We present the first data on the concentration of sea-salt aerosol throughout
most of the depth of the troposphere and over a wide range of latitudes,
which were obtained during the Atmospheric Tomography (ATom) mission.
Sea-salt concentrations in the upper troposphere are very small, usually less
than 10 ng per standard m
Sea-salt particles are the largest aerosol component in the atmosphere by mass (Liao et al., 2006). They represent about 30 % of global column optical depth due to aerosols (Bellouin et al., 2013), a somewhat smaller percentage than for mass because of their relatively large size compared to other aerosols. Global climate models disagree on future changes in sea-salt aerosol due to changes in wind speed and sea ice in a warming climate (Liao et al., 2006; Jones et al., 2007; Boucher et al., 2013). Given the large contribution of sea salt to the global aerosol optical depth, this could represent a significant climate feedback, with uncertainty even between different scenarios in the same model (Hoose et al., 2008). Sulfuric acid, nitric acid, and some other acids can displace halogens in salt particles. This provides both a sink for sulfate and nitrate and a source of reactive chlorine, bromine, and iodine to the atmosphere (Chameides and Stelson, 1992; Finlayson-Pitts and Hemminger, 2000).
There is a large literature on the source of sea-salt aerosol as a function of wind speed (e.g., Gong, 2003; Lewis and Schwartz, 2004; Grythe et al., 2014). There has been increased recognition of the importance of submicron salt particles to aerosol number (Kreidenweis et al., 1998; Clarke et al., 2003, 2006). These submicron sea-salt particles are enriched in organics compared to sea water, although the amount of enrichment is not consistent and may vary under differing conditions (Middlebrook et al., 1998; Modini et al., 2010; Vignati et al., 2010; Ovadnevaite et al., 2011; Gantt and Meskhidze, 2013).
Almost all of the sea-salt aerosol literature considers measurements within the marine boundary layer and even there consists mostly of surface measurements. There have been few measurements of how sea salt varies with altitude. Shinozuka et al. (2004) presented profiles of nonvolatile aerosol, presumed to be sea salt, up to about 2 km altitude for one region in the Southern Ocean and one region in the tropical Pacific.
We present here the first measurements of the concentration of sea-salt aerosol over a wide range of altitudes and latitudes. We consider the sea-salt vertical transport, wet removal, and compositional variability. These data provide strong constraints on aerosol transport efficiency out of the marine boundary layer and are a useful tool in evaluating aerosol removal in large-scale models.
We quantify sea-salt aerosol by merging measured size distributions with the fraction of particles in each size range identified as sea salt by single particle mass spectrometry. Sea-salt particles were identified from mass spectra of single aerosol particles using the Particle Analysis by Laser Mass Spectrometry (PALMS) instrument (Thomson et al., 2000). Particles enter a vacuum and cross a split continuous laser beam. The transit time between the beams provides the particle velocity, used to determine its aerodynamic diameter. The aerosol inlet to PALMS is controlled to about 35 mbar, with a small dependence on outside pressure because the pressure transducer used for control was not positioned to capture the full effect of a jet downstream of the first critical orifice. Transit times were calibrated to known particle sizes at laboratory pressure (about 820 mbar) before and after every field deployment. An excimer laser is triggered when a particle arrives at the second laser beam and ions are produced when the 193 nm pulse hits the particle. Either positive or negative ions are analyzed with a time-of-flight mass spectrometer, with the polarity switched every few minutes. For these data, about 60 % of the time was spent acquiring positive ion spectra.
The PALMS instrument was flown on the NASA DC-8 during the Atmospheric
Tomography Mission (ATom) as well as earlier missions. The ATom mission
consisted of several series of flights between about 85
Particle size distributions were measured by an ultra-high-sensitivity
aerosol spectrometer (UHSAS; Kupc et al., 2018) and a laser aerosol
spectrometer (LAS, TSI Incorporated). In the merged size distribution, UHSAS
data were used for particles less than 0.51
The PALMS instrument has substantial biases in sampling efficiency for
different size particles, with higher efficiency for particles about 1 to
3
PALMS, the LAS, and the UHSAS all sampled from a University of Hawaii inlet
owned by NASA Langley on the DC-8 (McNaughton et al., 2007). About 1.5 m
of 0.25 in. outside diameter stainless steel tubing with a usual volumetric
flow rate of 3.5 to 7 L min
The University of Hawaii inlet on the DC-8 has been shown to quantitatively transmit
particles as large as 3.1
In-cloud data are excluded from the results shown here. Huebert et al. (1990) showed that up to 90 % of the largest sea-salt particles in the marine boundary layer can deposit to the walls of an inlet. Cloud droplets or ice crystals impacting a forward-facing aircraft inlet can act like a high-pressure washer to dislodge some of that salt, potentially leading to large sea-salt artifacts in clouds. During both ATom and previous missions (Murphy et al., 2004), PALMS observed anomalous particles in clouds, both sea salt and other compositions, reinforcing the decision to exclude in-cloud data.
Mass spectra of
Sea-salt particles were identified in the positive ion mass spectra using a
combination of peaks, starting with a large
The main concern in identification is that at extremely low concentrations
of sea-salt aerosol, as found over land or at high altitude, there may be a
contribution from other particles, particularly dust from salt flats that is
chemically similar to oceanic sea salt. A manual review was made of
Sea-salt particles can also be identified from negative ion spectra using chloride ions and cluster ions containing Na. Data in this paper are from positive ion spectra because in aged sea-salt particles the chloride can be almost entirely displaced by sulfate and nitrate, making identification more difficult with negative ions. In regions with fresh sea salt, results from negative ion spectra were nearly identical to those from positive ion spectra.
The number and mass of sea-salt particles were calculated for every 5 min of flight time in order to acquire enough mass spectra for a statistically significant normalization to the UHSAS and LAS size distributions. During climbs and descents, 5 min represents about 2.5 km in altitude. Mass spectra of sea-salt particles were typically acquired at a rate of more than one per second in the marine boundary layer and less than one per minute at high altitude.
Figure 2 shows concentrations of sea-salt aerosol at low altitudes measured
by the PALMS and LAS combination compared to filter measurements of sodium (Dibb
et al., 1999). The filter samples indicate more sea-salt mass, which is
expected because the inlet to the filter sampler transmitted larger
particles than the inlet to PALMS. The good correlation adds confidence to
the PALMS measurements in the upper troposphere, which are much more
sensitive than the filter sampler. Because PALMS samples each particle at a
particular time, its composition can be associated with a particular
altitude, aerosol concentration, water vapor concentration, and so forth.
This allows for very long averaging times in similar air. For example, the
average concentration of sea-salt aerosol in air with 10 to 20 ppmv of water
can be calculated from collecting many such stretches of flight data even
though they are not contiguous in time and might even be on different days.
Furthermore, the PALMS single particle data are easier to screen for short
periods of cloud and other artifacts than the extended filter samples are.
One can eliminate short cloud encounters without losing the data before and
after the cloud. The internal consistency of the data suggests that the
detection limit for sea salt is better than 10 ng m
Sea-salt aerosol mass at low altitudes derived from PALMS compared to filter samples of aerosol sodium. The inlet for the filter sampler transmits larger particles than the PALMS inlet; hence, PALMS is expected to sample somewhat less mass. The 1 : 1 line is displaced because not all sea-salt mass is sodium. The cutoff at 85 % relative humidity is imposed because comparing the different inlet cut points becomes especially problematic when the particles are enlarged due to water uptake at high humidity.
The concentration of sea-salt aerosol in the marine boundary layer measured
by PALMS was usually between 0.3 and 3
Concentration of sea-salt aerosol during the ATom1 and ATom2 flights
over the Arctic, Pacific, and Southern oceans. The blank region between about
60 to 70
There are already extensive measurements of sea-salt aerosol in the marine
boundary layer (Lewis and Schwartz, 2004). The novel data here are the
concentrations at higher altitudes. Figure 3 shows a latitude–altitude
cross section of sea-salt aerosol concentrations over the Pacific Ocean.
Even though they were in different seasons, systematic differences between
ATom1 and ATom2 are not visible on a log color scale so they are combined in
Fig. 3. More subtle differences depending on season are discussed below. The
Pacific Ocean is shown because there was less mineral dust to complicate the
analysis of very low concentrations; concentrations over the Atlantic Ocean
were similar. Shinozuka et al. (2004) inferred sea-salt aerosol
concentrations from nonvolatile particles. The results shown here agree
that this was valid for their measurements at
A salient property of the distribution of sea-salt aerosol in Fig. 3 is
that the concentration falls off rapidly with altitude, by about a factor of
10 for every 2 km. Above 6 km, the concentrations were almost always less
than 10 ng m
Figure 4 shows size distributions of sea-salt particles from PALMS during
ATom1. Most of the sea-salt mass is in particles larger than 1
Size distributions of sea-salt particles in the marine boundary
layer. Panel
Sea-salt aerosols were not confined to areas with open water. Significant
concentrations of sea-salt aerosol were also observed over ice-covered
regions of the Arctic Ocean during ATom. For a portion of a flight north of
Alaska on 19 February 2017, the nearest large areas of open water were about
1000 km away. Yet significant concentrations of sea-salt aerosol
(
Mass spectra of sea-salt particles over the ice-covered Arctic Ocean were depleted in Na relative to Mg, K, and Ca compared to particles at lower latitudes (Fig. 5). The Na depletion over the Arctic was due to an increased population of particles with low Na rather than every particle having less Na. It is robust in the following sense: we used the NH Pacific as a reference case for ATom2 because that is the region with the most data in the marine boundary layer. We then compared data from other ocean regions with the NH Pacific. The magnitude of the difference in the average Na signal between the Arctic Ocean and the NH Pacific was much larger than the differences between the NH Pacific and any of the other open ocean regions, showing that the Arctic sea-salt aerosol is distinct from open ocean sea-salt aerosol.
Average spectra of sea-salt particles larger than 1
Similar Na depletion was also observed during ARCPAC when comparing mass
spectra of particles over Arctic Ocean compared to test flights over the
Gulf of Mexico. In contrast, significant Na depletion over the Arctic Ocean
was not observed during ATom1, which was flown during August when much of
the Arctic Ocean had some open water. Unfortunately, occasional detector
saturation by the
Na depletion in polar sea-salt aerosol is consistent with Wagenbach et al. (1998) and Hara et al. (2012), who found
Simple altitude profiles of PALMS sea-salt aerosol in various latitude bands will be presented in model–measurement comparison papers (Yu et al., 2019; Bian et al., 2019; Zhang et al., 2019). An alternative way of presenting the concentration of sea-salt aerosol is as a correlation with water vapor (Fig. 6). Sea salt is water soluble, so one might expect that its removal would approximately scale with removal of water via precipitation. Figure 6 shows that this is the case, at least when considered as an average over many vertical profiles. Because the concentration of water vapor falls off rapidly with altitude, the correlations in Fig. 6 are in a sense vertical profiles. However, the concentration of sea-salt aerosol is often better correlated with water vapor than with altitude itself.
Measured average sea-salt mass concentrations as a function of water
vapor. Latitude bands for the Northern and Southern hemispheres are 20 to
65
The log–log slopes are not far from one, indicating that sea-salt aerosol is
removed with water: when 90 % of the water is removed about 90 % of
the sea-salt aerosol is removed. Labels indicate water mixing ratios beyond
which most clouds are ice or most are liquid water, based on saturation vapor
pressures of
In many ways Fig. 6 shows a remarkably simple picture of sea-salt aerosol concentrations by season and hemisphere. Since the ATom1 and ATom2 deployments were roughly 6 months apart and covered both hemispheres, it is possible to distinguish seasonal and hemispheric differences. Concentrations in the tropical atmosphere show little seasonal dependence. The two summer hemispheres are fairly similar to the tropics. The two winter curves are shifted up and to the left. Especially in the Northern Hemisphere, the sea-salt concentration near the ocean surface (at the top right of each curve) is not all that different in winter and summer. Instead, a similar amount of sea-salt aerosol is emitted into a lower absolute humidity in the colder winter air. This suggests two reasons more sea-salt aerosol can reach the upper troposphere in winter than in summer. The main reason is that more sea-salt particles can survive into the upper troposphere in winter simply because in winter there is less water available to wash out the aerosol. Second, removal of sea-salt aerosol in ice clouds may be less efficient than in liquid water clouds.
Model results for the correlation between sea-salt aerosol and water vapor are shown in Fig. 7. For simplicity only a subset of the regions in Fig. 6 are shown. The Community Earth System Model with the Continuous-time Autoregressive Moving Average (CESM-CARMA) couples a sectional aerosol model (Yu et al., 2015; Toon et al., 1988) with the National Science Foundation and Department of Energy CESM. CARMA uses 20 size bins for sea spray aerosols which are composed of salt, marine sulfate, and marine organics. The Goddard Earth Observing System Model version 5 (GEOS-5) simulates meteorological fields to drive the online Goddard Chemistry Aerosol Radiation and Transport (GOCART) aerosol model (Colarco et al., 2010; Bian et al., 2013). GOCART sea-salt aerosol is emitted using an upgraded emission algorithm (Gong, 2003; Bian et al., 2019) and removed by warm cloud from convective updraft and large-scale rainout and washout, as well as by dry deposition and sedimentation (Chin et al., 2002). A humidified sea-salt particle size (Gerber, 1985) is used for computations of particle fall velocity, deposition velocity, and optical parameters. The detailed description of the GEOS-5–GOCART sea-salt aerosol simulation for this work is given in Bian et al. (2019).
Model correlations for the CESM-CARMA and GEOS-5 models. ATom2 and tropical curves are omitted for simplicity. GEOS-5 model output was sampled along the flight tracks and CESM-CARMA at all altitudes in a curtain along the flight tracks. Also shown is one curve from the CESM-CARMA model before a revised convective aerosol removal scheme was implemented (Yu et al., 2019).
Both models reproduce the strong correlation between sea-salt aerosol and
water vapor. Both models also capture the difference between the winter and
summer hemispheres in the correlation. The GEOS-5 model may be removing
aerosol too slowly in ice clouds. The comparison to these data
uncovered an error in aerosol removal in the CESM model in which sea-salt
aerosol was originally overestimated by a factor of over 100 in the upper
troposphere. One example is shown in Fig. 7. The overestimate was traced
to aerosols not being properly removed from air transported in the sub-grid
convective parameterization. A detailed analysis and new removal
parameterization are described by Yu et al. (2019). An interaction between
removal and entrainment parameterizations was also identified as an issue in
the Community Atmosphere Model version 5 (CAM5) by Wang et al. (2013), based on black carbon data in the upper
troposphere. The CARMA bin microphysics also reproduces the similar shape of
the size distribution of sea-salt aerosol at different altitudes (Fig. 4b). In the model, only particles larger than about 3
One consequence of the small concentrations of sea-salt aerosol in the upper troposphere is that it contributes little to chemical reactivity. In particular, Wang et al. (2015) postulated an important role for upward transport of sea-salt aerosol followed by release of bromine into the upper troposphere. These data show that de-bromination of sea salt cannot be a significant source in the upper troposphere. There was almost always less than 10 ppt of sea-salt aerosol by mass and often less than 1 ppt. Given that sea salt is very roughly 0.1 % by mole bromine, there would be parts per quadrillion of bromine available from transported sea-salt aerosol. If bromine from sea-salt aerosol is to significantly affect the upper troposphere, it would have to be released at low altitude and transported in the gas phase, although it is not clear if there are any suitable gas-phase bromine compounds that would survive wet scavenging.
The small concentrations of sea-salt aerosol in the upper troposphere also
provide a strong constraint on the influence of salt on the
gas-phase reactive-nitrogen budget. Even complete replacement at altitude of sea-salt chlorine by
nitrate would be a very small sink for nitrate compared to the hundreds of
pptv of
These are the first measurements of sea-salt aerosol over a wide range of
altitudes and latitudes. Data are available from near the surface to about
12 km altitude from about 65
Sea-salt aerosol's only source is at the surface and its only sink is by scavenging (i.e., sea-salt particles do not evaporate). That makes sea-salt aerosol a powerful tool to study wet removal of aerosol. The data here indicate that removal of sea-salt aerosol is very approximately proportional to the removal of water over a wide range of absolute humidity, with possibly more efficient removal in liquid water clouds than in ice clouds.
Data are publicly available at
KDF, GPS, and DMM provided PALMS data. CAB, AK, MD, BW, and CJW provided size distribution data. JED and EMS provided filter data. JPD and GD provided water vapor data. HB and PY provided model results. DMM wrote the paper with assistance from all authors.
The authors declare that they have no conflict of interest.
The participation of PALMS in the Atmospheric Tomography Mission flights (ATom1 and ATom2) was supported by NOAA climate funding. The mission as a whole was supported by NASA's Earth System Science Pathfinder Program EVS-2 funding. Charles A. Brock and Christina J. Williamson were supported by award NNH15AB12I and by NOAA's Health of the Atmosphere and Atmospheric Chemistry, Carbon Cycle, and Climate Programs. Agnieszka Kupc was supported by the Austrian Science Fund's Erwin Schrodinger Fellowship J-3613. The CESM project is supported by the National Science Foundation and the Office of Science (BER) of the US Department of Energy. Bernadett Weinzierl and Maximillian Dollner have received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program under grant agreement no. 640458 (A-LIFE) and from the University of Vienna. We thank the ATom team and crews for making the flights possible, and Bruce Anderson of NASA Langley for the use of the University of Hawaii inlet.
This paper was edited by Yafang Cheng and reviewed by two anonymous referees.