A model study of Circumpolar Deep Water on the West Antarctic Peninsula and Ross Sea continental shelves

A model study of Circumpolar Deep Water on the West Antarctic Peninsula and Ross Sea continental shelves

Deep-Sea Research II 58 (2011) 1508–1523 Contents lists available at ScienceDirect Deep-Sea Research II journal homepage: www.elsevier.com/locate/ds...

4MB Sizes 1 Downloads 50 Views

Deep-Sea Research II 58 (2011) 1508–1523

Contents lists available at ScienceDirect

Deep-Sea Research II journal homepage: www.elsevier.com/locate/dsr2

A model study of Circumpolar Deep Water on the West Antarctic Peninsula and Ross Sea continental shelves Michael S. Dinniman a,n, John M. Klinck a, Walker O. Smith Jr.b a b

Center for Coastal Physical Oceanography, Old Dominion University, Norfolk, VA 23529, USA Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA 23062, USA

a r t i c l e i n f o

a b s t r a c t

Available online 16 December 2010

Transport of relatively warm, nutrient-rich Circumpolar Deep Water (CDW) onto continental shelves around Antarctica has important effects on physical and biological processes. However, the characteristics of the CDW along the shelf break, as well as what happens to it once it has been advected onto the continental shelf, differ spatially. In the present study high resolution (4–5 km) regional models of the Ross Sea and the West Antarctic Peninsula coastal ocean are used to compare differences in CDW transport. The models compared very well with observations from both regions. Examining the fluxes not only of heat, but also of a simulated ‘‘dye’’ representing CDW, shows that in both cases CDW crosses the shelf break in specific locations primarily determined by the bathymetry, but eventually floods much of the shelf. The frequency of intrusions in Marguerite Trough was ca. 2–3 per month, similar to recent mooring observations. A significant correlation between the along shelf break wind stress and the cross shelf break dye flux through Marguerite Trough was observed, suggesting that intrusions are at least partially related to short duration wind events. The primary difference between the CDW intrusions on the Ross and west Antarctic Peninsula shelves is that there is more vigorous mixing of the CDW with the surface waters in the Ross Sea, especially in the west where High Salinity Shelf Water is created. The models show that the CDW moving across the Antarctic Peninsula continental shelf towards the base of the ice shelves not only is warmer initially and travels a shorter distance than that advected towards the base of the Ross Ice Shelf, but it is also subjected to less vertical mixing with surface waters, which conserves the heat available to be advected under the ice shelves. This difference in vertical mixing also likely leads to differences in the supply of nutrients from the CDW into the upper water column, and thus modulates the impacts on surface biogeochemical processes. & 2010 Elsevier Ltd. All rights reserved.

Keywords: Circumpolar Deep Water Antarctic continental shelves Antarctic Peninsula Ross Sea Ocean-Shelf exchange

1. Introduction The core of the Antarctic Circumpolar Current (ACC) that surrounds the Antarctic continent consists of Circumpolar Deep Water (CDW), which is a mixture of deep water from all of the world’s oceans (Orsi et al., 1995). This water is relatively warm (T40 1C) and saline (4 34.6). CDW is typically divided into two types, Upper Circumpolar Deep Water (UCDW), characterized by relatively low oxygen and high nutrient concentrations, and Lower Circumpolar Deep Water (LCDW), which is characterized by higher salinities. These differences are due to the source regions of these water masses, as UCDW is thought to originate in the Indian and Pacific Oceans (Callahan, 1972) and LCDW is derived from modified North Atlantic Deep Water (Whitworth and Nowlin, 1987). CDW is found near the shelf break around the entire continent, except the Weddell Sea (Hofmann and Klinck, 1998), but its

n

Corresponding author. Tel.: + 1 757 683 5559; fax: + 1 757 683 5550. E-mail address: [email protected] (M.S. Dinniman).

0967-0645/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr2.2010.11.013

characteristics differ depending on location (Whitworth et al., 1998). The two locations in our study are the continental shelves along the West Antarctic Peninsula and the Ross Sea. Orsi et al. (1995) define the southern terminus of UCDW as the poleward boundary of the ACC, and in the Bellingshausen Sea this abuts the continental shelf break and contains water with temperatures up to 1.8 1C. The CDW along the shelf break in the Ross Sea is thought to split off from the warm ACC CDW, which is north of the Ross Sea in this area and removed from the shelf break, and flow along as the southern limb of the Ross Gyre where it cools as it travels from east to west (from 1.6 1C at 1301W to o1.2 1C around 1601E; Whitworth et al., 1998), at least partially due to mixing with cold outflows from the central and western shelves of the Ross Sea (Orsi and Wiederwohl, 2009). Transport of CDW onto Antarctic continental shelves is critically important for several physical and biological processes. For example, CDW is thought to supply the majority of the heat needed to maintain the subsurface temperatures of the Antarctic Peninsula shelf region (Klinck, 1998; Klinck et al., 2004; Martinson et al., 2008). Advection of this warm water across the continental shelf is also thought to supply most of the heat involved in the basal melt of

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

several of the ice shelves along the coast in the Amundsen (specifically Pine Island Glacier: Jacobs et al., 1996; Jenkins et al., 1997; Hellmer et al., 1998; Walker et al., 2007) and Bellingshausen Seas (Potter and Paren, 1985; Talbot, 1988; Jenkins and Jacobs, 2008), although some recent work (Holland et al., 2010) suggests that much of the interannual variability in the basal melt of Bellingshausen ice shelves may result more from changes in surface waters than from CDW from off the shelf. In the Ross Sea the CDW supplied to the continental shelf is greatly modified on the shelf and appears as a new water mass called Modified Circumpolar Deep Water (MCDW; Jacobs and Giulivi, 1999; Gordon et al., 2000; Orsi and Wiederwohl, 2009). CDW or MCDW is thought to be important to many processes, including the dynamics of the Ross Sea Polynya (Jacobs and Comiso, 1989; Fichefet and Goosse, 1999; Reddy et al., 2007), the formation of Antarctic Bottom Water in the Ross Sea (Orsi et al., 2001; Gordon et al., 2009; Orsi and Wiederwohl, 2009), and supplying much of the heat involved in the basal melting (although less than along the Antarctic Peninsula) of the Ross Ice Shelf (Jacobs et al., 1979, 1985; Smethie and Jacobs, 2005). CDW transport onto Antarctic continental shelves also has important biological impacts. For example, Pre´zelin et al. (2000) showed that upwelling of CDW near the shelf break along the west Antarctic Peninsula brought nutrients into the euphotic zone, increased primary productivity and resulted in the dominance of the phytoplankton assemblage by diatoms. On the Ross continental shelf, primary production is thought to be seasonally limited by iron (Sedwick and DiTullio, 1997; Sedwick et al., 2000; Olson et al., 2000), and there have been suggestions (Hiscock, 2004; Peloquin and Smith, 2007) that MCDW supplies a significant amount of iron and thus stimulates primary productivity. It has been demonstrated that MCDW commonly occurs at depths of 100–200 m (Orsi and Wiederwohl, 2009) and it was hypothesized that mesoscale processes, such as those described by Hales and Takahashi (2004), were important in enhancing iron input into the surface layer (Peloquin and Smith, 2007). However, Reddy and Arrigo (2006) argue that not enough MCDW (and MCDW-supplied iron) is entrained into the surface waters during summer to significantly impact productivity. Direct observations of these processes remain elusive. In the present study a regional circulation model of the Bellingshausen Sea/West Antarctic Peninsula (WAP) is used to study the timing and forcing mechanisms of transport of CDW onto the continental shelf. Then, the evolution of the CDW once onto the shelf, and how this affects the heat transport to the base of ice shelves, is compared with CDW intrusions in a regional circulation model of the Ross Sea. Finally, the potential impacts on biological processes of these intrusions are discussed.

1509

column and, where needed, the draft below mean sea level of the ice shelf. The ice shelf draft is taken from the BEDMAP gridded digital model of ice thickness and subglacial topography for Antarctica (Lythe et al., 2001). In Dinniman et al. (2007), the sea floor topography also came from BEDMAP, but here much of the sea floor topography not under the RIS came from a digitized version of an updated Ross Sea bathymetry (Davey, 2004). Both bathymetric surfaces were slightly smoothed with a modified Shapiro filter which was designed to selectively smooth areas where the changes in the ice thickness or bottom bathymetry are large with respect to the total depth (Wilkin and Hedstr¨om, 1998). Initial fields of temperature and salinity are taken from the World Ocean Atlas 2001 (WOA01), and the values just north of the RIS are extrapolated southward to represent the below-ice shelf initial conditions. Winds are taken from forecasts from the Antarctic Mesoscale Prediction System (AMPS; Powers et al., 2003; Bromwich et al., 2005). AMPS uses a mesoscale meteorological model to compute high resolution atmospheric forecast fields for operational use in Antarctica. The model grid used provided winds at 30-km horizontal spacing over much of the Southern Ocean (including our entire model domain). The rest of the model atmospheric forcing fields (air pressure, humidity, air temperature and clouds) are the same as in Dinniman et al. (2007). Ocean tides are not included in this model. In place of a fully dynamic sea ice model, ice concentrations from the Special Sensor Microwave Imager (SSM/I) are imposed and ice melt or freeze is calculated from the imposed ice distribution. The model surface heat flux is calculated as a linear combination of heat flux due to ice cover and the open-water heat flux, with the ratio determined by the ice concentration in that grid cell (Markus, 1999). This works well in simulating water mass formation on the Ross Sea continental shelf (Dinniman et al., 2007) because of the difficulties in accurately simulating polynya formation in the western Ross Sea due to inaccuracies in large-scale wind products resolving katabatic winds along ¨ the Victoria Land coast (Petrelli et al., 2008; Husrevo˘ glu, 2008). The open water heat flux was calculated with the COARE version 2.0 bulk flux algorithm (Fairall et al., 1996). The fresh water flux (imposed as a salt flux in the model) is also calculated as a linear combination of open water evaporation minus precipitation and the flux due to ice melting or freezing (Markus, 1999). The only relaxation term in the surface forcing is a very weak (relaxation timescale of 3 years) restoration of the surface salinity to the monthly WOA01 values. Underneath the RIS, the model includes the mechanical and thermodynamic effects of ice shelves on the waters beneath as described in Dinniman et al. (2007) except that now the heat and salt transfer coefficients are no longer constant but are functions of the friction velocity (Holland and Jenkins, 1999). The large icebergs that appeared early in the decade in the Ross Sea (Dinniman et al., 2007) are not included in the model.

2. Circulation models and experiments 2.1. Ross Sea model

2.2. West Antarctic Peninsula model

The Ross Sea model used is that described in Dinniman et al. (2007) with minor modifications. The model uses the Regional Ocean Modeling System (ROMS), which is a primitive equation, finite difference model with a terrain-following vertical coordinate system (Shchepetkin and McWilliams, 2005; Haidvogel et al., 2008; Shchepetkin and McWilliams, 2009). The model domain (Fig. 1) extends from well north of the shelf break (67.51S) southward to 851S and includes almost the entire cavity beneath the Ross Ice Shelf (RIS). The horizontal grid spacing is 5 km and there are 24 vertical levels whose thickness vary with the water column depth but are concentrated towards the top and bottom surfaces. The model simulates the interaction between the floating Ross Ice Shelf and the water cavity underneath and thus two bathymetric surfaces must be defined for the model: the bottom of the water

The circulation model for the Antarctic Peninsula area also uses ROMS and is similar to the Ross Sea model. The model domain starts in the Bellingshausen Sea near Thurston Island in the west, continues eastward along the west side of the Antarctic Peninsula and extends into the Scotia Sea (Fig. 2). The horizontal resolution is 4 km, and there are again 24 vertical levels spaced so that there is enhanced resolution in the surface and bottom layers. The model bathymetry (bottom of the water column and draft below mean sea level of any floating ice shelves) is defined from several sources, including ETOPO2v2 global 2 min resolution bathymetry (Smith and Sandwell, 1997), BEDMAP gridded data (Lythe et al., 2001), a gridded high-resolution bathymetry for the Marguerite Bay area of the Antarctic Peninsula (Bolmer, 2008) and measurements of the ice shelf thickness and bed elevation in the George VI Ice Shelf

1510

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

Fig. 1. Domain for the Ross Sea model. The contour interval for the bathymetry is every 100 m up to 1000 m depth and 250 m deeper than 1000 m. The shaded areas represent the extent of the ice shelf. The contours below the ice shelf represent water column thickness. The arrows represent the mean flow of the westward shelf break current (along the shelf break) and the southern limb of the Ross Gyre.

area (Maslanyj, 1987). Both bathymetric surfaces were slightly smoothed in the same manner as in the Ross Sea model. The primary difference between the Peninsula and the Ross Sea models is that a dynamic sea ice model (Budgell, 2005) is now used to prognostically calculate ice concentration and thickness instead of imposing ice conditions from satellite observations. The Budgell sea ice model that has been added to ROMS is based on ice thermodynamics described by Mellor and Kantha (1989) and ¨ Hakkinen and Mellor (1992). Sea ice is represented by two layers, which allows temperature gradients within the ice. A snow layer is included, which acts as an insulating layer and changes the surface albedo. A molecular sub-layer (Mellor et al., 1986) between the ice bottom and the upper ocean has been added, allowing for more realistic freezing and melting based on heat exchange across the narrow layer. Ice dynamics are based on an elastic–viscous–plastic rheology (Hunke and Dukowicz, 1997; Hunke, 2001). This model also includes the mechanical and thermodynamic effects of ice shelves as in the Ross model.

In our previous models with imposed sea-ice concentrations (e.g. Dinniman et al., 2003; the Ross Sea model in the previous section, Dinniman et al., 2007), vertical momentum and tracer mixing were computed using the K-profile parameterization (KPP) mixing scheme (Large et al., 1994) implemented in ROMS with a modification (found to be necessary due to extreme stabilizing effect of melting ice) where the surface boundary layer depth under stabilizing conditions was set to a minimum depth, equal to the directly wind forced minimum depth under stable conditions in a Kraus/Turner bulk mixed-layer model (Niiler and Kraus, 1977; Dinniman et al., 2003). With the use of a dynamic sea-ice model (and also, perhaps, due to use of higher temporal resolution wind forcing), it was found to be necessary only to set the surface boundary layer depth under stabilizing conditions when the surface shortwave flux was non-zero. Initial fields of temperature and salinity are computed from the Simple Ocean Data Assimilation (SODA version 1.4.2) ocean reanalysis (Carton and Giese, 2008), and initial fields of ice concentration are from an SSM/I climatology. Winds are from AMPS forecasts. Most of

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

the remaining atmospheric conditions needed (air temperature, humidity, sea-level pressure and precipitation) are taken from a monthly climatology derived from AMPS forecasts. Cloud cover is from the ISCCP cloud climatology. Open ocean momentum, heat and fresh water (imposed as a salt flux) fluxes for the model are calculated based on the COARE 3.0 bulk flux algorithm (Fairall et al., 2003) and there is no relaxation of surface temperature or salinity. Open boundaries are handled as in Dinniman and Klinck (2004), except that the temperature, salinity and depth-averaged velocity for the boundaries are now monthly climatologies derived from the SODA reanalysis and ice concentration (from SSM/I) is now needed on the boundaries. Tides are not included in this model. 2.3. Simulations Both models are initialized in mid-September and run for 6 years with a 2-year repeating cycle of daily winds and monthly climatologies for all other forcing. At the end of the spin-up, dye representing CDW is placed in water off the continental shelf (no dye to start with on the shelf) with an initial concentration of 100.0.

1511

For the Ross Sea, CDW is initially defined as water off the continental shelf (defined by the 800 m isobath) at any depth that has a temperature greater than 0.0 1C. For the WAP, CDW is initially defined as water off the continental shelf (defined here by the 1200 m isobath) with a temperature greater than 0.0 1C and below 200 m (due to the presence of surface water in the northern part of the model domain that is warmer than 0.0 1C). In the Ross Sea experiment, the simulation is continued from September 15, 2001 to September 15, 2003 (forced by daily AMPS winds over this period), and the CDW dye is allowed to advect and diffuse over the entire model domain. Because the AMPS 30-km model domain does not fully cover our WAP model domain until November 2002, the WAP experiment simulation covers the period September 15, 2003–2005. This model is forced by twice daily AMPS winds over this period and the CDW dye is again allowed to advect and diffuse over the entire model domain. Note that twice daily wind forcing for the WAP model is different from the daily frequency used for the Ross Sea model; this was done because we wanted the WAP model to be forced by the greatest available temporal resolution of winds since one of the main efforts there was to study the timing of the CDW intrusions. As both models run, there is a continuous source of dye into the model domain as the boundary conditions also contain dye in off-shelf CDW waters. There is no flux of dye across the ocean surface or bottom and there are no sinks of dye other than advection out of the model domain at the open boundaries. A summary of differences between the models can be seen in Table 1. Model computer code setup files, animations of model results and extra figures are available as supplemental material (http://www.ccpo.odu.edu/ msd/DSRpaper).

3. Model results and validation 3.1. CDW Transport onto the West Antarctic Peninsula continental shelf

Fig. 2. Domain for the West Antarctic Peninsula (WAP) model. The contour interval for the bathymetry is every 100 m up to 1000 m depth and 250 m deeper than 1000 m. The shaded areas represent the extent of the ice shelves. The contours below the ice shelf represent water column thickness. ‘‘AI’’ is Anvers Island. Other islands referred to in the text are labeled in Fig. 4. The dashed lines represent the locations of the ACC fronts as defined by Orsi et al. (1995): SBDY—Southern Boundary of the ACC, SACCF—Southern ACC Front, PF—Polar Front.

The initial pathways of CDW onto the continental shelf can be seen from the dye concentrations on the shelf (Fig. 3; also see animation in supplemental material). High concentrations of dye early in the simulation at a constant model level (depth varies with bathymetry, but this layer is ca. 350 m over much of the shelf) match the locations where observations have shown intrusions of CDW. For example, observations (Klinck et al., 2004; Pre´zelin et al., 2004; Martinson et al., 2008; Moffat et al., 2009) have consistently showed increased values of Tmax (maximum temperature below the depth of the deepest mixed layer) at the mouth of Marguerite Trough (MT, Fig. 4). Shipboard ADCP data (Savidge and Amft, 2009) also show strong shoreward flow along the northeastern side of Marguerite Trough. These intrusions not only continue across the shelf into Marguerite Bay along the northeastern side of MT (Moffat et al., 2009), but also transport heat to the northeast towards

Table 1 Summary of differences between model simulations.

Model grid resolution Ocean initial conditions and lateral boundary conditions Time period of simulations Sea-Ice model Modifications to KPP mixing active COARE algorithm Wind forcing frequency (AMPS) Air temperature, atmospheric humidity, sea level pressure source Surface tracer relaxation

WAP simulation

Ross Sea simulation

4 km SODA 1.4.2 September 15, 2003–2005 Dynamic Sea-Ice model (Budgell, 2005) Stable conditions v3.0 12 h AMPS forecasts Temperature: none, salinity: none

5 km WOA01 September 15, 2001–2003 Imposed Sea-Ice (Markus, 1999) Stable conditions and surface shortwave flux40 v2.0 24 h ERA-40 reanalysis Temperature: none, salinity: WOA01 (trelax ¼ 3 years)

1512

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

least as related to CDW intrusions, and these solutions are a good representation of the circulation patterns within the WAP. Hydrographic surveys (20 km resolution or coarser) of the area suggest that these intrusions have horizontal scales on the order of tens of kilometers (Klinck et al., 2004). However, Moffat et al. (2009) suggest, based on mooring data, that the spatial scale of the intrusions may only be on the order of the internal radius of deformation on the shelf (about 4–5 km). With a resolution of 4 km, the model does not fully resolve baroclinic eddies on the shelf, but the narrow horizontal scale (8–20 km in width) of the initial dye patches on the shelf supports the Moffat et al. scale rather than that of the hydrographic surveys. The frequency of UCDW intrusions onto the WAP continental shelf is less well known than the locations. Based on the broadscale hydrography from four cruises in 2001 and 2002, Klinck et al. (2004) estimated that there are 4–6 UCDW intrusions annually. Moffat et al. (2009), using mooring data in MT, estimated that the intrusions have a much higher frequency (approximately 4 per month) and were of short duration (typically 1–3 days, with some events lasting up to 7 days) with no evident seasonality. One-day averages of the advective flux of dye across the entrance to MT (Figs. 4 and 5) show that the dye flux is almost always towards the coast (negative values) and consists of several short-period pulses. There is also no evident seasonality in the modeled intrusions. The minimum flux value used as the definition of an ‘‘intrusion’’ controls the resulting model intrusion frequencies in MT. Moffat et al. (2009) used a threshold of 1.5 1C, which they found was exceeded 15% of the time at mid-depths at the mooring on the eastern/northern side of the trough. If we define an intrusion as an

Fig. 3. Dye concentration on model level 4 of the WAP 60 days (top) and 350 days (bottom) after the dye is allowed to advect from the open ocean onto the shelf.

Lavoisier Island and the southern end of Renaud Island (Martinson et al., 2008). This same path can clearly be seen in the model dye distribution (Fig. 3). Moffat et al. (2009) highlight another intrusion to the south of MT that enters the shelf near 67.51S (also seen in Pre´zelin et al., 2004), and the model also reproduces this incursion. Later in the simulation (Fig. 3, also see supplemental material), there are still areas of high dye concentration showing the intrusion locations, but there is also dye over much of the shelf. There now appears to be considerably more dye on the shelf north of Alexander Island than south of it. Hydrographic observations from instrumented seals were used to create maps of Tmax at 20 km resolution for much of the Bellingshausen Sea continental shelf (Hofmann et al., 2009). Their Tmax plot (Fig. 6, Hofmann et al., 2009) also shows that Tmax is significantly higher on the shelf north of Alexander Island than farther south, suggesting that there is significantly more CDW on the shelf north of Alexander Island. Hence, the model faithfully reproduces the observed currents, at

Fig. 4. Enlargement of area showing where Marguerite Trough intersects the continental slope. The contour interval for the bathymetry is 50 m from 250 to 1000 and 250 m at depths greater than 1000 m (and from 0 to 250 m). The blue line represents the cross section that is used to compute the flux of dye into Marguerite Trough. The green/red line represents the cross section that is used for computing transport along the slope. The black line near the bottom is the cross section shown in Fig. 9. The black line extending northwestward from ‘‘Ad.I.’’ is the cross section shown in Fig. 10. ‘‘MT’’ is the entrance to Marguerite Trough. ‘‘Ad.I.’’ is Adelaide Island. ‘‘L.I.’’ is Lavoisier Island. ‘‘R.I.’’ is Renaud Island (for interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

1513

event having a peak onshore (negative) flux magnitude greater than 45  106 dye units m3 s  1 (and use the Moffat et al. (2009) definition of when events begin/end), we find a mean of 3.0 intrusions per month, with a typical duration of 2–4 days (Fig. 6). If we increase the intrusion threshold to 57  106 dye units m3 s  1, which is exceeded 15% of the time (the same percentage as the Moffat et al. (2009) results), an average of 1.8 intrusions occur per month, with a typical duration of 1–4 days (Fig. 6). Note that the model results are 1-day average fields, while the mooring data are hourly. In both cases the frequency and duration of the intrusions are much closer to those estimated by mooring observations than those estimated from broad-scale hydrography. It should be noted that the dye used in the model to define CDW represents water that contains both UCDW and LCDW, while the Moffat et al. (2009) results presented above are just for UCDW. However, the mean velocities at the mooring in the northeastern entrance to MT are very barotropic (Moffat, 2007) and there was no observed correlation between the UCDW and LCDW inflows (with the LCDW inflows having a significantly longer time scale). Moffat et al. (2009) also show that near the shelf break most of the CDW in the water column is UCDW. Therefore, the model’s high frequency intrusion behavior should primarily represent UCDW.

3.2. CDW Transport onto the Ross Sea continental shelf

Fig. 5. Model dye flux (dye units—Sverdrups) through the gate across Margeurite Trough. Negative values represent flux onto the continental shelf. The dashed lines at  45 and  57 dye units-Sv are explained in the text.

Fig. 6. Histogram of intrusion event duration for the 45 and 57 dye units Sv thresholds.

The pathways of CDW onto the Ross Sea continental shelf can be seen from the dye concentrations on the shelf (Fig. 7, also see animation in supplemental material). Early in the simulation, high concentrations of dye on a constant model level (level 12, approximately half way through the water column) show intrusions onto the continental shelf in several locations, most of which are along the east side of troughs, including Glomar Challenger (intersecting the shelf break around 1761W), Joides (1781E) and Drygalski (1731E). All these locations match well with observations of locations of MCDW intrusions (Jacobs and Giulivi, 1999; Budillon et al., 2003; Orsi and Wiederwohl, 2009). Later in the simulation (Fig. 7, also see supplemental material), areas of high dye concentration remain near the intrusion locations, but there is also dye over almost the entire shelf not underneath the RIS. As can be seen from the animations, it takes longer for dye to cover the shelf in the Ross Sea than in the WAP implying a longer residence time for the Ross Sea. 620 days after the dye has been released, some of the dye has been advected underneath the RIS, primarily entering near Ross Island, but also near 1751W. Observations along the RIS Front (Jacobs and Giulivi, 1999; Smethie and Jacobs, 2005; Loose et al., 2009) clearly show MCDW along the ice shelf edge at 1751W. There is less observational evidence of MCDW entering the RIS cavity on the west side, but there are suggestions that it does (Loose et al., 2009). It is difficult to compare the frequency of the model intrusions onto the Ross shelf with observations. The model does not include tides, which can be strong (velocities up to 1 m s  1; Padman et al., 2003, 2009) near the shelf break in the northwestern Ross Sea. While the effect of the strong tides on CDW intrusion timing is unknown, the tides have been shown to mix MCDW and HSSW near the shelf break (Whitworth and Orsi, 2006; Muench et al., 2009) and have a ‘‘profound’’ (Padman et al., 2009) impact on dense outflows of Antarctic Bottom Water off of the continental shelf (Gordon et al., 2004; Muench et al., 2009; Padman et al., 2009). As such, they can be expected to have a significant impact on the CDW dynamics at the shelf break. In addition, there are no mooring or hydrographic observations that we know of that investigate the timing of the intrusions. The ANSLOPE mooring data presented in Whitworth and Orsi (2006) could possibly be used for such a study, but that is beyond the scope of this paper.

1514

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

Fig. 8. Top panel: Model temperature section across the Ross Ice Shelf front. Solid bold contour line is  1.4 1C (representative of MCDW at depth) and dashed bold contour lines are below  1.90 1C (below surface freezing point and thus representative of ISW). Bottom panel: Temperature section across the Ross Ice Shelf front from the Orsi and Wiederwohl (2009) climatology.

Fig. 7. Dye concentration on model level 12 of the Ross Sea 100 days (top) and 620 days (bottom) after the dye is allowed to advect from the open ocean onto the shelf.

3.3. Transport of heat into ice shelf cavities and basal melt It is important to demonstrate that the models are realistically simulating the basal melt rates and transport in and out of the ice shelf cavities. In the Ross Sea simulation, the annual average basal melt rate of the RIS is 15.3 cm year  1 (also see mean basal melt rate figure in supplemental material), close to the previous simulation value of 13.4 cm year  1 (Dinniman et al., 2007) and within the range of observations (12–22 cm year  1; Shabtaie and Bentley, 1987; Lingle et al., 1991; Jacobs et al., 1992; Loose et al., 2009). Tidal currents under the RIS are typically o5 cm/s except in some areas near the east side of the cavity (Padman et al., 2003) and the lack of tides in the model may lead to an underestimation of the melt rate in these areas. A section of the modeled temperature averaged over summer 2000–2001 (Fig. 8) across the front of the RIS shows MCDW (temperatures greater than 1.4 1C at depth) entering the RIS cavity near 1741W. The section also shows water with temperatures below the surface freezing point (o 1.9 1C) at about 177.51W. Due to the depression of the freezing point of seawater with pressure, water with these temperatures must have been created below the RIS. This water mass (Ice Shelf Water; ISW) is advected northward from under the ice shelf. The model can be compared with a cross section of observed temperature (Fig. 8) from

Fig. 9. Cross section of model annual average velocity in the cross ice shelf front direction at the northernmost model grid points underneath George VI Ice Shelf. Positive values (solid contours) represent flow out from under the Ice Shelf northward into Marguerite Bay. Contour interval is 1 cm s  1.

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

a new high resolution annual climatology of the Ross Sea (Orsi and Wiederwohl, 2009), which is representative of summer conditions when most observations were collected. Both the climatology and model show MCDW in the same location with about the same crosssectional area. There is significantly more cross-sectional area of ISW in the observations than in the model, but the location is the same. The similarity in the temperature cross sections and mean basal melt rates strongly suggests that the model is reasonably simulating the transport of heat into and out of the RIS cavity. In the Antarctic Peninsula simulation the annual average basal melt rate of the George VI Ice Shelf (GVIIS) is 6.0 m year  1 (also see mean basal melt rate figure in supplemental material). This is somewhat greater than the estimates from observations: 2.1 m year  1 (Potter and Paren, 1985), 2.8 m year  1 (Corr et al., 2002) and 3.1–4.8 m year  1 (Jenkins and Jacobs, 2008). It should be noted though that all of these observational values are estimates over different spatial and temporal scales. A cross-section of modeled annual average velocity across the north entrance of the GVIIS cavity (Fig. 9) shows primarily inflow on the eastern side, with the highest inflow speed at depth, and outflow on the western side, with the highest speed just below the base of the ice shelf. The model has a net through-flow from south to north of 0.11 Sv (1 Sv¼106 m3 s  1) through the cavity (which is open on both ends, Fig. 2). The high outflow speed on the western side near the ice shelf base has been observed (Potter and Paren, 1985) and is due to less dense melt water at the base of the ice shelf riding up the ice shelf, and being forced towards the west due to the Coriolis force, before leaving the ice shelf cavity as a plume. A cross section of estimated velocity from observations (Fig. 8 in Jenkins and Jacobs, 2008) shows a similar pattern of strong inflow near-bottom on the east side and stronger

Fig. 10. Cross section of modeled temperature in the WAP (see Fig. 4 for location) in summer (top) and winter (bottom).

1515

outflow near the top on the western side. Jenkins and Jacobs (2008) also estimate that the net flow through the ice shelf cavity is from south to north, although their value of 0.17–0.27 Sv is higher than that predicted by the model. It should be noted, however, that the Jenkins and Jacobs estimate essentially covers a 1 week period, whereas the model average over 1 year contains substantial variability (standard deviation from 5-day model averages¼0.14 Sv). Although the mean basal melt rate in the model is slightly higher than estimates from observations, given the close agreement between the modeled and observed velocity estimates, it appears that the model is reasonably simulating the transport of heat into and out of at least the northern end of the GVIIS cavity. Note that the model bathymetry around the southern entrance to GVIIS (Ronne Entrance area) is much less well defined than that around the northern entrance—Marguerite Bay area (Bolmer, 2008; Padman et al., 2010), which is why we focus on the northern end of the cavity. 3.4. Seasonal changes in vertical structure on the continental shelves There are several differences between the models (Table 1) which can make it difficult to compare their results against each

Fig. 11. Top panel: Salinity in the Ross Sea at 300 m from the Orsi and Wiederwohl climatology (2009). Bottom panel: Ross Sea model average salinity for summer 2001– 2002 at 300 m.

1516

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

other. However, if both models can be shown to accurately simulate the seasonal changes in vertical structure on the two continental shelves, then there can be some confidence that the two models can be used to compare the vertical mixing between the two areas (as shown in Section 4.2). Comparisons of modeled temperature cross sections on the shelf in the WAP in summer and winter (Fig. 10) with observations in the same location (summer: Pre´zelin et al., 2000; winter: Klinck et al., 2004) suggest that the model represents the seasonal vertical mixing reasonably well. Both the model and observations show a near freezing 50–100 m deep surface mixed layer in winter, a shallow warm surface layer with a remnant winter water layer below it in summer and a seasonally invariant pycnocline below. Few winter observations are available in the Ross Sea, and as a result it is difficult to make seasonal comparisons of modeled temperature or salinity distributions with observations. However, one test of the model’s representation of seasonal vertical mixing is the modeled mid-depth and deep salinity structure on the shelf. There is a strong east/west gradient in the deep salinity structure on the Ross Sea shelf (Fig. 11 and Jacobs et al., 1985; Jacobs and Giulivi, 1999; Orsi and Wiederwohl, 2009) and the High Salinity Shelf Water (HSSW) in the west cannot be maintained if the physical processes that create deep vertical mixing are not ¨ well represented (Husrevo˘ glu, 2008). Comparison of the mean model salinity at 300 m (Fig. 11) with the annual salinity climatology of Orsi and Wiederwohl (2009) demonstrates that the model generates the observed salinity structure on the shelf and therefore gives us confidence that the model well represents vertical mixing.

4. Discussion 4.1. Mechanisms for CDW intrusions onto the West Antarctic Peninsula shelf Moffat et al. (2009) suggest that the intrusions they measured in the mooring record resulted from isolated eddies being advected through MT with a horizontal scale of 4–5 km, but do not address the formation process of the intrusions. The horizontal resolution of our model (4 km) does not well resolve baroclinic eddies on the continental shelf. However, as the intrusion frequency and duration match the observations, the model can provide insights into the formation process. In both the Moffat et al. (2009) and our results, the duration of the intrusions corresponds to the synoptic (weather) frequency, generally about 2–8 days. The component of the pseudo-stress !! (product of the wind vector and the wind speed ¼9 u 9 u ) perpendicular to the cross section across which the dye flux was calculated (i.e. the direction of the dye flux) has no significant correlation to the dye flux. However, there is a significant correlation between the component of the pseudo-stress parallel to the cross section and the dye flux (Fig. 12). The correlation is at a maximum when the CDW flux lags the pseudo-stress by 3 days (r ¼0.426; Table 2). The large-lag standard error for the correlation (Sciremammano, 1979) is 0.065, so this result is easily significant at p¼ 0.01 (2.6s). There is no significant correlation between either the along-slope or across-slope volume transport across a section (combined red and green lines: Fig. 4) upstream of MT (Fig. 5) and the dye flux across the MT entrance. However, the along-slope transport for the inner part of the slope (red section of the cross-slope line: Fig. 4) does have a strong correlation with dye flux, with a maximum value of r¼0.638 (large-lag standard error of 0.081) when the along-slope transport leads the dye flux by 2 days. If the correlation between the alongslope velocity and cross MT dye flux is due to an advective effect, the mean advective speed would be 20 cm s  1 given the distance between the centers of the two locations (34 km) and the lag of 2 days. The significant correlation and lag imply that changes in the

100

Pseudo-Stress(*0.4) Dye Flux(*-1.0)

50

0

-50

-100 Sep. 2003

Dec. 2003

Mar. 2004

Jun. 2004

Sep. 2004

100

50

0

-50

-100 Mar. 2004

Date Fig. 12. Time history of model pseudo-stress (red dashed line) parallel to the blue line in Fig. 5 that defines the cross-section for dye flux (thus, perpendicular to the dye flux) and model dye flux anomaly (blue solid line). The pseudo-stress has been multiplied by 0.4 and the units are m2 s  2. The dye flux anomaly has had its sign reversed and the units are dye units-Sv. The black arrows show the events in Fig. 13 when there is (October 2003) and is not (April 2004) an intrusion.

along slope total transport on the inner slope lead to changes in the dye flux across the entrance to MT. The correlation between the along-slope pseudo-stress over the inner slope section and the along slope transport is stronger than the correlation between the pseudostress over the dye transport cross-section and the cross-MT dye flux, with a maximum value of r¼0.549 (large-lag standard error of 0.066) when the pseudo-stress leads the volume transport by 2 days. Finally, there is also a significant correlation between the pseudostress over the inner slope section and the cross-MT dye flux with a maximum value of r¼0.438 (large-lag standard error of 0.070) when the pseudo-stress leads the dye flux by 3 days. The along-slope pseudo-stress is nearly in the same direction as the component of the pseudo-stress parallel to the cross section, and the correlations between each of these and the dye flux are almost identical (0.438 vs. 0.426) which demonstrates how coherent the wind forcing is over the small distance between the two locations (which are only about one AMPS model grid point apart). It appears that changes in the wind along the continental slope drive changes in the volume transport along the inner continental slope with the maximum effect lagged by approximately two days. Then, the high volume transport events on the inner shelf parallel to the slope and upstream of the entrance to MT lead to large intrusions of CDW into MT. The 2-day lag between the alongshore wind stress and the upper slope current is a little longer than the typical lag of 0.5–1.0 days observed in areas where a directly forced relationship has been found (e.g. Noble and Ramp, 2000; Skagseth

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

1517

Table 2 Correlations between different quantities from Section 4.1. The ‘‘Time Shift of Max. Correlation’’ is for the first item listed in the ‘‘Comparison’’ column with respect to the second item. Comparison

Maximum significant correlation

Time shift of max. correlation

Large-lag standard error

Wind along trough vs. dye flux Wind across trough vs. dye flux Transport along shelf slope vs. dye flux Transport along inner shelf slope vs. dye flux Wind along inner shelf break vs. transport along inner shelf break Wind along inner shelf break vs. dye flux

No significant correlation 0.426 No significant correlation 0.638 0.549 0.438

N/A 3 day N/A 2 day 2 day 3 day

N/A 0.065 N/A 0.081 0.066 0.070

and Orvik, 2002), but with the model ACC along the shelf break here, the response on the upper slope may not just be due to local wind stress. At times when no intrusion is present (e.g. April 28, 2004, Fig. 13), the depth-averaged velocity in the MT entrance area shows a strong flow along the inner slope upstream of the MT entrance. This flow separates from the slope near the upstream edge of MT, crosses the entrance to MT while over the continental shelf and then turns back towards the slope along a ridge at the downstream edge of the entrance. When there is a large intrusion (e.g. October 20, 2003, Fig. 13), the flow along the inner slope upstream of the MT entrance is stronger. The depth-averaged current turns more sharply into MT, with little direct return back to the slope, and flows southward along the eastern (downstream) edge of the trough. One possible explanation is momentum advection (note that the flow on the shelf below the surface layer is very barotropic). The continental slope curves onshore upstream of the MT entrance (Fig. 4) before turning offshore near the center of the entrance. Momentum advection would make the along-slope flow continue in a straight line just at the upstream entrance to MT where the continental slope starts to curve away from the coast. When the flow is relatively fast, this flow will continue across the entrance to MT in the direction (almost due east) of the upstream continental slope. When the flow is relatively slow, it may still separate from the continental slope at the entrance to MT, but with a less severe angle (the flow is more able to turn with the curvature of the continental slope at the entrance) and conservation of potential vorticity forces it to curve back offshore when the flow impinges upon the ridge to the north of MT. This corresponds with concepts presented by Dinniman et al. (2003) and Dinniman and Klinck (2004), where CDW was shown to be forced onto the shelf in specific locations, due at least partially to momentum advection and the curvature of the shelf break, and then the general shelf circulation takes the CDW into the interior. An alternate explanation is to consider the entrance to MT to be similar to that of a submarine canyon. Allen (1996) showed that when geostrophic flow is increased along the slope in an upwelling region (shallow bathymetry to the right of the flow in the southern hemisphere) due to a sudden wind event, the cyclonic vorticity generated at the upstream edge of a submarine canyon entrance due to vortex stretching is advected into the canyon. This creates a cyclonic flow over the canyon entrance with a strong ageostrophic on-shore velocity that can, depending on the width of the canyon, travel along the downstream wall. When the upstream flow is slower (‘‘non-intrusion’’), the vorticity stretching happens closer to the upstream rim, which sets up the cyclonic flow further upstream, which in turn provides the flow more of a chance to be forced by the effects of the anti-cyclonic circulation due to vorticity ‘‘squishing’’ as the flow impinges upon the downstream ridge. However, this is feasible only for situations where the flow along the slope is in the upwelling sense (opposite direction as coastal trapped waves or rightbounded in the Southern Hemisphere). There are observed intrusions

lead lead lead lead

in the Ross and Amundsen Seas that occur in locations where the shelf break flow is downwelling (left-bounded), which leads us to question this explanation. A model study of the Pine Island Bay area (Thoma et al., 2008) showed that increasing westerly winds are associated with increased intrusions. This could be associated with upwelling, but it could also be associated with accelerations of the flow along the shelf break and increased intrusion frequency due to flowtopography interaction or increased topographically induced meanders. Thoma et al. (2008) also mentioned that a sensitivity study forced with constant winds showed little shelf-slope exchange below the surface layers.

4.2. Differences in CDW transport and mixing on the two shelves The temperature of the CDW at the shelf break along the Antarctic Peninsula is somewhat warmer than the CDW along the Ross Sea shelf break (  1.8 vs. 1.0 1C). By the time this water is advected to the ice shelf fronts in the two regions, the MCDW has been cooled much more from the source CDW temperature in the Ross Sea (final temperature 1.3 1C; Jacobs and Giulivi, 1999) than along the WAP shelf (e.g. temperature of 1.0 1C at the George VI Ice Shelf front; Jenkins and Jacobs, 2008). Estimated average basal melt rates underneath GVIIS range from 2–5 m year  1 (Potter and Paren, 1985; Corr et al., 2002; Jenkins and Jacobs, 2008) and are an order of magnitude greater than the estimated range of 12–22 cm year  1 for the RIS (Shabtaie and Bentley, 1987; Lingle et al., 1991; Jacobs et al., 1992; Loose et al., 2009). If the heat from the MCDW is a major source of the heat to the base of the ice shelves, then differences in the modification of CDW while on the shelf could be critical in explaining the large differences in basal melt rates of ice shelves of similar thicknesses. Another possibility is that the much greater basal melt rate of the Antarctic Peninsula ice shelves could be due to a greater volume of oceanic water being advected to the shelves, rather than just to a difference in the temperature of the advected water. These models have no heat flux through the bedrock into the ocean and the horizontal diffusion of heat is very small. Therefore, the change in heat of the water in an ice shelf cavity is primarily the sum of the surface heat flux due to basal melt/freeze and the horizontal advection of heat into/out of the cavity. If the advective heat flux into the RIS cavity and GVIIS cavity is compared (Fig. 14), there is actually more heat advected into the cavity underneath GVIIS, even though the volume of water in the RIS cavity (1.3  105 km3) is much greater than in the GVIIS cavity (7.3  103 km3). In response to the greater advection of heat, there is slightly greater total basal melt (integrated over the entire ice shelf) beneath GVIIS than RIS. Since the area of the GVIIS (2.0  104 km2) is much smaller than that of the RIS (4.7  105 km2), this results in a much greater melt rate for GVIIS. The mean dye concentration (Fig. 15) in the two ice shelf cavities is comparable, but slightly greater underneath GVIIS. This means that the

1518

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

Fig. 14. Heat flux into the cavity beneath Ross (blue) and George VI (red) ice shelves. The solid line is the heat advected into the cavity and the dashed line represents the heat exchange between the water surface and the ice. The top panel covers the first year of each simulation (WAP: 9/03–9/04, Ross Sea: 9/01–9/02) and the bottom panel is for the second year (WAP: 9/04–9/05, Ross Sea: 9/02–9/03).

Fig. 13. Depth averaged velocity at the entrance to Marguerite Trough during a period when there is no intrusion (top panel) and when there is an intrusion (bottom panel) of CDW dye. The blue and green/red lines are the same as in Fig. 4 (for interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).

concentration of water underneath the ice shelves that originates from off the shelf is only slightly greater under GVIIS. Therefore, the flux per unit volume of offshore water into the cavities is not greatly different between the two, and the significant difference in the advected heat per unit volume is primarily due to differences in the temperature of the local source water rather than differences in the volume flux. It should be noted that not all the heat advected into each cavity originates from off the shelf, as heating of the surface water near the ice shelf front may play an important role in the basal melt of both the Ross (Dinniman et al., 2007) and George VI (Holland et al., 2010) ice shelves. However, if the heating of nearby surface water was the dominant term in the total heat advection, then the heat advection would be overwhelmed by the summer flux. This is clearly not the case for GVIIS (Fig. 14) and, although important, the summer flux is also not dominant for RIS.

Differences in the surface concentration of dye show why there are such great differences in the temperature of the MCDW that reaches the ice shelves. In both models there is no CDW dye in the surface layer (over the shelf or the abyssal ocean) at the start of the simulation. One year after dye introduction (Fig. 16), there is a significant amount of dye in the surface layer (concentrations 415.0) over much of the Ross Sea continental shelf, with even higher concentrations in the west. In contrast, after 1 year into the simulation on the WAP shelf (Fig. 16), there is very little dye in the surface layer. This difference reflects the large difference in vertical mixing between the two regions. A more quantitative measure of this difference on the two shelves can be obtained by comparing the concentration of dye in the top 100 m and the entire water column in the open Ross Sea continental shelf vs. the ice shelf free area of the WAP continental shelf from Charcot Island to Anvers Island. The mean dye concentration over the entire water column averaged over the first year on the Ross continental shelf (12.8 dye units, not shown) was less than that for the WAP (17.6 dye units), showing that the ratio of on-shelf net transport of CDW to total water volume on the open shelf is lower for the Ross Sea. However, the mean concentration over just the top 100 m was more than four times greater on the Ross continental shelf (9.33 dye units; also see surface dye concentration figure in supplemental material) than that for the WAP (2.05 dye units). This indicates that there is much more vigorous vertical mixing of the MCDW on the continental shelf in the Ross Sea than along the WAP. This is not surprising given that the western Ross Sea is known to be a site of strong vertical mixing and HSSW formation due to extensive sea-ice formation and brine rejection over large areas that have reduced ice cover in winter due to strong katabatic winds ¨ (Jacobs et al., 1985; Budillon and Spezie, 2000; Husrevo˘ glu, 2008).

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

1519

Mean Dye Concentration (Ice Shelf Cavities) 10 8

RIS

concentration

GVI

6 4 2 0

Sep

Dec

Mar

Jun

Sep

Dec

Mar

Jun

Sep

10

concentration

8 6 4 2 0 Sep

Fig. 15. Time history of mean dye concentration in the cavity beneath Ross (blue/ solid) and George VI (red/dashed) ice shelves.

Conversely, there is no HSSW water formed on the shelf in the WAP region (Hofmann and Klinck, 1998; Martinson et al., 2008), and Howard et al. (2004) found little vertical mixing through the pycnocline of the UCDW. The difference in the vertical mixing of the MCDW on the continental shelf between the Ross Sea and the WAP might even be underestimated in the models due to the lack of tides. Modeling (Padman et al., 2002) and observations (e.g. Beardsley et al., 2004) show that in the WAP the tides at the shelf break and on the shelf are generally much weaker than in the Ross Sea. The large tidal excursions along the shelf break in the northwestern Ross Sea are thought to be important in helping dense water flow off the continental shelf (Gordon et al., 2004; Muench et al., 2009; Padman et al., 2009) and thus could conceivably increase the amount of CDW to flow onto the shelf. Also, the tides on the shelf itself, while not excessively strong (10–20 cm s  1 maximum velocities over most of the shelf; Padman et al., 2003), could increase the amount of vertical mixing of MCDW with surface waters. The lack of these two processes in our Ross Sea simulation may cause the model to underestimate the surface concentrations of CDW over the shelf. We can estimate the contribution to the vertical heat flux from CDW by computing a scaled heat flux where 5-day averages of the vertical heat flux into each model grid cell (computed internally by the model) are multiplied by the 5-day average dye concentration/ 100 for the same cell. A comparison of the scaled heat flux through 200 m for the open Ross Sea continental shelf and the Marguerite Bay area (Fig. 17) shows that there is considerably more heat lost on average from CDW on the Ross shelf ( 0.96 W m  2) than on the WAP shelf (0.31 W m  2). That is, water on the open shelf that originated as CDW loses much more heat due to vertical mixing in the Ross Sea than in the west Antarctic Peninsula. Therefore, the on-shelf

Fig. 16. Dye concentration in the surface layer of the WAP (upper) and Ross Sea (lower) models 1 year after the dye is allowed to advect from the open ocean onto the shelf. Note that the scale here is different from the plots of dye concentration at depth.

modification of the CDW appears to be substantially more important than differences in cross-shelf transport fluxes in determining the heat advected to the ice shelf cavities (and subsequent basal melt). 4.3. Biological implications of differences of vertical mixing of CDW This difference in vertical mixing also has potentially important implications for the biology in the two areas. For example, as the CDW off the Antarctic Peninsula has roughly similar macro- and micronutrient characteristics as it does near the Ross Sea, there will be significantly less nutrient input from off-shelf CDW into the surface layer on the WAP shelf compared to that in the Ross Sea. In both regions there is significant winter mixing (and hence nutrient input), but the intrusions may provide additional nutrients (especially micronutrients) that fuel a greater seasonal productivity in those areas influenced by

1520

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

50 g C m  2 year  1 for the Ross Sea shelf. The algal assimilation ratio used here is not conservative (and probably only appropriate for Phaeocystis antarctica), so these estimates are clearly upper bounds. Estimated net annual primary productivity for the Ross shelf ranges from 100 to 200 g C m  2 year  1 (Arrigo et al., 2008; Smith and Comiso, 2008), so the estimates of the productivity supported by MCDW intrusions of Fe suggest that CDW may be a significant source of dissolved Fe for primary production in the Ross Sea. Estimated net annual primary production in the shallow coastal waters near Palmer Station on the WAP shelf is 47–351 g C m  2 year  1 (Ducklow et al., 2006) and integrated shelf productivity is ca. 60 g C m  2 year  1 (Smith and Comiso, 2008). As such, our estimates of iron inputs to support productivity suggest that CDW may not be a significant source of Fe in this area.

Vertical Diffusive Heat Flux (Dyed Shelf Water below 200m) 1

Heat Flux (Watts m-2)

0 -1 -2 -3

Ross Sea shelf Marguerite Bay area

-4 -5 Sep

4.4. Potential future changes

Dec

Mar

Jun

Sep

1

Heat Flux (Watts m-2)

0 -1 -2 -3 -4 -5 Sep

Dec

Mar

Jun

Sep

Fig. 17. Vertical diffusive heat flux across 200 m scaled by (dye concentration/100) in each model grid cell for the open shelf adjacent to the Ross Ice Shelf (blue/solid) and the open shelf from Marguerite Bay to the shelf break in the WAP (red/dashed) model.

intrusions. For example, the western Ross Sea is more strongly impacted by intrusions than the eastern region, and it has been shown that seasonal productivity in at least one year was greater in the west than east (Hiscock, 2004). Interannual variations in phytoplankton dynamics may also reflect the nutrient input from MCDW intrusions (Peloquin and Smith, 2007). While observations of iron concentrations early in the growing season are limited, the potential for influencing seasonal productivity via trace-metal inputs remains. There is no comparable experimental verification of iron limitation in the WAP, although Pre´zelin et al., (2000) found that upwelling of CDW resulted in a change of phytoplankton composition from flagellates to diatoms, and this change could have resulted from Fe (or other micronutrient) inputs. Since the model can calculate the flux of CDW into a given area, and if an initial dissolved Fe concentration of CDW is assumed, it is possible to estimate how much primary production can be supported by CDW flux of dissolved Fe into the euphotic zone. If we assume an initial dissolved Fe concentration of 0.5 nM in CDW (P. Sedwick, personal communication) and a C:Fe algal assimilation ratio of 450,000 mol mol  1 (Tagliabue and Arrigo, 2005), then scaling from the total (advective and diffusive) CDW flux into the upper 100 m of the open shelf over a growing season (defined as 11/1–2/28 of the first year of the model run) gives a Fe flux that can support primary production of o1 g C m  2 year  1 for the WAP shelf and 17 g C m  2 year  1 for the Ross Sea shelf. If we assume that dissolved Fe that advects or diffuses into the top 100 m at any time of the year is bio-available during the growing season, then the estimates of enhanced MCDW-fueled productivity increase to 10 g C m  2 year  1 for the WAP shelf and

There has been a shift towards positive polarity in the Southern Annular Mode (SAM) in recent years (Marshall, 2003) and this is associated with a strengthening and poleward displacement of the circumpolar westerlies. For the Antarctic Peninsula, positive SAM index has been related to an increase in the frequency of mesoscale cyclones and a shift in the storm tracks to favor more east-bound trajectories, consistent with the strengthening of the westerlies (Lubin et al., 2008). If the intrusions, at least on the WAP shelf, are related to the strength and frequency of the high frequency storm events, then this could indicate an increase in the frequency and amount of CDW and heat transported onto the continental shelf. These changes in the winds may also lead to more vertical mixing and loss of this heat to the atmosphere. However, if this water moves under ice shelves quickly enough to retain most of its excess heat, then it can increase the melt rate of the bottom of the ice shelves. Biologically, increased winds on the shelf would likely result in a decreased primary productivity, as well as a shift in phytoplankton composition from diatoms to flagellates as has been observed in the past two decades (Montes-Hugo et al., 2009). Such a decrease would have substantial ecological and biogeochemical effects (e.g. reduced upper trophic level productivity; altered vertical flux patterns), but the full extent of these impacts is difficult to ascertain at this time. Changes in the winds also might be predicted to impact the Ross Sea. For example, increased westerlies to the north of the Ross Sea continental shelf can reasonably be expected to increase the strength of the Ross Gyre, whose western-traveling southern limb forms the shelf break current. Hence, increased frequency, duration and strength of intrusions might follow, and this could in turn have a direct impact on micronutrient inputs onto the shelf. Thus, the biological responses of MCDW intrusions would increase, which could potentially induce a trophic cascade within the food web. For example, MCDW intrusions on the Ross Sea shelf are apparently linked to the locations and reproduction of Antarctic krill (Euphausia superba) (Sala et al., 2002). Antarctic krill are found only in association with the shelf break in the Ross Sea, and are fed upon by various predators such as penguins and Antarctic toothfish (Smith et al., 2007); indeed, Ade lie penguins in the northern Ross Sea apparently target Antarctic krill, and thus may be dependent on MCDW intrusions (P. Lyver, personal communication). Changes in the frequency of MCDW intrusions might result in a number of unexpected changes in the trophic linkages on the Ross Sea continental shelf.

5. Summary Regional circulation models for the West Antarctic Peninsula and the Ross Sea were used to examine the dynamics of Circumpolar Deep Water intrusions onto the continental shelf. Both models

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

accurately simulate the locations and, at least for the WAP model, frequency of these intrusions. The Ross Sea model simulates well the annual average basal melt rate beneath the Ross Ice Shelf (15.3 cm year  1) and appears to properly simulate the pathways of MCDW into and ISW out of the cavity underneath the ice shelf. The WAP simulation of the annual average basal melt rate beneath George VI Ice Shelf (6.0 m year  1) is somewhat higher than the estimates from observations, but the model simulates the net volume transport through and the velocities at the north end of the cavity beneath the ice shelf. Both models calculate the seasonal changes in the density structure on the shelf. The model suggests that the CDW intrusions in the WAP are relatively frequent (approximately 2–3 per month) with no apparent seasonality, as opposed to hydrographic based estimates of 4–6 intrusions per year. The model intrusions have a significant correlation between the along-shelf break wind stress and the CDW flux (as represented by a modeled dye) through MT suggesting that intrusions are at least partially related to short-duration wind events. The mechanism for these events may be due to momentum advection of the more intense flow on the inner shelf (due to the wind), but there also may be other causes. A more detailed study of these processes will require smaller grid spacing to better resolve the vorticity generated at the trough entrance that may result in eddying behavior. The primary difference between the CDW intrusions on the Ross and WAP shelves is that there is more vigorous mixing of the CDW with the surface waters in the Ross Sea, especially in the west. The CDW moving onto the WAP shelf towards the base of the ice shelves not only starts out warmer and travels a shorter distance than that advected towards the base of the Ross Ice Shelf, but it is also subjected to less vertical mixing and exchange with surface waters. This allows for more of the heat originally contained in the CDW to advect underneath the ice shelves in the WAP area and contribute to their melting. This difference in vertical mixing of CDW between the two areas also likely leads to differences in nutrient supply into the upper water column. Estimates based on model CDW flux into the top 100 m of the water column give upper bounds to the primary productivity that can be supported by CDW Fe flux of 50 g C m  2 year  1 for the Ross shelf and 10 g C m  2 year  1 for the WAP shelf. CDW may be a significant source of dissolved Fe for primary production in the Ross Sea, but is not likely to be one for the WAP shelf. Potential changes in the frequency, strength and duration of intrusions also may have significant impacts on the trophic linkages and biogeochemistry of the continental shelves.

Acknowledgments This research was supported by the National Science Foundation under Grant numbers ANT-0523172 and OPP-03-37247. The AMPS data was provided by John Cassano and AMPS is supported by US National Science Foundation support to NCAR, Ohio State University and the University of Colorado. We thank Laurie Padman, Stan Jacobs, Tom Bolmer and the BEDMAP Consortium for help with digital bathymetry. We also thank Alex Orsi and Chrissy Wiederwohl for the use of their climatology of the Ross Sea. Comments from the editor and two very insightful reviews contributed greatly to improving the paper. This is US GLOBEC contribution no. 691 and VIMS contribution 3127.

References Allen, S.E., 1996. Topographically generated, subinertial flows within a finite length canyon. Journal of Physical Oceanography 26, 1608–1632.

1521

Arrigo, K.R., van Dijken, G., Long, M., 2008. Coastal Southern Ocean: a strong anthropogenic CO2 sink. Geophysical Research Letters 35 (L21602). doi:10.1029/2008GL035624. Beardsley, R.C., Limeburner, R., Owens, W.B., 2004. Drifter measurements of surface currents near Marguerite Bay on the western Antarctic Peninsula shelf during austral summer and fall, 2001 and 2002. Deep-Sea Research II 51, 1947–1964. Bolmer, S.T., 2008. A note on the development of the bathymetry of the continental margin west of the Antarctic Peninsula from 651 to 711S and 651 to 781W. Deep-Sea Research II 55, 271–276. doi:10.1016/j.dsr2.2007.10.004. Bromwich, D.H., Monaghan, A.J., Manning, K.W., Powers, J.G., 2005. Real-time forecasting for the Antarctic: an evaluation of the Antarctic Mesoscale Prediction System (AMPS). Monthly Weather Review 133, 579–603. Budgell, P., 2005. Numerical simulation of ice-ocean variability in the Barents Sea region. Ocean Dynamics 55, 370–387. Budillon, G., Spezie, G., 2000. Thermohaline structure and variability in the Terra Nova Bay polynya, Ross Sea. Antarctic Science 12, 493–508. Budillon, G., Pacciaroni, M., Cozzi, S., Rivaro, P., Catalano, G., Ianni, C., Cantoni, C., 2003. An optimum multiparameter mixing analysis of the shelf waters in the Ross Sea. Antarctic Science 15, 105–118. Callahan, J.E., 1972. The structure and circulation of deep water in the Antarctic. Deep-Sea Research 19, 563–575. Carton, J.A., Giese, B.A., 2008. A reanalysis of ocean climate using SODA. Monthly Weather Review 136, 2999–3017. Corr, H.F.J., Jenkins, A., Nicholls, K.W., Doake, C.S.M., 2002. Precise measurement of changes in ice-shelf thickness by phase-sensitive radar to determine basal melt rates. Geophysical Research Letters 29, 10.1029/2001GL014606. Davey, F.J., 2004. Ross Sea Bathymetry, 1:2,000,000, Version 1.0, Institute of Geological & Nuclear Sciences Geophysical Map 16. Institute of Geological & Nuclear Sciences Limited, Lower Hutt, New Zealand. Dinniman, M.S., Klinck, J.M., Smith Jr., W.O., 2003. Cross-shelf exchange in a model of the Ross Sea circulation and biogeochemistry. Deep-Sea Research II 50, 3103–3120. Dinniman, M.S., Klinck, J.M., 2004. A model study of circulation and cross shelf exchange on the west Antarctic Peninsula continental shelf. Deep-Sea Research II 51, 2003–2022. Dinniman, M.S., Klinck, J.M., Smith Jr., W.O., 2007. The influence of sea ice cover and icebergs on circulation and water mass formation in a numerical circulation model of the Ross Sea, Antarctica. Journal of Geophysical Research 112, C11013. doi:10.1029/2006JC004036. Ducklow, H.W., Fraser, W., Karl, D.M., Quetin, L.B., Ross, R.M., Smith, R.C., Stammerjohn, S.E., Vernet, M., Daniels, R.M., 2006. Water column processes in the West Antarctic Peninsula and the Ross Sea: interannual variations and foodweb structure. Deep-Sea Research II 53, 834–852. Fairall, C.W., Bradley, E.F., Rogers, D.P., Edson, J.B., Young, G.S., 1996. Bulk parameterization of air-sea fluxes for Tropical Ocean-Global Atmosphere Coupled-Ocean Atmosphere Response Experiment. Journal of Geophysical Research 101, 3747–3764. Fairall, C.W., Bradley, E.F., Hare, J.E., Grachev, A.A., Edson, J.B., 2003. Bulk parameterization of air-sea fluxes: updates and verification for the COARE algorithm. Journal of Climate 16, 571–591. Fichefet, T., Goosse, H., 1999. A numerical investigation of the spring Ross Sea polynya. Geophysical Research Letters 26, 1015–1018. Gordon, A.L., Zambianchi, E., Orsi, A., Visbeck, M., Giulivi, C.F., Whitworth III, T., Spezie, G., 2004. Energetic plumes over the western Ross Sea continental slope. Geophysical Research Letters 31, L21302. doi:10.1029/2004GL020785. Gordon, A.L., Orsi, A.H., Muench, R., Huber, B.A., Zambianchi, E., Visbeck, M., 2009. Western Ross Sea continental slope gravity currents. Deep-Sea Research II 56, 796–817. Gordon, L.I., Codispoti, L.A., Jennings, J.C., Millero, F.J., Morrison, J.M., Sweeney, C., 2000. Seasonal evolution of hydrographic properties in the Ross Sea, Antarctica, 1996–1997. Deep-Sea Research II 47, 3095–3117. Haidvogel, D.B., Arango, H., Budgell, W.P., Cornuelle, B.D., Curchitser, E., Di Lorenzo, E., Fennel, K., Geyer, W.R., Hermann, A.M., Lanerolle, L., Levin, J., McWilliams, J.C., Miller, A.J., Moore, A.M., Powell, T.M., Shchepetkin, A.F., Sherwood, C.R., Signell, R.P., Warner, J.C., Wilkin, J., 2008. Ocean forecasting in terrain-following coordinates: formulation and skill assessment of the Regional Ocean Modeling System. Journal of Computational Physics 227, 3595–3624. doi:10.1016/ j.jcp.2007.06.016. ¨ Hakkinen, S., Mellor, G.L., 1992. Modeling the seasonal variability of a coupled arctic ice-ocean system. Journal of Geophysical Research 97, 20285–20304. Hales, B., Takahashi, T., 2004. High-resolution biogeochemical investigation of the Ross Sea, Antarctica, during the AESOPS (US JGOFS) Program. Global Biogeochemical Cycles 18, GB3006. doi:10.1029/2003GB002165. Hellmer, H.H., Jacobs, S.S., Jenkins, A., 1998. Oceanic erosion of a floating Antarctic glacier in the Amundsen Sea. In: Jacobs, S.S., Weiss, R.F. (Eds.), Ocean, Ice, and Atmosphere: Interactions at the Antarctic Continental Margin, AGU Antarctic Research Series, vol. 75, pp. 83–99. Hiscock, M., 2004. The regulation of primary productivity in the Southern Ocean. Ph.D. Dissertation, Duke University, Durham, NC, 150pp. Hofmann, E.E., Klinck, J.M., 1998. Hydrography and circulation of the Antarctic continental shelf: 1501E to the Greenwich Meridian. In: Robinson, A.R., Brink, K.H. (Eds.), The Sea, The Global Coastal Ocean, Regional Studies and Synthesis, vol. 11. John Wiley & Sons, New York, pp. 997–1042. Hofmann, E.E., Costa, D.P., Daly, K., Dinniman, M.S., Klinck, J.M., Marrari, M., Padman, ˜ ones, A., 2009. Results from US Southern Ocean GLOBEC Synthesis Studies. L., Pin GLOBEC International Newsletter 15 (1), 43–48.

1522

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

Holland, D.M., Jenkins, A., 1999. Modelling thermodynamic ice-ocean interactions at the base of an ice shelf. Journal of Physical Oceanography 29, 1787–1800. Holland, P.R., Jenkins, A., Holland, D.M., 2010. Ice and ocean processes in the Bellingshausen Sea, Antarctica. Journal of Geophysical Research 115, C05020, doi: 10.1029/2008JC005219. Howard, S.L., Hyatt, J., Padman, L., 2004. Mixing in the pycnocline over the western Antarctic Peninsula shelf during Southern Ocean GLOBEC. Deep-Sea Research II 51, 1965–1979. Hunke, E.C., Dukowicz, J.K., 1997. An elastic–viscous–plastic model for sea ice dynamics. Journal of Physical Oceanography 27, 1849–1867. Hunke, E.C., 2001. Viscous–plastic sea ice dynamics with the EVP model: linearization issues. Journal of Computational Physics 170, 18–38. ¨ Husrevo˘ glu, Y.S., 2008. Modeling the seasonal sea ice cycle in the Ross Sea, Antarctica. Ph.D, Dissertation, Old Dominion University, 172 pp. Jacobs, S.S., Gordon, A.L., Ardai Jr., J.L., 1979. Circulation and melting beneath the Ross Ice Shelf. Science 203, 439–443. Jacobs, S.S., Fairbanks, R.G., Horibe, Y., 1985. Origin and evolution of water masses 16 near the Antarctic continental margin: evidence from H18 2 O2/H2 O2 ratios in seawater. In: Jacobs, S.S. (Ed.), Oceanology of the Antarctic Continental Shelf, AGU Antarctic Research Series, vol. 43, pp. 59–85. Jacobs, S.S., Comiso, J.C., 1989. Sea ice and oceanic processes on the Ross Sea continental shelf. Journal of Geophysical Research 94, 18,195–18,211. Jacobs, S.S., Hellmer, H.H., Doake, C.S.M., Jenkins, A., Frolich, R.M., 1992. Melting of ice shelves and the mass balance of Antarctica. Journal of Glaciology 38, 375–387. Jacobs, S.S., Hellmer, H.H., Jenkins, A., 1996. Antarctic ice sheet melting in the Southeast Pacific. Geophysical Research Letters 23, 957–960. Jacobs, S.S., Giulivi, C.F., 1999. Thermohaline data and ocean circulation on the Ross Sea continental shelf. In: Spezie, G., Manzella, G.M.R. (Eds.), Oceanography of the Ross Sea, Antarctica. Springer, Milan, pp. 3–16. Jenkins, A., Vaughn, D.G., Jacobs, S.S., Hellmer, H.H., Keys, J.R., 1997. Glaciological and oceanographic evidence of high melt rates beneath Pine Island Glacier, west Antarctica. Journal of Glaciology 43, 114–121. Jenkins, A., Jacobs, S.S., 2008. Circulation and melting beneath George VI Ice Shelf, Antarctica. Journal of Geophysical Research 113, C04013. doi:10.1029/ 2007JC004449. Klinck, J.M., 1998. Heat and salt changes on the continental shelf west of the Antarctic Peninsula between January 1993 and January 1994. Journal of Geophysical Research 103, 7617–7636. Klinck, J.M., Hofmann, E.E., Beardsley, R.C., Salihoglu, B., Howard, S., 2004. Water mass properties and circulation on the west Antarctic Peninsula continental shelf in austral fall and winter 2001. Deep-Sea Research II 51, 1925–1946. Large, W.G., McWilliams, J.C., Doney, S.C., 1994. Oceanic vertical mixing: a review and a model with nonlocal boundary layer parameterization. Reviews of Geophysics 32, 363–403. Lingle, C.S., Schilling, D.H., Fastook, J.L., Paterson, W.S.B., Brown, T.J., 1991. A flow band model of the Ross Ice Shelf, Antarctica—response to CO2-induced climatic warming. Journal of Geophysical Research 96, 6849–6871. Loose, B., Schlosser, P., Smethie, W.M., Jacobs, S., 2009. An optimized estimate of glacial melt from the Ross Ice Shelf using noble gases, stable isotopes, and CFC transient tracers. Journal of Geophysical Research 114, C08007. doi:10.1029/ 2008JC005048. Lubin, D., Wittenmyer, R.A., Bromwich, D.H., Marshall, G.J., 2008. Antarctic Peninsula mesoscale cyclone variability and climatic impacts influenced by the SAM. Geophysical Research Letters 35, L02808. doi:10.1029/2007GL032170. Lythe, M.B., Vaughan, D.B., BEDMAP Consortium, 2001. BEDMAP: a new ice thickness and subglacial topographic model of Antarctica. Journal of Geophysical Research 106, 11335–11351. Markus, T., 1999. Results from an ECMWF-SSM/I forced mixed layer model of the Southern Ocean. Journal of Geophysical Research 104, 15,603–15,620. Marshall, G.J., 2003. Trends in the Southern annular mode from observations and reanalyses. Journal of Climate 16, 4134–4143. Martinson, D.G., Stammerjohn, S.E., Iannuzzi, R.A., Smith, R.C., Vernet, M., 2008. Western Antarctic Peninsula physical oceanography and spatio-temporal variability. Deep-Sea Research II 55, 1964–1987. Maslanyj, M.P., 1987. Seismic bedrock depth measurements and the origin of George VI Sound, Antarctic Peninsula. British Antarctic Survey Bulletin 75, 51–65. Mellor, G.L., McPhee, M.G., Steele, M., 1986. Ice seawater turbulent boundary-layer interaction with melting or freezing. Journal of Physical Oceanography 16, 1829–1846. Mellor, G.L., Kantha, L., 1989. An ice-ocean coupled model. Journal of Geophysical Research 94, 10,937–10,954. Moffat, C., 2007. Ocean circulation and dynamics on the West Antarctic Peninsula continental shelf. Ph.D. Dissertation, MIT/WHOI Joint Program, 186pp. Moffat, C., Owens, B., Beardsley, R.C., 2009. On the characteristics of Circumpolar Deep Water Intrusions to the west Antarctic Peninsula Continental Shelf. Journal of Geophysical Research 114, C05017. doi:10.1029/2008JC004955. Montes-Hugo, M., Doney, S.C., Ducklow, H.W., Fraser, W., Martinson, D., Stammerjohn, S.E., Schofield, O., 2009. Recent changes in phytoplankton communities associated with rapid regional climate change along the Western Antarctic Peninsula. Science 323, 1470–1473. doi:10.1126/science.1164533. Muench, R., Padman, L., Gordon, A., Orsi, A., 2009. A dense water outflow from the Ross Sea, Antarctica: mixing and the contribution of tides. Journal of Marine Systems 77, 369–387.

Niiler, P.P., Kraus, E.B., 1977. One-dimensional models of the Upper Ocean. In: Krauss, E.B. (Ed.), Modeling and Prediction of the Upper Layers of the Ocean. Pergamon, New York, pp. 143–172. Noble, M.A., Ramp, S.R., 2000. Subtidal currents over the central California slope: evidence for offshore veering of the undercurrent and for direct, wind-driven slope currents. Deep-Sea Research II 47, 871–906. Olson, R., Sosik, H., Chekalyuk, A., Shalapyonok, A., 2000. Effects of iron enrichment on phytoplankton in the Southern Ocean during late summer: active fluorescence and flow cytometric analyses. Deep-Sea Research II 47, 3181–3200. Orsi, A.H., Whitworth III, T., Nowlin Jr., W.D., 1995. On the meridional extent and fronts of the Antarctic Circumpolar Current. Deep-Sea Research I 42, 641–673. Orsi, A.H., Jacobs, S.S., Gordon, A.L., Visbeck, M., 2001. Cooling and ventilating the Abyssal Ocean. Geophysical Research Letters 28, 2923–2926. Orsi, A.H., Wiederwohl, C.L., 2009. A recount of Ross Sea water. Deep-Sea Research II 56, 778–795. Padman, L., Fricker, H.A., Coleman, R., Howard, S., Erofeeva, S., 2002. A new tidal model for the Antarctic ice shelves and seas. Annals of Glaciology 34, 247–254. Padman, L., Erofeeva, S., Joughin, I., 2003. Tides of the Ross Sea and Ross Ice Shelf cavity. Antarctic Science 15, 31–40. doi:10.1017/S0954102003001032. Padman, L., Howard, S.L., Orsi, A.H., Muench, R.D., 2009. Tides of the northwestern Ross Sea and their impact on dense outflows of Antarctic Bottom Water. Deep-Sea Research II 56, 818–834. Padman, L., Costa, D.P., Bolmer, S.T., Goebel, M.E., Huckstadt, L.A., Jenkins, A., McDonald, B.J., Shoosmith, D.R., 2010. Seals map bathymetry of the Antarctic continental shelf. Geophysical Research Letters 37, L21601. doi:10.1029/ 2010GL044921. Peloquin, J.A., Smith Jr., W.O., 2007. Phytoplankton blooms in the Ross Sea, Antarctica: interannual variability in magnitude, temporal patterns, and composition. Journal of Geophysical Research 112, C08013. doi:10.1029/ 2006JC003816. Petrelli, P., Bindoff, N.L., Bergamasco, A., 2008. The sea ice dynamics of Terra Nova Bay and Ross Ice Shelf Polynyas during a spring and winter simulation. Journal of Geophysical Research 113, C09003. doi:10.1029/2006JC004048. Potter, J.R., Paren, J.G., 1985. Interaction between ice shelf and ocean in George VI Sound, Antarctica. In: Jacobs, S.S. (Ed.), Oceanology of the Antarctic Continental Shelf, AGU Antarctic Research Series, vol. 43, pp. 35–58. Powers, J.G., Monaghan, A.J., Cayette, A.M., Bromwich, D.H., Kuo, Y.-H., Manning, K.W., 2003. Real-time mesoscale modeling over Antarctica: the Antarctic Mesoscale Prediction System (AMPS). Bulletin of the American Meteorological Society 84, 1533–1545. Pre´zelin, B.B., Hofmann, E.E., Mengelt, C., Klinck, J.M., 2000. The linkage between Upper Circumpolar Deep Water (UCDW) and phytoplankton assemblages on the west Antarctic Peninsula continental shelf. Journal of Marine Research 58, 165–202. Pre´zelin, B.B., Hofmann, E.E., Moline, M., Klinck, J.M., 2004. Physical forcing of phytoplankton community structure and primary production in continental shelf waters of the western Antarctic Peninsula. Journal of Marine Research 62, 419–460. Reddy, T.E., Arrigo, K.R., 2006. Constraints on the extent of the Ross Sea phytoplankton bloom. Journal of Geophysical Research 111, C07005. doi:10.1029/ 2005JC003339. Reddy, T.E., Arrigo, K.R., Holland, D.M., 2007. The role of thermal and mechanical processes in the formation of the Ross Sea summer polynya. Journal of Geophysical Research 112, C07027. doi:10.1029/2006JC003874. Sala, A., Azzali, M., Russo, A., 2002. Krill of the Ross Sea: distribution, abundance and demography of Euphausia superba and Euphausia crystallorophias during the Italian Antarctic Expedition (January–February 2000). Scientia Marina 60, 123–133. Savidge, D.K., Amft, J.A., 2009. Circulation on the West Antarctic Peninsula derived from 6 years of shipboard ADCP transects. Deep-Sea Research I 56, 1633–1655. Sciremammano Jr., F., 1979. A suggestion for the presentation of correlations and their significance levels. Journal of Physical Oceanography 9, 1273–1276. Sedwick, P., DiTullio, G., 1997. Regulation of algal blooms in Antarctic shelf waters by the release of iron from melting sea ice. Geophysical Research Letters 24, 2515–2518. Sedwick, P., DiTullio, G., Mackey, D., 2000. Iron and manganese in the Ross Sea, Antarctica: seasonal iron limitation in Antarctic shelf waters. Journal of Geophysical Research 105, 11,321–11,336. Shabtaie, S., Bentley, C.R., 1987. West Antarctic ice streams draining into the Ross Ice Shelf: configuration and mass balance. Journal of Geophysical Research 92, 1311–1336. Shchepetkin, A.F., McWilliams, J.C., 2005. The regional oceanic modeling system (ROMS): a split-explicit, free-surface, topography-following coordinate oceanic model. Ocean Modelling 9, 347–440. Shchepetkin, A.F., McWilliams, J.C., 2009. Correction and commentary for ‘‘Ocean forecasting in terrain-following coordinate: formulation and skill assessment of the regional ocean modeling system’’ by Haidvogel et al., J. Comp. Phys. 227. Journal of Computational Physics 228, 8985–9000. doi:10.1016/j.jcp.2009. 09.002. Skagseth, Ø., Orvik, K.A., 2002. Identifying fluctuations in the Norwegian Atlantic Slope Current by means of empirical orthogonal functions. Continental Shelf Research 22, 547–563. Smethie Jr., W.M., Jacobs, S.S., 2005. Circulation and melting under the Ross Ice Shelf: estimates from evolving CFC, salinity and temperature fields in the Ross Sea. Deep-Sea Research I 52, 959–978. Smith, W.H., Sandwell, D.T., 1997. Global sea floor topography from satellite altimetry and ship depth soundings. Science 277, 1956–1962.

M.S. Dinniman et al. / Deep-Sea Research II 58 (2011) 1508–1523

Smith Jr., W.O., Ainley, D.G., Cattaneo-Vietti, R., 2007. Trophic interactions within the Ross Sea continental shelf ecosystem. Philosophical Transactions of the Royal Society B 362, 95–111. Smith Jr., W.O., Comiso, J.C., 2008. Influence of sea ice on primary production in the Southern Ocean: a satellite perspective. Journal of Geophysical Research 113, C05S93. doi:10.1029/2007JC004251. Tagliabue, A., Arrigo, K.R., 2005. Iron in the Ross Sea: 1. Impact on CO2 fluxes via variation in phytoplankton functional group and non-Redfield stoichiometry. Journal of Geophysical Research 110, C03009 doi: 03010.01029/02004JC002531. Talbot, M.H., 1988. Oceanic environment of George VI Ice Shelf, Antarctic Peninsula. Annals of Glaciology 11, 161–164. Thoma, M., Jenkins, A., Holland, D., Jacobs, S., 2008. Modelling Circumpolar Deep Water intrusions on the Amundsen Sea continental shelf. Geophysical Research Letters 35, L18602. doi:10.1029/2008GL034939. Walker, D.P., Brandon, M.A., Jenkins, A., Allen, J.T., Dowdeswell, J.A., Evans, J., 2007. Oceanic heat transport onto the Amundsen Sea shelf through a submarine

1523

glacial trough. Geophysical Research Letters 34, L02602. doi:10.1029/ 2006GL028154. Whitworth III, T., Nowlin Jr., W.D., 1987. Water masses and currents of the Southern Ocean at the Greenwich Meridian. Journal of Geophysical Research 41, 629–641. Whitworth III, T., Orsi, A.H., Kim, S.-J., Nowlin Jr., W.D., Locarnini, R.A., 1998. Water masses and mixing near the Antarctic Slope Front. In: Jacobs, S.S., Weiss, R.F. (Eds.), Ocean, Ice, and Atmosphere: Interactions at the Antarctic Continental Margin, AGU Antarctic Research Series, vol. 75, pp. 1–27. Whitworth III, T., Orsi, A.H., 2006. Antarctic Bottom Water production and export by tides in the Ross Sea. Geophysical Research Letters 33, L12609. doi:10.1029/ 2006GL026357. ¨ Wilkin, J., Hedstrom, K.S., 1998. User’s manual for an orthogonal curvilinear gridgeneration package. Technical Report, Institute of Marine and Coastal Sciences, Rutgers University, New Brunswick, NJ, 33pp.