High-resolution inventory to capture glacier disintegration in the Austrian Silvretta

Eastern Alpine glaciers have been receding since the LIA maximum, but the majority of glacier margins could be delineated unambiguously for the last Austrian glacier inventories. Even debris-covered termini, changes in slope, colour or the position of englacial streams enabled at least an in situ survey of glacier outlines. Today the outlines of totally debriscovered glacier ice are fuzzy and raise the theoretical discussion if these glaciogenic features are still glaciers and should be 10 part of the respective inventory – or part of an inventory of transient cryogenic landforms. A new high-resolution glacier inventory (area and surface elevation) was compiled for the years 2017 and 2018 to quantify glacier changes for the Austrian Silvretta region in full. Glacier outlines were mapped manually, based on orthophotos and elevation models and patterns of volume change of 1 to 0.5 m spatial resolution. The vertical accuracy of the DEMs generated from 6 to 8 LiDAR points per m2 is in the order of centimetres. calculated in relation to the previous inventories dating from 2004/2006 (LiDAR), 2002, 1969 15 (photogrammetry) and to the Little Ice Age maximum extent (moraines). Between 2004/06 and 2017/2018, the 46 glaciers of the Austrian Silvretta lost -29±4% of their area and now cover 13.1±0.4 km2. This is only 32±2% of their LIA extent of 40.9±4.1 km2. The area change rate increased from -0.6%/year (1969-2002) to -2.4%/year (2004/06-2017/18). The annual geodetic mass balance showed a loss increasing from -0.2±0.1 m w.e./year (1969-2002) to -0.8 m ±0.1 w.e./year (2004/062017/18) with an interim peak in 2002-2004/06 at -1.5±0.7 m w.e./year. Identifying the glacier outlines offers a wide range 20 of possible interpretations of former glaciers that have evolved into small and now totally debris-covered cryogenic geomorphological structures. Only the patterns and amounts of volume changes allow us to estimate the area of the buried glacier remnants. To keep track of the buried ice and its fate, and to distinguish increasing debris cover from ice loss, we recommend inventory repeat frequencies of three to five years and surface elevation data with a spatial resolution of one metre. 25 https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c © Author(s) 2021. CC BY 4.0 License.


Introduction
As a reaction to climate warming, mountain glaciers all over the world are changing at an increasing pace (IPCC, 2019) and at unprecedented rates (Zemp et al., 2015), but with significant regional variability (IPCC, 2019). It is widely accepted that in 30 some mountain ranges a significant part of today's glaciers may disappear within this century (e.g. Zemp et al., 2019a, Huss andFischer, 2016a). A loss of almost all mountain glaciers by 2300 seems possible (Marzeion et al., 2012). For tackling climate change using glaciers as essential climate variables (Bojinski et al., 2014), the precise and detailed monitoring of glacier recession is essential (Zemp et al., 2019b). It can also reveal regionally different response times (Zekollari et al., 2020), various uncertainties (Huss et al., 2014) and serve as a basis for specific scenarios (e.g. Zekollari et al., 2019). 35 Glacier inventories are amongst the most valuable tools for glacier monitoring on a regional scale (Haeberli et al., 2007, Gärtner Roer et al., 2019, and for remote and large glaciers. After the pioneering World Glacier Inventory (WGI), compiled by WGMS and NSIDC (1999), several initiatives, such as GLIMS (Kargel et al., 2014, Racoviteanu et al., 2009, have published global glacier inventories like the Randolph glacier inventory (Pfeffer et al., 2014). International consortia like Globglacier work out guidelines for mapping glaciers, e.g. Paul et al. (2009). At the same time, smaller regional studies work on new methods (e.g. 40 Paul et al, 2020) and have responded to regional phenomena and demands, for example, debris-covered glaciers (Nagai et al., 2016) or the large and often cloudy and snow-covered Patagonian glaciers (Meier et al., 2018). Regional inventories at high resolution can also serve as validation for large-scale semi-automatic remote sensing products.
Airborne LiDAR has been a valuable tool for glacier studies for nearly 20 years (e.g. Stötter, 2002, Pellikka andRees, 2009). Acquisition technologies, processing and analysis have been significantly enhanced since the early years to reach 45 a few cm nominal vertical accuracy in flat areas. High point densities allow processing gridded elevation data with a spatial resolution of 0.5 m or even higher. Early scientific work included the investigation of the potential of (repeat) LiDAR data to map geomorphological processes (Höfle and Rutzinger, 2011) and glacier/rock glacier extent (Abermann et al. 2010). In 2004, federal authorities in Austria initiated the compilation of the first federal DEMs based on LiDAR which were used to update the photogrammetric Austrian glacier inventories . Now a repeat federal LiDAR DEM is available for 50 several regions in Austria.
In this article, we present a new LiDAR-based glacier inventory of the Austrian Silvretta (Figure 1), located in the federal states of Tyrol and Vorarlberg. This inventory presents the glacier area and surface elevation for the years 2017 (Vorarlberg) and 2018 (Tyrol). Here, for the first time, a regional glacier inventory was derived from two high-resolution LiDAR surveys for a period of beginning glacier downwaste. Not only was it necessary to include the volume changes to delineate the debris-55 https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License. covered glacier margins as proposed by Abermann et al. (2009), the volume change was used to estimate the geodetic mass balance of the 46 glaciers.  Figure 4 for Litzner glacier. The direct mass balance of Jamtalferner has been strongly negative for the last two decades, with an ELA above summits for most years and several years with zero accumulation area. Beginning in the year 1892, annual length change measurements have 65 been taken at the glaciers Jamtalferner, Southern Totenfeldferner, Bieltalferner, Vermunt Gletscher, Ochsentaler Gletscher, Schneeglocken Gletscher, and Klostertaler Gletscher M (Fischer et al., 2018, Fischer et al., 2016b. Some time series, however, have been abandoned: at Larainferner in 1993 (dead ice body at the undefined terminus), at Schattenspitz Gletscher (debris cover) and Klostertaler Gletscher S in 1995, at Klostertaler Gletscher N in 2003 and at Litzner Gletscher in 2013 (debris cover all over the terminus). 70 The three Schnapfenkuchl glaciers as they present themselves today ( Figure 2) cannot at first glance be identified even during a field survey, as bare ice is rarely visible (Figure 3), and the geomorphological structure of the surface is not dominated by ice dynamics as could be expected for lager debris-covered valley glaciers. https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License.

75
Fischer), which is typical for the current state for the glaciers of Austrian Silvretta studied in this paper: The glaciers are small, with a minimised accumulation area and increased debris cover, so that bare ice is rarely exposed. This raises the question if these transient cryogenic landforms are still glaciers and how we can monitor at which point the glaciers can actually be defined as 'gone'. https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License.  We analyse Schnapfenkuchl V and H as if they were glaciers, because these structures were clearly identified as glaciers in past inventories: exposed bare ice, accumulation areas, an englacial drainage system and crevasses as indicators for ice dynamics were present in 2006 and before ( Figure 4). In recent years these traditional and evident properties of a glacier became hidden by debris. In the orthophotos of 2002, 2009 and 2015 it is evident that Schnapfenkuchl glaciers could be clearly identified as glaciers in 2002, while in 2009 and 2015 we would hardly map any glaciers at these locations. 95 The increase in debris cover is not restricted to the few small glaciers like Schnapfenkuchl H and V. It is a widespread phenomenon in the Austrian Silvretta. For example, length change measurements of Litzner glacier were abandoned because the identification of the glacier margin was hampered by the increase in debris cover. melt rates of more than 1 m w.e. even close to the summits, the flat glacier disintegrates, as measured on Jamtalferner (Fischer et al., 2016c). In the course of disintegration and surface elevation lowering, debris and rock fall cover the ice surface.
We will very likely face a rapid recession of mountain glaciers in the coming decades, with large glacier systems disintegrating 110 into smaller glaciers. This has already been happening in the Silvretta for the last hundred years. The now very small and rapidly changing glaciers of the Austrian Silvretta are therefore a perfect test site for analysing the potentials and limitations of repeat LiDAR as a high-resolution airborne remote-sensing method for monitoring glacier fade out in qualitative and quantitative terms. Even small glaciers contribute to sea level rise (Bahr and Radic, 2012) and can be significant for local hydrological and hazard management. The precise monitoring of a glacial landscape evolving into a postglacial one is also 115 important for the interpretation and dating of paleoglacial landforms, as studied in the Austrian Silvretta by Braumann et al., 2020. By definition, the new glacier inventory aims at tackling the changes in area and volume of all glaciers in the region. This  (Würländer, 2019). In the final report for the Tyrolean part (Rieger,2019) the uncertainty estimate of the LiDAR data processed with the OPALS software is estimated by the comparison to control 145 areas. The resulting standard deviation of elevation at the control areas is 0.032 m. The control area located in the subsample we used for the study showed a standard deviation of 0.030 m. All control areas are located on stable ground outside glaciers.

Areas and surface elevations in previous glacier inventories
For the Austrian part of the Silvretta range, we compiled glacier area (A) changes to the LIA maximum from mapping 150 moraines, for 1969, 1996 and 2002 from orthophotos, and for 2004 (Vorarlberg) and 2006 (Tyrol) from existing glacier inventories . For all inventories apart from the LIA inventory, not only glacier areas but also glacier surface elevations are available. The glacier margins were delineated manually with an uncertainty of the resulting area (σA) of ± 1.5% for glaciers larger than 1km² and ±5% for smaller ones (Abermann et al. 2009).  The identification of areas with subsidence by melt is a clear advantage of this technology and allows mapping glacier areas that are hard to identify in orthophotos or VIS remote-sensing data. In a validation step, orthophotos were used to check the 175 LiDAR-derived outlines.
We applied no minimum size to mapping the glaciers. Only four of the 46 glaciers mapped were larger than 1 km² in 2002, and two of the glaciers were smaller than the 0.01 km² recommended by Paul et al. (2009) as a practical lower limit for mapping mountain glaciers by remote sensing.

2.4 Comparison of nominal relative uncertainties of all LiDAR data used in the study
LiDAR surveys using different instruments and intended point densities (Table 1) produce different representations of the infinitesimally accurate 'real surface', even without real surface changes over time. Based on achieved point densities, the gridding method and resolution may add misalignment of different DEM of the same area. For this, LiDAR data is validated 185 at defined reference areas. Nevertheless, there is a tradition in glaciological remote sensing and photogrammetry to crosscheck the DEM accuracy for glacier-covered areas in potentially stable areas without surface changes. To estimate the final uncertainty with respect to changing point densities and methods applied for generating DEMs (Table 2), we analysed surface elevation changes (Δz) not only at glaciers, but also at rock glaciers and for a buffer of 1000 m and between 1000 to-2000 m around all glaciers and rock glaciers. Although these areas are only partly representative for glaciers in terms of slope ( Figure 7) and roughness, we consider these numbers a very conservative estimate for the uncertainty of the 195 Δz at glaciers. For the rough and changing rock glaciers, we found a mean Δz of -0.4 m with a standard deviation of 1.1 m. At the buffer, excluding the unstable paraglacial areas (1000-2000m), we found a mean elevation difference of 0.0±0.6 m. The standard deviation is twice the uncertainty found by Abermann for LiDAR DEMs. We took the standard deviation as error in Δz for further error propagation. Studies on the derivation of DEMs from LiDAR point clouds reveal that a slope steeper than about 40° potentially exhibits larger deviation from the 'true' surface (Sailer et al., 2014). Although the algorithm applied to convert point clouds to gridded 205 data plays a major role for the representation of a specific surface (elevation and shape), the representation of the smooth glacier surfaces is a bit more resilient to low resolution than very rough geomorphological features. Sailer et al. (2014) claim that the cell size for analysing glacier changes could be even between 5 and 10 m. Sailer et al. (2014) recommend cell sizes below 1 m for terrain steeper than 40°, which we rarely find on the glaciers of the Austrian Silvretta. There, 90% of the glacier area presents slopes below 40° (Figure 7). In any case, the spatial resolution of 210 the LiDAR DEMs analysed in this study fulfils the criteria above.  buffer regions are steeper than the glaciers, so that the elevation difference in the stable buffer zones can be considered an upper limit for the uncertainty.

2.4
Volume change 220 The volume change for each pixel was calculated by multiplying the Δz with the 1x1 m² pixel area A. For the total volume change ΔV of a glacier, the volume changes of all pixels within the glacier margins of the first date (t1) of the period t1 to t2 (2) 230

Geodetic mass balance
The geodetic mass balance Bgeo was calculated from volume change, assuming a constant glacier density of 850 kg/m³ (Equation 3).

= ∆ *
Calculating the geodetic mass balance from volume change requires assumptions on the stability of the glacier bed and on the 235 density of the volume lost or gained. Erosion and deposition of sediments at the glacier bed was neglected for this study, as research on the quantification of volumes is still ongoing. Previous studies on Austria's mass balance glaciers used a constant density of 850 kg/m³, so did Fischer et al. (2015) for the Swiss glaciers.
In recent years, the firn cover has melted. Thus the density at the surface may be higher.
The annual area-averaged specific geodetic mass balance bgeo is then calculated by Equation 5, dividing the geodetic mass 245 balance Bgeo by the area of the glacier at the beginning of period t1 and by the number of years (t2-t1).

Total area and volume changes
From the LIA maximum to 2017/18, the Austrian Silvretta lost 68% of its glacier area ( Table 3). The mean annual area loss in the latest period was -2.4%, which is more than twice the loss of the period before. The 10 totally debris-covered glaciers cover an area 0.303 km² (Table S3). For three glaciers (ID 13006, Fluchthornferner S and Litzner Gletscher E), neither bare ice nor 255 signs of motion or a drainage system were visible, mean thickness changes between 2004/06 and 2017/18 were smaller than 2.6 m without a clear thickness change pattern to indicate an ice margin. From that we can conclude that the subsurface ice merely melted.
Area changes were calculated for all glaciers for the periods LIA maximum -1969LIA maximum - and 1969LIA maximum - -2002

265
The annual specific geodetic mass balance (Table 4)   with highest ablations in the past. This is also evident from the fact that the maximum specific direct mass balance has been measured in 2015 (Fischer et al., 2016), despite an increase of mass loss at most of the (remaining) stakes.  Glaciers with smaller areas present the highest variability. In contrast to the short and first warm inventory period 2002-2004/06, when some of the smaller glaciers were still quite stable, the range of geodetic mass balance for small glacier sizes is extremely high, from losses of more than 1 m w.e./ year to a few quite stable conditions, which are related to debris cover (or loss of ice).

Challenges for mapping area changes of disintegrating glaciers 295
With ongoing glacier disintegration, mapping glacier outlines becomes ever more ambiguous even if using high-resolution volume change data. Major points to discuss are: what exactly are the properties that make a cryogenic feature a glacier which should be included in a glacier inventory, 300 -should we introduce inventories of all cryogenic features including glaciers, permafrost and rock glaciers or -do we need to define a point of glacier disappearance?
All glaciers in this study were mapped first in previous inventories at times when bare ice was largely visible. For the now totally debris-covered glaciers, it is hard to decide using only optical data if there is any ice left. In the absence of bare ice or 305 stable surface structures, such as ponds or crevasses, and with soft slopes and low potential velocities, surface velocities do not help to distinguish buried glaciers from rock glaciers and permafrost dynamics. As measurements in bore holes in rock glaciers show, sliding debris and rocks on the ice can account for a major part of the total surface velocity (Krainer et al., 2015). If a subsurface runoff system exists at the terminus, it is necessary to analyse if it is fed by groundwater, melt of seasonal snow, permafrost ice or glacier ice. This can be done by analysing seasonality of the amount of runoff (e.g. Brighenti 310 https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License. et al., 2019), chemical composition and ecological properties (e.g. Tolotti et al., 2020) as well as by isotope analysis of the meltwater (e.g. Wagenbach et al., 2012).
Careful analysis is needed to decide whether a formerly well-defined glacier still fulfils the criteria of a glacier. Taking glaciers off inventories prematurely is to be avoided, as they may still contribute to glacial runoff in the basin, can force debris slide or serve as essential climate variables. Without keeping them in inventories, we lose track of these transient states. 315

Uncertainties in mapping totally debris-covered glaciers
Mapping glaciers that have become fully covered with debris is uncertain for two reasons. First, mapping areas of possible ice covered with debris by volume changes only allows tackling the presence of melting ice during the inventory period, but it 320 does not prove that any ice is left at the end of the period. Second, the presence of melting ice is not restricted to debris-covered glaciers but is also true for permafrost. The Schnapfenkuchl glaciers V and H (Figure 9) present glacier ice at locations where debris flows exposed the ice and past inventories showed bare ice (see the supplement).
The Schnapfenkuchl glaciers are embedded in an environment adjacent to a number of rock glaciers. The delineation of buried glaciers in the presence of permafrost and mass movements upon and from the glacier needs a high temporal frequency of 325 inventory data to arrive at the detection of glacier ice. High spatial resolution is needed to distinguish between ice loss and ice dynamics and to track the geomorphological processes and features related to volume change.

Distinguishing rock glaciers from small totally debris-covered glaciers
Past definitions of glaciers focussed on the mass of ice. From the earliest definitions, e.g. by Walcher (1773) and Tyndall 340 (1860), to more modern ones within glaciology (Klebelsberg, 1948) and in neighbouring fields (Dexter, 2013), scientists have obviously been investigating glaciers closer to equilibrium than those in our study: "Diese Eisberge werden Ferner genennet, welches Wort… das Eis bedeutet welches mit Schnee vermenget gesammelt hat" (Walcher, 1773) [These mountains of ice are called "Ferner", a word …that denotes ice mixed with snow which has accumulated] 345 "At its origin then a glacier is snow -at its lower extremity it is ice." (Tyndall, 1860)

Gletscher sind Massen körnigen Firns und Eises, die aus Schneeansammlungen hervorgehen und sich dahin bewegen, wo sie abschmelzen oder verdunsten können….scheinbarer oder auch wirklicher Mangel an Bewegung schliesst aber doch die
Bezeichnung Gletscher nicht aus, übergeordnetes Merkmal ist die Körnerstruktur." (Klebelsberg 1948) [Glaciers are masses of granulated firn and ice that have evolved out of accumulations of snow and which move towards 350 where they can melt or evaporate… however, an apparent or actual lack of movement does not preclude it being labelled a glacier, the predominant characteristic is the granular structure.] "A glacier is a mass of relatively slow moving ice created by the accumulation of snow" (Dexter, 2013) Although englacial and supraglacial debris was present on and in glaciers even during the LIA, ice and snow dominated at the time. In our glacier inventory of 2017/18, 10 of 43 glaciers are predominantly covered by debris, not snow, with only few 355 remnants of bare ice visible. Mean thickness changes range from -1.5±0.1 m to -12.3±0.6 m for the total period. At Garnera Gletscher and glaciers #12018, #13005, no bare ice is visible. On the surface of all other glaciers, bare ice is visible in parts.
Several of the debris-covered glaciers, for example, the Schnapfenkuchl glaciers (Figure 9), are located between active rock glaciers captured in the Tyrolean rock glacier inventory (Krainer and Ribis, 2012). This means that two independent groups of researchers, glaciologists and geologists, mapped the same site, for example, the easternmost Schnapfenkuchl glacier/rock 360 glacier, in inventories of different landforms. A continuum from glacier to debris-covered glacier to rock glacier was recently discussed by Anderson et al. (2018). Kellerer-Pirklbauer and Kaufmann (2018) analysed the glacial history of an Austrian site of long-term rock glacier monitoring and found evidence of post-LIA deglaciation to permafrost formation. For the broader scientific investigation of this phenomenon, international cooperation for comparing processes and regimes in other mountain regions are vital, since a case study can only be a first step towards deriving more general empirical and theoretical knowledge. 365 More concise definitions of the term glacier are needed to reflect current conditions on the ground. We suggest:

1)
Glaciers are bodies of sedimentary ice, firn and snow, formed by densification of snow AND en-and supraglacial sediments of all grain sizes.
or 2) Glaciers are exposed bodies of sedimentary ice, firn and snow formed by densification of snow with signs of deformation 370 and an en-or supraglacial drainage system. https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License.
Definition (1) includes all sizes of ice bodies which have been formed as part of glaciers, even dead debris-covered glacier ice.
It understands debris as a natural part of the glacier system, also in the calculation of volume changes, and sets down a clear base for mapping glaciers in inventories and calculating geodetic balances. It does away with the need for investigating 375 drainage systems or deformation on buried ice bodies to find out if a given structure is a glacier. The drawback is that volume changes are not necessarily ice volume changes, but it accounts for the 'real world' problem that it is not possible to assess the amount of englacial debris anyway.
Definition (2) excludes dead and buried glacier ice, which might have advantages for mapping, especially in low resolutions, 380 but leaves an undefined gap between glaciers and permafrost ice, thereby introducing a third class, i.e. buried glacier ice.

Comparison of changes in the Austrian Silvretta with other glacier regions
For the Austrian Silvretta, the mean annual geodetic balance is -0.8 ±0.1m w.e./yr for the period from 2004/06 to 2017/18. This is less negative than the -1.03 m w.e. reported for the Glarus and Leopoldine Alps by Sommer et al. (2020) for the period 385 from 2000 to 2014, but but a greater loss than than the -0.62 m w.e./yr calculated by Fischer, Huss and Hoelzle (2015) for the Swiss Alps between 1980 and 2010. They also reported a variability of geodetic balances on the catchment scale, ranging from -0.52 to -1.07 m w.e. Our data confirm the highly variable sensitivity of small glaciers to warming as found by Huss and Fischer (2016).
The annual rate of area losses in the Austrian Silvretta (1969-2017/18) of -1.13% is larger than the -0.52% reported for a 390 slightly shorter period  in the Caucasus (Tielidze et al., 2020), but similar to the -1.1% reported by Paul et al. (2020) for the Alps between 2003 and 2015/16 excluding glaciers smaller than 0.01 km².

395
Geodetic mass balance has been analysed for potential uncertainties, for example, unaccounted seasonal snow interpreted as ice volume change or changing glacier beds, and real mass changes differing from surface mass balance, for example, as a result of refreezing, firn density changes, or crevasse volume (e.g. Zemp et al., 2013). For the very small and debris-covered glaciers we analysed, an additional source of uncertainty is the amount of the accumulated debris on the surface. This volume is not part of the hydrological cycle so that, from a hydrological perspective, we would wish to exclude this volume from the 400 analysis. This would be possible only if we could keep track of the erosion rates in the source areas. This would necessitate a totally different monitoring system to track the steep headwalls. In terms of geomorphology, debris and rocks are part of the glacier and therefore there is no need to distinguish old from new supra-and englacial debris and rock.
The rougher surface at steep debris-covered glaciers with avalanche activity seems to encourage the formation of perennial snow patches on the debris-covered glacier. This is not relevant for delineating glaciers but raises the question on the effects 405 of these perennial snow volumes on geodetic mass balance. This question could be also answered with a specific monitoring effort tackling the now stored volumes and water equivalents. In this effort first hints on potential ice formation in such structures could be gathered.

Summary and Conclusions 410
The annual change rates of area and volume change indicate an increasing pace of glacier retreat in the Austrian Silvretta. A growing number of nunataks, the disintegration of larger glaciers into smaller ones and the accumulation and relocation of debris on the glacier surface all indicate an additional need for compiling glacier inventories -if we want to keep track of buried glacier ice. Figure 10 outlines the procedure applied in this study. For most of the 46 glaciers, the procedure of 415 delineating the glacier surface by surface roughness and volume change validated by orthophotos worked out well, even for the ten glaciers totally covered by debris. However, for some of the analysed glaciers, for example Schnapfenkuchl V, future analysis of areas will be difficult, as the surface is now totally covered with debris, and accumulation and relocation of the debris is still ongoing. Therefore, we cannot expect a distinct pattern of thickness change in future, with a maximum close to the terminus. In the next decade we thus need a scientific discussion on how to proceed with as yet undefined subsurface 420 glacier remnants.
https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License. Figure 10: Workflow for high-resolution glacier mapping and mass balance calculation as applied in this study. It does lead to quantitative results for most of the glaciers but also shows up the need of scientific discussion where 425 geomorphological processes other than ice melt contribute significantly to volume changes.
The technical requirements for mapping small vanishing glaciers depend on glacier size, annual rate of thickness changes and accumulation rates of debris. The small glacier structures that remain in the Austrian Silvretta are typically just a few tens of metres wide. With increasing pixel size, the accuracy of mapping vertical changes decreases, which makes it difficult to 430 distinguish geomorphological processes like accumulation and relocation of rocks and debris from ice volume change.
Therefore, we recommend 1 m spatial resolution. The vertical accuracy needed to represent the geomorphological processes of debris relocation depends on the steepness of the area and on the temporal interval chosen. For a period of 10 years and a slope of up to 40° for Alpine glaciers, a vertical resolution of 1/10 of the spatial resolution is sufficient to distinguish volume changes by ice melt from erosion and deposition. 435 https://doi.org/10.5194/tc-2020-376 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License.
Of the now 43 glaciers of the Austrian Silvretta, only three are larger than 1 km², 19 are smaller than 0.1 km² (of these, 13 are smaller than 0.05 km²). Applying minimum sizes for glacier area with thresholds at 0.1 or 0.05 km² would exclude 0.82 km²/6.2% and 0.35 km²/2.7% of the total area. This is higher than /in the same magnitude as the nominal uncertainty of 0.4 km² in the total area. Such thresholds for very small glaciers would make many of them disappear from inventories and hamper any efforts to tackle the hazard potential of deglaciation, as even small glacier remnants could be relevant here. 440 There is probably no hard limit for surveying deglaciation in terms of glacier size, as the monitoring strategies for rock glaciers and permafrost can take over. Handing over what is left from glaciers to the scientific networks of the permafrost community could be an emerging and exciting new playground for both fields.
This regional study can only point out the specific challenges and limitations for tackling glacier change in the Austrian Silvretta. These will differ from region to region depending on the climatic regime, lithology, topography, glacier types and 445 many other factors. We would therefore appreciate an international effort to compare the need for tackling deglaciation in other regions, as a common monitoring framework will be essential in an ever warmer future.

Author Contributions
Andrea Fischer designed this study, worked on glaciers in the Silvretta range for over a decade and wrote the text. Kay 450 Helfricht, Bernd Seiser and Martin Stocker-Waldhuber analysed the geodetic data and contributed to discussion and text.

Funding
Both LiDAR DEMs were provided by the TIRIS section of the federal administration of Tyrol and the federal administration of Vorarlberg.

Acknowledgments 455
The federal administrations of Tyrol and Vorarlberg are acknowledged for providing geodata. We thank the community of Galtür, the Lorenz family at Jamtalhütte and Oswald Heis for supporting logistics. Kati Heinrich helped with the figures, Brigitte Scott checked the English -thank you!