Upper ocean distribution of glacial meltwater in the Amundsen Sea, Antarctica

15 Pine Island Ice Shelf, in the Amundsen Sea, is losing mass due to increased heat trans16 port by warm ocean water penetrating beneath the ice shelf and causing basal melt. Trac17 ing this warm deep water and the resulting glacial meltwater can identify changes in melt 18 rate and the regions most affected by the increased input of this freshwater. Here, op19 timum multi-parameter analysis is used to deduce glacial meltwater fractions from in20 dependent water mass characteristics (standard hydrographic observations, noble gases 21 and oxygen isotopes), collected during a ship-based campaign in the eastern Amundsen 22 Sea in February-March 2014. Noble gases (neon, argon, krypton and xenon) and oxy23 gen isotopes are used to trace the glacial melt and meteoric water found in seawater and 24 we demonstrate how their signatures can be used to rectify the hydrographic trace of glacial 25 meltwater, which provides a much higher resolution picture. The presence of glacial melt26 water is shown to mask the Winter Water properties, resulting in differences between 27 the water mass analyses of up to 4 g kg−1 glacial meltwater content. This discrepancy 28 can be accounted for by redefining the ”pure” Winter Water endpoint in the hydrographic 29 glacial meltwater calculation. The corrected glacial meltwater content values show a per30 sistent signature between 150 400 m of the water column across all of the sample lo31 cations (up to 535 km from Pine Island Ice Shelf), with increased concentration towards 32 the west along the coastline. It also shows, for the first time, the signature of glacial melt33 water flowing off-shelf in the eastern channel. 34 Plain Language Summary 35 Pine Island Ice Shelf in the Amundsen Sea, Antarctica, is melting due to warm ocean 36 waters. The glacial meltwater that is produced is less salty and carries essential food for 37 biological organisms, so where the glacial meltwater goes once it leaves the front of the 38 ice-shelf is important to understand: less salt in the ocean at the surface makes it eas39 ier to form sea ice, and increased productivity from biological organisms can help draw 40 carbon down into the ocean from the atmosphere. We use noble gases to identify where 41 this glacial meltwater is, as the signature that the meltwater leaves in the gases is unique 42 like a fingerprint. We use the noble gas meltwater signature to improve our identifica43 tion of glacial meltwater using temperature, salinity and dissolved oxygen (hydrographic 44 observations), which are easier and cheaper to collect so cover a larger area. Using the 45 improved signature from hydrographic observations we identify the presence of glacial 46 meltwater between 150-400 m depth everywhere across the continental shelf. We also show, 47 for the first time, glacial meltwater from the ice-shelf flowing off-shelf in the easternmost 48 channel. These results are important as they show where glacial meltwater is affecting 49 the ocean column most. 50

treat. Since the 1990s, multiple field campaigns have taken place in this region, oper- ). This location bias is mainly due to the reliability associated with the tracers used 84 to identify GMW, as it was unknown how reliable conservative tracers (and pseudo-conservative 85 tracers such as dissolved oxygen concentration) would be with increasing distance from 86 the ice shelves (Jenkins, 1999). 87 Recent work has shown that up to 500 km from PIIS, hydrographic tracers (con-88 servative temperature, absolute salinity and dissolved oxygen concentrations) identify . 96 The heavier noble gases, krypton (Kr) and xenon (Xe), are undersaturated in GMW, and 97 so are used as additional 'fingerprints' to identify the GMW (Loose & Jenkins, 2014). 98 The signature of GMW from noble gases has some variability associated with physical 99 effects (such as air content in the ice), but this is relatively small compared to the vari-  In this study, we present hydrographic, noble gas and oxygen isotope data collected 109 from the Amundsen Sea as part of the 2014 iSTAR research cruise (Section 2). We cal-110 culate freshwater distribution from oxygen isotope ratios (Section 3) and the distribu-111 tion of glacial meltwater using noble gases (Section 4). The hydrographic GMW calcu- 112 lations are compared with the noble gases and improved using the noble gas GMW con-113 tent as ground-truth, revealing a higher spatial resolution and more extensive dataset 114 of GMW content (Section 5). Finally, we combine the GMW content with current ve-115 locity data to identify meltwater pathways across the eastern Amundsen Sea (Section

118
The analysis included in this paper was conducted using data and water samples  (Table 1; Figure 2a). The GMW appears as a warmer, more saline and 132 less oxygenated water mass than the WW due to its admixture with mCDW, but as a 133 pure water mass GMW is cold, fresh and highly oxygenated (Table 1). All water mass 134 content is reported as g kg −1 , which is comparable to . Four sections are focused on  Figure S1 (Section B).

137
Water samples for oxygen isotope analysis were taken at 53 stations, and for no-138 ble gas (helium, neon, argon, krypton and xenon) analysis at 31 stations (Figure 1, Fig-139 ures S2-5), with the two techniques coinciding at 19 stations.  analytical reproducibility was <0.04 on duplicates. 161 We use current velocity data from a RDI 300kHz Workhorse Lowered Acoustic Doppler 162 Current Profiler unit fitted to the CTD rosette frame. We are using LADCP velocity pro-163 files that are co-located with the CTD stations and tracers collected.

177
where δ 18 O mCDW represents the oxygen isotope ratio endpoint for mCDW, F is the wa-178 ter mass fraction, and δ 18 O obs is the observed oxygen isotope ratio.

179
The mCDW that is present on the eastern Amundsen Sea shelf has a δ 18 O of 0.05  Table 1). 192 We use Monte Carlo simulations to estimate the uncertainties in the water mass 193 calculation. Each endpoint is perturbed around the reported endpoint (Table 1)   As the oxygen isotope ratios cannot be used to distinguish glacial meltwater from 251 local precipitation, we use other tracers measured during the fieldwork. By using a sim-252 ilar method to the one used for oxygen isotope ratios, we identify different water masses 253 in the Amundsen Sea using noble gas concentrations.
262 where χ n,k is the noble gas tracer n of water mass k and F k is the water mass frac- and persistent across the continental shelf.

306
Following the same method as with the oxygen isotope water mass fractions, we 307 assess the spatial distribution of NG GMW by calculating column inventories between 308 150 -700 m (Figure 4f). This depth range is selected in order to compare values more 309 easily with the hydrographic GMW content, which cannot be used in the upper 150 m 310 due to atmospheric interaction and SIM (Jenkins, 1999). This shows high NG GMW con- The GMW content is recalculated using the hydrographic tracers and mCDW, pWW 376 and GMW endpoints (Figure 2a,b, Figure 6). Previously, the hydrographic calculation the core of higher GMW content (Figure 8c,d).

466
At the edge of the continental shelf, there is no significant increase in GMW con-467 tent towards the surface (Figure 9a,c). Across the central channel, 535 km from PIIS, 468 the GMW content is lower than in the eastern channel and all values below 150 m depth 469 are less than 3.9 g kg −1 (Figure 9a). This section is also different to the previous sec-

507
The use of noble gases to quantify GMW provides reliable estimates that are used 508 as a ground-truth for our other water mass calculations. The noble gases revealed a per-509 sistent signature between 150 -400 m depth of GMW across all of the stations sampled, 510 which has not been reported in the eastern Amundsen Sea before. It is likely that close 511 to PIIS there is a significant GMW content that is excluded from this study between the 512 surface and 150 m, but due to atmospheric effects these depths have been excluded. This  We have demonstrated the value of oxygen isotope ratios and noble gas concentra-575 tions in determining freshwater distribution across Amundsen Sea. Noble gas concen-576 trations enable a reliable calculation of GMW content that is used as a ground-truth for 577 hydrographic water mass calculations to be tuned to, using the pWW endpoint.