3.1. SGI1973 and corresponding DEM

The first complete Swiss Glacier Inventory (SGI1973) was compiled from aerial photographs taken in September 1973. Glacier outlines were manually delineated using stereo-photogrammetric interpretation and transferred onto topographic map sheets at a scale of 1 : 25 000 (Müller and others, Reference Müller, Caflisch and Müller1976). The glacier outlines were digitised and later published in a consistent digital format (Maisch and others, Reference Maisch, Wipf, Denneler, Battaglia and Benz2000; Paul, Reference Paul2004). The inventory was subsequently revised and quality-checked for consistency and homogeneity of the glacier-specific identifiers, resulting in an updated, standardised version of SGI1973 (GLAMOS, 2020a). As the inventory was originally based on analogue data, some uncertainties remain. These include potential errors due to misinterpretation of outlines, inaccuracies during digitisation and imprecise georeferencing of map sheets (Paul, Reference Paul2004). There is no reliable overall uncertainty estimate for SGI1973; however, based on the descriptions of possible errors by Paul (Reference Paul2004), the uncertainty in total glacier area can be approximated at about ±5%.

Glacier delineation in the SGI1973 followed the guidelines outlined in the UNESCO manual Perennial Snow and Ice Masses (UNESCO, 1970). According to these recommendations, all perennial snow and ice bodies were to be included if they could be distinguished from seasonal snow and surrounding terrain in aerial imagery. This included active glaciers, glacierets, perennial snow patches and ice-cored moraines, as well as inactive ice bodies such as aprons above bergschrunds or debris-covered disconnected dead ice in front of the glacier terminus. Although the UNESCO guidelines did not prescribe a strict minimum area, a practical lower limit of approximately 0.01 km² emerged due to the mapping scale and limitations in visual interpretation. However, this threshold was not applied uniformly, and many very small glaciers below this size threshold were still mapped and included. In addition, many of these small ice patches did not receive a unique identifier but were instead assigned the ID of a neighbouring larger glacier. Strictly speaking, they are thus not considered individual glaciers and would only be referred to as vanished if their parent glacier also completely vanished.

For topographical analysis, the SGI1973 was combined with the DHM25 digital elevation model provided by swisstopo (swisstopo, 2005). This 25×25 m DEM was derived from interpolated contour lines of the 1 : 25 000 topographic maps and includes spot heights and break lines (Rickenbacher, Reference Rickenbacher1998). The first version of this terrain model refers to the years between ca. 1960 and 1990 (Fischer and others, Reference Fischer, Huss and Hoelzle2015), with most acquisitions around 1985. As glacier change in the Swiss Alps was observed to be relatively limited during this time period, the surface elevations can be assumed to be consistent with the outlines of the SGI1973 (e.g. Paul, Reference Paul2004).

3.2. SGI2016

The Swiss Glacier Inventory 2016 (SGI2016) represents the latest fully updated nationwide dataset of glacier outlines and attributes referring to the years 2013–18 (Linsbauer and others, Reference Linsbauer2021). It was compiled using high-resolution aerial imagery (swisstopo, 2020a) and digital elevation models (swisstopo, 2018), in collaboration with the Federal Office of Topography (swisstopo) and Glacier Monitoring in Switzerland (GLAMOS), combining topographical and glaciological knowledge. The inventory maintained the hydrologically-based ID system introduced with the SGI1973 and includes standardised attribute data such as area, elevation range, aspect, slope and debris cover (GLAMOS, 2020b). The overall uncertainty in total glacier area has been estimated at ±2.3% (Linsbauer and others, Reference Linsbauer2021).

The glacier outlines were mapped according to the definition by Cogley and others (Reference Cogley2011), where a glacier is defined as 'a perennial mass of ice, and possibly firn and snow […] showing evidence of past or present flow'. In addition, the SGI2016 applied a clearly defined set of inclusion criteria. Only perennial ice bodies larger than 0.01 km² were considered (Leigh and others, Reference Leigh, Stokes, Carr, Evans, Andreassen and Evans2019), and evidence of glacier-like flow structures was required. Very small or inactive glacierets, snow patches and ice remnants without dynamic behaviour were excluded. Particular attention was paid to debris-covered glacier areas, which were consistently delineated based on visual appearance, surface features and additional reference data such as elevation differences and surface classification from swisstopo (swisstopo, 2020b). Generally, SGI2016 outlines lie within SGI1973 boundaries, as expected with glacier retreat. In some rare cases, however, SGI2016 shows slightly larger glacier extents, mostly in the accumulation area. In 1973, snow and firn present at high elevation during the summer season made it difficult to distinguish glaciers from seasonal snow, leading to locally underestimated outlines, whereas the more recent imagery with reduced snow cover allowed clearer delineation up to the highest reaches.

3.3. Vanished glacier polygon layers

To identify glacier areas that were present in the SGI1973 but no longer existed in the SGI2016, the two inventories were overlaid and polygons from SGI1973 were selected that did not intersect with any polygon in SGI2016 (inverted spatial join). The resulting polygons, however, do not directly correspond to the total number of vanished glaciers. Polygons smaller than 0.01 km² were excluded, as they would not have been part of SGI2016 even if some ice remained. In addition, some small glacier polygons without an individual ID but attributed to the same ID as a neighbouring larger glacier were also excluded from the final count. Finally, polygons with the same glacier identifier were merged into single entities, resulting in a refined layer of vanished glaciers between the SGI1973 and SGI2016 used for all subsequent analyses (Fig. 2). The uncertainty in the vanished glacier area was estimated using Gaussian error propagation from the reported uncertainties of the SGI1973 (±5%) and the SGI2016 (±2.3%), resulting in a combined uncertainty of approximately ±6%.

Figure 2. Detailed views illustrating glacier area loss and the methodology used to identify vanished glaciers. (a) Close-up of the triple watershed region between the Rhine, Rhone, and Po catchments, showing substantial glacier area loss between SGI1973 and SGI2016 (dark blue areas). (b) Methodological classification of vanished glacier polygons: green polygons correspond to glacier areas from SGI1973 that share an ID with glaciers still existing in the SGI2016, yellow polygons represent ice bodies smaller than the SGI2016 minimum size threshold of 0.01 km², pink polygons indicate vanished glaciers, i.e. ice bodies with a unique ID in 1973 that are no longer present in the SGI2016.

4.1. Number and area of vanished glaciers

The 3051 glacier polygons contained in the SGI1973 correspond to 2732 distinct glacier entities when dissolved by SGI-ID (1311 km²). Excluding polygons smaller than 0.01 km² leaves 2379 glacier entities (1309 km²). In comparison, SGI2016 includes 1834 polygons, representing 1400 glacier entities with unique identifiers (961 km²). This corresponds to an overall area change of −350 km² or −26.7% (−0.6% a^–1) between 1973 and 2016, as reported by Linsbauer and others (Reference Linsbauer2021).

The inverted spatial join between SGI1973 and SGI2016 initially yielded 1652 polygons (corresponding to 1468 SGI-IDs) from SGI1973 that do not spatially overlap with any glacier polygon in the SGI2016. After applying the 0.01 km² threshold (excluding 506 polygons) and removing 84 polygons belonging to still-existing glacier IDs, the resulting dataset (Section 3.3, Figs. 1 and 2) contains 1019 vanished glaciers, covering a total area of 47±3 km². This accounts for 13% of the total area loss reported between 1973 and 2016.

The 1019 glaciers that vanished between the SGI1973 and the SGI2016 are unevenly distributed across both size classes and major river catchments (Table 1). The vast majority of vanished glaciers were small, with nearly 90% having an area of less than 0.10 km². In terms of area, however, the largest contributions to glacier loss came from the intermediate size classes. Glaciers between 0.05 and 0.20 km² together accounted for over 50% of the total area lost, despite representing only about 25% of the vanished glaciers in terms of their number. In contrast, the smallest glaciers (0.01–0.05 km²), although numerous, contributed less than 40% of the total area loss.

Table 1 Number and total area of vanished glaciers in Switzerland between the SGI1973 and the SGI2016, categorised by glacier size and main river catchments. Percentages are given relative to the total number and area of vanished glaciers in each catchment.

The regional distribution of glacier loss highlights distinct differences in both magnitude and character. The Rhine catchment experienced the highest number of vanished glaciers (423), contributing 18.0 km² to the total glacier area reduction between the SGI1973 and SGI2016 in that hydrological basin. In relative terms, the Po (58%) and Danube (55%) catchments show the highest rate of vanished glaciers, underlining the high vulnerability of small and peripheral ice bodies in these regions. In contrast, the Rhone basin—which is dominated by large valley glaciers—exhibits the highest overall area loss between 1973 and 2016 (162.7 km²), but only 10% (16.0 km²) of this area change can be attributed to vanished glaciers, reflecting the resilience of large ice masses. In the Swiss part of the Adige catchment, no glaciers vanished during the observation period, since no glaciers were present in either SGI1973 or SGI2016, highlighting the pronounced variability between the major river basins. At the scale of the Swiss Alps, 13% (47±3 km²) of the total glacier area loss between SGI1973 and SGI2016 is due to vanished glaciers. This underlines their substantial role in the overall decline, despite their small individual area. Because vanished glaciers no longer appear in later inventories, studies that consider only glaciers present in both inventories may overlook their contribution to total glacier loss (see, e.g. Parkes and Marzeion, Reference Parkes and Marzeion2018). This can lead to a systematic underestimation of total glacier change.

4.2. Locations and examples of glacier disappearance

Most of the vanished glaciers are located in regions where peak elevations lie well below 4000 m and extensive glacier systems are absent. These areas include sections along the main Alpine divide between the Rhone and Po (Ticino) catchments, the Rhine–Po (Ticino) divide, as well as the inner-Alpine regions of Grisons and along the Rhine–Danube (Inn) watershed (see Fig. 1). These mountain ranges are not among the highest in the Swiss Alps but play a key role in separating major drainage basins. In the 1973 inventory, many small glaciers in these regions were still mapped, although they contributed only marginally to the overall glacierised area (Fig. 3). Due to their limited extent, such ice bodies were particularly prone to complete disappearance over the following decades.

Figure 3. Close-up to three exemplary regions illustrating glacier changes and prominent cases of glacier disappearance using swisstopo aerial images from 1973 (top) overlain with SGI1973 glacier outlines (green), and 2016 (bottom) with outlines from both SGI1973 (green) and SGI2016 (blue). The examples are located in the regions of (a) Piz da l’Aua, (b) Zapporthorn, and (c) Piz Jenatsch (see Figure 1 for location).

To illustrate the characteristics of vanished glaciers, we present three representative regions (Fig. 3), each centered around a summit at approximately 3000 m a.s.l.: Piz da l’Aua (2992 m), Piz Jenatsch (3240 m) and Zapporthorn (3142 m) (see Fig. 1 for peak locations). All sites are located along the main Alpine divide and were still surrounded by several glaciers in 1973, particularly on north-facing slopes.

The region around Piz da l’Aua (Fig. 3a) underwent particularly pronounced changes. By 2016, nearly all glacier ice had vanished. Only a small, heavily debris-covered remnant of Vadret da Lischana remained and was still included in SGI2016. This ice body is barely recognizable as a glacier and marks the final stage of deglaciation. Since 1973, the landscape has undergone a fundamental transformation, now dominated by bare rock and scree surfaces and alpine lakes. Several glaciers, e.g. Vadret da Rims and Vadret d’Immez, that were still mapped in 1973 have completely vanished and are now part of the vanished glacier inventory.

In the Zapporthorn region (Fig. 3b), the shaded northern flank remained glacierised in 2016, although with substantial area loss. In contrast, the south- and east-facing slopes experienced more dramatic changes. Most glaciers on this side, for example, Ghiacciaio de Mucia, have completely vanished.

The northern flank of Piz Jenatsch, which was glacierised in 1973, is now entirely ice-free (Fig. 3c). Vadret d’Err, which previously covered this slope, had an area of 0.55 km² in SGI1973. This glacier, located in the Rhine catchment, is the largest glacier to have completely vanished between 1973 and 2016. The remaining ice patches in 2016 were too small and fragmented to meet the SGI2016 inclusion criteria.

The largest vanished glacier in the Rhone basin was Mettligletscher with an area of 0.31 km² in 1973 (Fig. 2a). In the Po basin, the largest vanished glacier was Ghiacciaio de Mucia (0.36 km²) south-east of Zapporthorn (Fig. 3b) and Vadret d’Albris in the Danube basin (0.42 km²). Together with Vadret d’Err in the Rhine basin (see above), these represent the four largest vanished glaciers in Switzerland (see Fig. 1 for locations of the four largest vanished glaciers). The complete demise of these glaciers with an area of almost 0.5 km² in the 1970s highlights the vulnerability of significant ice bodies if they are located on the slopes of rather low-elevation summits close to the climatic equilibrium line altitude.

4.3. Topographic influences

To investigate the characteristics of the glaciers vanished in the Swiss Alps between 1973 and 2016 and the influences on their disappearance, we analysed our inventory with respect to different topographical factors (Fig. 4). All 1019 vanished glaciers were split into four predefined classes of (i) their area according to the SGI1973, (ii) their median elevation, (iii) their mean surface slope and (iv) their dominant aspect. In the same way, we also classified all glaciers larger than 0.01 km² contained in the SGI1973. This allows a quantification of the relative loss of glaciers in classes of the above-mentioned variables with respect to the total initial glacier population. The complete loss of individual glaciers is discussed in terms of the number per class, as well as the total area of the vanished glaciers.

Figure 4. Comparison of the count (grey bars, black numbers) and the area (blue bars and numbers) of glaciers vanished between 1973 and 2016 in four classes with respect to the initial glacier cover. The relative share of vanished glaciers is classified for (a) area, (b) median glacier elevation, (c) mean slope, and (d) dominant aspect.

When considering glaciers with an area between 0.01 km² and 0.05 km² in the year 1973, we note that 74% of them have vanished, relative to their original number (Fig. 4a). This indicates that the effects of climate change of the last four decades have resulted in the demise of all but a quarter of these very small glaciers. For larger glacier size classes, the relative loss is much more limited and accounts for only 3% of glaciers with an area above 0.2 km² by 1973. The total number of glaciers has been reduced by about 40% throughout most classes of median elevation (Fig. 4b). Only glaciers whose median elevation was above 3100 m a.s.l. were lost somewhat less frequently. Interestingly, however, vanished glaciers with a very low median elevation (<2500 m a.s.l.) contributed more to the overall area loss of the corresponding class. Similarly, for mean slope, no clear dependency is recognisable with respect to the fraction of vanished glaciers: for both gently-sloping and steep glaciers, 30–40% of the initial population has vanished (Fig. 4c). Vanished glaciers, however, make up for a non-negligible share of the initial area (18%) for steep glaciers, while they comprise a negligible share (1%) of glaciers with a mean slope below 20 degrees. This is explained by the concentration of the overall glacier area in large and flat glaciers, whereas steep glaciers are often small. A very clear relation of the share of vanished glaciers is visible for the dominant aspect: While only 20–30% of the inventoried north- and west-facing glaciers vanished, 70% of all east- and south-facing glaciers were lost between 1973 and 2016 (Fig. 4d). This demonstrates how significantly vanishing glaciers can affect glacier distribution in a mountain range. In terms of area lost, vanishing glaciers have only a minor influence in most aspects, while they account for 10% of the total glacier area exposed to the South.

4.4. Comparison to overall glacier area loss

We found that the area loss that can be attributed to vanished glaciers amounts to 47±3 km² between SGI1973 and the SGI2016 (Table 1). Considering that more than 40% of all glaciers inventoried in 1973 have vanished, this area loss is relatively limited, which is explained by the typically very small extent of the glaciers that have vanished (Figs. 2 and 4a). To analyse the altitudinal distribution of the area that becoming ice-free between 1973 and 2016, we summed the areas using the DEM corresponding to the SGI1973 (see Section 3.1), along with the hypsometry of the vanished glaciers.

Most surface area change caused by vanishing glaciers was observed at around 2600 m a.s.l., while remaining glaciers lost the biggest share of their area around an elevation of 2800 m a.s.l. (Fig. 5). This clearly indicates that the demise of glaciers is focused on rather low elevations, where very small glaciers—most prone to early disappearance—are abundant. The area loss of vanished glaciers is spread over a relatively narrow elevation range of between 2200 and 2950 m a.s.l. (containing 90% of the observed area change). For glaciers that are still inventoried in the SGI2016, 30% of the area change, however, occurred above an elevation of 3000 m a.s.l.

Figure 5. Altitudinal distribution of the area becoming ice-free between the SGI1973 and the SGI2016 in 50 m elevation bands for all glaciers (dashed grey), the glaciers still existing according to the SGI2016 (blue) and the vanished glaciers (red). The percentage of the lost glacier area attributed to the vanished glaciers is indicated for 200 m elevation segments on the right.

Even though vanished glaciers are predominantly very small and account for a limited area with respect to the total glacierised area, their loss may represent an important share in certain elevation ranges. Between 2300 and 2550 m a.s.l., more than 30% of the overall glacier area loss from all Swiss glaciers is explained by vanished glaciers, peaking at 44% at an elevation of 2450 m a.s.l. (Fig. 5). This can be explained by two factors: first, the vulnerability of ice located at low elevation due to high annual mean air temperatures; and second, the characteristics of small glaciers, which typically lack of a high-elevation accumulation zone and are strongly influenced by topographical effects (snow redistribution, shading, debris cover). However, such topographical effects can contribute to both the disappearance and the temporary preservation of ice at rather low altitude (Kuhn, Reference Kuhn1995; DeBeer and Sharp, Reference DeBeer and Sharp2009; Izagirre and others, Reference Izagirre2024). These findings emphasise the importance of overall area losses attributed to vanished glaciers in certain elevation zones, even for regional-scale assessments.

Considering 194 hydrological basins in Switzerland according to Müller and others (Reference Müller, Caflisch and Müller1976) that contained some glaciers in 1973, we find that in 15 of these catchments all glaciers have vanished and in more than half in 36% of the catchments. The hydrological basins that lost at least 50% of their glaciers are exclusively characterised by only small ice areas, in total only accounting for 7% of the glacierised area in 1973. Interestingly, the basins with a large share of vanished glaciers show the same mean glacier elevation as those basins with a low number of vanished glaciers. However, glaciers extend to substantially higher elevations in basins with low glacier disappearance. We find glaciers, on average, do not reach above 3070 m a.s.l. in basins with >50% glacier loss between 1973 and 2016 and maximally reach 2940 m a.s.l. in basins where >75% of the glaciers vanished. This indicates that glacier disappearance is clearly focused on peripheral mountain regions with peaks only slightly reaching above the climatic equilibrium line altitude.