The team focused on the Nivlisen Ice Shelf, a 7,600-square-kilometre slab of floating ice fed mainly by the Potsdam Glacier. The shelf has a long history of surface meltwater activity, with large lakes and channels documented nearly every summer since the early satellite era. Yet while supraglacial lakes have attracted considerable attention, because ponded water can drive hydrofracturing and catastrophic ice-shelf collapse, the channels that connect and drain those lakes have remained surprisingly understudied. That gap matters, because channels control where meltwater is stored, where it is routed, and whether it escapes the ice surface altogether.

To fill it, the researchers turned to freely available satellite imagery from the European Sentinel-2 and American Landsat-8 missions, covering every January from 2017 to 2024, plus a detailed seasonal sequence through the 2018–2019 melt season. Surface water was detected using a spectral technique called the normalised difference water index for ice, which exploits the fact that meltwater reflects blue light strongly and red light weakly. Channels were then traced by hand as centrelines, verified against the Reference Elevation Model of Antarctica to confirm that each feature followed a downslope path consistent with gravitational flow. From these maps the team derived four key metrics: surface water area, number of channels, total channel length, and the number of distinct drainage components, defined as isolated channels, connected channel systems, or channels linked to lakes.

The results reveal extraordinary variability. In total, the team mapped 710 drainage components comprising 5,513 individual channel segments. In the peak summer of 2018, the ice shelf hosted roughly 172 square kilometres of surface water, 1,650 channels organised into 186 drainage components, and a combined channel length exceeding 1,400 kilometres. By January 2022, in stark contrast, no visible surface water could be detected at all. In 2023 and 2024 the system partially recovered, but channel lengths and water areas remained far below pre-2021 levels. Statistical tests confirmed that these interannual differences in water depth and elevation were significant, even though median depths hovered near 0.4 metres and most drainage components sat at around 70 metres above sea level throughout the record.

The seasonal analysis of 2018–2019 captured the drainage system in a remarkable act of self-organisation. In early December, the ice shelf carried only sparse, isolated pools of water. By late December, short fragmented channels had appeared near the grounding line, where the ice first begins to float. By late January, the network had reached its maximum extent: 924 channels totalling 967 kilometres, with surface water peaking at about 191 square kilometres. Crucially, while the number of channels grew, the number of separate drainage components fell from roughly 170 to about 100, showing that previously independent systems were merging into one integrated network. By early March, the system had fragmented again, with total channel length collapsing below 35 kilometres.

This seasonal pulse has a clear directional signature. Meltwater that began the season ponded in lakes at elevations of up to about 400 metres near the grounding line was progressively transferred downslope toward the ice front, roughly 20 kilometres away, where elevations drop to around 50 metres. As the season advanced, water at high elevations became less apparent, lakes there emptied and in some cases vanished, and the drainage network migrated toward lower ground. The researchers interpret this as a transition from an inefficient, fragmented system to an efficient, connected one, in which fast-flowing channels evacuate water from the upper parts of the shelf and deliver it to lower regions, where it ponds, refreezes, or soaks into the firn, the compressed snow layer that blankets the ice.

Perhaps the most striking finding is that the plumbing is permanent. Channels repeatedly reform in the same locations, aligned with surface flow-stripes, the subtle longitudinal structures etched into the ice by long-term flow. Even more dramatically, the main western channel makes an abrupt 90-degree turn to the east after merging with two other major channels, a diversion the team traced to a local surface high above a basal pinning point, a bump of grounded ice beneath the shelf that warps the surface upward. This topographic steering funnels water from all three channels into a single terminal lake, which covered 4.35 square kilometres in January 2019 and expanded to 10.1 square kilometres in January 2020, sitting just 18 kilometres from the ice front. Radar imagery suggests that when the surface water disappears in cold years, the frozen traces of these channels persist through winter, forming impermeable ice conduits that reactivate the following summer.

What drives the year-to-year swings? The answer, overwhelmingly, is temperature. Correlating the drainage metrics against climate variables from the MAR regional climate model, the ERA5 reanalysis, and local weather stations, the team found that mean summer air temperature from ERA5 was tightly coupled to every measure of the drainage system, with correlation coefficients above 0.9 and statistical significance at the 95 percent level. Total channel length tracked temperature most closely, with a correlation of 0.99. Shorter-term analysis showed the network responding to cumulative melt within one to two weeks. Snowfall, by contrast, had a variable and sometimes masking effect: exceptionally high summer snowfall in 2022 buried the surface under fresh, bright snow, hiding any meltwater from optical satellites and contributing to that year’s apparent total absence of surface water.

The implications for ice-shelf stability are nuanced. On the one hand, the Nivlisen drainage system currently acts as a closed circuit: meltwater is redistributed across the shelf but, in the observation period, was not exported off it, instead being retained in lakes, refrozen, or absorbed into the firn. The melt-over-accumulation ratio, an index of how much melt the snowpack must absorb, stayed well below the threshold associated with firn saturation, suggesting the shelf retains considerable buffering capacity. On the other hand, the study shows that warm summers can move meltwater over tens of kilometres and concentrate it in specific places, including a large lake near the ice front. If future warming increases melt production or erodes the firn’s storage capacity, these pre-existing, topographically anchored pathways could route water toward structurally vulnerable zones, raising local hydrofracturing risk. Historical Landsat imagery hints that systems comparable to the 2019–2020 network appeared as far back as 1976 and 2008, meaning such episodes are episodic rather than new, but their frequency may change as Antarctica warms.

What makes the study especially valuable is its methodological honesty and its contrast with Greenland. On the Greenland Ice Sheet, supraglacial drainage expands up-ice through the summer as the melt zone climbs to higher elevations, and water exits the system through moulins and crevasses. Nivlisen behaves in the opposite way: a floating, closed surface system that migrates downslope and retains its water. The authors caution that their correlations rest on only seven years with visible water and that satellite-derived depths remain unvalidated in the field. Even so, by quantifying how a major East Antarctic ice shelf builds, connects, and dismantles its river network each summer, the work provides a baseline against which future change, and the growing threat of surface melt to the ice shelves that hold back the Antarctic Ice Sheet, can be measured with confidence.

Subject of Research: Interannual and seasonal evolution of supraglacial meltwater drainage systems on the Nivlisen Ice Shelf, East Antarctica

Article Title: Interannual and seasonal evolution of supraglacial drainage systems on Nivlisen Ice Shelf, East Antarctica

Article References: Interannual and seasonal evolution of supraglacial drainage systems on Nivlisen Ice Shelf, East Antarctica. (n.d.). https://doi.org/10.5194/tc-20-5785-2026

Image Credits: AI Generated

DOI: 10.5194/tc-20-5785-2026

Keywords: Antarctica, ice shelves, supraglacial drainage, meltwater channels, supraglacial lakes, remote sensing, Sentinel-2, Landsat-8, climate warming, hydrofracturing, The Cryosphere, East Antarctic Ice Sheet