Nitrous oxide variability at sub-kilometre resolution in the Atlantic sector of the Southern Ocean

Introduction

Nitrous oxide (N₂O) is a strong greenhouse gas and currently the third largest contributor to radiative forcing after carbon dioxide (CO₂) and methane (CH₄) (Hartmann et al., 2013). Furthermore, N₂O is an important source for stratospheric NO_x, which is involved in catalytic ozone depletion (Crutzen, 1970; Ravishankara, Daniel & Portmann, 2009). The oceans, including coastal zones, estuaries and rivers, are estimated to contribute approximately 25% to global N₂O emissions (Ciais et al., 2013). The Southern Ocean alone is estimated to account for 5% of global emissions (0.9 Tg a⁻¹ N (nitrogen equivalents), (Nevison et al., 2005)). However, measurements of oceanic N₂O concentrations in this region are scarce and these emission estimates are based on atmospheric measurements at Cape Grim and a few seawater measurements in the Southern Ocean from 1977–1993 (Nevison, Weiss & Erickson, 1995; Nevison et al., 2005 and references therein).

In the ocean N₂O is produced as a byproduct during nitrification and an intermediate during nitrifier-denitrification in oxygenated waters, and as an intermediate during denitrification in anoxic waters. Nitrification is a two-step process during which ammonia oxidation is followed by nitrite oxidation. The first step is carried out by ammonia oxidising bacteria (AOB) and archaea (AOA). N₂O is a by-product of this step that can be produced during hydroxylamine or nitric oxide oxidation (Ward, 2008 and references therein). Only recently both steps of nitrification, ammonia oxidation and nitrite oxidation, have been observed in one organism (Daims et al., 2015; Van Kessel et al., 2015).

Nitrifier-denitrification is the process of ammonia oxidation to nitrite, followed by nitrite reduction via NO and N₂O to N₂ (Wrage et al., 2001).

Denitrification takes place under anoxic conditions where NO${}_3^{-}$ acts as an electron acceptor in the absence of oxygen ((Devol, 2008) and references therein). NO${}_3^{-}$ is reduced to NO${}_2^{-}$, NO and N₂O. At the core of oxygen minimum zones, N₂O is further reduced to N₂ (Bange, 2008 and references therein).

The solubility of N₂O in seawater has been described by Weiss & Price (1980). N₂O is a soluble gas, similar to CO₂. Due to its high solubility physical processes such as bubble effects do not have a strong effect on surface N₂O saturations. Warming and cooling of surface waters also has only a small effect (4% per degree centigrade). Due to the non-linearity of saturation curves for water mass mixing, diapycnal mixing, upwelling and entrainment of deeper water masses into the surface only has an impact on N₂O saturations if there is a large concentration gradient. Examples for high N₂O surface saturations due to these processes are rising isopycnals at the Southern Antarctic Circumpolar Current Front (SACCF, see below, Chen et al., 2014 and this work) and upwelling along the California coast (Lueker et al., 2003).

The Scotia Sea in the Atlantic sector of the Southern Ocean is confined by the Scotia Ridge to the north, east and south and the Drake Passage to the west (Atkinson et al., 2001). South Georgia is part of the North Scotia Ridge in the path of the Antarctic Circumpolar Current (ACC) and within the Antarctic Zone (AAZ) south of the Polar Front (PF).

The ACC is driven by strong, westerly winds between approximately 45–55°S (Trenberth, Large & Olson, 1990) and connects the Atlantic, Pacific and Indian Ocean. Isopycnals rise step-wise towards the south, characterising the ACC fronts and their resulting strong surface currents (Orsi, Whitworth & Nowlin, 1995 and references within): the Subantarctic Front (SAF), Polar Front (PF) and the Southern ACC Front (SACCF), accounting for most of the ACC transport. The SAF is characterised by the rapid subduction of the salinity minimum of Antarctic Intermediate Water (AAIW) to the north (Whitworth & Nowlin, 1987). Subduction of the temperature minimum of the AASW marks the position of the PF (Gordon, 1967). The SACCF is then characterised by a subsurface horizontal temperature gradient as Upper Circumpolar Deep Water (UCDW) shoals to the south. The Southern Boundary of the ACC (SB) can be identified by a loss of UCDW properties (temperature, salinity, oxygen minimum) to the south. To the north of the SB the UCDW isopycnal shoals and UCDW can be entrained into the Surface Mixed Layer (SML) (Orsi, Whitworth & Nowlin, 1995). The Weddell-Scotia Confluence separates the ACC from the waters of the Weddell Gyre (Gordon, 1967) and is a regional feature of the Scotia Sea.

Around South Georgia, the path of the ACC is strongly influenced by the bathymetry (Atkinson et al., 2001 and references therein): The South Georgia Ridge to the west of the island deflects the ACC to the north, resulting in strong large- and mesoscale variability. Additionally, the wide shelf seems to influence temperature and salinity in the waters around the island. It was observed that in summer the water is slightly warmer and fresher than in the surrounding ocean, possibly due to solar heating of the shallow water column and local runoff, respectively (Priddle, Heywood & Theriot, 1986; Brandon et al., 2000).

Waters around South Georgia are characterised by high abundances of phytoplankton, zooplankton and vertebrate predators (Atkinson et al., 2001), whereas most other areas of the Southern Ocean are dominated by High Nutrient Low Chlorophyll (HNLC) conditions (Martin, 1990). A relatively stable water column and benthic iron input support productivity in the vicinity of the island (Holeton et al., 2005; Korb et al., 2005).

To the south of the Scotia Sea and the ACC, the Weddell Sea is characterised by a large gyre, flowing eastwards until 20–30°E, returning westwards along the continental margin and following the eastern coast of the Antarctic Peninsula northwards (Deacon, 1979). Seasonal blooms in the Weddell Sea are associated with the Antarctic shelf and the ice edge (El-Sayed & Taguchi, 1981; Kristiansen, Syvertsen & Farbrot, 1992; Nelson et al., 1989; Smith Jr & Nelson, 1990). Furthermore, drifting icebergs stimulate productivity by input of terrigenous iron through melt water (Biddle et al., 2015; Smith et al., 2007).

Generally, the Southern Ocean has the potential for both production and removal of N₂O (Rees, Owens & Upstill-Goddard, 1997): Solubility of N₂O increases at lower temperatures, and together with downwelling areas associated with deep-water formation and convergences in the Antarctic frontal zones, wide areas could function as sinks. On the other hand, upwelling of deep and intermediate waters could be a source of biologically produced N₂O to the atmosphere. Artificial iron fertilisation of the Southern Ocean may stimulate biological CO₂ uptake, but could potentially increase N₂O production, offsetting the benefits of CO₂ sequestration in terms of radiative forcing (Fuhrman & Capone, 1991; Jin & Gruber, 2003). While an iron fertilisation experiment in the Australasian sector of the Southern Ocean showed N₂O accumulation in the pycnocline (Law & Ling, 2001), no increase in N₂O concentrations was observed during a similar experiment in the Atlantic subpolar sector (Walter et al., 2005).

Here, we present high-resolution measurements of ocean surface N₂O concentrations from the Scotia Sea and Weddell Sea. Using laser off-axis Integrated Cavity Output Spectroscopy (ICOS) (Arevalo-Martinez et al., 2013; Grefe & Kaiser, 2014) combined with wind-speed gas exchange parameterisations we can resolve small-scale variability in N₂O fluxes and capture the impact of frontal structures and changes in weather conditions on emissions from the Southern Ocean.

Methods

N₂O concentrations in surface waters were measured during the annual Western Core Box (WCB) krill survey in the Scotia Sea between 28 December 2011 and 16 January 2012 (JR260B) and in the Weddell Sea from 20 January to 2 February 2012 (JR255A/GENTOO Gliders: Excellent New Tools for Observing the Ocean) on board RRS James Clark Ross. The measurement region for both cruises is shown in Fig. 1.

The setup and performance of the coupled analyser-equilibrator system is described in Grefe & Kaiser (2014). Briefly, a percolating glass bed equilibrator was connected to a N₂O/CO analyser (N₂O/CO-23d, Los Gatos Research Inc.). Artificial air mixtures (21% O₂, 79% N₂) with nominal N₂O mole fractions of 300, 320 and 340 nmol mol⁻¹ (BOC) were used as reference gases. These gas mixtures were compared with IMECC/NOAA standards to determine the exact values of (297.6 ± 0.1), (325.3 ± 0.1) and (344.2 ± 0.1) nmol mol⁻¹ (NOAA-2006 scale). Reference gases and marine background air were measured twice a day for 20 min each during JR260B. Due to sufficient analyser stability during this cruise, calibration was reduced to once a day during JR255A/GENTOO. To ensure complete flushing of the cavity, only the last 5 min of each 20-min gas measurement were evaluated. Correspondingly, the first 15 min of equilibrator data following the reference gas and air measurements were discarded. Precision for single reference measurements (standard deviation of measurements at 1 Hz over 5 min) was 0.4 nmol mol⁻¹. The standard deviation of the uncorrected reference gas measurements throughout JR260B was 1.1 nmol mol⁻¹ (n = 29) and 0.8 nmol mol⁻¹ (n = 19) for JR255A/GENTOO. To estimate the long-term repeatability of the measurements after calibration, we used the standards with the lowest and highest N₂O mole fraction to calibrate the standard with the middle N₂O mole fraction. The mean difference between resulting calibrated N₂O mole fraction and the actual value of 325.3 nmol mol⁻¹ was (0.2 ± 0.1) nmol mol⁻¹, corresponding to a precision better than 0.1%. The flow rate of the headspace gas through the analyser was 400 mL min⁻¹ (293 K, 1 bar) resulting in a 95% relaxation time (t₉₅ = 3τ) of approximately 7 min (Grefe & Kaiser, 2014). Data were acquired at a rate of 1 Hz; which were binned into 60 s averages for data evaluation purposes. Water flow through the equilibrator was held constant at 1.8–1.9 L min⁻¹, using a flow regulator ($\frac{1}{2}$ inch diameter tap tail flow regulator; Robert Pearson & Company Ltd, Warminster, UK). Water temperature in the equilibrator was measured with a Pt-100 temperature probe (Omega Engineering Limited, Manchester, UK), calibrated against a mercury reference thermometer to within ± 0.06 °C, and recorded using a RTD temperature recorder and USB Datalogger interface (both Omega Engineering Limited).

Concentration of dissolved N₂O (c) was calculated from the dry mole fraction measured in the equilibrator headspace (x), water temperature in the equilibrator (T_eq), salinity (S) and atmospheric pressure (p_air) using the solubility function F as described by Weiss & Price (1980), Eq. (1). (1)${\displaystyle c=xF\left(T_{\mathrm{eq}},S\right)p_{\mathrm{air}}.}$ N₂O saturation in surface waters (s) was calculated by comparing x with the atmospheric mole fraction x_air and the respective equilibrium concentrations for (T_eq) and temperature at the seawater intake (T_in) Eq. (2). (2)${\displaystyle s=\frac{xF\left(T_{\mathrm{eq}},S\right)}{x_{\mathrm{air}}F\left(T_{\mathrm{in}},S\right)}.}$ Values for p_air, S and T_in were from the ship’s surface water and meteorological monitoring system Surfmet (http://www.bodc.ac.uk).

Sea-to-air flux was calculated using wind speeds at 10 m above sea level from the CCMP Wind Vector Analysis Product (http://www.remss.com/measurements/ccmp). CCMP data have been interpolated to match cruise dates and positions for JR260B and JR255A/GENTOO. The gas transfer coefficient (k_w) was calculated, using the parameterisation of Nightingale et al. (2000) (Eq. (3)) and k_w was adjusted for N₂O with the Schmidt number Sc calculated following (Wanninkhof, 2014), Eq. (3). (3)${\displaystyle \frac{k_w}{m{d}^{-1}}=0.24\left[0.222{\left(\frac{u_{\text{CCMP}}}{\mathrm{m}{\mathrm{s}}^{-1}}\right)}^2+0.333\frac{u_{\text{CCMP}}}{\mathrm{m}{\mathrm{s}}^{-1}}\right]{\left(\frac{Sc}{600}\right)}^{-0.5}.}$ The air-sea flux (Φ) was calculated from k_w and the difference between N₂O concentrations in seawater c and air saturation concentrations (c_sat), Eq. (4): (4)${\displaystyle \Phi =k_w\left(c-c_{\mathrm{air}}\right)=k_w\left[c-x_{air}F\left(T_{in},S\right)p_{\mathrm{air}}\right].}$ Ship-based underway measurements of sea surface temperature and salinity were used to identify frontal structures for JR260B and JR255A/GENTOO. The location of these areas was then compared to the climatological front positions of Orsi, Whitworth & Nowlin (1995) to ensure positions based on measured salinity and temperature fall within expected geographical locations.

Satellite products for chlorophyll a and climatology data for nitrate (NO${}_3^{-}$) concentrations were compared with in situ N₂O saturation data. Chlorophyll a concentrations were obtained from MODIS-Aqua (https://oceandata.sci.gsfc.nasa.gov) for December 2011 and January 2012. The level-3 product was downloaded at a 4 km resolution. Climatological mean dissolved NO${}_3^{-}$ concentrations were downloaded from the World Ocean Atlas 2013 (Garcia et al., 2013) for boreal winter (https://www.nodc.noaa.gov/OC5/woa13/woa13data.html). This climatology data was gridded 1°longitude by 1° latitude for 37 standard depths. Depth profiles taken during station work south of the PF and ACC for JR260B and JR255A/GENTOO respectively suggested a mixed layer depth (MLD) of approximately 60 m, similar to MLD derived from Argo float profiles January to March (Dong et al., 2008). NO${}_3^{-}$ concentrations from this depth were extracted from the climatology and interpolated along the cruise track for comparison with N₂O saturations.

Results and Discussion

N₂O concentrations and saturations in the surface ocean

N₂O dry mole fractions measured in marine air were on average (323.6 ± 0.6) nmol mol⁻¹ during JR260B and (324.0 ± 0.7) nmol mol⁻¹ during JR255A/GENTOO. These values agree within measurement uncertainties with data from the Advanced Global Atmospheric Gases Experiment (AGAGE) for Cape Grim, Tasmania in January 2012 (40.68°S 144.69°E, (323.9 ± 0.5) nmol mol⁻¹). Surface water saturations and sea-to-air-flux were calculated based on the mean atmospheric mole fractions measured during JR260B and JR255A/GENTOO.

JR260B

Concentrations and saturations of N₂O in the surface ocean for JR260B are shown in Figs. 2A and 2B, respectively. Average N₂O concentrations were (14.0 ± 0.7) nmol L⁻¹, ranging from 11.2 nmol L⁻¹ to 16.2 nmol L⁻¹. Average N₂O saturation was (104.3 ± 3.4)%, ranging from 99.4% to 126.1%.

For the purposes of the discussion of the data, they were divided into four different areas based on differences in biogeochemical conditions and oceanographic settings:

1. Measurements between the Falkland Islands and the SAF in comparatively warm, saline surface waters (θ >7 °C, S 34.1–34.2) were located in the Subantarctic Zone (SAZ).

2. The Polar Frontal Zone (PFZ) covered the rapidly cooling and freshening surface waters (θ declining from 7 to 3 °C, S declining from 34.2 to 33.9) between the SAF and PF.

3. The Antarctic Zone (AAZ) was identified by the relatively homogeneous cold and fresh T-S signature of AASW.

4. Stromness Bay is discussed separately, due to the unique influence of the island of South Georgia in this area.

Subantarctic Zone (SAZ).

In the Subantarctic Surface Waters (SASW) close to the Falkland Islands to the east of the Subantarctic Front (SAF), average surface concentrations were low, while saturations were similar to those observed in the Antarctic Zone (AAZ) (11.4 ± 0.2) nmol L⁻¹, (102.4 ± 0.7)%, see below). Water temperature and salinity were higher than in the Antarctic waters south and east of the PF, so the difference in concentrations was mostly due to solubility effects. Zhan & Chen (2009) observed higher saturations (110 to 135%) to the north of the SAF in the Indian Sector of the Southern Ocean. The cause of this difference between the two sectors of the Southern Ocean is not known and highlights the need for more research to better understand marine N₂O cycling.

Polar Frontal Zone (PFZ).

N₂O concentrations across the PFZ between the SAF and PF were highly variable with values ranging from 11.4 to 16.2 nmol L⁻¹ (101.7 to 126.1% saturation) while water temperature and salinity were decreasing towards the east. Average concentrations and saturations across the PFZ were (13.8 ± 0.9) nmol L⁻¹ and (108.8 ± 5.2)%, respectively.

The most important sources of N₂O to surface waters are upwelling of old water masses with preformed high N₂O concentrations and in situ production during remineralisation/nitrification or denitrification. High N₂O saturations across the PFZ could neither be attributed to a specific water mass (identifiable from temperature, salinity and density of the surface water), nor to a point on the mixing line between Subantarctic Surface Water (SASW) and Antarctic Surface Water (AASW) across the PFZ (Fig. 3A). Therefore, it is unlikely that upwelling and mixing processes or regional oceanographic features are the source of the high N₂O saturations across the PFZ.

Frontal systems can supply nutrients and iron to surface waters with large phytoplankton blooms forming across the PFZ, potentially fuelled by iron input from the Antarctic Peninsula archipelago, the Scotia Ridge and Georgia Rise (De Baar et al., 1995; Wadley, Jickells & Heywood, 2014). Satellite observations showed high chlorophyll a concentrations in proximity to high N₂O supersaturations in the PFZ (Fig. 4A, ‘Comparison of N₂O surface saturations to chlorophyll a and nitrate concentration’). Remineralisation of sinking bloom biomass could lead to enhanced in situ N₂O production across the PFZ, resulting in high saturations.

In contrast to the high surface N₂O concentrations and saturations in the PFZ during JR260B, data from the Indian sector of the Southern Ocean showed minimum N₂O inventories within the PFZ (Farías et al., 2015). Farías et al. identify uptake of N₂O as an alternative substrate during biological nitrogen fixation as a potential sink process, presumably stimulated by iron input from the sediments of the Kerguelen archipelago. While both islands, South Georgia in the Atlantic and the Kerguelen Islands in the Indian Sector, are positioned within the PFZ and sediments are a source of iron to the water column, the data from the Kerguelen region were collected to the west of the islands, whereas JR260B data were collected to the east of South Georgia. The geographic location of the measurements in relation to islands could affect the biogeochemical properties of the water column and the community composition of N₂O cycling bacteria and archaea (Santoro et al., 2011; Löscher et al., 2012) and therefore result in major differences in the local source–sink behaviour of the ocean.

Antarctic Zone (AAZ).

Average N₂O concentrations in the AAZ to the west of the PF near South Georgia were substantially higher than in the SAZ (14.1 ± 0.4) nmol L⁻¹. Saturations, however, were only slightly increased to (103.9 ± 3.1)%, similar to values of (103.0 ± 2.0)% previously observed in the region (Weiss et al., 1992). The lower water temperature and salinity of Antarctic Surface Water AASW increased N₂O solubility, resulting in lower saturation values. Upwelling of deep water masses would not be expected away from the ACC and denitrification is not likely to take place in the oxygenated surface waters of the AAZ. Nitrification and nitrifier-denitrification would be more likely sources of in situ N₂O production. High nitrification rates (>30 mmol m⁻² d⁻¹) were suggested for the Pacific sector of the Southern Ocean south of the PF (Sambrotto & Mace, 2000) and could also account for N₂O supersaturation observed in the AAZ of the Atlantic sector. Extensive phytoplankton blooms were observed in the vicinity of South Georgia and across the Scotia Sea, extending to the southern limit of the Polar Front (Korb, Whitehouse & Ward, 2004; Korb et al., 2005). These blooms can develop due to a stable water column over the shelf and iron input from the shelf sediments and island runoff (De Baar et al., 1995; Holeton et al., 2005; Korb et al., 2005; Wadley, Jickells & Heywood, 2014). This accumulation of biomass would supply ample substrate to sustain high nitrification rates and N₂O production could lead to the observed supersaturation close to the island, despite the absence of a clear correlation between chlorophyll a, climatological NO${}_3^{-}$ concentrations and N₂O saturations (‘N₂O sea-to-air flux’). Not much is known about nitrifier-denitrification in polar waters and additional research is required to estimate the importance of this production pathway for N₂O accumulations in surface waters.

Stromness bay.

N₂O concentrations and saturations in coastal Stromness Bay, South Georgia, were higher than in the open waters of the AAZ ((14.8 ± 0.3) nmol L⁻¹ and (108.1 ± 2.6)%, respectively) with highest saturations observed where fresh meltwater runoff from the island was mixing with AASW (Fig. 3A). Terrestrial runoff can transport iron and biomass from land into the sea and stimulate productivity and subsequent remineralisation where N₂O is produced during nitrification. In addition, fur seals (Arctocephalus gazella) and macaroni penguins (Eudyptes chrysolophus) have large breeding colonies on South Georgia, re-distributing nitrogen from their hunting grounds to the island (Whitehouse et al., 1999).

While no NO${}_3^{-}$ concentrations are available for JR260B, Rees et al. (2016) observed high NO${}_3^{-}$ values of 25.8 µmol L⁻¹ at the base of the mixed layer north of South Georgia in austral summer 2013. Concentrations would be expected to be even higher in Stromness Bay due to land run-off and the high abundance of penguins and seals observed at the shore and in the water at the time of measurements. High nitrogen loads in coastal waters are expected to lead to high N₂O saturations (Bange, Rapsomanikis & Andreae, 1996) and would explain the consistently high supersaturations observed in Stromness Bay during JR260B.

Nitrifier-denitrification in anoxic sites of suspended particles can be an additional source of N₂O (Ostrom et al., 2000). As the water depth in Stromness Bay is shallow (60 m) and well mixed at the anchoring site, N₂O produced by denitrification at the sediment interface could have diffused into the water column, contributing to the high concentration at the surface.

JR255A/GENTOO

During JR255A/GENTOO, average N₂O surface concentrations were (14.9 ± 1.2) nmol L⁻¹, corresponding to saturations of (103.1 ± 3.6)% (range: (10.6–17.0)nmol L⁻¹; (96.5–116.0)% saturation). While N₂O concentrations were higher than for JR260B, mainly due to the lower water temperatures, average saturations were slightly lower.

As for JR260B, data have been discussed for three areas, based on substantial differences in biogeochemical conditions and oceanographic settings:

1. Data collected between the Falkland Islands and the South Scotia Ridge were discussed below as Drake Passage, crossing the SAF (θ >7 °C to the north), PF (θ decreasing from 7 °C to 3 °C, S from 34.2 to 33.8), SAACF (identified by a small salinity change in surface waters from 33.8 to 34) and SB (surface waters influenced by UCDW, S increases from 34.4 to 34.6);

2. South of the South Scotia Ridge, measurements were designated as Weddell Sea data, including a standing eddy (identified by a decrease in surface salinity from 34.4 to 34.2 and an increase of surface temperature at the edges from 0.5 to 1 °C, Thompson et al., 2009) forming over the ridge and data collected from the Antarctic shelf;

3. Measurements influenced by the sea ice edge in the Weddell Sea (θ <0 °C and salinity <33.6).

Drake Passage.

Average N₂O concentrations at the beginning of the cruise were (13.7 ± 1.8) nmol L⁻¹ ((102.9 ± 4.2)% saturation). Decreasing temperatures during transit across Drake Passage to the Eastern Antarctic Peninsula partially accounted for increasing N₂O concentrations (Fig. 5A), resulting in a wide range of concentrations across the region, whereas saturations were on average only slightly above equilibrium with the atmosphere. Water masses across the Drake Passage were Subantarctic Surface Water to the north of the SAF and Antarctic Surface Water to the south of the PF (SASW and AASW, respectively) (Fig. 3B). As these surface waters are in constant contact with the atmosphere, they are expected to be in equilibrium with the atmosphere in absence of biological N₂O sources. N₂O saturations above 100% would therefore be due to in situ production during remineralisation of biomass.

For comparison, Rees, Owens & Upstill-Goddard (1997) observed slightly lower saturations for Drake Passage ((99.7 ± 3.0)%), while Weiss et al. (1992) reported values close to the data presented here (Ajax 2: (102.3 ± 0.9)%, JR255A/GENTOO: (102.9 ± 4.2)%). Higher saturation values for Weiss et al. (1992) and JR255A/GENTOO could be due to interannual and seasonal variability. JR255A/GENTOO and Ajax 2 took place in January and February, later in the austral summer season than the November/December cruise of Rees, Owens & Upstill-Goddard (1997). Nitrification as part of the remineralisation of sinking biomass could have increased N₂O accumulations in the mixed layer over the summer, resulting in higher values later in the growing season. No direct link to chlorophyll a concentrations was found (‘Comparison of N₂O surface saturations to chlorophyll a and nitrate concentrations’), and was unlikely to exist due to the temporal and spatial decoupling.

Just south of the Southern Boundary of the Antarctic Circumpolar Current (SB), concentrations as well as saturations reached the highest values observed during JR255A/GENTOO. Temperature and salinity characteristics of the surface water indicated an influence of Upper Circumpolar Deep Water (UCDW) from below (Fig. 3B, Orsi, Whitworth & Nowlin, 1995). UCDW is an old water mass, high in remineralised nutrients and N₂O as a byproduct of nitrification. Depth profiles in the Indian sector of the Southern Ocean displayed N₂O maxima in the oxygen minimum layer of the UCDW and potential transport of these waters to the surface (Chen et al., 2014; Zhan et al., 2015). Entrainment of UCDW across the thermocline, as inferred from a decrease in surface temperature and an increase in salinity, could explain the high saturations across the SB during JR255A/GENTOO. Additionally, iron input from the Scotia Ridge, as observed by Klunder et al. (2014) could enhance in situ productivity, supplying substrate in form of sinking particles for in situ N₂O production.

Weddell sea.

Surface N₂O concentrations of the open waters of the Weddell Sea were on average (15.2 ± 0.2) nmol L⁻¹ ((102.2 ± 1.6)% saturation). South of the ACC, the surface waters were colder than across Drake Passage, increasing the solubility of N₂O and resulting in slightly lower saturations despite higher concentrations. Small but persistent supersaturations during JR255A were in contrast to N₂O data from the Indian sector, where surface waters were mainly undersaturated (Chen et al., 2014; Zhan & Chen, 2009; Zhan et al., 2015). Meltwater low in N₂O in combination with strong stratification, preventing exchange across the thermocline were identified as possible reasons for N₂O undersaturations. While underaturations were observed during JR255A/GENTOO close to the sea ice edge (see below), meltwater does not seem to have a strong impact on the N₂O saturations of surface waters in the Weddell Sea. Further studies are required to identify the differences between the Atlantic and Indian sector of the Southern Ocean in terms of meltwater contribution to the mixed layer and strength of stratification and mixing across the thermocline.

Higher-than-average N₂O saturations for the Weddell Sea during JR255A/GENTOO were observed on the Antarctic shelf off the tip of Joinville Island ((104.2 ± 1.7)%) and during the section across the large standing eddy forming over the South Scotia Ridge, centred on 62°S and 54°W (Thompson et al., 2009) ((110.5 ± 1.2)%) (Figs. 5A and 5B). Iron from the sediments, as well as land run-off could have stimulated productivity in these areas (Klunder et al., 2014; Sañudo Wilhelmy et al., 2002; Thompson & Youngs, 2013; Wadley, Jickells & Heywood, 2014). In addition, the sampling region was more sheltered from the circumpolar winds by the Antarctic Peninsula, stabilising the water column over the shallow bathymetry. The combination of ample supply of trace nutrients and a stable water column, keeping phototrophic organisms in the euphotic zone, presumably resulted in comparably high productivity and subsequent N₂O production during remineralisation of sinking biomass. In the vicinity of dynamic features, such as frontal zones, average chlorophyll concentrations for January 2012 show an increase in chlorophyll a over the shelf and across the standing eddy where N₂O saturations are highest (Fig. 4: 25 and 29 Jan). In areas where physical transport and mixing is reduced, N₂O, as well as chlorophyll, can accumulate and show the expected relationship between high productivity and N₂O production during nitrification as sinking biomass is remineralised.

Sea ice edge.

Lowest saturations of on average (98.8.0 ± 1.3)% were observed close to the sea ice edge in the south east of the survey region (Fig. 5B). Average N₂O concentrations at the ice edge were (15.2 ± 0.2) nmol L⁻¹, similar to the average of (15.4 ± 0.2) nmol L⁻¹ observed in the other open ocean regions in the Weddell Sea. Saturations, however, were lower due to lower temperature and salinity, resulting in higher N₂O solubility. Randall et al. (2012) observed under-saturations of N₂O within sea ice due to loss of dissolved gases during brine rejection. Mixing of seawater with under-saturated melt water from sea ice that has undergone brine rejection during formation would decrease surface concentrations, which was not observed during JR255A. However, it is possible that end-members were not captured in these measurements and the mixing line for temperature and salinity indeed indicates dilution of Weddell Sea surface waters with melt water low in N₂O.

Comparison of N₂O surface saturations to chlorophyll a and nitrate concentrations

Satellite products and climatologies are readily available datasets, covering pigment and nutrient distribution for wide areas of the global ocean. To investigate potential links of N₂O surface saturations to productivity and remineralisation, we have tried to find correlations between N₂O, chlorophyll a concentrations (as proxy for biomass accumulation) and nitrate concentrations (as proxy for integrated remineralisation). A tentative link between chlorophyll a concentrations from satellite products and NO${}_3^{-}$ concentrations from climatologies could also open up these data sources to infer global and long-term N₂O saturations without relying on in situ data or N₂O models.

Surface saturations of N₂O were compared to monthly average chlorophyll a concentration for December 2011 and January 2012 (Figs. 5 and 6). There was no simple correlation between N₂O saturations and chlorophyll a concentrations, neither for specific areas of the two cruises, nor for the complete dataset (Figs. 6A and 6B). However, average chlorophyll was high in proximity to the crossings of the PFZ during JR260B (Fig. 4A: 30 Dec, 15 Jan).

In addition to satellite chlorophyll values, NO${}_3^{-}$ concentrations from the World Ocean Atlas climatology (https://www.nodc.noaa.gov/OC5/woa13/woa13data.html) were calculated for the mixed layer (MLD, 60 m) and were compared to N₂O saturations for JR260B and JR255A/GENTOO to test whether NO${}_3^{-}$ as end-product of nitrification could be used as a proxy for mixed layer N₂O production (Figs. 7A and 7B). However, again no clear correlation was observed between N₂O saturations and NO${}_3^{-}$ concentrations, neither for specific survey areas and oceanographic features, nor for the entire dataset.

Chlorophyll as a proxy for productivity and NO${}_3^{-}$ as the endproduct of nitrification (a major N₂O-producing pathway) might be expected to show similar behaviour to N₂O mixed layer saturations. However, there are reasons why we do not observe simple correlations between these parameters.

Firstly, there is the methodological issue of comparing near real-time N₂O data with monthly averages (chlorophyll) and climatology data (NO${}_3^{-}$). The location of frontal systems and upwelling water masses in the Southern Ocean are not fixed. Mixing and transport of water will influence observations during JR260B and satellite and climatology data.

Secondly, biomass (represented by chlorophyll) is not immediately available to nitrifying organisms. Breaking down of sinking biomass and release of NH${}_4^{+}$, the substrate for nitrification, introduces a time delay for N₂O production. NH${}_4^{+}$ and NO${}_3^{-}$ are not exclusively controlled by the activity of nitrifying bacteria and archaea. They are also important nutrients for phytoplankton, therefore uptake by algae is further complicating the use of NO${}_3^{-}$ as a proxy for N₂O saturations.

The lack of a clear, quantitative link between N₂O saturations and observations of chlorophyll a and NO${}_3^{-}$ makes it difficult to infer s(N₂O) (unlike CO₂) by multi-linear regression (Wallace, 1995) or prognostic neural network models (Landschützer et al., 2014). Despite the absence of a quantitative correlation, high N₂O saturations were accompanied by an increase in chlorophyll a concentrations on the Antarctic shelf and across the standing eddy.

Conclusions

High-resolution (1 min averages of measurements at 1 Hz) N₂O data were collected from the surface waters of the Atlantic sector of the Southern Ocean during transit and station work, covering highly varied areas across the ACC and the AAZ. These data from the austral summer 2011/2012 add substantially to the limited observations of N₂O emissions from the Southern Ocean, increasing observations from the Scotia Sea by 20,674 values (188 data points from 1989 previously published by Weiss et al. (1992)) and by 9730 from the Weddell Sea (250 data points from 1984 previously published by Weiss et al. (1992)), and show unprecedented fine-scale variability compared to the existing data (supplementary Figures 1 and 2 in Weiss et al., 1992).

Both, the Scotia Sea and the Weddell Sea, were a source of N₂O to the atmosphere. The source strength was highly variable in space and time with rare areas of negative sea-to-air flux.

The following areas were identified as strong N₂O sources by high sea-to-air flux at the time of the survey and/or high supersaturations:

• shallow bathymetry/ridges: North and South Scotia Ridge,

• shelf and coastal areas: The Antarctic shelf and coastal South Georgia and

• the Polar Frontal Zone: To the east of Drake Passage,

whereasN₂O sources were associated with water mass transport at

• the Southern Boundary of the ACC, where shoaling UCDW was influencing surface waters of the mixed layer.

The following areas were identified as sinks for atmospheric N₂O by negative sea-to-air flux at the time of the survey:

N₂O sinks through physical processes:

• Sea ice edge: N₂O sink due to increased solubility or brine rejection.

N₂O sinks through other processes, presumably seasonal cooling and temporal lag between heat and gas fluxes and/or biological nitrogen fixation:

• Falkland shelf to the south of the islands.

• To the east of the PF in the Scotia Sea.

Neither chlorophyll a, nor NO${}_3^{-}$ concentrations showed any correlation to N₂O saturations. Assuming that the average flux values of 2.3 µmol m⁻² d⁻¹ and 0.7 µmol m⁻² d⁻¹ calculated for JR260B and JR255A are representative for the Scotia Sea and Weddell Sea over time, the combined area would contribute 0.04 Tg a⁻¹ N (nitrogen equivalent) to the global N₂O source. While this value is relatively low, also compared to model estimates of 0.9 Tg a⁻¹ N (nitrogen equivalents) for the entire Southern Ocean (Nevison et al., 2005), areas with a strong variability in N₂O saturation and sea-to-air flux were observed for both cruises. Short-term average flux for the Scotia Sea was almost twice as high as the global average. These findings indicate the importance of high-resolution data to accurately estimate the source strength of mesoscale features, such as frontal systems and eddies. This study highlights the need for further high-resolution data from other sectors of the Southern Ocean to resolve variability in N₂O emissions and improve global emission estimates.

Supplemental Information