The ice volume preserved on glaciers is a key variable to quantify, among others, their importance as future water resources and their potential contribution for sea level rise. The most used and easy to apply the method to retrieve the ice volume is the scaling Volume-Area approach. This method relies on the empirical relationship between the area of the glacier and its volume of ice. The accuracy of this approach depends on the fitness of the empirical relationship and the accuracy of the glacier mapping. The former is assessed by local ice thickness measurements and volume estimation. Meanwhile, the way glaciers are charted or inventoried is hardly evaluated.

Here we present the ice volume calculation for all glaciers along the Southern Andes using the simple Volume-Area approach and two independent glacier inventories, the well-known Randolph Glacier Inventory (RGI version 6.0) and the new publicly available National Glacier Inventories of Argentina and Chile. We apply the same method to both inventories with the aim to analyze their effect on the ice volume calculation.

Our preliminary results showed that the RGI slightly overestimate (~4%) the ice volume along the Southern Andes compared to the local inventories. Although this could be assumed to be related to the slightly higher (<1%) glacier area in the RGI compared to the local inventories, we found substantial differences between the RGI and the local inventories which best explains the bias. We saw an overestimation of glaciers largest than 1km² and an underestimation of very small glaciers (<0.5 km²), debris-covered glaciers and rock glaciers at RGI against local inventories. Differences may not have a substantial impact for potential sea level rise calculation, but they will have a significant effect on future water resources calculation, especially where rock-glaciers and debris-covered glaciers are more common than debris-free glaciers.

My research focuses on understanding how rock debris accumulating on the surface of a glacier affects its behaviour.

In particular:

What controls if, where, and how fast rocks form a surface cover over a shrinking glacier?

Will it accelerate or retard the glacier recession in the coming years?

Is this effect significant?

Should water planners consider it in their projections of glacier runoff?

How will it affect glacier hazard potential?

The answers to all of these are complicated, and in part it depends on the climate the glacier is in, its surrounding terrain, and where it sits on the evolutionary pathway between a healthy clean ice glacier and, after a lot of glacier shrinkage, a much reduced glacier with a very well developed debris cover.

I use field measuremetns, numerical modelling and remote sensing to try and enage in multidisciplinary work to address these questions.

Does it matter that glaciers get dirtier as they shrink?

As mountain glaciers shrink, many of them become more dirty. This is because the glaciers are losing snow and ice, but the amount of rocks and dust melting out of the ice or falling or blowing onto the glacier from the from the surroundings is increasing at the same time.

This layer of rocks on the surface changes how the glacier melts, and my research is about how this affects the overall glacier behaviour. I want to find out:

What controls if, where, and how fast rocks form a surface cover over a shrinking glacier?

Will such a rock layer make glaciers shrink faster or slower?

Will it change the amount of meltwater coming from the glaciers?

Is it important to include the effect of surface rocks in forecasts of water availablility and glacier change?

How will it affect glacier hazard potential?

The answers to all of these are complicated, and in part it depends on the climate the glacier is in, its surrounding terrain, and where it sits on the evolutionary pathway between a healthy clean ice glacier and, after a lot of glacier shrinkage, a much reduced glacier with a very well developed debris cover.

Glaciers are becoming greener – what sounds here quite picturesque is a warning signal for the environmental status of our glaciers. With this saying we understand a coupling effect on supraglacial surfaces which has long been neglected in the awareness: Reflectivity (Albedo) is one of the key factors for the increase or decrease of ice masses. Albedo can be reduced substantially by organic or inorganic particles which absorb solar energy and contribute to the warming of the surface. Supraglacial areas are also habitat for mainly microbial living communities such as bacteria, algae or fungi. Algae can protect themselves from damaging UV-radiation by darker pigmentation which results in a darkening of the surface. The darker coloration is increasing the availability of liquid water, hence again the growth of microbial communities. This biologically induced impact on albedo is called “bioalbedo” which has never been taken into account in climate models. So far we have most information on bioalbedo on arctic glaciers which is quite a shame that literally nothing is known about alpine glaciers. Hence, the motivation to work on alpine sites is even larger.

The aim of this interdisciplinary study is a quantification and qualification of organic and inorganic particles on an alpine glacier (Jamtalferner). Moreover, we aim to establish a non-invasive method called L.I.F.E. (Laser induced fluorescent emission) to quantify phototrophic pigments in high resolution. Results show an increasing autotrophic and heterotrophic activity with increasing level of coverage which might be due to the effect of bioalbedo.

A distributed, empirical, mass balance model has been developed for Brewster Glacier (Southern Alps, New Zealand) to accurately assess the ongoing changes of alpine glaciers, a key hydrological resource in the South Island of New Zealand. The performance of the model to reconstruct mass balance is considered by comparing to distributed in-situ glaciological measurements between 2004 and 2018. Model components are addressed in-depth, identifying the various challenges of modelling glacier mass balance in a data-sparse environment.

Downscaled meteorological datasets can provide accurate temperature and precipitation data, facilitating the running of empirical models outside the temporal extent of surface observations. Accumulation components introduce large uncertainties into the model due to the lack of accurate snowfall data to calibrate a precipitation phase transition. Calculated extraterrestrial irradiance allows for a temperature-index ablation component to consider seasonal variations in incoming shortwave radiation. The ablation component is calibrated to point-scale measurements of the surface mass balance based on the surface energy balance. Large discrepancies arise when the relative proportions of the surface energy balance vary, such as during warm and moist conditions, when the turbulent heat fluxes dominate the energy balance.

The spatial distribution of parameters calibrated at the point scale produces significant uncertainty, resulting in annual ablation at the lowest elevations of the glacier being underestimated by 2000 mm w.e., while underestimating accumulation at the highest elevations of the glacier by at least 1000 mm w.e. Altitudinal variation due to a temperature lapse and shaded topography does not capture observed variation in mass balance, demonstrating the importance of the spatial variation of energy fluxes. The calibration of empirical factors from a single point on a glacier may not be suitable for representing the conditions over the entire glacier, raising questions regarding the accuracy of distributed mass balance models and the method of parameter calibration.

One of the challenges in resolving changes in seasonal snow and glaciers in the coming century in mountainous regions across the world may in some cases be linked to separating regional variability or trends in large-scale atmospheric circulation from background global warming. For example, the well documented advance of some fast-responding glaciers in New Zealand between 1983 and 2008 clearly showed that regional variability in atmospheric circulation can temporarily counteract the effects of global warming. There is evidence from climate models that the frequency of certain weather patterns in the New Zealand region may not remain stationary throughout the 21^st century, with projections for increased anticyclonic conditions in winter months. Therefore, any future prediction of the response of glaciers in the Southern Alps to climate change may be required to consider the possibility of future trends in circulation. How much faith can we put into climate model predictions of weather patterns in the coming century and do they warrant further analysis? Do we anticipate that the increase in air temperatures globally will be so dominant that any future trends in weather patterns will be masked out by the drastic decline of glaciers, and reduction in seasonal snow?

Traditionally, future estimates of snow are based on dedicated snow/hydrological models forced by climate projections, which, however, are computationally intensive and which decouple hydrology from climate forcing. Recently, regional climate models (RCM) have been used as an alternative, although snow is only an auxiliary parameter in RCMs and not as accurately represented compared to dedicated snow models. Nonetheless, RCMs encompass the climate-hydrology feedbacks, cover large areas, and have become available in moderate horizontal resolutions. However, RCMs have deficiencies in mountain areas (e.g. cold bias in the Alps) and suffer from orographic smoothing. Reliable estimates of future conditions thus require some observational data to correct the bias. But observations that cover a large spatial extent and the whole altitudinal gradient are difficult to acquire in the Alps because of the heterogenous topography and national borders confining exchange between meteorological services. As an alternative, we use observations from remote sensing (snow cover from the MODIS satellites) to correct the bias in four RCMs of the EURO-CORDEX initiative, which provide snow cover as output. The goal is to have a regional climate model ensemble view of future snow cover for the different RCP scenarios based on the CMIP5 simulations. These are the first results of CliRSnow, a project that aims at providing bias corrected and downscaled projections of snow cover for the whole Alpine region until 2100. Future activities include downscaling and comparison to the traditional approach (dedicated snow/hydrological model). This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 795310.

Addressing climate change impacts on mountain cryosphere raises several methodological questions: i) inference of robust trends in data series which include systematic observations, historical and paleo-environmental archives, ii) identification of the factors feeding the highlighted changes and their respective weight, iii) generating future trajectories in an explicit multi-model, multi-scenario ensemble framework, iv) possible distinction of the specific behavior of extreme events. Our work investigates how innovative statistical processing techniques can be developed to tackle these by combining different modelling techniques including extreme value theory and spatio-temporal models, mostly within a hierarchical Bayesian framework. In term of processes, we focus on snow extremes, snow avalanches and glaciers and, for these, we are looking on long range trends, how these relate to environmental drivers including climate change, and how to produce realistic projections for the future. Our applications concern mostly the French Alps due to the availability of data, but other high mountain areas and/or wider scale applications are also considered. Significant changes having occurred mostly over the recent decades but also over longer, centennial time scales could be highlighted. Whereas climate change is clearly a primary explanation for these, other factors such as changes in land use and social practices should not be neglected. This contribution will share our experience and main results with the community as it could presumably be useful for other areas/processes. In addition to innovative findings of wide geophysical relevance, this should in fine contribute to design efficient adaptation strategies in mountain territories (e.g. design of snow loads or avalanche hazard mapping eliminating the questionable stationarity assumption).

Precise knowledge of glacier extents is mandatory for nearly all calculations related to glaciers (e.g. mass balance, future evolution, run-off, sea level contribution) as their area is either used as a multiplicative factor or to spatially constrain observations. The quality of the related glacier outline datasets thus directly impacts on the quality of the derived products. Due to debris cover and clouds, however, delineation of glaciers is not automatic and requires careful and laborious manual editing.

Accordingly, most applications and models use glacier outline datasets that are readily available from the (multi-temporal) GLIMS database or the RGI. While both are continuously updated and improved by the scientific community, not all datasets have the quality required for local to regional scale glaciologic or hydrologic applications. Moreover, updates are often by chance and consistency in the interpretation is lacking across analysts. The latter is in part due to the so far used 30 m resolution Landsat images that have indeed limits in interpretation.

However, with today’s 10 m resolution Sentinel-2 images visual interpretation has much improved and the quality of glacier outlines is in general much higher. Additionally, the 5-day repeat cycle offers unprecedented opportunities to acquire cloud-free images near the end of the ablation period and trace fast processes such as the temporal evolution of snow cover or surging glaciers in full detail. Along with an increasing use of automated processing lines, a true monitoring of glaciers and other cryospheric elements is now possible. However, data dissemination and links to the hydrologic modelling community can still be improved and related possibilities should be discussed at the workshop.

Climate change and dramatic glacier mass loss in the tropical Andes has transformed downstream hydrology, while increasing demands for water by multiple endusers even located beyond the watershed has complicated the assessment of vulnerability. Our collaborative research has coupled multiscalar observations of changes in glacier volume, hydrology, water quality, and land usage with social and economic data about perceptions of and responses to environmental change. Here, we synthesize work conducted in proglacial valleys of glacierized mountain ranges in different regions of the tropical Andes that are in proximity to growing water usage from urban sectors, agriculture, hydroelectric generation, and mining. Our approach has focused on tracing glacier melt to runoff below glaciers in streams and groundwater exchanges. We demonstrate the importance of utilizing mixed methodologies of both physical and social sciences in international partnership. We have maintained embedded sensor observations, utilized innovative aerial platforms, integrated the analysis of both water availability and water use. Natural water-rock interactions as well as human use pose threats to water quality. Social, economic and demographic changes are likely to have equal influence on glacier water resources. Future scenarios of hydro-climatic vulnerability call for better understanding of coupled human and natural hydrodynamics, involving both multiscale process studies and more robust models of glacier-climate interactions.

The biggest future challenge of climate and glacier change to Central Asia (CA) is impact of glacier retreat on water recourses. The rivers of CA start in the mountains and supply up to 90% of water for domestic, industrial and agricultural use on the plain. Changes in streamflow in response to changes in glacierization have been modelled for a number of catchments in CA suggesting that an initial growth in discharge will occur followed by a decline after the 2040s. The uncertainty about the extent and timing of the projected changes remains high because they depend on relative contributions of different sources of water to total runoff such as seasonal snow pack, glacier ice, rock glaciers and permafrost. These vary between catchments and have not been reliably quantified. This study presents an assessment of the observed and future streamflow in the glacierized catchments of CA and aims to constrain uncertainty in modelling changes of water availability and quality using data on isotopic composition of streamflow and different sources of water. The data are supplied by the newly developed Central Asian Research and Adaptation Water Network (CARAWAN) comprised of six catchments: Ulken and Kishi Almaty in Kazakhstan, Chon- Kyzyl-Suu and Ala-Archa in Kyrgyzstan, Varzob-Kafarnigan in Tajikistan and Pskem in Uzbekistan.

Predictions of glacier volume and contribution to downstream hydrology rely on modelling studies, driven by atmospheric forcings. These atmospheric forcings vary in terms of resolution, period covered, and downscaling approach. The choice of forcing and initial conditions influence the prediction. Here, three methodologies to estimate changes in glacier volume and runoff by the end of the century (2086-2100) at Peyto basin, a 21 km² glacierized alpine catchment in the Canadian Rockies, are compared. A semi-distributed, physically based glacier hydrology model with unprecedented process representation was used for all simulations. First, bias-corrected statistically downscaled outputs from the Canadian Regional Climate Model (CanRCM) over 2000-2100 were used to drive the glacier hydrology model. At a resolution of 25 km, with daily outputs, this meteorological forcing was continuous but did not explicitly calculate convection and only coarsely represented orographic precipitation and wind fields. Second, bias-corrected outputs from a pseudo-global warming Weather and Research Forecasting (WRF-PGW) simulation, obtained by adding a monthly climate change signal to a 2000-2015 WRF model forcing simulation, were used to drive the glacier hydrology model for the 2086-2100 period. This produces a dynamically downscaled, convection permitting, “future weather” at high spatial and temporal resolution (4x4 km, hourly), but initial conditions for 2086 are uncertain. Third, the two methods were combined, using the continuous CanRCM run to obtain initial conditions for the 2086-2100 simulation with the WRF-PGW forcing. The results are analyzed to calculate changes in the driving meteorology, specifically the increased air temperature and the shift in precipitation phase and amount, as well as the resulting impact snow processes, glacier mass balance and glacier hydrology in the catchment. This approach provides insight into the uncertainty in projected changes caused by the meteorological forcings and modelling approach and provides an unprecedented examination of the impacts of future mountain weather on glacier hydrology.

Cotopaxi has several studies about the historical eruptions and the hazards that it could produce. Glacier Inventory which includes, among others, changes of glacier coverage by aerial photographs, one determination of the glacier volume by ground penetrating radar (GPR), and the interaction between the glacier and the volcanic activity by ice cracking (Métaxian et al., 2003)

Cotopaxi has 19 glaciers which are very important, for mitigating their hazards in the case of an eruption., A recession of the surface of the glaciers was the 30% it has been detected by photogrammetry between 1956 and 1997 (Jordan et al., 2005). In 2015 the volcano showed very strong signals of reactivation. The Geophysics Institute of Ecuador (IG) has been monitoring the volcano some years. It reported the increase of seismic activity, deformation, ash plumes, and ash fall, SO2 emissions, fumarolic activity, and lahars.

The research analyzes the changes of the thickness and the volume of the glacier based in historical information and collection of data by remote sensing technics, ground penetration radar, geodesic methods, and field trip work. It is in process an updating of the measuring of the thickness of the glacier in the northern and eastern areas. The data combinate with high digital elevation model to detect changes of the glacier after the eruptive process started in 2015.

Permafrost degradation is a key parameter driving climate change induced rock slope instabilities in high mountain areas. The temperature sensitivity of thermal and mechanical rock properties demand for novel high-grade monitoring and modelling campaigns to deduce possible hazards. Electrical resistivity tomography (ERT) became the first choice tool for permafrost monitoring in the alpine space in recent decades.

Here, we present the effects of seasonal and extreme weather events on the annual permafrost distribution at the Zugspitze within an 11-year ERT-monitoring program. The quantitative interpretation of ERT-data for frozen rock temperatures is based on the resistivity-temperature laboratory calibration on Wettersteinkalk (Zugspitze) by Krautblatter et al. (2010).

Our preliminary results are:

1. The natural thermal regime of steep rock walls is reproduced by ERT-temperature data and validated via a climate-driven thermal model.
2. Solar radiation forcing affects rock wall temperatures with a signal propagation time of ca. 2 months, largely via conductive energy transport.
3. 11 years of ERT data validate the resistivity temperature relationship for a natural rock wall environment. Moreover, we can distinguish periodical and uniquely effects due to annual seasonality and extreme weather events.
4. The summers of 2018 and 2015 record the highest yet measured permafrost core temperatures at the Zugspitze. Highly fractured zones display specific warming patterns of the adjacent rock mass.

We further present an approach to the thermo-geophysical modelling of permafrost temperatures in steep rock walls with high spatial resolution linking apparent resistivities, ground thermal properties and meteorological data. Further we intent to model past and future permafrost / rock temperature conditions at the Zugspitze in the realm of climate change.

Krautblatter, M., Verleysdonk, S., Flores-Orozco, A. & Kemna, A. (2010): Temperature-calibrated imaging of seasonal changes in permafrost rock walls by quantitative electrical resistivity tomography (Zugspitze, German/Austrian Alps). J. Geophys. Res. 115: F02003.

In the alpine region, the information about snow cover and snow depth has high socio-economic importance not only for monitoring local water resources and storage but also for predicting climate-related hazards and effects on winter tourism. The snow cover and snow depth exhibit high spatial and temporal variability mainly caused by the highly diverse topographic characteristics of alpine terrain and its influence on wind and radiation conditions. Therefore, for deriving snow information with high spatial and temporal resolution, we aim at investigating the feasibility of using outdoor webcam images, like those provided by ski resorts. Specifically, we test the use of webcam data as reference for (i) increasing the spatial resolution and accuracy of snow cover maps derived from Sentinel-1 and Sentinel-2 data and (ii) assessing the snow depth from image matching data (e.g. from Very High Resolution Pléiades Images and aerial images from Unmanned Aerial Vehicle). The approach to derive snow cover maps and snow depth from webcam images is applied in different study areas located in Austria and Switzerland, selected according to the topography, forest coverage and the availability of the Sentinel data. This work shows that webcam images represent a rich data source to complement and improve satellite-derived snow information in cloudy conditions, steep mountainous terrain and shadowed areas, and some webcams can be used as reference to assess the interannual snow depth.

Water security requires adequate hydrological conditions to sustain economic activity, human water use and ecological services. In watersheds were most water supply is provided by snow and glacier melt, global warming threatens existing water management arrangements by significantly altering the statistical properties of flow regimes. This effect is expected to be enhanced during drought conditions, when socio-hydrological systems see themselves subject to increased stress. In this work we examine the hydrological conditions to be expected under drought events in future plausible climates. Through physically-based mass and energy balance modeling, we investigate snowpack dynamics, glacier ice and melt evolution, and their contribution to seasonal river flows in the Maipo River basin, in central Chile. Under drought conditions, we find that important changes in snowpack extent and duration may affect ecosystems along the critical zone in the transition between rain- and snow-dominated landscapes. In late summer, when river flows are disproportionately influenced by glacier melt, we discuss the emergence of new flow statistics, which can affect water management schemes that rely on probability-of-exceedence during drought conditions for the definition of water rights.

Based on long-term climate predictions, cryosphere resources in the Alps are projected to decline, until almost complete extinction of many glaciated areas within a few decades. However, for the management and planning of water resources, it is crucial to rely on accurate predictions of the evolution of ice and snow resources on the seasonal scale.

This is one of the objectives of the MEDSCOPE project in the framework of the ERA4CS initiative, where a case study is being developed in which the seasonal climate forecasts produced by the project for the Mediterranean region is downscaled for selected glaciated areas in the Italian Alps. The downscaled climate variables will be used to force two distinct numerical models, namely a physically-based multi-layer snowpack model and an empirical glacier model calibrated with historical data. Model outputs will be evaluated and compared with observational data for glacier mass balance and length change, and with snow depth and snow water equivalent measurements by automatic stations in the study areas.

The first results of this work will be presented at the Conference.

Runoff dynamics of alpine streams are controlled by varying contributions of snowmelt, glacier ice melt, rain, and groundwater. Nivo-meteorological variables such as air temperature, precipitation, and the amount of snow available for melting generate complex controlling factors. In this study, we suggest an interdisciplinary approach of characterizing runoff, temperature, and precipitation regimes and their inter- and intra-annual variability by means of wavelet analysis and tracer-based end-member mixing models. The study areas were located in the Saldur catchment (LTSER site), the Sulden catchment, and the Plima catchment. In the Saldur catchment, a stream gauge in the lower reach (catchment area: 62 km²) was operating from 2009 to 2013 and a second stream gauge was installed in the upper reach (catchment area: 11 km²) in 2011. Since 2014, runoff data are available also from a monitoring station in the Plima catchment (catchment area: 22 km²), which is operative from May to October of each year, and in the Sulden catchment (catchment area: 130 km²) as well. At the latter station, electrical conductivity was continuously measured and an automatic water sampler collected daily water samples from May to October 2014 to 2018 for isotopic analysis. Air temperature, precipitation, and snow depth from locations within or near the study areas were acquired from weather stations of the Hydrographic Office of the Autonomous Province of Bozen-Bolzano and EURAC Research. Daily MODIS composite data were used as proxy to identify the ablation period in each catchment and thus the onset of ice melt contributions to the stream from the lower parts of the glaciers. The wavelet spectra of runoff and nivo-meteorological data will help to identify intra-annual changes in the dominant runoff component, evaluate the effect of glacier area proportion, and assess how heat waves, cold spells, and dry periods influence runoff dynamics of glacierized catchments.

Despite of their importance for sea-level rise, seasonal water availability, and as source of geohazards, mountain glaciers are one of the few remaining sub-systems of the global climate system for which no globally applicable, open source, community-driven model exists. Here we present the Open Global Glacier Model (OGGM, http://www.oggm.org), developed to provide a modular and open source numerical model framework for simulating past and future change of any glacier in the world. The modelling chain comprises data downloading tools (glacier outlines, topography, climate, validation data), a preprocessing module, a mass-balance model, a distributed ice thickness estimation model, and an ice flow model. Thanks to the modular framework, modules of various complexity can be added to the codebase, allowing to run new kinds of model intercomparisons in a controlled environment. OGGM spans a wide range of applications, from ice-climate interaction studies at millenial time scales to estimates of the contribution of glaciers to past and future sea-level change. It has the potential to become a self-sustained, community driven model for global and regional glacier evolution.

Indirect effects of climate change include the increasing frequency and magnitude of natural hazards in the vicinity of glaciated mountains. In this regard, studies aiming at understanding the interspersed consequences of climate change on glacier and high-altitude landslide dynamics are sparse in Africa, especially in the remote and rugged Rwenzori Mountains (>5000 m), located along the equator at the border between Eastern Congo and Uganda. Here we report on a multi-disciplinary research approach including field mapping, remote sensing, UAV-based photogrammetry and rock fall trajectory modelling, to assess Rwenzori glacier recession and its potential links with surficial mass movement in the region.

Sierra Nevada still contain alpine permafrost and a small rock glacier over 3000 meters altitude that are Little Ice Age relics. These ice masses are receding, and this process is controlled primarily by snow cover duration, air and surface temperature. Snow cover and sea ice are among the Earths most dynamic features (Hall et al. 2006). Field measurements are hardly interpolated to the full extension of snow cover, especially in mountainous regions. Therefore, the monitoring of snow cover through satellite sensors is nowadays the most feasible way to study its evolution, and helps us to better understand its dynamics.

This work aims to understand how the recent climate conditions are affecting snow cover distribution and duration in Sierra Nevada, and to identify possible trends within its high inter and intraanual variability. For that purpose 14 hydrological years of Landsat 5TM and 7ETM+ imagery were analyzed in order to map snow extension and frequency.

A total of 162 Landsat images were used from September, 2000 to August, 2014 to map snow cover. Two ratios and one index were evaluated: R₄₅ (Hall et al., 1987), R₃₅ (Rott, 1994) and the Normalized Difference Snow Index (NDSI) (Dozier, 1989). As discussed in Santos et al., (2012) the NDSI showed to be the most suitable index to map snow cover in Sierra Nevada.

The analysis of all snow maps generated (up to 170) display a slight reduction of the snow cover (sf of -2.2Km²/year) during the 14 years analyzed and a more important decrease rate on spring time (sf of -12.35Km²/year), even though a high variability in snow cover is observed. Additionally, when analyzed the annual snow frequency, it is perceived that during the driest year (2001-02) only 0.12% of snow cover remains 10-12 months/year, while in 2008-09, the wettest year, this value reaches 0.34%.

Melting snow in the Andes is a main water resource for the population of Chile. Even though several studies have been performed to model snowpack evolution in different catchments in the semi-arid Andes with different models, no comparisons have been made between them. The sensitivity of the models SNOWPACK and SnowModel to its input parameters and parametrization has been assessed. SNOWPACK is especially designed for forecasting avalanches in the Alps and SnowModel is often used to model wind transport of snow. The models are evaluated on a point scale with data from the AWS Tapado (4306 m.a.s.l.) in the year 2017.

The snow models have been calibrated with an ensemble approach and the snow roughness lengths have been tested. Then, a Monte Carlo approach has been applied to the models with variations in the forcing data. This has been done for the variables TA, RH, WS, WD, P, S↓, L↓ and precipitation.

The biggest difference between the models is found in their sublimation rates. Sublimation contributes for 72% of the total ablation by SNOWPACK, whereas this contribution is only 42% by SnowModel even though their atmospheric corrections are calibrated to the same extend.

Secondly, the albedo parameterizations and microstructure of the modelled snowpacks are completely different, causing different melt rates at the end of the winter season. The albedo parametrization of SNOWPACK corresponds better to the measured albedo before melting starts and thus this model is favourable to simulate the snowpack at Tapado AWS.

The sensitivity of both models to the forcing data is in the same order of magnitude and highly influenced by the precipitation uncertainties. The precipitation measurements in this study were subjected to a lot of corrections and cause the biggest uncertainty in the total snow accumulation at Tapado.