Freshwater budget of the Canada Basin, Arctic Ocean, from salinity, d 18 O, and nutrients

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113,, doi:10.1029/2006jc003858, 2008 Freshwater budget of the Canada Basin, Arctic Ocean, from salinity, d 18 O, and nutrients M. Yamamoto-Kawai, 1 F. A. McLaughlin, 1 E. C. Carmack, 1 S. Nishino, 2 and K. Shimada 2 Received 1 August 2006; revised 13 September 2007; accepted 6 November 2007; published 12 January 2008. [1] The contribution of freshwater components (e.g., meteoric, sea ice, and Pacific water) in the Canada Basin is quantified using salinity, d 18 O, and nutrient data collected in 2003 and 2004. The penetration depth of sea ice meltwater is limited to the upper 30 m, and brine, rejected during sea ice formation, is observed from 30 to 250 m depth. The fraction of meteoric water is high in the upper 50 m and decreases with depth. Pacific water entering via Bering Strait is the main source of freshwater below 50 m depth. Bering Strait throughflow, which transports Pacific water with salinity 32.5 together with meteoric water supplied upstream of the Bering Strait, contributes up to 75% of freshwater input (>3200 km 3 a 1 ) to the Canada Basin. The mean residence time of Pacific water in the Canada Basin is estimated to be 11 years. Precipitation and river runoff from both North American and Eurasian continents add >800 km 3 a 1 and sea ice formation removes <900 km 3 a 1 (<0.6 m a 1 ) of fresh water. The export of ice and liquid fresh water from the Canada Basin contributes 40% of the freshwater flux from the Arctic Ocean to the North Atlantic Ocean. Citation: Yamamoto-Kawai, M., F. A. McLaughlin, E. C. Carmack, S. Nishino, and K. Shimada (2008), Freshwater budget of the Canada Basin, Arctic Ocean, from salinity, d 18 O, and nutrients, J. Geophys. Res., 113,, doi:10.1029/2006jc003858. 1. Introduction [2] The Canada Basin stores a large quantity of fresh water (relative to a reference salinity of 34.85 [Aagaard and Carmack, 1989]), comprising sea ice meltwater, meteoric water (precipitation and runoff), and low-salinity Pacific water that enters via Bering Strait. This freshwater storage subsequently plays a role in global climate by affecting airice-ocean heat exchange and convective processes in the North Atlantic [cf. Häkkinen, 1999; Proshutinsky et al., 2002; Karcher et al., 2005; Dukhovskoy et al., 2006]. It is therefore desirable to know where and how long each component of fresh water is stored in the ocean in order to understand how they may be affected by climate variability. [3] The oxygen isotope ratio (d 18 O) and salinity (S) of water have been used in three-component mixing models to estimate the fractional composition of sea ice meltwater, meteoric water, and saline or oceanic water [e.g., Bauch et al., 1995; Macdonald et al., 2002]. Although identification of the sea ice meltwater and meteoric water end-members is straightforward, the choice of the saline end-member has varied among studies. For example, Macdonald et al. [1995, 2002] used the polar mixed layer (S = 32.2) and middle halocline water (S = 33.1), respectively, although these 1 Department of Fisheries and Oceans, Institute of Ocean Sciences, Sidney, British Columbia, Canada. 2 Institute of Observational Research for Global Change, Japan Agency for Marine-Earth Science and Technology, Yokosuka, Kanagawa, Japan. Copyright 2008 by the American Geophysical Union. 0148-0227/08/2006JC003858 layers already contain sea ice meltwater and meteoric water. Melling and Moore [1995] and Yamamoto-Kawai et al. [2005] used Atlantic water (S 34.9) as the saline endmember because Pacific water can roughly be expressed as a linear mixture of Atlantic water and meteoric water on a d 18 O-S diagram. This later approach, however, does not distinguish between Pacific and meteoric waters. [4] To estimate the contribution of Pacific water, Bauch et al. [1995] and Ekwurzel et al. [2001] included the nutrient tracers Si and PO 4 *, respectively, because nutrient concentrations are much higher in Pacific than in Atlantic waters. However, Si and PO 4 * undergo significant change while crossing the Bering and Chukchi shelves, due to biological uptake in summer and regeneration of organic matter in winter, and thus nutrient concentrations are relatively high in winter and low in summer [Cooper et al., 1997, 1999]. Accordingly, the use of winter values of Si and PO 4 * may underestimate the contribution of Pacific and overestimate that of Atlantic water. To avoid these effects, Jones et al. [1998] used the nitrate-phosphate relationship, itself, as a tracer of Pacific water. Because nitrate and phosphate usually change at a ratio of 16:1 during biological production and regeneration of organic matter, the slope of the nitrate/phosphate relationship is 16 for both Pacific and Atlantic waters. The intercepts are significantly different, however, owing to the denitrification that occurs in the Pacific Ocean and in bottom sediments on the Bering and Chukchi shelves [Devol et al., 1997; Tanaka et al., 2004; Yamamoto-Kawai et al., 2006] and this then causes a nitrate deficit in Pacific water flowing into the Arctic Ocean. Assuming that the only cause of a shift in the nitratephosphate line away from the Atlantic-origin water line is 1of12

CO 2 -H 2 O equilibration unit with precision of 0.03%. Samples from the CCGS Louis S. St-Laurent 2004 cruise were analyzed at Oregon State University with a precision of 0.04%. Figure 1. Map of stations occupied during R/V Mirai cruises in 2000 and 2002 (open circles) and CCGS Louis St. Laurent cruises in 2003 and 2004 (solid circles). Bathymetric contours at 1000, 2000, and 3000 m are shown in gray. The shaded area indicates the region of the Canada Basin defined by bottom depths greater than 1000 m from 70 N to85 Nand 176 W to 100 W. mixing with Pacific water, Jones et al. [1998] estimated the fraction of Pacific water found in the surface layer of the Arctic Ocean. Similarly, Taylor et al. [2003] combined d 18 O and nitrate-phosphate data to obtain mixing ratios of Atlantic water, Pacific water, sea ice meltwater, and meteoric water in waters flowing from the Arctic Ocean through Fram Strait. [5] Here we use nitrate, nitrite, ammonium, and phosphate data, together with d 18 O and S data, to quantitatively estimate the distributions of the four main sources waters to the Canada Basin: Atlantic water, Pacific water, sea ice meltwater, and meteoric water. From these distributions, the volume, flux and residence time are also estimated. 2. Data [6] Samples for nutrient analysis were collected during cruises on the Japanese R/V Mirai, in 2000 and 2002 and Canadian CCGS Louis S. St-Laurent in 2003 and 2004. R/V Mirai cruises occupied the continental shelf/slope areas primarily and the CCGS Louis S. St-Laurent cruises extended into the Canada Basin interior (Figure 1). Nitrate (NO 3 ), nitrite (NO 2 ), ammonium (NH + 4 ), and phosphate (PO 3 4 ) were analyzed onboard the R/V Mirai using a Bran & Luebbe TRAACS 800 system with a precision of: NO 3 = 0.06, NO 2 = 0.01, NH + 4 = 0.03, PO 3 4 = 0.02 mmol kg 1. Nitrate and phosphate for the CCGS Louis S. St-Laurent were measured on a three channel Technicon Auto Analyzer and precisions are NO 3 = 0.08 and PO 3 4 = 0.01 mmol kg 1. Oxygen isotope (d 18 O) samples collected during the CCGS Louis S. St-Laurent 2003 cruise and from the Bering Sea during Mirai cruises were analyzed at the International Arctic Research Center (IARC), Fairbanks Alaska with a Finnigan MAT-252 mass spectrometer connected with a 3. Nitrate Versus Phosphate in the Canada Basin [7] Nitrate and phosphate data collected from the Chukchi shelf/slope region during the R/V Mirai cruises, when plotted together with the pure-atlantic and pure-pacific water lines defined by Jones et al. [1998], show good agreement with the Atlantic but not the Pacific water line (Figure 2a). Using the approach of Jones et al. [1998], it would appear that surface waters, almost depleted in nitrate, contain Atlantic water that has been mixed with low-salinity water (S < 32); however, this is not the case. This depletion is due to biological processes in Pacific water and is found only in summer [Codispoti et al., 2005]. Most of the R/V Mirai data from the surface to the depth of the nutrient maximum (at S 33) lie to the left of the Pacific line but some data lie to the right. The later have high ammonium concentrations, up to 7 mmol/kg (Figure 2a), and were collected in the Chukchi Sea [Nishino et al., 2005]. High ammonium content has been observed during summer in bottom waters on highly productive shelves such as the Bering and Chukchi [Saino et al., 1983; Codispoti et al., 2005]. Once this high-ammonium water leaves the shelf and enters the Canada Basin, nitrification converts ammonium to nitrate. Therefore the nitrate-phosphate relationship of Pacific water is changed, not only by mixing with Atlantic water, but also by nitrification of ammonium. To incorporate these shelf-water effects we use total dissolved inorganic nitrogen (DIN: NO 3 +NO 2 +NH 4 + ) instead of nitrate since DIN does not change during nitrification and export to the basin (Figure 2b). Note that nitrite was lower than 0.3 mmol/kg for all stations except those near the Mackenzie River mouth (NO 2 =upto0.6mmol kg 1 ). Comparison of Figures 2a and 2b illustrates that the difference between Chukchi Shelf waters, found to the right of the Pacific line, and other Canada Basin waters, found to the left (Figure 2a), is due to nitrification of ammonium and not from a difference in Pacific water content. Importantly, only waters between the nutrient maximum and waters on the Atlantic water line are a true mixture of Pacific and Atlantic waters. [8] Applying a linear regression to only those data that lie above the nutrient maximum (S 33) and excluding data from surface waters affected by summer productivity (PO 4 3 < 1.0), we determined a new equation for the Pacific water line, DIN ¼ 13:957 PO 3 4 11:306 ðr ¼ 0:98Þ: ð1þ The difference between the Pacific water line defined here and that given by Jones et al. [1998] may be due either to their use of nitrate instead of DIN, further denitrification or both. Their data were obtained in the Chukchi slope region near 170 W, during the 1994 Arctic Ocean Section. Along the same section, Wheeler et al. [1997] measured ammonium contents as high as 2.6 mmol kg 1 in bottom waters on the Chukchi shelf; however, over the slope, ammonium concentrations were not sufficiently high to explain the difference between these two Pacific water lines. Seasonal 2of12

Figure 2. Nitrogen versus phosphorus data from (a, b) R/V Mirai 2000 and 2002, (c) R/V Mirai 2004, and (d) CCGS Louis S. St-Laurent 2003 and 2004 cruises. The y axis is nitrate in Figures 2a and 2d, and is total dissolved inorganic nitrogen (DIN = nitrate + nitrite + ammonium) in Figures 2b and 2c. The thick gray line (A) and black dashed line (P) indicate pure-atlantic water line and pure-pacific water line, respectively, as reported by Jones et al. [1998]. The thick black line is the regression line determined in this study for the DIN-phosphate relation at S 33 and phosphate 1.0. The data points are colored by ammonium concentration (Figure 2a), S (Figures 2b and 2d) or station locations (Figure 2c: open and solid circles indicate data from stations with bottom depth <200 m and >200 m, respectively). variability is not an explanation since both constructs used summer data, and Devol et al. [1997] found no strong seasonal signal in the denitrification rate, at least in the coastal region near Pt. Barrow, despite the presumed variability in primary production. [9] Spatial variability in DIN-phosphate is a possible explanation for the difference between the two Pacific lines since the Jones et al. [1998] section was located west of the region studied from the R/V Mirai in 2000 and 2002. To investigate this, we examine DIN-phosphate data collected from the R/V Mirai in 2004 [Shimada, 2004], in the Chukchi shelf/slope region between 180 E and 145 W. Data from shallow stations (<200 m) show more scatter in the DIN-phosphate relationship than data from deeper slope and basin stations where data lie on the new Pacific water line defined using R/V Mirai 2000 and 2002 cruises (Figure 2c). This scatter is likely due to local denitrification. It is thus possible that Jones et al. [1998] data were collected from waters altered by locally enhanced denitrification. Likewise, interannual variability in denitrification on the Bering and Chukchi shelves or in the DIN-phosphate relationship of inflowing North Pacific water is also possible. Nevertheless, the fact that most of the water in both the western and eastern Chukchi slope in 2004 (Figure 2c), as well as water 3of12

Table 1. End-Member Values Used in This Study ATW PW SIM MW Salinity 34.87 ± 0.03 32.5 ± 0.2 4 ± 1 0 d 18 O, % 0.24 ± 0.05 0.8 ± 0.1 2 ±1.0 20 ± 2 in the Canada Basin interior (Figure 2d), lie along the new Pacific water line above the nutrient maximum imply that seasonal and interannual variability is small and support the reliability of our new line as the appropriate DIN-phosphate relationship for Pacific water. Note that ammonium and nitrite contents are negligible (DIN nitrate) in basin waters. Figure 2d also shows that mixtures containing Atlantic water are only found below the nutrient maximum (S > 33) in the Canada Basin. Effects of mixing Pacific water with fresh water from rivers, precipitation and sea ice, which would shift the DIN-phosphate relationship toward the Atlantic water line [Jones et al., 1998], are negligible here because the range of Pacific water used here to obtain equation (1) is constrained between 31 < S 33. 4. Calculations [10] In this section we combine S, d 18 O and nutrient data to quantify the contribution of each freshwater source. Assuming that water in the Canada Basin is a mixture of saline end-member (SE), meteoric water (MW), and sea ice meltwater (SIM), the fractions of each of these waters are calculated using the following mass balance equations for salt and d 18 O, f SE þ f SIM þ f MW ¼ 1; f SE S SE þ f SIM S SIM þ f MW S MW ¼ S obs ; f SE d SE þ f SIM d SIM þ f MW d MW ¼ d obs ; where f, S, and d refer to the fraction, salinity, and d 18 O, respectively. S obs and d obs are the observed values from each seawater sample. Sea ice formation, which injects brine into the underlying seawater, will be represented by a negative f SIM [cf. Östlund and Hut, 1984]. [11] According to the nutrient analysis presented in section 3, Pacific water (PW) is the SE for S 33 waters. In S > 33 waters, the SE is a mixture of PW and Atlantic water (ATW), and the mixing ratio (R PW ) is calculated using the Jones et al. [1998] approach but applying the new equation (1). It should be noted that this partition at S = 33 is only applicable in the Canada Basin. In other areas of the Arctic Ocean, the S of the partition must be determined and effects of biological processes and freshwater inputs carefully considered. For each observed nitrate value, two phosphate values are determined, one from the Atlantic water line (P ATW ) and one from the Pacific water line (P PW ) [Jones et al., 1998]. Then the ratio of Pacific water (R PW )is determined from the observed (P obs ) and calculated phosphate values (R PW =(P obs P ATW )/(P PW P ATW )) for S > 33 waters. Assuming the Atlantic and Pacific stoichiometric lines are parallel, the relative distance of any data point from ð2þ ð3þ ð4þ the two lines indicates the mixing ratio of Pacific and Atlantic waters. However, as the Atlantic and the Pacific lines are not strictly parallel, there is a potential error in R PW which we estimated to be <0.14 when P PW and P ATW are constrained between 1.5 2.3 mmol kg 1 and 0.6 1.1 mmol kg 1, respectively (see Figure 2d). [12] For S > 33 waters, S SE and d SE are calculated as follows to solve equations (2) (4), S SE ¼ S PW R PW þ S ATW ð1 R PW Þ ð5þ d SE ¼ d PW R PW þ d ATW ð1 R PW Þ: ð6þ The obtained f SE is a sum of f PW and f ATW and their mixing ratio is R PW :(1 R PW ). Therefore f PW ¼ f SE R PW ð7þ f ATW ¼ f SE ð1 R PW Þ: ð8þ For S 33 waters, S SE =P SE, d SE = d SE, and R PW =1. [13] The S and d 18 O values for the four end-members used in the following calculations are summarized in Table 1 and shown in Figure 3 together with observed data from the Canada Basin and the Bering shelf. The value of d MW is taken from Cooper et al. [2005]. S SIM is from Ekwurzel et al. [2001] and d SIM is set to be 2% to represent the d 18 O value of sea ice in the Canada Basin [Eicken et al., 2002; Pfirman et al., 2004]. Values for ATW are means of ATW data collected in the Eurasian Basin, upstream of the Canada Basin [Yamamoto-Kawai et al., 2005]. [14] Woodgate et al. [2005] found that the mean nearbottom S in Bering Strait (50 m depth) is 32.5 ± 0.3 and accordingly we use this value to define PW. The d 18 Oof PW is 0.8 ± 0.1%, based on the d 18 O of the Bering Shelf water at S = 32.5 ± 0.3 (Figure 3). Although our data are from the eastern Bering Shelf, Cooper et al. [1997, 2006] observed similar d 18 O values for S 32.5 water in Anadyr Strait and in the central Bering Strait. Bering Strait throughflow also transports additional freshwaters in the Alaskan Coastal Current, in stratification in the central Bering Strait and in sea ice transport [Woodgate and Aagaard, 2005]. Therefore S and d 18 O of the throughflow vary both seasonally and spatially. The S- d 18 O diagram (Figure 3) shows that the summer Bering shelf waters roughly follow a mixing line connecting PW and MW. Thus the throughflow water with S < 32.5 can be expressed as a mixture of PW and MW (PW + MW) upon entering the Canada Basin. In winter, injection of brine from the formation and growth of sea ice increases S of the throughflow [cf. Woodgate et al., 2005] and thus winter Bering Strait water with S > 32.5 are expressed as PW SIM. Accordingly, MW and SIM transported through Bering Strait to the Canada Basin cannot be distinguished from direct inputs in the Arctic Ocean. Our estimates of MW and SIM include, in fact, those from the Bering Sea. In the following, we first quantify the volumes of PW, MW and SIM in the Canada Basin and then use the transport estimates of Woodgate and Aagaard [2005] to quantify the volumes of MW and SIM that enter the Canada Basin as direct inputs. 4of12

Figure 3. (a) Relationship between S and d 18 O of water in the Canada Basin (black dots) and on the Bering Shelf in the upper 5 m (light blue crosses) and from 5 to 50 m (blue dots). Open squares indicate ATW and PW end-members. Arrows indicate mixing with MW, SIM, and brine. (b) Station locations on the Bering Shelf are shown on the map: blue dots indicate stations where samples were collected from the surface to the bottom, and light blue crosses indicate stations where only the surface was sampled. (c) Full range S and d 18 O relationship indicating end-member values for MW and SIM. Grey lines in Figures 3a and 3c are mixing line between PW and MW. [15] To quantitatively compare freshwater sources, estimated f PW and f SIM are subsequently converted into fractions of freshwater equivalent (f FWeqPW and f FWeqSIM ), relative to the S of ATW, 34.87, f FWeqPW ¼ f PW ð34:87 S PW Þ=34:87 ð9þ f FWeqSIM ¼ f SIM ð34:87 S SIM Þ=34:87: ð10þ The uncertainties of f MW, f PW, f ATW, f SIM, f FWeqPW, and f FWeqSIM estimates are ±0.03, 0.14, 0.14, 0.03, 0.03, and 0.03, respectively, and are due to uncertainties in the range of end-member values (Table 1), d 18 O analysis, and the uncertainties in the calculation of R PW. 5. Composition of Water in the Canada Basin [16] In S-d 18 O diagram (Figure 3), surface waters in the Canada Basin with S < 28 lie to the left of a mixing line connecting PW and MW, indicating mixing with SIM. Waters in the S range of 30 34.5, most notably at salinities near 33, lie to the right. This deviation cannot be explained by mixing of source waters. Instead, injection of brine from growing sea ice causes this deviation, in that the S increases owing to brine injection but the d 18 O value remains virtually unchanged. Plots of S and fractions of the four source waters (Figures 4a and 4b), as quantitatively estimated from S, d 18 O and nutrients, show that positive values of f SIM are found only above 30 m and brine is found between 30 m and 250 m. The depth of 30 m corresponds to the bottom of the summer mixed layer in the Canada Basin [McLaughlin et al., 2004]. The f MW is 0.1 at the surface, decreases rapidly with depth to 0.05 at 50 m, and then gradually decreases to 0 at 300 m. Figure 4b shows that SE is the major source of the Canada Basin water and that, together, MW and SIM are less than 0.2. From the surface to 100 to 150 m, PW is the SE source. Deeper in the water column, the PW decreases and ATW increases and below 150 to 250 m ATW is the SE source. The transition depth, where mixing between PW with ATW begins, is nonuniform 5of12

Figure 4. Vertical distributions of: (a) S; (b) the f MW (green), f SIM (gray), f PW (blue) and f ATW (red); (c) f FWeqSW ; and (d) f MW (green), f FWeqSIM (gray), and f FWeqPW (blue). The thick black line in Figure 4a indicates the S of ATW. Error bars for f MW,f SIM,f PW, and f ATW are also indicated in Figure 4b. Note two scales are shown in Figure 4b. across the basin, being shallower in the north and deeper in the south. [17] Figure 4c shows the distribution of the freshwater equivalent of seawater (FWeqSW) when referenced to S of ATW, 34.87, f FWeqSW ¼ ð34:87 S obs Þ=34:87: ð11þ The f FWeqSW is highest (0.2) at the surface, steeply decreases to 0.1 at 50 m, and then gradually decreases to 0 at 300 m. The composition of this fraction when plotted by source (Figure 4d) shows that MW is the largest contributor of fresh water between the surface and 50 m; at the surface, where f FWeqSW 0.20, MW, FWeqPW and FWeqSIM contribute 0.12, 0.06, and 0.02, respectively. Below 50 m, PW is the main source of fresh water in the Canada Basin. The f FWeqPW is relatively constant at about 0.06 from the surface to a depth of 100 m in the north and to a depth of 200 m in the south (Figure 4d; see also Figure 5 for horizontal distribution). 6. Horizontal Distributions of Freshwater Sources in the Canada Basin [18] The fractions of freshwater sources, calculated at individual stations, are integrated from the surface to 300 m, and these inventories are mapped to show their horizontal distributions (Figure 5). Changing one end-member value at a time, by the maximum error value reported in Table 1, results in differences of 2.1 m in each of FWeqSW, MW, FWeqPW and FWeqSIM, 21 m in PW, and 2.4 m in SIM. The mean, range, and standard deviation of inventories from all observations are listed in Table 2. [19] The inventory of FWeqSW is highest at 20 m in the southern part of the Canada Basin and decreases to 14 m toward the northwest (Figure 5a). The MW inventory is highest (>15 m) near the Mackenzie River, is 14 m in the south and <10 m farther north (Figure 5b). The inventory of FWeqPW is 14 m in the south and 10 m in the north with a mean of 13 m (Figure 5c). The inventory of PW is >200 m in the south and <150 m in the north (Figure 5e) and the mean is 190 m. This is much higher than reported by Ekwurzel et al. [2001], who estimated <60 m using data from the 1994 Arctic Ocean Section but using PO 4 * instead of nitrate and phosphate to estimate the amount of Pacific water. This difference can be explained by difference in sampling location, well out of the core of the Beaufort Gyre, the corresponding change in the depth of Pacific water [cf. McLaughlin et al., 2004], and also by the use of PO 4 * which underestimates PW (see section 1). 6of12

Figure 5. Maps showing the distribution of (a) FWeqSW, (b) MW, (c) FWeqPW, (d) FWeqSIM, (e) PW, and (f) SIM as water column equivalents in meters, integrated from the surface to the 300 m depth. [20] The inventory of SIM is negative over our entire study area and this indicates that fresh water is removed from the Canada Basin by sea ice formation and export (Figure 5f). The mean equivalent height of fresh water removed as sea ice is 8 m, with higher values ( 10 m) near the northern Chukchi and Beaufort shelves and lower values ( 6 m) near the Alaskan coast (Figure 5d). However, since brine and SIM can be laterally transported by shelf- basin interaction, this distribution does not necessarily reflect in situ processes but instead reflects the history of the water mass formation and spreading and shelf-basin exchange. Thus it is not possible to distinguish between processes that take place within the basin or on the surrounding shelves. [21] The mean inventories of MW, FWeqPW, and FWeqSIM for the study area are 12.6 m, 13.0 m, and 7.9 m, 7 of 12

Table 2. Mean, Minimum, Maximum, and Standard Error of Inventory (Equivalent Height) for Each Freshwater Source Calculated at 45 Stations (61 Stations for FWeqSW Calculations Using S Only) in the Canada Basin a PW SIM MW FWeqPW FWeqSIM FWeqSW Summer Estimate (Observed) Minimum, m 111 14.6 6.8 7.6 12.9 13.0 Maximum, m 237 3.5 17.9 16.1 3.1 23.8 Standard Error 4.4 0.4 0.5 0.3 0.4 0.3 Observed area mean, m 192 8.9 12.6 13.0 7.9 17.8 Canada Basin mean, m 172 7.3 10.5 11.7 6.5 16.3 Winter Estimates Minimum, m 112 15.2 6.1 7.6 13.4 12.6 Maximum, m 238 4.7 17.9 16.2 4.2 22.3 Standard Error 4.5 0.4 0.4 0.3 0.4 0.3 Observed area mean, m 193 9.7 12.3 13.1 8.6 16.9 Canada Basin mean, m 173 7.8 10.4 11.7 6.9 15.7 a Mean inventories for the 21 stations (33 stations for FWeqSW calculations using S only) in the central Canada Basin (75 N 80 N) are also listed. respectively (Table 2). Because all inventories decrease from south to north, we also calculated mean inventories for the central region of the Canada Basin, between 75 N and 80 N, and assume that these inventories represent mean values for the entire Canada Basin. The mean inventories of MW, FWeqPW, and FWeqSIM in the central Canada Basin are 10.5 m, 11.7 m, and 6.5 m, respectively (Table 2). These estimates are based on summer observations and, because inventories of MW and SIM are lower in winter, we also estimated inventories in winter by assuming that surface freshening in summer reaches only to 30 m depth because SIM is positive in the top 30 m of the water column (Figure 4b). Freshwater fractions at 30 m in summer are then used to integrate from 0 to 30 m to represent the winter mixed layer. Results show that seasonal variability in freshwater inventories (1 m) is much smaller than spatial variability (10 m; see Table 2 and Figure 5). [22] In summary, fresh water in the Canada Basin consists mainly of MW and FWeqPW, and they contribute almost equally (10.5 m and 11.7 m, respectively) to the FWeqSW found in the top 300 m of the water column. Sea ice formation removes about 6.5 m, one third of the fresh water contributed by MW and PW combined. 7. Storage, Residence Time, and Flux of Fresh Water [23] Mean inventories for the central Canada Basin (75 N 80 N), with an uncertainty of ±2 m, are used to calculate the storage, flux, and residence time of fresh water in the upper 300 m of the Canada Basin. Multiplying by the surface area of the Canada Basin (1.6 km 2 ), the volume of fresh water added by MW and FWeqPW are estimated to be 16,800 and 19,200 km 3 respectively, and the formation of sea ice has removed 10,400 km 3. Freshwater stored in the Canada Basin is 25,600 km 3 (Table 3) and this corresponds to about 1/3 of the total fresh water stored in the entire Arctic Ocean as estimated from observations and models [Zhan and Zhang, 2001, and references therein]. The volumes of MW and FWeqPW that are stored in the ocean also depend on the fraction of fresh water removed as sea ice. In summer MW and PW are found in a 1:9 ratio in the upper 50 m (Figure 4b). Assuming that sea ice is formed equally from both sources, 1000 km 3 will be removed from MW volume (10% of 10,400 km 3 ) and 9,400 km 3 from FWeqPW (90% of 10,400 km 3 ). Thus the net storage of MW and FWeqPW are estimated to be 15,800 km 3 (16,800 1000) and 9800 km 3 (19,200 9400), respectively. [24] Residence times are estimated as follows. The mean Bering Strait transport of water with mean S = 32.5 is 0.8 ± 0.1 Sv [Woodgate and Aagaard, 2005] and this inputs 1700 ± 300 km 3 a 1 of fresh water (relative to S = 34.87) into the Arctic Ocean. Bering Strait throughflow also includes other fresh water transport via the Alaskan Coastal Current, seasonal stratification and sea ice [Woodgate and Aagaard, 2005]. However, because we define PW as water with S = 32.5, we assume that the transport of FWeqPW is 1700 ± 300 km 3 a 1, and therefore any additional fresh water entering the Canada Basin thorough Bering Strait will be identified in either the MW or SIM component. Assuming that all PW enters the Canada Basin interior, we estimate that the mean residence time of PW and FWeqPW in the Canada Basin is 11 ± 4 years (dividing the volume of fresh water added by FWeqPW (19,200 ± 3200 km 3 ) by the annual input (1700 ± 300 km 3 a 1 )). [25] For MW and FWeqSIM, which include waters transported through Bering Strait as well as input from other sources, both residence time and fluxes to the Canada Basin are unknown. MW, FWeqSIM and FWeqPW comprise the upper part of the water column and if they move together within the Canada Basin, their residence times should be identical. If this were the case, we could apply 11 years of residence time to the estimated volumes of MW and FWeqSIM to obtain fluxes of 1500 km 3 a 1 (16,800/11), and 900 km 3 a 1 ( 10,400/11), respectively. However, since the vertical distributions of MW and FWeqSIM are not similar to PW but are instead high in the surface and decrease with depth (see Figure 4d), their residence times will be shorter than 11 years. Thus the flux of MW is larger than 1500 km 3 a 1 and the flux of SIM is smaller (less negative) than 900 km 3 a 1. 8. Discussion [26] The Canada Basin receives >3200 km 3 of fresh water from PW (1700 km 3 ) and MW (>1500 km 3 ) every year, and 8of12

Table 3. Summary of Estimates Determined in This Study a MW FWeqPW FWeqSIM FWeqSW FW inventory (0 300 m), m 10.5 ± 2 12 ± 2 6.5 ± 2 16 ± 2 Volume of FW added, km 3 16,800 ± 3200 19,200 ± 3200 10,400 ± 3200 25,600 ± 3200 Volume of FW stored in water, km 3 15,800 ± 3500 9,800 ± 6000 0 25,600 ± 3200 FW flux to the basin (0 300 m), km 3 a 1 >1500 ± 800 1700 ± 300 < 900 ± 600 b 2300 ± 1100 FW flux to the basin (0 300 m), m a 1 >0.9 ± 0.5 1.1 ± 0.2 < 0.6 ± 0.4 b 1.4 ± 0.7 a MW includes river water and precipitation input into the Arctic Ocean and those transported by Bering Strait throughflow. b Flux of FWeqSIM is smaller (less negative) than indicated value. <900 km 3 of fresh water is removed by sea ice formation (Table 3). On the basis of these estimates, we first discuss the contribution of fresh water imported through Bering Strait. This will be followed by a discussion of the composition of fresh water exported from the Canada Basin. [27] In sections 4 and 7 it was noted that Bering Strait transports not only PW but also MW and SIM. The export of fresh water as sea ice into the Arctic Ocean through Bering Strait is 115 km 3 a 1 [Aagaard et al., 2006], and when this sea ice melts within the Arctic Ocean it increases the fraction of SIM. Although SIM derived from melting in the Bering Sea may also be transported to the Canada Basin, the SIM component of Bering Sea surface water is not as large as the MW component even in spring shortly after melting [Cooper et al., 1997, 2006]. Thus the net effect of Bering Strait throughflow on the FWeqSIM flux in the Canada Basin is small compared to the large uncertainty of the flux estimate (<900 ± 600 km 3 a 1 ). Woodgate and Aagaard [2005] estimate that summer surface water and the Alaskan Costal Current transport 220 420 km 3 a 1 and 350 km 3 a 1 of fresh water, respectively, into the Arctic and these are incorporated in our estimates of MW in the Canada Basin. Assuming that MW transported through the Bering Strait is 700 km 3 a 1, the transport of total fresh water through Bering Strait will be 2400 km 3 a 1 (1700 + 700) and the flux of MW directly into the Canada Basin will be >800 km 3 a 1 (>1500 700). Thus the Bering Strait throughflow contributes up to 75% of freshwater enters the Canada Basin from PW and MW (2400/3200). [28] To estimate the export of fresh water from the Canada Basin we examine the rate of supply of each component. The net formation rate of sea ice estimated here is 0.6 m a 1 (900 km 3 a 1 /1.6 10 6 km 2 ). This is lower than the net sea ice growth of 2 3 ma 1 observed in the Beaufort Sea [Melling and Moore, 1995; Melling and Reidel, 1996; Macdonald et al., 1995] but is similar to the 0.5 m a 1 value estimated from observations [Östlund and Hut, 1984] and from models [Steele and Flato, 2000] of the Arctic Ocean interior. The estimated sea ice flux of <900 km 3 a 1 from the Canada Basin accounts for <40% of the sea ice flux through Fram Strait and the Canadian Archipelago (2460 km 3 a 1 [Serreze et al., 2006]). Therefore >60% or >1560 km 3 a 1 of the sea ice is exported from other regions of the Arctic Ocean, probably the Eurasian shelf seas. A freshwater export of <900 km 3 a 1 as sea ice requires that the remaining >2300 km 3 a 1 of freshwater input to the Canada Basin (>3200 km 3 a 1 )is exported in the liquid phase. According to recent estimates of the freshwater budget of the Arctic Ocean [Serreze et al., 2006], the amount of liquid fresh water exported through Fram Strait and the Canadian Archipelago is 2400 and 3200 km 3 a 1, respectively, and the flux from the Canada Basin accounts for more than 40% of this export. Our calculations indicate that the export of sea ice and liquid fresh water from the Canada Basin contributes 40% of the freshwater flux exiting the Arctic Ocean. [29] The MW flux to the Canada Basin estimated in this study is >800 km 3 a 1, excluding MW imported through Bering Strait. Lammers et al. [2001] estimated the discharge from North American rivers directly into the Canada Basin is 418 km 3 a 1 (excluding runoff entering via the Bering Strait). This suggests that a near-equal amount of MW from other sources is required to account for the MW flux to the Canada Basin. This is consistent with findings by Yamamoto-Kawai et al. [2005] that alkalinity values found in the southern Canada Basin require a significant input of MW other than North American runoff. If we assume that both FWeqPW and MW entering via Bering Strait have an alkalinity value of 930 mmol kg 1, then North American runoff and precipitation contribute 20% and 10% of >800 km 3 a 1 input of MW, respectively [see Yamamoto- Kawai et al., 2005]. The remaining 70% must derive from Eurasian runoff. Fluxes of North American runoff, precipitation and Eurasian runoff to the Canada Basin are thus estimated to be >160, >80 and >560 km 3 a 1, respectively, for a total of >800 km 3 a 1. We estimate that the flux of North American runoff that enters the Canada Basin to be more than 40% (160/418 km 3 a 1 ). The total river discharge into the Arctic Ocean is 3200 km 3 a 1 [Serreze et al., 2006]. Our calculations indicate that >25% of this total discharge from Arctic rivers enters the Canada Basin. It must be noted, however, that the uncertainty of the flux calculations in this study are high and that estimates from Yamamoto- Kawai et al. [2005] are based on alkalinity data from a limited region in the southern Canada Basin. [30] We estimate the residence time of PW (or FWeqPW) in the upper 300 m in the Canada Basin to be 11 ± 4 years, assuming that all PW enters the Canada Basin interior. This underestimates the residence time if some PW exits the Arctic Ocean without entering the basin interior, for example, by following a direct pathway along the upper continental slope and through the Arctic Archipelago. The residence time for MW, SIM and FWeqSW will be shorter than 11 years because their fractions are high in upper layer and decrease with depth. A residence time of <11 years of water in the Canada Basin is consistent with tritium-helium ages of <4 years at the surface and 18 years at 300 m reported by Smethie et al. [2000] and with tritium ages of 10 16 years for the freshwater component of seawater reported by Östlund [1982]. 9of12

Figure 6. Schematic of the freshwater budget of the Canada Basin. Fluxes of FWeqPW and freshwater storage in sea ice obtained from the literature are indicated in parentheses as a and b; other values are determined in this study. [31] Some of the fresh water from MW and PW sources will be exported as sea ice, and the residence time of fresh water in sea ice will be different than in the ocean. Assuming that a mean thickness of sea ice for the Canada Basin is 4 m [Thomas et al., 1996] and that the S SIM is 4, then the volume of fresh water stored in sea ice is found to be 5700 km 3 and the residence time of sea ice is >6 years (5700/<900). In the Rigor and Wallace [2004] model simulation the age of sea ice in the 1980s was older than 10 years and, after the 1990s, decreased to less than 5 years in most of the Canada Basin. Our residence time estimate is bracketed by these model results. The storage of fresh water in the Canada Basin, as both sea ice and liquid fresh water, is 31,300 km 3 (25,600 + 5700), and thus the residence time of all freshwater components in the basin is <10 years (31,300/>3200). In summary, fresh water and sea ice reside in the Canada Basin for <11 years and >6 years, respectively, and the mean residence time of fresh water in the Canada Basin, including both water and sea ice, is <10 years. 9. Summary [32] We developed a method to quantify the mixing ratio of ATW, PW, MW and SIM in the Canada Basin. Our approach combines the method of Jones et al. [1998], who found that Pacific and Atlantic waters have different nitratephosphate relationships and this difference can be used to estimate the mixing ratio of ATW and PW, and the mass balance equations for S and d 18 O to estimate MW and SIM contributions. Using nutrient data from the Chukchi shelf/ slope region and the Canada Basin, we revised the equation defining Pacific Water. We find that mixing of ATW and PW occurs only in waters with S > 33 in the Canada Basin. Therefore fractions of source water components can be estimated in the Canada Basin using S and d 18 O for waters with S 33, and using S, d 18 O and nutrients for S > 33 waters. [33] This approach is used to describe distributions of source waters over much of the Canada Basin in 2003 and 2004. Amounts of source waters integrated for 0 300 m of the basin together with PW flux through the Bering Sea [Woodgate and Aagaard, 2005] are then used to estimate the freshwater budget of the basin (Figure 6). The transport of fresh water through Bering Strait (2400 km 3 a 1 ) accounts for up to 75% of total freshwater input to the Canada Basin. Precipitation and river inflow to the Arctic Ocean add more than 800 km 3 a 1 of fresh water as MW to the Canada Basin. Contributions from Eurasian rivers are required to account for the total MW input to the Canada Basin. Of this combined freshwater influx that enters the Canada Basin, <900 km 3 a 1 (<0.6 m a 1 ) is exported as sea ice, and the remainder is exported as liquid water. The mean residence time of PW in the Canada Basin is estimated to be 11 ± 4 years, and somewhat less for MW and SIM. The export of sea ice and liquid fresh water from the Canada Basin contributes 40% of the freshwater flux exiting the Arctic Ocean. [34] These estimates use data collected in 2003 and 2004 and assume steady state. The Canada Basin, however, is currently in transition [McLaughlin et al., 2004]. The position of the Pacific water front retreated from the Lomonosov Ridge to the Alpha-Mendeleyev Ridge in 1990s [McLaughlin et al., 1996], presumably altering the volume of PW in the Arctic Ocean. Also, warmer Atlantic water arrived in the western Canada Basin in the late 1990s [McLaughlin et al., 2002], and spread a eastward in the early 2000s [Shimada et al., 2004]. These changes and associated changes in freshwater content have been attributed to changes in atmospheric forcing in that the cyclonicity increased during the early 1990s and decreased in the late 1990s [Proshutinsky et al. [2002]. Although atmospheric forcing has partly returned to pre-1990s conditions, the sea ice cover in the Canada Basin continued to decrease owing to the transport of warm Pacific water offshore and a feedback mechanism involving delayed sea ice formation, 10 of 12

reduced ice stresses and more efficient coupling of windforcing on the ocean [Shimada et al., 2006]. Accordingly, the freshwater budget illustrated in Figure 6 presents a snapshot of conditions in the Canada Basin during the early 2000s. [35] Acknowledgments. We would like to thank A. Proshutinsky as a principal investigator of the project, funded by NSF, and for his valuable comments. We are grateful to B. van Hardenberg and S. Zimmerman (chief scientists aboard the CCGS Louis S. St.-Laurent), other scientists and technicians who collected data during cruises, together with the officers and crews of the CCGS Louis S. St. Laurent and R/V Mirai. We also thank Kelly Falkner, Christopher Guy, and two reviewers who provided constructive comments that helped to improve our manuscript. References Aagaard, K., and E. C. Carmack (1989), The role of sea ice and other fresh water in the Arctic circulation, J. Geophys. Res., 94, 14,485 14,498. Aagaard, K., T. J. Weingartner, S. L. Danielson, R. A. Woodgate, G. C. Johnson, and T. E. Whitledge (2006), Some controls on flow and salinity in Bering Strait, Geophys. Res. Lett., 33, L19602, doi:10.1029/ 2006GL026612. Bauch, D., P. Schlosser, and R. Fairbanks (1995), Freshwater balance and sources of deep and bottom water in the Arctic Ocean inferred from the distribution of H 2 18 O, Prog. Oceanogr., 35, 53 80. Codispoti, L. A., C. Flagg, V. Kelly, and J. H. Swift (2005), Hydrographic conditions during the 2002 SBI process experiments, Deep Sea Res., Part II, 52, 3199 3226. Cooper, L. W., T. E. Whitledge, J. M. Grebmeier, and T. Weingartner (1997), The nutrient, salinity, and stable isotope composition of Bering and Chukchi seas waters in and near the Bering Strait, J. Geophys. Res., 102, 12,563 12,573. Cooper, L. W., G. F. Cota, L. R. Pomeroy, J. M. Grebmeier, and T. E. Whitledge (1999), Modification of NO, PO, and NO/PO during flow across the Bering and Chukchi shelves: Implications for use as Arctic water mass tracers, J. Geophys. Res., 104, 7827 7836. Cooper, L. W., R. Benner, J. W. McClelland, B. J. Peterson, R. M. Holmes, P. A. Raymond, D. A. Hansell, J. M. Grebmeier, and L. A. Codispoti (2005), Linkages among runoff, dissolved organic carbon, and the stable oxygen isotope composition of seawater and other water mass indicators in the Arctic Ocean, J. Geophys. Res., 110, G02013, doi:10.1029/ 2005JG000031. Cooper, L. W., L. A. Codispoti, V. Kelly, G. G. Sheffield, and J. M. Grebmeier (2006), The potential for using Little Diomede Island as a platform for observing environmental conditions in Bering Strait, Arctic, 52, 129 141. Devol, A. H., L. A. Codispoti, and J. P. Christensen (1997), Summer and winter denitrification rates in western Arctic shelf sediments, Cont. Shelf Res., 17, 1029 1050. Dukhovskoy, D., M. Johnson, and A. Proshutinsky (2006), Arctic decadal variability from an idealized atmosphere-ice-ocean model: 2. Simulation of decadal oscillations, J. Geophys. Res., 111, C06029, doi:10.1029/ 2004JC002820. Eicken, H., H. R. Krouse, D. Kadko, and D. K. Perovich (2002), Tracer studies of pathways and rates of meltwater transport through Arctic summer sea ice, J. Geophys. Res., 107(C10), 8046, doi:10.1029/ 2000JC000583. Ekwurzel, B., P. Schlosser, R. Mortlock, R. Fairbanks, and J. Swift (2001), River runoff, sea ice meltwater, and Pacific water distribution and mean residence times in the Arctic Ocean, J. Geophys. Res., 106, 9075 9092. Häkkinen, S. (1999), A simulation of thermohaline effects of a Great Salinity Anomaly, J. Clim., 12, 1781 1795. Jones, E. P., L. G. Anderson, and J. H. Swift (1998), Distribution of Atlantic and Pacific waters in the upper Arctic Ocean: Implications for circulation, Geophys. Res., Lett., 25, 765 768. Karcher, M., R. Gerdes, F. Kauker, C. Köberle, and I. Yashayaev (2005), Arctic Ocean change heralds North Atlantic freshening, Geophys. Res. Lett., 32, L21606, doi:10.1029/2005gl023861. Lammers, R. B., A. I. Shiklomanov, C. J. Vorosmarty, B. M. Fekete, and B. J. Peterson (2001), Assessment of contemporary Arctic river runoff based on observational discharge records, J. Geophys. Res., 106, 3321 3334. Macdonald, R. W., D. W. Paton, E. C. Carmack, and A. Omstedt (1995), The freshwater budget and under-ice spreading of Mackenzie River water in the Canadian Beaufort Sea based on salinity and 18 O/ 16 O measurements in water and ice, J. Geophys. Res., 100, 895 919. Macdonald, R. W., F. A. McLaughlin, and E. C. Carmack (2002), Fresh water and its sources during the SHEBA drift in the Canada Basin of the Arctic Ocean, Deep Sea Res., Part I, 49, 1769 1785. McLaughlin, F. A., E. C. Carmack, R. W. Macdonald, and J. K. B. Bishop (1996), Physical and geochemical properties across the Atlantic/Pacific water mass front in the southern Canada Basin, J. Geophys. Res., 101, 1183 1197. McLaughlin, F., E. Carmack, R. Macdonald, A. J. Weaver, and J. Smith (2002), The Canada Basin, 1989 1995: Upstream events and far-field effects of the Barents Sea, J. Geophys. Res., 107(C7), 3082, doi:10.1029/ 2001JC000904. McLaughlin, F. A., E. C. Carmack, R. W. Macdonald, H. Melling, J. H. Swift, P. A. Wheeler, B. F. Sherr, and E. B. Sherr (2004), The joint roles of Pacific and Atlantic-origin waters in the Canada Basin, 1997 1998, Deep Sea Res., Part I, 51, 107 128. Melling, H., and R. M. Moore (1995), Modification of halocline source waters during freezing on the Beaufort Sea shelf: Evidence from oxygen isotopes and dissolved nutrients, Cont. Shelf Res., 15, 89 113. Melling, H., and R. A. Reidel (1996), Development of seasonal pack ice in the Beaufort Sea during the winter of 1991 1992: A view from below, J. Geophys. Res., 101, 11,975 11,991. Nishino, S., K. Shimada, and M. Itoh (2005), Use of ammonium and other nitrogen tracers to investigate the spreading of shelf waters in the western Arctic halocline, J. Geophys. Res., 110, C10005, doi:10.1029/ 2003JC002118. Östlund, H. G. (1982), The residence time of the freshwater component in the Arctic Ocean, J. Geophys. Res., 87, 2035 2043. Östlund, H. G., and G. Hut (1984), Arctic Ocean water mass balance from isotope data, J. Geophys. Res., 89, 6373 6381. Pfirman, S., W. Haxby, H. Eicken, and M. Jeffries (2004), Drifting Arctic sea ice archives changes in ocean surface conditions, Geophys. Res. Lett., 31, L19401, doi:10.1029/2004gl020666. Proshutinsky, A., R. H. Bourke, and F. A. McLaughlin (2002), The role of the Beaufort Gyre in Arctic climate variability: seasonal to decadal climate scales, Geophys. Res. Lett., 29(23), 2100, doi:10.1029/ 2002GL015847. Rigor, I. G., and J. M. Wallace (2004), Variations in the age of Arctic seaice and summer sea-ice extent, Geophys. Res. Lett., 31, L09401, doi:10.1029/2004gl019492. Saino, T., H. Otobe, E. Wada, and A. Hattori (1983), Subsurface ammonium maximum in the northern North Pacific and the Bering Sea in summer, Deep Sea Res., 11, 1157 1983. Serreze, M. C., A. P. Barrett, A. G. Slater, R. A. Woodgate, K. Aagaard, R. B. Lammers, M. Steele, R. Moritz, M. Meredith, and C. M. Lee (2006), The large-scale freshwater cycle of the Arctic, J. Geophys. Res., 111, C11010, doi:10.1029/2005jc003424. Shimada, K. (2004), R/V Mirai Cruise Report MR04-05, edited by K. Shimada, S. Nishino, and M. Itoh, report, Jpn. Agency for Mar.-Earth Sci. and Technol. (JAMSTEC), Yokosuka, Japan. Shimada, K., F. McLaughlin, E. Carmack, A. Proshutinsky, S. Nishino, and M. Itoh (2004), Penetration of the 1990s warm temperature anomaly of Atlantic Water in the Canada Basin, Geophys. Res. Lett., 31, L20301, doi:10.1029/2004gl020860. Shimada, K., T. Kamoshida, M. Itoh, S. Nishino, E. Carmack, F. McLaughlin, S. Zimmermann, and A. Proshutinsky (2006), Pacific Ocean inflow: Influence on catastrophic reduction of sea ice cover in the Arctic Ocean, Geophys. Res. Lett., 33, L08605, doi:10.1029/2005gl025624. Smethie, W. M., P. Schlosser, G. Bönisch, and T. S. Hopkins (2000), Renewal and circulation of intermediate waters in the Canadian Basin observed on the SCICEX 96 cruise, J. Geophys. Res., 105, 1105 1121. Steele, M., and G. M. Flato (2000), Sea ice growth, melt, and modeling: A survey, in Fresh Water Budget of the Arctic Ocean, edited by E. L. Lewis, pp. 549 587, Kluwer Acad., Norwell, Mass. Tanaka, T., L. Guo, C. Deal, N. Tanaka, T. Whitledge, and A. Murata (2004), N deficiency in a well-oxygenated cold bottom water over the Bering Sea shelf: Influence of sedimentary denitrification, Cont. Shelf Res., 24, 1271 1283. Taylor, J. R., K. K. Falkner, U. Schauer, and M. Meredith (2003), Quantitative considerations of dissolved barium as a tracer in the Arctic Ocean, J. Geophys. Res., 108(C12), 3374, doi:10.1029/2002jc001635. Thomas, T., S. Martin, D. Rothrock, and M. Steele (1996), Assimilating satellite concentration data into an Arctic sea ice mass balance model, 1979 1985, J. Geophys. Res., 101, 20,849 20,868. Wheeler, P. A., J. M. Watkins, and R. L. Hansing (1997), Nutrients, organic carbon and inorganic nitrogen in the upper water column of the Arctic Ocean: Implications for the sources of dissolved organic carbon, Deep Sea Res., Part II, 44, 1571 1592. Woodgate, R. A., and K. Aagaard (2005), Revising the Bering Strait freshwater flux into the Arctic Ocean, Geophys. Res. Lett., 32, L02602, doi:10.1029/2004gl021747. 11 of 12

Woodgate, R. A., K. Aagaard, and T. J. Weingartner (2005), Monthly temperature, salinity and transport variability of the Bering Strait throughflow, Geophys. Res. Lett., 32, L04601, doi:10.1029/ 2004GL021880. Yamamoto-Kawai, M., N. Tanaka, and S. Pivovarov (2005), Freshwater and brine behaviors in the Arctic Ocean deduced from historical data of d 18 O andalkalinity(1929 2002A.D.),J. Geophys. Res., 110, C10003, doi:10.1029/2004jc002793. Yamamoto-Kawai, M., E. C. Carmack, and F. A. McLaughlin (2006), Nitrogen balance and Arctic throughflow, Nature, 443(43), doi:10.1038/ 443043a. Zhan, X., and J. Zhang (2001), Heat and freshwater budgets and pathways in the Arctic Mediterranean in a coupled ocean/sea-ice model, J. Oceanogr., 57, 207 234. E. C. Carmack, F. A. McLaughlin, and M. Yamamoto-Kawai, Department of Fisheries and Oceans, Institute of Ocean Sciences, Sidney, BC, Canada V8L 4B2. (kawaim@pac.dfo-mpo.gc.ca) S. Nishino and K. Shimada, Institute of Observational Research for Global Change, Japan Agency for Marine-Earth Science and Technology, 2-15 Natsushima, Yokosuka, Kanagawa 237-0061, Japan. 12 of 12