Hostname: page-component-78c5997874-dh8gc Total loading time: 0 Render date: 2024-11-10T10:19:06.060Z Has data issue: false hasContentIssue false

Sea-ice concentration retrieval in the Antarctic based on the SSM/ I 85.5 GHz polarization

Published online by Cambridge University Press:  14 September 2017

Stefan Kern
Affiliation:
Institute of Environmental Physics, P.O. Box 330440, D-28334 Bremen, Germany
Georg Heygster
Affiliation:
Institute of Environmental Physics, P.O. Box 330440, D-28334 Bremen, Germany
Rights & Permissions [Opens in a new window]

Abstract

Using data from the 85 GHz channels of the Special Sensor Microwave/ Imager (SSM/I) allows a resolution improvement by at least a factor of four compared to the other channels. Consequently higher-resolution sea-ice concentration data can be obtained which in turn can be used to improve the results of numerical weather-prediction (NWP) and global circulation models. The proposed new sea-ice concentration retrieval algorithm (SEA LION algorithm) uses the polarization at 85 GHz (p). Emission from atmospheric water and scattering at the wind-roughened sea surface (weather effect) decrease p and cause an overestimate of the sea-ice concentration. We quantify the weather effect with a radiative transfer model and atmospheric data obtained from NWP models and the other SSM/I channels, and correct p for this effect. Tie points of open water and sea ice are determined for each month separately from daily gridded 85 GHz SSM/I brightness temperatures. Sea-ice concentrations are calculated with the new algorithm for the entire Southern Ocean for each day of the period 1992−98 with a spatial resolution of 12.5 × 12.5 km2. Comparisons of these ice concentrations with Operational Linescan System visible images reveal convincing results concerning the monitoring of coastal polynyas and the break-up of the pack ice in spring. SEA LION sea-ice extents and areas, and comparisons between SEA LION sea-ice concentrations and ship observations, agree with those obtained by the NASA Team and the Bootstrap algorithms:

Type
Remote Sensing of Sea-Ice and Snow-Cover Characteristics
Copyright
Copyright © the Author(s) [year] 2001

1. Introduction

In this paper, we describe and test a new algorithm, which exploits the higher spatial resolution of the 85 GHz channels of the Special Sensor Microwave/Imager (SSM/I) to retrieve higher-resolution daily sea-ice concentration data than are currently provided by other algorithms. The SSM/I sensor, first launched in 1987 as part of the Defense Meteorological Satellite Program (DMSP), scans the Earth’s surface conically with a constant ground-surface incidence angle of 53.1° and is equipped with channels operating at 19.35, 37.0 and 85.5 GHz with both vertical and horizontal polarization, and at 22.235 GHz with vertical polarization alone (Reference Hollinger, Lo and PoeHollinger and others, 1987). Among others, the Bootstrap (Reference Comiso, Grenfell, Lange, Lohanick, Moore, Wadhams and CarseyComiso and others, 1992) and the NASA Team (Reference CavalieriCavalieri and others, 1991) algorithms, which are based on the 19 and 37 GHz channels, are generally used for routine sea-ice concentration retrieval. However, both algorithms suffer from the sensor’s large footprint of 69 × 43 km2 (at 19 GHz) and of 37 × 28 km2 (37 GHz). The footprint at 85 GHz (15 × 13 km2) allows a much better spatial resolution. Reference Svendsen, Mätzler and GrenfellSvendsen and others (1987) were the first to use frequencies near 90 GHz for sea-ice concentration retrieval. Their results are based on the brightness-temperature polarization difference (BTPD) and have been validated for clear-sky winter (Reference Lomax, Lubin and WhritnerLomax and others, 1995) and overcast summer conditions (Reference Lubin, Garrity, Ramseier and WhritnerLubin and others, 1997).

We use the normalized BTPD (NBTPD) (also called polarization) at 85 GHz instead of the BTPD to minimize the influence of changing physical temperatures of the radiating portion of the sea ice. The other SSM/I channels are not used to retrieve the sea-ice concentration in this study.

Sea ice significantly reduces the exchange of sensible and latent heat as well as momentum between the polar ocean and atmosphere (Reference MaykutMaykut, 1978). Gaining an accurate knowledge of the associated surface fluxes is very important for modeling atmospheric dynamics and thermodynamics with numerical weather-prediction (NWP) models. Over sea ice, the quality of the modeled fluxes is largely determined by the sea-ice concentration, which can be calculated from space-borne remote-sensing data, as well as its thickness and roughness, quantities that are more difficult to estimate from these data. The spatial resolution of space-borne remote-sensing data covers several orders of magnitude, from ∼25 m for the synthetic aperture radar (SAR) to > 25 km for the SSM/I. Though SAR is an excellent source for detailed information on type and structure of a sea-ice cover (Reference Drinkwater and JeffriesDrinkwater, 1998), larger- (hemispherical-) scale geophysical parameters, such as sea-ice concentration obtained from SSM/I data, are more convenient for use in NWP and global circulation models.

The following sections will focus on the new algorithm and possible error sources. The effects of snow cover and intervening atmosphere, which vary both spatially and temporally, are discussed and accounted for by the use of temporal tie points and a radiative transfer model, respectively. The new algorithm is used to calculate daily sea-ice concentration data for the period 1992−98. Examples are compared with independent data for validation purposes.

2. Methods

The basis of the algorithm presented, henceforth referred to as SEA LION, is as follows. The brightness temperature tp, emitted at polarization p from a partly sea-ice-covered ocean area equal to unity, with the fractions of sea ice c and open water (1 −c), can be written as:

(1)

where the emissivities of open water and sea ice at polarization p are given by ϵpw and ϵpi, respectively, and tsw and tsi are the physical temperatures of the sea surface and of the radiating sea-ice portion, respectively. This approach follows that of Reference Svendsen, Mätzler and GrenfellSvendsen and others (1987) but omits multi-year ice since Antarctic sea ice primarily consists of first-year (FY) ice. The NBTPD, which is also used in the NASA Team algorithm, is defined as:

(2)

where tv and Th are the brightness temperatures at vertical and horizontal polarization, respectively. The NBTPD calculated from emissivities, which have been measured in situ at 90 GHz in the Weddell Sea (Reference Comiso, Grenfell, Lange, Lohanick, Moore, Wadhams and CarseyComiso and others, 1992), is small over sea ice but quite large over open water (Table 1). Inserting Equation (1) into Equation (2) yields:

(3)

where a is given by

(4)

Table 1. Surface emissivities ϵv and eh and their standard deviations av and ah measured in situ at 90 GHz in the weddell sea, antarctica (Reference Comiso, Grenfell, Lange, Lohanick, Moore, Wadhams and Carseycomiso and others, 1992), and the nbtpd p and corresponding standard deviations (p) as calculated from these measurements

The triplets pw, tvw, thw and pi tvi thi are the tie points of open water and sea ice, respectively (section 2.1), which we calculate prior to the retrieval of c (section 2.3). Since Equation (3) is valid for surface measurements only, the use of SSM/I data requires a consideration of the atmospheric effect (section 2.2).

2.1. Tie points

The wavelength at 85 GHz is small compared to the other SSM/I channels. This has two consequences. One is that sea ice becomes radiometrically opaque at a smaller thickness (Reference GrenfellGrenfell and others, 1998). Secondly, emission and, more important, scattering in the snow provide a significant contribution to the radiometric signal of snow-covered sea ice (Reference Grenfell, Comiso, Lange, Eicken and WensnahanGrenfell and others, 1994, Reference Grenfell1998) and may smooth different signals associated with different sea-ice types. However, the snow-cover properties (e.g. the liquid-water content, the grain-size and the density) exhibit a large spatial and temporal variability, and can change on daily to seasonal time-scales due to snow metamorphism and precipitation (Reference Garrity and CarseyGarrity, 1992; Reference Massom, Lytle, Worby and AllisonMassom and others, 1998; Reference Sturm, Morris, Massom and JeffriesSturm and others, 1998).

Taking this into account, the emissivities, which are given in Table 1 and which have been measured in the Weddell Sea in winter and spring only (Reference Comiso, Grenfell, Lange, Lohanick, Moore, Wadhams and CarseyComiso and others, 1992), are not representative for the sea-ice conditions of the entire Antarctic all the year round. Considering this lack of information and the highly varying snow properties, we decided to derive sea-ice tie points for each month separately as follows. Monthly averages of the NBTPD, PAVE, and the temporal variability (variance), a\, are calculated for each pixel using the U.S. National Snow and Ice Data Center (NSIDC) daily gridded SSM/I brightness temperatures. The averages of p (not shown) allow the separation of sea ice (PAVE < 0.05) from open water (PAVE > 0.12). However, over open water, PAVE is significantly smaller than p obtained from in situ measurements of ϵV and ϵh (Table 1), due to the monthly-averaged weather effect, which extends also into the sea-ice zone. The variances of p (not shown) increase towards the marginal ice zone (MIZ), i.e. pixels that are closer to the MIZ exhibit a larger temporal variability than pixels from the inner pack. This can be explained by an increasing variability of c and an increase of the direct (atmospheric water content) and indirect (precipitation, melt-freeze cycles, flooding) weather influences towards the MIZ. Over polynyas, PAVE is smaller and 4 is larger than in other pack-ice areas, since c and atmospheric water contents are more variable.

The sea-ice tie points Tvi, Thi and pi are calculated by averaging only values of Tv, th and p from areas where PAVE < 0.05 and a% < 25 × 10−6 p < 0.005; Table 1) for the given month. Thus, almost all pixels with c < 100% and with a highly variable polarization, due to either changing surface properties or weather conditions or both, are excluded. The high limit chosen for PAVE ensures that nilas (Table 1) is considered.

Open-water tie points Tvw, Thw and pw are also calculated for each month separately from NSIDC daily gridded SSM/I brightness temperatures within clear-sky areas, which have been extracted from daily maps of the integrated cloud liquid-water content l (l ≤15 gm−2 = clear sky) (Reference Karstens, Simmer and RuprechtKarstens and others, 1994; Reference HeygsterHeygster and others, 1996) and which simultaneously exhibit a surface wind speed v < 10 ms"1 and an integrated water-vapour content W<10kgm−2. Finally, the remaining direct weather effects on the tie points over open water of v and w, and over sea ice of w only, are quantified and subtracted (section 2.2).

2.2. Atmospheric effects

Compared to the Arctic, the atmospheric effect on SSM/I measurements is much more pronounced in the Antarctic sea-ice zone due to its proximity to the Southern Ocean, which is a substantial source of atmospheric heat and moisture and a site of significant cyclogenesis (Reference King and TurnerKing and Turner, 1997). The increase of the SSM/I brightness temperatures due to emission from atmospheric water, which mainly consists of w and l, is larger at 85 GHz than at the other SSM/I channels (Reference Ulaby, Moore and FungUlaby and others, 1981). Though this increase is less significant over most sea-ice-covered areas due to their high surface emissivities, it becomes larger with decreasing surface emissivities (e.g. over the MIZ: Reference FuhrhopFuhrhop and others, 1998). Figure 1 shows the increase of c due to w and l as calculated from uncorrected 85 GHz SSM/I brightness temperatures with Equation (3) for a calm sea surface (salinity: 34 ppt; surface temperature: 0°C; c= 0%). Over open water, c could increase to 90% or more. Over 100% of cold FY ice (Table 1), this increase (not shown) is still one-tenth of that over open water.

Fig. 1. Increase of cine to w and l as calculated with equation (3) from uncorrected 8.5 ghz ssm/i brightness temperatures over a calm sea surface with c = 0% and the tie points of sea ice and open water as given in the upper left corner.

In order to quantify the effect of w and l, we have modeled brightness temperatures with the radiative transfer model MWMOD (MicroWave MODel) (Reference FuhrhopFuhrhop and others, 1998) for emissivities of 0.44−0.98 in increments of 0.01 and typical values of w and l. For each emissivity, a two-dimensional polynomial fit yields a set of coefficients that allows the subtraction of the quantified atmospheric water effect from the SSM/I brightness temperatures if w, l and the emissivities are known. While w is taken from the operational NWP model of the European Centre for Medium-range Weather Forecasts (ECMWF) over both open ocean and sea ice, this is not possible for l. Over open water, l has been calculated with the method given by Reference Karstens, Simmer and RuprechtKarstens and others (1994), a method which cannot be applied over sea ice. Reference Miao, Johnsen, Kern, Heygster and KunziMiao and others (2000) developed what they call the R-factor method to identify regions where l > 100 g m−2. We used this method to successfully mask out about 75% of such regions over open water and sea ice. The remaining 25% with l > 100 g m−2 account for < 5% of the total area covered by the NSIDC grid used. Now we use daily sea-ice concentration maps, which are based on the NASA Team algorithm, as a mask to predefine water- and ice-covered areas. Over sea ice, we set l to its monthly-average value obtained from daily data of l within an approximately 100 km wide open-water area adjacent to the sea-ice edge. These averages vary between 40 and 80 g m−2 in winter and summer, respectively. This was done considering the high percentage of clouds covering the Southern Ocean (70% according to the results of the International Satellite Cloud Climatology Project (ISCCP)), and the lack of reliable data of l.

Oversea ice, the ratios ofTvi–dTvi(O2) orThi–dThi(O2) and the monthly-averaged ECMWF surface temperatures are taken as the emissivity. The quantities dTvi(O2) and dThi(02) denote brightness-temperatures contributions due to oxygen absorption as quantified with MWMOD. Over open water, the emissivities are altered by the surface wind (Reference Ulaby, Moore and FungUlaby and others, 1981). In particular, v > 10 ms"1 significantly decrease p at 85 GHz, again causing an overestimation of c if neglected. We used MWMOD to calculate sea-surface emissivities and to model the change of Tv and Th for typical wind speeds, which have also been taken from the ECMWF model. The emissivities are put into a look-up table. For a given v, this table provides the correct sea-surface emissivity required by the correction for the effect ofw and l on tv and th. The changes of tv and Th according to v are subtracted from the brightness temperatures after this correction by using a set of coefficients, obtained with a one-dimensional polynomial fit from the modeled brightness temperatures.

2.3. Algorithm

Equation (3) is the basis of the SEA LION algorithm, which iteratively calculates and minimizes the difference between p obtained from the uncorrected NSIDC daily gridded 85 GHz SSM/I brightness temperatures and p obtained from brightness temperatures that have been modeled for the given atmospheric conditions and the retrieved c. The first iteration step is to calculate a first-guess sea-ice concentration cfg using Equation (3) and uncorrected SSM/I data in all regions which have not been masked out using the r-factor method. In order to prove cfg, brightness temperatures are modeled with MWMOD according to the values of w, l, v and cfg. If p obtained from these modeled brightness temperatures differs from the measured one by < 0.001, which is about 1% change in c, then cfg is taken as the actual sea-ice concentration, and the iteration is stopped. Otherwise the weather effect given by w, l and v is quantified and subtracted from the SSM/I brightness temperatures. A first-order weather-corrected p is calculated from these corrected SSM/I brightness temperatures and is used to calculate a new first-guess sea-ice concentration, similar to the first iteration step. The new cfg is proved as described above and so on.

The iteration is continued until the difference between the modeled p, reflecting the same atmospheric conditions but different values of cfg at each iteration step, and p calculated from the uncorrected SSM/ I brightness temperatures is < 0.001. This is usually achieved within the first 10 iterations. Data gaps caused by masking out areas with a large l value are filled using temporal linear interpolation of c obtained for the day before and after the considered one. The accuracy of c would be close to 1% if determined only by the difference threshold used for the minimization, but the coarser spatial resolution of the involved atmospheric data and the variability of the sea-ice tie points due to varying sea-ice and snow properties limit the accuracy to about 10%.

3. Results

Figures 2 and 3 show monthly 85 GHz sea-ice tie points averaged over the years 1992−98. Clearly tvi and Thi are highly variable throughout the year. Freeze-up coincides well with a sharp increase of tvi and Thi in March/April (Fig. 2). The onset of melt is marked by a decrease of tvi and Thi from November to January. Remarkable are the low values of Thi of about 200 K in summer. This corresponds to a surface emissivity of about 0.7 and may be caused by old refrozen coarse-grained snow. The gradual increase of tvi and Thi between August and November is probably a result of a growing liquid-water fraction in the snow, which increases the emissivities at both polarizations (Reference Garrity and CarseyGarrity, 1992). Values of p, vary between 0.021 January) and 0.028 (October) and are fairly constant within one standard deviation, which has been calculated from the standard deviations of Tvi and Thi. Based on these sea-ice tie points, the SEA LION algorithm has been used to calculate daily Antarctic sea-ice concentrations C85 for 1992−98 from NSIDC daily gridded brightness temperatures. Hereafter, NASA Team sea-ice concentrations obtained with the extended weather correction (Reference HeygsterHeygster and others, 1996) are denoted by CNT, and Bootstrap sea-ice concentrations determined with seasonal coefficients by CBO.

Fig. 2. Monthly 85ghz sea-ice tie points tvi andthi averaged over the period 1992−98. the error bars denote one standard deviation. subscripts v and h refer to vertical and horizontal polarization, respectively.

Fig. 3. Monthly 85 ghz sea-ice tie point pi averaged over the period 1992−98. the error bars denote one standard deviation.

Figure 4 shows the beginning of the annual sea-ice decay in the Ross Sea in November 1997. More details can be identified in the images on the right side showing cs5 than in those on the left showing CNT. Figure 5 shows Operational Line-scan System (OLS) visible images of the Ross/Amundsen Sea region overlaid by the 15%, 60% and 90% isolines of c85 (Fig. 5a) and CNT (Fig. 5b), on 16 November 1996. About 16 original OLS pixels have been averaged for one OLS pixel (10 × 10 km2) shown here. The isolines are mapped onto the NSIDC 12.5 × 12.5 km2 grid. Several coastal polynyas and a decaying sea-ice cover typical for spring can be identified. The 15% isoline of CNT appears only in one polynya (Fig. 5b) whereas c85 is <15% in almost every polynya (Fig. 5a). This seems to be quite reasonable since, during November, solar radiation accompanied by rising air temperatures stops the sea-ice production in coastal polynyas and keeps them open (Reference Markus, Kottmeier, Fahrbach and JeffriesMarkus and others, 1998). First polynyas within the decaying pack ice (e.g. in the upper left (Amundsen Sea) and the lower right quarter (Ross Sea)) can be identified quite well by the 60% and sometimes even the 15% isoline of c85 (Fig. 5a). However, there is almost no correspondence between these areas and the 60% isoline of cnt (Fig. 5b).

Fig. 4. Evolution of the sea-ice cover in the ross sea, november 1997. left panels: CNT; right panels: cs5. sea-ice concentrations <15% have been set to zero.

Fig. 5. Ols visible images overlaid with ssm/i sea-ice concentration isolines of 16 november 1996: (a) c85, and(b) CNT. the south pole is at the upper right corner, and the right (top) image borders are along 180° w (90° w).

Sea-ice concentrations derived from in situ observations from the bridge of the research vessel .nathaniel b. palmer in 1994−98 (CSHIP) are compared to SSM/I sea-ice concentrations CSSMI, which have been mapped onto the NSIDC 25 × 25 km2 grid and rounded to the nearest tenth. Linear correlation coefficients for CSSMI and CSHIP He between 0.528 (C85) and 0.576 (CB0). Average differences CSSMI – CSHIP amount to –5% (CBO), −10% (C85) and −18% (CNT). The most convincing linear regression is obtained with C85: C85 = 17.0 + 0.7 × CSHIP

Daily sea-ice areas and extents are calculated from CSSMI for 1992−98 (not shown). The annual and interannual variability of SEA LION sea-ice areas and extents shows reasonable agreement with those obtained from cbo and CNT, but for the SEA LION study period sea-ice extents and areas fall below those obtained from cbO. The extent difference is 0.5 × 106 to 1.5 × 106 km2. The areal difference is up to 0.5 × 106 km2 in summer and 0.5 × 106 to 1.5 × 106 km2 in winter. SEA LION sea-ice extents also fall below NASA Team sea-ice extents (by 106km2 in summer and 0.5 × 106 km2 in winter). SEA LION sea-ice areas are similar to NASA Team sea-ice areas in summer but exceed the latter by 0.5 × 106 to 1.5 × 106 km2 in winter.

4. Discussion and Conclusions

The SEA LION algorithm uses the SSM/I 85 GHz polarization together with monthly sea-ice and open-water tie points, and a weather correction scheme which is based on radiative transfer modeling. It provides daily Antarctic sea-ice concentration data with a spatial resolution of 12.5 × 12.5 km2 Daily sea-ice extents and areas obtained from sea-ice concentration data obtained with the NASA Team (CNT), the Bootstrap (CBO) and the SEA LION (C85) algorithm show reasonable agreement concerning the annual and interannual variability. However, in the regions studied, SEA LION generally provides a smaller sea-ice extent (by about 106 km2) throughout the year. The SEA LION sea-ice areas are similar to those obtained from sea-ice concentrations computed with the NASA Team and Bootstrap algorithms in summer but lie between them in winter.

A comparison of OLS visible images with coincident isolines of c85 and CNT shows convincing results using SEA LION, particularly concerning the areal coverage by polynyas. Sea-ice concentration gradients provided by SEA LION are better resolved than those derived from the NASA Team algorithm. This may explain why the NASA Team or Bootstrap sea-ice extents are larger than the SEA LION sea-ice extent (more open-water pixels within the pack ice), while the NASA Team and Bootstrap sea-ice areas remain similar to the SEA LION sea-ice area (smaller net amount of sea-ice covered pixels, but higher average sea-ice concentration). We have also compared c85 with 850 ship observations of the sea-ice concentration for 1994−98. The results agree with those obtained with the NASA Team and Bootstrap algorithms.

The quality of the NWP model data, which are used for the correction of the atmospheric effect, and in particular of the cloud liquid-water content l, has a large impact on the retrieval of c85 and may cause significant errors. Further improvements of the atmospheric input parameters are necessary. Monthly sea-ice tie points seem to account for the annual evolution of the microwave signal of almost the entire Antarctic sea-ice cover, but smooth out regional variations. Dividing the ice pack into zones depending on the state of snow metamorphism, and estimating regional sea-ice tie points may be one step towards accounting for these variations. This area requires further research, especially in the light of Figure 2 and when considering the findings of Reference Drinkwater and JeffriesDrink-water (1998) concerning the seasonal backscatter variability of Antarctic sea ice.

The upcoming Advanced Microwave Scanning Radiometer (AMSR) on board the Adeos II and EOS Terra satellites operates at frequencies close to those of the SSM/ I and is additionally equipped with 6 GHz channels. Due to technological advances, its data will allow an improvement of current sea-ice concentration products, mainly due to the enhanced spatial resolution of about 10 × 10 km2, and probably in a more economical way than the method presented here, since a simple weather correction seems to be sufficient. The AMSR will provide 89 GHz data with a higher spatial resolution than the low AMSR frequencies. These data can be used to derive higher-resolved and thus more realistic sea-ice concentrations, at least under clear-sky conditions.

Acknowledgements

This work was supported by the European Union, Brussels, Belgium, under contract ENV4−CT97−0415. SSM/I data were provided by the Earth Observing System Distributed Active Archive Center at the NSIDC, University of Colorado, kern and heygster: sea-ice concentration retrievalin the antarctic Boulder. Other data were provided by the Antarctic Meteorology Research Center (Space Science Engineering Center, Madison, WI), the ECMWF (Reading, U.K) and Deutsches Klimarechenzentrum (Hamburg, Germany). We would like to thank C. Haas, C. Garrity and R. Ramseier for useful discussions. The help of two anonymous referees and the scientific editor R. Massom is gratefully acknowledged.

References

Cavalieri, D.J. and 6 others. 1991. Aircraft active and passive microwave validation of sea ice concentration from the Defense Meteorological Satellite program special sensor microwave imager. J. Geophys. Res., 96(C12), 21,989−22,008.Google Scholar
Comiso, J. C., Grenfell, T. C., Lange, M., Lohanick, A.W., Moore, R. K. and Wadhams, P.. 1992. Microwave remote sensing of the Southern Ocean ice cover. Passive microwave signatures of sea ice. In Carsey, ED. and 7 others, eds. Microwave remote sensing of sea tee. Washington, DC, American Geophysical Union, 243−259. (Geophysical Monograph Series 68.)Google Scholar
Drinkwater, M. R. 1998. Active microwave remote sensing observations of Weddell Sea ice. In Jeffries, M. O., ed. Antarctic sea ice: physical processes, interactions and variability Washington, DC, American Geophysical Union, 187−212. (Antarctic Research Series 74.)Google Scholar
Fuhrhop, R. and 6 others. 1998. A combined radiative transfer model for sea ice, open ocean, and atmosphere. Radio Sci., 33(2), 303−316.Google Scholar
Garrity, K. 1992. Characterization of snow on floating ice and case studies of brightness temperature change during the onset of melt. In Carsey, F. D. and 7 others, eds. Microwave remote sensing of sea ice. Washington, DC, American Geophysical Union, 313−328. (Geophysical Monograph Series 68.)Google Scholar
Grenfell, T. C., Comiso, J. C., Lange, M. A., Eicken, H. and Wensnahan, M. R.. 1994. Passive microwave observations of the Weddell Sea during austral winter and early spring. J. Geophys. Res., 99(C5), 9995−10,010.Google Scholar
Grenfell, T. C. and 11 others. 1998. Evolution of electromagnetic signatures of sea ice from initial formation to the establishment of thick first-year ice. IEEE Trans. Geosci. Remote Sensing, GE-36(5), 1642−1654.CrossRefGoogle Scholar
Heygster, G. and 12 others. 1996. PELICON – Project for Estimation of Long-term variability in Ice Concentration. Final report. Bremen, University of Bremen. European Communities Environment Programme. (Contract EV5V-CT93−0268.)Google Scholar
Hollinger, J. P., Lo, R. and Poe, G.. 1987. Special sensor microwave/imager user’s guide. Washington, DC, U. S. Naval Research Laboratory.Google Scholar
Karstens, U., Simmer, C. and Ruprecht, E.. 1994. Remote sensing of cloud liquid water. Meteorol. Atmos. Phys., 54,157−171 Google Scholar
King, J. C. and Turner, J.. 1997. Antarctic meteorology and climatology. Cambridge, Cambridge University Press.Google Scholar
Lomax, A. S., Lubin, D. and Whritner, R. H.. 1995. The potential for interpreting total and multiyear-ice concentrations in SSM/ I 85.5 GHz imagery. Remote Sensing Environ., 54(1), 13−26.CrossRefGoogle Scholar
Lubin, D., Garrity, C., Ramseier, R. O. and Whritner, R. H.. 1997. Total sea ice concentration retrieval from the SSM/I 85.5 GHz channels during the Arctic summer. Remote Sensing Environ., 62(1), 63−76.Google Scholar
Markus, T, Kottmeier, C. and Fahrbach, E.. 1998. Ice formation in coastal polynyas in the Weddell Sea and their impact on oceanic salinity. In Jeffries, M. O., ed. Antarctic seaice: physical processes, interactions and variability Washington, DC, American Geophysical Union, 273−292. (Antarctic Research Series 74.)Google Scholar
Massom, R. A., Lytle, V. I., Worby, A. P. and Allison, I.. 1998. Winter snow cover variability on East Antarctic sea ice. J. Geophys. Res., 103(C11), 24,837−24,855.Google Scholar
Maykut, G A. 1978. Energy exchange over young sea ice in the central Arctic. J. Geophys. Res., 83(C7), 3646−3658.Google Scholar
Miao, J., Johnsen, K.-E, Kern, S., Heygster, G. and Kunzi, K.. 2000. Signature of clouds over Antarctic sea ice detected by the Special Sensor Microwave/Imager (SSM/I). IEEE Trans. Geosci. Remote Sensing, GE-38(5), 2333−2345.Google Scholar
Sturm, M., Morris, K. and Massom, R.. 1998. The winter snow cover of the West Antarctic pack ice: its spatial and temporal variability, In Jeffries, M. O., ed. Antarctic sea ice: physical processes, interactions and variability Washington, DC, American Geophysical Union, 1−18. (Antarctic Research Series 74.)Google Scholar
Svendsen, E., Mätzler, C. and Grenfell, T. C.. 1987. A model for retrieving total sea ice concentration from a spaceborne dual-polarized passive microwave instrument operating near 90 GHz. Int. J. Remote Sensing, 8(10), 1479−1487.Google Scholar
Ulaby, F. T., Moore, R. K. and Fung, A. K.. 1981. Microwave remote sensing, active and passive. Vol. 1. Fundamentals and radiometry Reading, MA, Addison-Wesley Publishing Co.Google Scholar
Figure 0

Table 1. Surface emissivities ϵv and eh and their standard deviations av and ah measured in situ at 90 GHz in the weddell sea, antarctica (comiso and others, 1992), and the nbtpd p and corresponding standard deviations (p) as calculated from these measurements

Figure 1

Fig. 1. Increase of cine to w and l as calculated with equation (3) from uncorrected 8.5 ghz ssm/i brightness temperatures over a calm sea surface with c = 0% and the tie points of sea ice and open water as given in the upper left corner.

Figure 2

Fig. 2. Monthly 85ghz sea-ice tie points tvi andthi averaged over the period 1992−98. the error bars denote one standard deviation. subscripts v and h refer to vertical and horizontal polarization, respectively.

Figure 3

Fig. 3. Monthly 85 ghz sea-ice tie point pi averaged over the period 1992−98. the error bars denote one standard deviation.

Figure 4

Fig. 4. Evolution of the sea-ice cover in the ross sea, november 1997. left panels: CNT; right panels: cs5. sea-ice concentrations <15% have been set to zero.

Figure 5

Fig. 5. Ols visible images overlaid with ssm/i sea-ice concentration isolines of 16 november 1996: (a) c85, and(b) CNT. the south pole is at the upper right corner, and the right (top) image borders are along 180° w (90° w).