Submarine groundwater discharge in Lützow-Holm Bay, Antarctica



[1] We measured submarine groundwater discharge (SGD) for the first time in Antarctica, at Lützow-Holm Bay, using a newly developed automated seepage meter. Measured SGD rates ranged from 10−8 to 10−6 m s−1 are substantially greater than previously observed discharge rates at similar depth. The obtained SGD rates are accurate to within 3%–5%. Using a fast Fourier transform, we found that variations in the power spectrum density had a predominant period of 12.8 h, similar to that of the M2 tide, with peak values differing from average SGD rates by a factor of three. The high SGD rates obtained at the Antarctic marginal ice zone may reflect the contribution of significant volumes of subglacial meltwater from the rugged subglacial mountains located about 20 km inland from the coast. We discuss a possible mechanism of meltwater discharge to this coastal region.

1. Introduction

[2] Studies undertaken in the 1980s revealed a large subglacial lake, Lake Vostok, in Victoria Land, Antarctica [Kapitsa et al., 1996], with north–south tidal motions [Wendt et al., 2005]. Recent geophysical data indicate the existence of 145 subglacial lakes in Antarctica [Siegert et al., 2005], with several being interconnected by subsurface channels to enable rapid water migration, possibly involving discharge into the ocean [e.g., Sawagaki and Hirakawa, 2002; Wingham et al., 2006]. These studies indicate that large amounts of water may exist at the bottom of the Antarctic ice sheet, with freshwater being discharged from the land to the coastal ocean as submarine groundwater discharge (SGD). To understand water-mass discharge from Antarctica, it is necessary to estimate not only ice mass balance from space observations but also to directly measure subsurface water flow to the coastal ocean.

[3] For direct measurements of SGD, in situ measurements by, for example, piezometer and Rn (Radon) methods, require a priori knowledge of related parameters; consequently, these methods cannot strictly be considered direct measurement methods [see Taniguchi et al., 2003]. Traditional manual seepage meters [Lee, 1977] were previously the only instrument employed in directly measuring SGD rates. Various types of automated seepage meters have recently been developed, such as ultrasonic groundwater seepage meters, dye-dilution meters, and autonomous electro-magnetic (EM) seepage meters, among which continuous-heat type automated seepage meters (ASMs) [Taniguchi and Iwakawa, 2001; Mwashote et al., 2010] can record long-term SGD trends at high resolution, enabling accurate estimates of SGD rates.

[4] The application of ASMs was previously limited to shallow water depths of 0.5–35 m [Taniguchi et al., 2002, 2006]; however, we developed a new continuous-heat type ASM for special application in deep water in Antarctica (Antarctic ASM). For details on this instrument and its calibration, see Mwashote et al. [2010].

2. Study Area and Methods

[5] The present study area comprises the sea-bottom region from Syowa Station (East Ongul Island) to Langhovde, Lützow-Holm Bay, East Antarctica (Figure 1a). In the eastern part of the study area, a continental ice sheet, including several glaciers, flows into Lützow-Holm Bay. Langhovde is one of several areas of rocky outcrop found scattered along the southern Sôya Coast. The maximum sea-floor depth along the coast (i.e., the Ongul Strait) ranges from 600 to 700 m, with the trough having formed by the scouring action of past glaciers [Moriwaki, 1979]. The bedrock in the study area is metamorphosed granite gneiss, and sea-bottom sediments recovered by core sampling consist largely of clay, silt, and sand [e.g., Igarashi et al., 2001].

Figure 1.

(a) Bathymetric map of Lützow-Holm Bay showing the locations of measurement sites at Stations 1–7 (black circles). The depth contour interval is 100 m. The dashed line P–P′–Q along the streamline of the Langhovde Glacier indicates the location of the profile shown in Figure 3. Gray areas indicate islands or bare-rock area in the ice sheet. (b) Deployment and mooring of the Antarctic ASM from on-ice, showing the sea-bottom setting and sensing unit (enlarged). CT: Conductivity–temperature sensor.

[6] Although data on in situ permeability are unavailable for the area, grain-size analysis of a 40-cm sea-floor sediment core recovered from Station 6 (see Figure 1a) revealed a silt content (grain size of 0.338–62.5 μm) of 45.8%–57.3%, clay content (grain size <0.338 μm) of 40.2%–51.5%, and sand content (grain size of 62.5 μm to 2 mm) of 2.5%–2.9% [Shimizu et al., 1988]. Glaciomarine sediments within a 117-cm core recovered from Station 5 [Sasaki, 1985] were immature, feldspathic, wacke-type muddy sand or sandy mud with an average grain size of 200 μm.

[7] The Antarctic ASMs were installed from on-ice during April–November 2005, once the sea ice had hardened. The ASMs were deployed using an electrical winch through a hole drilled in the ice (area, 0.4 m2; wall thickness, 0.5–1.5 m; see Figure 1b), and were moored using a float and rope. Locations of ASMs deployment-site were carefully selected depending on the geomorphology and accessibility. Arrival of the ASMs at seafloor was made sure by rope, and the sediment of the seafloor is soft enough to install the chamber with good sealing because of the weight of the chamber itself. SGD measurements were made hourly (Locations 3–7; see Figure 1a) or every 2 h (Locations 1 and 2). The environmental temperature of seawater was measured using a temperature sensor. During the period of ASM measurements, continuous tidal data were recorded at Syowa Station (see Figure 1).

[8] The vented benthic chambers of the ASMs have a sectional area of 0.14 m2. They are designed to gather the upward water flux into a guide tube to enable measurement of the temperature gradient between the downstream and upstream thermal resistors using a heating element located on the downstream side (see the sensing unit shown in Figure 1b). The temperature difference between the two thermal resistors (separated by a distance of 17 cm) is greatest during periods of no water flow, and decreases systematically with increasing water-flow velocity; this ASM cannot measure inverted flow. ASM measurements are based on the principle of heat dissipation [e.g., Taniguchi and Iwakawa, 2001; Mwashote et al., 2010] in one-dimensional water flow; the relationship between temperature gradient by voltage output and fluid flux (the V–F relationship) must be calibrated based on laboratory experiments.

[9] The present SGD measurements were performed under stable seawater temperatures of between −1.7 and −1.5°C, with a standard deviation of less than 0.05°C. Controlled laboratory experiments revealed that the V–F relationship was a function of the environmental temperature surrounding the sensors. A 0.1°C fluctuation in the environmental temperature is considered to result in 0.30%–0.35% uncertainty in the voltage output, corresponding to 3%–5% uncertainty in measurements of flow rate, as shown in Figure 4b of Mwashote et al. [2010].

3. Measurement Results

[10] Table 1 summarizes the measurement data obtained at each location, including site data. The average SGD rates obtained at Lützow-Holm Bay ranged from 0.85 × 10−7 to 9.5 × 10−7 m s−1, with the highest rate observed west of Langhovde (Location 6).

Table 1. SGD Data Recorded Using a Seepage Meter at Seven Sites in Lützow-Holm Bay, Antarctica, in 2005
 Location 1Location 2Location 3Location 4Location 5Location 6Location 7
  • a

    Standard deviation, representing the range in values over the measurement period rather than the error of the average estimate of submarine groundwater discharge (SGD).

  • b

    Seawater temperature.

Period8–14 Apr.22–26 Apr.17–22 Nov.11–16 Nov.4–11 Nov.14–18 Oct.5–12 Oct.
Water depth12 m12 m665 m260 m660 m470 m60 m
SGD Average0.85 × 10−7 m s−10.94 × 10−7 m s−11.70 × 10−7 m s−11.08 × 10−7 m s−10.86 × 10−7 m s−19.5 × 10−7 m s−10.89 × 10−7 m s−1
S.D.a0.10 × 10−7 m s−10.44 × 10−7 m s−12.80 × 10−7 m s−10.08 × 10−7 m s−10.19 × 10−7 m s−116.7 × 10−7 m s−10.20 × 10−7 m s−1
Sea Temp.b−1.60°C−1.69°C−1.51°C−1.60°C−1.58°C−1.55°C−1.57°C

[11] Figure 2a shows clear semidiurnal changes in hourly averaged SGD rates at Location 3 and tidal variations at Syowa Station (Location 1) from 17 to 23 November 2005. The observed SGD rate increases with decreasing tidal sea level. Moreover, the amplitude of SGD variations decreases with decreasing tidal amplitude; that is, SGD variations are larger during spring tide than during neap tide. The maximum hourly SGD rate (5.1 × 10−7 m s−1) is higher than the average value (1.7 × 10−7 m s−1) by a factor of three. The fast Fourier transform (FFT) power spectrum density (PSD) for the SGD variations during the same period shows a peak at a period of 12.8 h, slightly longer than the period of the M2 tide.

Figure 2.

(a) Temporal variations in the rate of submarine groundwater discharge (SGD) at Location 3 (thin line; see Figure 1) compared with tidal variations at Syowa Station (thick line). (b) Rate of submarine groundwater discharge (SGD) measured by seepage meters at mid-latitude sites (open circles) in previous studies, which show an overall decrease with increasing seawater depth D; these values were scaled to the unit of meters per second. Solid circles indicate data from the present study at Lützow-Holm Bay. Data numbers correspond to the locations in Table 1.

4. Discussion

4.1. Observed SGD Values

[12] SGD consists of the total submarine discharge of fresh groundwater from land to the sea (SFGD) and the discharge of recirculated saline groundwater (RSGD) [e.g., Burnett et al., 2006]. Recent studies have reported that SGD increases sharply from neap to spring tide, with a pseudo-semi-monthly period. This phenomenon in shallow waters (depths to about 100 m) has been explained previously in terms of tidal pumping [e.g., Taniguchi et al., 2006]. When the sea level rises with the tide, SFGD decreases because the hydraulic head difference between land and sea is reduced, and RSGD becomes smaller with increasing water pressure. It is interesting that our data obtained in deep waters (e.g., Location 3 at 665 m depth) show similar characteristics to those obtained in shallow waters. However, because the ratio of tidal height change (1.5 m) to sea depth (665 m) is only 0.2%, the effect of tidal pumping in terms of surface water should be reduced by an order of magnitude compared with ratios of 15%–40% for a shallow sea depth of 10 m.

[13] Figure 2b shows that the observed SGD rates obtained by seepage meters in previous studies at other sites of mid-latitude generally decrease with increasing water depth (D) [Taniguchi et al., 2002]. These lower SGD rates for greater water depths (larger D) were interpreted from the potential difference between land and sea, which decreases with increasing water pressure [e.g., Burnett et al., 2006; Taniguchi et al., 2006].

[14] However the present SGD data from the Antarctic coast show no clear relation with seawater depth (solid circles in Figure 2b). Moreover, the data are larger by one or two orders of magnitude than the range of existing data. Unfortunately, conductivity measurements were not successful during the present study because of malfunctioning of the conductivity sensor inside the chamber; consequently, we were unable to separate SGD into SFGD and RSGD based on conductivity data. However, in the deep sea-floor setting of the study area, sedimentary layers are thin, and under those geological setting, RSGD is difficult to imagine. Assuming that for these reasons, RSGD is small or negligible, a terrestrial source is likely. Here, we propose a potential origin and mechanism for terrestrially-derived SGD.

4.2. Potential Origin of the Fresh Groundwater Discharging Into Lützow-Holm Bay

[15] Figure 3 shows a schematic of the sub-ice topography along the line P–P′–Q in Figure 1a, based on BEDMAP [Lithe et al., 2000]. Nishio et al. [1989] calculated that the basal temperature along the flow line of Shirase Glacier (located 200 km south of the area shown in Figure 1a) is at pressure melting point between 20 and 260 km from the coast. Radio-echo sounding data reported by Mae and Yoshida [1987] for this region support the occurrence of melting and a small amount of water in the ice mass or the presence of a wet-film at the ice–rock interface.

Figure 3.

Schematic cross-section of the ice sheet/bedrock topography along the dashed line in Figure 1a. Meltwater is believed to accumulate on the foreside slope of the bedrock, provided the condition of γ/α > 11 is met, where γ is bed slope and α is the ice surface gradient. Locations of P, P′, Q and Station 5 are shown in Figure 1.

[16] A previous simulation using an equi-potential hydrostatic model [Sawagaki and Hirakawa, 2002] showed that basal meltwater might accumulate in bedrock depressions, provided the relationship 11α < γ is satisfied [e.g., Shoemaker, 1991], where α is the ice surface gradient (positive if the ice surface slopes in a downglacier direction) and γ is the bed slope (positive if the bed is ascending in a downglacier direction) (see Figure 3). Once the subglacial meltwater has been produced, the water first fills the nearest subglacial basin. Once the upper basins are filled with water, the subglacial water overflows to the nearest lower basin or to the ice margin. Once all the basins are full, steady-state drainage occurs.

[17] For the schematic profile shown in Figure 3, the ratio of average bedrock topography (tanγ ∼1.4 × 10−2) to the average ice sheet topography (tanα ∼5.0 × 10−3) yields γ/α ∼ 3, which does not meet the condition of γ/α > 11. However, the bare-rock region in Lützow-Holm Bay possesses a rugged topography [e.g., Geographical Survey Institute, 2002] with variations in slope by a factor of three to six over short distances; accordingly, we infer that the surrounding subglacial region has similar variations in slope. Under these conditions, melt-water generation would occur in the study area.

[18] Sawagaki and Hirakawa [2002] estimated that the size of the largest subglacial pond extends for 6 km in length and 300 m in depth, located inland of steep mountains situated 22 km inland from the Sôya Coast. This model requires preferential groundwater flow paths in the form of channel and stria in the bedrock. High SGD represents heterogeneous discharge zone, and may be common along the coastal margin of the Antarctic ice sheet.

5. Conclusions

[19] We estimated SGD rates, for the first time, in the Antarctic coastal zone with an accuracy of 3%–5%. Hourly averaged SGD rates measured using a newly developed ASM were between 10−8 and 10−6 m s−1 at several sites in Lützow-Holm Bay at seawater depths greater than 100 m. These values are up to two orders of magnitude higher than the empirically determined SGD rates at mid-latitude sites, which show a decreasing trend with increasing water depth. The maximum hourly SGD rate was higher than the average value by a factor of three; in contrast, the tidal range was only 0.2% of seawater depth. As a possible mechanism that explains the unexpectedly high SGD rates obtained in the present study, we considered the results of simulations performed using an equi-potential hydrostatic model [Sawagaki and Hirakawa, 2002]. The rugged subglacial bedrock topography across the marginal ice sheet is consistent with the conditions that produce meltwater upon the foreside slope of bedrock, which may discharge in the ocean if channels and stria are present.


submarine groundwater discharge for a unit coastline length, m s−1.


voltage output from a seepage meter, volts.


fluid flux, equivalent to SGD, m s−1.


power spectral density, dimensionless.


seawater depth, m.


[20] We thank the members of the 46th Japanese Antarctic Research Expedition for their assistance. This research was supported in part by Grant-in-Aid 16310015 (Principal Investigator: K. Shibuya) for scientific research awarded by the Japan Society for the Promotion of Science.

[21] The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.