Active sea-ice production in Antarctic coastal polynyas causes dense water formation, finally leading to Antarctic Bottom Water (AABW) formation. This study gives the first mapping of sea ice production in the Antarctic Ocean, based on heat-flux calculation with ice thickness data derived from satellite data. The highest ice production occurs in the Ross Ice Shelf Polynya region. The ice production there decreased by ∼30% from the 1990s to the 2000s, which can be one candidate for causing the recent freshening of AABW. The Cape Darnley polynya in East Antarctica is found to be the second highest production area, suggesting a possible AABW formation area. According to our estimation, around 10% of Southern Ocean sea ice is produced in the major Antarctic coastal polynyas. The mapping provides surface heat- and salt-flux conditions in the ice-covered region, which have not been well understood.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The sinking of dense water in the polar regions drives the global thermohaline circulation, leading to heat and material exchange between the atmosphere and deep ocean. It is widely recognized that dense water formation caused by brine rejection due to high ice production in Antarctic coastal polynyas is responsible for the formation of Antarctic Bottom Water (AABW), which is the densest water mass in the world ocean and the most important source of the abyssal circulation. Antarctic coastal polynyas are formed by divergent ice motion due to prevailing winds and/or oceanic currents, and most of the area is covered with thin ice [Pease, 1987]. During winter, heat loss over thin ice is one or two orders of magnitude larger than that over thick ice [Maykut, 1978], and thus coastal polynyas are regarded as the ice production factories. However, quantitative estimation of ice production in Antarctic coastal polynyas has not been made due to the difficulty of direct in-situ measurements. The purpose of this study is to estimate the spatial distribution and interannual variability of ice production in coastal polynyas over the entire Southern Ocean.
 In this study, we estimate the ice production which contributes to deep water formation through satellite remote sensing. Although spatial and temporal changes of coastal polynyas are quite large, passive microwave data (e.g., Special Sensor Microwave Imager [SSM/I] data) have the potential to estimate thin ice thickness over the entire Southern Ocean on a twice daily basis regardless of darkness or cloud cover. By using the ice thickness data, ice production over the thin ice areas can be estimated from heat budget analysis, if all the heat loss is assumed to be used for ice formation.
 In some polynyas, there have been studies that estimate ice production from heat flux calculation using the thin ice algorithm from passive microwave data [Markus et al., 1998; Renfrew et al., 2002; Martin et al., 2007]. In their studies, there still remain problems in using the passive microwave data for quantitative estimation of ice production. First, their passive microwave algorithms are mainly for the detection of thin ice areas and are not suitable for the estimation of ice thickness itself in the Southern Ocean, though Martin et al. [2004, 2005] proposed the algorithm that provides thin ice thickness in an Arctic polynya. This is mainly due to the lack of validation of ice thickness from in-situ data in Antarctic coastal polynyas. Since surface heat loss is considerably sensitive to the ice thickness, an algorithm that provides ice thickness itself is required for quantitative estimation of ice production. The second problem is the misinterpretation of signal from ice shelves, fast ice, and icebergs with that from the thin ice area. These types of ice generally exist adjacent to Antarctic coastal polynyas, and their microwave characteristics are similar to those of thin ice. Furthermore, the ice shelves, fast ice, and icebergs areas are changeable due to calving, break-up, and drifting. The third problem is relatively low spatial resolution of the passive microwave data (15–30 km), which cannot detect narrow coastal polynyas.
 To overcome the first problem, we use a newly developed algorithm [Tamura et al., 2007] that estimates the thickness for thin sea ice (<0.2 m). The algorithm was developed based on a comparison between polarization ratios of brightness temperatures in the 85-GHz and 37-GHz channels of SSM/I and ice thickness estimated from Advanced Very High Resolution Radiometer (AVHRR) data. The AVHRR ice thickness was validated by ground truth data observed in a coastal polynya during the Antarctic Remote Ice Sensing Experiment (ARISE) 2003 [Tamura et al., 2006]. To overcome the second problem, ice shelves, fast ice, and icebergs are discriminated from thin sea ice by using a scatter plot of vertical versus horizontal polarization brightness temperatures at 85-GHz [Tamura et al., 2007]. Signals of ice shelves, fast ice, and icebergs are correlated with those of the ice sheet on the scatter plot. The discrimination is carried out using this characteristic. To minimize the third problem (low resolution), the thermal ice thickness is used instead of algebraically averaged ice thickness for each grid cell when the AVHRR ice thickness is mapped onto the SSM/I grid cell [Tamura et al., 2007]. The thermal ice thickness is defined by the thickness for which the calculated total heat loss from AVHRR data can be realized under the assumption of a uniform ice thickness in the SSM/I grid cell. In other words, the thermal ice thickness is a thickness that is suitable for the heat flux calculation in a bulk fashion for a given grid cell, whereas in reality the ice thickness is not uniform.
2. Data and Methods
 Ice production is estimated from heat flux calculation during the freezing period (from March to October) by assuming that all of the heat loss at the surface is used for ice formation [Ohshima et al., 2003]. Specifically, the volume of ice production V is given by V = H/ρiLf, where H is the heat loss at the thin ice surface, ρi (= 900 kg m−3) is the ice density, and Lf (= 0.302 MJ kg−1) is the latent heat of fusion of sea ice. The oceanic heat flux from below is expected to be small in the Antarctic coastal polynyas because the whole water column is close to the freezing point during winter in the continental shelf region [Muench and Gordon, 1995; Jacobs and Giulivi, 1998; Williams and Bindoff, 2003]. Thus, the water temperature under ice is assumed to be at the freezing point (271 K). Heat loss is obtained assuming that the sum of radiative and turbulent fluxes at the ice surface is balanced by the conductive heat flux in the ice, whose thickness is derived from the thin ice algorithm using the SSM/I Equal Area Scalable Earth-Grid (85-GHz and 37-GHz) data. For the SSM/I data, no-data pixels and pixels that are judged to be affected by atmospheric water vapor are filled using spatial and temporal interpolation [Tamura et al., 2007]. In the algorithm, the 85-GHz data that are judged to suffer from atmospheric effects through the comparison with the 37-GHz data are excluded. Each heat flux component is calculated by the formula that is suitable for the Antarctic sea-ice zone [Nihashi and Ohshima, 2001]. As atmospheric input data, we use air and dew-point temperatures at 2 m, wind at 10 m, and surface sea level pressure both from the European Centre for Medium-Range Weather Forecasts Re-Analysis data (ERA-40; 1992–2001) and the National Centers for Environmental Prediction/Department of Energy Re-Analysis data (NCEP2; 1992–2005), with the resolutions of 1.125° × 1.125° and 1.9° × 1.9°, respectively. These data are interpolated onto the SSM/I grid (12.5 km × 12.5 km). The calculation is performed twice a day over the entire Southern Ocean.
 Based on the procedure described in the previous section, we provide the mapping of ice production in the Antarctic Ocean (Figure 1). The high ice production is confined to the coastal polynya areas. The high ice production areas appear on the west side of the peninsula and glacier tongue which is a downstream region of the Antarctic Coastal Current. This feature is particularly eminent along the coast of East Antarctica (such as the Mertz Glacier Polynya; see Figure 1 (right)). The largest ice production area exists in the Ross Sea. In contrast, in the Weddell Sea, which is considered to be a major source region of AABW formation [Orsi et al., 2002], ice production is not prominent. There is a possibility that Weddell Sea ice production is somewhat underestimated due to the difficulty in detection of narrow coastal polynyas which are dominant there. Even so, the ice production is expected to be smaller than that in other areas. It is likely that the dense water over the Weddell Sea shelf is formed by the accumulated brine rejection along the Antarctic Coastal Current from the east [Markus et al., 1998], rather than solely by the local ice production. The mapping also provides the surface salinity flux by the ice formation: annual ice production of 1 m corresponds to salinity flux of 21.7 kg m−2 yr−1, assuming water and thin ice salinity of 35 and 10.85 [Martin and Kaufmann, 1981], respectively.
 Among the 13 polynyas of high ice production, the Ross Ice Shelf Polynya has by far the highest ice production (Table 1). This is consistent with the fact that AABW with the highest salinity is formed in the Ross Sea [Orsi et al., 1999]. What we found in the mapping is that the second highest production is in the Cape Darnley polynya located west of the Amery Ice Shelf (the polynya name follows Massom et al. ). The third highest one is the Mertz Glacier Polynya, which produces dense shelf water as an ingredient of the Adélie Land Bottom Water (ALBW) [Rintoul, 1998]. Since high ice production is the primary condition for the formation of dense shelf water or AABW, the Cape Darnley polynya can be a formation site of AABW. Actually, from the in-situ observation [Jacobs and Georgi, 1977], sinking of dense water with high dissolved oxygen was shown on the shelf slope just west of this polynya, downstream along the path of the Antarctic Coastal Current. This study promotes us to investigate this less-observed area of the Cape Darnley polynya as a possible dense water or AABW formation area. Motivated by this result, the Japanese program of the International Polar Year plans to focus on this polynya.
Table 1. Mean Values of Annual Cumulative Sea-Ice Production for the Major 13 Antarctic Coastal Polynyas with their Standard Deviations and Trendsa
Ice Production, km3
The calculation was performed for 1992–2001 using the ERA-40 data. The locations of the polynyas are indicated in Figure 1.
390 ± 59
181 ± 19
120 ± 11
110 ± 14
92.0 ± 16
84.6 ± 25
80.0 ± 19
75.5 ± 11
73.3 ± 9.9
68.2 ± 5.8
59.2 ± 10
42.6 ± 6.7
33.7 ± 6.1
1410 ± 75
 The Mertz Glacier Polynya is a relatively well-observed Antarctic polynya. Its sea ice production rate averaged from July through August 1999 was estimated to be 5.8 cm day−1 through a salt budget analysis in the ocean [Williams and Bindoff, 2003]. The ice production rate during the same period in this study, 4.9 cm day−1, is close to this value. From the ice-ocean coupled model [Marsland et al., 2004], the ice production rate in this polynya averaged over winter months (from May to September) is 4.9 cm day−1. The production rate averaged over the same periods in this study is 5.0 cm day−1. Our estimation is considered to be within a realistic range.
 The maximum ice area in the Southern Ocean is ∼17 × 106 km2 while the minimum is ∼4 × 106 km2 [Zwally et al., 2002]. Assuming a mean ice thickness of 1 m, the annual ice production becomes 13 × 103 km3. According to the total of Table 1, around 10% of Southern Ocean sea ice is produced in the 13 major Antarctic coastal polynyas, though the total area of the 13 polynyas is only ∼1% of the maximum ice area. In Table 1, it is also noted that more than the half of the total ice production is occurring in the East Antarctica (40–160°E).
 In the Ross Ice Shelf Polynya with the highest ice production, the production was decreased by ∼30% from the average of the 1990s to that of the 2000s (Figure 2a). This is also indicated by a largest negative trend value in ice production (Table 1). The ice production reduction of 120 km3 corresponds to a mass gain of ∼70 billion metric tons of fresh water. The observational results showed that, during the second half of the 20th century, bottom water which is thought to be formed in the Ross Sea and offshore of Adélie Land has gradually freshened and that AABW in the Australian Antarctic Basin originating from these bottom waters has also gradually freshened [Jacobs et al., 2002; Whitworth, 2002; Rintoul, 2007]. The causes of the freshening are considered to be glacial ice melting, an increase in precipitation, or a decrease in sea-ice formation [Jacobs, 2004; Rintoul, 2007]. Such freshening seems to occur more rapidly from the 1990s to the 2000s [Rintoul, 2007], which is our analysis period. In-situ observations showed that the salinity of the Ross Sea Bottom Water (RSBW) significantly decreased from 1997 to 2001 [Bergamasco et al., 2004]. The ALBW, which is formed by both the dense shelf water in the Mertz Glacier Polynya and the inflowing high salinity RSBW [Rintoul, 1998], has also freshened from the mid-1990s to the early 2000s [Aoki et al., 2005]. Further, the AABW in the Australian Antarctic Basin freshened rapidly from the mid-1990s to the mid-2000s [Rintoul, 2007]. From our results, the ice production in the Mertz Glacier Polynya was not decreased (Table 1). In this study, we propose that the decrease in ice production in the Ross Ice Shelf Polynya is a plausible candidate for causing recent freshening of RSBW, ALBW, and AABW in the Australian Antarctic Basin.
 There are two candidates for causing the ice production decrease of the 2000s in the Ross Ice Shelf Polynya (Figure 2a): decrease in the polynya area and atmospheric warming. During the winters of 2000 and 2002, the development of the Ross Ice Shelf Polynya was suppressed by giant icebergs, B-15 and C-19, calved from the Ross Ice Shelf [Martin et al., 2007]. The sea-ice algorithm used in this study can detect these giant icebergs [Tamura et al., 2007]. Our analyses show that the polynya area and the ice production significantly decreased in 2000 and 2002 (Figure 2b). The causes of the small ice production in 2001 and 2003–2005 are considered to be both reduction of polynya area and atmospheric warming. Although air temperature had already risen since 1996 (Figure 2c), larger polynya area kept relatively high ice production during the period of 1996–1999 (Figure 2b).
 Since the ice production estimation is performed with the same thin ice algorithm and objective analysis data over the entire Southern Ocean, spatial distribution and interannual variability can be discussed, although uncertainties in absolute value of ice production cannot be excluded. The uncertainties mainly arise from errors in ice thickness data and atmospheric input data (mainly wind speed and air temperature). The sensitivity tests of ice production averaged over the 13 coastal polynyas during 1992–2001 are carried out as follows. The sensitivity of the estimated ice production to the ice thickness is checked by changing the thickness by ± 0.05 m (this value is based on the error of the algorithm [Tamura et al., 2007]). The annual cumulative ice production is decreased by ∼24% (increased by ∼27%) when the ice thickness is increased (decreased) by 0.05 m. The sensitivities to the atmospheric data are checked by using the difference in data set between ERA-40 and NCEP2, averaged over the ocean area between 65°S and 80°S. The changes in estimated annual cumulative ice production due to the changes in air temperature (1.5 K) and wind speed (15%) are 8.4% and 15%, respectively. Furthermore, the ice production calculated using the NCEP2 data is higher by 26% than that using the ERA-40 data. Although these tests are not exact error analyses, they give measures of uncertainties in the estimated ice production.
 Finally, ambiguities in the estimation of ice production are discussed. Although oceanic heat flux from below has been neglected in this study, this is not always the case. Ice production was estimated during the freezing period (from March to October). In March and October, ambiguity in heat loss at the thin ice surface becomes relatively large, because of relatively large shortwave radiation for which the treatment of the absorption into the ice interior has large uncertainty. Even so, this ambiguity does not affect the annual cumulative sea ice production much because the production in both months is not large. The thin ice algorithm used in this study does not always detect the coastal polynya and lead when the area is smaller than the SSM/I resolution in the 85-GHz channel (12.5 km × 12.5 km) as in the case of the Weddell Sea, described in Section 3. It is noted that our estimation does not include the ice production in thick ice regions with thicknesses > 0.2 m. However, the ice production rate there seems much smaller than in thin ice region. Although such ambiguities exist, the ice production mapping in this study gives the basic information for the surface heat- and salt-flux conditions in the ice-covered sea, which has not been well understood. These will be useful for validating and providing boundary conditions for coupled atmosphere-ice-ocean models.
 The SSM/I data were provided by the National Snow and Ice Data Center (NSIDC), University of Colorado. We sincerely acknowledge Shigeru Aoki and Yasushi Fukamachi for their instructive comments. We express our appreciation to anonymous reviewers for their useful comments. This work was supported by the 21st Century Center of Excellence Program funded by the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT), the fund from the Research Revolution 2002 of the Project for Sustainable Coexistence of Human, Nature and the Earth within the MEXT, and the fund from Core Research for Evolutional Science and Technology of Japan Science and Technology Agency.