Debris-covered glaciers generally exhibit large,
gently sloping, slow-flowing tongues. At present, many of these glaciers
show high thinning rates despite thick debris cover. Due to the lack of
observations, most existing studies have neglected the dynamic interactions
between debris cover and glacier evolution over longer time periods. The
main aim of this study is to reveal such interactions by reconstructing
changes of debris cover, glacier geometry, flow velocities, and surface
features of Zmuttgletscher (Switzerland), based on historic maps, satellite
images, aerial photographs, and field observations. We show that debris
cover extent has increased from
Debris-covered glaciers have been observed to show a delayed adjustment of their length to climatic changes (e.g. Ogilvie, 1904; Scherler et al., 2011; Banerjee and Shankar, 2013). This behaviour can be explained by melt reduction due to insulation by the debris layer, which commonly increases in thickness towards the terminus (Östrem, 1959; Nakawo et al., 1986; Nicholson and Benn, 2006; Anderson and Anderson, 2018), and is expected to distinctly prolong the glacier's response time (Jóhannesson et al., 1989). Several studies showed that debris-covered glaciers exhibit similar thinning rates to debris-free glaciers in the Himalayas (e.g. Kääb et al., 2012; Gardelle et al., 2012; Nuimura et al., 2012). Explanations proposed for this behaviour include the emergence of surface features with exposed ice (e.g. ice cliffs, water flow channels, ponds), enhanced thinning further up-glacier that compensates for the debris-induced melt reduction, reduced mass flux from the accumulation areas and decreasing emergence velocities at the tongue, and englacial ablation with subsequent roof collapsing (e.g. Pellicciotti et al., 2015; Vincent et al., 2016; Ragettli et al., 2016; Banerjee, 2017; Brun et al., 2018). Ice cliffs and ponds can enhance ablation in comparison to debris-covered surfaces and even debris-free ice and are common features on debris-covered glacier tongues (Benn et al., 2012; Brun et al., 2016). Especially during periods of negative mass balance, the down-glacier increase in debris cover thickness reduces ablation through its insulating effect and can lead to a lower – and even reversed – mass balance gradient (Nakawo et al., 1999; Benn and Lehmkuhl, 2000; Benn et al., 2012; Ragettli et al., 2016; Rounce et al., 2018). Over time, this reduction in ablation can lead to a decrease in surface slope and, consequently, driving stress and ice flow velocity (Kääb, 2005; Bolch et al., 2008b; Quincey et al., 2009; Jouvet et al., 2011; Rowan et al., 2015; Dehecq et al., 2019). Furthermore, with an increase in equilibrium line altitude (ELA) the englacial debris melts out earlier, leading to an extended debris cover further up-glacier (Benn et al., 2012; Kirkbride and Deline, 2013; Carturan et al., 2013). As a consequence of reduced ablation and driving stress, heavily debris-covered glaciers often have long, gently sloping, low-lying tongues with low flow velocities or even stagnant parts (Benn et al., 2012). Current research on many debris-covered glaciers is mostly focussed on processes, such as ablation beneath the debris cover and in areas of ice cliffs and ponds, thinning of glacier tongues, surface flow velocities, or changes in debris cover over time (e.g. Hambrey et al., 2008; Bolch et al., 2012; Benn et al., 2012; Dobhal et al., 2013; Ragettli et al., 2016; Gibson et al., 2017). A few studies have started to integrate the existing understanding and interactions of some of these processes into numerical models for the evolution of such glaciers (Jouvet et al., 2011; Rowan et al., 2015; Anderson and Anderson, 2016; Banerjee, 2017).
However, most studies face difficulties leading to persistent shortcomings
in our understanding of the development of debris-covered glaciers: (i) time
series are often too short to inform on the full response time of larger
glaciers; (ii) investigations are often local because repeated tongue-wide
data are sparse, especially for longer periods; (iii) by considering only
one or a few variables, the mutual influences of changing debris cover
and glacier geometry cannot be assessed; (iv) long-term (
To better understand how a changing debris cover affects glacier geometry, flow velocities, and surface features, and how debris cover is in turn affected by these variables, it is necessary to consider the long-term development of glaciers beyond their potential response times. Few studies have investigated the temporal evolution of debris cover on glaciers (overview in Kirkbride and Deline, 2013), or the evolution of debris-covered glaciers over time (e.g. D'Agata and Zanutta, 2007; Capt et al., 2016). Several studies observed an increase in debris cover on glaciers during negative mass balance periods (e.g. Bolch et al., 2008a; Quincey and Glasser, 2009; Kirkbride and Deline, 2013). Because of the overall negative mass balance trend, glaciers are and will be increasingly affected by debris cover. It is important to understand the magnitude of this increase and how it affects the geometry and dynamics of the glaciers in order to simulate their future development (Jouvet et al., 2011; Anderson and Anderson, 2016).
Large debris-covered glaciers in the Himalayas and Karakoram are not ideal
candidates for long-term investigations due to their long response times
(
In this study we aim to understand the long-term geometric evolution and dynamics of debris-covered glaciers through the study of Zmuttgletscher in the Swiss Alps. This medium-sized valley glacier has been going through the transition from a mostly debris-free glacier in the late 1850s to one that is almost completely debris-covered in its ablation area (2017). We quantify the evolution of geometry (length, area, elevation, slope), mass balance, flow, and debris cover at a high spatial and temporal resolution since the end of the Little Ice Age (LIA) around 1850. We use these data to investigate interactions between geometry evolution, ice flow, and debris cover and the related drivers. We further analyse the occurrence of ice cliffs and their role in long-term glacier evolution.
Zmuttgletscher (
Zmuttgletscher is located in a relatively dry region at the main divide of
the Alps; thus it receives precipitation from both northern and southern
weather systems. There are no direct measurements at higher elevations, but
model estimates suggest values between 0.8 and 1.5 m (MeteoSwiss, 2014).
Glaciological mass balance measurements at the nearby debris-free
Findelgletscher (15 km distant) suggest end-of-winter accumulation around
0.8–1.5 m water equivalent (w.e.) (Sold et al., 2016). However, at
Zmuttgletscher, avalanching additionally contributes to accumulation. When
including contributing rock walls and lateral moraines, the total area
available for accumulation is up to 22 km
Zmuttgletscher has several independent and connected tributaries
(Fig. 1). The major accumulation area in the
south – Tiefmattengletscher – reaches almost up to the summit of Dent
d'Hérens. The western accumulation area – Stockjigletscher – is a
relatively flat area above two distinct icefalls between Tête Blanche
(3710 m), Stockji (3092 m), and Wandfluehorn (3588 m).
Schönbielgletscher is a tributary from the north reaching up to
Our analyses are based on topographical maps, aeroplane-based and UAV-based aerial images, and satellite images (Table 1), in addition to various field observations, and long-term air temperature measurements. The use of these data is discussed in the respective sections below. Areal images were available from (i) post-war mapping flights by the American military, (ii) national flight campaigns for topographic map production, (iii) specific flight campaigns for glacier monitoring purposes conducted by Swisstopo, (iv) fixed-wing UAV flights (using a Sensefly eBee Classic; Sensefly SA) in 2016 and 2017, and (v) Pleiades tri-stereo images from 2017.
Topographic maps and satellite and aerial images used in this study. Abbreviations are as follows: dH: elevation change; DTM: digital terrain model; OP: orthophoto; obl. aer.: oblique aerial; Ple.: Pleiades. Entries denoted with an asterisk are used as a final product. All data except 1894, 2016, and 2017 are from Swisstopo (2018a). Oblique photos from 1894 were taken by Reid (1894).
Glacier surface information was extracted from stereo-photogrammetric DTMs generated from aerial images and the Siegfried map (Table 1). To obtain a DTM representing the glacier tongue in 1879 the Siegfried map from 1880 (Swisstopo, 2010) was georeferenced using ground control points (GCPs). Subsequently, the contour lines were semi-automatically digitised by separating the differently coloured (blue is on-glacier; brown is off-glacier) contour lines from other symbols (Siedler, 2011). Their elevation information was extracted and interpolated using the Topo-to-Raster tool in ArcGIS. The point of origin for elevation measurements, situated at the shore of Lake Geneva, changed in the 1930s from 376.2 to 373.6 m (Swisstopo, 2018a) and the elevations derived from the Siegfried map were corrected accordingly.
The DTMs from 1946 to 2017 (apart from 2010 and the Pleiades DTM from
2017) were created with photogrammetric methods using structure-from-motion
software (Agisoft LLC, 2016; Pix4D, 2016). For each date, a point cloud was
produced from the available number of input images (4–29) and then
georeferenced using a set of 10–15 GCPs, i.e. reference points in the images
that could be referred to points on stable ground, taken from Swissimage
2013. The quality of the DTMs is comparable to DTMs from traditional
photogrammetric software (see Mölg and Bolch, 2017). Their uncertainty
in elevation – defined as the standard deviation over stable terrain around
the glacier tongue – mostly lies within 2 m depending on the resolution of
the aerial images, the number of images, and image quality factors (somewhat
higher values were obtained for the DTMs from 1879 and 1946;
Table 1). The uncertainties were derived for
terrain with comparable steepness to the glacier surface (
Orthophotos were generated by rectification of the stereo images using the respective DTM.
Glacier areas were measured from the Dufour map (1859), the Siegfried map (1879), all available orthophotos, and the Swissimage (2005, 2007, 2010, 2013; Swisstopo, 2010; Table 1) by manual digitisation. The Dufour map (map sheet 22, section 8, number 485) from 1859 was the first map containing elevation information (in the form of contour lines) and distances acquired with modern methods (Wolf, 1879; Graf, 1896; Rastner et al., 2016). The extent of the glacier and the supraglacial debris could be extracted from the map, whereas its elevation information was disregarded due to strong, non-linear, horizontal distortions.
The time series of maps and orthophotos resulted in glacier area values for each corresponding date since 1859. The mapping quality is often lower in debris-covered areas compared to debris-free areas (Paul et al., 2013), but the high resolution of these images allowed correct interpretation of the glacier margin. The glacier boundary in the accumulation area to the west was taken from the 2010 ice divide and was kept constant over time. The hanging glaciers at the north face of Dent d'Hérens have been included in the glacier area since they contribute mass to the main glacier through frequent ice avalanches.
Front variation measurements were conducted by using the glacier outlines for each date. Along the ice front – perpendicular to the flow – the change was measured at distances of 100 m and then averaged (Koblet et al., 2010; Bhambri et al., 2012). For the comparison to other glaciers we used GLAMOS length variations (GLAMOS, 2018), which were acquired in the field with the same concept of several parallel measurements, equidistant by 50 m. At Zmuttgletscher, GLAMOS measurements have regularly been conducted between 1892 and 1997, but with serious data issues before 1946. The GLAMOS data for Zmuttgletscher since 1946 are almost identical to our own relative length change record (Fig. S1 in the Supplement) and are therefore not further used. To interpret the observed length changes and the potential influence of debris cover, we compared the variations from our measurements to the GLAMOS data of 10 other Swiss glaciers.
The uncertainty of both area and length results is estimated to lie within
Debris cover extent for the orthophotos and historic maps was manually digitised (Fig. 2a, b). We mapped only areas of continuous debris cover, thus avoiding the inclusion of sparse debris cover. The debris extent of the Siegfried map (1879) was verified using two photographs taken in 1894 (Fig. 2c). This information was valuable to limit the debris extent at and up-glacier of the confluence of Tiefmattengletscher and Schönbielgletscher, which was the region of the strongest changes (Fig. 2d).
Example of mapped debris cover on
Ablation was measured at seven points on the glacier tongue (Fig. 1) during summer 2017 to better understand the influence of debris on melt and to complement the long-term elevation change data. These point measurements were conducted on bare ice and on surfaces with variable debris thickness at elevations between 2300 and 2600 m using 2 m long PVC stakes that were connected by zip ties and drilled into the ice. Debris cover was removed for the drilling and repositioned after inserting the stakes. To get a rough estimate of ablation on ice cliffs, the horizontal backwasting of a south-facing and north-facing ice cliff was measured using horizontally inserted ablation stakes. All stakes were measured in intervals of several weeks over the course of the 2017 ablation season.
Debris thickness data were collected in the field by manual excavation along
and in between three transects perpendicular to the glacier flow direction
in September 2017 (for transect locations see Fig. 1). Each data point represents the average of three measurements
Homogenised time series of air temperature measurements by MeteoSwiss
(Füllemann et al., 2011) were used to interpret the observed glacier
evolution. The selected climate stations had to be as close to the study
site as possible, lie at similar elevations, and cover a long period. Since
no single station fulfilled all requirements, we used the stations on Col du
Grand St. Bernard (2472 m,
Surface flow velocities provide important information about the glaciers'
dynamical state and its change over time. Automatic feature tracking methods
were not feasible for imagery acquired before 2005 because the time
differences and, thus, displacements of the features were too large.
Therefore we manually tracked boulders to infer flow velocities along the
debris-covered part of the main tongue as well as on the lower
debris-covered part of Schönbielgletscher. The tongue was divided into
four sections and 11 subsections according to differences in dynamic
state and ice flow units (Fig. 8). The individual
measurements were averaged for every time period and subsection to achieve
a comprehensive picture of the dynamic changes. For the periods 2005–2007
and 2016–2017 we extracted flow velocity fields using the feature tracking
module IMCORR in SAGA GIS (Fahnestock et al., 1992; Scambos et al., 1992;
Conrad et al., 2015). The results were filtered using a visually defined
threshold of correlation quality (“strength”
Between 22 and 24 August 2017 the tongue of Zmuttgletscher was observed with a
terrestrial radar interferometer (GPRI) developed by Gamma Remote Sensing AG (Werner et al.,
2008). The GPRI was installed on a hill about 3 km away from the terminus in
2017 (Fig. 1). Measurements were acquired every
minute for 1.5 d with a final range resolution of 3.75 m and an azimuth
resolution of 7 m at a slant range of 1 km. The interferograms were
determined with a standard workflow following Caduff et al. (2015) using the
Gamma software, were stacked over a window of 8 h to reduce noise, and
were afterwards unwrapped using a stationary point on the ground. The
unwrapped phases were then converted to line-of-sight displacements
according to Werner et al. (2008), whereby negative velocities were
considered to be noise and filtered out. The results were georeferenced by
rotating the displacement map to best match with the DEM25 (Swisstopo,
2005). Afterwards, the data were resampled into the new grid using nearest-neighbour interpolation. To assess the uncertainties in the velocities we
looked at the difference from zero in measured displacement of 10 stationary
points. This results in an uncertainty of the stacked velocity maps of
Ice cliffs, exposed ice at supraglacial meltwater channels, and lakes were
extracted in a semi-automatic process using an object-based approach with
Trimble eCognition (eCognition Essentials 1.3, 2016); see Kraaijenbrink et
al. (2016). The location and area of these surface features was determined
on all orthophotos from 2013. A primary segmentation divides the
image into polygons based on pixel intensity and image texture. Ice cliffs
often consist of a lower, steeper section of bare ice and a flatter section
of ice covered with sand and pebble-sized debris particles in the upper
part, where the maximum solar angle seems to define the slope of pole-facing
cliffs (Sakai et al., 2002; Buri et al., 2016). These changing slope areas
were often separated by segmentation, and were in a second step manually
selected and combined into one polygon per ice cliff or lake. Supraglacial
channels cover only small areas and are thus incorporated into the term “ice
cliffs”. The described approach is effort efficient and assures a low
uncertainty level, which is on the order of
The time series of surface elevation change over the glacier tongue was
restricted to the overlapping area of all available DTMs from 1879 to 2017
(11 dates), which reaches up to
Glacier-wide geodetic mass balance estimates were calculated between dates
for which DTMs covered large parts of the glacier surface (1879, 1946, 1977,
1988, 2001, 2010, 2017). In areas of data voids or artefacts in the DTM,
especially at higher elevations, no surface elevation change values could be
calculated. In order to reduce the sensitivity to data voids and outliers,
100 m elevation bins (starting from 2100 m) were used for the elevation
change calculation. In elevation bins without DTM coverage, the average of
three extrapolation methods was used as the final mass balance value. The
following three methods were used to fill the data voids: (1) a linear
relationship between elevation and elevation change (e.g. Kohler et al.,
2007; Kääb, 2008), which is based on the strong respective
correlation up to
The total uncertainty of the surface elevation estimates and the mass
balance consists in (i) the accuracy of the individual DTMs, (ii) the
filling of the empty elevation zones, (iii) the glacier volume density, (iv) the debris volume changes, and (v) the DTM co-registration. Ice density
changes are on average assumed to be negligible over a longer time span but
we included an uncertainty of
Zmuttgletscher's long-term mass balance from 1879 to 2017 was
Glacier-wide surface elevation changes and geodetic mass balance of Zmuttgletscher.
From close to the end of the LIA (
Cumulative length change of Zmuttgletscher with data points (vertical red lines) from this study, and several other Swiss glaciers (GLAMOS, 2018). For a list of total length and area changes see Fig. S1.
At the end of the LIA
Evolution of total glacier area, debris-covered glacier area, and the share of the debris-covered glacier area.
Generally, the debris cover extent has expanded up-glacier in
Tiefmattengletscher and Schönbielgletscher. The extension was pronounced
along the surface of Schönbielgletscher and in the central part of the
main glacier tongue (Fig. 6). In both areas the
debris extent has expanded to above the icefall into the former
accumulation areas and now is close to the foot of the contributing rock
walls of Dent d'Hérens (Tiefmattengletscher) and Dent Blanche
(Schönbielgletscher). The debris extent also strongly expanded at the
glacier margins below Stockji, due to further input from moraines and the
disconnection of a contributing tributary. By 2010, the icefall at
Stockjigletscher had thinned sufficiently to disconnect and expose the rock
wall beneath, from which a small rock fall detached between 2010 and 2013
(Fig. 6). This debris mound covered an area of
approximately
Current extent of debris cover and its evolution since 1859
(
Field investigations revealed a homogeneous debris cover in some regions
with stone sizes in the topmost layer mainly between 10 and 30 cm in
diameter, and a much more heterogeneous debris cover in other regions, with
pebble-sized stones to metre-scale boulders (Fig. S4). The typical base
layer of the debris consists of fine-grained material (sand and even silt),
and is overlain by a few centimetres of pebbles. The thickness of the debris
layer along each transect varied from less than 5 cm to greater than 70 cm.
The thickest areas were found on the elongated ridge, especially on the
steep, southern slope (Fig. 7a, transect length
400–500 m). The average thickness along the upper transect was
Evolution of surface flow since 1961 for the different sections of the glacier tongue as labelled in the inset map. After an increase in the 1970s and 1980s a rapid slowdown occurred in all observed regions. SBG: Schönbielgletscher; TMG: Tiefmattengletscher.
Ablation measurements from seven locations and over a 7-week period in
summer 2017 show an “Östrem-like” behaviour with respect to debris
thickness (Fig. 7b; Östrem, 1959). These data
indicate that melt is strongly dependent on debris thickness and much less
on elevation. Compared to the reference stake at 2606 m on debris-free ice,
field measurements in summer 2017 showed a reduction of
Overall, there is a trend of decreasing velocities since the first
measurement period of 1961–1977 until the most recent one (2010–2017), but with
a clear phase of acceleration in the late 1970s and 1980s. Velocities in the
lowest section of the glacier tongue (yellow) as well as the lower
Schönbielgletscher have decreased from 10 to 20 m yr
Influence of ice cliffs on total elevation change rate over the
lowest area of the glacier tongue for eight time periods from 1961 to 2017.
The displacements from radar interferometry from the 1.5 d period in
summer 2017 yield similar results and confirm the quasi-stagnation of the
lower tongue with line-of-sight flow speeds
Even though the glacier has become more debris-covered, we did not find any
clear long-term trend in area and location of ice cliffs
(Table 3). They are almost exclusively located in
the lowest part of the glacier tongue (yellow area in
Fig. 9). The total ice-cliff area decreased by
over 40 % in the 1970s and early 1980s but increased again afterwards. The
topographically corrected ice-cliff area amounts to less than 0.5 % of the
total glacier area and less than 1.8 % of the debris-covered glacier area
(Table 3). As only a few small ponds have been
detected (
Evolution of elevation change over the glacier tongue since 1879.
Ice-cliff areas are shown in red. Note how the “wave” of mass flux moves
down-glacier from
Evolution of ice-cliff area from 1946 to 2013; dc.: debris-covered.
Over the entire period, the average elevation change of ice-cliff areas
(cliffs
Average annual elevation change as a function of surface elevation (in 50 m elevation bins) for seven time periods since 1879.
Since the end of the LIA, the surface elevation of the Zmuttgletscher tongue
below 2750 m a.s.l. has almost continuously decreased
(Fig. 10). Between 1879 and 1946, the average
elevation change was negative at
The relation between surface elevation change and surface elevation has strongly shifted over time (Fig. 11). During the periods from 1879 to 1988 thinning was stronger in lower elevations, with an almost stable surface above 2500 m in the period of 1977–1988. After 1988 the thinning has in general become stronger but also more homogeneous along the glacier tongue.
The DTMs used for calculating mass balance and surface elevation changes
have uncertainties of
The time series of glacier surface flow velocities along the lowest
The uncertainty of the mapped debris extent is equal to that of the glacier
and ice-cliff mapping at
The automatic detection of ice cliffs using topographic and geometric
variables as applied by others (e.g. Brun et al., 2016; Kraaijenbrink et
al., 2016) showed no satisfying results. Many steep areas (
On the basis of their dynamic behaviour we distinguish four distinct phases
over the entire observation period since 1879. In the first phase, between
1879 and 1977, as a response to the atmospheric warming in the mid-19th
century, the mass balance of Zmuttgletscher turned negative
(Fig. 3), first only slightly (1879–1946:
In a second phase between 1977 and 1988, atmospheric cooling and increased
precipitation brought a shift towards less negative mass balance
(
In the third phase between 1988 and 2001 the mass balance became more negative again – along with an abrupt rise in air temperature (Fig. 3a, b) – and flow velocity reductions were pervasive across the glacier tongue, first higher up, then also in the lowest regions (Figs. 8, 12b). The debris-covered area substantially increased – especially during the 1990s – while the glacier area was stable and the central part of the terminus even advanced for some metres, leaving behind a small moraine (Fig. S18). The delayed response to the preceding period with less negative mass balance is, however, still detectable in the slight elevation gain right at the terminus in the period 1995–2001, whereas it was already highly negative on the rest of the tongue (Fig. 10f).
In the most recent period (2001 to 2017), the glacier has continued to
develop a more negative mass balance (approx.
Zmuttgletscher's debris extent has increased by
To understand the potential long-term effects of a debris cover on glacier
evolution, regional comparisons provide a useful context. Nevertheless, one
has to be careful in comparing length and area changes of glaciers for a
given climate history because of differences in geometry and hence response
times (Jóhannesson et al., 1989). The retreat of Zmuttgletscher is
relatively modest compared to that of other large Swiss glaciers
(Fig. 4) and it has shown little terminus
fluctuation, even during the climatically favourable period in the 1970s and
1980s. Other glaciers with similarly subdued fluctuations are
Unteraargletscher and Glacier de Zinal, which are also debris covered in
their lower reaches. Unlike smaller and debris-free glaciers, neither
Unteraargletscher nor Glacier de Zinal advanced in the 1980s and 1990s
(Fig. 4). Aletschgletscher (
Area changes from the beginning of the 1970s until
The relatively uniform and reduced thinning over much of Zmuttgletscher's tongue can be attributed to the debris cover and is observed at several debris-covered glaciers in Switzerland. This suppression of thinning by debris cover on the lower ablation area is best illustrated when plotting the elevation change for one period (1980–2010) for 11 larger Swiss glaciers along 10 vertical elevation bins of equal size (Fig. 13). For all debris-covered glaciers, including Zmuttgletscher, the thinning becomes almost independent of elevation for the lower elevation bins, whereas for debris-free glaciers, it continues to increase towards the terminus. The non-linear relationship between thinning and elevation causes the slower adaptation of debris-covered glacier length and area and hence more extended tongues. A similar flattening or even reversal of the elevation change gradient towards debris-covered glacier termini has also been observed in other regions, e.g. in the central Himalayas (Inoue, 1977; Bolch et al., 2008a; Benn et al., 2012; Ye et al., 2015; Ragettli et al., 2016) and the Tien Shan (Pieczonka and Bolch, 2015).
Elevation change between 1980 and 2010 against normalised elevation for a selection of Swiss glaciers. Solid lines: mostly debris-free glaciers; dashed lines: debris-covered glaciers. Data from Fischer et al. (2015). The lowest section of Aletschgletscher (yellow line) is actually also slightly debris covered.
The evolution of glacier-wide mass balance of Zmuttgletscher since 1879 is comparable to centennial trends in other debris-free glaciers in the Swiss Alps, which are closely related to variations in climate (Fig. 3; Schmidli, 2000; Bauder et al., 2007; Huss et al., 2010; Hirschi et al., 2013; World Glacier Monitoring Service, 2017). For Zmuttgletscher, the strong connection of mass balance to climate is exemplified by the almost balanced conditions during colder phases before 1920 and in the 1970s and 1980s as well as by the strongly negative mass balance in the 1940s and 1950s and in recent decades (Fig. 3; Bauder et al., 2007; Escher-Vetter et al., 2009). This clearly demonstrates that the glacier-wide mass balance of Zmuttgletscher has foremost been governed by climatic changes and has not been strongly influenced by debris cover and changes thereof. This is in contrast to the case of Glacier de Miage in Italy, for which Thomson et al. (2000) related the positive mass balance on the tongue over the 20th century partly to debris cover. The debris cover of Glacier de Miage is, however, thicker than at Zmuttgletscher.
The relatively thin debris at Zmuttgletscher may on the one hand limit the decoupling from climatic influences, and on the other hand still be sufficient to reduce glacier thinning and terminus retreat. This results in an extended and stagnating glacier tongue that increases the area of ablation (relative to a debris-free glacier tongue), which in turn enhances the sensitivity of the glacier-wide mass balance to climatic changes.
The observed temporal variations in ice flow and thickness on the main glacier tongue of Zmuttgletscher also closely reflect variations in climate as exemplified by the acceleration and thickening in the period of more positive mass balance in the 1970s and 1980s. Positive mass balances impacting flow velocities were also observed at debris-covered Glacier de Tsarmine (Capt et al., 2016) and a number of other debris-free glaciers in Switzerland (Glacier de Giétro, Glacier de Corbassière, Mattmark; Bauder, 2017), Austria (Hintereisferner, Kesselwandferner; Stocker-Waldhuber et al., 2018), and France (Glacier d' Argentière; Vincent et al., 2009). Also at Glacier de Miage, a kinematic wave migrating down-glacier was observed between the 1960s and 1980s that led to a small terminus advance in the late 1980s (Thomson et al., 2000). Also the long-term (1960s until 2017) reduction in flow velocities on the tongue of Zmuttgletscher is in line with observations from all of the above-mentioned glaciers.
All these examples show a direct reaction of flow velocities and hence ice flux to climatic changes. This link is consistent with the principle of mass conservation, regardless of debris coverage, but breaks down when the glacier tongue starts to stagnate as a result of the debris cover slowing down the retreat. Such a dynamical stagnation and decoupling has in the last decade also been observed at the tongue of Zmuttgletscher and is characteristic of strongly debris-covered glaciers (e.g. Bolch et al., 2012).
On large debris-covered glaciers, lower flow velocities are often the result of a decreased driving stress due to flat tongues that result from a sustained reversal of the mass balance gradient (Bolch et al., 2008a; Anderson and Anderson, 2016). At Zmuttgletscher, the elevation change gradient has indeed decreased to almost zero during recent decades. Nevertheless, the glacier surface slope has slightly increased over time; therefore we attribute the decrease in dynamic activity to the reduction in ice thickness and, hence, mass flux and climate. This conclusion is coherent with findings by Dehecq et al. (2019) from glaciers in south and central Asia.
Examples of superficial and englacial ablation and their
consequences.
Whilst the presence of a debris mantle may not substantially influence the
overall mass loss rate of Zmuttgletscher, we find that the main reason for
the observed heterogeneous thinning pattern is the increasingly extensive
and also likely thicker debris cover that is heterogeneously distributed
over the tongue. The heterogeneity is further increased by the presence of
ice cliffs. Field visits and patterns of surface lowering suggest a close
association between ice-cliff formation and the presence of supraglacial
meltwater channels. Surface meltwater often runs off superficially
over substantial distances and water flow channels are abundant all over the
glacier tongue outside of crevassed areas. Inside and alongside these
channels bare ice becomes exposed when water washes away the debris or
laterally cuts into the ice (Fig. 14a) and the
debris slides off the oversteepened channel walls. The location of
supraglacial meltwater channels on Zmuttgletscher seems closely related to
areas of compressional flow – in flat and stagnating areas – and has been
most pronounced in the lowest
Large ice cliffs also exist at the terminus and are responsible for
exacerbated retreat compared to where the terminus is gently sloping and
debris mantled. The situation of a terminal cliff at the glacier mouth
combined with terminus retreat is also found at Gangotri glacier for example
(Bhambri et al., 2012; Bhattacharya et al., 2016), whereas some glaciers
with a stable terminus (e.g. Khumbu, Miage) do not show such terminal cliffs
(Bolch et al., 2008a; Diolaiuti et al., 2009). Remarkably, depressions and
irregular surface topography near the terminus were already indicated in the
Siegfried map from 1879. Certain ice cliffs on Zmuttgletscher have reached
The strong expansion of debris cover might suggest a high and increasing importance of ice cliffs. However, we found that ice cliffs are only responsible for approximately 5 % of glacier-wide volume loss because (i) their total area is small (Table 3; also compared to other glaciers: e.g. Sakai et al., 2000; Juen et al., 2014; Ragettli et al., 2016) and (ii) the debris is relatively thin on Zmuttgletscher.
Even though high flow velocities may be expected to subdue the emergence of ice cliffs, we could not find a clear velocity threshold linked to their occurrence, e.g. to designate areas of potential ice-cliff formation. However, the upper boundary of the ice-cliff zone was moved downstream by increased flow velocities in the 1970s and 1980s (Fig. 12). Further, during this period the total ice-cliff area was clearly reduced compared to both periods with lower flow speeds before and after (Table 3). This suggests a clear link between dynamic state and the occurrence of ice cliffs and would imply expanding ice-cliff areas on stagnating tongues, consistent with the general interpretation of observations (e.g. Pellicciotti et al., 2015).
Increased flow velocities also led to a local down-glacier movement of the
upper boundary of the debris extent (Fig. 12).
Similarly, Deline (2005) documented a decrease in debris-covered area for
Glacier de Miage between
This study presents a
In general, flow velocities have strongly decreased in the last 2 decades similar to other glaciers in the Alps, and the lower 1.5 km of the tongue is almost stagnant (2017), even though the glacier tongue has slightly steepened over time. The low flow velocities are due to thinning and reduced ice flux, and to a minor degree influenced by debris cover. Higher flow velocities between 1977 and 1988 were triggered by a more positive surface mass balance and the related increased mass flux from the accumulation area, which also caused local glacier thickening. This increase led to a slight down-glacier migration of the debris cover boundary, followed by a strong debris cover increase when velocities dropped in the 1990s and rising air temperature led to increasingly negative mass balance. Higher flow velocities also moved the upper boundary of the ice-cliff zone down-glacier, temporarily reduced the ice-cliff area, and eventually led to a slight advance in the 1990s.
These findings suggest a clear and direct influence of flow dynamics on debris extent and ice-cliff formation. The above-described processes and feedbacks are likely valid and relevant for other debris-covered glaciers in the Alps and elsewhere at potentially different rates and magnitudes. In the context of global warming, it is therefore crucial to include these findings in models for glacier projections.
Most datasets derived in this study are under the educational license and any third-party transfer is restricted. The original stereo images and historic maps can be purchased from SwissTopo (
The supplement related to this article is available online at:
NM, TB, and AV designed the study. NM generated the DTMs and Orthophotos and performed all analysis. AW performed radar measurements and processing and provided the respective text section. NM wrote the draft of the paper. All authors contributed to the final version of the paper.
The authors declare that they have no conflict of interest.
The study profited from the effort of a number of fieldworkers: Alessandro Cicoira, Philipp Rastner, Martin Schulthess, James Ferguson, Jan Beutel, Rémy Mercenier, Johann Junghardt, Luisa von Albedyll, and Florentin Brendler. Additionally, Rahel Ganarin conducted and supported the comprehensive debris thickness measurements. Mauro Fischer and Horst Machguth provided the much appreciated elevation change dataset of all Swiss glaciers. We thank Owen King for proofreading a previous version of the paper. A special thank you goes to the 3G group for the constructive discussions. We are grateful for the long-term homogenised temperature data series provided by MeteoSwiss. Finally, we acknowledge the thorough, stimulating, and fast reviews by David Rounce and Philip Kraaijenbrink and the additional comments by the scientific editor, which were of great help when revising the paper and have significantly improved its quality.
This research has been supported by the Swiss National Science Foundation “Understanding and quantifying the transient dynamics and evolution of debris-covered glaciers” (grant no. 200021_169775).
This paper was edited by Etienne Berthier and reviewed by David Rounce and Philip Kraaijenbrink.