CR1.3

Crunching satellite imagery or Digital Elevation Models (DEMs) is part of your weekly routine?

You are desperate to calculate glacier volume changes despite gappy observations?

Your head blows up just trying to provide those errors bars for your mass balance estimates?

You think science should be fully reproducible?

Python is one of your favourite programming languages?

If you answered yes to any two of those questions, you should definitely attend this presentation!

Remote sensing is becoming increasingly important in our understanding of global glacier changes. Dozens of studies each year aim at estimating geodetic glacier mass balance from the regional to the global scale, providing factual numbers behind glacier retreat. But how reliable are those numbers? Current approaches raise several problems:

• External data, such as glacier outlines can be updated regularly, as is done e.g. with the RGI outlines, potentially making previous estimates obsolete.
• Data processing techniques, such as DEM coregistration or gap-filling, evolve over time in the community.
• Many results are not reproducible or cannot even be updated, because access to the data or code is not granted.

All these issues make the comparison and validation of older vs newer studies challenging and questions the reliability of glacier change estimates. Why not team-up and create the tools we all dream of?

Here we present xDEM, an open-source, community-built and easy-to-use set of tools for DEMs postprocessing and volume change calculation. The tool is designed as a set of Python modules, built on top of popular libraries (rasterio, geopandas, GDAL). It will ultimately provide all that is needed to turn individual raw DEMs into a geodetic volume change and its uncertainties: coregistration, bias correction, gap-filling, volume change calculation and spatial statistics (e.g. variograms). The concept behind xDEM is:

Ease of use: Python modules developed by glaciologists, (mostly) for glaciologists.

Flexibility and modularity: We offer a set of options, rather than favouring a single method and make it straightforward to combine them.

Reproducibility: Version-controlled; releases saved with DOI; test-based development ensures our code always performs as expected.

The progress of the project can be followed at https://github.com/GlacioHack/xdem.

We illustrate the use of xDEM for various test cases and on-going projects to post-process DEMs obtained from ~1930 terrestrial images of the Swiss Alps, American reconnaissance KH-9 satellite images, modern ASTER and Pleiades images or the recent RAGMAC intercomparison experiment.

How to cite: Dehecq, A., Mannerfelt, E., Hugonnet, R., and Tedstone, A.: xDEM - A python library for reproducible DEM analysis and geodetic volume change calculations, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5781, https://doi.org/10.5194/egusphere-egu22-5781, 2022.

We give an overview of the Instructed Glacier Model (IGM) -- a new framework to model the evolution of glaciers at any scale by coupling ice dynamics, surface mass balance, and mass conservation. The key novelty of IGM is that it models the ice flow by a Convolutional Neural Network (CNN), which is trained from physical high-order ice flow mechanical models. Doing so has major advantages in both forward and inverse modelling.

In forward modelling, the most computationally demanding model component (the ice flow) is substituted by a very cheap CNN emulator. Once trained with representative data, IGM permits to model individual mountain glaciers several orders of magnitude faster than high-order ones on CPU with fidelity levels above 90 % in terms of ice flow solutions leading to nearly identical transient thickness evolution. Switching to Graphics Processing Unit (GPU) permits additional significant speed-ups, especially when modelling large-scale glacier networks and/or high spatial resolutions.

In inverse modelling, the substitution by a CNN emulator does not only speed up but facilitates dramatically the data assimilation step, i.e. the search for optimal ice thickness and ice flow parameter spatial distributions that match spatial observations at best (such as ice flow, surface topography or ice thickness profiles) while being consistent with the high-order ice flow mechanics. The reason is that inverting a CNN can take great benefit from the tools used for its training such as automatic differentiation, stochastic gradient methods, and GPU.

IGM is an open-source Python code (https://github.com/jouvetg/igm), which deals with two-dimensional gridded input and output data. Together with a companion library of trained ice flow emulators, IGM permits user-friendly, computationally highly-efficient, easy-to-customize, and mechanically state-of-the-art glacier forward and inverse modelling at any scale. We illustrate its potential by replicating a simulation of the great Aletsch Glacier, Switzerland, from 1880 to 2100, based on a Stokes model. The complete workflow (data assimilation and 220 years long forward modelling) at 100 m of resolution takes about 1-2 min on the GPU of a laptop and can be replicated and adapted easily using an online Colab notebook.

How to cite: Jouvet, G., Cordonnier, G., Kim, B., Luethi, M., Vieli, A., and Aschwanden, A.: IGM, a glacier evolution model accelerated by deep-learning and GPU, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7271, https://doi.org/10.5194/egusphere-egu22-7271, 2022.

Concepts and ideas related to implementation of calving in large-scale ice-sheet models are presented and discussed, and new model verification experiments proposed. For unconfined ice shelves, any calving law where the calving rate increases with cliff height (free board) must lead to an unstable advance or retreat. No other solutions are possible and all calving front positions are always unstable. If in contrast, calving rate is a monotonically decreasing function of cliff height, both stable and unstable positions are possible. An example of such a configuration and simple analytical solution for the transient evolution of the calving front is provided, which can be used for numerical verification purposes. It is argued that cliff-height based calving laws are, at least for the case of buttressed ice shelves, arguably unphysical as they can result in a multi-valued function for the calving rate as a function of local state of stress. Implementation of a new variational form of the level-set method, involving forward-and-backward diffusion, for capturing the evolution of calving fronts is discussed and several applications to Pine Island and Thwaites glacier shown.

How to cite: Gudmundsson, G. H.: Implementation of calving processes in large-scale ice sheet models, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-4263, https://doi.org/10.5194/egusphere-egu22-4263, 2022.

Energy budget-based distributed modelling in glacierized catchments is important to examine glaciological-hydrological regimes and compute flow rates in current and projected scenarios. Trends in ablation of snow and glaciers retreat depend upon snow and ice reserves, meteorological parameters and geographical features which vary across sub-basins in Upper Indus Basin. This study attempts to address these issues by employing [1] regional climate models (RCMs) and the Physical based Distributed Snow Land and Ice Model (Ranzi and Rosso 1991; Grossi et al. 2013) in the Naltar catchment (area of 242.41 km2, with 42 km2 glacierized), located in the Hunza river basin (Upper Indus Basin), to project snow and glacier melt and daily streamflow. The calibration and validation of the model were successfully carried out using observed historical meteorological data at hourly time resolution from high altitude meteorological stations (Liaqat et al. 2021). For each of the climate simulations, projections of near future (2040-2059) and far future (2080-2099) under three Representative Concentration Pathways (RCPs) namely RCP2.6, RCP4.5, and RCP8.5 are presented[2]  with respect to corresponding present climate (1991-2010). We used all relevant meteorological variables from an ensemble of 37 simulations in total, which were performed by 3 RCMs driven by 11 different global climate models (GCMs) and were developed under the CORDEX Experiment, (Giorgi et al. 2009)-South Asia initiative. RCMs often present systematic biases and, despite their rather high spatial resolution (here approximately 50km x 50km) they are still too coarse for hydrological impact assessments. In order to produce localized and unbiased climate projections, we scaled the observed climate according to the simulated changes by means of the delta change method as described in Räisänen and Räty (2013) and Räty et al. (2014). Correction factor [3]  in the mean and standard deviation for all for all meteorological variables were obtained for the near and far future periods compared to the historical period (1991-2010) for each simulation. [4] T[5] he joint analysis of climate projections and hydrological modelling, spanning different scenarios and other sources of uncertainty is essential to predict future changes in water resources availability to satisfy mainly irrigation demand in the downstream areas.

References

Giorgi F, Jones C, Asrar GR (2009) Addressing climate information needs at the regional level: the CORDEX framework World Meteorological Organization Bulletin 58:175

Grossi G, Caronna P, Ranzi R (2013) Hydrologic vulnerability to climate change of the Mandrone glacier (Adamello-Presanella group, Italian Alps) Advances in water resources 55:190-203

Liaqat MU, Grossi G, Ansari R, Ranzi R Modeling Hydrological Vulnerability to Climate Change in the Glacierized Naltar Catchment (Pakistan) Using a Distributed Energy Balance Model. In: AGU Fall Meeting 2021, 2021. AGU,

Räisänen J, Räty O (2013) Projections of daily mean temperature variability in the future: cross-validation tests with ENSEMBLES regional climate simulations Climate dynamics 41:1553-1568

Ranzi R, Rosso R (1991) Physically based approach to modelling distributed snowmelt in a small alpine catchment IAHS PUBL, IAHS, WALLINGFORD:141-150

Räty O, Räisänen J, Ylhäisi JS (2014) Evaluation of delta change and bias correction methods for future daily precipitation: intermodel cross-validation using ENSEMBLES simulations Climate dynamics 42:2287-2303

How to cite: Liaqat, M. U., Casanueva, A., Grossi, G., and Ranzi, R.: Future climate and runoff projections in the Naltar Catchment, Upper Indus Basin from CORDEX-South Asia regional climate models and hydrological modelling, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6030, https://doi.org/10.5194/egusphere-egu22-6030, 2022.

The number and extent of glacial lakes in mountain regions worldwide has increased over recent decades as glaciers have lost mass. These ice-contact lakes modify the dynamic response of glaciers to climate change, presenting a challenge to projecting their future evolution. In High Mountain Asia (HMA) glacial lakes have expanded by more than 45% in the last 30 years. Previous studies have demonstrated the contrasting dynamic evolution of lake- and land-terminating glaciers in the Eastern Himalaya, although it was previously unclear if this was a localised phenomenon. Using existing and manually derived datasets, we observed glacier surface velocity, surface elevation, terminus position and glacial lake area change across HMA’s differing climatic regimes over a twenty-year period (2000–2020) to investigate the dynamic evolution of ~60 lake- and land-terminating glaciers.

Our results show that lake-terminating glaciers in the Himalaya, Karakoram and Pamir experienced faster ice flow in the ablation zone, significant surface thinning and extensive terminus recession in comparison to land-terminating glaciers over the same period. The majority of lake-terminating glacier population experienced a glacier-wide increase in velocity during the twenty-year observation period, including 58% of individual glaciers. In comparison, 62% of land-terminating glaciers experienced a decrease in velocity during the same period. This result suggests that lake-induced dynamic changes are occurring irrespective of the regional climatic regime. Our observations also revealed that lake-terminating debris-covered glaciers experienced a greater magnitude of change in velocity, surface elevation and terminus position, than their clean ice counterparts. These results are important for making projections of future glacier change in HMA where many debris-covered glaciers are pre-disposed to the development of terminal lakes in the next few decades.

How to cite: Scoffield, A., Rowan, A., and Sole, A.: The influence of ice-contact lakes and supraglacial debris on glacier change in High Mountain Asia, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7754, https://doi.org/10.5194/egusphere-egu22-7754, 2022.

In the northern Japanese Alps, more than 100 perennial snow patches exist (Higuchi and Iozawa, 1971)．Recently, several groups measured the ice thickness and horizontal flow velocity of seven perennial snow patches in the region, finding them to be active glaciers (e.g., Arie et al., 2019). As they are less than 0.5 km² in area, they are classified as very small glaciers (VSGs). According to Arie et al. (2021), who observed the mass balance using geodetic methods from 2015 to 2019, 1) the fluctuation of the annual mass balance of Japanese VSGs was highly dependent on yearly fluctuation in accumulation depth, 2) the mass balance amplitude was the largest of all glaciers in the world recorded by WGMS, 3) VSGs can be formed only in terrains where avalanches and snowdrifts can acquire more than double the snowfall. However, for avalanches and snowdrifts in 3), the specific topographic conditions that indicate the magnitude of these contributions are not clear. Hughes (2009) found that the contribution of avalanches to the glacier is large where the "avalanche ratio," which is the ratio of total avalanche discharge area to total glacier area, is high.
 
In this study, we compared the avalanche ratio, distribution altitude, and slope direction of the seven confirmed VSGs, seven large perennial snow patches (over 10,000 m²), and three small perennial snow patches (under 1000 m²) to show the topographic conditions for the formation of glaciers and perennial snow patches in the northern Japanese Alps. As a result, there was a positive correlation between the average snow depth of VSGs calculated by the geodetic method from 2015 to 2021 and the avalanche ratio. A negative correlation was seen between the avalanche ratio and distribution altitude in the VSGs, and the lower the altitude, the higher the avalanche ratio. In addition, the relationship between avalanche ratio and distribution altitude showed that the avalanche ratio of VSGs and large perennial snow patches were larger than that of small perennial snow patches at the same altitude. The avalanche ratio of Ikenotan Glacier, which is the only glacier on the windward slope with no snowdrift, was more than twice as large as that of VSGs at the same altitude. These results suggest that the magnitude of the contribution of avalanche and snowdrift deposition and the distribution altitude determine the size of glaciers and perennial snow patches.
 
Arie, K., Narama, C., Fukui, K., Iida, H. and Takahashi, K.: Ice thickness and flow of the Karamatsuzawa perennial snow patch in the northern Japanese Alps, Journal of the Japanese Society of Snow and Ice, 81(6), 283–295, doi:10.5331/seppyo.81.6_283, 2019.
Arie, K., Narama, C., Yamamoto, R., Fukui, K. and Iida, H.: Characteristics of mountain glaciers in the northern Japanese Alps, cryosphere, 1–28, doi:10.5194/tc-2021-182, 2021.
Higuchi, K. and Iozawa, T.: Atlas of perennial snow patches in central Japan, Water Research Laboratory. Faculty of Science, Nagoya University., 1971.
Hughes, P. D.: Twenty-first Century Glaciers and Climate in the Prokletije Mountains, Albania, Arct. Antarct. Alp. Res., 41(4), 455–459, 2009.

How to cite: Arie, K. and Narama, C.: Topographic conditions for the formation of glaciers and perennial snow patches in the northern Japanese Alps, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-131, https://doi.org/10.5194/egusphere-egu22-131, 2022.

Alpine glacier retreat has increased markedly since the late 1980s and is commonly linked to the effects of rising air temperature on surface melt. Less considered are processes associated with glacier snout-marginal surface collapse. A survey of 22 retreating Swiss glaciers suggests that collapse events have increased in frequency since the early 2000s, driven by ice thinning and reductions in glacier-longitudinal ice flux.

Detailed measurement of a collapse event at one glacier with Uncrewed Aerial Vehicles and ablation stakes showed 0.02 m/day vertical surface deformation above a meandering main subglacial channel, the planform of which was mapped with Ground Penetrating Radar measurements. However, with low rates of longitudinal flux (<1.3 m/year), ice creep was insufficient to close the channel in the snout marginal zone. We hypothesize that an open channel maintains contact between subglacial ice and the atmosphere, allowing greater incursion of warm air up-glacier, thus enhancing melt from below. The associated meandering of subglacial channels at glacier snouts leads to surface collapse due to erosion and internal melt as well as removal of ice via fluvial processes.

How to cite: Egli, P., Belotti, B., Ouvry, B., Irving, J., and Lane, S.: Subglacial channels, climate warming and increasing frequency of alpine glacier snout collapse, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-3638, https://doi.org/10.5194/egusphere-egu22-3638, 2022.

In most models of large-scale ice-sheet dynamics, horizontal density variations within the ice are largely ignored. Ice-sheets typically comprise a core of meteoric ice, and an overlying layer of lower-density firn of variable thickness. This gives rise to spatial variation in the average density of the ice at each point on the surface, which in principle will modify the glacial dynamics.  A common approach to handle density-variation in the ice is to adjust the thickness of the glacier to the equivalent height of constant-density meteoric ice. We refer to this as the density-to-thickness (D2T) adjustment method. While this approximation preserves the total mass of the ice-column at each spatial coordinate, it introduces additional unwanted terms in the momentum equations, and misses other correction terms. 

In this study, we investigate the D2T adjustment approximation in detail, and consider a number of alternative formulations to handle the density variation in the ice-sheet, based around the modified field equations that we derive in the presence of a variable density field. The alternative formulations include: a static density distribution in which accumulation and compactfication of the firn layer counteracts the advection of the density field so that the time-evolution of the density field can be ignored; or alternatively a time-evolving density distribution with advects with the ice-flow, such that the material derivative of the density field is zero. 

These different formulations are studied in detail within the framework of perturbation analysis. We derive transfer functions for the induced perturbations in both the glacial thickness and velocity, in response to a small perturbation in the density field. We study the frequency profile of the response and its temporal evolution. This helps us gain a deeper understanding of the different formulations, and their impact on glacial dynamics. Within the numerical ice-flow model Úa, we compare the D2T adjustment method to an approach which explicitly includes the density variations, applied to numerical simulations of the Western Antarctic region containing Pine Island and Thwaites Glaciers.

How to cite: Schelpe, C. and Gudmundsson, H.: Treatment of Density Variations in Ice-Flow Models using the Shallow Ice Stream Approximation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-4937, https://doi.org/10.5194/egusphere-egu22-4937, 2022.