Introduction

The Weddell Sea is a source of sea ice that flows northward into the path of the easterly Antarctic Circumpolar Current. The broad features of pack-ice movement and of the Weddell Sea circulation have been discussed by Reference CarmackCarmack (1986) but details of both the pack-ice motion in winter and of the effects of large-scale weather systems crossing the Weddell Sea are missing.

In this study we are concerned with that area of the Weddell Sea bounded by the great circles joining the tip of the Antarctic Peninsula (63.3 °S, 55.0°W) to a point 65°S, 30°W and thence to Kapp Norvegia (71°S, 12°W), with an area of 2 x 10⁶ km². The central Weddell Sea is usually devoid of direct meteorological observations and conventional synoptic analyses can be suspect. Numerical analyses do, however, incorporate satellite-sounder data which improve analysis quality, although errors do still occur. Further improvement can be made by inserting ice-strengthened buoys into the pack ice to transmit meteorological data via polar-orbiting satellites. WWSP 86 provided an opportunity to use co-ordinated buoy deployment to improve the meteorological analyses and to determine the influence of atmospheric and oceanic forcing on sea-ice motion. These are also discussed in companion papers (Reference Rowe, Sear, Morrison, Wadhams, Limbert and CraneRowe and others, 1989; Reference Wadhams, Sear, Crane, Rowe, Morrison and LimbertWadhams and others, 1989). In this paper, we concentrate on gross ice motion over periods of days to months.

Buoy 0534 (WMO number 71513) was deployed on 29 December 1985 at 76°S, 33°W, and sensor data and position fixes were processed by System Argos from 2 January 1986. From mid-March 1986, the data were circulated via the WMO Global Telecommunication System (GTS) to the meteorological analysis centres at the U.K. Meteorological Office and the European Centre for Medium Range Weather Forecasting. The buoy ceased transmitting on 24 April 1987 at 62°S, 43°W.

Long-Term Drift Pattern

The drift of buoy 0534 is shown in Figure 1. Initially, buoy 0534 paralleled the line of the coast and the ice shelf between 76°S, 33°W and 77°S, 46°W before turning and travelling north to reach 74°S by June (E and F on Figure 1). It took a further 2 months to reach 71°S (G) and from there reached 69 °S (H) in October. The buoy remained at this latitude until November before moving north again to reach 68°S in December. It stayed close to 66°S from January to March 1988 (J) before moving off rapidly north-east. The letters A to J in Figure 1 identify short-term, apparently anomalous deviations from the overall drift pattern. These fluctuations can be associated with specific weather sequences. Although the net distance covered during the identified oscillations and loops was often small, the actual distance travelled on a day-to-day basis was often quite large (Table I). The ratio of total distance travelled to the displacement of the buoy is called the “meander coefficient” (see Table I) after Reference Dunbar and WittmanDunbar and Whitman (1962). We note that the meander coefficient must be defined with respect to the overall lime between the beginning and end of the observations, and the time step (t) between successive measurements of position. In this paper t = 1 d but a shorter time step which revealed finer details of the buoy’s motion would clearly give a larger meander coefficient for the same overall time.

Buoy Motion and Weather Systems

Eight periods have been selected to illustrate the influence of large-scale weather systems on the pack-ice motion evidenced by buoy 0534. The first three A, B, and C all show that well-defined looping occurred during the summer when there was more or less open pack (5/10 to 7/10 ice cover). Thereafter, satellite imagery indicated greater concentrations (9/10 to 10/10). Surface-pressure and temperature data measured by the buoy became available to the analysis centres during period C and a brief description of the synoptic meteorology for that and five subsequent periods follows.

Fig. 1.

Tracks of buoy 0534 (from 2 January 1986 to 31 March 1987), of Endurance (1915-16), and Deutschland (1912) (after Reference AckleyAckley, 1979). The letters indicate periods of buoy movement discussed in the text and in Table I.

Pack-Ice Motion and the Geostrophic Wind

Quantifying pack-ice motion in response to wind forcing is possible even though the anemometer on the buoy proved inadequate as a guide to wind stress. The geostrophic wind was derived from the U.K. Meteorological Office numerical surface-pressure analyses for 00.00 GMT. The wind speed and direction so obtained are equivalent to the wind between 600 and 1000 m above the surface. The 10 m wind speed is approximately 20% less than this and is deflected by about 20 to the right depending on surface roughness. Reference BrenneckeBrennecke (1921) showed that on average the pack ice moved 34° to the left of the surface wind. Thus, the pack-ice movement is within 15° of the geostrophic wind and is indistinguishable from directions determined from the isobars of surface pressure. The daily mean displacement of the buoy was determined by successive daily positions closest to 12.00 GMT. Although not ideal, the 00.00 GMT analyses do represent the average atmospheric pattern over the 24 h between successive location fixes, even for mobile weather systems. Only the first 120 d of charts have been analysed and until mid-March no buoy data were included in the numerical analyses. The results are summarized in Table II. The surface wind speeds are 80% of the estimated geostrophic wind.

The term “gyre” is used to indicate the overall long-term current system of the Weddell Sea. The pack-ice motion was assigned as moving predominantly in one of eight compass directions. Winds assisting this motion could be within 60° of the direction of motion. Sea-ice velocity can only be accurately determined when the meander coefficient is close to 1 or when the sampling rate is less than 3 h (Reference ThorndikeThorndike, 1986). However, this study is restricted to net displacements of the sea ice on time-scales of days to months. A study of ice motion with a finer time-scale resolution is given in Rowe and others (1989).

Movement has been grouped into five categories for four 30 d periods:

1. With the gyre and with the local wind.

2. With the gyre and against the local wind.

3. Against the gyre and with the local wind or with variable winds.

4. Against the gyre and against the local wind.

5. With the gyre and with light or variable winds.

Comparison of these five categories (Table II) shows that the motion in the direction of the main current assisted by the wind (category 1) is the dominant feature and is about 30% greater than when the wind and gyre are opposed (category 3). However, some of the wind speeds in category 3 are low, particularly in the first 30 d after deployment. Internal pack-ice resistance may have determined local-scale ice motion in some cases. Certainly, most of category 3 motion and categories 2 and 4 occurred during relaxation of the pack following prolonged periods of category 1 motion which compacted the pack ice. There may also be some categorization errors because, until mid-March, the synoptic analyses did not contain the buoy’s data. These errors would be associated with slack pressure gradients and low wind speeds. Truly calm conditions or light variable winds allow the pack to respond to the gyre only and category 5 is thus a good estimate of the gyre/coastal current in the southern Weddell Sea, which here corresponds to a mean displacement of 4.3 ± 1.7 kmd⁻¹.

Annual Pack-Ice Drift

The overall pack-ice movement is made clearer if the westward and northward components of the buoy’s motion are presented separately (Fig. 3). The movement west after the first 100 d is in a series of discrete steps, as a consequence of the changing meteorological regimes described earlier. Considering the length scales of the atmospheric systems in the region (see, for example, Reference ThorndikeThorndike (1986)), a shift north of 5-10 must affect a large proportion of the central Weddell Sea.

The mean northward drift of buoy 0534 is revealed by a straight-line fit by regression on Figure 3a between Day 100 1986 and Day 59 1987. The regression equation is:

LATITUDE = −80.42 + (0.0372 x DAY)

where DAY is in the range 100-425. The standard error of the intercept is 0.0738 and that of the slope is 2.65 x 10⁻⁴.

Fig. 3.

Components of translation of buoy 0534 from 2 January 1986 to 28 February 1987. Northward translation is scaled by latitude and eastward translation is in kilometres.

This shows that the 95% confidence limits about the slope are small and the northward drift is close to that given by the regression-line slope. The standard error of the regression is 0.449. The overall northward movement is 4.12 ± 0.03 km d⁻¹. That net speed is similar to the value determined for category 5 when the buoy was farther south, close to the ice shelf.

Northward motion, after near-stationary or westerly movement, ranged from 4.3 to 6.1 kmd⁻¹ over periods of 30-80 d. During Days 208-214, the pack moved north at the much higher rate of 14-35 kmd⁻¹ as compression was released after prolonged stable-blocking high pressure was replaced by a strong cyclonic circulation. Therefore, when comparing long-term drift from different years, periods in excess of 80 d are to be preferred. The calculation of the large-scale features of pack-ice motion over long periods may, because of significantly large meandering, conceal a truer general speed of movement, just as a displacement over a few days conceals the daily meandering and hourly velocities (Reference ThorndikeThorndike, 1986). It is, however, this broad-scale motion which determines the net transport of sea ice around the Weddell Sea and its export into the Southern Ocean. For example, buoy 0534 in the two periods between day numbers 86/05 and 86/185 had total displacements averaging 4.07 and 3.49 km d \ whereas an average of the 10 d displacements during the same periods gave speeds of 5.0 and 4.9 km d⁻¹ (Table III). Which of these is the truer representation of the pack-ice gross speed? 5.0 km d⁻¹ for the first period makes due allowance for the change in direction of the inshore current as it followed the ice front and was about half the value of the daily displacements A, B, and C shown in Table I. During the second period, the buoy followed a snake-like course which reduced the net displacement. However, the consistency of 4.9 km d⁻¹ with the previous period suggests that this is a reasonable estimate of the mean wind-assisted movement.

Table 3

Net displacements, in kmd⁻¹, for buoys deployed during wwsp 86

The subsequent predominantly northward net displacement represented the best estimate of mean motion of the underlying ocean current. Averaged over the whole 220 d, from 4 July 1986 to 9 February 1987, the net speed of movement is 4.43 km d⁻¹. As already demonstrated, the total distance travelled by an individual floe is much greater than its displacement during a specific period. The displacement of buoy 0534 between day numbers 86/75 and 87/59 was 1500 km but the total distance travelled was 3395 km. This gives an overall meander coefficient of 2.26, about half of that of the shorter, more active periods shown in Table I.

Spatial and Secular Variation of Pack-Ice Motion

When the drift of buoy 0534 is compared with those of the ice-beset ships Endurance and Deutschland (Fig. 1), there is a remarkable similarity with the track of Endurance from July onwards. Further information is obtained from a number of transmitters dropped on to the pack ice in December 1978 and February 1980 in the south-western Weddell Sea (Reference AckleyAckley, 1981) and from a series of buoys deployed in the north-eastern Weddell Sea as part of WWSP 86 (Reference Hoeber and Gube–LenhardHoeber and Gube-Lenhardt, 1987; Reference Schnack-SchielSchnack-Schiel, 1987).

The gross features of the WWSP 86 drifting buoys are shown in Figure 4 and a summary of their net displacements is given in Table III. The northward movement of these buoys, Ackley’s buoys, and the two beset ships are tabulated in Table IV, which divides the area into 5° longitude bands between 35° and 60 W.

The buoys deployed in the north-eastern Weddell Sea were grouped in threes which moved in concert. Each group exhibited three distinct phases. The northern group moved west for 80 d and then west-south-west with a large anti-clockwise loop which reduced the net displacement; the group then turned north. The southern group followed the coastal current south-west for 100 d then, after performing a smaller anti-clockwise loop in concert with the northern group, it moved west-north-west before it too turned north for the last 100 d. This latter group closely followed the 2000 m bathymetric contour until reaching 45°W. Both buoy 0534 and Endurance followed this contour after leaving the continental shelf at 72.5°S, 50°W. Their motions would suggest a measure of steering by the under-water topography.

Fig. 4.

Schematic showing the gross features of the tracks of the buoys listed in Table III, 1986-87.

Table 4

Available data for the northward drift of pack ice in the weddell sea using buoy, air-dropped transmitter, and beset ship tracks. the data are grouped into four equal longitudinal zones. the track of endurance, which moved between two different zones, is summarized in each zone it occupied

None of the buoys deployed north of 75°S penetrated south of that latitude. Buoy 0534, placed south of 75°S, followed the coastline until it turned north near Berkner Island. Apart from the greater resistance to westward progress caused by the quasi-permanent sea ice closer to the Antarctic Peninsula, there may be an outflow of water northward from beneath the Filchner Ice Shelf as part of a cyclonic circulation beneath the ice shelf (Reference Carmack and FosterCarmack and Foster, 1975).

The track of one other buoy (74512) is also shown in Figure 4 to illustrate the divergence which takes place near 68°S, 12°W. The timing of the southernmost point of this buoy on 30 January 1987 coincides with the southernmost positions of the other two groups of buoys. This south-west motion over 20 d extended over the whole pack-ice area between 67°S, 10°W and 75°S, 34°w, and represents the effect on the summer pack of an area of low pressure centred over the Weddell Sea. The subsequent looping occurred as high pressure developed and the circumpolar trough moved north to 65°S.

Comparing the northward motion of pack ice in the central and western Weddell Sea (Table III) and noting the discharge rates from the different longitude bands, we obtain the following relationships:

in 1912 A >C, in 1915 C >D, in 1979 C ≪ E, in 1980 B = C, in 1987 A ≫ C.

From this we can deduce that the rates of discharge from the areas are in the following order:

A<B ≃ C > D > E

so that the fastest discharge lies between 40° and 50 W. Assuming that in most years Β - C, we can deduce the following:

in band A, 1912 > 1987;

in B and C combined, 1912 > 1987 > 1980 > 1979.

The track of Endurance crossed from C to D but may still be compared to the track of buoy 0534 and shows that the discharge rate for 1915 ≃ discharge rate for 1986. That discharge rate lies between those for 1979 and 1980; thus the ranking of discharge rates for each year is:

1912 > 1987 > 1980 > 1915 ≃ 1986 > 1979.

From satellite imagery, we know that pack ice extended north of South Georgia (54°S, 35 °W) in both 1980 and 1987 — both years of high discharge rate — and was much farther south in 1979 and 1986 with slower discharge rates. It is tempting to relate the ice-discharge rates to mean annual temperature in the South Orkney Islands; Reference BuddBudd (1975) has shown that there is about a 2°northward extension of pack ice associated with an annual temperature reduction of 1°C at that latitude. However, both 1912 and 1915 were cold years in the South Orkney Islands (Reference Jones and LimbertJones and Limbert, 1987), although the discharge rates were widely different.

It is to be expected that when low pressure is concentrated in the eastern Weddell Sea within 5° latitude of 70°S, and between 20° and 35°W, the ice-discharge rates are likely to be higher than average, with cold southerly winds predominating at the South Orkney Islands. Such situations occur from time to time when high pressure extends south from South America and blocking occurs near the northern half of the Antarctic Peninsula. Such blocking in the region is common (Reference Trenberth and MoTrenberth and Mo, 1985). Conversely, frequent passage of depressions across the northern tip of the Antarctic Peninsula into the Scotia Sea between 60° and 65°S, concomitant with high pressure extending north into the Weddell Sea, will retard ice movement northward. Such situations developed from time to time in 1986, but in early 1987 blocking anti-cyclones close to the tip of the Antarctic Peninsula were more common. This suggests that 1987 began with relatively high discharge out of the Weddell Sea with cold southerly winds.

Weddell Sea Sea-Ice Budget

The term budget is used here to describe the changes of the pack-ice area. The loss of ice cover from an area is a visible sign that there is a mass loss by under-water melting throughout the whole pack. Correspondingly, the increase in area indicates pack-ice thickening. From the hull-temperature record of buoy 0534, we know that growth was occurring from March through to October and that under-water melting occurs in the northern Weddell Sea during the 4 months November-February. This was similar to results found by Reference LimbertLimbert (1968) at Halley Bay (75°S, 26°W).

The equation of areal continuity for the Weddell Sea could be written as:

Change in Ice Area = Areal Outflow – Areal Inflow – Local Formation.

In this paper we have presented data on outflow (discharge speed). The areal extent of sea ice is known from satellite imagery and ice influx from the east can be estimated and so, if the width of the front through which northward discharge occurs is known, an estimate of the summer melt can be made.

Additional WWSP 86 buoys were placed close to 67°S, 0°W (Reference Schnack-SchielSchnack-Schiel, 1987), and drifted northward and then turned east. Their divergence from those placed further west implies that there is little net inflow or outflow near 65 °S, 30 W. In 1986 and early 1987, the mean northward discharge speed, ignoring the slow-moving boundary region D, was 5.0 kmd⁻¹ between 35° and 50°W at 65°S. The areal discharge rate of sea ice northward is thus of the order of 5000 km² d⁻¹. This represents a total outflow of 1.8 x 10⁶ km² year⁻¹, which represents approximately 90% of the defined area of the Weddell Sea. If this were to be balanced by an influx of 8.0 km d⁻¹ over the 625 km front extending from Kapp Norvegia to 67.5°S, 25°W, then the equation of continuity would be satisfied. However, from limited information available which suggests divergence north of 70°S at 20°W, the influx of ice is possibly less than the total required for balance.

Assuming an annual discharge of 90% of the defined area at a constant rate, the seasonal budgets are as follows:

Winter (March-October, 8 months)

(a) 60% of the Weddell Sea sea-ice is discharged into the Southern Ocean.

(b) This must be replaced by formation of new ice in polynyas and by influx from the east, south of 70°S and near Kapp Norvegia.

Summer (November-February, 4 months)

• a. 30% of the preceding winter sea ice is discharged.

• b. 40% of the area remains covered (Reference Zwally, Comiso, Parkinson, Campbell, Carsey and GloersenZwally and others, 1983); and so

• c. 30% of the sea ice melts in situ.

The analysis is consistent with the evidence from satellite passive microwave observations for the 1973-76 period as presented by Reference Zwally, Comiso, Parkinson, Campbell, Carsey and GloersenZwally and others (1983). They generated maps of monthly changes in sea-ice concentration which show the major areas of melt within the Weddell Sea.

Conclusions

The movements of satellite-tracked buoys and of vessels beset in ice have been analysed to study the sea-ice motion of the Weddell Sea. It is clear that the major vacillations of the sea-ice field in the region result from the movement of the atmospheric circumpolar trough, through its influence on the motions of individual weather systems. We have demonstrated the type of influence of these weather systems in the region. The spatial and temporal consistency of the northward motion in the central and western Weddell Sea indicates the strength of the underlying northward current which forms a part of the Weddell Sea circulation. There are notable year-to-year variations in the overall northward translation. In 1986, in the western sector, it was 4.2 km d⁻¹(Fig. 2) but the mean value for the various years is closer to 5.0 kmd⁻¹. Using all available ice-drifter data, it has been possible to determine a sea-ice budget for the Weddell Sea. It has been found that each winter approximately 60% of the Weddell Sea sea ice is moved northward into the Southern Ocean, and that this amount is replaced by influx from the east and by the formation of new ice in leads and coastal polynyas. In summer, about 30% of the previous winter's ice load is discharged to the north, a further 30% melts in situ, leaving the remaining 40% covered by ice. In the light of this areal budget, it would seem unlikely that any mobile pack ice in the Weddell Sea is more than 2 years old, and that the majority is first-year ice.

Future Work.

Three main questions arise. How consistent is the Endurance/0534 track, what area of pack ice moves simultaneously under the influence of atmospheric forcing, and how real is the apparent quasi-stationary zone at 65°S between 10° and 30°W? To answer the first, a regular annual deployment of a buoy at 76°S between 30° and 35°W is in progress. To answer the second requires an array of buoys between 70° and 76°S, and between 30° and 35°W. The third question could best be answered by a line of buoys from 65° to 70°S between 15° and 25°W. It would be difficult to implement the latter two deployments without the aid of aircraft.

Acknowledgements

We are very grateful for the material help and collaboration given throughout WWSP 86 by V. Squire, then at Scott Polar Research Institute, Cambridge, and to E. Augstein of the Alfred-Wegener-Institut für Polar- und Meeresforschung, Bremerhaven. Especially, thanks are due to the latter scientist and to H. Hoeber of Meteorologisches Institut der Universität Hamburg, and to C. Kottmeir of the Institut für Meteorologie und Klimatologie der Universität Hannover, for allowing us to use the position data for the buoys deployed from FS Polarstern during WWSP 86. The U.K. Meteorological Office numerical analyses provided a vital source of information. The analysis charts are published by permission of the Director General of the U.K. Meteorological Office.