Bathymetry and geology
The outer approximately first third of the 116 km-long Victoria Fjord was bathymetrically mapped using a multibeam echo sounder; a high concentration of icebergs prevented the icebreaker from progressing farther into the fjord (Supplementary Fig. 4). Information about the seafloor in the unmapped inner part was therefore acquired by obtaining spot depth soundings using a single-beam echosounder deployed from a helicopter (Fig. 2a and Supplementary Fig. 5). The multibeam bathymetry together with all spot soundings have been used to update the most recent version the International Bathymetric Chart of the Arctic Ocean (IBCAO) Version 5.023 digital terrain model (DTM), which in turn uses BedMachine Version 5 for the under ice topography and inner fjord bathymetry24(see “Methods” and Supplementary Fig. 6).
Fig. 2: Detailed bathymetric maps and data from Victoria Fjord and the Lincoln Sea.
The alternative text for this image may have been generated using AI.
a, b Maps showing seafloor bathymetry from multibeam mapping and spot soundings (pink crosses with recorded depths). The background bathymetry, shown in a blue colour scheme, is based on the International Bathymetric Chart of the Arctic Ocean (IBCAO) Version 5.023, updated in this study with the new depth data acquired during the GEOEO 2024. Locations of temperature profiles acquired with CTD and XBT instruments are marked with filled circles. Radio echo-sounding (RES) profiles acquired from C.H. Ostenfeld Gletsjer by helicopter are shown with orange lines. The 1917 ice-tongue margin in Victoria Fjord, mapped by Lauge Koch, is shown in turquoise. Schematic subsurface AW flow inferred from temperature profiles is indicated by red arrows. c Selected temperature profiles from Victoria Fjord and Lincoln Sea; locations are shown on the map by black and orange filled circles, respectively. Locations of bathymetric profiles shown in Supplementary Fig. 1 are indicated with black lines. The approximate deepest passages across the sills, along with their depths, are indicated on the map.
At the mouth of Victoria Fjord, a prominent bathymetric outer sill was identified. The sill spans the full width of the fjord between the eastern fjord wall and Stephenson Island and extends approximately 30 km along its axis (Fig. 2a). The fjord west of Stephenson Island was not mapped. The seafloor morphology is rough and irregular, with exposed structural ridges overprinted by glacial landforms, consistent with past grounded ice flow at this point along the fjord. In overall shape and relief, it resembles the outer sill of nearby Sherard Osborn Fjord9, although the sill in Victoria Fjord is slightly deeper, with a threshold depth of ∼512 m, compared to ∼486 m in Sherard Osborn Fjord (Fig. 2a and Supplementary Fig. 1). Further south into the fjord, inside this outer sill, the seafloor deepens to a maximum of ∼1100 m. This is approximately 100 m deeper than the depths found north of the sill. The spot soundings indicate the presence of a shallow inner sill in Victoria Fjord, but it does not span the entire width of the fjord (Fig. 2a). A distinct, deep channel through this sill begins around the 400 m isobath and is more than 7 km wide. Based on our spot soundings, the deepest point of this passage has a water depth of ∼565 m (Fig. 2a and Supplementary Fig. 1). In comparison, the previously mapped inner sill in Sherard Osborn Fjord has only a ∼1 km wide channel on its eastern side, where the water depth reaches a maximum of ∼390 m9 (Supplementary Fig. 1c).
Inland of the inner sill in Victoria Fjord, the fjord deepens again. The bathymetry of the innermost section, about 44 km from the ice margin, is constrained by only four spot soundings (Fig. 1b, 628, 376, 622 and 514 m, N–S). Three additional measurements indicated depths greater than 450 m, but a problem in the echosounder configuration limited the recordings below this depth threshold during the deployment.
Comparing the spot soundings with IBCAO 5.0, which incorporates BedMachine data in unsurveyed areas such as Victoria Fjord, reveals large differences. The average depth difference is −342 m (IBCAO minus spot soundings), with a median difference of −361 m. The IBCAO/BedMachine bathymetry is systematically shallower in the central fjord and significantly underestimates the basin depth with an observed maximum difference of −865 m. The large discrepancies are not unexpected. In previously unsurveyed fjords, BedMachine constrains seabed elevation, i.e. water depth, at the grounding line using ice thickness derived from radar data and a mass-conservation approach. Fjord bathymetry is then interpolated by fitting a minimum-curvature surface, typically relying on sparse depth soundings located outside the fjord.
When the new bathymetry is merged with available geological maps, it becomes strikingly clear that the geological formations of north-western Greenland align with the mapped bathymetric sills in Victoria, Sherard Osborn and Petermann fjords (Fig. 3). The sills are situated within the main fold and thrust belt of the Devonian–Carboniferous Ellesmerian orogen exposed on Greenland9. The outer sill in Victoria Fjord aligns with the same bedrock units as those at the outer sill of Sherard Osborn Fjord, comprising steeply dipping beds of deformed Silurian deep water turbidites and older deposits of the Franklinian Basin9 (Fig. 3). The inner sill in Victoria Fjord corresponds to a belt of carbonate reefs belonging to the Silurian Washington Land Group that form a former shelf edge25,26 (Fig. 3). These reefs are dolomitized and silicified, making them less likely to be eroded away, which leads to topographic highs on land and sills in the fjords. The same durable bedrock forms the base of a shallow inner sill that nearly continuously extends across Sherard Osborn Fjord and greatly restricts the inflow of AW. This is evidenced by a decrease in AW temperature of approximately 0.2 °C landward of the sill9 (Supplementary Fig. 1d). The Silurian Washington Land Group extends westward beyond Sherard Osborn Fjord and aligns with the prominent sill at the entrance of Peterman Fjord (Fig. 3). This bathymetric shoal forming the sill has a maximum water depth of about 443 m27 and is generally deeper than the inner sill in Sherard Osborn Fjord. The bathymetric profile across the sill at the mouth of Petermann Fjord is similar in overall geometry to that of the inner sill in Victoria Fjord, with comparable cross-sectional shape, although the deepest passage is located to the west rather than the east (Supplementary Fig. 1).
Fig. 3: Geological formations on north-western Greenland aligning with mapped bathymetric sills in Victoria (VF), Sherard Osborn (SOF) and Petermann fjords (PF).
The alternative text for this image may have been generated using AI.
Carb. carbonate, Silic. siliciclastic, F&T fold and thrust, SE shelf edge. Dashed line marks the approximate position of the Washington Land Group. Geological information is from Hopper and Ineson25.
The dynamics of the outlet glaciers in Sherard Osborn and Petermann fjords, both at present and over longer timescales, have been shown in previous studies to be strongly influenced by these sills9,27,28. This suggests that regional geology exerts a fundamental control on the fjord bathymetries, and thereby on glacier behaviour. In the following, we assess whether the oceanographic data from Victoria Fjord indicate that sill geometry similarly influences present conditions at C.H. Ostenfeld Gletsjer.
Oceanography
During the GEOEO 2024, temperature and salinity profiles were collected from IB Oden in the Lincoln Sea, and throughout Victoria and Petermann fjords using a CTD (conductivity, temperature, depth) probe (Fig. 2 and Supplementary Fig. 1). In the southern two-thirds of Victoria Fjord, inaccessible to IB Oden, temperature data were acquired using XBTs (expendable bathythermographs) launched from a helicopter. To compare the ocean temperature structures in Victoria and Sherard Osborn fjords, we use the oceanographic data from the Ryder 2019 Expedition9,29, while the comparison with Petermann Fjord is based on the data collected during the GEOEO 2024. Subsurface water masses with temperatures above 0 °C are inferred to be AW, as there is no other known source of warm water in the region.
In 2024, waters in the Lincoln Sea and the outer part of Victoria Fjord reached temperatures above 0 °C at depths greater than about 200–275 m (Fig. 2c). Water with a temperature above 0 °C was observed at slightly greater depths of ∼290 m in the inner fjord of Victoria Fjord, i.e. landward of the inner sill, indicating the full intrusion of AW up to the ice margin of C.H. Ostenfeld Gletsjer (Fig. 2c). This is consistent with the bathymetric data from Victoria Fjord, which, as previously described, show a >400 m deep and ~7 km wide passage in the inner sill. In contrast, Sherard Osborn Fjord has a nearly continuous inner sill spanning the fjord, and temperatures there first reach 0 °C at much deeper depths of ~350–375 m landward of this sill (Supplementary Fig. 1d).
Figure 4 shows the temperature–salinity relationship for AW in the Lincoln Sea, and Victoria and Petermann fjords for selected stations taken during the GEOEO 2024; observations taken near the front of Ryder Gletsjer in 2019 are also shown. The maximum AW temperature in Lincoln Sea, observed in 2024, was slightly above 0.4 °C, and AW temperatures reaching 0.35 °C were encountered in the Victoria Fjord CTD and XBT measurements. In Petermann Fjord, the temperatures around the sill level (~440 m) were colder than at similar depths in the Lincoln Sea and Victoria Fjord (Supplementary Fig. 1d). However, the temperature increases with depth in Petermann Fjord, reaching ~0.3 °C at around 1000 m, indicating that inflows of warmer AW from the Lincoln Sea occasionally spill over the sill. Similarly, the cooler temperatures in the deep Victoria Fjord presumably reflect past inflow events, which carried colder but more saline AW into the deep fjord basin.
Fig. 4: Temperature–salinity relationship for Atlantic Water (AW) in the Lincoln Sea and Victoria, Sherard Osborn and Petermann fjords.
The alternative text for this image may have been generated using AI.
The temperature-salinity data are based on CTD stations shown in Fig. 2a, b and Supplementary Fig. 1. Station numbers are show next to the profiles in corresponding colours. The inner part of Victoria Fjord, beyond the reach of IB Oden and only sampled by XBT, which lacks salinity data, is not included in the comparison. Grey lines show potential density, σθ = ρθ − 1000 kg m−3.
The maximum temperature in the inner Sherard Osborn Fjord, near the front of Ryder Gletsjer ice tongue, was notably colder than the AW temperature maximum outside the inner sill in the main fjord in 2019. The reason for this is the sill geometry in Sherard Osborn Fjord: as the AW flows towards the glacier over the relatively shallow inner sill, it mixes with colder and fresher outflowing water that has been modified by glacier–ocean interactions9,30. Indeed, Supplementary Fig. 1d shows that the water deeper than the sill in the inner basin of Sherard Osborn Fjord, which reaches the Ryder Gletsjer ice tongue, is colder than the AW outside the fjord in the Lincoln Sea. Furthermore, Fig. 4 shows that the densest water, occupying the deep inner basin of Sherard Osborn Fjord, is fresher and less dense than the warmest Lincoln Sea AW, which is typically encountered at depths around 300–400 m (Fig. 2c). Importantly, the flow and its associated mixing over the inner sill provides a stabilising dynamical feedback to ocean warming: AW temperature changes in the Lincoln Sea yield a muted temperature response at the grounding line of Ryder Gletsjer9,30.
The observed AW features in Victoria Fjord suggest that the C.H. Ostenfeld Gletsjer is more susceptible to changes in AW temperature than Ryder Gletsjer. The C.H. Ostenfeld Gletsjer is exposed to AW from the Lincoln Sea that is apparently only slightly cooled along its flow into Victoria Fjord while Ryder Gletsjer is, as mentioned, protected by the prominent inner bathymetric sill.
The Lincoln Sea CTD stations taken during the Ryder 2019 and GEOEO 2024 expeditions have AW temperature maxima in the range 0.35–0.44 °C. Notably, AW temperature maxima have been reported in the range 0.35–0.8 °C from ice-tethered measurements in the Northern Lincoln Sea, near or on the continental slope31. These measurements cover the period 1991–2009, with a data gap between 1997 and 2002 and reveal temporal as well as spatial variability, including a noticeable bi-modal pattern with particularly high AW temperatures in 2004, 2005 and 2008. The apparent bi-modal variability of AW temperatures in the Lincoln Sea may reflect variability in upstream circulation pathways, with warmer anomalies potentially associated with AW advected along the Lomonosov Ridge. Corresponding measurements of AW in the Southern Lincoln Sea, closer to Victoria Fjord, are lacking. It therefore remains uncertain whether the anomalously warm AW (~0.8 °C) observed on the Northern Lincoln Sea continental slope occasionally reached Victoria Fjord in the past. Nevertheless, the documented spatial and temporal variability of AW in the Lincoln Sea indicates that episodic warming cannot be excluded. On the Northeast Coast of Greenland, interannual variability in basal melt of Nioghalvfjerdsfjorden Gletsjer (79 N Glacier) has been shown to correlate with changes in AW temperature, demonstrating that relatively modest AW variability can influence glacier–ocean interactions elsewhere in Greenland32. Recent observations also indicate that cooling of AW in the Fram Strait can reduce submarine melt in Northeast Greenland, highlighting the sensitivity of glacier–ocean interactions to upstream AW variability33.
Ice margin geometry from radio-echo sounding and drone measurements
Radio echo-sounding profiles at the margin of C.H. Ostenfeld Gletsjer indicate a grounded calving cliff during the survey time, 21–26 August 2024 (Fig. 5), except for along a ~4 km stretch in the easternmost part where short floating extensions of about 0.4–0.9 km are present (Fig. 5 and Supplementary Fig. 2b). In the C.H. Ostenfeld Gletsjer central sector, the margin is grounded at a depth of ∼360 m and its subaerial face rises to ∼35 m above mean sea level (defined here relative to GEOID1634; Fig. 5). Grounding line depths vary along the margin and are generally shallower at the western section of C.H. Ostenfeld Gletsjer. The bed elevation derived from the RES profiles shows a poor match in places with the BedMachine Version 5 under-ice topography, particularly in the westernmost and easternmost sectors (Supplementary Fig. 6c).
Fig. 5: Radio-echo sounding (RES) and uncrewed aerial vehicle (UAV) observations at C.H.
The alternative text for this image may have been generated using AI.
Ostenfeld Gletsjer, constraining the ice margin during the GEOEO 2024 (August 2024). a RES data acquisition (red lines) at C.H. Ostenfeld Gletsjer 21st–26th August 2024. Backdrop is pansharpened Landsat 9 true colour satellite imagery acquired 23rd August 2024. The black line is the 2019 ice margin position5. Orange dots are RES-derived ice margin (21st–26th August 2024) based on termination of bright basal reflection in the radargrams. The yellow frame shows the location of Fig. 5b. Blue and pink lines represent the locations of radargrams shown in Fig. 5c, e, respectively. bUAV-derived digital elevation model of the ice surface on 27th August, 2024. Backdrop is pansharpened Landsat 9 true colour acquired on 23rd August 2024. UAV DEM is referenced to GGEOID1634, with elevations expressed relative to the local mean sea level datum. The map is rotated 33° clockwise relative to Fig. 5a. c Arctic DEM corrected RES data along the trunk of C.H. Ostenfeld (from A to B). Elevation correction performed using 2 m Arctic DEM mosaic (v4.1), so ice surface profile is not fully representative of the ice topography at the time of RES data acquisition. d Ice surface elevation profile across UAV-derived DEM with the resolution of 49 cm (relative to GGEOID16). Extent corresponds to Fig. 5e. e UAV DEM corrected RES data along the first ~2 km of the C.H (from C to D) Ostenfeld trunk. Extent corresponds to Fig. 5d. Note the correspondence between the end of bright bed reflection and the break in ice surface slope at the time of survey.
The height of the calving cliff along the central section of the eastern margin of C.H. Ostenfeld Gletsjer was determined in detail from topographic mapping by an uncrewed aerial vehicle (UAV) (Fig. 5). The UAV imagery additionally allows calving dynamics at C.H. Ostenfeld to be deduced from the characteristics of recently calved icebergs. In the vicinity of the terminus, there are many large, tabular icebergs (up to ∼3 × 0.7 km in size), the presence of which suggests a calving style associated with a floating terminus35. In addition, there are several icebergs closer to the terminus on the shallower western side of the glacier, taken to be the most recent calving events in this area. Here, there has been a so-called ‘bottom-out’ rotation of the newly calved icebergs, i.e. capsizing due to buoyancy adjustment. Such rotation has been previously observed in Greenland and described as slab capsize or buoyant flexure calving, thought to be indicative of glaciers nearing/at flotation35. These icebergs are up to 200 m long, suggesting they represent the full thickness of the glacier in this region. Together with an inspection of Landsat imagery dating back to 2002, when the main break-up began (Supplementary Fig. 5), these observations suggest that the terminus of C.H. Ostenfeld Gletsjer may develop a small floating extension during winter, which then disintegrates seasonally via the calving of tabular icebergs upon the break-up of ice mélange in spring/summer. This intermittent and spatially limited flotation contrasts with the former tens-of-kilometres-long, persistent ice tongue and underscores the transition to a fundamentally different terminus regime. After the removal of the small floating section, there is a shift in calving style to buoyant flexure calving, indicating a grounded terminus near flotation. This kind of seasonal shift is not unprecedented in Greenland. For example, this was suggested following UAV surveys at Store Glacier36. Furthermore, the Landsat imagery shows that by 2012 the margin was for the most part less than 1 km away from that of 2024, apart from along the western side, showing how there has been little net movement of the terminus in over a decade5.
Taken together, the radargrams and UAV surveys indicate that Victoria Fjord has undergone a long-term regime shift from hosting an outlet glacier ending with a tens-of-kilometres-long floating ice tongue to a glacier with a near-grounded calving cliff. Such a transition is comparable to the shift observed at Ilulissat when Sermeq Kujalleq evolved from ice-tongue calving to ice-cliff calving15.