1. Introduction
Global warming is resulting in unprecedented ice mass loss and the disappearance of glaciers worldwide (IPCC, Reference Masson-Delmotte, Masson-Delmotte, Zhai, Pirani, Connors, Péan, Berger, Caud, Chen, Goldfarb, Gomis, Huang, Leitzell, Lonnoy, Matthews, Maycock, Waterfield, Yelekçi, Yu and Zhou2021; the GlaMBIE Team, 2025). The ongoing changes of the cryosphere have led to rising sea level and an enhanced risk of geohazards, affecting ecosystems and biodiversity, the tourism industry and hydropower production (IPCC, Reference Pörtner, Masson-Delmotte, Zhai, Tignor, Poloczanska, Mintenbeck, Alegría, Nicolai, Okem, Petzold and Rama2019). Glacier loss also affects societies in many other ways, including the cultural heritage and national identity (e.g., Björnsson, Reference Björnsson2017; Howe and Boyer, Reference Howe and Boyer2025). At the turn of the twenty-first century, more than 300 glaciers in Iceland were catalogued and assigned with specific names (Sigurðsson and RS, Reference Sigurðsson and Williams2008). A follow-up survey conducted in 2017 revealed that 56 of the smaller glaciers had either disappeared or become inactive (Hannesdóttir, Reference Hannesdóttir2020), and according to the Iceland Glacier Newsletter in 2025, the number of vanished glaciers had increased to 67. The most widely renowned vanished glacier in Iceland is Okjökull (Ok glacier: 64°36′N, 20°52′W), which once covered the upper northern slope and the crater of the shield volcano Ok (1141 m a.s.l.) in western Iceland (Fig. 1). In a local TV program Landinn in September 2014 Oddur Sigurðsson declared that Okjökull did no longer meet the criteria for a glacier, as it had become too thin to flow under its own weight. It gained international attention in 2019 when a memorial plaque was installed to explain its fate to future generations (Howe and Boyer, Reference Howe and Boyer2024).
Iceland (a) and the location of Okjökull and Hofsjökull eystri (red triangles), the locations of two weather stations, from which summer temperatures shown in Figure 4, are marked with yellow plus symbols. On the larger maps the established maximum extent of Okjökull (b) and Hofsjökull eystri (c), in ∼1890 is depicted as bright green dashed lines, and the shaded area between the lines displays the area where uncertainty of the location is large and the minimum and maximum extent is estimated. Blue lines represent the extent in1945. Okjökull extent in 2021 (GNSS field survey by Tómas Jóhannesson, pers. comm.) and Hofsjökull eystri extent in 2024 are indicated with white areas. Flutes and striae identified on aerial photographs are indicated with brown lines and moraines are drawn on the maps as red lines.
The fate of Okjökull (Fig. 2) awaits several other glaciers in Iceland in the coming decades. These include some of the valley glaciers on Tröllaskagi, north Iceland, Kaldaklofsjökull (1.5 km2 in 2023) and Torfajökull (7.3 km2 in 2023) in south Iceland, as well as a few glaciers located east of Iceland’s largest ice cap, Vatnajökull. Hofsjökull eystri (Figs 1 and 2), located at 64°37′N, 15°03′W, is the second largest glacier in this complex, with Þrándarjökull being the largest (12.6 km2 in 2023). After considerations of criteria such as elevation range, size, climatological settings and accessibility of the rapidly shrinking glaciers in Iceland, the glaciological community in Iceland agreed to list Hofsjökull eystri on the Global Glacier Casualty List (Boyer and Howe, Reference Boyer and Howe2024), a website launched in Reykjavík in August 2024 that visualises recently lost and near-vanished glaciers.
(a) View to the southwest across Okjökull in 2003 (oblique aerial photograph by Oddur Sigurðsson). (b) View to the south over Hofsjökull eystri in 2006 (oblique aerial photograph by Snævarr Guðmundsson).
Hofsjökull eystri resides on an extensive 1200 m high mountain crest, with an altitudinal range of 960–1080 m a.s.l. Both Hofsjökull eystri and Okjökull are located at 64.4°N, and their elevation span was similar, but they are in different climatological settings; Hofsjökull eystri is close to the southeast coast, receiving high amounts of precipitation from the prevailing wind direction, whereas Okjökull is located in the rain shadow of Langjökull and Þórisjökull in west Iceland (Massad and others, Reference Massad, Petersen, Þórarinsdóttir and Roberts2020).
This study has three main objectives: to assess the Little Ice Age (LIA) maximum extent of Hofsjökull eystri and Okjökull; to trace their evolution through the 20th century until present, including Okjökull’s disappearance; and to project how long Hofsjökull eystri will survive in the coming decades, based on thinning rates observed during the past two decades.
2. Data
To establish and evaluate the maximum extent of Hofsjökull eystri and Okjökull at the end of the LIA, a map from Icelandic surveyor Björn Gunnlaugsson (Reference Gunnlaugsson1844) and a geological map prepared by the geologist Þorvaldur Thoroddsen (Reference Thoroddsen1901), as well as the descriptions provided by Þorvaldur Thoroddsen (Reference Thoroddsen1883, Reference Thoroddsen1895, Reference Thoroddsen1911) from his journeys to Hofsjökull eystri in 1882 and 1894, and the records from the land surveys launched at the turn of the 20th century by the Danish General Staff (DGS, later Geodætisk Institute) were used. The land surveys continued until 1940 (Böðvarsson, Reference Böðvarsson1996) and give indications of glacier extent, although glacier margins were not always updated between the survey years, they are used here to provide an estimate of the glacier changes during the first half of the 20th century.
To quantify changes in extent, area, elevation and volume since the mid-20th century, we used orthomosaics, satellite imagery, and a series of Digital Elevation Models (DEMs). The datasets are listed in Table 1. These include aerial-photograph DEMs and orthoimages of Hofsjökull eystri in 1946, 1967, 1990 (Belart, Reference Belart2020), a lidar DEM (2012), ArcticDEMs (Hofsjökull eystri 2024; Ok 2013, 2017, 2024), and DEMs generated for this study from aerial photographs of Ok (1960, 1978), ASTER imagery (2000) and SPOT5 imagery of Hofsjökull eystri (2003). In addition we use glacier outlines from previous studies for Hofsjökull eystri in 1946, 1967, 1990 and 2012 (Belart, Reference Belart2020) and for Okjökull in 1945 (Hannesdóttir, Reference Hannesdóttir2020).
Data used to obtain DEMs and orthoimages in this study. The aerial photographs used are from the Natural Sciences Institute of Iceland (NSII), formerly the National Land Survey of Iceland (www.Lmi.Is). The spatial resolution of the orthomosaic (GSD Ortho) and DEM (GSD DEM) is presented as the Ground Sampling Distance. See data availability statement for further details on data access.
Certain steps in the processing of the stereo images required the use of external DEMs as a reference. For this purpose we also used the Copernicus Digital Elevation Model (GLO30), as well as the IslandsDEM (https://dem.gis.is), a seamless, country-wide mosaic of ArcticDEM (Porter and others, Reference Porter2022) co-registered to the Copernicus Digital Elevation Model (GLO30) and to local lidar data (Jóhannesson and others, Reference Jóhannesson2013). Finally, we also used Bing Maps (©2025 Microsoft Corporation), visualised in QGIS (Version 3.4), for parts of the photogrammetric processing.
To measure the current thickness and map the bedrock topography of Hofsjökull eystri a low frequency (5 MHz) radio echo sounding (RES) survey was conducted in May 2025. The measured tracks extend a total of 13 km and they are shown in Fig. 3f.
The glacier thickness (color coded images) and extent (red lines) of Hofsjökull eystri (a–f and upper color scale) and Okjökull (g–k and lower color scale). Underlying elevation contour maps and relief images show the ice-free areas in 2024 from ArcticDEM (Porter, Reference Porter2022) and bedrock beneath Hofsjökull eystri interpolated from traced bed reflections in RES-profiles (locations shown with black lines in f). Glacier area, volume, maximum and mean thickness are given in the corner insets of a–k. Panel l shows the area (red) and volume evolution (blue) of the two glaciers, including two projections on how Hofsjökull eystri may develop in the near future (dashed and dotted lines).
3. Methods
3.1. Tracing the little ice age extent
High-resolution aerial imagery was used to map the outermost glaciogeomorphological features associated with the LIA maximum (Eythorsson, Reference Eythorsson1935; Thorarinsson, Reference Thorarinsson1943; Grove, Reference Grove2004; Hannesdóttir and others, Reference Hannesdóttir2020), including moraines, striations, and flutes (Fig. 1). For both glaciers, the traces of the LIA maximum glacier extent are vague because the subglacial surface is mainly solid bedrock with limited loose material and the glaciers were thin. Although relatively little is known about their extent during the 19th century, it is reasonable to assume that both glaciers were at or near their historical maximum extent toward the end of that century, like the majority of glaciers in Iceland (Hannesdóttir and others, Reference Hannesdóttir2020). In this study, a “reinterpreted” glacier extent is established and evaluated based on mapping of flutes, striations and moraines, supported by historical descriptions (Thoroddsen, Reference Thoroddsen1883, Reference Thoroddsen1895, Reference Thoroddsen1911).
3.2. Aerial photographs and satellite stereoimages
The processing of the ASTER and SPOT5 DEMs was carried out using the Ames StereoPipeline v 3.6 (Beyer and others, Reference Beyer, Alexandrov and McMichael2018; Shean and others, Reference Shean2016; Alexandrov and others, Reference Alexandrov2025). The processing of ASTER was done using the following routines in Ames StereoPipeline: (1) aster2asp, (2) mapproject, (3) parallel_stereo, and (4) point2dem. The processing of SPOT5 was done using the following routines in Ames StereoPipeline: (1) add_spot_rpc (2) mapproject, (3) parallel_stereo, and (4) point2dem. In both cases, the processing required the use of a reference DEM, which was the Copernicus GLO30. The DEM was co-registered to the lidar (for Hofsjökull eystri) and to the October 2024 ArcticDEM (for Ok). We used the co-registration algorithm of Nuth and Kääb (Reference Nuth and Kääb2011), implemented in Python in demcoreg (https://github.com/dshean/demcoreg, Shean and others, Reference Shean2016). The horizontal shifts calculated were then applied to the orthoimage obtained.
The processing of the orthoimages of Okjökull from 1960 and 1978 was done using Agisoft Metashape (Version 2.1, 2025). A series of Ground Control Points (GCPs) were hand-picked from a recent orthomosaic of Iceland from Bing Maps (@2025 Microsoft Corporation). The elevations were extracted from the IslandsDEM. The Bing orthomosaic can have horizontal uncertainties up to 10 m, and the elevations from the IslandsDEM are accurate within 0.5 m. We carefully configured enough flexibility in the GCPs to account for these uncertainties during the bundle adjustment, and after deleting and relocating GCPs with large residuals, we ran the processing in two iterations.
3.3. Glacier outlines
The glacier outlines presented in this study are all shown in the Supplementary Material section (Fig. S1). These include previously published glacier outlines of Hofsjökull in 1946, 1960, 1990, and 2012 (Belart and others, Reference Belart2020), and the outlines of Okjökull in 1945 (Hannesdóttir and others, Reference Hannesdóttir2020). The glacier margin of Hofsjökull eystri in 2003 and 2024 and Okjökull in 1960, 1978, 2000, 2013, and 2017 were delineated specifically in this study to be coherent with the DEMs presented. These vary only slightly from previously published data (Helgadóttir, Reference Helgadóttir2017; Hannesdóttir and others, Reference Hannesdóttir2020). This was done based on orthorectified aerial or satellite images (lacking in the case of the DEMs from the ArcticDEM archive), hill-shade and contour map representations of the DEMs, and elevation change maps (e.g. Fischer and others, Reference Fischer, Schwaizer, Seiser, Helfricht and Stocker-Waldhuber2021) relative to the 2024 DEMs of Ok and Hofsjökull eystri (Figs S2 and S3). The elevation change maps were particularly useful to distinguish winter snow patches, typically only a few meters thick, from the actual glacier (see Fig. S2b,c,e,f).
3.4. Glacier thickness maps
To calculate glacier thickness maps of Okjökull and Hofsjökull eystri at different times (Fig. 3), bedrock DEMs are required. In the case of Okjökull (Fig. 3g–k), the bedrock is assumed to correspond to the latest DEM in October 2024, despite some snow and ice patches covering less than 5% at that time, and unlikely to exceed more than a few meters in thickness. In the case of Hofsjökull eystri, the bedrock of the current glacier was mapped with low-frequency (5 MHz) radio echo sounding (RES) in May 2025. The profile locations are shown in Fig. 3f. The processing of the RES profiles and bedrock tracing is identical to previous bedrock mapping by Magnússon and others (e.g., Masson-Delmotte and others, Reference Masson-Delmotte, Masson-Delmotte, Zhai, Pirani, Connors, Péan, Berger, Caud, Chen, Goldfarb, Gomis, Huang, Leitzell, Lonnoy, Matthews, Maycock, Waterfield, Yelekçi, Yu and Zhou2021). The bedrock DEM was calculated using kriging interpolation in Surfer 13 (© Golden Software, LLC) using the elevation of traced bedrock reflections as well as the elevation at the glacier margin in 2024 (ArcticDEM) as inputs. The resulting bedrock DEM was mosaicked with the DEM of the glacier-free areas in 2024. This mosaic was used for calculating the ice thickness maps of Hofsjökull eystri (Fig. 3a–f). The glacier volumes represented in Fig. 3 are derived with integration of the presented thickness maps.
3.5. Glacier thinning rates and projected future development of Hofsjökull eystri
From the difference between DEMs we mapped the elevation change rate of the glaciers (Fig. S4). These maps were used to calculate the average lowering rate for each period defined by the dates of the DEMs (Fig. 4). We also use these maps to project the future development of Hofsjökull eystri (Fig. 3l). This is done for two scenarios, assuming the lowering rates of (a) the period 2012–2024 (Fig. S5) and (b) the period 2003–2012 (Fig. S6). The ice thickness map for 2024 was used to determine the timing of ice disappearance at a given location. For both scenarios the glacier is lowering in all grid points, but the lowering was faster in 2003–2012 (Figs 4 and S6).
The glacier area loss rate in percentage per year (red lines) relative to the maximum extent (∼1890) for (a) Hofsjökull eystri and (b) Okjökull, for periods defined by dates of presented glacier outlines (Figs 3 and S1). Light-red boxes indicate uncertainty. The mean lowering rate of the glacier surface (black lines) calculated over the common glacier area of two consecutive DEMs (Fig. S4). Gray boxes indicate uncertainty. The measured temperature (blue lines, Björnsson and others, Reference Björnsson2018) averaged over the summer months (May–September) at stations operated since the 19th century closest to Hofsjökull eystri (Teigarhorn) and Okjökull (Stykkishólmur) (see Fig. 1 for locations). Dash-line indicate values for each summer, bold solid line the average summer temperature for the same periods as presented for the area loss and lowering rates.
3.6. Uncertainties
We assume an uncertainty of 5% (Paul and others, Reference Paul2017) for the observed area of Hofsjökull eystri and Okjökull in 1945/46 and later. For errors contributing to the uncertainty of the derived volumes, we assume ±1 m as possible bias in elevation change maps relative to the 2024 DEMs of Hofsjökull eystri and Okjökull. This can be considered a generous estimate compared to geostatistical uncertainty estimates for similar datasets (Magnússon and others, Reference Magnússon, Belart, Pálsson, Ágústsson and Crochet2016a; Belart and others, Reference Belart2020). Additionally, the uncertainty of the 2024 volume of Hofsjökull eystri derived from RES (inclusive in all volume estimates of Hofsjökull eystri) is estimated to be 10%. This is higher than the 5% uncertainty typically assumed for similar datasets obtained with the same survey approach (Magnússon and others, Reference Magnússon2016b; Magnússon and others, Reference Magnússon, Pálsson, Jarosch, van Boeckel, Hannesdóttir and Belart2021). However, since the measured maximum ice thickness of Hofsjökull eystri is only 55 m, the derived volume is more sensitive to possible elevation offsets than for a substantially thicker glacier, making 10% a more reasonable uncertainty estimate.
4. Results
The ice thickness of the two glaciers during the time span of available DEMs, along with a graph extending the volume and area, portray the evolution from the 1890s (Fig. 3). The mean lowering rate (based on the elevation change rates presented in Fig. S4) and the area loss rate relative to the maximum extent, are compared to the summer temperature at two weather stations (see location in Fig. 1) close to each glacier (Fig. 4) confirming a clear correlation between the glacier evolution and temperature.
4.1. Hofsjökull eystri LIA extent
Based on the depicted terminal moraines, the elevation span of Hofsjökull eystri at its maximum extent ranged from ∼710 to 1180 m a.s.l. The terminus extended east to the Hofsvötn lakes (Fig. 1). The end moraine lies across the larger lake, and a series of moraine ridges are evident along its western shore, indicating repeated advances following the onset of the glacier retreat. In the late 19th century the margin of Hofsjökull eystri merged with the Tungutindajökull glacier in the saddle south of the mountain. The extent cannot be precisely determined without further field investigation, so the shaded area is large at the southern end, as well as at the northwestern edge of the glacier. The area span calculated from the inner and outer boundary of the shaded region in Fig. 1 is 12.6–14.6 km2; this range is presented as the area uncertainty (7%) shown in Fig. 3l.
4.2. Hofsjökull eystri post LIA
Hofsjökull eystri was initially surveyed in 1940 and then again in 1946. A comparison of the maximum extent in the 1890s to the mid-20th century extent reveals that the ice cap had decreased in size by 40–50%. At that time, the glacier had retreated approximately 700 m from the Hofsvötn lakes. On its southwest side an outlet glacier known as Morsárjökull (now vanished) retreated by ∼500 m between the 1890s and 1940, and on the northwest margin the glacier retreated ∼200 m. Böðvarsson’s 1940 map (NLSI, 2025) shows that the glacier had thinned significantly since its maximum extent, and by that time, no ice remained on the highest peak or the southern slopes of the mountain.
In 1946 the area had reduced to 7.1 ± 0.4 km2 and the volume to 0.34 ± 0.01 km3 with a maximum thickness of ∼130 m (Fig. 3a). In 1967 its maximum thickness had reduced by ∼15 m, and its volume declined to 0.25 ± 0.01 km3 (Fig. 3b), and it continued to lose area, volume, and thickness. Hofsjökull eystri was entered into the Global Glacier Casualty list (Boyer and Howe, Reference Boyer and Howe2024) in 2024. Its area has diminished to 2.1 ± 0.1 km2 and volume to 0.039 ± 0.004 km3, corresponding to 30% and 11% of its 1946 area and volume, respectively. A majority of both the area and volume change relative to 1946 took place after 1990 (Fig. 3l).
4.3. Hofsjökull eystri future evolution
Two near-future scenarios for Hofsjökull eystri based on the observed lowering rate are depicted in the Supplementary Material (Figs S5 and S6). The elevation loss in the period 2012-2024 is maintained until 2065, when the area may have reduced to 0.03 km2 and the maximum thickness to 5 m (Fig. S5). If the lowering rate observed in 2003–2012 is continued the loss is faster; in 2050 the area is projected to have reduced to 0.04 km2 and maximum thickness to 9 m (Figs 3l and S6). Based on these scenarios, we conclude that in current climate conditions it is likely that Hofsjökull eystri will follow in the footsteps of Okjökull and vanish completely within the next 30–45 years.
4.4. Okjökull LIA extent
The LIA extent of Okjökull is based on the outermost glacial features (Fig. 1). The geomorphological records are vague on all sides, but a distinct end moraine only remains to the northeast. The presented LIA maximum extent spans an area range of 9.1–10.7 km2 from the inner to the outer boundary depicted in Fig. 1 and shown as area uncertainty in Fig. 3l.
4.5. Okjökull post LIA
The outlines of Okjökull surveyed in 1910 and 1939 show the retreat of the glacier during this period. The 1910 map indicates that the glacier, or possibly a persistent snowfield, extended approximately 500 m south of the crater, while in the 1939 map the glacier margin is at the crater rim. The eastern part of the glacier is not depicted on the 1910 map, rendering its full extent uncertain. Furthermore, the 1910 map suggests that by that time Okjökull at least extended as far north as it did in 1945 or even further. It may therefore have remained at its end moraines throughout the early 20th century. However, based on the contour lines, the glacier appears to have terminated at an elevation of approximately 900 m. In contrast, by 1945, the north margin remained at 850 m a.s.l., while its lateral margins were confined within the 1910 boundaries. The 1910 map also indicates that the maximum elevation of the glacier surface reached an elevation of 1200 m a.s.l. at the northern rim of the crater, suggesting an ice thickness of approximately 40–80 m. By 1939, the glacier surface had lowered to around 1180 m, and by 1945, to ∼1120 m north of the crater (NLSI, 2025). Okjökull had reduced its size to 7.5 ± 0.4 km2 in 1945 and had reduced to almost half that by 1960 (4.4 ± 0.2 km2) (Fig. 3g). At that time the volume of Okjökull was 0.128 ± 0.004 km3, with a maximum ice thickness of 69 m.
Since the 1960s, Okjökull has continued its decline toward complete disappearance (Fig. 3). In 2013, a year before the glacier was declared vanished, its maximum thickness had decreased to 18 m, its area to 0.47 ± 0.02 km2, and its volume to 0.0022 ± 0.0001 km3 (Fig. 3j), less than 2% of its 1960 volume. In 2017 remnants of ice still prevailed in several isolated patches covering in total ∼0.1 km2. These patches contained at that time up to 11 m thick ice, with a total volume of ∼0.0003 km3 (Fig. 3k). Part of these ice patches still prevailed in 2021 and covered an area of ∼0.03 km2 (Outlined with GNSS field survey; Tómas Jóhannesson, pers. comm. 2021), in 2 trips in August and September 2025 small remnants of ice patches in same locations could be found, so complete disappearance of the last remnants of Okjökull has to date not been confirmed.
5. Discussion
Uncertainty remains regarding the LIA maximum extents of Hofsjökull eystri and Okjökull, due to vague or poorly preserved glacial remnants. As a result, surface area estimates are approximate. Gunnlaugsson’s (Reference Gunnlaugsson1844) and Thoroddsen’s (Reference Thoroddsen1901) maps show extensive ice coverage at Okjökull, though likely exaggerated due to persistent snowfields or firn. The outlines presented in Fig. 1 are reconstructed from available data, including historical maps, and geomorphological evidence. Evidence suggests Hofsjökull eystri exceeded 12 km2 at its maximum extent, and written contemporary descriptions by Th. Thoroddsen in 1882 and 1894 indicate it likely occurred in the 1890s (Thoroddsen, Reference Thoroddsen1883, Reference Thoroddsen1895).
Geomorphological evidence suggests Okjökull was relatively thin and slow flowing, with poorly defined moraines and striations except along its northern flank. Thoroddsen (Reference Thoroddsen1911) described it in 1898 as a substantial glacier dissolving into snow patches extending well below the firn line, but he noted that outlet glaciers had never been observed there. It is possible that persistent winter snow and long-lasting firn fields may have contributed to an overestimation of the glacier’s extent. The ∼1890 extent is therefore shown as a range on the diagram in Fig. 3l.
Okjökull’s shield-shaped topography resulted in greater ice thickness at the center, where striated and fluted beds indicate movement. The ice gradually thinned toward the glacier’s margins and lost its capacity to deform and move. Peripheral ice became largely stagnant. As thinning progressed, the crater ice mass detached from the main body. By the mid-20th century, the crater rim had emerged, and by 2007, a crater lake was identified (Malmquist and others, Reference Malmquist, Ingimarsson, Ingvason, Stefánsson and Þ2013).
The relative area loss and the mean lowering rate of both glaciers are shown in Fig. 4 (based on maps in Fig. S4). Both show strong correlation with the summer temperatures at the nearby stations. It is however clear that changes in summer temperature do not fully explain variations in the area loss and thinning rate of the two glaciers. For Hofsjökull eystri the lowering rate is higher in 2003–2012 than in 2012–2024, even though the averaged temperature at Teigarhorn over same periods is slightly higher in the latter period. Hofsjökull eystri is located in a wetter climate than Okjökull. Average annual precipitation in 1981–2010 was close to 3000 mm a−1 at Hofsjökull whereas at Okjökull it was close to 2000 mm a−1 (Björnsson and others, Reference Björnsson2018). This disparity in precipitation likely contributed to greater accumulation, thereby prolonging the survival of Hofsjökull eystri in contrast to the now-vanished Okjökull.
The fact that Hofsjökull eystri has persisted longer than Okjökull may be attributed to differences in topography in addition to the prevailing weather patterns. Although Hofsjökull eystri and Okjökull resided at similar elevations and latitudes, the topography is different. Both glaciers rest on bedrock, but while Okjökull covered a convex substratum, Hofsjökull resides partly in shallow troughs or cirques. The map series of Hofsjökull eystri (Fig. 3a–f) illustrates that the bed topography has a significant role in its future prospects. The radio-echo sounding data reveal that the glacier is thinnest along the rim of the underlying troughs and currently measuring only 10–20 m in thickness. At present, the remaining ice is on average only 18 m thick and therefore either stagnant or close to stagnation, lacking the dynamic flow characteristic of an active glacier. Based on current trends, we project that in 10–15 years the glacier will have thinned so much that it has broken up into three ice patches (Supplementary Figs S5 and S6).
The maximum thickness of Hofsjökull eystri in 2024 was merely 55 m. The uncertainty due to the assumptions of negligible surface mass balance-elevation feedback and changes in ice dynamics is therefore small in comparison to the uncertainty related to the inherent assumption of our projections that the climate conditions in the coming decades will be similar to those in 2003–2024. Further warming will result in Hofsjökull eystri likely disappearing faster than predicted here.
6. Conclusion
By utilising various datasets, we provide an assessment of the evolution of Hofsjökull eystri and Okjökull from their late-19th-century maximum extent to the present. Although uncertainty remains regarding their precise extent and area around 1890, Hofsjökull eystri is estimated to have covered 12.6–14.6 km2, and Okjökull 9.1–10.7 km2.
More reliable data reveal that in the mid-20th century the area of Hofsjökull eystri was 7.1 ± 0.4 km2, with a volume of 0.34 ± 0.01 km3 and a maximum thickness of ∼130 m. Okjökull, at that time, measured 7.5 ± 0.4 km2 in area, 0.128 ± 0.004 km3 in volume, and reached a maximum thickness of 60 m.
Both glaciers continued to decline in the latter half of the 20th century. By 2014, Okjökull had disappeared, leaving only small isolated ice patches. Hofsjökull eystri had diminished to 2.1 ± 0.1 km2 in area and 0.039 ± 0.004 km3 in volume. Results from a 2025 radio-echo sounding (RES) survey and the observed thinning rates indicate that Hofsjökull eystri is likely to meet the same fate. If thinning continues at a similar rate as in the past two decades Hofsjökull eystri will completely vanish in the period 2055–2070.
The fate of the two glaciers in Iceland reflects a global trend: glaciers are disappearing at an alarming rate. Documenting the evolution of these glaciers, their names, extents and geomorphic legacy since the Little Ice Age is essential for understanding landscape change and the broader impacts of climate change.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/aog.2025.10026.
Data availability statement
The aerial photographs used in this study are available through the webmap https://loftmyndasja.gis.is and the https://www.lmi.is/is/vefsjar/korta-og-loftmyndasofn/loftmyndasafn. The ASTER data is available through https://www.earthdata.nasa.gov/. The SPOT5 data is available through https://regards.cnes.fr/. The ArcticDEM data is available through https://www.pgc.umn.edu/data/arcticdem/. The processed DEMs and orthomosaics of this study are available upon request to the authors.
Acknowledgements
The authors greatly appreciate the constructive comments by the anonymous reviewers and the editors of “Annals of Glaciology” that helped to improve the paper. S.G. wants to thank the South East Iceland Nature Research Center for the financial commitment, E.M., H.H. and G.A. thank Horizon Europe funding under the ICELINK Grant Agreement No. 101184621 (Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union) Landsvirkjun supported the field survey on Hofsjökull Eystri in spring 2025. Sveinbjörn Steinþórsson and Andri Gunnarsson are thanked for their work on that field survey.
Open access
