Positive Mass Balance of the Ross Ice Streams, West Antarctica

See allHide authors and affiliations

Science  18 Jan 2002:
Vol. 295, Issue 5554, pp. 476-480
DOI: 10.1126/science.1066875


We have used ice-flow velocity measurements from synthetic aperture radar to reassess the mass balance of the Ross Ice Streams, West Antarctica. We find strong evidence for ice-sheet growth (+26.8 gigatons per year), in contrast to earlier estimates indicating a mass deficit (−20.9 gigatons per year). Average thickening is equal to ∼25% of the accumulation rate, with most of this growth occurring on Ice Stream C. Whillans Ice Stream, which was thought to have a significantly negative mass balance, is close to balance, reflecting its continuing slowdown. The overall positive mass balance may signal an end to the Holocene retreat of these ice streams.

Over the past several decades there has been concern that the marine-based West Antarctic Ice Sheet might collapse within the next several centuries, raising sea level by 5 to 6 m (1). Underlain by a thick layer of marine sediments, this ice sheet has exhibited considerable change in flow over the last millennium, particularly along the Siple Coast in the Ross Sea Sector (2), and since the last glacial maximum, the grounding line (the point where the ice sheet loses contact with its bed and begins to float) has retreated nearly 1300 km along the western side of the Ross Embayment. The chronology established for this retreat suggests mean grounding-line migration rates of 120 m/year (3). Although other processes might intervene, extrapolation of these rates has been used to predict a 4000-year lifetime for the West Antarctic Ice Sheet (4). Although this is a much longer period than earlier estimates (5) that predict a collapse over a few centuries, it does imply a sea-level rise of 12.5 to 15 cm per century.

Hypotheses of continued grounding-line retreat and possible collapse have been supported by an estimate of −20.9 ± 13.7 Gton/year (6, 7) for the mass balance of the Ross Ice Streams (A to F) (Fig. 1). This negative value by Shabtaie and Bentley (S&B) (6) implies that ice discharge (loss) exceeds accumulation by ∼25%, causing the ice sheet to thin and the grounding line to retreat. Much of this imbalance has been attributed to Whillans Ice Stream (formerly known as Ice Stream B), but negative imbalances were also found for the other Ross Ice Streams, with the exception of Ice Stream C, which stagnated 150 years ago (8). More recent analyses (9) based on similar data have estimated comparable negative imbalances on Whillans Ice Stream.

Figure 1

Ice-flow velocity (colors) over radar imagery from the RADARSAT Antarctic Mapping Project Mosaic (41). Flow velocity at 100 m/year intervals is contoured with thin black lines. White vectors show subsampled velocity vectors in fast-moving areas. Catchment boundaries (12) for individual ice streams are plotted with thick black lines. Flux gates used in discharge calculations are shown with red lines. Green lines show the locations of the S&B flux gates (6), and the light blue line shows the outline of the S&B catchment. The white box on the inset map shows the location of the study area.

The ice-discharge estimates in earlier studies relied on relatively sparse in situ measurements of ice-flow velocity (10). For some ice streams, the S&B discharge estimates were based on only one or two velocity measurements (6). We used spatially dense estimates of ice-flow velocity (Fig. 1) afforded by Interferometric Synthetic Aperture Radar (InSAR) (11) to reassess the mass balance of the Ross Ice Streams.

We began by determining the catchments for individual ice streams (12). As in earlier studies, we then estimated the ice-discharge flux through a “gate” located near the grounding line at the downstream end of each catchment (13). The total accumulation for each catchment area above its respective gate was determined by integrating gridded accumulation data (14). The difference between the accumulation (input) and discharge (output) gives the mass balance for the catchment. The results with uncertainties are tabulated in Table 1.

Table 1

Discharge and accumulation fluxes for the Ross Ice Streams. The summed total excludes the first three entries because the combined flux through these gates is included by the wider gate farther downstream (see Fig. 1). All accumulation numbers for individual ice streams are based on the average of two accumulation maps (14), except where noted. The last row shows the S&B results for their catchment shown in Fig. 1.

View this table:

Our results indicate that, contrary to earlier estimates, the mass balance of this sector of the West Antarctic Ice Sheet is positive (26.8 ± 14.9 Gton/year). Ice Streams D and E and the combined outflow of Ice Stream A and Whillans Ice Stream are in balance to within the 2σ limits of the uncertainty. This is consistent with radar altimetry observations that show little thickening or thinning over the catchments of Ice Streams D and E (15). Ice stream F is significantly positive, but its contribution to the overall total is small relative to that of the other ice streams. Stagnant Ice Stream C has a strongly positive mass balance because of its negligible outflow, and it is the major contributor to the overall positive mass balance for the region. Thus, the positive imbalance is driven not by climate-related changes in accumulation or melt, but rather by the internal ice-stream dynamics that led to the stoppage of Ice Stream C.

There is a 47.7 Gton/year difference in our estimate of net balance (+26.8 Gton/year) compared with the S&B estimate (−20.9 Gton/year) (6). The largest component of this difference is a change in estimated discharge fluxes (27.1 Gton/year). A breakdown of the discharge-flux differences is given in Table 2. Several studies have documented a deceleration on Ice Stream A and Whillans Ice Stream (9) with the most current estimate yielding an average deceleration rate of 5.0 m/year2 from 1974 to 1997 (16). As indicated in Table 2, the 23% deceleration over this period accounts for 10.6 Gton/year (38%) of the total difference in discharge. For Ice Streams D to F, there is little evidence of temporal variation in velocity over this period. The differences in discharge estimates for these ice streams are attributable to a variety sources, as shown in Table 2.

Table 2

Comparison of InSAR discharge estimates (seeTable 1) with S&B estimates (6). Most of the difference in discharge for Ice Stream A and Whillans Ice Stream can be attributed to velocity change and accumulation between gates. Much of the remaining difference may be the result of measurement error and additional flow from glaciers that discharge to the ice shelf near the southern end of the S&B gate (see Fig. 1). We estimate with a large uncertainty that the S&B gate for Ice Streams D and E was too wide by 17% and contains a substantial amount of slow-moving ice. Accounting for this difference and accumulation between gates yields a remaining difference with opposite sign. This negative difference is likely the result of some combination of remaining uncertainty in the S&B gate width, diversion of flow between gates (e.g., toward the S&B F gate), and basal melting on the floating part of the shelf (40). The difference at the S&B Ice Stream F gate is largely attributable to flux from Ice Streams D and E that passes through this gate.

View this table:

An additional difference of 3.5 Gton/year can be accounted for by the inclusion of basal melt beneath the grounded ice in the S&B estimate. We believe basal melt to be about an order of magnitude smaller than this amount and exclude it from our calculations (17).

As illustrated in Fig. 1, the overall catchment boundary we have derived differs from that used by Shabtaie and Bentley, making comparison of our basinwide accumulation estimate with the S&B estimate difficult. We performed a detailed comparison of accumulation over the S&B catchment area that leads us to conclude that the S&B results underestimate accumulation (18). This accounts for most of the revision in mass balance that is attributable to differences in accumulation. A much smaller contribution to this revision is attributable to our use of the average of two accumulation maps.

Although Whillans Ice Stream is nearly in balance at present, it was clearly out of balance in the past, as indicated by the roughly 23% decrease in velocity from 1974 to 1997 (16). It is unclear how long the deceleration has been underway, but it is likely that the discharge could have been greater still before 1974 (16). Although deceleration of Whillans Ice Stream has helped to restore its mass balance, changes in the ice surface geometry and resulting drainage patterns have also contributed. The surface of Whillans B2 (Whillans Branch 2) is substantially lower than that of Ice Stream C. This difference can be attributed, in part, to a thickening on Ice Stream C since stagnation (19) and a thinning of Whillans B2 (9), which is consistent with a past negative mass balance. As indicated in earlier work (19) and to greater extent by the catchments shown in Fig. 1, the drainage for Whillans B2 has expanded to include some of the area that formerly fed Ice Stream C. Furthermore, on the southern side of the B2 catchment, the boundary extends to the edge of a tributary feeding Whillans B1 (Whillans Branch 1), suggesting capture of part of its catchment by Whillans B2. This potential “ice piracy” may explain the negative mass balance for Whillans B1 and what is likely a relatively recent shift to a positive mass balance on Whillans B2. Our results represent an average over the catchment and, thus, are not inconsistent with estimates of local thinning on Whillans Ice Stream (9).

Critical to prediction of how this sector of the ice sheet will evolve over the next few centuries is an understanding of the future behavior of Ice Stream A and Whillans Ice Stream. These ice streams are currently in balance so that the observed deceleration could indicate that the ice sheet is responding as needed to compensate for a past imbalance. Field measurements collected in 1998 suggest that the deceleration may be continuing unabated (20). If so, then any restoration of balance is likely to “overshoot” and yield at least a temporary shift to a positive mass balance. Such a shift might be well within the degree of ice-flow variability that has been inferred for the region over the last 1000 years (2).

The deceleration could also indicate more than a simple restoration of mass balance. Thermodynamic models of ice-stream evolution show that ice streams may be inherently cyclic in their behavior, with switches between active (“purge”) and inactive (“binge”) phases occurring with a periodicity of several thousand years (21, 22). If deceleration on the ice plain fed by Ice Stream A and Whillans Ice Stream continues at its current rate, then outflow will cease in 70 to 80 years. Thus, these ice streams may be undergoing a transition from the purge to the binge phase, as seems to have occurred on neighboring Ice Stream C about 150 years ago (8). Modeling of ice-stream behavior indicates that ice-stream stoppages may be triggered by a prolonged period of negative mass balance, which leads to ice-stream thinning that, in turn, switches the basal thermal regime from melting to freezing (21, 22). Freezing removes water from the subglacial zone, thereby increasing basal resistance and decreasing ice-stream velocity. Through nonlinear positive feedback between basal freezing and shear heating, freeze-on-driven ice-stream slowdown may be a runaway process that, once initiated, leads to complete ice-stream shutdown (23). A paucity of direct constraints on the rates of sub–ice stream meltwater production and redistribution (24, 25) makes it difficult to rule out the possibility that the observed slowdown of Whillans Ice Stream is just a decadal-scale fluctuation. The recent stoppage of Ice Stream C (8), however, coupled with the results of thermodynamic ice-stream models, makes it plausible that the current slowdown of Whillans Ice Stream could continue to a complete stagnation. This would decrease outflow by 30.3 Gton/year, reduce collective discharge to 42% of accumulation, and yield an average thickening rate of 8.3 cm/year over the total catchment area shown in Fig. 1.

Even with a complete shutdown of Ice Stream A and Whillans Ice Stream, the direct impact on sea level is not large. If this stagnation were to occur, a positive imbalance of 57 Gton/year would lower sea level by 0.16 mm/year compared to a rise of 0.06 mm/year with the S&B estimate (6). For comparison, this positive imbalance would be nearly equivalent to completely blocking the flow of the Missouri River (72 Gton/year at U.S. Geological Service gage in Hermann, Missouri). Of more importance, our results, although providing only a limited “snapshot of mass balance,” suggest a reduced probability for scenarios that involve continued inland migration of the grounding line, a near-term catastrophic collapse, and a substantial rise in sea level. This analysis covers only the Ross Sea sector of the ice sheet, and negative imbalances are observed in other areas of West Antarctica such as Pine Island and Thwaites Glaciers (26).

A reduction in discharge could have important implications for the near-term stability of the Ross Ice Shelf, which exists largely as a result of high discharge by ice streams into the Ross Embayment. The combined stagnation of Whillans Ice Stream and Ice Streams A and C would reduce discharge across the grounding line of the Ross Ice Shelf by roughly 25% from that before the shutdown of Ice Stream C. Such a reduction in discharge could cause the ice shelf to thin and could trigger a retreat and/or break-up. Additional impetus for retreat/break-up may come from future climatic warming that appears to have helped to destabilize some smaller ice shelves along the Antarctic Peninsula (27). Over time scales on the order of decades to centuries, ice shelves may represent the most vulnerable element of the West Antarctic ice sheet/shelf system. Break-up of the Ross Ice Shelf alone would expose ∼400,000 km2 of new shallow sea surface area and could have important implications for exchange of energy and water between the ocean and the atmosphere/ice-sheet system in the region. Moreover, brine exclusion during sea-ice formation could turn this newly exposed polar continental shelf into a key source of bottom ocean water. This strengthened Antarctic bottom-water formation could outcompete the North Atlantic source of bottom water and switch the global ocean into a new mode of thermohaline circulation, with global climatic implications (28).

The positive imbalance we observe and the trend toward a potentially larger imbalance are evocative of an ice sheet in advance rather than in retreat. There is ample evidence for a large retreat of the West Antarctic ice sheet over the last several thousand years (3,4). The observed positive imbalance developed within just the last two centuries as a result of the stoppage of Ice Stream C and slowdown of Whillans Ice Stream. If the current positive imbalance is not merely a part of decadal- or century-scale fluctuations, it represents a reversal of the long-term Holocene retreat.

  • * E-mail: ian{at}, tulaczyk{at}


Stay Connected to Science

Navigate This Article