Introduction

Surface melting of sea ice is triggered by solar radiation during the spring and summer. Complete melting of snow over sea ice increases surface pond formation (Reference Fetterer and UntersteinerFetterer and Untersteiner, 1998). Additionally, surface ponds form by seawater inundation, particularly around pressure ridges (Reference Ackley and SullivanAckley and Sullivan, 1994; Reference Haas, Thomas and BareissHaas and others, 2001; Reference Schnack-Schiel, Thomas, Haas, Dieckmann and AlheitSchnack-Schiel and others, 2001; Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002). The formation of a surface pond, a pool of water on the sea ice, dramatically affects the energy and mass balance (e.g. Reference Ebert and CurryEbert and Curry, 1993) and the optical properties of the underlying sea ice (e.g. Reference Perovich, Grenfell, Light and HobbsPerovich and others, 2002). Surface pond formation can also alter the sea-ice structure and lead to under-ice meltwater pond formation (Reference EickenEicken, 1994; Reference Eicken, Krouse, Kadko and PerovichEicken and others, 2002). Moreover, surface pond formation affects the chemical and biological characteristics of the sea-ice–water system (Reference GradingerGradinger, 1996; Reference Herborg, Thomas, Kennedy, Haas and DieckmannHerborg and others, 2001; Reference Schnack-Schiel, Thomas, Haas, Dieckmann and AlheitSchnack-Schiel and others, 2001; Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002).

In summer, up to 50% of the frozen surface of the Arctic Ocean is covered by surface ponds (Reference Fetterer and UntersteinerFetterer and Untersteiner, 1998), indicating the importance of understanding how surface ponds might relate to climate. The Surface Heat Budget of the Arctic Ocean (SHEBA) project, an interdisciplinary series of sea-ice field experiments focusing on mechanisms affecting the surface properties of Arctic sea ice, has contributed greatly to our knowledge of surface ponds (Reference PerovichPerovich and others, 1999). Reference Taylor and FelthamTaylor and Feltham (2004) developed a model of surface pond formation on sea ice to clarify the sequential processes from sea-ice formation to its erosion in summer. Reference Rösel and KaleschkeRösel and Kaleschke (2011) have quantified changes in the area occupied by surface ponds throughout the seasonal cycle using satellite data on Arctic sea-ice coverage.

These studies of surface ponds, however, have focused mainly on ponds on Arctic sea ice, although surface pond formation on Antarctic sea ice is not uncommon (Reference Ackley and SullivanAckley and Sullivan, 1994; Reference Haas, Thomas and BareissHaas and others, 2001; Reference Schnack-Schiel, Thomas, Haas, Dieckmann and AlheitSchnack-Schiel and others, 2001; Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002). Flooding of the snow/ice interface is a common feature of sea ice in Antarctica, where precipitation is high and sea ice is relatively thin. As a result, the snow cover usually does not melt completely and surface ponds do not form (Reference MassomMassom and others, 2001; Reference Sturm, Massom, Thomas and DieckmannSturm and Massom, 2010). However, even in Antarctica, surface ponds on sea ice are found in areas of multi-year landfast ice near the continental coast (Reference WadhamsWadhams, 2000) and around pressure ridges (Reference Ackley and SullivanAckley and Sullivan, 1994; Reference Haas, Thomas and BareissHaas and others, 2001; Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002). In addition, melting of wide areas (∼10 km) of multiyear fast ice and formation of surface ponds in LützowHolm Bay, East Antarctica, have been observed in satellite data (Reference Enomoto, Nishio, Warashina and UshioEnomoto and others, 2002). Although biogeochemical studies of surface ponds are important for understanding not only summertime ecology but also chemical cycles in atmosphere–ice–ocean systems, little information is available on the biogeochemistry of sea-ice surface ponds, particularly in Antarctica.

In this study, as a first step to understanding Antarctic surface ponds, we carried out a program of field measurements focusing primarily on the biogeochemical components of surface ponds. We also obtained a sea-ice core from the bottom of a surface pond and compared its physicochemical properties with those of an ice core collected away from the surface pond in order to clarify how surface pond formation affects the properties of the underlying sea ice.

Materials and Methods

Surface pond water was collected from nine ponds on multiyear landfast ice in Lützow-Holm Bay from 29 January to 5 February 2010 (Fig. 1; Tables 1 and 2). During this period, snow accumulation was minimal on the downwind side of Iwa-zima Island and an iceberg, allowing the formation of a bare-ice region and surface ponds on the sea-ice surface (Fig. 1b). The water surface of all surface ponds was covered by thin ice (Table 1).

Fig. 1. (a) Location map of the sampling area in Lützow-Holm Bay, East Antarctica. The dotted line indicates the ice edge in February 2010. (b) The study region, showing the surface pond locations, near the coast of the Antarctic continent. This Advanced Land Observing Satellite (ALOS)/Panchromatic Remote-sensing Instrumentfor Stereo Mapping (PRISM) image was obtained on 26 January 2010.

Table 1. Surface pond width, length, area, depth and ice thickness at the top of the pond. nm: not measured

Table 2. Temperature, salinity, Chl-a concentrations, DIC and nutrients concentrations for surface pond and under-ice water. ns: not sampled

To sample the water in a surface pond, a 0.05 m × 0.05 m hole was made in a thin ice surface with a hand-saw. Then surface pond water was pumped out through the hole with a diaphragm pump (EWP-01, As One Corporation, Osaka, Japan) from 0.3 m depth (Fig. 2a), and collected into a 12 mL glass screw-cap vial (Nichiden-Rika Glass Co. Ltd, Kobe, Japan) for measurement of salinity, a 500 mL Nalgene polycarbonate bottle (Thermo Fisher Scientific Inc., Waltham, MA, USA) for measurement of chlorophyll a (Chl a), a 120 mL amber glass vial (Maruemu Co. Ltd, Osaka, Japan) for measurement of the dissolved inorganic carbon (DIC) concentration, and a 10 mL polyethylene screw-cap vial (Eiken Chemical Co. Ltd, Tokyo, Japan) for measurement of inorganic nutrient concentrations. A saturated mercuric chloride (HgCl₂) solution (200 μL) was added to the DIC samples to stop biological activity. Samples were stored in a refrigerator (+4°C) until measurement of salinity and DIC, and in a freezer (−30°C) until measurement of nutrient concentrations. Samples for Chl-a measurement were immediately filtered through a 25 mm Whatman GF/F filter in the laboratory, which was near the sampling station. Chlorophyll pigments on the filters were then extracted with dimethylformamide (Reference Suzuki and IshimaruSuzuki and Ishimaru, 1990) for 24 hours at approximately −80°C.

Fig. 2. (a) Schematic section of a surface pond in the sea ice, illustrating the sampling of surface pond water and pond-ice and bare-ice core collection. (b, c) Photographs of the top portions of the split ice cores in transmitted light: (b) pond-ice core and (c) bare-ice core.

Sea-ice cores and under-ice water were also collected during the study period. One ice core was collected from the ice beneath a surface pond (pond-ice core) and another from a bare-ice site near the surface pond (bare-ice core) (Fig. 2a). Details of methods for sea-ice core sampling and under-ice water sampling, and of the analytical procedures used to determine the physical and chemical properties of the core and water samples, are given in the Appendix.

The temperature of surface pond water was measured at 0.3 m depth in each surface pond, and that of under-ice water was measured at 1 and 7 m below the bottom of the sea ice with a needle-type temperature sensor (Testo 110 NTC, Brandt Instruments, Inc., Prairieville, LA, USA) at each sampling time. In addition, a thermistor probe (TR-5106, T&D Corp., Nagano, Japan) and logger (RTR-52A, T&D Corp.) were installed in surface pond P1 at 0.3 m depth to monitor daily variations in the water temperature from 30 January to 5 February 2010.

Surface dimensions of surface ponds were measured with a tape measure. Areas were then calculated with ImageJ software version 1.43u (W. Rasband, National Institutes of Health, Bethesda, MD, USA; http://rsb.info.nih.gov/ij). To measure the depth distribution, the surface was divided into a grid (0.5–1.0 m). At each gridpoint, a 0.05 m × 0.05 m hole in the thin ice at the surface was made with a hand-saw, and surface pond depth was measured with a ruler. The thickness of the thin ice at each point was also measured.

Results

Surface pond dimensions

Surface pond dimensions are shown in Table 1. Most surface ponds were relatively long and narrow (Table 1). Their areas varied greatly and ranged from 4.1 to 67.5 m². Maximum depth ranged from 0.17 to 0.70 m, and correlated with surface pond area (n = 7, r ² = 0.6, p < 0.04) (Fig. 3). In addition, surface pond area was correlated with salinity (n = 9, r ² = 0.7, p < 0.01). Bathymetric mapping of depth distributions (not shown) showed that the surface ponds had basin structures.

Fig. 3. Relationships between (a) surface pond depth and area and (b) surface pond area and salinity.

Daily surface pond temperature variations

Time series of surface pond water temperature, solar radiation and air temperature are shown in Figure 4. Surface pond temperatures fluctuated from +0.0 to +1.4°C on a daily cycle, in synchrony with the fluctuations in solar radiation rather than air temperature.

Fig. 4. Time series of (a) surface pond P1 water temperature, (b) solar radiation and (c) air temperature. Solar radiation and air temperature were measured at Syowa station, <1 km from our surface pond study area (Japan Meteorological Agency; http://www.data.jma.go.jp/obd/stats/etrn/index.php).

Biogeochemical properties of surface pond water and under-ice water

Table 2 shows the results of the temperature, salinity, Chl-a concentration, and DIC and nutrient concentration measurements of surface pond water and under-ice water. In general, the temperatures of the surface pond water were higher than those of the under-ice water. Salinity and DIC and nutrient concentrations in surface pond water were much lower than those in under-ice water, whereas Chl-a concentrations in surface pond water were lower than or the same as those in under-ice water.

Relationships between biogeochemical parameters (DIC and nutrient concentrations) and salinity in surface pond and under-ice water are shown in Figure 5. In general, DIC, NO₃ + NO₂ and PO₄ concentrations in surface pond water were below the dilution line (Fig. 5a–c), except for PO₄ concentrations at low salinity (<1.7), which were above the dilution line (Fig. 5c). NO₃+NO₂ concentrations were depleted in surface pond water (Fig. 5b). No enrichment or depletion with respect to the dilution line was observed in surface pond SiO₂ concentrations (Fig. 5d).

Fig. 5. Plots of (a) DIC, (b) NO₃ + NO₂, (c) PO₄ and (d) SiO₂ concentrations versus salinity for surface pond water and under-ice water. Shaded areas indicate dilution lines predicted from values in under-ice water.

Comparisons of physico-chemical properties between the pond-ice and the bare-ice core

Photographs of the pond-ice and bare-ice core sections (Fig. 2b and c) showed remarkable differences: unlike the bare-ice core, the pond-ice core contained many voids and had an irregular edge. These results indicate that the pond-ice core was more porous than the bare-ice core.

Figure 6 shows depth profiles of bulk ice density and salinity in the pond-ice and bare-ice cores. Bulk ice density was 525.9 ± 64.5 kg m⁻³ in the pond-ice core and 825.1 ± 64.6 kg m⁻³ in the bare-ice core and increased with depth in both cores (Fig. 6a). Bulk ice salinity tended to increase with depth in the bare-ice core, from zero at the top, to 1.6 at the bottom of the sea ice (Fig. 6b). In the pond-ice core, bulk ice salinity was lower at each depth than the corresponding value in the bare-ice core (Fig. 6b).

Fig. 6. Depth profiles of bulk ice (a) density and (b) salinity in the pond-ice and bare-ice cores.

Depth profiles of bulk Chl-a and nutrient concentrations in the pond-ice and bare-ice cores are shown in Figure 7. The bulk Chl-a concentration in the bare-ice core was close to zero at the top and in the middle of the sea ice, but at the bottom it increased to 6.5 μg L⁻¹. In the pond-ice core, the bulk ice Chl-a concentration at each depth was lower than that at the corresponding depth in the bare-ice core (Fig. 7a). Bulk NO₃ + NO₂ concentrations were almost constant (1.2 ± 0.5 μmol L⁻¹) in the sea ice of the bare-ice core, whereas in the pond-ice core they were clearly higher and tended to decrease with depth, from a maximum of 9.6 μmol L⁻¹ near the top of the sea ice to a minimum of 3.2 μmol L⁻¹ near the bottom (Fig. 7b). In contrast, the bulk PO₄ and SiO₂ concentrations tended to increase with depth in both cores, and their profiles were similar to that of the bulk Chl-a concentration. Bulk PO₄ and SiO₂ concentrations in the pond-ice core were lower, however, than those at corresponding depths in the bare-ice core (Fig. 7c and d).

Fig. 7. Depth profiles of bulk ice (a) Chl a, (b) NO₃ + NO₂, (c) PO₄ and (d) SiO₂ concentrations in the pond-ice and bare-ice cores.

Discussion

The surface ponds examined were on the downwind side of an island and an iceberg, where snow accumulation was minimal, resulting in the formation of a bare-ice region. Snow is a good reflector, whereas bare ice is an effective absorber, of solar radiant energy (Reference Fetterer and UntersteinerFetterer and Untersteiner, 1998). Thus, surface ponds tend to form on bare sea ice. Surface ponds form every year in summer at this study site (personal communication from S. Kudo, 2009). Furthermore, in Lützow-Holm Bay, melting of multi-year fast ice over a wide area (∼10 km) and the formation of surface ponds have been observed by satellite (Reference Enomoto, Nishio, Warashina and UshioEnomoto and others, 2002).

We examined surface ponds that varied greatly in size and other characteristics, and obtained a wide range of information about their geometry, physics, chemistry and biology. Nearly all the studied surface ponds were approximately rectangular (Table 1), with their longer dimension aligned with the dominant northeast–southwest wind direction in this area. Snow surface conditions are changed by wind, causing sea-ice surface roughness (e.g. sastrugi) (Reference GoodwinGoodwin, 1990; Reference Sturm, Eicken, Gradinger, Salganek, Shirasawa, Perovich and LeppärantaSturm, 2009). Thus, these observations suggest that the changes in snow depth changed the melting rate of the snow, and that the surface ponds formed according to the snow surface roughness. Surface ponds were also formed by the flooding of ice-floe surface, leading to further snow and ice melt at the surface (Reference Ackley and SullivanAckley and Sullivan, 1994). The surface pond salinity varied widely from almost zero to sea-water values, depending on the dominant process for surface pond formation (Reference Ackley and SullivanAckley and Sullivan, 1994).

Surface pond area showed good correlation with both depth and salinity (Fig. 3), suggesting that it might be possible to predict surface pond depth and salinity from pond area information obtained from the images (pictures) taken from the ship, helicopter and satellite. Surface pond salinity increased with area (Fig. 3b), suggesting that it also increased with depth. Because the sea ice sampled from beneath a surface pond (pond-ice core) was more porous than that in the core collected outside the surface pond (bare-ice core) (Figs 2b and c and 6a), water likely passes from the surface pond down through the porous sea ice, and under-ice water probably moves up through the sea ice into the surface pond (Reference Eicken, Krouse, Kadko and PerovichEicken and others, 2002), so the extent of this exchange will depend on the thickness of the sea ice beneath the surface pond (core length), and thus on surface pond depth.

The results of the long-term measurement of surface pond temperature suggest that it varied with solar radiation rather than with air temperature (Fig. 4), indicating that solar radiation penetrates into the surface ponds through the thin ice cover, raising the water temperature in the surface pond to above the freezing point. Solar radiant energy absorbed by the meltwater will then melt additional ice. Surface ponds have been observed to increase in size and deepen, and to merge with adjacent surface ponds (Reference PerovichPerovich, 1999; Reference Eicken, Krouse, Kadko and PerovichEicken and others, 2002). These processes have been successfully modeled by a one-dimensional thermodynamic surface-pond–sea-ice model (Reference Taylor and FelthamTaylor and Feltham, 2004).

Comparison of surface pond water properties with dilution lines (Fig. 5) suggested that DIC and nutrients (NO₃ + NO₂ and PO₄) were removed by biological productivity from the water during pond formation. NO₃ + NO₂ concentrations were completely depleted (see Fig. 5b), suggesting that nitrogen was a limiting factor of biological productivity, even though other nutrients (PO₄ and SiO₂) were available (Fig. 5c and d). The nitrate depletions were also found in the surface ponds (Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002) and surface gap layer in the Weddell Sea (Reference Papadimitriou, Thomas, Kennedy, Kuosa and DieckmannPapadimitriou and others, 2009).

The algal production in surface ponds during pond formation was examined based on the changes in the NO₃ + NO₂ concentrations. If NO₃ + NO₂ concentrations behaved conservatively during ice and surface pond formation, the initial NO₃ + NO₂ concentrations could be calculated from the relationships between the NO₃+NO₂ concentrations and salinity shown in Figure 5b (dilution line). Because NO₃ + NO₂ concentrations in surface ponds were depleted over the sampling period, the initial NO₃ + NO₂ concentrations (0.0–6.1 μmol L⁻¹; NO₃ + NO₂ concentration above the dilution line for each salinity in Fig. 5b) could be used as the measure for algal production during surface pond formation. Algal production (as particulate organic carbon (POC)) was calculated using a Redfield C: N molar ratio of 6.6, and results indicated that 0.0–40.3 μmol of POC were produced by algal production during pond formation. This value might be a lower limit because it was assumed that there was no exchange of water between surface ponds and the under-ice. Earlier studies indicated that the salinities of surface ponds (7–30; Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002) and gap layers (20–33; Reference Papadimitriou, Thomas, Kennedy, Kuosa and DieckmannPapadimitriou and others, 2009) were higher than those in the surface pond in this study (0.1–8.6; Table 2). Therefore, the mixing with under-ice/ambient water occurred frequently and supplied the nutrients for the algal production, although nitrate was depleted in the water (Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002). This is probably one reason why Chl-a concentrations in the surface pond in this study were so much lower than those in previous studies (Reference Kennedy, Thomas, Kattner, Haas and DieckmannKennedy and others, 2002; Reference Papadimitriou, Thomas, Kennedy, Kuosa and DieckmannPapadimitriou and others, 2009). It would be reasonable to assume that the surface melting is the dominant process for formation of surface ponds in this study area.

In the pond-ice core, the trend of the bulk ice NO₃ + NO₂ concentration profile differed from those of the bulk ice PO₄ and SiO₂ concentration profiles (Fig. 7). Bulk ice NO₃ + NO₂ concentrations in the upper part of the pond-ice core were high (up to 9.6 μmol L⁻¹), and much higher than the concentration in surface pond water (Figs 5b and 7b; Table 2). Furthermore, NO₃+NO₂ concentrations in the pond-ice core were above the dilution lines (not shown). These results suggest that remineralization of organic matter occurred at the bottom of the surface pond (upper part of the pond-ice core). The particulate matter, formed as a result of biological productivity in the surface pond, sank to the bottom of the surface pond, where the organic matter was remineralized. In the Antarctic sea ice, the highest bacterial production was found in saline ponds on the surface of floes and in pore water in the interior of floes throughout the ice floes (Reference Kottmeier and SullivanKottmeier and Sullivan, 1990). Therefore, high dissolved organic carbon was measured in surface ponds in the Weddell Sea during summer (Reference Herborg, Thomas, Kennedy, Haas and DieckmannHerborg and others, 2001). In addition, Reference Schnack-Schiel, Thomas, Haas, Dieckmann and AlheitSchnack-Schiel and others (2001) found grazers in Antarctic surface ponds. The grazers are contributed and influenced by the biogeochemical cycles in the surface ponds (Reference Schnack-Schiel, Thomas, Haas, Dieckmann and AlheitSchnack-Schiel and others, 2001).

The pond-ice core was porous and characterized by low bulk ice density (Figs 2b and c and 6a), suggesting high porosity and permeability. Therefore, particulate and dissolved components could easily percolate downward through the ice. Because high amounts of NO₃+NO₂ were preserved in the pond ice (Fig. 7b), the depletion of NO₃+NO₂ probably occurred in the surface pond water (Fig. 5b; Table 2).

The trends of the bulk ice PO₄ and SiO₂ profiles in the pond-ice core increased with depth, especially below ∼1 m depth, similar to the bulk ice salinity trend, and the PO₄ and SiO₂ concentrations were generally lower than those in the bare-ice core (Figs 6b and 7c and d). These results suggest that the behaviors of PO₄ and SiO₂ were the same as that of salinity. No clear effect of the remineralization of organic matter on the PO₄ concentration in the pond-ice core was observed (Fig. 7c). However, excess (enrichment of) PO₄ was observed at very low salinity in the surface pond water (Fig. 5c), suggesting that some mineralized PO₄ at the bottom of the surface pond was diffused into the surface pond water. The low bulk ice SiO₂ concentration in the pond-ice core (Fig. 7d) and the salinity–SiO₂ concentration relationship in the surface pond water (Fig. 5d) suggested that biogenic opal (diatom) production and dissolution were balanced, or that the contribution of biogenic opal to the variation in the SiO₂ concentration in the pond ice and the surface pond water was minor because it is said that photosynthetic flagellates, which do not utilize SiO₂ during photosynthesis, dominated in the upper part of the Antarctic fast sea ice in McMurdo Sound (Reference Stoecker, Gustafson, Black and BaierStoecker and others, 1998, Reference Stoecker, Gustafson, Baier and Black2000) and near Davis Station, East Antarctica (Reference Thomson, McMinn, Kiessling, Watson and GoldsworthyThomson and others, 2006).

Bulk ice Chl-a concentrations in the bare-ice core were high in the lower part of the core (Fig. 7a). It is well known that assemblages of ice algae occur at the bottom of sea ice and contribute significantly to primary production (Reference LegendreLegendre and others, 1992; Reference Arrigo, Mock, Lizotte, Thomas and DieckmannArrigo and others, 2009). In contrast to the bare-ice core, Chl-a concentrations throughout the pond-ice core were low. The bottom concentration of 0.1 μg L⁻¹ was approximately 1/65 that of the bare-ice core (6.5 μg L⁻¹). These results suggest that surface pond formation led to a decrease in the Chl-a concentration of the sea ice. This decrease suggests that meltwater flushes the ice algae (which contain the Chl a) out of the porous sea ice into the under-ice water, and that the removal of ice algae living at the bottom of the sea ice occurred due to ice melting. Comparisons of the different lengths of the pond-ice and bare-ice cores with surface pond depth suggest that ice melting occurred at the bottom of the pond-ice more than the bare-ice core over the sampling period.

The physico-chemical properties of ice cores (Figs 2b and c and 6 and 7) showed remarkable differences between the pond-ice and the bare-ice core. However, the results obtained might be affected by short-range spatial variability in sea-ice properties (e.g. Reference Granskog, Leppäranta, Kawamura, Ehn and ShirasawaGranskog and others, 2004). No data were collected that would allow the heterogeneity of these sea-ice parameters to be evaluated. Therefore, it might be difficult to extrapolate the significance of the findings clearly. For future study, replicate samplings for sea-ice cores are needed.