Interactions of atmospheric aerosols with clouds and precipitation are a vital, yet uncertain, part of the climate system. Indirect aerosol radiative forcing effects are associated with large uncertainties in global climate models (Boucher et al., 2013), and they are poorly constrained, with a factor of 5 variation across different models (Quaas et al., 2009). An effect that is especially relevant to these clouds is the first indirect or Twomey effect (Twomey, 1974), which accounts for changes to cloud albedo resulting from perturbations in aerosol concentration and therefore the cloud condensation nuclei (CCN) availability. Remote marine clouds are particularly susceptible to perturbations in aerosol concentrations because they are relatively optically thin, and the background aerosol concentrations in the remote MBL are low (Bony and Dufresne, 2005; Carslaw et al., 2013; Quaas et al., 2009; Wood, 2005). However, by itself, the Twomey effect is insufficient in explaining all indirect effects (Wood, 2007). Factors such as aerosol suppression of precipitation can also be important (Wood, 2005; Wood et al., 2015), highlighting the need for measurement of aerosol and cloud properties in the remote MBL.

Sources of CCN in the MBL have been at the center of much research for several decades. Early studies have hypothesized that dimethyl sulfide (DMS), a volatile organic compound (VOC) produced by phytoplankton, is a major source of secondary sulfate aerosols, which then dominate the marine CCN budget (Charlson et al., 1987). This came to be known as the CLAW hypothesis, which posits a feedback loop between ocean biochemistry, marine cloud properties and climate (Charlson et al., 1987). More recently, it has been shown that there are factors, such as additional sea salt sources of CCN in the MBL or complex interactions between CCN and aerosol–cloud interactions, that prevent DMS-derived sulfate from being directly involved in a climate feedback mechanism (Quinn and Bates, 2011). Sea salt sources of CCN that compete with DMS-derived sulfate include primary organic species emitted together with sea salt upon bubble bursting, which is a mixture of fragments of marine biota, their exudates, and other simple proteins, lipids, and carbohydrates (Quinn and Bates, 2011). However, it has also been shown that sea salt contributes only 30 % to the CCN budget in midlatitudes, and secondary, possibly DMS derived, non-sea-salt sulfate can dominate most MBL CCN budgets (Quinn et al., 2017). This secondary source of CCN has been shown to be important in the ENA region (Zheng et al., 2018). Because the emissions of marine VOCs and chemistry of both primary and secondary marine organic aerosol are controlled by ocean ecosystems, they are likely impacted by climate change. Sea surface temperatures drive marine phytoplankton diversity, and the distributions of phytoplankton species in the warming ocean are likely to shift in the near future (Flombaum et al., 2013; Righetti et al., 2019). A recent modeling study by Wang et al. (2018) found significant radiative effects from such broad community shifts. Understanding complex relationships between ocean diversity, aerosol chemistry and MBL clouds is of critical importance in the changing climate.

Apart from local oceanic sources of aerosols, such as DMS oxidation and ejection of sea salt, the remote MBL can also be impacted by long-range continental transport. Biomass burning plumes can be effectively transported between the continents (Brocchi et al., 2018), as the aerosols are injected into the free troposphere or the stratosphere. Secondary aerosols of continental origins, such as secondary organic aerosol (SOA) produced by oxidation of VOCs, can also potentially reach the remote atmosphere. Recent studies have shown that SOA may be more resistant to evaporation than previously thought, increasing the SOA lifetime and the distances it can be transported (Shrivastava et al., 2013, 2015; Vaden et al., 2010, 2011; Zelenyuk et al., 2012). Furthermore, recent research has shown that while the lifetime of most biogenic SOA is shorter than their mechanical removal timescales, there exists a fraction of non-photolabile SOA that could potentially have long lifetimes and be effectively transported into the remote regions (O'Brien and Kroll, 2019; Zawadowicz et al., 2020). In summary, the MBL CCN budget is composed of primary ocean emissions (sea salt and primary organics), secondary aerosols derived from ocean biogenic VOCs and aerosols transported remotely from the continents. The specific local variations of those sources, their seasonality, and the exact mixing states of MBL aerosols are difficult to predict due to a lack of detailed chemical measurements of aerosol in remote marine regions yet crucial for understanding cloud properties and radiative forcing in these areas.

This study focuses on chemical measurements of non-refractory aerosol and trace gas composition vertical profiles, from the Aerosol and Cloud Experiment in the Eastern North Atlantic (ACE-ENA), a US Department of Energy (DOE) airborne measurement campaign. The site of these measurements is the Azores archipelago in the eastern North Atlantic (Fig. 1), which is uniquely suited for characterization of both marine and long-range-transported aerosol properties and various meteorological conditions favorable to both of these CCN composition regimes. Graciosa Island in the Azores is also the location of a permanent DOE Atmospheric Radiation Measurement (ARM) user facility measurement site. Because the Azores straddle the boundary between subtropics and midlatitudes, they experience a wide range of meteorological conditions throughout the year (Wood et al., 2015). There is a marked seasonality in the wind patterns near the Graciosa site: in the winter, there is a strong gradient of surface pressure between the Icelandic low and the Azores high-pressure systems, and the winds tend to have high average speeds and come from the southwest (Wood et al., 2015). In the summer, the Icelandic low disappears and the Azores high-pressure system strengthens, which is associated with winds predominantly from the northwest with lower average wind speeds (Wood et al., 2015). The high wind speeds are responsible for higher sea salt contributions to the ENA CCN budget in the winter (Zheng et al., 2018). This pattern is also responsible for the seasonal peak in total cloud fraction, which occurs the winter (Wood et al., 2015). This dynamic complexity makes it difficult to attribute sources of aerosols in the ENA region because the MBL air masses are continually diluted with free-tropospheric air on a timescale of several days (Wood et al., 2015). As observed during the Clouds, Aerosol, and Precipitation in the Marine Boundary Layer (CAP-MBL) campaign, marine air masses can have continental features due to the entrainment from the free troposphere (Wood et al., 2015). Previous measurements on Pico Island also indicate a seasonal summer peak in carbon monoxide due to long-range transport from North America (Val Martin et al., 2008). Additionally, single-scattering albedo measurements during CAP-MBL indicate that the aerosols are more absorbing during springtime, also consistent with transport of North American biomass burning aerosols (Logan et al., 2014).

The atmospheric chemistry of the North Atlantic region, including the Azores, has been studied in previous oceanographic cruises (Andreae et al., 2003) and aircraft campaigns. Early work on sulfur cycle in the eastern North Atlantic has been summarized by Galloway et al. (1992). Notable campaigns in the region include the Atlantic Stratocumulus Transition Experiment/Marine Aerosol and Gas Exchange (ASTEX/MAGE) (Blomquist et al., 1996) and the second Aerosol Characterization Experiment (ACE-2) (Raes et al., 2000). Of note are also studies on the west coast of Ireland at the Mace Head observatory, which frequently encounter North Atlantic air masses, as described in Dall'Osto et al. (2010) and Ovadnevaite et al. (2014).

More recent field measurements focusing on the aerosol chemistry of the North Atlantic region include the NASA North Atlantic Aerosols and Marine Ecosystems Study (NAAMES), which included both aircraft and shipborne observations focused on marine biological productivity of the western North Atlantic (Behrenfeld et al., 2019), and the African biomass-burning-focused NASA Observations of Aerosols above Clouds and their Interactions (ORACLES), which included aircraft deployments of a suite of instruments, including the aerosol mass spectrometer (AMS), in the southeastern Atlantic. The AMS has a long field deployment history on a variety of platforms (Zhang et al., 2007), but particle composition measurements in remote marine environments are of recent interest. Between 2011 and 2012, an AMS was deployed aboard the German research vessel Polarstern during four cruises in the Atlantic Ocean (Huang et al., 2017, 2018). The average aerosol composition was found to be ∼50 % sulfate and ∼20 % organic (Huang et al., 2018). The organic component was further analyzed with positive matrix factorization, and it was determined to be a mixture of primary and secondary (DMS- and amine-derived) marine aerosol and aerosol transported from North America (Huang et al., 2018). Between 2016 and 2017, AMS was also deployed aboard the NASA DC-8 aircraft during the Atmospheric Tomography (ATom) missions in the remote atmosphere, including the North Atlantic region (Hodshire et al., 2019). In 2015, AMS was also deployed on aircraft as a part of the Network on Climate and Aerosols: Addressing Key Uncertainties in Remote Canadian Environments (NETCARE) project in the Canadian high Arctic (Abbatt et al., 2019; Willis et al., 2019). A unique feature of the ACE-ENA aircraft deployments is the seasonally resolved measurements during summer and winter in the eastern North Atlantic region.

In this study, we focus on chemical characterization of aerosol and trace gases during ACE-ENA. We present average concentrations of sulfate, total organics, ammonium, and nitrate aerosol components and mixing ratios of trace gases such as methanol, acetone, DMS, isoprene, toluene and benzene. In addition, concentrations of particle-phase methanesulfonic acid (MSA), an oxidation product of DMS, were derived using aircraft measurements and laboratory calibrations. We also discuss vertical profiles of these quantities in context of other G-1 measurements, meteorology and Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT) (Stein et al., 2015) back trajectories to elucidate the sources of ENA aerosols.

The ACE-ENA campaign was conducted around the Azores archipelago in the eastern North Atlantic from 1 June 2017 to 28 February 2018. During this period, the DOE G-1 research aircraft, based out of Lajes Field on Terceira Island (Fig. 1), was deployed for two intensive operation periods (IOPs), consisting of flights around the Graciosa Island ENA ARM research site. The first aircraft deployment period occurred in the summer, from 1 June to 31 July 2017, and the second occurred in the winter, from 1 January 2018 to 28 February 2019. During the summertime deployment period, the aircraft completed 20 research flights, and during the winter, it completed 19 research flights (Table S1 in the Supplement). Figure S1 in the Supplement shows flight tracks for all research flights. In general, the flight plans were focused on multiple L-shaped transits at different altitudes with the ENA site as the focal point, with occasional excursions to other cloud layers. Each research flight also typically included at least one spiral profile through the atmosphere to characterize the boundary layer structure. The locations of spiral profiles are shown in Fig. S1c and d and included in Table S1. This study uses the vertical profiles to explore the marine and continental influences on aerosol composition.

Examination of average concentrations of trace gases and aerosols, their vertical profiles, and external factors such as meteorology, back trajectories and biological productivity reveals an interplay of local and transported emissions giving rise to CCN in the ENA region. Two tracers of biological ocean productivity were measured during ACE-ENA: DMS and particulate MSA. Both of these tracers are prominent in the marine boundary layer but generally only in the summer. The synchronization in concentrations of DMS and MSA is expected, as MSA is produced through atmospheric oxidation of DMS. In the marine boundary layer, DMS is a by-product of metabolism of some phytoplankton species, such as coccolithophores and dinoflagellates (Keller et al., 1989). Because its production is biogenic in nature, it undergoes seasonal cycles, which tend to be strongly latitude dependent (Galí and Simó, 2015; Polimene et al., 2012). At polar and subpolar latitudes, which are light-limited in the winter, concentrations of ocean DMS peak synchronously with phytoplankton biomass, but at subtropical latitudes, peak DMS emissions lag behind peak biomass by as much as a few months (Galí and Simó, 2015; Lana et al., 2012; Polimene et al., 2012). The cause of this lag is the way seasonality in solar irradiation drives diversity of phytoplankton taxa, with the DMS-producing species preferring more irradiated conditions (Galí and Simó, 2015; Polimene et al., 2012). Figure S5 shows distributions of ocean chlorophyll a measured by MODIS aboard the Aqua satellite (NASA Goddard Space Flight Center, 2018) during the two intensive periods during ACE-ENA. While the summer coincides with greater biological productivity at the subpolar latitudes north of the Azores, the chlorophyll a concentrations at the ENA site are higher in the winter (0.27 mg m-3) than in the summer (0.12 mg m-3), which is an opposite trend to the aircraft DMS and MSA measurements. This could be due to the lag in seasonal production of DMS in the Azores, which straddle midlatitude and subtropical regimes. The anti-correlation between phytoplankton biomass and ocean DMS concentrations has been shown to persist in the -20 to 40∘ latitude band in both hemispheres, which includes the Azores (Lana et al., 2012). The seasonal surface DMS concentration has been shown to peak in the summer in the eastern North Atlantic using a global model (Kloster et al., 2006). Previous measurements of surface seawater DMS concentrations compiled in the Global Surface Seawater DMS Database (https://saga.pmel.noaa.gov/dms/, last access: 2 May 2020) (Kettle et al., 1999) also indicate higher concentration around the Azores in the summer (mean 2±2 nM L-1) than in the winter (0.5±0.4 nM L-1) (Fig. S6). An alternative explanation is transport of DMS-rich air masses north of the site, but the HYSPLIT (Stein et al., 2015) back trajectories shown in Fig. S7 do not suggest strong transport from the north at low altitudes where DMS signal is most prominent.

Local production of MSA and sulfate from oceanic emissions is a major source of aerosol mass in the summer. Sulfate is highly hygroscopic, and therefore this aerosol will be efficient as CCN, which explains higher on average CCN concentrations close to the ocean surface in the summer (Fig. 13). There is evidence from analysis of ENA site data that accumulation-mode particles entrained from the free troposphere, as well as growth of Aitken-mode particles into accumulation mode in the MBL, are the largest contributors to the CCN budget in the ENA region (Zheng et al., 2018). In particular, the latter source can represent as much as 60 % of the CCN budget in the summer (Zheng et al., 2018). Aitken-mode particles, in this case, are entrained from the free troposphere, and they continue to grow by condensation of DMS oxidation products, such as MSA and sulfuric acid, into CCN-relevant sulfate particles (Zheng et al., 2018). Annual mean contributions of sea salt to accumulation and Aitken modes have been found to be 21 % and less than 10 %, respectively (Zheng et al., 2018), which suggests that AMS is sensitive to the majority of the ENA CCN budget composition.

Vertical profiles presented in Figs. 8–10 show strong stratification in aerosol chemistry. Below ∼1000 m, the particles are MSA-influenced and strongly sulfate dominated and acidic; at 1000–3000 m, the particles contain no MSA, are more neutralized (though still acidic), and can contain organics, ammonium and black carbon. HYSPLIT back trajectories in Fig. S7 corroborate the low-altitude air masses as strongly influenced by marine conditions. The high-altitude features are also consistent with long-rage transport of North American continental emissions. Research flight no. 19 on 19 July 2017 shows the strongest continental transport influence at altitudes between 1000 and 2500 m. The peak in aerosol abundance, organic and black carbon concentrations, and higher pH (Fig. 6) is accompanied by a peak in CO, methane and acetone concentrations (Figs. 11 and 12). Those characteristics suggest a biomass burning plume, but the lack of an AMS m/z 60 levoglucosan marker points to highly processed biomass burning aerosol or anthropogenic emissions. In order to constrain the source of the emissions, HYSPLIT back-trajectory analysis of this flight was carried out as shown in Fig. 14. An ensemble of 14 d back trajectories was generated from the site of the first vertical profile (Table S1) at 2000 m altitude. The back trajectories reach >500 m altitude over the east of the United States on 12–15 July. The Fire Inventory from NCAR (FINNv1) (Wiedinmyer et al., 2011) was then used to provide locations of major fires on these 4 d, as shown in Fig. S8. Two fire clusters on those days occur in southwestern Canada and the southeastern United States. The southeastern US location in South Carolina, where FINN predicts the highest biomass burning emissions, was used to start 10 d forward trajectories in HYSPLIT in Fig. 14. Those reach the Azores at the correct time and altitude, which suggest that southeastern US could be the location for biomass burning emissions seen in the ENA region on 19 July 2017. A similar back-trajectory analysis was carried out for the southwestern Canadian fire location (Fig. S9), and those trajectories do not appear to reach the correct location. However, uncertainties in trajectory analysis and factors such as rapid mixing of free-tropospheric and MBL air (Wood et al., 2015) do not allow ruling it out as a source completely. Anthropogenic emissions such as traffic also cannot be ruled out completely as a source of 19 July elevated aerosol concentrations, other polluted layers at ENA, or the highly oxidized background organic aerosol. Such polluted layers were observed relatively frequently during the campaign, and they were often accompanied by large increases in CCN concentrations (Fig. 13). This shows that periodic long-range transport of pollution into the ENA region can significantly perturb the local CCN budget.

Long-range-transported aerosols (organic, ammonium, rBC) and reactive gases (methanol, acetone) at ENA show strong seasonality with high concentrations in the summer and low in the winter. This can be partially attributed to the seasonality of North American wildfires, which occur in the summer and winter. However, Fig. 12 suggests an opposite trend for CO, a non-reactive, water-insoluble gas, which on average shows higher concentrations in the winter. Analysis of 3 years of data at the ENA ARM site suggests a peak of CO and ozone concentrations occurring in the spring (March) and a minimum during the summer (July) (Zheng et al., 2018). This trend coincides with the CO and ozone seasonality seen during ACE-ENA. One explanation is that CO undergoes more photolytic destruction in the summer due to higher OH concentrations. Another is that continental transport into the Azores might actually be more prominent during the winter, but the particles and reactive gases are maybe locally scavenged by clouds. Figure S5 shows MODIS Aqua observations of cloud fraction over the North Atlantic during the two intensive observation periods during the ACE-ENA campaign. The mean cloud fraction over the Azores is 0.56 and 0.77 during summer and fall, respectively. Additionally, Fig. S10 shows that on average the winds were stronger in the winter, and the HYSPLIT back trajectories in Fig. S7 show more long-range transport during the winter at all altitudes. The seasonal trend in CO can also be attributed to local emissions to some extent, as higher emissions in the winter coincide with residential heating and power use. The EDGAR-HTAP V2 (https://edgar.jrc.ec.europa.eu/htap_v2/, last access: 4 May 2020) gridded emissions inventory (Janssens-Maenhout et al., 2015) was used to estimate local CO emissions in the Azores (Fig. S11), showing increased emissions in the power and residential sectors. However, local Azores anthropogenic emissions are still rather low, and the trends in local emissions and transported pollution are likely superimposed.

During the ACE-ENA campaign, 39 research flights in the vicinity of the Graciosa Island in the Azores were carried out during two seasons, summer 2017 and winter 2018. Aerosol and trace gas chemistry were characterized using the AMS and PTR-MS, as well as a suite of other sensors. AMS measurements reveal clean conditions with <1 µg m-3 non-refractory aerosol abundance and strong seasonality, where the summer aerosol concentrations were ∼4 times higher than in the winter. Trace reactive gas concentrations were often <1 ppb throughout the campaign, with similar seasonality. Particles sampled in the MBL showed contributions from MSA, which is a tracer for ocean biogenic activity. The summertime clean MBL at ENA is characterized by secondary sulfuric acid aerosol produced through DMS oxidation and some highly oxidized organic aerosol, which could be entrained from the free troposphere or ejected from the ocean. Secondary sulfate aerosol is also hygroscopic, which leads to a high baseline concentration of CCN above the ocean (100 cm-3 for 0.1 % supersaturation and 200 cm-3 for 0.3 % supersaturation). In the winter, less DMS is emitted due to seasonal shifts in ocean productivity, and consequently aerosol and CCN concentrations in the boundary layer also decrease (∼50 cm-3 for both supersaturations sampled).

Examination of vertical profiles and HYSPLIT back trajectories revealed numerous remote transport events that influenced the aerosol sampled in the free troposphere, above 1000 m. Both anthropogenic pollution and North American wildfires are a likely sources, with especially high emissions observed on 19 July 2017. Such long-range-transport events were accompanied by large plumes of CCN in the free troposphere (Fig. 13). The 19 July 2017 event was in particular traced to possible fires in the North American southeastern US region, and it was associated with a large plume of CCN (200–300 cm-3 for 0.1 % supersaturation and 300–400 cm-3 for 0.3 % supersaturation). This work shows that the CCN budget in the MBL in the eastern North Atlantic region is a result of complex interactions between long-rage transport of North American wildfire and anthropogenic emissions and the local marine biogenic sources.