ABSTRACT

The possibility that recent Antarctic sea ice expansion resulted from an increase in freshwater reaching the Southern Ocean is investigated here. The freshwater flux from ice sheet and ice shelf mass imbalance is largely missing in models that participated in phase 5 of the Coupled Model Intercomparison Project (CMIP5). However, on average, precipitation minus evaporation (PE) reaching the Southern Ocean has increased in CMIP5 models to a present value that is about greater than preindustrial times and 5–22 times larger than estimates of the mass imbalance of Antarctic ice sheets and shelves (119–544 ). Two sets of experiments were conducted from 1980 to 2013 in CESM1(CAM5), one of the CMIP5 models, artificially distributing freshwater either at the ocean surface to mimic iceberg melt or at the ice shelf fronts at depth. An anomalous reduction in vertical advection of heat into the surface mixed layer resulted in sea surface cooling at high southern latitudes and an associated increase in sea ice area. Enhancing the freshwater input by an amount within the range of estimates of the Antarctic mass imbalance did not have any significant effect on either sea ice area magnitude or trend. Freshwater enhancement of raised the total sea ice area by 1 × 106 km2, yet this and even an enhancement of was insufficient to offset the sea ice decline due to anthropogenic forcing for any period of 20 years or longer. Further, the sea ice response was found to be insensitive to the depth of freshwater injection.

1. Introduction

Sea ice is a critical component of Earth’s climate, controlling ocean–atmosphere heat exchange and driving deep ocean convection (Vaughan et al. 2013). It plays an important role in the global climate because of the sea ice–albedo feedback, which has been a major factor in the rapid decline in Arctic sea ice extent (Screen and Simmonds 2010). Earth is warming (Vaughan et al. 2013), including the upper 700 m of the Southern Ocean (Gille 2008), although sea surface temperatures are not increasing everywhere (Fan et al. 2014). Observations show Antarctic sea ice extent has expanded around 75% of the continent’s perimeter over the past three decades (Turner et al. 2009; Zunz et al. 2013). However, in contrast, phase 5 of the Coupled Model Intercomparison Project (CMIP5) models have a decline in Antarctic sea ice due to climate forcing over this period (Maksym et al. 2012; Zunz et al. 2013). Recently there has been some debate over the statistical significance of the observed increase in sea ice extent (Eisenman et al. 2014), because a change in the satellite sensor in December 1991 was not accounted for correctly in one of the main data products arising from use of NASA’s bootstrap algorithm. Nonetheless, the annual mean sea ice extent is certainly not decreasing in the Antarctic as it is in the Arctic.

Antarctic sea ice cover is strongly influenced by both winds and sea surface temperature (SST), and the coupled trio of sea ice, winds, and SST exhibit large interannual and decadal variability (e.g., Fan et al. 2014; Holland and Kwok 2012; Renwick et al. 2012). The sea ice variability is linked to distant regions through atmospheric teleconnections (e.g., Stammerjohn et al. 2008; Ding et al. 2011; Li et al. 2014). Some authors have argued that natural variability could be responsible for the recent sea ice expansion (e.g., Polvani and Smith 2013; Zunz et al. 2013). However, it is unclear if natural variability can explain the detailed pattern of sea ice trends correctly or whether any one explanation can capture the sea ice trends in all regions at once.

Identification of a missing mechanism responsible for the inconsistency between models and observations has been the subject of much recent work. Mechanisms that have been explored include wind changes (Holland and Kwok 2012; Turner et al. 2013; Holland et al. 2014), ice–ocean feedback (Goosse and Zunz 2014), and the freshwater flux from ice shelf melt (Bintanja et al. 2013, 2015; Swart and Fyfe 2013), but none have conclusively explained the discrepancy.

Here we focus on the hypothesis that freshening the Southern Ocean could explain the recent Antarctic sea ice expansion. The effect of such surface freshening has been studied in coupled ocean–sea ice models (e.g., Beckmann and Goosse 2003; Hellmer 2004), and earth system models of intermediate complexity (e.g., Aiken and England 2008; Swingedouw et al. 2008). These studies have indicated that artificially enhancing the freshwater input to the Southern Ocean is effective at increasing ocean stratification, which inhibits the vertical transport of warmer water from depth to the ocean surface, and in all cases SSTs cool, resulting in increased sea ice formation.

In more recent studies with an earth system model, Bintanja et al. (2013, 2015) added freshwater amounts that were intended to replicate current sources from Antarctic basal ice shelf melt. In the first of the two studies, Bintanja et al. (2013) achieved increases of up to 10% in sea ice concentration over a 31-yr period with the EC-Earth model under constant year 2000 forcing. Their freshwater flux of was distributed nearly uniformly around the Antarctic coast and uniformly throughout the year. Bintanja et al. (2015) then showed additional experiments required as little as to reverse the modeled sea ice area trend in a representative concentration pathway radiative forcing (RCP8.5) scenario.

Swart and Fyfe (2013) used the University of Victoria (UVic) Earth System Climate Model (ESCM) (a coupled ocean–sea ice model with an energy balance model atmosphere) to investigate the effects of surface freshwater fluxes that increased from 0 to and 0 to over periods of 47 and 29 yr, respectively. With wind forcing fixed to isolate the effects of the freshwater input, they performed each of these runs with freshwater either distributed uniformly around the Antarctic coast or concentrated in the Amundsen Bay. They found that none of their freshwater scenarios reversed the sea ice loss in the model, although all of their scenarios reduced the amount of sea ice loss relative to their control integrations from 1970 to 2020 using historical and RCP8.5 forcing. This significantly different result from that of Bintanja et al. (2013) and Bintanja et al. (2015) suggests there are differences between models that produce very different responses to similar forcing.

The studies of Bintanja et al. (2013, 2015) and Swart and Fyfe (2013) based their artificial freshwater amounts on estimates of the mass imbalance of the grounded ice of Antarctica (see Fig. 1), citing recent altimetric and gravimetric estimates from satellites by Rignot et al. (2011), Shepherd et al. (2012), and King et al. (2012).1 Such methods estimate the grounded ice loss to the ocean and therefore Antarctica’s contribution to sea level rise. Such data say nothing about the fate of the ice once it is afloat (as an ice shelf or iceberg), and therefore using only the values for the grounded ice sheet for ice shelf meltwater is an unusual assumption that neglects the additional freshwater input from the current mass imbalance of ice shelves (Shepherd et al. 2010; Rye et al. 2014; Paolo et al. 2015). Hence, the studies of Bintanja et al. (2013, 2015) and Swart and Fyfe (2013) not only disagree, but the studies of Bintanja et al. (2013, 2015) managed to cause the sea ice to expand in response to far less freshwater than equals estimates of the current mass imbalance of Antarctica’s grounded ice and ice shelves, as discussed later in this paper.

Fig. 1.

The components of the mass budget for (a) the grounded ice on Antarctica, (b) the Antarctic ice shelves, and (c) the representation of Antarctica in CESM1(CAM5). The snow water equivalent (S.W.E.) is indicated.

Fig. 1.

The components of the mass budget for (a) the grounded ice on Antarctica, (b) the Antarctic ice shelves, and (c) the representation of Antarctica in CESM1(CAM5). The snow water equivalent (S.W.E.) is indicated.

Perhaps even more surprising, we show in section 4a that the freshwater enhancements used by Bintanja et al. (2013, 2015) and Swart and Fyfe (2013) are insignificant relative to the amount of precipitation minus evaporation (PE) falling on the Southern Ocean and much less than the increase in PE over the Southern Ocean and Antarctica from preindustrial times to present day in these same models. Furthermore, in reality about half of the meltwater leaving Antarctica enters the Southern Ocean at the depth of the ice shelf front (Rignot et al. 2013; Depoorter et al. 2013). The potential mixing as the buoyant meltwater rises from the ice shelf front depth (~100–200 m) has been ignored in these studies and in many other artificial freshwater enhancement studies.

In this paper, we first discuss the differences between the model representation of the Antarctic mass budget and reality and cast the mass budget calculations in a new, consistent notation. We discuss what is known about the freshwater input to the Southern Ocean from Antarctica and the current mass imbalance of the grounded ice versus the ice shelves. We compare plausible trends in these sources to precipitation (minus evaporation) falling directly into the Southern Ocean.

We examine the influence of ice shelf processes on Antarctic sea ice extent through introduction of freshwater to the Community Earth System Model, version 1 (Community Atmosphere Model, version 5) [CESM1(CAM5)], which is a fully coupled earth system model and a member of the CMIP5 ensemble. We conduct a set of experiments that artificially enhance freshwater input to the model to investigate the effect on the local ocean and sea ice. It is important to note that this work is purely an experiment to determine the response of the climate system to an additional forcing, rather than an attempt to bring the model closer to reality. Two different sets of experiments are presented: one with the freshwater added at the ocean surface and distributed according to the meltwater input from icebergs in the GFDL model (Martin and Adcroft 2010), and the other with the freshwater added north of Antarctic ice shelves and at the depth of the ice shelf front. Finally, we discuss whether the model sensitivity to freshwater is plausible.

2. Antarctic mass budget for ice shelves and grounded ice

The mass budget of Antarctica’s grounded ice and ice shelves is governed by processes shown in Figs. 1a and 1b. Only the mass imbalance of the grounded ice can directly influence sea level rise. Because of its importance to society, it has been measured by many recent studies (see Table 1), and we consider this portion first. This is the only contribution considered by Bintanja et al. (2013, 2015) and Swart and Fyfe (2013). However, later we show that, although it is relevant to sea level rise, it is currently an insignificantly small source of freshwater to the Southern Ocean.

Table 1.

Summary of recent gravimetry- and altimetry-based estimates of Antarctic grounded ice mass imbalance (GMI), ice shelf mass imbalance (IMI) and total mass imbalance (GMI + IMI).

Summary of recent gravimetry- and altimetry-based estimates of Antarctic grounded ice mass imbalance (GMI), ice shelf mass imbalance (IMI) and total mass imbalance (GMI + IMI).
Summary of recent gravimetry- and altimetry-based estimates of Antarctic grounded ice mass imbalance (GMI), ice shelf mass imbalance (IMI) and total mass imbalance (GMI + IMI).

The mass budget for the grounded ice sheet (see Fig. 1a), including all the sources and sinks of mass, yields the following equation:

 
formula

where is the air–ice surface mass exchange rate (taking into account meltwater refreezing), is the mass flux across the grounding line, is the density of ice, is the horizontal area of the grounded ice, and is the rate of change of height of the grounded ice. At present the surface meltwater is thought to mostly refreeze within the snow cover (Liston and Winther 2005). The term is considered the mass imbalance, and it may be positive or negative depending on whether the ice sheet is gaining or losing ice, respectively. In a steady climate, the mass imbalance may be near zero if the averaging period is long enough (i.e., over several centuries) to make the contribution from natural variability negligible.

Recent estimates of the mass imbalance of the grounded ice of Antarctica range from to , where negative values indicate mass loss, and are summarized in Table 1. The very wide range of estimates, even for similar averaging periods, indicates the difficulty in obtaining these numbers. Nonetheless, Sutterley et al. (2014) note the imbalance of the grounded ice is accelerating, suggesting this imbalance will play an increasingly important role in future global climate change.

While the grounded ice mass imbalance is key to sea level rise, neither it nor any other term in Eq. (1) directly reach the Southern Ocean as freshwater, as almost all meltwater refreezes. To influence the freshwater influx, the grounded ice mass must first cross the grounding line and become part of the ice shelves.

Studies such as those of Depoorter et al. (2013) and Rignot et al. (2013) attempt to quantify each of the components that make up the mass budget for the Antarctic ice shelves. Their estimates are calculated using a combination of satellite data and modeling and provide values for basal melt rates, iceberg calving rates, surface mass balance, dynamic thinning, and flux of ice into the ice shelves at the grounding line. These studies both identify basal melting of ice shelves as the largest ice loss mechanism for the Antarctic ice shelves ( and , respectively), closely followed by iceberg calving ( and , respectively). These two loss mechanisms dominate the mass loss of the Antarctic continent. There is evidence that basal melt may have increased on some ice shelves in response to an increase in upwelling Circumpolar Deep Water (CDW) along the continental shelf, particularly near the Bellingshausen/Amundsen Sea region (e.g., Jacobs et al. 2011; Sutterley et al. 2014; Paolo et al. 2015). An increase in ice mass loss of the ice shelves is related to, but is by no means equal to, the mass imbalance of the grounded ice sheet.

The components of the mass budget for the Antarctic ice shelves are related by

 
formula

where is the grounding line flux, the air–ice surface mass exchange rate, the iceberg calving rate, the basal mass exchange rate with the ocean, and the dynamic ice thinning rate, given as the rate of change of height with time multiplied by the ice density and the horizontal area of the ice shelf . Positive (negative) values imply addition (removal) of mass to (from) the shelf, and the term is considered the mass imbalance, as in Eq. (1).

Estimates of the mass imbalance of Antarctic ice shelves are similarly varied as for the grounded ice. Using mixed methods, Shepherd et al. (2010) estimated the ice shelf imbalance at for 1994–2004, where we have multiplied their volume rate of change by the density of solid ice, . In contrast, Paolo et al. (2015) used only radar altimetry to estimate what they considered a lower bound for the ice shelf imbalance of for 2003–12 (after applying the same unit conversion factor). Importantly, Paolo et al. (2015) also found more than an order of magnitude increase in the mass imbalance between 1994–2003 and 2003–12.

If the mass imbalance of grounded ice and/or the ice shelves has increased over the last few decades or centuries, then the freshwater flux to the Southern Ocean from Antarctica would also have increased by an amount equal to the increase in the total mass imbalance. To estimate an “extra” yearly freshwater input at present relative to a hypothetical time of ice balance, we sum the central values of the largest estimates of grounded ice and shelf imbalance to arrive at . Likewise, if we sum the lowest estimates, the amount is . The true increase in freshwater flux from Antarctica over the last few decades is clearly highly uncertain, and we do not claim that it lies within these rough estimates, although the study of Rye et al. (2014) calculates the same sum to get an estimate of ~350 ± 100 Gt yr−1, which lies within our range.

None of the earth system models in the CMIP5 ensemble include ice shelf cavities at present (Flato et al. 2013), and for many the ice shelves are represented as land. The model we used in our experiments, CESM1(CAM5), has this simple representation, where the entire Antarctic continent, including ice shelves, is treated as land with a maximum allowed snow cover of 1 m. Figure 1c shows the model representation of the Antarctic continent and the components of its mass budget. When the snow thickness exceeds 1 m, it is immediately dumped at the coast as runoff (Oleson et al. 2013). In fact, the model does not capture all the processes in Eq. (2). Instead, it represents the mass budget of Antarctica as

 
formula

where is the runoff from the continent, , and is the rate of change of height of snow water equivalent with respect to time, with here denoting the density of water, and the constraint that . The grounding line flux and ice thinning rate are not represented because ice sheet dynamics are not included in the model, while the basal mass balance and calving flux are not included because of the lack of realistic ice shelves in the model. Because surface melt is rare, , so we have

 
formula

In other words, an increase in PE over Antarctica in CMIP5 models is essentially equal to an increase in freshwater flux to the Southern Ocean.

In summary, we have cast the mass budget calculations in a consistent notation, which makes comparison of values measured or calculated by different studies for different components easier to understand. The mass budget in earth system models represents a greatly simplified version of reality and means that the models are unable to capture any mass imbalance.

3. Methods

a. The model

The model used in this study is CESM1(CAM5) (Hurrell et al. 2013). The simulations were run with the Parallel Ocean Program, version 2 (POP2), ocean model; the Community Ice Code, version 4 (CICE4), sea ice model; the Community Land Model, version 4, (CLM4) land component; and the Community Atmosphere Model, version 5, (CAM5) atmosphere component. These stand-alone components were coupled by the CESM1 coupler, version 7 (CPL7), infrastructure. The model was run at approximately horizontal resolution in all components for all simulations, with 60 vertical layers in the ocean and 30 in the atmosphere.

Our experiments were run from January 1980 to December 2013, with twentieth-century transient forcing until December 2005 and using the RCP8.5 thereafter. This represents the “high emissions scenario” for greenhouse gas emissions in the models (Taylor et al. 2012). We branch our experiments in 1980 from four different ensemble members of the CESM1(CAM5) Large Ensemble (LENS) project (Kay et al. 2015). The 30 ensemble members of the LENS have the same model configuration and forcing scenarios as used in this study (without the extra freshwater forcing), where each ensemble member has the sea surface temperature (SST) in 1920 perturbed by K, where N is the number of the ensemble member (i.e., –30). This perturbation is enough for the climate state to have diverged by 1980 to produce an ensemble with which statistical comparisons can be made. We show the 30 ensemble members in Fig. 2, and the four randomly chosen ensemble members (labeled A–D) that form our sensitivity experiments in Table 2. To compare the response to freshwater scenarios independent of initial conditions, we branched each of the freshwater scenarios that we tested (described next) from LENS member A. We also investigated the sensitivity to the initial conditions by varying the LENS member from which we branched for select freshwater scenarios.

Fig. 2.

Seasonal total sea ice area for the 30 members of the CESM1(CAM5) LENS. Our experiments were branched from the colored trajectories. We shall refer to the member highlighted in blue as run A, red as run B, pink as run C, and green as run D. The IFW experiments were branched from A and B, while the SFW experiments were branched from A, C, and D. A, B, C, and D correspond to members 25, 20, 26, and 27 of the ensemble, respectively.

Fig. 2.

Seasonal total sea ice area for the 30 members of the CESM1(CAM5) LENS. Our experiments were branched from the colored trajectories. We shall refer to the member highlighted in blue as run A, red as run B, pink as run C, and green as run D. The IFW experiments were branched from A and B, while the SFW experiments were branched from A, C, and D. A, B, C, and D correspond to members 25, 20, 26, and 27 of the ensemble, respectively.

Table 2.

The experiments discussed in this paper. The A–D suffix on the experiment name indicates multiple ensemble members, and the CESM1(CAM5) LENS member counterpart, as shown in Fig. 2. Mask source gives the data source used to construct the distribution of freshwater input.

The experiments discussed in this paper. The A–D suffix on the experiment name indicates multiple ensemble members, and the CESM1(CAM5) LENS member counterpart, as shown in Fig. 2. Mask source gives the data source used to construct the distribution of freshwater input.
The experiments discussed in this paper. The A–D suffix on the experiment name indicates multiple ensemble members, and the CESM1(CAM5) LENS member counterpart, as shown in Fig. 2. Mask source gives the data source used to construct the distribution of freshwater input.

b. Surface freshwater experiments

To simulate freshwater input from either ice shelf basal melt or iceberg melt in excess of the normal way that CESM1(CAM5) deals with the mass balance of Antarctica [described in Fig. 1c and Eq. (4)], we enhanced the freshwater entering the Southern Ocean. In the first set of experiments, we added surface freshwater (SFW) to investigate the response as if all the freshwater missing in our model (and other CMIP5 models) were from an increase in the iceberg flux. Since the ocean in CESM1(CAM5) conserves volume, and direct addition of freshwater is not possible, we parameterize the freshwater input as a negative salinity forcing by multiplying the freshwater flux by minus the reference salinity of the ocean, psu. After discovering our model had a very weak response to freshwater flux estimates of the current Antarctic mass imbalance, we chose to introduce larger amounts of freshwater enhancement; specifically, we input 1000, 2000, or 3000 of additional freshwater in an attempt to determine how much freshwater is required to have a significant effect on the sea ice area trend. We acknowledge that these freshwater inputs are much larger than estimates of the combined ice shelf/ice sheet mass imbalance. Three experiments were conducted with to test for reproducibility (see Table 2). To distribute the freshwater realistically around the Antarctic coast, we used the 100-yr monthly mean global meltwater distributions from icebergs in the GFDL-ESM runs (Martin and Adcroft 2010), regridded onto the CESM grid (see Fig. 3a). The freshwater flux was introduced at an annually periodic rate using the GFDL iceberg distribution, because of the lack of current knowledge of the seasonality of freshwater flux from iceberg calving. Although several papers have shown that the latent heat associated with melting icebergs has a significant impact on the hydrography and sea ice in the Southern Ocean (e.g., Jongma et al. 2009), we have not taken it into account because our purpose is to isolate the effects of freshwater alone to compare more directly with the studies of Bintanja et al. (2013, 2015) and Swart and Fyfe (2013).

Fig. 3.

Maps showing the (top) SFW and (bottom) IFW distributions. The surface freshwater distribution based on iceberg drift is on a logarithmic scale in order to resolve input far from the coast. The interior freshwater distribution based on ice shelf location shows the location of the grid cells in which the negative salinity forcing was input; the color scale indicates the depth at which the forcing was input.

Fig. 3.

Maps showing the (top) SFW and (bottom) IFW distributions. The surface freshwater distribution based on iceberg drift is on a logarithmic scale in order to resolve input far from the coast. The interior freshwater distribution based on ice shelf location shows the location of the grid cells in which the negative salinity forcing was input; the color scale indicates the depth at which the forcing was input.

c. Interior freshwater experiments

In a second set of experiments, freshwater was added at the ice shelf fronts to investigate the response as if all the freshwater missing in our model (and other CMIP5 models) were from an increase in the basal melt of ice shelves. This applies a constant reduction in salinity to the specified vertical level. We injected the freshwater in front of ice shelves and at the depth of the front (see Fig. 3b). The ice shelf location and depth were derived from the RTopo-1 dataset (Timmermann et al. 2010). These were then regridded onto the CESM1(CAM5) grid and checked for mismatches between the RTopo-1 and CESM bathymetry, which arose because of the large resolution difference between the dataset and the model [the RTopo-1 dataset is much higher resolution than the CESM1(CAM5) grid]. In some cases, the interpolated depth of the ice shelf front from RTopo-1 was deeper than the CESM1(CAM5) bathymetry. In these cases, the problem was resolved by manually raising the vertical layer in the model into which the freshwater was input to the lowest level within the ocean. Other issues arose when islands were present in the middle of an ice shelf, causing false identification of the ice shelf front. These cells were manually inspected and removed. The freshwater flux was then divided evenly among the grid cells, and the forcing was input uniformly throughout the year.

Three interior freshwater (IFW) experiments were conducted, denoted as IFW167A, IFW2000A, and IFW2000B (see Table 2). The IFW167A experiment simulated a freshwater input within our calculated range of estimates of present total ice mass imbalance for Antarctica. The IFW2000A and IFW2000B experiments were conducted after preliminary results from the surface freshwater experiments suggested this magnitude of freshwater input () was necessary in order to see a significant change in the annual mean sea ice area over the duration of the experiments.

Figure 4 shows a comparison of the depth of freshwater input with the modeled seasonal mixed layer depth from the CESM1(CAM5) LENS mean. We see that, for the shallower mixed layer depths of summer and autumn, the depth of interior freshwater input is predominantly below the mixed layer, while in winter and spring about half the input cells lie within the mixed layer. This is important since freshwater that is input directly into the mixed layer will be immediately mixed with the ambient water, while that input below the mixed layer will take longer to be mixed.

Fig. 4.

Difference between the modeled mixed layer depth and the depth of freshwater input for the interior freshwater experiments for each season. A positive value indicates the depth of input is within the mixed layer.

Fig. 4.

Difference between the modeled mixed layer depth and the depth of freshwater input for the interior freshwater experiments for each season. A positive value indicates the depth of input is within the mixed layer.

In summary, we have two sets of experiments to test the effect of freshwater input either due to iceberg calving (surface experiments) or basal melt (interior experiments). It should be noted that, in both sets of experiments, we are only considering the freshening effect of the meltwater and that we do not apply any explicit cooling to the model.

4. Results

a. CMIP5 freshwater budget

To put the amount of artificial freshwater enhancement used in our experiments and those of others in context, we first examined the sources of freshwater to the Southern Ocean from PE falling on the Antarctic continent and the Southern Ocean in the CMIP5 ensemble (Taylor et al. 2012) (http://cmip-pcmdi.llnl.gov/cmip5/) and in the Modern-Era Retrospective Analysis for Research and Applications (MERRA) and ECMWF reanalyses (ERA). Recall that on the continent PE is approximately equal to the amount of meltwater from the continent, which is the sole source of freshwater in the models (see Fig. 1c). In the Southern Ocean, PE either adds freshwater directly to the ocean surface or it accumulates on sea ice and subsequently melts some time later.

To calculate PE on Antarctica for CMIP5 models, we summed over grid cells using the land masks from individual models. For the Southern Ocean, PE was summed over all grid cells south of 50°S; then the values for the continent were subtracted to leave the total for the ocean.

Over the Southern Ocean, PE on average from 1994 to 2013 was 21 000 Gt yr−1 from MERRA and 27 700 Gt yr−1 from ERA-Interim. The CMIP5 models have an across-model ensemble mean of 23 108 Gt yr−1 and a standard deviation of (Fig. 5a). Over Antarctica, PE over the same period is an order of magnitude smaller; it is 2480 Gt yr−1 from MERRA and 2580 Gt yr−1 from ERA-Interim. The across-model mean for CMIP5 models for that period was with a standard deviation of (Fig. 5b). If Antarctica’s ice sheets and shelves were in mass balance, the meltwater from Antarctica (mainly from basal melt and iceberg calving) would equal PE falling over Antarctica averaged over a few decades. Thus, the mean combined freshwater input to the Southern Ocean from PE and Antarctic meltwater in CMIP5 models is about 25 700 Gt yr−1. Further, the change in PE since preindustrial times over the Southern Ocean and Antarctic continent combined in CMIP5 models, taken as the difference between the average over 1994–2013 and the average over 1861–1890, is , with a standard deviation of (Fig. 5c). The contribution to the increase in PE from over Antarctica alone in CMIP5 models is , which lies above the wide-ranging estimates of the total present mass imbalance of Antarctic ice from observations (roughly 119–544 ).

Fig. 5.

The net precipitation (precipitation − evaporation) for a selection of the models used in the CMIP5 ensemble for (a) the Southern Ocean averaged over 1994–2013, (b) the Antarctic continent averaged over 1994–2013, and (c) the difference between 1994–2013 and 1861–90 over the Southern Ocean and Antarctic continent combined. The 1994–2013 averages also include the ERA and MERRA estimates of PE over the Southern Ocean and Antarctic, respectively (red). The CMIP5 model used in this study [CESM1(CAM5)] is highlighted in cyan.

Fig. 5.

The net precipitation (precipitation − evaporation) for a selection of the models used in the CMIP5 ensemble for (a) the Southern Ocean averaged over 1994–2013, (b) the Antarctic continent averaged over 1994–2013, and (c) the difference between 1994–2013 and 1861–90 over the Southern Ocean and Antarctic continent combined. The 1994–2013 averages also include the ERA and MERRA estimates of PE over the Southern Ocean and Antarctic, respectively (red). The CMIP5 model used in this study [CESM1(CAM5)] is highlighted in cyan.

In summary, the largest source of freshwater to the Southern Ocean is the PE falling directly onto the ocean. The PE, and hence runoff [see Eq. (4)], from the Antarctic continent is an order of magnitude smaller, and is coincidentally of similar magnitude to the increase in PE falling on the Southern Ocean since preindustrial times and the largest of our artificial freshwater enhancement experiments ().

b. Ocean response difference between interior and surface freshening

The freshwater input scenario in our experiments is quite different for the surface and interior cases (see Fig. 3). When freshwater is injected in the interior, it enters exclusively at the Antarctic coast, while at the surface it is introduced over a much wider area. In this section, we present the ocean response to freshwater enhancement and describe the extent to which the point of origin of the freshening influences the results. We compare only the response of the ensemble means of surface freshwater experiments with freshwater enhancement and the interior freshwater experiments. With regard to the other experiments, the ocean response appears to be roughly linear in the magnitude of freshwater input, though the response to adding just was not significant.

Examining all experiments with freshwater enhancement, it takes only a few years after we begin to artificially add freshwater in 1980 before the upper-ocean salinity decreases substantially south of about 40°S (Fig. 6) (here and henceforth we compare our experiments to the 30-member ensemble mean of the LENS at an equivalent point in time). The response within the mixed layer, which is from the surface to ~100-m depth, between ~40° and 75°S shows little evidence of the point of origin of the freshening. The stabilizing effect of the desalination extends year-round to the northernmost reach of the sea ice cover.

Fig. 6.

The response of the zonal mean salinity in the ensemble mean of (top) SFW2000A, C, and D and (bottom) IFW2000A and B for 1994–2013 (all months) compared to the LENS. Blue (red) color denotes a decrease (increase) in zonal mean temperature relative to the LENS.

Fig. 6.

The response of the zonal mean salinity in the ensemble mean of (top) SFW2000A, C, and D and (bottom) IFW2000A and B for 1994–2013 (all months) compared to the LENS. Blue (red) color denotes a decrease (increase) in zonal mean temperature relative to the LENS.

The increased stratification of the water column inhibits sinking near the coast of Antarctica and upwelling farther north in the Southern Ocean. The resulting weaker meridional overturning circulation reduces the exchange of heat between the intermediate-depth ocean and the surface mixed layer. The temperature response shows upper-ocean cooling over a large domain, except for patches of warming below about 100 m south of ~60°S (Fig. 7). The maximum cooling in the zonal mean is over 0.5°C at the surface just beyond the winter sea ice extent. The sea surface temperature response is nearly the same for the surface and interior freshwater experiments. The temperature response at depth differs more between freshwater forcing scenarios. The coastal subsurface warming results from a reduction in sinking of cold continental shelf waters, while the subsurface warming at ~70°S results from a reduction in upwelling in the vicinity of a temperature inversion (i.e., the ocean is warmer below the mixed layer at ~70°S).

Fig. 7.

The response of the zonal mean temperature in the ensemble mean of (top) SFW2000A, C, and D and (bottom) IFW2000A and B for 1994–2013 (all months) compared to the LENS. Blue (red) color denotes a decrease (increase) in zonal mean temperature relative to the LENS.

Fig. 7.

The response of the zonal mean temperature in the ensemble mean of (top) SFW2000A, C, and D and (bottom) IFW2000A and B for 1994–2013 (all months) compared to the LENS. Blue (red) color denotes a decrease (increase) in zonal mean temperature relative to the LENS.

Because the interior freshwater experiments concentrate the freshwater flux near the coast of Antarctica, the coastal subsurface warming is greater, while the maximum warming in the surface freshwater experiments is at ~70°S. The apparent greater magnitude of warming in the interior freshwater experiments may be because signals are concentrated on smaller-latitude circles at high southern latitudes.

We diagnose the cooling rate by this mechanism in an analysis similar to that used by Ferreira et al. (2015). A key component of the temperature tendency is from advection by the residual mean vertical upwelling rate and the sum of the Eulerian and parameterized eddy-induced vertical velocities acting on the mean vertical temperature gradient . The reduction in upwelling results in an advective tendency response from the residual mean upwelling anomaly acting on the temperature gradient from the mean state:

 
formula

where indicates an anomaly. The expression in Eq. (5) does not include a vertical velocity gradient term, which Ferreira et al. (2015) have demonstrated is of second-order importance.

Figure 8 shows the advective temperature tendency response [using Eq. (5) with from the ensemble mean of the LENS] in our experiments. There is predominantly a cooling tendency in the Southern Ocean at 100–200-m depth, which is evidence of the reduction in upwelling of warmer water from below. There is no clear dependence in the response of the advective temperature tendency on whether the freshwater is injected in the interior or added at the surface (see Fig. 8).

Fig. 8.

The response in vertical advection for the (top) SFW2000A, C, and D and (bottom) IFW2000A and B experiments calculated from Eq. (5). Positive (negative) values denote an increase (decrease) in upward advection.

Fig. 8.

The response in vertical advection for the (top) SFW2000A, C, and D and (bottom) IFW2000A and B experiments calculated from Eq. (5). Positive (negative) values denote an increase (decrease) in upward advection.

It is interesting to note that the most negative temperature tendencies by advection in Fig. 8 appear far to the south of greatest cooling in Fig. 7. We attribute this disparity to the fact that we have only examined the response to vertical advection. Heat transport occurs mainly along isopycnals, which are more horizontal in midlatitudes. An anomalous northward ocean heat transport was found in support of this explanation (not shown).

An interesting result of our freshwater enhancement experiments is that freshwater added at the surface tends to reduce the mixed layer depth relative to the LENS mean at most times of the year, while the interior freshwater enhancement caused the mixed layer to become deeper as shown in Fig. 9. As seen in Fig. 4, when injected in the interior, most of the freshwater enters at the base of the mixed layer. Since the density of the water is dominated by the salinity, the freshwater is buoyant, which drives convective overturning and deepens the mixed layer.

Fig. 9.

Seasonal mean mixed layer depth, averaged over all longitudes and all latitudes south of 60°S.

Fig. 9.

Seasonal mean mixed layer depth, averaged over all longitudes and all latitudes south of 60°S.

In summary, injecting freshwater at depth does drive greater mixing, which significantly deepens the mixed layer (see Fig. 9) and leads to a greater reduction in salinity at 100–200-m depth at the Antarctic coast. Nonetheless, to a large extent, the upper-ocean salinity and temperature response is independent of the two methods we employed for adding freshwater, especially in ways that are likely to be important to the sea ice cover.

c. Sea ice response to artificial freshwater enhancement

Given the weak sensitivity of the surface ocean to the depth of freshwater injection, it is not surprising that the trend in sea ice area is also insensitive to the method by which we added freshwater. However, the response in the 1994–2013 annual mean of the total sea ice area does depend on the amount of freshwater input. After a 5–10-yr adjustment period from the start of freshwater enhancement, only the total area in those cases enhanced by or more (SFW2000A, C, and D; IFW2000A and B; and SFW3000A) lie outside the spread of the LENS members (see Fig. 10). In each of these cases, the total area is significantly larger than the distribution of the LENS in the last 20 years of the experiments (1994–2013) in every season. From 1994 to 2013, the total sea ice area in the ensemble mean of the freshwater enhancement cases compared to the LENS mean is significantly larger in winter and spring (by about a factor of 2) than in summer and autumn. The response in the magnitude of sea ice area for the IFW167A experiment stays well within the range of the LENS.

Fig. 10.

The 5-yr running mean total sea ice area for each of the experiments, as well as for the 30 individual LENS members and their mean. Note the different vertical axis scale between summer/autumn and winter/spring.

Fig. 10.

The 5-yr running mean total sea ice area for each of the experiments, as well as for the 30 individual LENS members and their mean. Note the different vertical axis scale between summer/autumn and winter/spring.

Figure 11 shows the slope of a linear fit to the time series of seasonal mean sea ice area for each of our artificial freshwater enhancement experiments. These are plotted on a histogram of the slopes of a linear fit to each of the members of the LENS for the period 1994–2013. We see that the trends for all of the experiments fall well within the range of the ensemble trends. This suggests that the introduction of large artificial freshwater enhancement causes no significant change in the trend in seasonal mean sea ice area.

Fig. 11.

Histogram of the slopes of a linear fit to the seasonal time series of total sea ice area for each of the CESM1(CAM5) LENS members overlaid with the IFW2000A and B (cyan); IFW167A (dotted black); SFW1000A (blue); SFW2000A, C, and D (pink); and SFW3000A (green) slopes. All were calculated as linear fits to the model output over the period 1994–2013 to avoid the transient initial response in the freshwater enhancement experiments.

Fig. 11.

Histogram of the slopes of a linear fit to the seasonal time series of total sea ice area for each of the CESM1(CAM5) LENS members overlaid with the IFW2000A and B (cyan); IFW167A (dotted black); SFW1000A (blue); SFW2000A, C, and D (pink); and SFW3000A (green) slopes. All were calculated as linear fits to the model output over the period 1994–2013 to avoid the transient initial response in the freshwater enhancement experiments.

In the trend analysis just described, we eliminate the first six years of our experiments, because during this time the sea ice in some of our experiments undergoes a rapid expansion before levelling off. We repeated our analysis for a range of different start and end dates with a period length of at least 20 years and found the results were unchanged.

Figure 12 shows spatial maps of the sea ice trend in the freshwater enhancement experiments branched from LENS run A (see Fig. 2) and for the ensemble mean of enhancement experiments compared to the LENS mean in individual seasons. In agreement with the trends in total area response in Fig. 11, there is no consistent spatial pattern in the trend response among the individual experiments in Fig. 12. Many anomalies persist over the seasonal cycle, which is expected because sea ice concentration anomalies exhibit persistence and reemergence for up to about a year (e.g., Holland et al. 2013). When averaged over a number of runs, these anomalies are removed.

Fig. 12.

(top) The LENS mean sea ice concentration and extent for 1994–2013, with the response to the freshwater forcing in sea ice concentration for the (upper middle) SFW2000A, (middle) SFW3000A, and (lower middle) IFW2000A experiments. We also show (bottom) the mean response of all experiments with . The response is the slope of a linear fit to the difference between the experiment and the LENS mean over the period 1994–2013 of the experiments. Note the differing color scales for (top) and (upper middle)–(bottom). In the lower four rows, blue (red) colors denote an increase (decrease) relative to the LENS mean.

Fig. 12.

(top) The LENS mean sea ice concentration and extent for 1994–2013, with the response to the freshwater forcing in sea ice concentration for the (upper middle) SFW2000A, (middle) SFW3000A, and (lower middle) IFW2000A experiments. We also show (bottom) the mean response of all experiments with . The response is the slope of a linear fit to the difference between the experiment and the LENS mean over the period 1994–2013 of the experiments. Note the differing color scales for (top) and (upper middle)–(bottom). In the lower four rows, blue (red) colors denote an increase (decrease) relative to the LENS mean.

5. Discussion

The freshwater inputs over the Southern Ocean in the CMIP5 ensemble (see Fig. 5), which include PE that falls directly into the ocean and that falls on Antarctica and generally becomes meltwater input to the Southern Ocean according to Eq. (4), give a useful benchmark with which to compare our freshwater forcings and those used in previous studies. It is also reassuring that these estimates agree well with the values obtained from the reanalyses. At most we are adding around 10% on top of the net amount of freshwater already received by the Southern Ocean from PE. Our most aggressive freshwater forcings are of a similar magnitude (i.e., ~100%) to the amount of additional PE entering the Southern Ocean at present compared to preindustrial values in CMIP5 models. In contrast, the previous studies of Bintanja et al. (2013, 2015) and Swart and Fyfe (2013) add at most 1%, 0.5% and 3% to the freshwater from PE that is received by the Southern Ocean. Importantly, in the latter two studies, the freshwater inputs were added to models forced with twentieth- and twenty-first-century scenarios, which therefore already have substantial increases in PE compared to the preindustrial period. Relative to this increase in PE, the enhancement was at most about 5% in Bintanja et al. (2015) and 30% in Swart and Fyfe (2013).

In response to artificially adding freshwater in our model, the upper 100–200 m freshens, and the upper 100 m cools south of about 65°S. The same near-surface response is described by other recent studies (Bintanja et al. 2013; Swart and Fyfe 2013, their supplement). The peak surface cooling in the zonal mean in Bintanja et al. (2013) is within the winter sea ice covered region about 65°S, while in our study it is shifted north by about 5° of latitude, which is nearly always beyond the sea ice cover. Even larger differences in the pattern of subsurface temperature prevail among recent studies (including ours). In Bintanja et al. (2013), the peak warming in the zonal mean occurs at about 42°S at ~300-m depth. In the other two studies (including ours), the peak warming in the zonal mean occurs south of 65°S at a similar depth.

It has been shown (Fig. 9) that the mixed layer response depends upon whether freshwater is added at the surface or interior, while sea ice and vertical advection do not. We suggest the dominant mechanism that limits the sea ice is the increased stratification of the ocean, where the density difference between the surface mixed layer and the ocean immediately below it is increased, inhibiting vertical transport of heat to the surface. Using salinity as a proxy for density, we see very little difference in this response between the two experiments (Fig. 6). We conclude that the behavior of the mixed layer depth, while interesting, does not determine the response in sea ice area.

Even though our model freshens and cools in the upper Southern Ocean, we see clearly in Fig. 11 that the trends in all of our artificial freshwater enhancement experiments fall well within the range of the trends of the LENS members (which had no artificial freshwater forcing). This suggests that even a very large artificial freshwater enhancement introduced at a constant annual mean rate is not sufficient to reverse the model’s trend in sea ice area over the last 34 years.

Although our artificial freshwater enhancement does not cause the sea ice to expand over time, our integrations do have a substantial ocean and 1994–2013 annual mean total sea ice area magnitude response. The sea ice total area is about 1 million square kilometers greater than in the LENS, and the sea surface temperature is cooler by as much as 0.5°C in the zonal mean. Interestingly, the IFW167A experiment, with a freshwater input that lies within our calculated range of estimates of total Antarctic ice mass imbalance has no significant effect on the sea ice area trend or magnitude.

The sea ice response in our experiments is more consistent with Swart and Fyfe (2013). When we added an amount of freshwater of a similar magnitude to theirs, we found no significant response in the sea ice total area in any season. We had to more than double the amount of freshwater used by Swart and Fyfe (2013) before the sea ice total area response was significant. Our results also agree with those of Zunz and Goosse (2015), who concluded that while freshwater input from melting plays some role in determining sea ice area, it appears not to be the dominant mechanism.

In contrast, Bintanja et al. (2013, 2015) had a significant response from an order of magnitude less freshwater forcing than was used in the artificial freshwater enhancement experiments in our model. In our experiments that have a significant sea ice response, the salinity response in the mixed layer also appears to be about 5–10 times greater than that of Bintanja et al. (2013). If we assume that no error was made in the estimate of freshwater inputs by Bintanja et al. or us, then it is difficult to understand why such a small freshwater enhancement had such a dramatic effect in the simulations presented in Bintanja et al. (2013, 2015). Our evaluation of the freshwater inputs into the Southern Ocean in Fig. 5 gives no indication why the results should differ so dramatically since our model [CESM1(CAM5)] and the model used by Bintanja et al. (EC-EARTH) are similar and both are in line with other CMIP5 models. We can only assume that the water column in the EC-EARTH model is weakly stratified so that the addition of a very small surface freshwater forcing is enough to cause significant surface cooling and thus reverse the trend in sea ice area.

6. Conclusions

We have investigated the hypothesis that recent freshening of the Southern Ocean might be the cause of recent Antarctic sea ice expansion. This mechanism has received attention in part because it involves meltwater from ice shelves and icebergs, which are not treated in GCMs, and therefore could be the missing mechanisms responsible for discrepancy in sea ice behavior in CMIP5 models and observations.

We began with an analysis of sources of freshwater that are included in CMIP5 from PE over the Southern Ocean and Antarctica. Given the simplifications to the surface mass balance of Antarctica in CMIP5 models, PE falling on Antarctica is roughly equal to the source of freshwater from Antarctica that reaches the Southern Ocean in CMIP5 models. We found PE directly falling on the Southern Ocean is about an order of magnitude higher than the PE that falls on Antarctica. Further, the increase (at present day relative to preindustrial) in this freshwater source to the Southern Ocean in CMIP5 models is on average. Thus, the increase in freshwater that has been accounted for in CMIP5 models is roughly 5–22 times larger than the sum of current estimates of the missing sources in CMIP5 models from the mass imbalance of the grounded ice sheet ( to ) and the ice shelves ( to ).

There are disagreements in the sensitivity of models to the missing freshwater sources from Antarctica among recent studies that have introduced artificial freshwater enhancements to the Southern Ocean. We not only explored the sensitivity in another model, but we asked how much freshening is needed to produce a significant response. We introduced freshwater enhancements to the Southern Ocean in the CESM1(CAM5) that ranged from 167 to , which at the high end is much larger than observational estimates suggest is reasonable. Freshwater input within the range of estimates of combined Antarctic ice sheet/ice shelf imbalance caused no significant effect on either the annual mean sea ice area magnitude or trend. In response to larger freshwater enhancement (), after an initial rapid adjustment, the sea ice area remained elevated by, at most, about 1 million square kilometers compared to integrations without freshwater enhancement. Despite the large freshwater input, the forcing we introduced was not sufficient to alter the trend in our model’s annual mean sea ice area after the initial rapid adjustment. Our weak response in sea ice area to this large forcing suggests a constant annual mean freshwater input is not wholly responsible for the observed increase in sea ice area over recent decades.

In addition to investigating the amount of freshwater needed to produce a significant sea ice response, we also explored whether the response depended on whether the freshwater was distributed as if all the meltwater was from iceberg melt or all from ice shelf basal melt. We anticipated that adding freshwater at depth might drive mixing that would compete with the ability of the freshwater to stratify the upper ocean. We found that injecting water at the depth of the front of ice shelf around Antarctica caused the ocean mixed layer to deepen, while adding freshwater at the surface caused the mixed layer to shoal. However, the overall response of the ocean and sea ice is not very sensitive to the difference, indicating that the likely mechanism by which Antarctica loses mass now and in the future will not affect the sea ice response.

A limitation of our experiments at depth is that we introduce freshwater at a constant rate in time over the length of the experiments, which is almost certainly not the case in nature. At present, little is known about the seasonality of meltwater from ice shelf melt, and the sensitivity of response to the time of freshwater input could be a useful area of future work.

The inconsistent response to artificial freshwater enhancement among different modeling studies suggests important mechanisms in the interaction between the ocean and sea ice are being misrepresented in models. An investigation into these interactions in models is needed to account for this discrepancy in response. A comparison of CMIP5 model responses to freshwater enhancement has been suggested by Bintanja et al. (2015) and seems a crucial step in identifying the source of discrepancy between models and observations and between models themselves.

Acknowledgments

We thank three anonymous reviewers whose detailed and thoughtful comments helped greatly in preparing the manuscript. We thank Dr. Torge Martin for providing the iceberg melt distributions from his earlier study and for helpful discussions. CMB gratefully acknowledges receiving a Fulbright U.S. Senior Scholar award along with the support of Fulbright New Zealand and funding from the National Science Foundation through Grant PLR-1341497. PJL and IJS were supported through University of Otago Research Grant 111030 and two subcontracts to NIWA: one for the Ministry of Business, Innovation, and Employment funded project “Climatic variability in the Ross Sea region of Antarctica and its potential influences on the marine ecosystem” (CO1X1226), and one from Crown Research Institute core funding. The authors wish to acknowledge the contribution of NeSI high-performance computing facilities to the results of this research. NZ’s national facilities are provided by the NZ eScience Infrastructure and funded jointly by NeSI’s collaborator institutions and through the Ministry of Business, Innovation and Employment’s Research Infrastructure programme. We thank Peter Maxwell in particular for his assistance in running CESM1(CAM5) on the cluster. We would like to acknowledge high-performance computing support from Yellowstone (ark:/85065/d7wd3xhc) provided by NCAR’s Computational and Information Systems Laboratory, sponsored by the National Science Foundation. AGP’s initial involvement in the project was supported by a University of Otago Summer Scholarship from the Polar Environments Research Theme, and his continued involvement through a Kelly Tarlton’s Antarctic Scholarship (awarded through the Antarctica New Zealand Postgraduate Research Scholarships Programme). We acknowledge the CESM1(CAM5) Large Ensemble Community Project and the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP. We thank the climate modelling groups (shown in Fig. 5 of this paper) for producing and making available their model output. For CMIP, the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals.

REFERENCES

REFERENCES
Aiken
,
C. M.
, and
M. H.
England
,
2008
:
Sensitivity of the present-day climate to freshwater forcing associated with Antarctic sea ice loss
.
J. Climate
,
21
,
3936
3946
, doi:.
Barletta
,
V. R.
,
L. S.
Sorensen
, and
R.
Forsberg
,
2013
:
Scatter of mass changes estimates at basin scale for Greenland and Antarctica
.
Cryosphere
,
7
,
1411
1432
, doi:.
Beckmann
,
A.
, and
H.
Goosse
,
2003
:
A parametrization of ice shelf–ocean interaction for climate models
.
Ocean Modell.
,
5
,
157
170
, doi:.
Bintanja
,
R.
,
G. J.
van Oldenborgh
,
S. S.
Drijfhout
,
B.
Wouters
, and
C. A.
Katsman
,
2013
:
Important role for ocean warming and increased ice-shelf melt in Antarctic sea ice expansion
.
Nat. Geosci.
,
6
,
376
379
, doi:.
Bintanja
,
R.
,
G. J.
van Oldenborgh
, and
C. A.
Katsman
,
2015
:
The effect of increased fresh water from Antarctic ice shelves on future trends in Antarctic sea ice
.
Ann. Glaciol.
,
56
,
120
126
, doi:.
Depoorter
,
M. A.
,
J. L.
Bamber
,
J. A.
Griggs
,
J. T. M.
Lenaerts
,
S. R. M.
Ligtenburg
,
M. R.
van den Broeke
, and
G.
Moholdt
,
2013
:
Calving fluxes and basal melt rates of Antarctic ice shelves
.
Nature
,
502
,
89
92
, doi:.
Ding
,
Q. H.
,
E. J.
Steig
,
D. S.
Battisti
, and
M.
Kuttel
,
2011
:
Winter warming in West Antarctica caused by central tropical Pacific warming
.
Nat. Geosci.
,
4
,
398
403
, doi:.
Eisenman
,
I.
,
W. N.
Meier
, and
J. R.
Norris
,
2014
:
A spurious jump in the satellite record: Has Antarctic sea ice expansion been overestimated?
Cryosphere
,
8
,
1289
1296
, doi:.
Fan
,
T.
,
C.
Deser
, and
D. P.
Schneider
,
2014
:
Recent Antarctic sea ice trends in the context of Southern Ocean surface climate variations since 1950
.
Geophys. Res. Lett.
,
41
,
2419
2426
, doi:.
Ferreira
,
D.
,
J.
Marshall
,
C. M.
Bitz
,
S.
Solomon
, and
A.
Plumb
,
2015
:
Antarctic ocean and sea ice response to ozone depletion: A two-time-scale problem
.
J. Climate
,
28
,
1206
1226
, doi:.
Flato
,
G.
, and Coauthors
,
2013
: Evaluation of climate models. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 126 pp. [Available online at https://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter09_FINAL.pdf.]
Gille
,
S. T.
,
2008
:
Decadal-scale temperature trends in the Southern Hemisphere ocean
.
J. Climate
,
21
,
4749
4765
, doi:.
Goosse
,
H.
, and
V.
Zunz
,
2014
:
Decadal trends in the Antarctic sea ice extent ultimately controlled by ice–ocean feedback
.
Cryosphere
,
8
,
453
470
, doi:.
Hellmer
,
H. H.
,
2004
:
Impact of Antarctic ice shelf basal melting on sea ice and deep ocean properties
.
Geophys. Res. Lett.
,
31
,
L10307
, doi:.
Holland
,
M. M.
,
E.
Blanchard-Wrigglesworth
,
J.
Kay
, and
S.
Vavrus
,
2013
:
Initial-value predictability of Antarctic sea ice in the Community Climate System Model 3
.
Geophys. Res. Lett.
,
40
,
2121
2124
, doi:.
Holland
,
P. R.
, and
R.
Kwok
,
2012
:
Wind-driven trends in Antarctic sea ice drift
.
Nat. Geosci.
,
5
,
872
875
, doi:.
Holland
,
P. R.
,
N.
Bruneau
,
C.
Enright
,
M.
Losch
,
N. T.
Kurtz
, and
R.
Kwok
,
2014
:
Modeled trends in Antarctic sea ice thickness
.
J. Climate
,
27
,
3784
3801
, doi:.
Hurrell
,
J. W.
, and Coauthors
,
2013
:
The Community Earth System Model: A framework for collaborative research
.
Bull. Amer. Meteor. Soc.
,
94
,
1339
1360
, doi:.
Jacobs
,
S. S.
,
A.
Jenkins
,
C. F.
Giulivi
, and
P.
Dutrieux
,
2011
:
Stronger ocean circulation and increased melting under Pine Island Glacier ice shelf
.
Nat. Geosci.
,
4
,
519
523
, doi:.
Jongma
,
J. I.
,
E.
Driesschaert
,
T.
Fichefet
,
H.
Goosse
, and
H.
Renssen
,
2009
:
The effect of dynamic–thermodynamic icebergs on the Southern Ocean in a three-dimensional model
.
Ocean Modell.
,
26
,
104
113
, doi:.
Kay
,
J. E.
, and Coauthors
,
2015
:
The Community Earth System Model (CESM) Large Ensemble Project: A community resource for studying climate change in the presence of internal climate variability
.
Bull. Amer. Met. Soc.
,
96
,
1333
1349
, doi:.
King
,
M. A.
,
R. J.
Bingham
,
P.
Moore
,
P. L.
Whitehouse
,
M. J.
Bentley
, and
G. A.
Milne
,
2012
:
Lower satellite-gravimetry estimates of Antarctic sea-level contribution
.
Nature
,
491
,
586
589
, doi:.
Li
,
X. C.
,
D. M.
Holland
,
E. P.
Gerber
, and
C.
Yoo
,
2014
:
Impacts of the north and tropical Atlantic Ocean on the Antarctic Peninsula and sea ice
.
Nature
,
505
,
538
542
, doi:.
Liston
,
G. E.
, and
J.-G.
Winther
,
2005
:
Antarctic surface and subsurface snow and ice melt fluxes
.
J. Climate
,
18
,
1469
1481
, doi:.
Maksym
,
T.
,
S. E.
Stammerjohn
,
S.
Ackley
, and
R.
Massom
,
2012
:
Antarctic sea ice—A polar opposite?
Oceanography
,
25
,
140
151
, doi:.
Martin
,
T.
, and
A.
Adcroft
,
2010
:
Parameterizing the fresh-water flux from land ice to ocean with interactive icebergs in a coupled climate model
.
Ocean Modell.
,
34
,
111
124
, doi:.
McMillan
,
M.
,
A.
Shepherd
,
A.
Sundal
,
K.
Briggs
,
A.
Muir
,
A.
Ridout
,
A.
Hogg
, and
D.
Wingham
,
2014
:
Increased ice losses from Antarctica detected by CryoSat-2
.
Geophys. Res. Lett.
,
41
,
3899
3905
, doi:.
Oleson
,
K. W.
, and Coauthors
,
2013
: Technical description of version 4.5 of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-503+STR, 233 pp., doi:.
Paolo
,
F. S.
,
H. A.
Fricker
, and
L.
Padman
,
2015
:
Volume loss from Antarctic ice shelves is accelerating
.
Science
,
348
,
327
331
, doi:.
Polvani
,
L. M.
, and
K. L.
Smith
,
2013
:
Can natural variability explain observed Antarctic sea ice trends? New modelling evidence from CMIP5
.
Geophys. Res. Lett.
,
40
,
3195
3199
, doi:.
Renwick
,
J. A.
,
A.
Kohout
, and
S.
Dean
,
2012
:
Atmospheric forcing of Antarctic sea ice on intraseasonal time scales
.
J. Climate
,
25
,
5962
5975
, doi:.
Rignot
,
E.
,
I.
Velicogna
,
M. R.
van den Broeke
,
A.
Monaghan
, and
J.
Lenaerts
,
2011
:
Acceleration of the contribution of the Greenland and Antarctic ice sheets to sea level rise
.
Geophys. Res. Lett.
,
38
,
L05503
, doi:.
Rignot
,
E.
,
S.
Jacobs
,
J.
Mouginot
, and
B.
Scheuchl
,
2013
:
Ice-shelf melting around Antarctica
.
Science
,
341
,
266
270
, doi:.
Rye
,
C. D.
,
A. C. N.
Garabato
,
P. R.
Holland
,
M. P.
Meredith
,
A. J. G.
Nurser
,
C. W.
Hughes
,
A. C.
Coward
, and
D. J.
Webb
,
2014
:
Rapid sea-level rise along the Antarctic margins in response to increased glacial discharge
.
Nat. Geosci.
,
7
,
732
735
, doi:.
Screen
,
J. A.
, and
I.
Simmonds
,
2010
:
The central role of diminishing sea ice in recent Artic temperature amplification
.
Nature
,
464
,
1334
1337
, doi:.
Shepherd
,
A.
,
D.
Wingham
,
D.
Wallis
,
K.
Giles
,
S.
Laxon
, and
A. V.
Sundal
,
2010
:
Recent loss of floating ice and the consequent sea level contribution
.
Geophys. Res. Lett.
,
37
,
L13503
, doi:.
Shepherd
,
A.
, and Coauthors
,
2012
:
A reconciled estimate of ice-sheet mass balance
.
Science
,
338
,
1183
1189
, doi:.
Stammerjohn
,
S. E.
,
D. G.
Martinson
,
R. C.
Smith
,
X.
Yuan
, and
D.
Rind
,
2008
:
Trends in Antarctic annual sea ice retreat and advance and their relation to El Niño–Southern Oscillation and Southern Annular Mode variability
.
J. Geophys. Res. Oceans
,
113
,
C03S90
, doi:.
Sutterley
,
T. C.
,
I.
Velicogna
,
E.
Rignot
,
J.
Mouginot
,
T.
Flament
,
M. R.
van den Broeke
,
J. M.
van Wessem
, and
C. H.
Reijmer
,
2014
:
Mass loss of the Amundsen Sea Embayment of West Antarctica from four independent techniques
.
Geophys. Res. Lett.
,
41
,
8421
8428
, doi:.
Swart
,
N. C.
, and
J. C.
Fyfe
,
2013
:
The influence of recent Antarctic ice sheet retreat on simulated sea ice area trends
.
Geophys. Res. Lett.
,
40
,
4328
4332
, doi:.
Swingedouw
,
D.
,
T.
Fichefet
,
P.
Huybrechts
,
H.
Goosse
,
E.
Driesschaert
, and
M.-F.
Loutre
,
2008
:
Antarctic ice-sheet melting provides negative feedbacks on future climate warming
.
Geophys. Res. Lett.
,
35
,
L17705
, doi:.
Taylor
,
K. E.
,
R. J.
Stouffer
, and
G. A.
Meehl
,
2012
:
An overview of CMIP5 and the experiment design
.
Bull. Amer. Meteor. Soc.
,
93
,
485
498
, doi:.
Timmermann
,
R.
, and Coauthors
,
2010
:
A consistent dataset of Antarctic ice sheet topography, cavity geometry, and global bathymetry
.
Earth Syst. Sci. Data
,
2
,
261
273
, doi:.
Turner
,
J.
, and Coauthors
,
2009
:
Non-annular atmospheric circulation change induced by stratospheric ozone depletion and its role in the recent increase of Antarctic sea ice extent
.
Geophys. Res. Lett.
,
36
,
L08502
, doi:.
Turner
,
J.
,
J. S.
Hosking
,
T.
Phillips
, and
G. J.
Marhsall
,
2013
:
Temporal and spatial evolution of the Antarctic sea ice prior to the September 2012 record maximum extent
.
Geophys. Res. Lett.
,
40
,
5894
5898
, doi:.
Vaughan
,
D. G.
, and Coauthors
,
2013
: Observations: Cryosphere. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 66 pp. [Available online at https://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter04_FINAL.pdf.]
Velicogna
,
I.
, and
J.
Wahr
,
2013
:
Time-variable gravity observations of ice sheet mass balance: Precision and limitations of the GRACE satellite data
.
Geophys. Res. Lett.
,
40
,
3055
3063
, doi:.
Williams
,
S. D. P.
,
P.
Moore
,
M. A.
King
, and
P. L.
Whitehouse
,
2014
:
Revisiting GRACE Antarctic ice mass trends and accelerations considering autocorrelation
.
Earth Planet. Sci. Lett.
,
385
,
12
21
, doi:.
Zunz
,
V.
, and
H.
Goosse
,
2015
:
Influence of freshwater input on the skill of decadal forecast of sea ice in the Southern Ocean
.
Cryosphere
,
9
,
541
556
, doi:.
Zunz
,
V.
,
H.
Goosse
, and
F.
Massonet
,
2013
:
How does internal variability influence the ability of CMIP5 models to reproduce the recent trend in Southern Ocean sea ice extent
.
Cryosphere
,
7
,
451
468
, doi:.
Zwally
,
H. J.
,
M. B.
Giovinetto
,
J.
Li
,
H. G.
Cornejo
,
M. A.
Beckley
,
A. C.
Brenner
,
J. L.
Saba
, and
D. H.
Yi
,
2005
:
Mass changes of the Greenland and Antarctic ice sheets and shelves and contributions to sea-level rise: 1992–2002
.
J. Glaciol.
,
51
,
509
527
, doi:.

Footnotes

1

It should be noted that Bintanja et al. (2013) justified their use of 250 Gt yr−1 based on Rignot et al. (2011). Interestingly, the value given by Rignot et al. (2011) for Antarctic ice sheet loss in 2006 was using the mass budget method ( is the net imbalance for Greenland.)