 4). Locally produced marine aerosols from breaking waves along the coast contribute to the relatively higher VWA concentrations and fluxes of Na+ at MHGT (Kunz et al. 2002). Relatively lower VWA concentrations at CDP is driven in part by dilution associated with the substantially larger annual precipitation amount (562 cm yr−1). The corresponding average annual wet-deposition flux of Na+ is similar to that at AMI. The relatively low VWA concentrations and wet-deposition fluxes of Na+ at TDP result from removal via wet and dry deposition during transport across the Andes Mountains that lie between the Pacific coast and the site.
b. Temporal variability
Temporal variability in precipitation composition and wet-deposition fluxes at a given location are driven by temporal variability in upwind emissions of precursor gases and aerosols from natural and anthropogenic sources and by meteorological processes that modulate emissions from many natural sources as well as the transport, chemical evolution, and deposition of precursor and condensable product species. These processes include large-scale atmospheric motion coupled with near-surface wind speed, temperature, solar radiation, cloud cover, and precipitation fields in fetch regions. All of these drivers vary over seasonal, interannual, multiyear, and climatological time scales.
1) Seasonal variability
Mean monthly precipitation amounts vary significantly over the year at BAT, AMI, LIJ1, LIJ2, and KAT, whereas those at most other sampling sites were more uniformly distributed (Table 5, Figs. 3a,b). Seasonal shifts in large-scale atmospheric circulation associated with the strength and location of the southern subtropical anticyclone over the Indian Ocean and the Bermuda high over the North Atlantic Ocean drive seasonal variability in precipitation amount at AMI (Moody et al. 1991) and BAT, respectively. Most precipitation at LIJ and KAT falls in association with seasonal monsoonal patterns that peak during the boreal and austral summers, respectively (Likens et al. 1987; Keene et al. 1989). The strong seasonal cycle in precipitation amount at KAT (Fig. 3) coupled with the relatively long period of record (Table 1) resulted in significant seasonal variability in monthly wet-deposition fluxes of all associated chemical constituents (Table 5). The lack of significance in monthly precipitation amount at LIJ1 is related in part to the relatively short duration of record for that site (Table 1). In addition, although minor to trace amounts of precipitation fell outside the monsoonal period at both LIJ1 and LIJ2, quality-assured chemical data were available for only May–October and June–October, respectively. The fact that months with measureable precipitation but no chemical data were excluded from the statistical evaluation contributed to the lack of significance in monthly wet fluxes of chemical species at the two LIJ sites.
Average monthly precipitation amounts at (a) marine sites and (b) near-coastal, continental, and high-elevation sites and corresponding monthly VWA concentrations of (c),(d) nss SO42−; (e),(f) CH3SO3−; (g),(h) NO3−; (i),(j) NH4+; (k),(l) Na+; (m),(n) HCOOHt; and (o),(p) CH3COOHt. Concentrations for sites designated with an asterisk in the legends are scaled to the second y axis and others are scaled to the y axis.
Citation: Journal of the Atmospheric Sciences 72, 8; 10.1175/JAS-D-14-0378.1
Seasonal variability in wind velocity and associated production of primary marine aerosols (Lewis and Schwartz 2004) drove significant variability in the corresponding monthly wet-deposition fluxes of Na+ at most marine-influenced sites (Table 5, Fig. 3). Significant seasonal variability in nss K+ at BTT and BAT and of nss Ca2+, NH4+, and NO3− at BAT (Table 5, Fig. 3) correspond to seasonal variability in the long-distance transport of crustal aerosol (nss K+ and nss Ca2+) and combustion products (nss K+, NH4+, and NO3−) to these sites from northern Africa and Europe (Savoie et al. 1989, 1992; Moody et al. 2014). Available evidence suggests that the significant seasonal pattern in monthly wet-deposition fluxes of NH4+ at AMI (Table 5, Fig. 3) was driven primarily by the seasonality in NH3 emissions from the surrounding ocean (Moody et al. 1991). The long-distance transport of NH3 (and associated particulate NH4+) emitted from terrestrial sources in southern Africa, particularly biomass burning which peaks during austral spring, may have contributed to relatively higher wet-deposition fluxes during that period (Fig. 3). The lack of known marine sources for NO3− coupled with detailed analysis of individual events based on air-mass history indicate that variability in monthly wet-deposition fluxes of NO3− at AMI were associated primarily with variability in the efficiency at which products from terrestrial sources over southern Africa (combustion, soils) and atmospheric sources (lightning) were transported to the site (Moody et al. 1991). The significant seasonality in monthly wet-deposition fluxes of CH3SO3− at two high-latitude marine sites (MHGT and AMI) was driven by seasonal variability in the production of precursor (CH3)2S by biota in the surface ocean and its subsequent emission to and oxidation in the atmosphere (Galloway and Gaudry 1984; Savoie et al. 2002).
2) Interannual variability
Although the relationships are nonlinear, on average, ionic strengths of higher-volume precipitation events are lower than those for lower-volume events (Galloway et al. 1989). In addition, as discussed above, greater precipitation amounts are generally associated with relatively greater wet-deposition fluxes. Consequently, CVs for interannual variability provide qualitative insight regarding the relative importance of precipitation amount (Fig. 4) as a driver of variability in annual VWA composition of precipitation and corresponding wet-deposition fluxes. If temporal distributions of precipitation events and the associated pools of atmospheric species that are scavenged by precipitation at a given location are similar from year to year, precipitation amount and the VWA concentrations and wet-deposition fluxes of constituents would exhibit similar CVs that are driven primarily by interannual variability in precipitation amount (Fig. 4). Alternatively, CVs for VWA composition and wet-deposition fluxes that are substantially greater than those for precipitation amount suggest that variability in source strengths contribute to the overall interannual variability in composition and deposition.
Annual precipitation amounts at (a) ADG (black bars) and MHGT (gray bars), (b) HRT (black bars) and BTT (gray bars), (c) BAT, (d) AMI, (e) TDP, (f) LIJ1 (black bars) and LIJ2 (gray bars), (g) KAT, and (h) MLO.
Citation: Journal of the Atmospheric Sciences 72, 8; 10.1175/JAS-D-14-0378.1
Table 6 lists CVs for the nine sites with periods of record greater than or equal to 6 yr. Excluding the high-elevation site at MLO, at which precipitation amount varies greatly from year to year (Fig. 4), CVs for precipitation amount at other sites range from 8.1 at ADG to 24.1 at MHGT. Curiously, the highest and lowest CVs correspond to the two Irish sites, which are in relatively close proximity. These results reflect the large spatial gradient in interannual variability in precipitation amount along the Irish coast (e.g., Rohan 1986). CVs for annual precipitation amount at the other sites vary by less than a factor of 2.
CVs for interannual variability in precipitation amount, VWA concentration, and wet-deposition flux (%). Results for SO42− at marine-influenced sites (italics) correspond to nss fractions.
Comparison of CVs for precipitation amount versus VWA concentration and wet-deposition fluxes of nss SO42− suggests that precipitation amount is the primary driver of interannual variability at ADG, BTT, BAT, and MLO. Precipitation amount and source strength account for roughly comparable fractions of interannual variability at HRT, TDP, and KAT, and source strength is the dominant cause of interannual variability at MHGT and AMI. Differences in results for the two sampling sites on Bermuda (BTT and HRT) were driven in part by the decreasing contributions of anthropogenic sources to nss SO42− in precipitation over the western North Atlantic Ocean during the period of record (discussed in more detail below). Comparison of CVs for CH3SO3− suggests that precipitation amount is the dominant source of interannual variability at MHGT, BTT, and TDP, whereas source strength is a relatively more important source of variability at BAT and AMI. Taken together, the above results imply that contributions of nss SO42− from anthropogenic sources at MHGT and from biogenic sources at AMI contribute significantly to the corresponding interannual variability in annual VWA concentrations and deposition fluxes at these sites.
Interannual variability in VWA concentrations and deposition fluxes of NO3− at MHGT, HRT, BTT, BAT, TDP, and MLO were driven primarily by precipitation amount, whereas those at ADG, AMI, and KAT were driven primarily by variability in source strength. Corresponding relationships for NH4+ were similar to those for NO3− at some sites but differed at others. Interannual variability in VWA concentrations and deposition fluxes of NH4+ at MHGT, BTT, BAT, and MLO were driven primarily by precipitation amount, whereas those at ADG, HRT, AMI, TDP, and KAT were driven primarily by variability source strength.
Relationships among CVs for carboxylic acids differed in some respects from those for the other species. For example, at several sites, CVs for interannual variability in wet-deposition fluxes of HCOOHt (HRT, BAT, AMI, KAT, and MLO) and CH3COOHt (MHGT, KAT, and MLO) were greater by factors of 2 or more relative to those for the corresponding VWA concentrations. In contrast, with one exception (NH4+ at BAT), CVs for interannual variability in VWA concentrations and corresponding wet-deposition fluxes for all other species at all sites agreed within less than a factor of 2. These results indicate that VWA concentrations versus deposition fluxes of carboxylic acids vary to a relatively greater degree than do those for other species; causes for such divergence are not known. More generally, CVs for annual variability in wet-deposition fluxes were often much greater than those for precipitation amount, suggesting substantially greater interannual variability in source strengths for the carboxylic species at many locations relative to those for other chemical constituents. The following interpretations are based primarily on CVs for wet-deposition fluxes. Comparison of CVs suggest that interannual variability in wet-deposition fluxes of HCOOHt at MHGT, ADG, HRT, BAT, AMI, TDP, and KAT were driven primarily by source strength; precipitation amount accounted for roughly half of total variability in wet fluxes only at BTT and MLO. Interannual variability in wet-deposition fluxes of CH3COOHt at ADG, HRT, BAT, AMI, TDP, and KAT were driven primarily by source strength, whereas precipitation amount was relatively more important at MHGT, BTT, and MLO. The large divergence in CVs for wet-deposition fluxes of both species at HRT versus BTT and at MHGT versus ADG (Table 6) suggest the possibility that contributions from local sources may have been significant at HRT and ADG.
3) Long-term trends
The following analysis focuses on S and N species measured at Bermuda, which were expected to vary in response to reductions in SO2 and NOx in the United States mandated by the Clean Air Act and associated amendments. As discussed in detail by Galloway et al. (1996), HCOOHt concentrations at TDP increased significantly over the period of record but processes responsible for this curious trend are not known.
To increase resolution in characterizing trends at Bermuda, we evaluated the utility of merging data from HRT and BTT based on two periods of parallel measures at different locations on the island. Galloway et al. (1988) compared the composition of precipitation sampled in parallel on an event basis from August 1982 to May 1984 at HRT and at High Point (HP), which is located approximately 200 m north of the BTT site. Identical protocols were used at both HRT and HP. Although results based on 66 samples collected concurrently at these sites suggest minor local influences on NH4+, NO3−, and nss SO42− concentrations in precipitation at HRT, the corresponding VWA concentrations for the two datasets were statistically indistinguishable. Precipitation composition at HRT and BTT was also measured in parallel from August 1988 to July 1989 (Table 1, Fig. 5). Because different sampling protocols were used at the two sites (event at HRT and daily at BTT), direct comparisons between paired data were not possible. Although monthly VWA concentrations of nss SO42− at the two sites differed significantly, absolute differences in VWA concentrations based on all data were small (1.5 μeq L−1) relative to the long-term trend (Fig. 5). Differences in NH4+ and NO3− were insignificant. Based on the above, we conclude that the datasets at HRT and BTT could be considered regionally representative and thus merged for evaluation of long-term trends over the 16-yr period of continuous record at Bermuda. Precipitation composition measured at BTT from 2006 through 2009 using identical procedures (see Keene et al. 2014) were included in the trend analysis to extend the duration of record yielding a total (albeit discontinuous) period of record of over 29 yr (late April 1980–June 2009).
Long-term trends in monthly VWA concentrations of (a) nss SO42−, (b) NO3−, and (c) NH4+ at HRT (solid black lines) and BTT (solid gray lines). Dashed lines depict SLRs for the entire period of record. For nss SO42−, slope = −0.474 μeq L−1 yr−1 and r2 = 0.23; for NO3−, slope = 0.018 μeq L−1 yr−1 and r2 = 0.002; and for NH4+, slope = −0.006 μeq L−1 yr−1 and r2 = 0.0003. VWA concentrations measured at both HRT and BTT during the period of parallel operation (26 Jul 1988–14 Aug 1989) are depicted but those for HRT during that period were not included in the regression calculations. Tick marks on x axis correspond to 1 Jan.
Citation: Journal of the Atmospheric Sciences 72, 8; 10.1175/JAS-D-14-0378.1
Based on the SLR for monthly VWA concentration versus time, between May 1980 and June 2009, nss SO42− concentrations in precipitation at Bermuda decreased by 14 μeq L−1 (0.47 μeq L−1 yr−1) or 85% (Fig. 5a). This trend was significant at α = 0.001. The decline was driven primarily by decreases in SO2 emissions across the United States (from 25.9 Tg SO2 yr−1 in 1980 to 9.08 Tg SO2 yr−1 in 2009 or 65%; EPA 2014) in response to the Clean Air Act and associated amendments. The smaller percentage reduction in emissions for the entire United States relative to the corresponding reduction in nss SO42− concentrations in precipitation on Bermuda was driven in part by the fact that a disproportionately large fraction of national reductions occurred in regions over the eastern United States from which air parcels were transported most frequently to Bermuda (e.g., Moody et al. 2014).
Although NOx emissions over the United States also decreased significantly during this period [from 27.1 Tg NOx (as NO2) yr−1 in 1980 to 15.7 Tg NOx yr−1 in 2009 or 42.1%; EPA (2014)], the corresponding NO3− concentrations in precipitation on Bermuda exhibited a very weak but positive trend [SLR slope = 0.018 μeq L−1 yr−1 that was significant (α = 0.036) (Fig. 5b)]. The smaller percentage decrease in NOx relative to SO2 emissions contributed to lower resolution in evaluating the trend but atmospheric processing may have also influenced temporal change. Relative to conventional HOx–NOx photochemistry, halogen radical chemistry in marine air accelerates the oxidation of NOx to soluble HNO3 (Keene et al. 2009). In addition, because of the pH dependence of its solubility, most HNO3 in marine air partitions with the relatively less acidic, shorter-lived, supermicron-diameter aerosol size fractions (Russell et al. 2003; Moody et al. 2014). The net effect of both processes is to accelerate wet- and dry-deposition rates of oxidized N in marine air, thereby limiting its long-distance transport over the ocean and attenuating corresponding impacts on the composition of marine precipitation downwind (Keene et al. 2014). However, NO3− concentrations in precipitation associated with atmospheric transport from North America are higher than those associated with flow from other regions (Keene et al. 2014). These results are consistent with the expectation that the higher concentrations associated with westerly flow reflect contributions from NOx emissions over North America. In addition, NO3− concentrations in precipitation and associated wet-deposition fluxes at Bermuda are substantially higher than those at more remote marine (BAT and AMI) and near-coastal locations (CDP and TDP; Figs. 2e,f), which suggest significant anthropogenic contributions. Consequently, it is unclear why trends in NO3− concentrations in precipitation at Bermuda do not reflect reductions in NOx emissions over North America.
The trend in NH4+ concentrations (Fig. 5c) was statistically indistinguishable from 0.0 (α = 0.168). The earliest NH3 emissions inventory for the United States prepared by the U.S. Environmental Protection Agency (EPA) was for 1990 but the available inventories (ranging from 4.32 Tg NH3 yr−1 in 1990 to 4.34 Tg NH3 yr−1 in 2009) indicate that NH3 emissions in the United States have remained relatively constant over the past 29 yr, which would be consistent with the lack of a significant trend in NH4+ concentrations in precipitation assuming that most NH4+ in precipitation at Bermuda originated from sources in North America (e.g., Keene et al. 2014; Moody et al. 2014).
The significant long-term trend of decreasing nss SO42− concentrations coupled with inconsequential trends in corresponding NO3− and NH4+ concentrations imply that the total amount of soluble acidity associated with the multiphase gas-aerosol system over the western North Atlantic Ocean has decreased over the period of record. This long-term change in atmospheric acidity is evident in the significant decreasing trend (30%) in monthly VWA H+ in precipitation at Bermuda over the period of record (SLR slope = −0.15 μeq L−1 yr−1, r2 = 0.03). As discussed in more detail by Keene et al. (2014), the long-term trend of decreasing atmospheric acidity over the western North Atlantic Ocean has important implications for pH-dependent chemical processes including the gas-aerosol phase partitioning and associated lifetimes against deposition for compounds with pH-dependent solubilities such as NH3.
c. Broader issues
The following analysis focuses primarily on the S and N species that have been the most closely connected to impacts on ecosystems via acidification, fertilization, and associated losses of biodiversity.
1) Anthropogenic contributions to wet-deposition fluxes in developed regions
Wet-deposition fluxes in remote regions provide relevant benchmarks against which to evaluate anthropogenic enhancements of deposition in populated and industrialized regions. For example, annual wet-deposition fluxes of nss SO42− at the three most remote marine-influenced sites (AMI, CDP, and TDP) and at the four remote continental sites with periods of record greater than 1-yr (PKF, LIJ1, LIJ2, and KAT) ranged from factors of 4 to 47 (median factor of 10) lower than the average annual wet-deposition flux over the mid-Atlantic region of the eastern United States between 2002 and 2004 (416 Eq ha−1 yr−1; NADP 2014; Table 4; Fig. 2). As discussed above, anthropogenic sources contribute to nss SO42− fluxes in remote regions and, consequently, we infer that these differences correspond to lower limits for the actual range in relative contributions of anthropogenic sources to wet-deposition fluxes of nss SO42− over the eastern United States.
The average annual wet-deposition fluxes of NO3− at these remote sites ranged from factors of 5 to 61 (median factor of 11) lower than those for the mid-Atlantic region of the eastern United States during this period (226 Eq ha−1 yr−1; NADP 2014; Table 4; Fig. 2). The corresponding average annual wet-deposition fluxes of NH4+ were 3–39 (median of 6) times lower than that for the mid-Atlantic region of the eastern United States during the same period (172 Eq ha−1 yr−1; NADP 2014; Table 4; Fig. 2). However, precipitation sampled by the NADP is not adequately preserved against microbial degradation between collection and analysis and, thus, wet fluxes of NH4+ reported by the NADP underestimate actual fluxes by factors of about 5%–15% (Keene et al. 2002). Because data reported herein correspond to samples that were adequately preserved against microbial activity, the actual differences between wet fluxes at remote sites and those for the mid-Atlantic region of the eastern United States are proportionately greater than inferred from comparison with NADP data as depicted in Fig. 2.
The combined wet fluxes of inorganic N (NO3− + NH4+) at these remote sites ranged from 8 to 83 (median 56) Eq ha−1 yr−1, which are factors of 5–49 (median of 7) lower than the total wet-deposition flux of inorganic N reported for the mid-Atlantic region of the eastern United States during 2002–04 (398 Eq ha−1 yr−1; NADP 2014; Fig. 2). Comparisons with wet-deposition fluxes in other developed regions of the world (e.g., Galloway et al. 2004) yield similar ranges in anthropogenic enhancements. Because of losses of NH4+ from NADP samples, the actual differences for the mid-Atlantic region of the United States are marginally greater. As summarized above and reported in detail elsewhere (e.g., Likens 2013), these large increases in the wet-deposition fluxes of S and N species within and downwind of industrialize regions have significantly perturbed the structure and functioning of ecosystems in those regions.
2) Measured versus modeled precipitation composition and wet-deposition fluxes
Over the past quarter century, a number of models have been developed to simulate spatial and temporal variability in deposition rates of atmospheric S and N species to Earth’s surface. These models are parameterized and validated based on measured wet-deposition fluxes in both populated and remote regions. Dentener et al. (2006) characterized performance of 23 global chemical-transport models (GCTMs) based on comparison between simulated and measured wet-deposition fluxes of nss SO42−, NO3−, and NH4+ for model year 2000. Results for Europe and North America indicate reasonably good agreement between simulated and measured fluxes; 60%–70% of the model-calculated wet-deposition rates agreed within ±50% of measurements. However, simulated and measured fluxes for remote regions (including 11 of the datasets presented in this paper) diverged to a greater degree (factor of 2 or more).
Two more recent studies also assessed model performance via both intercomparison of simulated results and comparison of simulated results with measurements. Lamarque et al. (2013) evaluated spatial and temporal variability in deposition fluxes of nss SO42−, NO3−, and NH4+ based on ensemble-mean simulations for model years 1850, 1980, and 2000 as well as predicted fluxes in 2030 and 2100 under various assumed scenarios for future emissions as part of the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP). Performance of 11 models was evaluated but not all models simulated fluxes for all chemical species so the number of simulations varied among species. Ensemble-mean simulations of wet-deposition fluxes of nss SO42− and TIN (NO3− and NH4+) for model years 1980 and 2000 at the sampling locations reported herein are compared with the corresponding annual-mean (±1 standard deviation) wet-deposition flux measured at those locations (Table 7).
Comparison of simulated and measured (meas.) wet-deposition fluxes (Eq ha−1 yr−1) of nss SO42− and TIN (NO3− + NH4+) for sampling locations reported herein.
In another study, Vet et al. (2014) reported spatial distributions for concentrations and deposition fluxes of nss SO42− and TIN (NO3− and NH4+) for model year 2001 based on ensemble-mean simulations by 21 GCTMs as part of Phase 1 of the Coordinated Model Studies Activities of the Task Force on Hemispheric Transport of Air Pollution (TF HTAP). Like ACCMIP, not all models simulated fluxes for all chemical species so the number of simulations varied among species. In addition, outlier results were selectively eliminated from the final ensemble results. The reported ranges in the ensemble means for the simulated wet-deposition fluxes that bracket sampling locations reported herein are compared with the corresponding annual-mean (±1 standard deviation) wet-deposition fluxes measured at those locations (Table 7).
Because 1) some of the same models were used in both ACCMIP and TF HTAP, 2) the measurement datasets with which models were compared in ACCMIP and TF HTAP were also similar, and 3) wet deposition in all models is parameterized using available measurement databases, it is not surprising that the results of the two studies are also similar. Most of the wet fluxes simulated at our sampling locations for 2000 reported by Lamarque et al. (2013) fall within the range reported for 2001 by Vet et al. (2014) and those that do not are very close to the limits of those ranges (Table 7). For most locations, the ensemble-mean wet-deposition fluxes of nss SO42− and TIN reported by Lamarque et al. (2013) fall within ±1 standard deviation of the corresponding mean for measured wet fluxes. Similarly, variability in means for most of the measured fluxes falls within or overlaps the corresponding ranges reported by Vet et al. (2014). The greater divergence evident at several locations including CDP, TDP, and LIJ probably reflects interrelated influences of large spatial gradients in atmospheric composition, precipitation amount, and associated wet removal. Such influences are not captured by relatively coarse model grids. For example, CDP and TDP are on the windward and leeward sides, respectively, of the Andes. The very high precipitation amounts along the Chilean coast at CDP (Table 1) effectively scavenge air flowing onshore, yielding relatively high wet fluxes, whereas the air parcels that produce precipitation sampled on the leeward side of the Andes at TDP had been scavenged by coastal precipitation upwind yielding relatively low wet fluxes. Such spatial variability decreases when relatively localized influences of this nature are averaged over larger spatial scales. Similarly, most precipitation at LIJ is associated with westerly and southwesterly flow from the Bay of Bengal and India over the Himalayas during the period of the Indian monsoon (Keene et al. 1989). Heavy precipitation on the windward side of the mountains scavenges air parcels upwind of the site, yielding low wet fluxes relative to simulations. ADG and HRT lie in regions with large spatial gradients in atmospheric composition. Relatively high measured deposition fluxes of nss SO42− compared to model simulations probably reflects the inability of models to reliably capture these gradients.
Although ensemble-mean simulations reproduce general features of wet deposition in remote regions (Table 7), the divergence between modeled and measured fluxes is typically much greater for remote relative to industrialized regions (Dentener et al. 2006; Vet et al. 2014). A number of issues contributed to this relatively larger divergence. Simulations based on specific model years represent discrete time slices whereas the measurements corresponded to wet fluxes measured over many years. In addition, many of the datasets evaluated herein do not include the modeled years with which they were compared and wet fluxes at all sites exhibited significant interannual variability. Finally, as discussed above, comparisons between measurements at specific locations versus simulated fluxes averaged over relatively coarse model grid cells are inherently problematic particularly in regions with large gradients such as coastal areas (e.g., Long et al. 2014). These comparisons underscore the need for models with higher spatial resolution coupled with observations at greater numbers of representative locations in remote regions.
3) Wet deposition versus total deposition
Total atmospheric deposition includes both wet deposition via precipitation and the dry deposition of gases and aerosols to surfaces. In some regions, the deposition of cloud and fog droplets to surfaces also contributes significantly to total deposition (Weathers et al. 1988). Reliable quantification of total deposition is required to evaluate associated impacts on ecosystems and to constrain biogeochemical cycles. Wet-deposition fluxes can be measured reliably and at relatively modest cost at many locations. However, reliable characterization of dry-deposition fluxes of gaseous and particulate S and N species is considerably more challenging and generally limited to short-term measurement campaigns that employ more sophisticated and labor-intensive instrumentation coupled with model calculations. A detailed review of this topic is beyond the scope of this study but a brief overview will provide relevant perspective.
Precipitation events efficiently scavenge soluble gases and aerosols from the air column and, consequently, wet deposition dominates total deposition in regions such as rain forests that experiences frequent precipitation. Conversely, dry-deposition processes are generally slower but operate continuously in the absence of precipitation and, consequently, dry deposition dominates total deposition in arid regions and during dry seasons in regions with distinct wet and dry cycles such as KAT. In most regions, available evidence suggests that both wet and dry pathways contribute significantly to total deposition (e.g., Likens 2013).
Physicochemical characteristics of the atmosphere also modulate the gas-aerosol phase partitioning, aerosol size distributions, and associated dry-deposition rates of S and N species. For example, because its solubility in aerosol solutions is pH-dependent, HNO3 partitions primarily into the gas phase in the highly acidic atmospheric system over the eastern United States, whereas it partitions primarily into the relatively less-acidic, supermicron aerosol size fractions in marine air (Russell et al. 2003; Fischer et al. 2006; Keene et al. 2009). Because larger size fractions of marine aerosols exhibit greater deposition velocities than HNO3, dry-deposition rates of total NO3 (HNO3 + particulate NO3−) to the sea surface increase as polluted air masses advect over the ocean thereby increasing the relative importance of dry- versus wet-deposition fluxes of total NO3 (e.g., Spokes et al. 2000; Russell et al. 2003). Similarly, NH3 partitions primarily into highly acidic submicron aerosol size fractions in polluted air over the eastern United States (e.g., Smith et al. 2007), whereas it partitions primarily into the gas phase in less-acidic agricultural regions associated with high NH3 emissions (Young et al. 2013). Because the dry-deposition velocities for NH3 are greater than those for submicron aerosol size fractions with which NH3 preferentially partitions (Smith et al. 2007), the dry-deposition fluxes of total NHx (NH3 + particulate NH4+) and the associated importance of dry versus wet deposition are disproportionately greater (relative to atmospheric concentrations) downwind of major agricultural emissions sources.
Many of the major physical (e.g., frequency of precipitation events) and chemical processes (e.g., size-resolved aerosol composition including pH) that modulate dry-deposition fluxes of major N and S species are not evaluated explicitly in current GCTMs. Consequently, processes that drive dry-deposition fluxes of gases and aerosols are typically implemented using highly parameterized formulations that are constrained by mass-balance considerations. GCTM simulations based on these parameterizations suggest that, globally, wet deposition accounts for 50%–70% of the total (wet + dry) deposition flux of SOx (SO2 + particulate nss SO42−), 40%–70% of the total deposition flux of NO3 (HNO3 + particulate NO3−), and 40%–80% of total deposition flux of NHx (NH3 + particulate NH4+) (Dentener et al. 2006).
4) Strategy to advance knowledge of deposition rates to remote regions
It is evident from the above that current understanding of atmospheric deposition fluxes in remote regions is poorly constrained. From an observational perspective, wet-deposition rates have been reliably quantified at relatively few locations and essentially no reliable data are available with which to estimate dry-deposition fluxes at given locations for periods of more than a month or two. Comparisons between model calculations and observations suggest that simulated wet-deposition fluxes of S and N species in remote regions are accurate to within a factor of only about 2. The accuracy of corresponding dry-deposition estimates cannot be reliably evaluated given the limited observational database and the highly parameterized treatment of dry deposition in GCTMs. In light of these uncertainties, we infer that current estimates of the total deposition of S and N species to remote regions is uncertain by factors of greater than 2. Total deposition rates of other chemical constituents including carboxylic acids and base cations are uncertain to greater degrees.
Our current understanding of and ability to predict atmospheric deposition and associated environmental implications in remote regions could be improved in several ways. Reliable measurements of precipitation composition and wet-deposition fluxes at a larger number of regionally representative continental and marine locations in remote regions would provide additional information with which to parameterize and evaluate the performance of GCTMs. Ideally, all samples would be analyzed in a single laboratory that employs rigorous quality-control procedures but, if not practical, multiple laboratories involved in such efforts should routinely intercompare performance to ensure that resulting data are comparable. Constraining dry-deposition velocities and corresponding fluxes will remain challenging (Mitchell et al. 2011). The high cost of generating the multiphase data required for credible estimates of dry fluxes prohibits routine application as part of measurement networks. However, compilation of available near-surface measurements of size-resolved aerosol composition; concurrent mixing ratios of SO2, HNO3, and NH3; and associated meteorological conditions (temperature, RH, and wind velocity) from past field campaigns in remote regions would provide useful context for constraining the associated dry fluxes to underlying surfaces. Finally, higher spatial resolution coupled with more realistic parameterization of thermodynamic and kinetic processes that regulate the phase partitioning of gases with size-resolved aerosols and the corresponding dry-deposition fluxes in GCTMs would also substantially improve the representativeness of simulated results.
4. Summary
Precipitation was sampled on an event or daily basis at 14 remote marine, near-coastal, continental, and high-elevation locations over a latitude range from 65°N to 51°S and subsequently characterized for major organic and inorganic chemical constituents and corresponding wet-deposition fluxes. The quality-assured database corresponds to 4958 individual samples.
S and N species transported from surrounding continents and emitted from ships at sea sustained generally higher VWA concentrations and wet-deposition fluxes of nss SO42−, NO3−, and NH4+ at sites in the North Atlantic basin relative to those at marine and near-coastal locations in the Southern Hemisphere. Marine biogenic sources accounted for 0.4 to 3.3 μeq L−1 of annual VWA nss SO42− in marine precipitation. VWA concentrations of SO42− in remote continental locations (2.9–7.7 μeq L−1) were generally higher and more variable. SO42− from anthropogenic sources contributed to these differences. Excluding relatively more polluted sites in the North Atlantic region, the high-elevation site at MLO, and SCL for which the data record was less than 1 year, annual wet-deposition fluxes of nss SO42− at remote sites ranged from factors of 4 to 47 (median of 10) less than that for the mid-Atlantic region of the eastern United States between 2002 and 2004.
VWA concentrations of NO3− at the three marine-influenced sites in the Southern Hemisphere (0.5–1.3 μeq L−1) were lower by factors of 2–14 relative to those at the North Atlantic sites. VWA NO3− concentrations at continental sites ranged from 1.4 μeq L−1 at LIJ1 to 4.8 μeq L−1 at KAT. The relatively higher concentrations at KAT were driven in part by contributions from combustion products of biomass burning in association with regional agriculture. Excluding sites in the North Atlantic region, MLO, and SCL, annual wet-deposition fluxes of NO3− at remote sites ranged from factors of 5–61 (median of 11) less than that for the mid-Atlantic region of the eastern United States during 2002–04.
VWA concentrations of NH4+ at marine-influenced sites in the Southern Hemisphere (0.5–2.6 μeq L−1) were generally lower than those at continental sites (2.3–4.2 μeq L−1). Excluding the North Atlantic sites, MLO, and SCL, annual wet-deposition fluxes of NH4+ at remote sites ranged from factors of 3 to 39 (median of 6) times less than that for the mid-Atlantic region of the eastern United States during 2002–04.
Excluding the North Atlantic sites, MLO, and SCL, annual wet-deposition fluxes of total inorganic N (NO3− + NH4+) ranged from 8 to 83 Eq ha−1 yr−1, which are factors of 5–49 (median of 7) less than the deposition flux of total inorganic N reported for the mid-Atlantic region of the eastern United States during 2002–04.
In contrast to S and N species, ranges in HCOOHt and CH3COOHt concentrations and wet-deposition fluxes at marine sites in the Northern and Southern Hemispheres overlapped, which implies that these acids originated primarily from natural marine-derived precursors and that the corresponding source strengths were relatively constant spatially. The greater variability in concentrations and depositions at near-coastal and continental sites suggests greater heterogeneity in natural terrestrial source strengths coupled in some regions with significant contributions from anthropogenic sources including agricultural burning at KAT.
Significant seasonal variability in wet-deposition fluxes was driven by corresponding variability in major physical and biological drivers. Wet fluxes at KAT and LIJ varied in response to the seasonal monsoon. Significant variability in Na+ deposition at all marine sites except BAT reflects seasonal variability in wind velocity and associated production and scavenging of marine aerosol. Significant differences in crustal and combustion-derived species at BAT result from a combination of seasonal variability in production of mineral aerosol over North Africa and its subsequent transport over the tropical North Atlantic coupled with seasonal shifts in transport from upwind source regions for combustion products in Europe and central Africa. Seasonal variability in CH3SO3− at the two high-latitude marine sites (MHT and AMI) reflects seasonality in emission of its biogenic precursor [(CH3)2S] from the surface ocean.
CVs provide qualitative insight regarding the relative importance of variability in precipitation amount versus source strength in driving interannual variability in wet-deposition fluxes. While both are important for most species at most sites, relative to those for S and N species, CVs for interannual variability in wet-deposition fluxes in carboxylic acids were typically much greater than those for precipitation amount. These results imply that interannual variability in source strengths for carboxylic acids at many locations is relatively greater than those for other chemical constituents.
Based on the SLR fit to VWA monthly concentrations versus time, nss SO42− concentrations in precipitation at Bermuda decreased significantly by 85% between May 1980 and June 2009. This trend was driven primarily by reductions in SO2 emissions over the United States in response to the Clean Air Act and associated amendments. Corresponding VWA concentrations of NO3− were marginally positive over the period of record despite large reductions (47%) in NOx emissions over the United States during this period. The trend in NH4+ was statistically indistinguishable from 0.0. These results imply that atmospheric acidity over the western North Atlantic Ocean has also decreased as reflected in the significant 30% decline in VWA H+ over the period of record. The changing atmospheric acidity has important implications for pH-dependent chemical processes including the gas-aerosol phase partitioning and lifetime against deposition for compounds with pH-dependent solubilities.
Measurements of precipitation composition and wet-deposition fluxes reported herein illustrate the value and utility of long-term data records (e.g., Lindenmayer and Likens 2010). This particular record is used widely by the research community to investigate the emission, transformation, transport, and deposition of atmospheric chemical constituents and their associated implications for earth systems including nutrient cycling and climate. Comparisons between these measurements and corresponding model simulations indicate that current predictive capabilities for the VWA composition of precipitation, corresponding wet-deposition fluxes, and associated environmental influences of atmospheric deposition in remote regions are uncertain by factors of 2 or more.
Acknowledgments
We thank D. Brown and G. Spain at MHGT; M. O’Sullivan at ADG; D. Connelly, A. Glasspoole, T. Jickells, and K. Simmons at HRT and BTT; C. Shea at BAT; A. Gaudry at AMI; B. Baker at CPE; J. Gonzales and C. Yanez at TDP; M. Cogan and C. Lasater at POK; staff at the Yunnan Environmental Monitoring Station at LIJ1 and LIJ2; H. Clark at SCL; J. Locke at KAT; and A. Yoshinaga at MLO for collecting the precipitation samples reported herein. K. Scott, E. Day, J. Montag and T. J. Wangerman at UVA assisted in supplying sites, analyzing samples, quality assuring data, and archiving results. H. Maring at the University of Miami (now at NASA) managed field operations during AEROCE. Support for the GPCP and WATOX was provided by National Oceanic and Atmospheric Administration through awards to UVA; support for AEROCE and subsequent measurements of precipitation composition at BTT was provided by the National Science Foundation (NSF) through Awards AGS-8701291, -9013128, -9414293, and -0541570 to UVA; and support for CPES was provided the Andrew W. Mellon Foundation through an award to UVA. NSF provided additional support for KNM through a REU award to UVA. JNG acknowledges his appreciation for visiting scientist position at the Marine Biological Laboratory where part of this analysis was performed. The authors appreciate the assistance of and helpful discussions with Richard Artz, Frank Dentener, Gillen Hollis, Jean-Francois Lamarque, Mike Lomas, and John Miller.
REFERENCES
Akagi, S. K., and Coauthors, 2012: Evolution of trace gases and particles emitted by a chaparral fire in California. Atmos. Chem. Phys., 12, 1397–1421, doi:10.5194/acp-12-1397-2012.
Albert, M. F. M. A., M. Schaap, G. de Leeuw, and P. J. H. Builtjes, 2010: Progress in the determination of the sea spray source function using satellite data. J. Integr. Environ. Sci., 7, 159–166, doi:10.1080/19438151003621466.
Altieri, K. E., M. G. Hastings, A. J. Peters, S. Oleynik, and D. M. Sigman, 2014: Isotopic evidence for a marine ammonium source in rainwater at Bermuda. Global Biogeochem. Cycles, 28, 1066–1080, doi:10.1002/2014GB004809.
Andreae, M. O., and P. Merlet, 2001: Emission of trace gases and aerosols from biomass burning. Global Biogeochem. Cycles, 15, 955–966, doi:10.1029/2000GB001382.
Andreae, M. O., and D. Rosenfeld, 2008: Aerosol–cloud–precipitation interactions. Part 1. The nature and sources of cloud-active aerosols. Earth-Sci. Rev., 89, 13–41, doi:10.1016/j.earscirev.2008.03.001.
Andreae, M. O., R. W. Talbot, T. W. Andreae, and R. C. Harriss, 1988: Formic and acetic acid over the central Amazon region, Brazil: 1. Dry season. J. Geophys. Res., 93, 1616–1624, doi:10.1029/JD093iD02p01616.
Artaxo, P., and Coauthors, 2002: Physical and chemical properties of aerosols in the wet and dry seasons in Rondonia, Amazonia. J. Geophys. Res., 107, 8081, doi:10.1029/2001JD000666.
Bates, T. S., J. A. Calhoun, and P. K. Quinn, 1992: Variations in the methanesulfonate to sulfate ratio in submicrometer marine aerosol particles over the South Pacific Ocean. J. Geophys. Res., 97, 9859–9865, doi:10.1029/92JD00411.
Bates, T. S., and Coauthors, 2012: Measurements of ocean derived aerosol off the coast of California. J. Geophys. Res., 117, D00V15, doi:10.1029/2012JD017588.
Bouwman, A. F., D. S. Lee, W. A. H. Asman, F. J. Dentener, K. W. Van Der Hock, and J. G. J. Olivier, 1997: A global high-resolution emission inventory for ammonia. Global Biogeochem. Cycles, 11, 561–587, doi:10.1029/97GB02266.
Chameides, W. L., and A. W. Stelson, 1992: Aqueous-phase chemical processes in deliquescent sea-salt aerosols: A mechanism that couples the atmospheric cycles of S and sea salt. J. Geophys. Res., 97, 20 565–20 580, doi:10.1029/92JD01923.
Church, T. M., J. N. Galloway, T. D. Jickells, and A. H. Knap, 1982: The chemistry of western Atlantic precipitation at the mid-ocean coast and on Bermuda. J. Geophys. Res., 87, 11 013–11 018, doi:10.1029/JC087iC13p11013.
Ciais, P., and Coauthors, 2014: Carbon and other biogeochemical cycles. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 465–570.
Clarke, A. D., and Coauthors, 2013: Free troposphere as a major source of CCN in the equatorial Pacific boundary layer: Long-range transport and teleconnections. Atmos. Chem. Phys., 13, 7511–7529, doi:10.5194/acp-13-7511-2013.
Corbett, J. J., P. S. Fischbeck, and S. N. Pandis, 1999: Global nitrogen and sulfur inventories for oceangoing ships. J. Geophys. Res., 104, 3457–3470, doi:10.1029/1998JD100040.
Cornell, S. E., T. D. Jickells, and C. A. Thornton, 1998: Urea in rainwater and atmospheric aerosols. Atmos. Environ., 32, 1903–1910, doi:10.1016/S1352-2310(97)00487-1.
Cornell, S. E., K. Mace, S. Coeppicus, R. Duce, B. Huebert, T. Jickells, and L. Z. Zhuang, 2001: Organic nitrogen in Hawaiian rain and aerosols. J. Geophys. Res., 106, 7973–7983, doi:10.1029/2000JD900655.
Dahl, E. E., S. A. Yvon-Lewis, and E. S. Saltzman, 2005: Saturation anomalies of alkyl nitrates in the tropical Pacific Ocean. Geophys. Res. Lett., 32, L20817, doi:10.1029/2005GL023896.
Davis, D., and Coauthors, 1998: DMS oxidation in the Antarctic marine boundary layer: Comparison of model simulations and field observations of DMS, DMSO, DMSO2, H2SO4(g), MSA(g), and MSA(p). J. Geophys. Res., 103, 1657–1678, doi:10.1029/97JD03452.
Dayan, U., J. M. Miller, W. C. Keene, and J. N. Galloway, 1985: An analysis of precipitation chemistry data from Alaska. Atmos. Environ., 19, 651–657, doi:10.1016/0004-6981(85)90044-7.
de Gouw, J. A., and Coauthors, 2005: Budget of organic carbon in a polluted atmosphere: Results from the New England Air Quality Study in 2002. J. Geophys. Res., 110, D16305, doi:10.1029/2004JD005623.
Dentener, F., and Coauthors, 2006: Nitrogen and sulfur deposition on regional and global scales: A multimodel evaluation. Global Biogeochem. Cycles, 20, GB4003, doi:10.1029/2005GB002672.
Duce, R. A., and Coauthors, 2008: Impacts of atmospheric anthropogenic nitrogen on the open ocean. Science, 320, 893–897, doi:10.1126/science.1150369.
EPA, 2014: National Emissions Inventory (NEI) air pollutant emissions trends data. Environmental Protection Agency, accessed 16 May 2014. [Available online at http://www.epa.gov/ttn/chief/trends/.]
Facchini, M. C., and Coauthors, 2008: Primary submicron marine aerosol dominated by insoluble organic colloids and aggregates. Geophys. Res. Lett., 35, L17814, doi:10.1029/2008GL034210.
Fischer, E., A. Pszenny, W. Keene, J. Maben, A. Smith, A. Stohl, and R. Talbot, 2006: Nitric acid phase partitioning and cycling in the New England coastal atmosphere. J. Geophys. Res., 111, D23S09, doi:10.1029/2006JD007328.
Freedman, D. A., 1981: Bootstrapping regression models. Ann. Stat., 9, 1218–1228, doi:10.1214/aos/1176345638.
Gabriel, R., L. Schäfer, C. Gerlach, T. Rausch, and J. Kesselmeier, 1999: Factors controlling the emissions of volatile organic acids from leaves of Quercus ilex L. (Holm oak). Atmos. Environ., 33, 1347–1355, doi:10.1016/S1352-2310(98)00369-0.
Galloway, J. N., and A. Gaudry, 1984: The composition of precipitation on Amsterdam Island, Indian Ocean. Atmos. Environ., 18, 2649–2656, doi:10.1016/0004-6981(84)90329-9.
Galloway, J. N., G. E. Likens, W. C. Keene, and J. M. Miller, 1982: The composition of precipitation in remote areas of the world. J. Geophys. Res., 87, 8771–8786, doi:10.1029/JC087iC11p08771.
Galloway, J. N., G. E. Likens, and M. E. Hawley, 1984: Acid precipitation: Natural versus anthropogenic components. Science, 226, 829–831, doi:10.1126/science.226.4676.829.
Galloway, J. N., J. J. Tokos Jr., A. H. Knap, and W. C. Keene, 1988: Local influences on the composition of precipitation on Bermuda. Tellus, 40B, 178–188, doi:10.1111/j.1600-0889.1988.tb00288.x.
Galloway, J. N., W. C. Keene, R. S. Artz, J. M. Miller, T. M. Church, and A. H. Knap, 1989: Processes controlling the concentration of SO4 =, NO3 −, NH4 +, H+, HCOOT and CH3COOT in precipitation on Bermuda. Tellus, 41B, 427–443, doi:10.1111/j.1600-0889.1989.tb00319.x.
Galloway, J. N., D. L. Savoie, W. C. Keene, and J. M. Prospero, 1993: The temporal and spatial variability of scavenging ratios for NSS sulfate, nitrate, methanesulfonate and sodium in the atmosphere over the North Atlantic Ocean. Atmos. Environ., 27A, 235–250, doi:10.1016/0960-1686(93)90354-2.
Galloway, J. N., W. C. Keene, and G. E. Likens, 1996: Processes controlling the composition of precipitation at a remote southern hemispheric location: Torres del Paine National Park, Chile. J. Geophys. Res., 101, 6883–6897, doi:10.1029/95JD03229.
Galloway, J. N., and Coauthors, 2004: Nitrogen cycles: Past, present, and future. Biogeochemistry,70, 153–226, doi:10.1007/s10533-004-0370-0.
Gantt, B., N. Meskhidze, and D. Kamykowski, 2009: A new physically-based quantification of marine isoprene and primary organic aerosol emissions. Atmos. Chem. Phys., 9, 4915–4927, doi:10.5194/acp-9-4915-2009.
Glasius, M., and Coauthors, 2000: Sources to formic acid studied by carbon isotopic analysis and air mass characterization. Atmos. Environ., 34, 2471–2479, doi:10.1016/S1352-2310(99)00416-1.
Glasius, M., and Coauthors, 2001: Relative contribution of biogenic and anthropogenic sources to formic and acetic acids in the atmospheric boundary layer. J. Geophys. Res., 106, 7415–7426, doi:10.1029/2000JD900676.
Gong, S. L., L. A. Barrie, and J.-P. Blanchet, 1997: Modeling sea-salt aerosols in the atmosphere: 1. Model development. J. Geophys. Res., 102, 3805–3818, doi:10.1029/96JD02953.
Gorzelska, K., J. N. Galloway, K. Watterson, and W. C. Keene, 1992: Water-soluble primary amine compounds in rural continental precipitation. Atmos. Environ., 26A, 1005–1018, doi:10.1016/0960-1686(92)90032-G.
Graedel, T. E., and T. Eisner, 1988: Atmospheric formic acid from formicine ants. Tellus, 40B, 335–339, doi:10.1111/j.1600-0889.1988.tb00107.x.
Graedel, T. E., and W. C. Keene, 1995: The tropospheric budget of reactive chlorine. Global Biogeochem. Cycles, 9, 47–78, doi:10.1029/94GB03103.
Granby, K., A. H. Egeløv, T. Nielsen, and C. Lohse, 1997: Carboxylic acids: Seasonal variation and relation to chemical and meteorological parameters. J. Atmos. Chem., 28, 195–207, doi:10.1023/A:1005877419395.
Hedin, L. O., J. J. Armesto, and A. H. Johnson, 1995: Patterns of nutrient loss from unpolluted, old-growth temperate forests: Evaluation of biogeochemical theory. Ecology, 76, 493–509, doi:10.2307/1941208.
Herlihy, L. J., J. N. Galloway, and A. L. Mills, 1987: Bacterial utilization of formic and acetic acid in rainwater. Atmos. Environ., 21, 2397–2402, doi:10.1016/0004-6981(87)90374-X.
Holben, B. N., A. Setzer, T. F. Eck, A. Pereira, and I. Slutsker, 1996: Effect of dry-season biomass burning on Amazon basin aerosol concentrations and optical properties, 1992–1994. J. Geophys. Res., 101, 19 465–19 481, doi:10.1029/96JD01114.
Jickells, T., A. Knap, T. Church, J. Galloway, and J. Miller, 1982: Acid rain on Bermuda. Nature, 297, 55–57, doi:10.1038/297055a0.
Jickells, T., A. R. Baker, J. N. Cape, S. E. Cornell, and E. Nemitz, 2013: The cycling of organic nitrogen through the atmosphere. Philos. Trans. Roy. Soc. London, B368, 20130115, doi:10.1098/rstb.2013.0115.
Jordan, C. E., and Coauthors, 2015: Origins of aerosol chlorine during winter over north central Colorado, USA. J. Geophys. Res. Atmos., 120, 678–694, doi:10.1002/2014JD022294.
Jourdain, B., and M. Legrand, 2001: Seasonal variations of atmospheric dimethylsulfide, dimethylsulfoxide, sulfur dioxide, methansulfonate, and non-sea-salt sulfate aerosols at Durmont d’Urville (coastal Antarctica) (December 1998 to July 1999). J. Geophys. Res., 106, 14 391–14 408, doi:10.1029/2000JD900841.
Kanakidou, M., and Coauthors, 2012: Atmospheric fluxes of organic N and P to the global ocean. Global Biogeochem. Cycles, 26, GB3026, doi:10.1029/2011GB004277.
Kawamura, K., L.-L. Ng, and I. R. Kaplan, 1985: Determination of organic acids (C1-C10) in the atmosphere, motor exhausts, and engine oils. Environ. Sci. Technol., 19, 1082–1086, doi:10.1021/es00141a010.
Keene, W. C., and J. N. Galloway, 1984: A note on acid rain in an Amazon rain forest. Tellus, 36B, 137–138, doi:10.3402/tellusb.v36i2.14883.
Keene, W. C., and J. N. Galloway, 1986: Considerations regarding sources for formic and acetic acids in the troposphere. J. Geophys. Res., 91, 14 466–14 474, doi:10.1029/JD091iD13p14466.
Keene, W. C., and J. N. Galloway, 1988: The biogeochemical cycling of formic and acetic acids through the troposphere: An overview of current understanding. Tellus, 40B, 322–334, doi:10.1111/j.1600-0889.1988.tb00106.x.
Keene, W. C., J. N. Galloway, and J. D. Holden Jr., 1983: Measurement of weak organic acidity in precipitation from remote areas of the world. J. Geophys. Res., 88, 5122–5130, doi:10.1029/JC088iC09p05122.
Keene, W. C., A. A. P. Pszenny, J. N. Galloway, and M. E. Hawley, 1986: Sea-salt corrections and interpretation of constituent ratios in marine precipitation. J. Geophys. Res., 91, 6647–6658, doi:10.1029/JD091iD06p06647.
Keene, W. C., G. Wu, R. S. Artz, J. N. Galloway, J. Liu, J. M. Harris, and J. M. Miller, 1989: Processes controlling the chemical composition of precipitation at a remote temperate continental site in south-central China. Proc. Int. Conf. on Global and Regional Environmental Atmospheric Chemistry, Beijing, China, U. S. Department of Commerce, 203.
Keene, W. C., R. Sander, A. A. P. Pszenny, R. Vogt, P. J. Crutzen, and J. N. Galloway, 1998: Aerosol pH in the marine boundary layer: A review and model evaluation. J. Aerosol Sci., 29, 339–356, doi:10.1016/S0021-8502(97)10011-8.
Keene, W. C., and Coauthors, 1999: Composite global emissions of reactive chlorine from natural and anthropogenic sources: Reactive Chlorine Emissions Inventory. J. Geophys. Res.,104, 8429–8440, doi:10.1029/1998JD100084.
Keene, W. C., J. A. Montag, J. R. Maben, M. Southwell, J. Leonard, T. M. Church, J. L. Moody, and J. N. Galloway, 2002: Organic nitrogen in precipitation over eastern North America. Atmos. Environ., 36, 4529–4540, doi:10.1016/S1352-2310(02)00403-X.
Keene, W. C., J. M. Lobert, P. J. Crutzen, J. R. Maben, D. H. Scharffe, T. Landmann, C. Hély, and C. Brain, 2006: Emissions of major gaseous and particulate species during experimental burns of southern African biomass. J. Geophys. Res., 111, D04301, doi:10.1029/2005JD006319.
Keene, W. C., and Coauthors, 2007: Chemical and physical characteristics of nascent aerosols produced by bursting bubbles at a model air-sea interface. J. Geophys. Res.,112, D21202, doi:10.1029/2007JD008464.
Keene, W. C., and Coauthors, 2009: Latitudinal variation in the multiphase chemical processing of inorganic halogens and related species over the eastern North and South Atlantic Oceans. Atmos. Chem. Phys., 9, 7361–7385, doi:10.5194/acp-9-7361-2009.
Keene, W. C., J. L. Moody, J. N. Galloway, J. M. Prospero, O. R. Cooper, S. Eckhardt, and J. R. Maben, 2014: Long-term trends in aerosol and precipitation composition over the western North Atlantic Ocean at Bermuda. Atmos. Chem. Phys., 14, 8119–8135, doi:10.5194/acp-14-8119-2014.
Koch, D., T. C. Bond, D. Streets, N. Unger, and R. G. van der Werf, 2007: Global impacts of aerosols from particular source regions and sectors. J. Geophys. Res.,112, D02205, doi:10.1029/2005JD007024.
Kunz, G. L., G. de Leeuw, E. Becker, and C. D. O’Dowd, 2002: Lidar observations of atmospheric boundary layer structure and sea spray aerosol plumes generation and transport at Mace Head, Ireland (PARFORCE experiment). J. Geophys. Res., 107, 8106, doi:10.1029/2001JD001240.
Lamarque, J. F., and Coauthors, 2013: Multi-model mean nitrogen and sulfur deposition from the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP): Evaluation of historical and projected future changes. Atmos. Chem. Phys., 13, 7997–8018, doi:10.5194/acp-13-7997-2013.
Langford, A. O., F. C. Fehsenfeld, J. Zachariassen, and D. S. Schimel, 1992: Gaseous ammonia fluxes and background concentrations in terrestrial ecosystems of the United States. Global Biogeochem. Cycles, 6, 459–483, doi:10.1029/92GB02123.
Lawler, M. J., B. D. Finley, W. C. Keene, A. A. P. Pszenny, K. A. Read, R. von Glasow, and E. S. Saltzman, 2009: Pollution-enhanced reactive chlorine chemistry in the eastern tropical Atlantic boundary layer. Geophys. Res. Lett., 36, L08810, doi:10.1029/2008GL036666.
Lawler, M. J., R. Sander, L. J. Carpenter, J. D. Lee, R. von Glasow, R. Sommariva, and E. S. Saltzman, 2011: HOCl and Cl2 observations in marine air. Atmos. Chem. Phys., 11, 7617–7628, doi:10.5194/acp-11-7617-2011.
Lawrence, M. G., and P. J. Crutzen, 1999: Influence of NOx emissions from ships on tropospheric photochemistry and climate. Nature, 402, 167–170, doi:10.1038/46013.
Lewis, E. R., and S. E. Schwartz, 2004: Sea Salt Aerosol Production: Mechanisms, Methods, Measurements, and Models - A Critical Review. Geophys. Monogr., Vol. 152, Amer. Geophys. Union, 413 pp, doi:10.1029/GM152.
Likens, G. E., 2013: Biogeochemistry of a Forested Ecosystem. 3rd ed. Springer-Verlag, 208 pp.
Likens, G. E., J. S. Eaton, and J. S. Galloway, 1977: Precipitation as a source of nutrients for terrestrial and aquatic ecosystems. Precipitation Scavenging (1974), R. G. Semonin and R. W. Beadle, Eds., Symposium Series 41, ERDA, 552–570.
Likens, G. E., W. C. Keene, J. M. Miller, and J. N. Galloway, 1987: Chemistry of precipitation from a remote, terrestrial site in Australia. J. Geophys. Res., 92, 13 299–13 314, doi:10.1029/JD092iD11p13299.
Lin, C. T., A. R. Baker, T. D. Jickells, S. Kelly, and T. Lesworth, 2012: An assessment of the significance of sulphate sources over the Atlantic Ocean based on sulphur isotope data. Atmos. Environ., 62, 615–621, doi:10.1016/j.atmosenv.2012.08.052.
Lindenmayer, D. B., and G. E. Likens, 2010: Effective Ecological Monitoring. CSIRO Publishing and Earthscan, 170 pp.
Liss, P., 1999: Take the shuttle—From marine algae to atmospheric chemistry. Science, 285, 1217–1218, doi:10.1126/science.285.5431.1217.
Long, M. S., W. C. Keene, D. J. Kieber, D. J. Erickson, and H. Maring, 2011: A sea-state based source function for size- and composition-resolved marine aerosol. Atmos. Chem. Phys., 11, 1203–1216, doi:10.5194/acp-11-1203-2011.
Long, M. S., W. C. Keene, R. Easter, R. Sander, X. Liu, A. Kerkweg, and D. J. Erickson, 2014: Sensitivity of tropospheric chemical composition to halogen-radical chemistry using a fully coupled size-resolved multiphase chemistry/global climate system: Halogen distributions, aerosol composition, and sensitivity of climate-relevant gases. Atmos. Chem. Phys., 14, 3397–3425, doi:10.5194/acp-14-3397-2014.
Macky, W. A., 1957: The rainfall and water supply of Bermuda. Bermuda Meteorological Station Tech. Note 1, 43 pp.
Miller, J. M., J. L. Moody, J. M. Harriss, and A. Gaudry, 1993: A 10-year trajectory flow climatology for Amsterdam Island, 1980–1989. Atmos. Environ., 27A, 1909–1916, doi:10.1016/0960-1686(93)90296-B.
Millet, D., 2012: Natural atmospheric acidity. Nat. Geosci., 5, 8–9, doi:10.1038/ngeo1361.
Milne, P. J., and R. G. Zika, 1993: Amino acid nitrogen in atmospheric aerosols: Occurrence, sources, and photochemical modification. J. Atmos. Chem., 16, 361–398, doi:10.1007/BF01032631.
Mitchell, M. J., and Coauthors, 2011: Comparisons of watershed sulfur budgets in southeast Canada and northeast United States: New approaches and implications. Biogeochemistry, 103, 181–207, doi:10.1007/s10533-010-9455-0.
Moody, J. L., and J. N. Galloway, 1988: Quantifying the relationship between atmospheric transport and the chemical composition of precipitation on Bermuda. Tellus, 40B, 463–479, doi:10.1111/j.1600-0889.1988.tb00117.x.
Moody, J. L., A. A. P. Pszenny, A. Gaudry, W. C. Keene, J. N. Galloway, and G. Polia, 1991: Precipitation composition and its variability in the southern Indian Ocean: Amsterdam Island, 1980–1987. J. Geophys. Res., 96, 20 769–20 786, doi:10.1029/91JD01921.
Moody, J. L., and Coauthors, 2014: Flow climatology for physicochemical properties of dichotomous aerosol over the western North Atlantic Ocean at Bermuda. Atmos. Chem. Phys.,14, 691–717, doi:10.5194/acp-14-691-2014.
Munger, J. W., S.-F. Fan, P. S. Bakwin, M. L. Goulden, A. H. Goldstein, A. S. Colman, and S. C. Wofsy, 1998: Regional budgets of nitrogen oxides from continental sources: Variations of rates for oxidation and deposition with season and distance from source regions. J. Geophys. Res., 103, 8355–8368, doi:10.1029/98JD00168.
NADP, 2014: NADP maps and data. National Acid Deposition Program, accessed 1 December 2014. [Available online at http://nadp.sws.uiuc.edu/data/.]
Neeb, P., F. Sauer, O. Horie, and G. R. Moortgat, 1997: Formation of hydroxymethyl hydroperoxide and formic acid in alkene ozonolysis in the presence of water vapor. Atmos. Environ., 31, 1417–1423, doi:10.1016/S1352-2310(96)00322-6.
Neff, J. C., E. A. Holland, F. J. Dentener, W. H. McDowell, and K. M. Russell, 2002: The origin, composition and rates of organic nitrogen deposition: A missing piece of the nitrogen cycle? Biogeochemistry, 57–58, 99–136, doi:10.1023/A:1015791622742.
Neuman, J. A., and Coauthors, 2006: Reactive nitrogen transport and photochemistry in urban plumes over the North Atlantic Ocean. J. Geophys. Res., 111, D23S54, doi:10.1029/2005JD007010.
Paulot, F., and Coauthors, 2011: Importance of secondary sources in the atmospheric budgets of formic and acetic acids. Atmos. Chem. Phys., 11, 1989–2013, doi:10.5194/acp-11-1989-2011.
Pierce, J. R., and P. J. Adams, 2006: Global evaluation of CCN formation by direct emission of sea salt and growth of ultrafine sea salt. J. Geophys. Res., 111, D06203, doi:10.1029/2005JD006186.
Prinn, R. G., and Coauthors, 2000: A history of chemically and radiatively important gases in air deduced from ALE/GAGE/AGAGE. J. Geophys. Res., 105, 17 751–17 792, doi:10.1029/2000JD900141.
Puxbaum, H., C. Rosenberg, M. Gregoi, C. Lanzerstorfer, E. Ober, and W. Winiwarter, 1988: Atmospheric concentrations of formic and acetic acid and related compounds in eastern and northern Austria. Atmos. Environ., 22, 2841–2850, doi:10.1016/0004-6981(88)90450-7.
Quinn, P. K., and T. S. Bates, 2011: The case against climate regulation via oceanic phytoplankton sulphur emissions. Nature, 480, 51–56, doi:10.1038/nature10580.
Quinn, P. K., T. S. Bates, K. S. Schulz, D. C. Coffman, A. A. Frossard, L. M. Russell, W. C. Keene, and D. J. Kieber, 2014: Contribution of sea surface carbon pool to organic matter enrichment in nascent sea spay aerosol. Nat. Geosci., 7, 228–232, doi:10.1038/NGEO2092.
Rodhe, H., 1999: Human impact on the atmospheric sulfur balance. Tellus,51B, 110–122, doi:10.1034/j.1600-0889.1999.00009.x.
Roelofs, G. J., 2008: A GCM study of organic matter in marine aerosol and its potential contribution to cloud drop activation. Atmos. Chem. Phys., 8, 709–719, doi:10.5194/acp-8-709-2008.
Rohan, P. K., 1986: Climate of Ireland. 2nd ed. The Stationary Office, 146 pp.
Russell, K. M., W. C. Keene, J. R. Maben, J. N. Galloway, and J. L. Moody, 2003: Phase-partitioning and dry deposition of atmospheric nitrogen at the mid-Atlantic U.S. coast. J. Geophys. Res., 108, 4656, doi:10.1029/2003JD003736.
Sanhueza, E., and M. O. Andreae, 1991: Emissions of formic and acetic acids from tropical savanna soils. Geophys. Res. Lett., 18, 1707–1710, doi:10.1029/91GL01565.
Savoie, D. L., J. M. Prospero, and E. S. Saltzman, 1989: Non-sea-salt sulfate and nitrate in trade wind aerosols at Barbados: Evidence for long-range transport. J. Geophys. Res., 94, 5069–5080, doi:10.1029/JD094iD04p05069.
Savoie, D. L., J. M. Prospero, S. J. Oltmans, W. C. Graustein, K. K. Turekian, J. T. Merrill, and H. Levy II, 1992: Sources of nitrate and ozone in the marine boundary layer of the tropical North Atlantic. J. Geophys. Res., 97, 11 575–11 589, doi:10.1029/92JD00894.
Savoie, D. L., R. Arimoto, W. C. Keene, J. M. Prospero, R. A. Duce, and J. N. Galloway, 2002: Marine biogenic and anthropogenic contributions to non-sea-salt sulfate in the marine boundary layer over the North Atlantic Ocean. J. Geophys. Res., 107, 4356, doi:10.1029/2001JD000970.
Shen, G. T., and E. A. Boyle, 1987: Lead in corals: Reconstruction of historical industrial fluxes to the surface ocean. Earth Planet. Sci. Lett., 82, 289–304, doi:10.1016/0012-821X(87)90203-2.
Shepson, P. B., E. Mackay, and K. Muthuramu, 1996: Henry’s Law constants and removal processes for several atmospheric ß-hydroxy alkyl nitrates. Environ. Sci. Technol., 30, 3618–3623, doi:10.1021/es960538y.
Simpson, R. M. C., S. G. Howell, B. W. Blomquist, A. D. Clarke, and B. J. Huebert, 2014: Dimethyl sulfide: Less important than long-range transport as a source of sulfate to the remote tropical Pacific marine boundary layer. J. Geophys. Res. Atmos., 119, 9142–9167, doi:10.1002/2014JD021643.
Sinreich, R., S. Coburn, B. Dix, and R. Volkamer, 2010: Ship-based detection of glyoxal over the remote tropical Pacific Ocean. Atmos. Chem. Phys., 10, 11 359–11 371, doi:10.5194/acp-10-11359-2010.
Smith, A. M., W. C. Keene, J. R. Maben, A. A. P. Pszenny, E. Fischer, and A. Stohl, 2007: Ammonia sources, transport, transformation, and deposition in coastal New England during summer. J. Geophys. Res., 112, D10S08, doi:10.1029/2006JD007574.
Spokes, L. J., S. G. Yeatman, S. E. Cornell, and T. D. Jickells, 2000: Nitrogen deposition to the eastern Atlantic Ocean. The importance of south-easterly flow. Tellus, 52B, 37–49, doi:10.1034/j.1600-0889.2000.00062.x.
Stallard, R. F., and J. M. Edmond, 1981: Geochemistry of the Amazon: 1. Precipitation chemistry and the marine contribution to the dissolved load at the time of peak discharge. J. Geophys. Res., 86, 9844–9858, doi:10.1029/JC086iC10p09844.
Stavrakou, T., and Coauthors, 2012: Satellite evidence for a large source of formic acid from boreal and tropical forests. Nat. Geosci., 5, 26–30, doi:10.1038/ngeo1354.
Swap, R., M. Garstang, S. Greco, R. Talbot, and P. Kallbert, 1992: Saharan dust in the Amazon basin. Tellus, 44B, 133–149, doi:10.1034/j.1600-0889.1992.t01-1-00005.x.
Turekian, V. C., S. A. Macko, and W. C. Keene, 2001: Application of stable sulfur isotopes to differentiate sources of size-resolved particulate sulfate in polluted marine air at Bermuda during spring. Geophys. Res. Lett., 28, 1491–1494, doi:10.1029/2000GL012296.
Turekian, V. C., S. A. Macko, and W. C. Keene, 2003: Concentrations, isotopic compositions, and sources of size-resolved, particulate organic carbon and oxalate in near-surface marine air at Bermuda during spring. J. Geophys. Res., 108, 4157, doi:10.1029/2002JD002053.
Vet, R., and Coauthors, 2014: A global assessment of precipitation chemistry and deposition of sulfur, nitrogen, sea salt, base cations, organic acids, acidity and pH, and phosphorus. Atmos. Environ., 93, 3–100, doi:10.1016/j.atmosenv.2013.10.060.
Vitousek, P. M., 2004: Nutrient Cycling and Limitation: Hawai’i as a Model System. Princeton University Press, 232 pp.
von Glasow, R., M. G. Lawrence, R. Sander, and P. J. Crutzen, 2003: Modeling the chemical effects of ship exhaust in the cloud-free marine boundary layer. Atmos. Chem. Phys., 3, 233–250, doi:10.5194/acp-3-233-2003.
Weathers, K. C., and Coauthors, 1988: Cloudwater chemistry from ten sites in North America. Environ. Sci. Technol., 22, 1018–1026, doi:10.1021/es00174a004.
World Weather Online, 2014: Adrigole monthly climate average, Ireland. Accessed 1 December 2014. [Available online at http://www.worldweatheronline.com/Adrigole-weather-averages/Cork/IE.aspx.]
Young, A., W. C. Keene, A. A. P. Pszenny, R. Sander, J. A. Thornton, T. P. Riedel, and J. R. Maben, 2013: Phase partitioning of soluble trace gases with size-resolved aerosols in near-surface continental air over northern Colorado, USA, during winter. J. Geophys. Res. Atmos., 118, 9414–9427, doi:10.1002/jgrd.50655.
Yu, H., L. A. Remer, M. Chin, H. Bian, Q. Tan, T. Yuan, and Y. Zhang, 2012: Aerosols from overseas rival domestic emissions over North America. Science, 337, 566–569, doi:10.1126/science.1217576.
Zhou, X., and Coauthors, 2008: Photochemical production of hydroxyl radical and hydroperoxides in water extracts of nascent marine aerosols produced by bursting bubbles from Sargasso seawater. Geophys. Res. Lett., 35, L20803, doi:10.1029/2008GL035418.
