The future evolution of the Antarctic Ice Sheet and its potential sea level contribution in a changing climate is highly uncertain , largely due to the uncertain response of ice shelves . Ice shelves play a critical role in buttressing the ice sheet and their weakening or eventual collapse can accelerate ice flow across the grounding line, which in turn contributes to sea level rise . It is therefore crucial to understand the processes that lead to ice shelf weakening to better constrain uncertainties in sea level rise projections.

Basal melting is one such process known to weaken ice shelves and has been linked to the collapse of Wordie Ice Shelf . Basal melting is often driven by the intrusion of Circumpolar Deep Water (CDW), a relatively warm water mass located at depth beyond the continental shelf . In some regions of Antarctica, the continental shelf's bathymetry allows CDW to flow over it, enabling this warm water to reach ice shelf cavities referred to as warm-cavity ice shelves,. This phenomenon occurs particularly in the Amundsen and Bellingshausen Seas – where the Wordie Ice Shelf was located – which experience some of the highest basal melt rates on the continent . Although basal melting is most pronounced near the grounding zone due to the pressure-dependent freezing point of seawater , the resulting meltwater plume usually travels from the grounding zone towards the ice shelf front, thereby carving basal channels in the ice . These channels, which are more prominent in warm-cavity ice shelves , represent potential zones of weakness due to their reduced thickness and elevated melt rates relative to their surroundings .

Basal channels are dynamic features that do not necessarily remain in a steady state. They can experience changes in melt intensity, fracturing, and re-routing. In warm-cavity ice shelves, the magnitude of melting within channels and across the entire ice shelf is primarily influenced by the volume and temperature of CDW entering the ice shelf cavity . These oceanic conditions can, in turn, be modulated by larger climatic phenomena such as the El Niño Southern Oscillation ENSO,. During El Niño events, the weakening of near-coastal easterly winds limits the shoreward supply of cooler surface waters and facilitates greater inflow of CDW onto the continental shelf, thereby intensifying basal melting . Increased availability or higher temperatures of CDW can thus lead to enhanced thinning through basal melting .

Basal channels can also be associated with transverse fracturing, a phenomenon observed on several ice shelves, including Pine Island, Nansen, Moscow University, and Totten . These fractures, driven by extensional stresses within the ice , have triggered major calving events on Nansen and Pine Island ice shelves . Although extensional stresses are considered a primary driver, variations in meltwater availability within channels may also contribute to fracture formation . In addition to transverse fracturing, longitudinal fractures within basal channels have been documented , potentially leading to full-thickness fractures and eventual ice shelf retreat .

The interactions between basal channels, fractures, and CDW-variations are highly complex, reflecting significant knowledge gaps in channel dynamics. Beyond changes in melt magnitude and fracturing, basal channels have been observed to migrate laterally or re-route entirely. Examples of such behavior include channels on the Getz and Roi Baudouin ice shelves, as well as other ice shelves in the Amundsen and Bellingshausen Sea sectors . Lateral channel migration, often in the direction favored by the Coriolis effect, has been linked to increased ocean heat availability . Similarly, the re-routing of a meltwater plume on Thwaites Ice Shelf has been observed to follow pre-existing fractures, underscoring the interplay between channel dynamics and ice shelf structure .

The importance of high-resolution basal channel melt rates on ice shelf weakening was emphasized using Basal melt rates Using the Reference Elevation Model of Antarctica (REMA) and Google Earth Engine BURGEE,. The high-resolution BURGEE melt data reveal a peculiar, and to our best knowledge so far unreported, channel on George VI Ice Shelf with high melt rates near extensive channelization (Fig. ). The channel appears to intersect and alter other channels in the area, suggesting significant changes in the channel system during the BURGEE period (2010–2022). Given the channel's location in a heavily channelized area and its proximity to the southern ice shelf front which has been persistently retreating since 1947 , this channel poses a potential risk of further retreat.

In this study, we examine the potential drivers behind the formation and evolution of this newly observed channel on the George VI Ice Shelf (Fig. ). Using high-resolution surface elevation and basal melt rate data, we assess temporal changes in channel morphology. Furthermore, we explore the roles of ice flow dynamics, ocean circulation, and ocean forcing as potential contributors to channel evolution. Finally, we employ a basal melt model with realistic ocean forcing and ice shelf geometries from two time periods (2010–2016 and 2016–2022) to investigate the possibility of channel re-routing within the observed network. This multi-faceted approach aims to advance our understanding of the processes driving rapid channel formation.

Our study area focuses on the region surrounding the newly identified channel on the George VI Ice Shelf located near the southern ice shelf front, as well as the two entrances to its ice shelf cavity where CDW flows in (Fig. ). The analysis spans the BURGEE period from 2010 to 2022.

George VI Ice Shelf is particularly vulnerable due to its already thin state and ongoing thinning . The ice shelf experiences significant melting both from its surface and its base , making it susceptible to structural weakening. Adding to this vulnerability, studies have shown that parts of the firn layer of the ice shelf are saturated to levels that could facilitate hydrofracturing, while other areas are approaching similar thresholds with only minor increases in temperature . This dual weakening mechanism highlights the sensitivity of George VI Ice Shelf to changes in climatic conditions.

The bathymetry of the region further amplifies this vulnerability. Troughs in the continental shelf guide CDW into the cavity beneath the ice shelf Fig. ,. Once inside the cavity, the CDW drives some of the highest basal melt rates observed in the Antarctic Peninsula .

In the following subsections, we describe the data and methods used to (i) generate high-resolution surface elevation and basal melt rates to analyze the temporal evolution of the channel (Sect. ); (ii) explore potential drivers behind the channel, utilizing both ice velocity observations (Sect. ) and model simulations of ocean circulation and forcing (Sect. ); (iii) model temporal changes in basal melt patterns near the channel driven by variations in ocean forcing and ice shelf geometry (Sect. ).

The basal melt rate from 2010 to 2022, shown in Fig. , reveals a general pattern consistent with previous studies : higher melt rates are observed in the southern region of the ice shelf compared to the north. Furthermore, as is typical for most ice shelves, the highest melt rates are concentrated near the grounding zone, with channels extending from these zones across mainly the southern part of the ice shelf. A closer examination of the newly identified channel highlights its position within a densely channelized area. This channel stands out with exceptionally high melt rates, reaching up to approximately 30 m yr⁻¹, which surpasses the melt rates in the surrounding channels. The channel appears to intersect existing channels, complicating the interpretation of the meltwater plume's pathway.

By examining the surface elevations over the study period, it becomes evident that the new channel has developed during this period; it began forming around 2015, reflected by a narrow surface depression in Fig. d. This depression deepens progressively throughout the study period by 23 m between 2013/2014 and 2022/2023, with two further key observations emerging. First, as the depression associated with the channel becomes more pronounced (Fig. j, around 11 km), the deepest part of the older channel becomes shallower (Fig. j, around 8 km), while the flanks of the older channel are lowering (Fig. j, around 4–8 and 9–10 km), thereby widening this pre-existing channel. These flanks are also associated with high melt rates (∼ 15 m yr⁻¹, Fig. ), as opposed to the deepest part of the channel (∼ 0 m yr⁻¹, Fig. ). The continuous lowering of the flanks suggests that the channel system is not in a steady state, with the closure of the deepest part possibly indicating channel re-routing, where the new channel may now serve as the primary basal melt pathway. Second, just downstream of the channel (Fig. j, around 12 km) a slight surface elevation bump appears from 2016/2017 onward. This type of bump, known as flanking uplift, is typically associated with fractures on ice shelves . However, narrower channels can cause similar flanking uplift . These findings suggest that the channel could either be a fracture, a basal melt channel, or a combination of both (i.e., a fracture that serves as a melt plume pathway or a channel that has begun to fracture). Optical imagery throughout the study period from Landsat 8 and Sentinel 2 also do not provide any clear answer as to whether it is a channel or fracture (Suppl. Movie). Since both basal channels and fractures can weaken ice shelves , the presence of either on George VI, a relatively thin ice shelf with a relatively warm atmosphere, likely contributes to its weakening .

To further investigate fracturing as a possible driver behind such a fast developing channel we explore changes in ice flow across the channel. Ice shelf fracturing is typically linked to variations in ice speed and divergence caused by stretching. Our analysis of ice speed and divergence across the channel show speeds along the transect fluctuating substantially in both 2014/2015 (minimum: 372 m yr⁻¹, maximum: 395 m yr⁻¹; Fig. a) and 2015/2016 (minimum: 375 m yr⁻¹, maximum: 397 m yr⁻¹; Fig. b) with an isolated peak in the ice speed at the channel location in 2015/2016. That peak is associated with sudden changes in divergence from -0.02 yr⁻¹ (compression) just upstream of the channel to 0.03 yr⁻¹ (stretching) just downstream of the channel (10–11 km, Fig. b), which could indicate fracturing as a potential driver of the channel. These changes, however, are rather subtle in comparison to the noise level and to observations across wider fractures (∼ 1–4 km) on e.g. Ross Ice Shelf, where divergences reaches 30–80 m yr⁻¹ at their maximum . The same pattern of a noisy divergence signal in the first few years followed by a limited divergence signal near the channel in the later years also becomes obvious from the spatial analysis of effective strain rates (Fig. ). The low magnitude of the changes could potentially be due to the coarse resolution of the velocity product (200 m) relative to the channel's width and depth. Finally, in the later years (2017 and 2020, Fig. c and d), both ice speeds and divergences are more stable, without any outstanding signals in the vicinity of the channel.

Focusing on changes in ocean heat as possible driver of channel changes, we investigate temporal changes in ocean temperature and salinity. Figure a–f illustrate temperature and salinity anomalies for the Northern, Middle, and Southern profiles (Fig. ), revealing a regime shift from cold and fresh conditions to warmer and saltier conditions below the surface layer across all profiles. At the Northern profile, this shift begins around 2013, intensifying in 2015/2016, after which sub-shelf temperatures remain above the temporal mean. Similar, albeit weaker, trends are observed in the Middle and Southern profiles. In the deeper ocean layers below -300 m, where CDW resides, both average temperature and salinity have increased over the study period across all three profiles (Fig. g–i). Notably, a jump in salinity is observed around 2016, with the most pronounced increase in the Northern profile (Fig. g), where the strongest temperature and salinity anomaly shifts also occur (Fig. a and d), although the average temperature at depth already starts increasing in 2011 (Fig. g). This shift in ocean regime aligns with the 2015 El Niño Southern Oscillation (ENSO) event, with its effects already having been shown to reach George VI . These changes in oceanic conditions, partially driven by ENSO, thus indicate more heat available for basal melting, likely intensifying and accelerating the meltwater plume, which in turn may cause higher melt rates and alterations in the plume pathway, influenced by the ice shelf's evolving geometry. Furthermore, in the Southern temperature anomaly profile, where several melt channels have their outflows (Fig. ), the upper ocean layers have become fresher since 2016 (Fig. f), possibly indicating increased meltwater outflow.

The increased meltwater outflow is supported by MITgcm-derived ice-shelf-wide and channel-area mean basal melt rates showing an increase between 2010 and 2020 (Fig. a), as does the mean melt rate within the channel area (Fig. b), thus supporting the increased meltwater outflow. The time series in Fig.  also indicate that the 2015 ENSO event coincides with an increase in the maximum ice-shelf-wide melt rate, and that the end of this ENSO event correlates with a shift towards higher mean melt rates both across the entire ice shelf and within the channel area. This shift also coincides with the onset of the most rapid deepening of the new channel (Fig. c). It is important to note that MITgcm uses a static ice-shelf geometry, which implies that changes in modelled basal melting do not reflect shifts in melt because of changing channel-geometry. However, for a new channel to arise or to modify an existing channel, elevated melt rates are likely needed to give the plume extra energy to cut new pathways. The amplified MITgcm melt rates after the 2015 ENSO event may therefore be indicative of more energy available for the plume to create new pathways.

Not all ENSO events are associated with the same changes in MITgcm-modelled melt rates, and not all shifts in modelled melt occur during ENSO phases. For example, the 1997/1998 El Niño was comparable in magnitude to the 2015 event and likewise accompanied by a peak in maximum ice-shelf-wide melt rates (Fig. a). Unlike 2015, however, the increase was short-lived and followed by a decline in both mean and maximum melt rates. One possible explanation is the rapid transition into a long La Niña phase immediately after 1997/1998, which likely counteracted the oceanic conditions favourable for sustained basal melting. In addition, ENSO events also vary in where the strongest sea-surface temperature anomalies develop across the tropical Pacific. The 1997/1998 event was a classic Eastern Pacific El Niño, whereas the 2015 event was initially Central Pacific but evolved into a mixed event with warming extending further eastward . Such differences in ENSO “flavour” can alter how tropical variability is transmitted to the Southern Ocean, leading to contrasting impacts on basal melting.

The LADDIE model outputs from the pre-emergence and post-emergence experiments allow us to zoom in on the channel at a finer resolution, enabling us to explore how changes in ice shelf geometry and ocean forcing affect basal melt patterns. The modelled melt rate patterns (Fig. c and f) align well with observations in areas outside the channel (Fig. g). Figure f shows that LADDIE suggests the channel could serve as a pathway for the meltwater plume, potentially contributing to the high melt rates within the channel itself. Moreover, both pre-emergence and post-emergence melt patterns show enhanced melt along the sides of the channel, similar to BURGEE, which explains the widening of the old channel seen in Fig. j. Notably, the observed BURGEE melt pattern (Fig. g) near the channel appears to combine elements from both the pre-emergence and post-emergence experiments, with channels following the “original” channel system as in the pre-emergence scenario combined with high melt rates within the new channel as in the post-emergence scenario. This agreement between observations and models suggests that the plume pathway may have shifted during the BURGEE observational period, now following a new route as indicated by the post-emergence experiment.

In this study, we uncovered a new channel within the channelized basal melting network of the George VI Ice Shelf, characterized by rapid changes in surface elevation, from 48 m in 2013/2014 to 25 m in 2022/2023. The onset of this channel coincides with both a shift towards a warmer ocean regime and increased modelled melt rates as well as subtle divergence changes across the channel, all aligning with the 2015 ENSO event . ENSO has already been linked to an intensified inflow of CDW onto the continental shelf during El Niño years , supporting the MITgcm results of a warmer regime and increased basal melt rates after 2015. The 2015 El Niño also stands out in terms of its atmospheric and surface impacts: in the George VI catchment it produced the strongest positive surface mass balance anomalies observed in recent decades, while continent-wide SMB changes during spring extended well beyond the range of background variability and penetrated further inland than during earlier extreme events . The same event has further been linked to the acceleration of glaciers feeding into the George VI Ice Shelf . Additionally, ENSO events have been shown to enhance basal melting on ice shelves in the Amundsen Sea Sector . While we cannot definitively conclude that ENSO caused the appearance of this channel, the timing is notable and the potential link to ENSO is important, as the projected increase in El Niño frequency suggests even further acceleration of ice-shelf basal melting in Antarctica in the future .

The observed new channel could represent a basal melt channel, a fracture, or a combination of both. Our investigations of the channel’s origin point towards either a basal melt channel or a combination of both. The uplift in the surface near the depression, coupled with signs of stretching in the divergence field downstream of the channel in 2015/2016, could suggest fracturing. However, these signals are subtle, leaving us unable to confirm or dismiss the possibility of fracturing with certainty. Furthermore, it should be noted that channels themselves can generate viscous flow independent of any fracturing , which implies that both the observed flanking uplift as well as the subtle divergence signal could also purely be due to channelization. At the same time, changes in ocean temperature, salinity, and circulation point to evolving and strengthening basal melt conditions that could encourage channel re-routing. Our LADDIE modelling results further support this possibility, indicating that even if the channel originated as a fracture, the meltwater plume may now be using it as a new pathway, which could further deepen the fracture, provided the initial fracture has a basal component. The new channel is narrow, whereas LADDIE has been primarily validated in broader channels with gently sloping sidewalls . Flow in narrow, steep-sided channels may involve additional complexity, including ageostrophic processes , that may not be fully captured in our setup. While we consider our current configuration sufficient for assessing plume pathways, caution is warranted when interpreting the exact melt rates and plume speeds in the channel.

To assess the potential impact of the new channel on the weakening of George VI Ice Shelf, we can draw parallels to other ice shelves where interactions between basal melt channels and fractures have led to structural instability. A notable example is Pine Island Ice Shelf, where a channel first observed in the 1970s progressively thinned and extended over time . This channel triggered both transverse fractures and fractures along the channel’s length, eventually leading to calving and retreat along the channel from approximately 2018 to 2022 . The new channel on George VI Ice Shelf is located just about 30 km from the southern edge, raising concerns that continued melting, thinning, and weakening in this region could drive significant structural changes. These changes may include enhanced fracturing, calving, and ultimately, retreat of the ice shelf in this highly channelized area. Given projections of increased ocean heat availability in the future , basal melting is expected to intensify, further amplifying the vulnerability of this portion of George VI Ice Shelf.

To definitively determine whether the channel is a fracture, a basal melt channel, or a combination of the two, field measurements, including basal melt rates, ice deformation monitoring, and radar surveys of ice thickness, are necessary. In parallel, a high-resolution, fully coupled model capable of simulating both ice flow, fracturing, and basal melting would be required to rigorously evaluate the interactions between channels and fractures and to assess causality in their formation. While our current results cannot unambiguously establish the mechanisms behind this channel, they provide a direct observation of weakening in a highly channelized, vulnerable portion of the ice shelf. These results underscore how quickly ice shelf channels – important for ice shelf integrity – can occur and how easily small-scale changes might go unnoticed. Closely observing the continued evolution of the ice shelf and its integrity will be crucial in understanding these weakening processes. Such knowledge could also be valuable for other heavily channelized ice shelves, like Pine Island and Totten, which both have a higher projected potential sea level rise contribution .

Our study highlights the rapid emergence of a significant channel on the George VI Ice Shelf, marked by a 23 m surface lowering over just nine years. The appearance of the channel aligns with both changes in ocean forcing, most notably increased ocean temperatures and salinity, and subtle changes in ice divergence, both of which coincide with the timing of a major ENSO event. While the exact link between ENSO and the development of this channel remains speculative, the temporal correlation suggests that large-scale climate patterns may have a role in amplifying basal melting and possibly in re-routing meltwater pathways on Antarctic ice shelves.

The presence of such a fast-evolving channel on an already thin and vulnerable ice shelf like George VI likely has destabilizing effects, accelerating its weakening through both basal melting and fracturing. The behavior of this channel on George VI may offer valuable insights into how sudden changes in ocean forcing could trigger similar destabilizing processes elsewhere.

Moving forward, continuous monitoring of this channel and its evolving impact on George VI is crucial. The lessons learned from tracking its development may provide critical information on the future behavior of other highly channelized ice shelves undergoing changes in ocean conditions. Understanding these processes is essential for better projecting potential ice shelf retreat and the associated contributions to global sea-level rise.