High‐resolution distribution pattern of surface water nitrous oxide along a cruise track from the Okhotsk Sea to the western Arctic Ocean

Nitrous oxide in the surface water was measured using an automatic underway system, along with measurements of surface water pCO2, dissolved oxygen, salinity, temperature, and sea ice coverage, along a cruise track through the Bering Sea and Chukchi Sea to the Canadian Basin during the 7th Chinese National Arctic Research Expedition. The results show that, overall, the regions along the cruise track are net sources of N2O to the atmosphere. Several N2O oversaturation maxima were observed along the cruise track, with an absolute maximum of approximately 60%. According to the hydrographic setting and the distribution patterns of pCO2, dissolved oxygen, and sea ice coverage in the study area, it was concluded that the N2O oversaturation maxima may result from hydrographic processes, such as mixing of different water masses, upwelling or convection, and possible production beneath the sea ice. Additionally, the lowest value of approximately 90% may result from dilution related to the melting of sea ice. An evaluation of the air–sea flux along the cruise track shows that the continental shelf and upwelling region are N2O sources, whereas the study area in the open Arctic Ocean does not show obvious source or sink characteristics.

Nitrous oxide is one of the more important greenhouse gases and has a greenhouse effect approximately 300 times greater than that of CO 2 on a per molecular basis; in addition, N 2 O is currently the ozone-depleting substance with the largest emission rate (Ravishankara et al. 2009). The ocean is among the most important N 2 O sources to the atmosphere and accounts for~21% of the global N 2 O (Bange et al. 2019). Since the first study on oceanic N 2 O carried out by Craig and Gordon (1963), this subject has been extensively studied (Bange et al. 2019), and the results show that open oceans are slightly supersaturated with N 2 O at 104-130% (Dore et al. 1998;Wilson et al. 2014). In the N 2 O hotspots in the eastern tropical South Pacific Ocean, eastern tropical North Pacific (ETNP) and the Arabian Sea, N 2 O supersaturation has been identified in the surface water and has been rigorously studied. Arévalo-Martínez et al. (2015) observed a maximum N 2 O supersaturation of 12,244% in the eastern tropical South Pacific (ETSP), which contributes 5-20% of the global budget. Kock et al. (2015) identified an accumulation of N 2 O (~850 nM) in the oxygen minimum zone, and Bourbonnais et al. (2017) observed a maximum N 2 O concentration of~190 nM in the surface water of the oxygen deficient zone of the same region. Furthermore, Babbin et al. (2015) revealed that the N 2 O turnover is approximately 20 times higher than that of atmospheric efflux in the ETNP. However, although N 2 O production is less active in the open ocean than in the hotspots, the open ocean contributes significantly to the global budget (Dore et al. 1998). Details on the N 2 O source and sink characteristics of hotspot regions, such as the Peruvian upwelling regions , and the open ocean, such as the Southern Ocean (Grefe et al. 2018), were further revealed with the development of high-resolution underway observation systems based on off-axis integrated cavity output spectroscopy (Arévalo-Martínez et al. 2013;Grefe and Kaiser 2014) and cavity ring down spectroscopy (Zhan et al. 2018).
The difference in N 2 O production rates between the open ocean and hotspot regions is due to the different mechanisms present at these locations. In oxic open ocean water, *Correspondence: zhanliyang@tio.org.cn This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Additional Supporting Information may be found in the online version of this article.
Special Issue: Biogeochemistry and Ecology across Arctic Aquatic Ecosystems in the Face of Change. Edited by: Peter J. Hernes, Suzanne E. Tank and Ronnie N. Glud nitrification is considered the dominant N 2 O production mechanism (Cohen and Gordon 1979;Bange and Andreae 1999). In the hotspot regions, N 2 O production is more intense than that in the open ocean because of the low oxygen conditions (oxygen minimum zone, OMZ) resulting from the remineralization of organic particulate matter associated with the high productivity in this region. Nitrification may be enhanced at low dissolved oxygen (DO) concentrations (Goreau et al. 1980); however, this phenomenon was not observed in more recent studies using isotopomer techniques (Frame and Casciotti 2010). Instead, denitrification was reported as a major contributor to N 2 O production under low oxygen conditions Bourbonnais et al. 2017), whereas Ji et al. (2015 revealed coupling of nitrification and denitrification in the ETSP OMZ region. In addition to canonical denitrification that utilizes NO 3 − , nitrifier denitrification is another denitrification process but only reduces NO 2 − to N 2 O. It may occur not only under low oxygen concentrations but also in the oxygenated lower euphotic zone and upper aphotic zone (Ostrom et al. 2000;Wilson et al. 2014). Among the studied areas, little is known about the N 2 O source and sink characteristics of the Arctic Ocean, and only limited studies have been carried out in this area (Hirota et al. 2009;Kitidis et al. 2010;Fenwick et al. 2017). These studies suggested that N 2 O oversaturation over the continental shelf may result from nitrification, denitrification or their coupled processes, whereas N 2 O undersaturation in surface water can be caused by ice melting (Randall et al. 2012). This work presents the application of an automated high-resolution underway N 2 O observation system in the Bering Sea, Chukchi Sea and Canadian Basin for the first time during the 7 th Chinese National Arctic Expedition (CHINARE), and the source and sink characteristics of the study area are discussed.

Method and study area
Study area and hydrographic setting During the 7 th CHINARE, which was conducted between early July and late September 2016, the N 2 O concentration and pCO 2 in the surface water between 49 N and 78 N along the cruise track were measured. The study area is shown in Fig. 1, and the cruise track is marked by a blue line. The R/V Xuelong left the Okhotsk Sea on the 16 th of July, entered the Bering Sea via the Aleutian Arc, and left the Bering Sea via the Bering Strait for the Chukchi Sea on the 24 th of July. After station sampling was conducted over the Chukchi Plateau, the collection of the data used in this work was completed on the 2 nd of Aug.
The current system in the study area is also shown in Fig. 1. The Alaskan Stream (light blue) flows westward along the southern flank of the Aleutian Islands, exchanging water with the Bering Sea via the main sills along the Aleutian Arc. One branch of the system flows along the eastern coast of the Bering Sea into the Arctic Ocean, whereas part of the system flows across the Bering Sea continental slope as the Bering Slope Current (purple), which is also the north limb of the Bering Sea Gyre. This current separates into two branches, one that flows northward and forms the Anadyr current (red) and the other that flows southward and is known as the Kamchatka Current (gray). The Anadyr Current and the Alaskan Coastal Current flow through the Bering Strait. North of the Bering Strait, the inflow waters separate into three main currents that flow through the Herald Canyon, Central Channel, and along the Alaskan coast, and all three currents turn eastward; part of the eastern branch turns west to form the Chukchi Slope Current at Barrow Canyon (light green) (Corlett and Pickart 2017). North of the Pacific-origin water is the eastward-flowing Atlantic-origin water.

Method for underway observations
Nitrous oxide and Carbon dioxide data in the air and surface water were continuously measured with an automated underway system consisting of an upstream device and Picarro 5101i N 2 O isotope and 2131i CO 2 isotope analyzers (Picarro). The setting of the instrument was as described in Zhan et al. (2018). Briefly, the automatic underway system was programmed to select gas samples from the equilibrator headspace, atmosphere, and two tanks of mixed N 2 O and CO 2 reference gas. Reference 1 contained 329.4 AE 0.4 ppb N 2 O and 409.2 AE 0.2 ppm CO 2 , and reference 2 contained 912.5 AE 0.9 ppb N 2 O and 685.1 AE 0.4 ppm CO 2 . These reference gases were provided by Beijing Huaxin Space & Sky Technology and were calibrated against NOAA reference gas with the 5101i instrument in the laboratory. Surface water was pumped from the water inlet 4.5 m below the surface and passed through the equilibrator at a rate of approximately 500 mL min −1 in the upstream device, whereas the closed loop of the gas line ran in the opposite direction at a rate of~500 mL min −1 . The gas sample then bubbled into the water sample and then passed through a Peltier dryer and Nafion tube to remove moisture. The analyzers were connected in parallel. The gas line was separated and joined upstream and downstream of the inlet and outlet of the two analyzers using tee connections. The sample gas flowed through the two analyzers at a flow rate controlled by a mass flow rate controller to meet the analyzers' required flow rates. The gas sample then returned to the equilibrator. The N 2 O measurement results were compared with those obtained using gas chromatography (GC) in the laboratory on land, and the results showed that the underway system data deviate by less than 3% from the data obtained by GC, suggesting that the two methods are consistent with each other. The precision of this method was better than 0.5%. The CO 2 data obtained using this method were compared to those obtained using the GO (General Oceanics) Automated Flowing pCO 2 Measuring System on the same cruise. The correlations between the data derived from these two methods resulted in a linear correlation of y = (1.0496 AE 0.0007)x + (−11.3887 AE 0.2220), with an R 2 of 0.9974.
The surface water temperature (SST) and salinity (SSS) were measured using an SBE21 instrument (Sea-Bird Electronics) deployed at the seawater inlet of the research vessel. The SST and SSS along the cruise track are shown in Figs. 2a and S1.
The surface DO was measured by an Aanderaa 4835 optode (Aanderaa Data Instruments AS), which was calibrated against discrete Winkler titrations.
The method for sea ice observation is described as follows. The sea ice concentration (in percentage, f ) was observed and documented every half-hour using the protocol of the Arctic Shipborne Sea Ice Standardization Tool established by the Climate-Cryosphere Arctic Sea Ice Working Group to characterize typical Arctic conditions. The sea ice observation was performed from the bridge of the R/V Xuelong for a local area with a diameter of 2 km, which might be reduced to 1 km on foggy days. For more detail, refer to Lei et al. (2017) and Hutchings and Faber (2018).

Air-sea flux calculation
The air-sea N 2 O flux along the cruise track was calculated using the air-sea equation (Liss and Merlivat 1986;Wanninkhof 1992): where F is the air-sea flux of the gas, k is the gas exchange coefficient, and ΔC is the difference between the observed concentration in and equilibrium concentration of the surface water calculated from the atmospheric partial pressure and in situ conditions, such as surface water temperature, salinity, and atmospheric pressure.
Since the above studies evaluated open water, the above equations were all multiplied by the percentage of open water, which was expressed as 1-f, where f represents the sea ice concentration.
The gas exchange coefficient of the whole cruise track was evaluated using the equations of Ho et al. (2006) and Nightingale et al. (2000).
U 10 is the wind speed above sea level and Sc is the Schmidt number, which is defined as the kinematic viscosity of the water divided by the diffusion coefficient of the gas and can be calculated following Wanninkhof (2014).
For the wind speed data, we used the remote sensing wind speed product from ASCAT (https://podaac-tools.jpl.nasa.gov) with a spatial resolution of 25 km. Taking into account the nonlinearity of the wind speed parameterization, the weighted gas exchange coefficient was calculated from the 60 daily wind speed data prior to the measuring time using the function developed by Reuer et al. (2007). The weighting is based on the fraction of the mixing layer ventilated on any given date (Fenwick et al. 2017). The mixed layer depth product was utilized, which could be obtained from global ocean physics reanalysis from CMEMS (Copernicus marine environment management service, http://marine.copernicus.edu/) with a high spatial resolution of 1/12 according to the collection dates.

Distribution of N 2 O along the cruise track
The distribution of N 2 O along the cruise track is shown in Fig. 2b. Except for several maxima along the whole cruise track, the N 2 O concentration shows a trend of increasing from 11.0 to~16.5 nM, which results from the increasing solubility of a gas with the decline in SST. When the N 2 O concentration was plotted against the SST (Fig. 3), most of the N 2 O concentration data are above the equilibrium line (red solid line), indicating that nearly all the surface water of the whole cruise track is oversaturated with N 2 O; in particular, at the high-and low-temperature ends, the surface water shows a state of N 2 O oversaturation.
To further reveal the source and sink characteristics of the study area, the saturation anomaly (SA) was calculated using the following equation: where C ob is the observed N 2 O concentration and C eq is the equilibrium concentration calculated from the in situ SST, SSS, atmospheric pressure, and atmospheric N 2 O mixing ratios. The derived SA of the whole cruise track was plotted against section distance (Fig. 2c). Except for the maxima along the cruise track, the rest of the surface water SA value was regarded as the SA baseline and can be divided into two parts: south of 4700 km, where the SA baseline ranged between 4% and 10%, which is consistent with other studies (Wilson et al. (2014) and references therein), and north of 4700 km, where the SA baseline is approximately 0%. The difference south and north of 4700 km probably results from different surface water properties. South of 4700 km, the surface water was influenced by different hydrographic structures, such as fronts, upwelling, and possible sediment release of N 2 O, which can reach or even outcrop the surface water, whereas north of 4700 km, the surface water is homogeneous in properties and well stratified due to the ice melting process, resulting in N 2 O near equilibrium in the surface water. This cruise was conducted in summer when the surface water was warming up. The SST of the cruise track on the 24 th of June and 24 th of July 2016 were selected by interpolating the SST data from AVHRR (Advanced Very High Resolution Radiometer, http://apdrc.soest.hawaii.edu). The 24 th of July was chosen because it is the middle date of this cruise track. The maximum SST difference (ΔT) is~4 C between these 2 months (Fig. 2d). Although the ΔT is a snapshot of the temperature change between June and July, it indicates that in the open water along the cruise track, the 4 C warming of the surface water at this latitude will lead to a maximum 15% decline in the N 2 O solubility according to the study on N 2 O solubility carried out by Weiss and Price (1980). However, the air-sea exchange was ongoing during this month, which replenished the surface water with atmospheric N 2 O. This exchange process is slower than that of air-sea heat exchange and therefore leads to seasonal N 2 O saturation anomalies (Butler et al. 1989). However, previous studies have shown that the oversaturation resulting from this seasonal effect should be less than 4% (Nevison et al. 1995). Therefore, for the open water on the cruise track, if a maximum seasonal effect of 4% was taken into account, the baseline of the SA is 0-6%, which indicates that the study region is a weak N 2 O source as a whole.
To further discuss the distribution pattern and the corresponding mechanisms, the cruise track was divided into three legs according to the characteristics of the study regions it passed through. Leg 1: The cruise track was located in the open subpolar ocean (between the starting point and 3000 km) and passed through part of the Okhotsk Sea, the Northwest Pacific, and the Aleutian Basin. The N 2 O in the surface water in these regions is generally regulated by hydrographic processes. Leg 2: The cruise track was located on the continental shelf (between 3000 and 4700 km), where possible sediment emissions may contribute N 2 O to the surface water (Hirota et al. 2009), and it passed through the Bering Sea continental shelf, Bering Strait, and Chukchi Sea. Leg 3: The cruise track was located in the open Arctic Ocean (north of 4700 km), where the N 2 O in the surface sea water is significantly influenced by the ice melting process. The mechanisms of N 2 O SA maximum formation are discussed according to these categories in the following sections.

Open sub-Arctic basin (leg 1)
Three maxima were observed on leg 1. The first maximum (maximum 1) was observed at 339 km between the Okhotsk Sea and the Pacific. The N 2 O concentration in the surface water of this maximum was 15.7 nM, which was~2 nM higher than that of the adjacent seas, and the N 2 O SA value was 20%. The water depth at this N 2 O SA maximum was more than 1000 m, and the SSS maximum present at this location suggested that this N 2 O oversaturation should not be contributed by continental shelf emission or terrigenous riverine discharge. It was reported that the Kamchatka Current flows southward into the Okhotsk Sea via paths between the Kuril Islands in this region (Tomczak and Godfrey 1994). This inflow current turns northeast after entering the Okhotsk Sea, which results in the southwestward transport of surface water due to the Coriolis effect and subsequent upwelling, which may bring N 2 O-enriched deep water to the surface and result in the N 2 O SA maximum 1. Deeper water enriched in N 2 O is generally also enriched with CO 2 , depleted in oxygen, and characterized by a relatively low temperature and high salinity. These signals were all observed at this location (Fig. 2a,e,f), indicating that upwelling may be a source that contributes to the N 2 O SA maximum 1.
Maximum 2 was a relatively weak maximum of approximately 13% (Fig. 2c). The distribution of pCO 2 over the same spatial range showed a remarkable increase in pCO 2 from 226 to 407 ppm north of 1000 km, and the oxygen saturation and SST values also showed turning points. Sharp changes in physical and chemical signals generally mark the presence of divergence, convergence or fronts. Therefore, a front may exist at this location. Butler et al. (1989), Grefe et al. (2018), and Rees et al. (1997) also show that N 2 O saturation may be elevated or variable at fronts. Elevated saturation levels may result from the water mixing process at the frontal structure (Rees et al. 1997), where a sharp change in temperature leads to a sharp change in gas solubility, or due to the presence of eddies (Rees et al. 1997), which uplift N 2 O-rich deeper water and lead to surface water N 2 O oversaturation. However, this variation or elevation of N 2 O saturation is not identified at all fronts that have been observed (Rees et al. 1997). With only limited high-resolution data available, the dynamics of N 2 O at fronts need to be further addressed.
The 3 rd maximum along leg 1 was located at the northwestern end of the cruise track and was among the most significant N 2 O SA maxima of the whole cruise track. The observed N 2 O SA at this maximum was~60%. This N 2 O SA maximum corresponded to a salinity and temperature minimum, which were approximately 1 C and 1.4 lower than the SST and SSS values of the adjacent region, respectively (Fig. 2a). Moreover, the corresponding pCO 2 at this location was the lowest of Leg 1, with a minimum of approximately 170 ppm. However, the DO did not show a remarkable supersaturation maximum. Decreasing SST and SSS values indicated that the cruise track may have coincided with the Kamchatka Current (Rogachev 2000), which may be enriched in N 2 O. When it flows southward, it becomes warmer and shows oversaturation of N 2 O. Another possibility is upwelling of subsurface water, which resulted from the passage of a cyclone over this location. The cyclone may have uplifted colder and N 2 O-enriched water to the surface layer, leading to surface water N 2 O oversaturation. However, subsurface water generally contains high concentrations of CO 2 , which may also lead to CO 2 oversaturation. One possible explanation for the pCO 2 minimum observed at this location may be the rapid consumption of CO 2 by photosynthesis.

Continental shelf (leg 2)
Over the continental shelf, four N 2 O SA maxima (4, 5, 6, and 7) ranging between 15% and 37% can be identified along leg 2. Since this leg is on the continental shelf, these maxima may be related to continental sediment emissions. Hirota et al. (2009) observed increasing concentrations of N 2 O with depth in the Bering Sea and concluded that the N 2 O over the Bering Sea continental shelf may be related to denitrification. The DO concentration in the water column is not low enough to initiate denitrification; thus, the sediment may provide a location for denitrification, which was previously observed in the Bering Sea sediment (Horak et al. 2013). Our results show that the SA maxima can be divided into two categories according to the corresponding pCO 2 and DO distribution patterns. The pCO 2 and DO distribution patterns are opposite in maxima between these two groups, with maximum 5 presenting a pCO 2 maximum and DO minimum and maxima 4, 6, and 7 presenting a pCO 2 minimum and DO maximum.
At nitrous oxide SA maximum 5 (SA~28%), the SST suddenly dropped from 11.3 C to 1.7 C, which was also shown by the NOAA AVHRR data (http://apdrc.soest.hawaii.edu/ data/data.php). A large area of low SSTs was observed north of St. Lawrence Island, and the SSS increased from 30.1 to 32.6; thus, this region presented the lowest temperature and highest salinity on the ice-free continental shelf. The corresponding pCO 2 maximum,~537 ppm, was the absolute maximum and the only source of CO 2 to the atmosphere along this cruise track. In addition, the only instance of DO undersaturation (~88%) was also observed at this location (Fig. 2f). This cold water region was reported by Nihoul (1986), who observed a front and upwelling west of the front. This upwelling can bring nutrient-rich bottom water to the surface in the western region north of St. Lawrence Island. Hence, it can be concluded that the upwelling at this site brought the bottom water, which was cold, contained high concentrations of N 2 O and CO 2 but was depleted of DO, to the near surface, resulting in the surface water N 2 O SA maximum.
The nitrous oxide SA maximum 6 (SA~21%) also corresponds to the SST minimum of~4.7 C and a salinity of~32.4. The variable SST and SSS around this maximum indicate that the cruise track probably crossed different water masses in the Chukchi Sea. According to Bates et al. (2006), the relatively fresh, warm, nutrient-depleted Alaskan Coastal Current is located in the east, and the relatively cold, salty, and nutrient-rich Anadyr Water (AW) is located in the west. Therefore, between these two water masses is a front, where the mixing process may enhance convection, bring N 2 O bottom water into the surface layer and result in this maximum. Furthermore, this bottom water is also nutrient-rich, promoting productivity at the front. Therefore, CO 2 is quickly consumed, and DO is produced, resulting in the pCO 2 minimum and DO maximum and resulting in pCO 2 and DO distribution patterns opposite those observed at maximum 5.
The nitrous oxide SA maximum 7, with an SA value of~37%, has a similar pCO 2 minimum and DO saturation pattern as maximum 6; however, the pCO 2 minimum and DO saturation maximum were the absolute minimum and maximum values of the whole cruise track, indicating the enhanced primary production present at this location. Moreover, the SST at this location experienced a remarkable decrease to approximately −0.4 C, which was close to the lowest SST value on the cruise track, and the SSS was as low as 21.0, which was also the lowest salinity of the whole cruise track. The N 2 O maximum of approximately 23 nM, together with the SA, indicated that this oversaturation does not result from the temperature effect. Regarding the sea ice distribution pattern (Fig. 2d), a high density of sea ice was observed for the first time on the cruise track at this site, indicating that this site was in the melting sea ice zone (MIZ).
Multiple processes and mechanisms may contribute to N 2 O oversaturation. Hydrographic processes are one of the processes leading to oversaturation. The existence of the eastward-flowing currents in the Chukchi Sea and the westward-flowing Chukchi Slope Current (Fig. 1) together may result in divergence at this location and bring subsurface water with high N 2 O concentrations (Zhang et al. 2015;Wu et al. 2017) to the surface layer, resulting in N 2 O oversaturation. However, the observed SST and SSS minima also indicate an ongoing ice melting process, which may dilute the surface water N 2 O concentration and even lead to N 2 O undersaturation. Wu et al. (2017) and Zhang et al. (2015) both observed N 2 O undersaturation at this location. This discrepancy between these studies may result from different observation times. Both Wu et al. (2017) and Zhang et al. (2015) carried out studies in late summer. When ice melting prevails, surface water is diluted by meltwater, and stratification develops, which nearly inhibits the exchange between subsurface and surface water. Hence, eddy diffusion can contribute only a very limited amount of N 2 O from subsurface water to surface water, and the surface water remains undersaturated or near equilibrium in terms of N 2 O. However, during early summer, when our underway observation was carried out, the ice melting process may not be significant enough to dilute the surface water, and the surface water is still in a state of N 2 O oversaturation.
Another possible process contributing to this oversaturation may be in situ production near the surface water or sea ice. A similar phenomenon of N 2 O oversaturation in the MIZ was also observed in Baffin Bay by Kitidis et al. (2010), who suggested it was the result of organic matter mineralization. Moreover, they also observed high N 2 O concentrations under multiyear ice, and this N 2 O may be emitted to the atmosphere when the sea ice retreats. Studies have suggested that production beneath sea ice may occur. Rysgaard and Glud (2004) suggested that denitrification in sea ice can also be a possible source for N 2 O; Baer et al. (2014) found nitrification to be stronger in winter than in summer, which may also result in N 2 O trapped beneath the sea ice.
In summary, the N 2 O SA maximum on leg 2 may result from uplift of N 2 O-rich bottom water over the continental shelf to the surface layer by upwelling or convection accompanying the frontal mixing process. Furthermore, the denitrification and nitrification processes in ice-covered areas may also be possible mechanisms that need to be explored.

Open Arctic Ocean (leg 3)
The nitrous oxide SA along leg 3 was close to zero, indicating that the N 2 O in the surface water throughout most of this leg was near equilibrium with the atmosphere. With the presence of sea ice, the SST remained relatively stable and ranged between −1.0 C and 1.2 C, whereas the salinity ranged between 23.60 and 27.60 (Fig. 2d). Except for N 2 O maximum 8, the N 2 O SA value along this leg centered on zero showed no obvious source or sink characteristics overall. The corresponding pCO 2 of this leg was approximately 370 ppm, and the DO in the surface water was in equilibrium with the atmosphere, suggesting that the primary production along  this leg was also low. The low primary production and net community production were described by Tremblay et al. (2015), and these conditions may not support significant N 2 O production in the subsurface water. The N 2 O SA maximum of approximately 34% corresponds to a pCO 2 minimum of approximately 313 ppm, which may also be MIZ-related N 2 O oversaturation. N 2 O undersaturation occurred with an SA value of approximately −9% at 4800 km, which corresponds to one of the temperature minima (−1.0 C) and a low salinity of approximately 25.02. Studies have shown that the N 2 O concentration in Arctic sea ice is only~6 nM (Randall et al. 2012); therefore, when the ice melts, it may dilute the surface water and result in surface water N 2 O undersaturation before returning to equilibrium again. Compared with the first two legs, leg 3 shows a different N 2 O SA distribution pattern, and the study area on this leg is the largest area with a near-equilibrium N 2 O concentration. The ice melting process is probably the main reason for the N 2 O saturation state observed in the surface water along this leg.

Air-sea flux along the cruise track
According to Fenwick et al. (2017), the air-sea flux evaluated from instantaneous wind speed data leads to a positive bias. Our result calculated from instantaneous wind speed also shows overestimation of the air-sea flux (result not shown). Therefore, to derive a more reasonable seasonal flux, the weighted air-sea exchange coefficient (Reuer et al. 2007) was calculated based on the functions suggested by Ho et al. (2006), Nightingale et al. (2000) and Van Der Loeff et al. (2014). With the weighted air-sea exchange coefficient, the air-sea fluxes derived from these methods are shown in Fig. 4. The gray dot and black dots are the results calculated from Ho et al. (2006) and Nightingale et al. (2000), respectively, which showed no difference (Fig. 4), whereas the air-sea N 2 O flux north of 4700 km with sea ice presence was also evaluated using the method put forward by Van Der Loeff et al. (2014), which resulted in values slightly lower than those of the above two methods. The average fluxes of different legs are shown in Table 1. As the result shows, the open sub-Arctic Ocean presents an average sea-to-air flux of~2.3 AE 2.2 μmol m −2 d −1 , while the continental shelf presents a higher average sea-to-air flux of approximately 3.7 AE 5.0 μmol m −2 d −1 . These values show that both the open sub-Arctic Ocean and the continental shelf are N 2 O sources, but the latter is a more important N 2 O source, especially at N 2 O maximum 5, where the highest N 2 O sea-to-air flux was identified. In the open Arctic Ocean, the air-sea fluxes derived from the equations suggested by Van Der Loeff et al. (2014), Ho et al. (2006) and Nightingale et al. (2000) are similar, and all three parameterizations indicate that the open Arctic Ocean shows no obvious source and sink characteristics during summer.
Compared with the results reported by Hirota et al. (2009) andFenwick et al. (2017), the air-sea flux values for the Bering Sea and Chukchi Sea reported here are higher than those reported by Fenwick et al. (2017) but similar to those reported by Hirota et al. (2009). A possible explanation for the difference between our result and that of Fenwick et al. (2017) over the continental shelf is that the high concentration of the surface water maximum may not be captured by the station sampling method since the distribution of N 2 O in the continental shelf sea water is inhomogeneous. However, these maxima are best captured by the high-resolution underway method. The similarity of the open Arctic Ocean N 2 O air-sea fluxes between Fenwick et al. (2017) and this study may be due to the relatively homogeneous distribution of surface water in this region; thus, the station sampling method can provide representative results.

Conclusions
In this study, N 2 O and CO 2 along a cruise track between the Bering Sea and Canadian Basin were observed using an underway method based on a cavity ring down system. The surface water along the cruise is a N 2 O source as a whole, and the results identified numerous N 2 O SA maxima and only one obvious region of undersaturation (~90% saturation), which may have resulted from dilution by melting ice. Several processes or formation mechanisms contribute to these N 2 O SA maxima: (1) certain hydrographic structures, such as convergence, divergence or fronts, which lead to elevated N 2 O SA values; (2) convection over the continental shelf, which brings N 2 O-rich bottom water to the surface and results in a N 2 O SA maximum and increased pCO 2 and decreased DO values; and (3) possible near-surface water N 2 O production. The third phenomenon can be observed in the MIZ, although further study is needed to reveal its mechanism. The air-sea flux evaluation results showed that the continental shelf presents the highest sea-to-air N 2 O flux along the cruise track, which is likely associated with N 2 O-rich deep water brought to the surface by upwelling and convection. The Bering Sea and Chukchi Sea continental shelf may be an important N 2 O source to the atmosphere, whereas the open Arctic Ocean did not show source or sink characteristics during the study season due to sea ice melting and the subsequent stratification of the water column.