3.1.1 Nitrate mass concentrations

Skin layer and atmospheric measurements of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations during the January 2017 ISOL-ICE campaign at DML are presented in Fig. 3. Aerosol mass concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ range from 0.5 to 19 ng m⁻³ and show a downward trend throughout January 2017 (R²=0.55; $p=<\mathrm{0.001}$; Fig. 3). In contrast, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in the skin layer increase during the month from 136 to 290 ng g⁻¹. Nitrate mass concentrations in both snow pits, which range from 23 to 142 ng g⁻¹, are substantially lower than those in the skin layer. Our measurements agree well with published measurements of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in snow pits at DML (Weller et al., 2004). While our January 2017 observations of atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations are 20–30 ng g⁻¹ lower than those observed in 2003 by Weller and Wagenbach (2007), which could be due to interannual variability of atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations, which varied by 30 ng g⁻¹ over summer between 2003 and 2005.

Figure 3January 2017 time series in Dronning Maud Land (DML) of (a) daily precipitation, (b) hourly wind direction and wind speed, (c) atmospheric and skin layer δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$, and (d) atmospheric and skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration. Error bars in panels (c)–(d) indicate the spatial variability in the site determined by multiple skin layer samples collected on 28 January 2017. The spatial variability exceeds the instrumental error, which is smaller than the symbol size. Meteorological data source: University of Utrecht (AWS9; DML05/Kohnen; $\mathrm{75}{}^{\circ }{\mathrm{00}}^{\prime }$ S, $\mathrm{00}{}^{\circ }{\mathrm{00}}^{\prime }$ E/W; ∼2900 m a.s.l.). Precipitation data source: RACMO2 (https://doi.org/10/c2pv, van de Berg et al., 2019).

Download

A comparison of Dome C and DML observations in skin layer, aerosol, and depth profiles is illustrated in Fig. 4, while archived ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values are reported in Table 2. Compared to Dome C, average annual atmospheric, skin layer, and snow pit mass concentrations are lower at DML, in agreement with observational and modelling studies where higher ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations are found at lower snow accumulation sites (Erbland et al., 2013). The ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration profile in the upper 50 cm of the snowpack at Dome C shows an exponential decrease with depth and becomes relatively constant at 35 ng g⁻¹ at 20 cm compared to 160–1400 ng g⁻¹ in the skin layer (Figs. 1 and 4; Erbland et al., 2013; Frey et al., 2009). While the highest ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in the snowpack at DML are also found in the skin layer, the mass concentration profile exhibits a different pattern. The sharp decrease in ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration occurs in the top ∼5 mm at which point the snow pit records interannual variability in the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration. Nitrate mass concentrations in DML snow pits exhibit a maximum in summer and minimum in winter.

Figure 4Comparison of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at Dronning Maud Land (DML) and Dome C in January 2017. ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration in (a) atmosphere, (b) skin layer, and (c) depth profiles. Insert: depth profile of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration highlighting seasonal variability. δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in (d) atmosphere, (e) skin layer, and (f) depth profiles. Grey bars indicate summer seasons for DML depth profiles.

Download

Table 2Summary of observed and simulated archived, aerosol, and skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations, δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ composition, and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass fluxes at Dronning Maud Land (DML) and Dome C. Abbreviation n.d.: no data. Base case refers to the TRANSITS simulation with a snow accumulation rate of 6 cm yr⁻¹ w.e. and an e-folding depth of 10 cm, while the 5 cm EFD case refers to a TRANSITS simulation with an observed snow accumulation rate that varied year to year between 6.0 and 7.1 cm yr⁻¹ w.e. and an e-folding depth of 5 cm.

^* Expected values for a site with an accumulation rate of 6 cm yr⁻¹ w.e. based on the spatial transect of Erbland et al. (2015).

Download Print Version | Download XLSX

While the precision of the IC measurement of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is better than 2 %, the spatial variability at DML in ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the skin layer exceeds this. During the sampling campaign, five skin layer samples were taken from an area of ∼2500 m² at the flux site (snow surface had sastrugi of up to 10 cm) to understand how representative the snow pit mass concentrations are of the greater study area. We found that the spatial variability in ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at DML was 10 % and 17 %, respectively (Fig. 3c–d). At Dome C, the spatial variability in ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations was between 15 % and 20 %. We note that this variability includes the natural spatial variability and the operator sampling technique.

Simulated TRANSITS results for the base case and 5 cm EFD case scenarios at the air–snow interface are illustrated in Fig. 5 along with TCO data (Fig. 5a). In the atmosphere, the TRANSITS model is forced with the smoothed profile of year-round atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ measurements from the DML site (Weller and Wagenbach, 2007) where the highest mass concentrations are in spring and summer with a maximum of 80 ng m⁻³ in November and a minimum of 2 ng m⁻³ in winter (Fig. 5b). Overall, the simulated values in the base case scenario are higher than the 5 cm EFD case in summer and autumn and converge to similar values in winter. In the skin layer, the simulated annual cycle of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations steadily rise in spring and reach a peak in January when they begin to decline to the lowest mass concentration in winter (Fig. 5d). The simulated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in the skin layer are an order of magnitude greater than our observations in January. The discrepancy between the significantly higher simulated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations than observations in the skin layer was also found at Dome C. Erbland et al. (2015) suggested that this discrepancy could be related to a sampling artefact, snow erosion, or a modelled time response to changes in past primary inputs. We provide an alternative explanation for the extremely high simulated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in the skin layer using daily measurements of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration in diamond dust and hoar frost collected from Polyvinyl chloride (PVC) sheets at Dome C in summer 2007/2008, i.e. new deposition. New deposition of diamond dust had ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations of up to 2000 ng g⁻¹, which is 4 times greater than that observed in natural snow from the skin layer at the same time (Fig. S4). Similarly, new deposition of hoar frost had ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations of up to 900 ng g⁻¹, which is 3 times greater than the skin layer snow. The formation of surface hoar frost occurs by co-condensation, i.e. the simultaneous condensation of water vapour and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ at the air–snow interface. Recent modelling suggests that co-condensation is the most important process explaining ${\mathrm{NO}}_{\mathrm{3}}^{-}$ incorporation in snow undergoing temperature gradient metamorphism at Dome C (Bock et al., 2016). Diamond dust can also scavenge high concentrations of HNO₃ at Dome C (Chan et al., 2018). Furthermore, the top layer of the snowpack is only 1 mm thick in the TRANSITS model, whereas our observations of the skin layer are 5 mm thick. Due to the photochemical loss of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations with depth, the highest ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations are expected in the top 1 mm layer, which is the layer best in equilibrium with the atmosphere. Here, extremely high mass concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ from new deposition from diamond dust and hoar frost are also found. In summary, it is likely that we do not measure such high ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in hoar frost and diamond dust in the skin layer because of sampling artefacts or blowing snow, which can dilute or remove the diamond dust and hoar frost. It is interesting to note that the higher simulated values in the skin layer do not impact the simulated depth profiles.

Figure 5ISOL-ICE observations and simulated annual cycle of skin layer and atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at Dronning Maud Land (DML) from the base case and 5 cm EFD case TRANSITS model simulations for January 2017. (a) Total column ozone: NIWA Bodeker combined dataset version 3.3 at DML averaged from 2000 to 2016 (http://www.bodekerscientific.com/data/total-column-ozone, last access: 15 January 2019). (b) Atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations are observations from Kohnen Station (Weller and Wagenbach, 2007) that are used as input into the model. ISOL-ICE observations and TRANSITS simulations of (c) atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$, (d) skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration, and (e) skin layer δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$.

Download

3.1.2 Isotopic composition of nitrate

The seasonal evolution of observed and simulated air–snow δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values are presented in Figs. 3 and 5, respectively. Atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ ranges from −49 ‰ to −20 ‰ at DML and −9 ‰ to 8 ‰ at Dome C during the January campaign and is depleted with respect to the skin layer, which ranges from −22 ‰ to 3 ‰ at DML (Fig. 3). The simulated atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in the base case for January are greater than our measurements, while the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in the 5 cm EFD case fall within the range of observations (Fig. 5). The annual cycle of simulated atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ for the 5 cm EFD case shows a 50 ‰ dip in spring to −42 ‰ from winter values, which coincides with the simulated atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration increase in spring (Fig. 5c). The highest simulated atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values (7 ‰) occur in winter, for both scenarios. While the simulated skin layer δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in January for the base case are ∼10 ‰ higher than our highest observations for that month, but the average January value in the 5 cm EFD case (−7 ‰) falls in the range of observed values (−10 ‰) (Fig. 5e). For the 5 cm EFD case, they begin to decrease by 30 ‰ in spring at the same time as atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values decrease. In October and November, the skin layer δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values begin to rise up to −11 ‰ in February in the 5 cm EFD case.

The δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in both snow pits at DML show extremely good reproducibility, with depth indicating there is little spatial variability within 1 km at the site (Fig. 4). The depth profiles of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values at DML exhibit large variability between seasons (−3 ‰ to 99 ‰) with more enriched values in spring and summer with respect to winter (Fig. 4). In comparison, the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in snow pits at Dome C do not preserve a seasonal cycle. However, in parallel with the exponential decay of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations with depth at Dome C, there is a strong increase in the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ with depth. At Dome C, δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ increases up to 250 ‰ in the top 50 cm; this increase is weaker at DML (up to 100 ‰ in the top 30 cm at which point seasonal cycles are evident). Although no annual cycle is preserved in the snowpack at Dome C, the year-round measurements of atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ show a decrease during sunlit conditions in spring and summer (Fig. 1). While the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the skin layer has a spring minimum that increases to a maximum at the end of summer (Fig. 1). Skin layer δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ is about 25 ‰ higher than atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$.

Sensitivity results of the depth profiles for the base case and 5 cm EFD case scenarios are discussed in Sect. 3.5.1, and we refer the reader to that section for an in-depth discussion of the TRANSITS sensitivity tests. We briefly describe differences between the depth profiles of the base case and 5 cm EFD case here to set the scene for the discussion. Overall, TRANSITS modelling shows that (i) the simulated δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in the base case scenario are higher than the 5 cm EFD case, (ii) the 5 cm EFD case falls within the range of observations for δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ but is significantly higher than the observed ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations, and (iii) TRANSITS modelling simulations using the observed e-folding depth of 5 cm (Sect. 3.3.2) are good fit with δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ observations.

3.1.3 Snow pit accumulation rate and nitrate mass fluxes

Annual layer counting of Na⁺ layers shows that snow pit A spans 8 years from autumn 2009 to summer 2017 and snow pit B spans 9 years from summer 2008 to summer 2017 with an age uncertainty of ±1 year at the base of the snow pit. The mean snow accumulation rate for the snow pits is estimated to be 6.3±1.4 cm yr⁻¹ w.e., consistent with published accumulation rates of 6.0–7.1 cm yr⁻¹ w.e. from snow pits and ice cores from DML (Sommer et al., 2000; Hofstede et al., 2004; Oerter et al., 2000).

Taking the simple mass balance approach, a schematic of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass fluxes for two scenarios is illustrated in Fig. 6. Scenario 1 is an average annual budget for DML (Fig. 6a). As the atmospheric campaign did not cover an entire annual cycle, we use estimates of atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass fluxes at DML reported by Pasteris et al. (2014) and Weller and Wagenbach (2007) of 43 and 45 pg m⁻² s⁻¹, respectively, as year-round dry deposition fluxes. Due to the linear relationship of ice core ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations with the inverse accumulation, the authors assume that the magnitude of the dry deposition flux is homogenous over the DML region. Mean annual mass concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in our snow pits suggest a total ${\mathrm{NO}}_{\mathrm{3}}^{-}$ deposition mass flux of 110 pg m⁻² s⁻¹ and therefore a wet deposition mass flux of 65 pg m⁻² s⁻¹.

Figure 6Schematic of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass fluxes at Dronning Maud Land (DML) for (a) annual mean scenario and (b) January scenario.

Download

However, at relatively low snow accumulation sites where photolysis drives the fractionation of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ from the surface snow to atmosphere (Frey et al., 2009), it is necessary to take into account the skin layer in the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass flux budget as this air–snow interface is where air–snow transfer of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ takes place. In scenario 2, we utilize the available ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations measured in aerosol, skin layer, and snow pits from the ISOL-ICE campaign to estimate the mass flux budget for January 2017 (Fig. 6b). The dry deposition mass flux of atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ during January 2017 at DML averages 64±38 pg m⁻² s⁻¹ (Table S2) and is greater than the annual mean flux estimated by Pasteris et al. (2014) and Weller and Wagenbach (2007), which is to be expected given the higher atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in summer (Fig. 5). Our wet deposition mass flux of 296 pg m⁻² s⁻¹ is also greater than the wet deposition flux calculated for the greater DML region by Pasteris et al. (2014). Like Dome C, the greatest deposition flux of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is to the skin layer, and it is 360 pg m⁻² s⁻¹; however only 110 pg m⁻² s⁻¹ of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is archived. Considering the active skin layer, only 30 % of deposited ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is archived in the snowpack, while 250 pg m⁻² s⁻¹ is re-emitted to the overlaying atmosphere.

Furthermore, the TRANSITS-simulated archived ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass flux at DML of 210 pg m⁻² s⁻¹ for the base case and 480 pg m⁻² s⁻¹ for the 5 cm EFD case overpredict the observed ${\mathrm{NO}}_{\mathrm{3}}^{-}$ archived mass flux due to the higher simulated archived ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations. Interestingly, the simulated archived mass flux at Dome C (88 pg m⁻² s⁻¹) is lower than DML, yet the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ deposition flux to the skin layer in January at Dome C is similar to DML. We continue our discussion focusing on the recycling and redistribution of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ that occurs in the active skin layer, emphasizing its importance.

3.2.1 Wet and dry deposition

Here we discuss the various processes in which ${\mathrm{NO}}_{\mathrm{3}}^{-}$ can be deposited to the skin layer at DML. Firstly, we first look at atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ deposition in relation to the source region of the air mass. The mean annual wind direction at the site is 65^∘ within the clean air sector (Figs. 3 and S5). There are two excursions from the predominant wind direction. The first excursion is between 19 and 22 January, where the wind direction switches to the southwest, i.e. atmosphere transport from the plateau. We do not see elevated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations during this period nor do we see a marked difference in isotopic signature that is similar to Dome C at this time (Fig. 4). This, in line with air mass back trajectories (not shown), suggests that transport of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ re-emitted from inland sites of the Antarctic, carrying a distinctively enriched δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signature, did not influence DML during our campaign. The second excursion occurs during a few short periods when the wind direction switches upwind of the station; however, there are no spikes in the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration or a change in the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signature, and so we can also rule out any downwind contamination from the station.

Secondly, we use modelled daily precipitation at the nearest Regional Atmospheric Climate Model (RACMO2; Van Meijgaard et al., 2008) grid point (75.0014^∘ S, 0.3278^∘ W; Fig. 3a) to identify the influence of cyclonic intrusions of marine air masses to wet deposition of ${\mathrm{NO}}_{\mathrm{3}}^{-}$. We observe that some peaks in the skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration are accompanied by fresh snow laden with relatively high sea salt aerosol mass concentrations and atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations, for example on 1, 13, and 18 January 2017 (Fig. S6). However, on other precipitation days, we observe lower atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and higher skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations that could be a result of HNO₃ scavenging. With only 1 month of data it is difficult to see the impact of wet deposition on the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration in the skin layer; i.e. whether fresh snowfall dilutes the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration in the skin layer or whether it scavenges HNO₃ (gas-phase), resulting in higher mass concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the skin layer.

Thirdly, we investigate daily changes in the atmospheric and skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ over the campaign to see the influence of dry deposition, by adsorption of atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the snow surface, on the high mass concentrations observed in the skin layer. Temporal variation in the mass concentration and isotopic signature of aerosol and surface snow at DML over January 2017 suggests atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is the source of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the skin layer. Throughout the month, the increase in the skin layer mass concentration of summer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ appears to be closely related to the decrease in the atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations (Fig. 3). There is a lag between atmospheric and skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$; i.e. atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations precede skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations by a day or two; however a longer time series is required to confirm this. The lag suggests that atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is a source of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the skin layer, in line with Dome C where the underlying snowpack is the dominant source of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the skin layer via photolytic recycling and redeposition. Furthermore, as atmospheric ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is deposited to the snow surface, ¹⁵N is preferentially removed first, leaving the air isotopically depleted relative to the isotopically enriched snow (Frey et al., 2009). Figures 3 and 4 illustrate that the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the atmosphere is depleted with respect to the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the skin layer snow. In the short time series, there are some periods where the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the snow and atmosphere are in phase, for example, 3–13 January 2017. During other periods, the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the snow and atmosphere switch to being out of phase, emphasizing ${\mathrm{NO}}_{\mathrm{3}}^{-}$ isotopic fractionation during those periods. Both HNO₃ and peroxynitric acid (HNO₄) can be adsorbed to the snow surface in tandem (Jones et al., 2014), and although we have no direct measurements of these during the campaign, based on previous studies we suggest that HNO₃ is dominantly adsorbed to the skin layer (Jones et al., 2007; Chan et al., 2018).

We conclude that ${\mathrm{HNO}}_{\mathrm{3}}^{-}$ scavenging, adsorption, and cyclonic intrusions of marine air masses deliver ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the skin layer at DML in summer. During the campaign, deposition is not influenced by the transport of air masses from the polar plateau, which carry a distinct atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signature. Interestingly, model results from Zatko et al. (2016), which account for transport of snow-sourced ${\mathrm{NO}}_{\mathrm{3}}^{-}$ emissions and deposition, show that the deposition of recycled ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to snow is lowest on the East Antarctic Plateau including the high-elevation DML region.

3.2.2 Temporal variability in nitrate deposition

The simulations in Fig. 5 and observations in Fig. 1 describe the seasonal evolution of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ deposition to the skin layer from the atmosphere at DML (Sect. 3.1.1 and 3.1.2). The annual cycle is consistent both (i) spatially across a vast area of Antarctica, i.e. South Pole, Dome C, Halley Station, and Neumayer Station (McCabe et al., 2007; Wolff et al., 2008; Erbland et al., 2013; Frey et al., 2009; Wagenbach et al., 1998), and (ii) temporally over last 7 years at Dome C (Fig. 1) (Erbland et al., 2015, 2013; Frey et al., 2009).

We also observe variability on shorter timescales. While not yet observed elsewhere on the Antarctic continent, over the short intensive sampling period at DML we observe significant variability in ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values that resembles a diurnal cycle. Over 4 h, the skin layer ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations varied by 46 ng g⁻¹, the skin layer δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ by 21 ‰, and the atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ by 18 ‰. Other coastal studies have attributed daily variability to individual storm events (Mulvaney et al., 1998; Weller et al., 1999). The sampling duration in this study is too short to confirm any diurnal patterns, but it would be interesting to investigate this further in future work. We note that due to post-depositional processes (Sect. 3.3) any short-term signals observed in the skin layer are unlikely to be preserved.

3.3.1 Nitrate redistribution

In corroboration with earlier work on the East Antarctic Plateau, we find clear evidence of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ redistribution via photolysis at DML and confirmation of our hypothesis that UV photolysis is driving ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling at DML. Firstly, the highly enriched δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values of snow at DML (−3 ‰ to 99 ‰) and the highly depleted atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values at DML (−20 ‰ to −49 ‰) are unique to post-depositional processes at low accumulation sites in Antarctica (Fig. S7) and lie outside the range of known anthropogenic, marine, or other natural source end members (e.g. Hastings et al., 2013; Kendall et al., 2007; Hoering, 1957; Miller et al., 2017, 2018; Yu and Elliott, 2017; Li and Wang, 2008; Freyer, 1991; Savarino et al., 2007).

Secondly, denitrification of the snowpack is seen through the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signature, which evolves from the enriched snowpack (−3 ‰ to 99 ‰), to the skin layer (−22 ‰ to 3 ‰), to the depleted atmosphere (−49 ‰ to −20 ‰), corresponding to mass loss from the snowpack (Figs. 4 and S7). Denitrification causes the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ of the residual snowpack ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to increase exponentially as ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations decrease.

Thirdly, sensitivity analysis with TRANSITS, where photolysis is the driving process, is able to explain the observed snow pit δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ variability when the e-folding depth is taken into account (Sect. 3.5).

Fourthly, enrichment of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ is observed in the top 30 cm of the snowpack at DML indicating ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolytic redistribution at DML in the photic zone of the snowpack (Fig. 4). In the photic zone, the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ observations closely match the simulated δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values from TRANSITS (Sect. 3.5).

Lastly, calculated fractionation constants (¹⁵ε_app) using our simulated results from the TRANSITS model base case (¹⁵ε_app average of −19 ‰ for the top 30 cm, i.e. active photic zone with an e-folding depth of 10 cm) fall in the range of expected ¹⁵ε_app values ($-\mathrm{59}{<}^{\mathrm{15}}{\mathit{\epsilon }}_{\mathrm{app}}<-\mathrm{16}$ ‰) within the “transition zone” characterized by snow accumulation rates typical of sites located between the Antarctic Plateau and coast (5–20 cm yr⁻¹ w.e.; Erbland et al. 2015). However, the ¹⁵ε_app for the 5 cm EFD case (¹⁵ε_app average of −11 ‰) is lower than predicted for a site with the same snow accumulation rate, highlighting the sensitivity of e-folding depth on ${\mathrm{NO}}_{\mathrm{3}}^{-}$ redistribution. Erbland et al. (2013) noted that uncertainties in the ¹⁵ε_app for snow pits in the transition zone were greater than coastal and plateau zones, indicating that the assumed single-loss Rayleigh model is not appropriate for transition zones. The discrepancy between our observed (12 ‰) and simulated (−19 ‰ and −11 ‰ for the base case and 5 cm EFD case, respectively) ¹⁵ε_app is due to the higher snow accumulation rate, which preserves seasonality, and with a noisy signal, there is no pure separation of the loss processes assuming Rayleigh isotopic fractionation.

3.3.2 Nitrate recycling

Only three studies have attempted to quantify the degree of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling between the air and snow (Davis et al., 2008; Erbland et al., 2015; Zatko et al., 2016). Erbland et al. (2015) used the TRANSITS model to estimate that ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is recycled four times on average before burial beneath the photic zone at Dome C, similar to the findings of Davis et al. (2008) for the same site. Using the approach of Erbland et al. (2015), we find that ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is recycled three times on average before it is archived at DML for the base case and two times on average for the 5 cm EFD case. Thus, a shallower e-folding depth reduces the recycling strength. Although these findings are consistent with spatial patterns of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling factors across Antarctica reported by Zatko et al. (2016), predictions for the DML region are almost double our estimates. As Dome C and DML lie on the same latitude (75^∘ S), incoming UV radiation (except for cloud cover) should not impact the efficiency of photolysis and thus recycling at the two sites. Below we provide some explanations for the weakened recycling at DML.

1. Higher snow accumulation rate. The TRANSITS modelling shows the influence of the snow accumulation rate on the depth profile of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$, including the preservation of a seasonal cycle at higher snow accumulation rates (Sect. 3.5.2). At low-accumulation sites, i.e. Dome C, the annual layer thickness is thinner so that ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in those layers is exposed to sunlight (and the actinic flux) and photochemical processes for longer, resulting in strong ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ enrichment in the snowpack. At DML, which has a higher snow accumulation rate than Dome C, the snow layers are buried more rapidly, leaving less time for HNO₃ to adsorb to the skin layer and less time for photolysis to redistribute snowpack ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the overlying air for re-adsorption to the skin layer. Therefore, photolysis-driven recycling of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ is largely dependent on the time that ${\mathrm{NO}}_{\mathrm{3}}^{-}$ remains in the snow photic zone.

2. Shallower e-folding depth. Based on measurements we derived an e-folding depth for DML ranging between 2 and 5 cm (Fig. S1). The standard deviations and variability between profiles (Table S3), reflect both systematic experimental errors and spatial variability in snow optical properties. e-folding depths at DML are similar to previous model estimates for South Pole (Wolff et al., 2002); however mean summer e-folding depths predicted for the DML region by Zatko et al. (2016) are overestimated by an order of magnitude. However, the e-folding depth at Dome C is considerably deeper, ranging between 10 and 20 cm depending on the snow properties (France et al., 2011). The origin of the reduced e-folding depth relative to Dome C is not known but is likely due to greater HUmic-LIke Substances (HULIS)/impurity content or different snow morphology (density and grain size of snow crystals) (Libois et al., 2013; Zatko et al., 2013; Brucker et al., 2010). In terms of published values, impurity concentrations are generally higher at DML, for example dust and major ion concentrations (Delmonte et al., 2019; Legrand and Delmas, 1988), due to proximity of marine sources. Yet station pollution is greater at Dome C (Helmig et al., 2020). Spatial patterns of modelled e-folding depths across Antarctica predict shallower e-folding depths in regions of relatively high black carbon concentrations located on the plateau in Antarctica (Zatko et al., 2016). In contrast, we observe a opposite pattern of higher black carbon concentrations and a deeper e-folding depth at Dome C compared to a shallower e-folding depth at DML. Therefore, the observed shallower e-folding depth at DML appears unrelated to black carbon concentrations as the modelling by Zatko et al. (2016) predicts a greater e-folding depth in the DML region where black carbon concentrations are lower. Furthermore, there is considerate variability in snow grain size across Antarctica. The larger e-folding depth in wind crust layers at Dome C is due to larger grain sizes in those layers (France et al., 2011). Snow grain size may be smaller at DML, which will increase scattering (Brucker et al., 2010), but further work is required to confirm if this is the dominate factor influencing the lower e-folding depth at DML. Sensitivity studies show that ${\mathrm{NO}}_{\mathrm{3}}^{-}$ impurities make a small contribution to the e-folding depth compared to scattering by snow grains which dominate (France et al., 2011; Chan et al., 2015; Zatko et al., 2013).

3. Lower photolysis rate. For the 1 to 14 January 2017 period, model estimates of $F_{{\mathrm{NO}}_{\mathrm{2}}}$ scaled approximately linearly with e-folding depth were 0.4, 1.0, and 1.9×10¹¹ molecule m⁻² s⁻¹ for e-folding depths of 2, 5, and 10 cm, respectively. Spatial variability in ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the top 30 cm of surface snow at DML based on snow pits A and B is on the order of 13 % inducing similar variability in the model estimates of $F_{{\mathrm{NO}}_{\mathrm{2}}}$. Estimates of $F_{{\mathrm{NO}}_{\mathrm{2}}}$ at Dome C, based on the same model during 1 to 14 January 2012, were larger with 1.2–7.3×10¹¹ molecule m⁻² s⁻¹ (Frey et al., 2013), mostly due to larger J(${\mathrm{NO}}_{\mathrm{3}}^{-}$) values observed above the surface as well as a larger e-folding depth (10 cm near the surface). It should be borne in mind that the above simple model estimates (Eq. 8) may significantly underestimate the real emission flux. Previous comparisons of $F_{{\mathrm{NO}}_{\mathrm{2}}}$ computed with Eq. (8) and $F_{{\mathrm{NO}}_x}$ measured at Dome C showed that observations can exceed model predictions by up to a factor 50 (Frey et al., 2015, 2013). While ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in snow, the surface actinic flux, and the e-folding depth were measured at the DML field site, the quantum yield of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolysis in surface snow (Φ${\mathrm{NO}}_{\mathrm{3}}^{-}$) was not but introduces significant uncertainty in the model estimates. Previous lab measurements on natural snow samples collected at Dome C showed Φ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to vary between 0.003 and 0.05 (Meusinger et al., 2014). As described above (Sect. 2.8) $J\left({\mathrm{NO}}_{\mathrm{3}}^{-}\right)$ used in Eq. (8) was calculated with Φ${\mathrm{NO}}_{\mathrm{3}}^{-}$ at −30 ^∘C ($=\mathrm{2}\times {\mathrm{10}}^{-\mathrm{3}}$) after Chu and Anastasio (2003), which is near the lower end of the observed range. Thus, up to half of the mismatch between Eq. (8) and Dome C observations can be explained by adjusting Φ${\mathrm{NO}}_{\mathrm{3}}^{-}$. Another factor contributing to larger fluxes and not included in Eq. (8) is forced ventilation.

   In the more sophisticated TRANSITS model, Erbland et al. (2015) found that the photolytic quantum yield was one of the major controls on archived flux and primary input flux at Dome C. Erbland et al. (2015) initially used a quantum yield of $\mathrm{2.1}\times {\mathrm{10}}^{-\mathrm{3}}$ at 246 K (France et al., 2011), but it underestimated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling and overestimated primary ${\mathrm{NO}}_{\mathrm{3}}^{-}$ trapped in snow. Adjusting the quantum yield to 0.026, within the range observed in the lab (Meusinger et al., 2014), gave more realistic archived δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values. However, at Dome C TRANSITS-simulated $F_{{\mathrm{NO}}_{\mathrm{2}}}$ fluxes were about a factor of 9–18 higher than observed $F_{{\mathrm{NO}}_x}$. Erbland et al. (2015) suggested that the discrepancy could result from the simplifications made in the TRANSITS model regarding the fate of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolysis products.

   Therefore, at DML, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolysis produces a lower snow emission flux of NO₂ to the atmosphere than at Dome C. This is due to (i) the shallower e-folding depth compared to Dome C, which implies reduced emission flux of NO_x, and (ii) the reduced UV exposure time of surface snow due to higher annual snow accumulation compared to Dome C. Furthermore, the large ¹⁵ε_app associated with ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolysis has been determined for snow at Dome C (Berhanu et al., 2014; Frey et al., 2009; Erbland et al., 2013) and DML. At both sites, δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ is enriched in the remaining skin layer. However, at DML, the ¹⁵ε_app is lower, which implies a weaker photolytic loss of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ associated with a higher snow accumulation rate. The lower snow emission flux of NO₂ and lower ¹⁵ε_app are evidence of a reduced recycling strength at DML relative to Dome C.

4. Lower nitrate uptake at warmer temperatures. The adsorption of HNO₃ on ice surfaces is temperature dependent with higher uptake at lower temperatures (Abbatt, 1997; Jones et al., 2014). Nitrate loss by evaporation is also dependent on temperature, with maximum ${\mathrm{NO}}_{\mathrm{3}}^{-}$ loss at higher temperatures (Thibert and Domine, 1998; Röthlisberger et al., 2000). The seasonal temperature difference at an individual site (i.e. DML or Dome C) could allow a seasonal dependence on the uptake and loss of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the skin layer, which results in the retention of a greater proportion of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in summer (Chan et al., 2018). However, there is only a relatively small temperature difference between Dome C and DML (Table 1), which is not enough to drive a large difference in HNO₃ uptake (Jones et al., 2014).

5. Lower export of locally produced nitrate. The degree of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling is also determined by atmospheric transport patterns across Antarctica. Export of locally produced NO_x on the Antarctic Plateau leads to greater enrichment in the depth profile of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ relative to the coast due to isotopic mass balance (Savarino et al., 2007; Zatko et al., 2016). Observations of enriched atmospheric δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at the coast suggest that NO_x has been sourced from in situ production on the Antarctic Plateau (Savarino et al., 2007; Morin et al., 2009; Shi et al., 2018). If there was less export of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ away from the DML site than Dome C, locally sourced NO_x would be redeposited back to the skin layer at the site, and the depth profile of the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ would not be as dramatically impacted as sites where there is substantial loss of ${\mathrm{NO}}_{\mathrm{3}}^{-}$.

3.5.1 Sensitivity of the ice core δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signal to e-folding depth

We measured an e-folding depth at DML between 2 and 5 cm, which is lower than that employed in the base case TRANSITS model simulation (10 cm). Furthermore, a range of e-folding depth values, between 3.7 and 20 cm, have been reported for Antarctica (Wolff et al., 2002; France et al., 2011; Zatko et al., 2016). Although the spatial trends predicted by the modelling of Zatko et al. (2016) are represented at Dome C and DML, an exception is the spatial pattern of the e-folding depth, where we observed a lower e-folding at DML than Dome C, opposite to what the model predicts. At DML, the Zatko et al. (2016) model results overestimate the archival time and recycling factor of ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and we suggest this is due to the lower observed than modelled e-folding depth. Furthermore, the positive bias of the TRANSITS base case simulation in archived δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at DML may be due to e-folding depth being smaller than at Dome C as indicated by direct observations. In order to test this assumption, the sensitivity of archived δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ to the e-folding depth parameter needs to be quantified, which has not been done before as far as we know. Zatko et al. (2016) modelled the e-folding depth over Antarctica and investigated the impact of snow-sourced NO_x fluxes but not on δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$.

Sensitivity results of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in the depth profiles for the base case and 5 cm EFD case scenarios are illustrated in Fig. 7. The ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ depth profiles for the base case show seasonal variability in the first year with a range of 380 ng g⁻¹ and 20 ‰, which decreases with depth to a range of 95 ng g⁻¹ and 10 ‰ in the fourth year. In comparison, in the 5 cm EFD case, the seasonality of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations in the first year ranges from 290 ng g⁻¹ and 40 ‰ to 75 ng g⁻¹ and 20 ‰ in the fourth year. Figure 7a shows that the e-folding depth has a large influence on the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ depth profile in terms of (i) depth of the photic zone and thus depth of the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ enrichment and (ii) the mean archived δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ value below the photic zone. A larger e-folding depth increases the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ enrichment in the photic zone and increases the archived mean δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ value. For example, an e-folding depth of 10 cm at DML gives δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ enrichment down to 30 cm and an archived mean δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ value of 125 ‰ in the snowpack compared to an e-folding depth of 20 cm, which enriches the snowpack down to 45 cm and more than doubles the archived mean δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ value to 320 ‰. Meanwhile, an e-folding depth of 2 cm gives minimal enrichment and a low archived mean δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ value of 25 ‰. In comparison to the base case simulation, which has an e-folding depth of 10 cm, a lower e-folding depth of 5 cm decreases the archived mean δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the snowpack to ∼50 ‰, closely matching our snow pit observations. Hence, a shallower e-folding depth in the range of that observed at DML, i.e. 2–5 cm can explain the more depleted δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ snow pit profile, relative to the base case simulation, as ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolysis occurs in a shallower depth. Therefore, e-folding depth knowledge is required to understand the sensitivity of archived δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at specific sites. We continue our sensitivity analysis using an e-folding depth of 5 cm and observed accumulation rate and refer to this scenario as our “5 cm EFD case”.

3.5.2 Sensitivity of the ice core δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signal to accumulation rate

The δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signal is also sensitive to the snow accumulation rate at DML. Plotted in Fig. 7b–c are the simulated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ depth profiles for accumulation rates of 2.5 and 11 cm yr⁻¹ w.e. for the 5 cm EFD case. As the accumulation rate increases, the annual layers of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ become thicker, the seasonal amplitude increases, the mean annual δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ value decreases, and there is less δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ enrichment in the photic zone (Fig. 7b). At very low snow accumulation rates, the seasonal cycle is smoothed, as in the case of Dome C (Fig. 7b). A similar pattern is observed for the simulated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations with depth: seasonal cycles of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations are more pronounced at higher snow accumulation rates, while interannual variability is smoothed at very low accumulation rates such as Dome C (Fig. 7c). The relationship between the snow accumulation rate and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ is nonlinear (Fig. 7b–c).

Even in the 5 cm EFD case, there is still an offset with the snow pit δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ depth profile below the active photic zone. To account for the offset, we investigated how the timing of snow deposition altered the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ depth profile. Rather than assuming a constant accumulation rate of 6 cm yr⁻¹ w.e., as in the 5 cm EFD case, we find that a variable snow accumulation rate, based on our observations from the snow pit, alters the depth of the summer and winter δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ peaks (Fig. 7b). Using the actual annual snow accumulation rate improves the model fit in the top 30 cm. Furthermore, the timing of the snow accumulation throughout the year has significant control over the amplitude of the seasonal δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ cycle. Snowfall at DML has a bimodal distribution with higher accumulation in austral autumn and early austral summer (Fig. S8). In Fig. 7d, we modified the timing of the snow accumulation during the year by depositing 90 % of the annual snowfall in (i) the first week of winter and (ii) the first week of summer, which represents the upper bound for snow accumulation in winter and summer, respectively. The remaining 10 % of the annual snowfall is distributed evenly across the rest of the weeks of the year. Summer snow accumulation results in a higher δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ enrichment compared to winter snow accumulation, as the exposure of summer layers to UV is longer, and thus ${\mathrm{NO}}_{\mathrm{3}}^{-}$ photolysis is stronger. Therefore, the timing and rate of snowfall can explain the misalignment between snow pit observations and 5 cm EFD case simulation, which shifts the depth and amplitude of the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ peaks in the depth profile.

On centennial to millennial timescales, the snow accumulation rate has varied in regions of Antarctica (e.g. Thomas et al., 2017), which could potentially modify the degree of post-depositional processing and thus impact the archival and temporal variability in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in ice cores. For example, the snow accumulation rate varied between 2.5 and 11 cm yr⁻¹ w.e. over the last 1000 years at DML (Sommer et al., 2000). At DML, higher snow accumulation rates would result in lower ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and more depleted δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in the skin layer, thus reducing the recycling strength and lowering the sensitivity of the UV proxy recorded in the ice over time and vice versa. TRANSITS modelling predicts that the upper and lower bounds of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values in a 1000-year ice core from DML that has an accumulation rate between 2.5 and 11 cm yr⁻¹ w.e. and e-folding depth of 5 cm to be between 30 ‰ and 140 ‰. Furthermore, δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values could range between 40 ‰ and 50 ‰ depending on the timing of snowfall and extreme precipitation events, which are known to play a dominant role in snowfall variability across Antarctica (Turner et al., 2019). At DML, snow pit observations suggest that the variation in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ between the polar day and polar night is 20 ‰. This seasonality is less than δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values expected for changes in snow accumulation rates over time. Therefore, any variation in snow accumulation will need to be accounted for in order to observe decadal-, centennial-, and millennial-scale trends in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$.

3.5.3 Sensitivity of ice core δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signal to TCO

Figure 8 shows the sensitivity of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ to variations in TCO. For each week, a constant amount of ozone (e.g. 100 DU) was added or subtracted from these present-day values. A decrease in TCO will increase UV radiation reaching the surface at an ice core site. As a result, stronger photolysis enhances ${\mathrm{NO}}_{\mathrm{3}}^{-}$ loss, redistribution, and recycling from the snowpack and ultimately decreases the archived ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentration. Furthermore, a decrease in TCO enriches the δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signature as the snow is exposed to a greater UV dose. We predict that a change of 100 Dobson Units (DU), i.e. the amount that ozone decreases each spring as a result of stratospheric ozone destruction processes, will result in a 10 ‰ change in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at DML. The variability in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ induced by TCO is less than the seasonal variability of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ recorded in the snow pit (20 ‰) and less than the predicted variability in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ due to changes in snow accumulation (110 ‰) or e-folding depth (100 ‰). As the above sensitivities have been evaluated individually, TCO depletion over many years may still be recoverable from ice core δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ if the other factors are constrained. For example, the e-folding depth at the DML site appears stable over the 8-year snow pit; the modelled δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ sensitivity of 100 ‰ represents an upper limit for changes in the e-folding depth ranging between 2 and 10 cm, and if the e-folding depth had changed recently, in an irregular manor, a regular annual cycle in δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ would not be evident (Fig. 4). Although additional studies of e-folding depth are required to confirm the variability in e-folding depth. The sensitivity of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ to TCO is greater at Dome C than DML (Fig. 8) due to the longer duration of surface snow exposure to UV radiation, stronger recycling and greater enrichment of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the photic zone. The sensitivity of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ to ${\mathrm{NO}}_{\mathrm{3}}^{-}$ recycling at DML is lower than expected from TRANSITS modelling for the same snow accumulation rate by Erbland et al. (2015), namely due to a lower e-folding depth than modelled, and thus the sensitivity of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ as a UV proxy is also lower than expected (Fig. 8). In addition, the oxygen isotopic composition of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ (δ¹⁷O-${\mathrm{NO}}_{\mathrm{3}}^{-}$) has been proposed as a proxy for stratospheric ozone at South Pole station (McCabe et al., 2007); however post depositional processes related to δ¹⁷O-${\mathrm{NO}}_{\mathrm{3}}^{-}$ need to be quantified to fully understand the sources and processes responsible for depositing and archiving the δ¹⁷O-${\mathrm{NO}}_{\mathrm{3}}^{-}$ signature in Antarctica.

Figure 8Expected response of archived δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ to changes in total column ozone at Dronning Maud Land (DML) and Dome C. Calculated sensitives represent an upper range as the real ozone hole lasts from September to November before recovery and not as modelled using the entire sunlit season. Archived DML δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ values were simulated using the observed accumulation rate, e-folding depth of 5 cm (5 cm EFD case), and present-day TCO values. These TCO values, which were used in all our calculations, vary weekly and can be found in Table S4. For each week, a constant amount of ozone (e.g. 100 DU) was added or subtracted from these present-day values. Dome C data source: Erbland et al. (2015).

Download

3.5.4 Implications for interpreting ice core δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$

Site-specific air–snow transfer studies provide an understanding of the mechanisms that archive δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ in ice cores, thus allowing for the interpretation of longer records of δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ from the site. Ice core records of archived ${\mathrm{NO}}_{\mathrm{3}}^{-}$ mass concentrations and δ¹⁵N-${\mathrm{NO}}_{\mathrm{3}}^{-}$ at DML are a result of two uptake and loss cycles that occur in the top 15 cm during sunlit conditions. While we do not observe further redistribution of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ in layers deeper than the photic zone, any further ${\mathrm{NO}}_{\mathrm{3}}^{-}$ diffusion within the firn or ice sections of an ice core can be constrained based on the temperature and snow accumulation rate at DML (Domine et al., 2008). This redistribution unlikely results in a loss of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ but could migrate ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to different layers, for example in acidic layers around volcanic horizons (Wolff, 1995).