pith. sign in

arxiv: 2605.21527 · v1 · pith:S3POQUU6new · submitted 2026-05-19 · 📡 eess.IV · cs.CV· cs.LG

CryoNet: A Deep Learning Framework for Multi-Modal Debris-Covered Glacier Mapping. A Case Study of the Poiqu Basin, Central Himalaya

Farzaneh Barzegar , Tobias Bolch , Norbert Kuehtreiber , Silvia L. Ullo This is my paper

Pith reviewed 2026-05-22 01:33 UTC · model grok-4.3

classification 📡 eess.IV cs.CVcs.LG
keywords debris-covered glaciersglacier mappingdeep learningmulti-modal remote sensingSentinel-2InSARconvolutional neural networkHimalaya
0
0 comments X

The pith

CryoNet fuses multi-modal satellite layers in a custom CNN to map debris-covered glaciers with over 90 percent IoU.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops CryoNet to automatically outline glaciers in high-mountain terrain where debris makes ice look like ordinary rock and soil from above. It stacks Sentinel-2 optical bands with elevation derivatives, radar coherence, texture measures, and other processed layers as input to a ResNet101 encoder-decoder network that includes nested skips and attention blocks. A sympathetic reader would care because reliable glacier boundaries matter for tracking freshwater loss and climate-driven change without labor-intensive field work. The model reports strong segmentation scores on the Poiqu Basin data and holds up when applied to an Alpine test region.

Core claim

CryoNet is an encoder-decoder CNN built on a ResNet101 backbone with nested skip connections and spatial-channel Squeeze-and-Excitation attention. It ingests a multi-modal stack of Sentinel-2 bands, DEM topographic variables, spectral indices, PCA, InSAR coherence and phase, tasseled-cap features, and GLCM texture to classify pixels as clean-ice glaciers, debris-covered glaciers, or glacial lakes. In the Poiqu Basin the network reaches an overall IoU of 90.52 percent and an IoU of 90.46 percent on the debris-covered class while exceeding the scores of DeepLabV3+, SegFormer, and U-Net.

What carries the argument

CryoNet, a ResNet101-based encoder-decoder CNN with nested skip connections and scSE attention, that fuses a stack of optical, topographic, radar, and texture layers to segment glacier surfaces.

If this is right

  • Large-scale glacier inventories become practical in regions where optical images alone cannot distinguish debris cover.
  • Consistent time-series mapping of glacier extent supports better estimates of ice-volume change and water-resource impacts.
  • The same trained weights can be applied to other mountain ranges, as shown by the transfer test to the Mont Blanc Massif.
  • Layer-importance analysis identifies which data sources contribute most, guiding efficient future data-collection choices.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Open multi-modal pipelines of this type could be extended to produce regularly updated global glacier maps from freely available archives.
  • If the model remains stable when some expensive layers are dropped, operational systems could run on lower-cost data streams.
  • Linking the output boundaries directly to glacier-flow or mass-balance models would let researchers test whether mapping errors propagate into climate projections.

Load-bearing premise

The auxiliary layers supply enough independent signal to separate debris-covered ice from rocks and soil that appear identical in ordinary optical images.

What would settle it

Apply the trained model to imagery where the InSAR or DEM layers are removed or misaligned and observe whether accuracy on debris-covered areas drops to the level of the single-modal baseline networks.

Figures

Figures reproduced from arXiv: 2605.21527 by Farzaneh Barzegar, Norbert Kuehtreiber, Silvia L. Ullo, Tobias Bolch.

Figure 1
Figure 1. Figure 1: Poiqu basin in WGS84 geographic coordinate [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: LST, PCA, TC, GLCM, and ITS-live velocity used [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The Reference Labels consistent glacier dataset for HMA. This dataset improves the original inventory by adjusting steep ice and snow￾covered slopes and shaded components, based on multi￾temporal Landsat imagery. It should be noted that this introduces a temporal mismatch between the reference glacier outlines and the Sentinel-2 imagery used in this study (2018). However, such discrepancies are common in g… view at source ↗
Figure 4
Figure 4. Figure 4: CryoNet architecture contrast between debris-covered ice and surrounding non￾glacial terrain. By leveraging multi-scale feature fusion and joint spatial–channel attention, the model enhances class sep￾arability and boundary delineation in heterogeneous mountain environments. The training configuration is detailed in the following section. D. Training and testing strategy To train CryoNet, we conducted a se… view at source ↗
Figure 5
Figure 5. Figure 5: Mont Blanc Massif Region. The black boxes show the areas of interest used in this study. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of overall accuracy between different [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Confusion matrices IV. DISCUSSION The experimental results demonstrate that integrating multi￾source data with the proposed attention-enhanced nested encoder-decoder architecture substantially improves the delin￾eation of debris-covered and clean-ice glaciers. Compared with DeepLabV3+, SegFormer, and the classical U-Net, CryoNet consistently achieved higher accuracy, precision, and IoU across all classes. … view at source ↗
Figure 8
Figure 8. Figure 8: Prediction results on an unseen subset: (a) the corresponding ground truth, (b) Predictions of CryoNet, (c) Predictions [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Prediction results on an unseen subset in Mont Blanc Massif, Alps: (a), (b), (c) [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Permutation-based channel importance for overall performance (top row), clean ice mapping (middle row), and [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
read the original abstract

Glaciers play a critical role as freshwater reserves and indicators of climate change, yet their automatic delineation, especially for debris-covered glaciers, remains challenging due to spectral similarity with surrounding terrain. This study introduces CryoNet, a deep learning framework that leverages a rich multi-modal dataset combining Sentinel-2 optical imagery, DEM-derived topographic variables, spectral indices, Principal Component Analysis (PCA), InSAR coherence and phase, tasseled-cap features, and GLCM texture to discriminate clean-ice glaciers, debris-covered glaciers, and glacial lakes. CryoNet is an encoder-decoder CNN with nested skip connections and spatial-channel Squeeze-and-Excitation (scSE) attention, built upon a ResNet101 encoder to capture hierarchical contextual and spatial features. The study is conducted in the Poiqu Basin in the central Himalaya, and transferability is evaluated by applying the trained model to the Mont Blanc Massif in the Alps. We additionally analyse the importance of each data layer in improving glacier mapping performance. The proposed model achieves an overall IoU of 90.52%, mean Recall of 98.08%, and mean Precision of 92.26%. For debris-covered glaciers specifically, CryoNet obtains an IoU of 90.46%, a recall of 95.79%, and a precision of 94.21%. Across both per-class and overall metrics, CryoNet surpasses DeepLabV3+, SegFormer, and U-Net, taken as state-of-the-art (SOTA) references, demonstrating its effectiveness for robust glacier mapping in complex high-mountain environments.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces CryoNet, a ResNet101-based encoder-decoder CNN with nested skip connections and scSE attention modules, for semantic segmentation of clean-ice glaciers, debris-covered glaciers, and glacial lakes. It fuses a multi-modal stack comprising Sentinel-2 optical bands, DEM derivatives, InSAR coherence and phase, tasseled-cap indices, GLCM textures, PCA, and spectral indices. The model is trained and evaluated in the Poiqu Basin (central Himalaya) and tested for transferability on the Mont Blanc Massif (Alps). Reported metrics include overall IoU of 90.52%, mean recall 98.08%, mean precision 92.26%, and debris-covered glacier IoU of 90.46% (recall 95.79%, precision 94.21%), outperforming DeepLabV3+, SegFormer, and U-Net baselines; an ablation study on input-layer importance is also presented.

Significance. If the performance claims are confirmed under controlled input conditions, the work would advance automated debris-covered glacier mapping in complex terrain, supporting freshwater and climate-change studies. The explicit transfer evaluation to an independent Alpine region and the layer-importance analysis constitute clear strengths that enhance reproducibility and interpretability.

major comments (2)
  1. [Results section (comparison to baselines)] Results section (comparison to baselines): The claim that CryoNet surpasses DeepLabV3+, SegFormer, and U-Net with a debris-covered IoU of 90.46% does not specify whether the baseline models received the identical multi-modal input stack (DEM derivatives, InSAR coherence/phase, GLCM, PCA, tasseled-cap) or only Sentinel-2 optical bands. If the baselines were trained on fewer channels, the performance delta cannot be attributed to the nested skips and scSE attention; the layer-importance analysis shows multi-modality helps but does not resolve this head-to-head fairness issue.
  2. [Methods section] Methods section: No information is provided on training-set size, number of patches or pixels, cross-validation procedure, or standard-error estimates for the reported IoU, precision, and recall values. Without these details the central quantitative claims (overall IoU 90.52%, debris-covered IoU 90.46%) cannot be fully assessed for robustness or statistical significance.
minor comments (2)
  1. [Abstract] The abstract states that the three baselines are 'taken as state-of-the-art (SOTA) references' but does not briefly motivate their selection relative to other recent glacier-mapping architectures.
  2. [Layer-importance analysis figure] Figure showing layer-importance results would be clearer if it included error bars or a statistical test supporting the ranking of input contributions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review, which has helped us identify areas where the manuscript can be clarified and strengthened. We address each major comment point by point below.

read point-by-point responses

  1. Referee: Results section (comparison to baselines): The claim that CryoNet surpasses DeepLabV3+, SegFormer, and U-Net with a debris-covered IoU of 90.46% does not specify whether the baseline models received the identical multi-modal input stack (DEM derivatives, InSAR coherence/phase, GLCM, PCA, tasseled-cap) or only Sentinel-2 optical bands. If the baselines were trained on fewer channels, the performance delta cannot be attributed to the nested skips and scSE attention; the layer-importance analysis shows multi-modality helps but does not resolve this head-to-head fairness issue.

    Authors: We thank the referee for identifying this important point on ensuring a fair comparison. All baseline models (DeepLabV3+, SegFormer, and U-Net) were trained and evaluated on the identical multi-modal input stack used for CryoNet, which includes Sentinel-2 optical bands, DEM derivatives, InSAR coherence and phase, tasseled-cap indices, GLCM textures, PCA components, and spectral indices. The observed performance improvements can therefore be attributed to CryoNet’s architectural components, specifically the nested skip connections and scSE attention modules. We will revise the Results section to explicitly state that the baselines received the full multi-modal input stack. revision: yes

  2. Referee: Methods section: No information is provided on training-set size, number of patches or pixels, cross-validation procedure, or standard-error estimates for the reported IoU, precision, and recall values. Without these details the central quantitative claims (overall IoU 90.52%, debris-covered IoU 90.46%) cannot be fully assessed for robustness or statistical significance.

    Authors: We agree that these details are necessary to allow readers to fully evaluate the robustness of the quantitative results. In the revised manuscript we will expand the Methods section to report the training-set size (number of patches and total pixels), the cross-validation procedure (5-fold cross-validation), and standard-error estimates for IoU, precision, and recall computed across the validation folds. These additions will directly address the concern regarding statistical assessment of the reported metrics. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical results on held-out and external test data are independent of model inputs

full rationale

The paper reports standard empirical performance metrics (IoU, precision, recall) for a trained encoder-decoder CNN on a held-out test set within the Poiqu Basin plus a fully separate Alpine transferability region. These quantities are measured outcomes, not quantities defined by construction from the training inputs or fitted parameters. No mathematical derivation chain exists that reduces a claimed result to a self-referential definition, fitted subset, or load-bearing self-citation. The architecture (ResNet101 + nested skips + scSE) and multi-modal feature stack are presented as design choices whose contribution is assessed via explicit layer-importance ablation, supplying an internal check that does not collapse into the target metrics. Baselines are invoked as external SOTA references without any indication that their training protocol is defined inside the present work.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central performance claim rests on the standard assumption that a supervised CNN can learn discriminative features from the listed remote-sensing layers; no additional free parameters or invented physical entities are introduced beyond ordinary model training.

axioms (1)
  • domain assumption Convolutional neural networks with attention can learn hierarchical spatial and channel features sufficient to discriminate debris-covered ice from surrounding terrain when supplied with the described multi-modal stack.
    Invoked implicitly by the choice of ResNet101 encoder plus scSE blocks and the decision to fuse optical, topographic, InSAR, and texture inputs.

pith-pipeline@v0.9.0 · 5851 in / 1373 out tokens · 43997 ms · 2026-05-22T01:33:41.480580+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

68 extracted references · 68 canonical work pages

  1. [1]

    Importance and vulnerability of the world’s water towers,

    W. W. Immerzeel, A. F. Lutz, M. Andrade, A. Bahl, H. Biemans, T. Bolch, S. Hyde, S. Brumby, B. Davies, A. Elmoreet al., “Importance and vulnerability of the world’s water towers,”Nature, vol. 577, no. 7790, pp. 364–369, 2020

    work page 2020

  2. [2]

    The concept of essential climate variables in support of climate research, applications, and policy,

    S. Bojinski, M. Verstraete, T. C. Peterson, C. Richter, A. Simmons, and M. Zemp, “The concept of essential climate variables in support of climate research, applications, and policy,”Bulletin of the American Meteorological Society, vol. 95, no. 9, pp. 1431–1443, 2014

    work page 2014

  3. [3]

    Future emergence of new ecosystems caused by glacial retreat,

    J. B. Bosson, M. Huss, S. Cauvy-Frauni ´e, J. C. Cl ´ement, G. Costes, M. Fischer, J. Poulenard, and F. Arthaud, “Future emergence of new ecosystems caused by glacial retreat,”Nature, vol. 620, pp. 562–569, 8 2023

    work page 2023

  4. [4]

    The ocean and cryosphere in a changing climate,

    H.-O. P ¨ortner, D. C. Roberts, V . Masson-Delmotte, P. Zhai, M. Tignor, E. Poloczanska, N. M. Weyeret al., “The ocean and cryosphere in a changing climate,”IPCC special report on the ocean and cryosphere in a changing climate, vol. 1155, pp. 10–1017, 2019

    work page 2019

  5. [5]

    The Randolph Glacier Inventory: A globally complete inventory of glaciers,

    W. T. Pfeffer, A. A. Arendt, A. Bliss, T. Bolch, J. G. Cogley, A. S. Gardner, J.-O. Hagen, R. Hock, G. Kaser, C. Kienholz, E. S. Miles, G. Moholdt, N. M ¨olg, F. Paul, V . Radi´c, P. Rastner, B. H. Raup, J. Rich, and M. J. Sharp, “The Randolph Glacier Inventory: A globally complete inventory of glaciers,”Journal of Glaciology, vol. 60, no. 221, pp. 537– 552, 2014

    work page 2014

  6. [6]

    Debris-covered glaciers,

    M. P. Kirkbride, “Debris-covered glaciers,” inEncyclopedia of snow, ice and glaciers. Springer, 2011, pp. 180–182

    work page 2011

  7. [7]

    Decision tree and texture analysis for mapping debris-covered glaciers in the Kangchenjunga area, eastern Himalaya,

    A. Racoviteanu and M. W. Williams, “Decision tree and texture analysis for mapping debris-covered glaciers in the Kangchenjunga area, eastern Himalaya,”Remote Sensing, vol. 4, pp. 3078–3109, 10 2012

    work page 2012

  8. [8]

    The state of rock debris covering Earth’s glaciers,

    S. Herreid and F. Pellicciotti, “The state of rock debris covering Earth’s glaciers,”Nature Geoscience, vol. 13, pp. 621–627, 9 2020

    work page 2020

  9. [9]

    Response of debris-covered glaciers in the Mount Everest region to recent warming, and implications for outburst flood hazards,

    D. Benn, T. Bolch, K. Hands, J. Gulley, A. Luckman, L. Nicholson, D. Quincey, S. Thompson, R. Toumi, and S. Wiseman, “Response of debris-covered glaciers in the Mount Everest region to recent warming, and implications for outburst flood hazards,”Earth-Science Reviews, vol. 114, pp. 156–174, 8 2012

    work page 2012

  10. [10]

    Randolph Glacier Inventory-A Dataset of Global Glacier Outlines, Version 7.0,

    RGI Consortium, “Randolph Glacier Inventory-A Dataset of Global Glacier Outlines, Version 7.0,” Boulder, Colorado, USA, 2023. [Online]. Available: https://doi.org/10.5067/f6jmovy5navz

    work page doi:10.5067/f6jmovy5navz 2023

  11. [11]

    Globally scalable glacier mapping by deep learning matches expert delineation accuracy,

    K. A. Maslov, C. Persello, T. Schellenberger, and A. Stein, “Globally scalable glacier mapping by deep learning matches expert delineation accuracy,”Nature Communications, vol. 16, 12 2025

    work page 2025

  12. [12]

    Automatic delineation of debris- covered glaciers using InSAR coherence derived from X-, C- and L- band radar data: A case study of Yazgyl Glacier,

    S. Lippl, S. Vijay, and M. Braun, “Automatic delineation of debris- covered glaciers using InSAR coherence derived from X-, C- and L- band radar data: A case study of Yazgyl Glacier,”Journal of Glaciology, vol. 64, pp. 811–821, 10 2018

    work page 2018

  13. [13]

    On the accuracy of glacier outlines derived from remote-sensing data,

    F. Paul, N. E. Barrand, S. Baumann, E. Berthier, T. Bolch, K. Casey, H. Frey, S. P. Joshi, V . Konovalov, R. L. Bris, N. M ¨olg, G. Nosenko, C. Nuth, A. Pope, A. Racoviteanu, P. Rastner, B. Raup, K. Scharrer, S. Steffen, and S. Winsvold, “On the accuracy of glacier outlines derived from remote-sensing data,”Annals of Glaciology, vol. 54, pp. 171–182, 7 2013

    work page 2013

  14. [14]

    Improving semi-automated glacier mapping with a multi-method approach: Applications in central Asia,

    T. Smith, B. Bookhagen, and F. Cannon, “Improving semi-automated glacier mapping with a multi-method approach: Applications in central Asia,”Cryosphere, vol. 9, pp. 1747–1759, 9 2015

    work page 2015

  15. [15]

    A consistent glacier inventory for Karakoram and Pamir derived from Landsat data: Distribution of debris cover and mapping challenges,

    N. M ¨olg, T. Bolch, P. Rastner, T. Strozzi, and F. Paul, “A consistent glacier inventory for Karakoram and Pamir derived from Landsat data: Distribution of debris cover and mapping challenges,”Earth System Science Data, vol. 10, pp. 1807–1827, 10 2018

    work page 2018

  16. [16]

    Automated delineation of debris-covered glaciers based on ASTER data,

    T. Bolch, M. F. Buchroithner, A. Kunert, and U. Kamp, “Automated delineation of debris-covered glaciers based on ASTER data,” inProc. 27th EARSeL-Symposium, 4.-7.6.07. Millpress Science Publishers, 2007, pp. 403–410

    work page 2007

  17. [17]

    A Systematic Review of Methodological Advances in Glacier-Velocity Retrieval with an Empha- sis on Debris-Covered Glaciers

    N. Norova, A. Samat, and J. Abuduwaili, “A Systematic Review of Methodological Advances in Glacier-Velocity Retrieval with an Empha- sis on Debris-Covered Glaciers.”Remote Sensing, vol. 18, no. 1, 2026

    work page 2026

  18. [18]

    Multi-decadal mapping of debris-covered glaciers in the Zanskar Himalaya using Landsat Time Series on Google Earth Engine,

    S. Ahmad and A. S. Jasrotia, “Multi-decadal mapping of debris-covered glaciers in the Zanskar Himalaya using Landsat Time Series on Google Earth Engine,”Journal of Earth System Science, vol. 135, no. 2, p. 64, 2026

    work page 2026

  19. [19]

    Deep learning in remote sensing applications: A meta-analysis and review,

    L. Ma, Y . Liu, X. Zhang, Y . Ye, G. Yin, and B. A. Johnson, “Deep learning in remote sensing applications: A meta-analysis and review,” pp. 166–177, 6 2019

    work page 2019

  20. [20]

    A review of remote sensing image segmentation by deep learning methods,

    J. Li, Y . Cai, Q. Li, M. Kou, and T. Zhang, “A review of remote sensing image segmentation by deep learning methods,”International Journal of Digital Earth, vol. 17, 12 2024

    work page 2024

  21. [21]

    A Comprehensive Review of Remote Sensing and Artificial Intelligence Integration: Advances, Applications, and Challenges,

    N. Kazanskiy, R. Khabibullin, A. Nikonorov, and S. Khonina, “A Comprehensive Review of Remote Sensing and Artificial Intelligence Integration: Advances, Applications, and Challenges,”Sensors, vol. 25, p. 5965, 9 2025

    work page 2025

  22. [22]

    SAU-Net: A Deep Learning Approach for Glacier Mapping Based on Multisource Remote Sensing Data,

    Y . Xiang, L. Zhao, J. Li, F. Gao, S. Bian, M. Hou, X. Luo, and C. Guo, “SAU-Net: A Deep Learning Approach for Glacier Mapping Based on Multisource Remote Sensing Data,”IEEE Access, vol. 13, pp. 32 087– 32 099, 2025

    work page 2025

  23. [23]

    Mapping Debris-Covered Glaciers Using High- Resolution Imagery (GF-2) and Deep Learning Algorithms,

    X. Yang, F. Xie, S. Liu, Y . Zhu, J. Fan, H. Zhao, Y . Fu, Y . Duan, R. Fu, and S. Guo, “Mapping Debris-Covered Glaciers Using High- Resolution Imagery (GF-2) and Deep Learning Algorithms,”Remote Sensing, vol. 16, 6 2024

    work page 2024

  24. [24]

    DL4GAM: A multi- modal deep learning-based framework for glacier area monitoring, trained and validated on the European Alps,

    C.-A. Diaconu, H. Zekollari, and J. L. Bamber, “DL4GAM: A multi- modal deep learning-based framework for glacier area monitoring, trained and validated on the European Alps,”Earth and Space Science, vol. 12, no. 9, p. e2025EA004197, 2025

    work page 2025

  25. [25]

    G-LEAU-Net: Attention-enhanced deep learning for accurate glacier mapping from satellite data,

    Y . Zhang, F. Han, J. Guan, S. Wang, and C. Chen, “G-LEAU-Net: Attention-enhanced deep learning for accurate glacier mapping from satellite data,”Science of Remote Sensing, p. 100434, 2026

    work page 2026

  26. [26]

    Debris covered glacier mapping using newly annotated multisource remote sensing data and geo-foundational model,

    S. Kaushik, L. Maurya, E. Tellman, G. Zhang, and J. K. Dharpure, “Debris covered glacier mapping using newly annotated multisource remote sensing data and geo-foundational model,”Science of Remote Sensing, vol. 12, p. 100319, 2025

    work page 2025

  27. [27]

    GlacierNet2: A hybrid Multi-Model learning architecture for alpine glacier mapping,

    Z. Xie, U. K. Haritashya, V . K. Asari, M. P. Bishop, J. S. Kargel, and T. H. Aspiras, “GlacierNet2: A hybrid Multi-Model learning architecture for alpine glacier mapping,”International Journal of Applied Earth Observation and Geoinformation, vol. 112, 8 2022

    work page 2022

  28. [28]

    GlacierNet: A Deep-Learning Approach for Debris- Covered Glacier Mapping,

    Z. Xie, U. K. Haritashya, V . K. Asari, B. W. Young, M. P. Bishop, and J. S. Kargel, “GlacierNet: A Deep-Learning Approach for Debris- Covered Glacier Mapping,”IEEE Access, vol. 8, pp. 83 495–83 510, 2020

    work page 2020

  29. [29]

    Susceptibility analysis of glacier debris flow by investigating glacier changes based on remote sensing imagery and deep learning: A case study,

    S. Yang, G. Mei, and Y . Zhang, “Susceptibility analysis of glacier debris flow by investigating glacier changes based on remote sensing imagery and deep learning: A case study,”Natural Hazards Research, vol. 4, pp. 539–549, 12 2024

    work page 2024

  30. [30]

    Accurate and automatic mapping of complex debris-covered glacier from remote sensing imagery using deep convolutional networks,

    R. Lin, G. Mei, and N. Xu, “Accurate and automatic mapping of complex debris-covered glacier from remote sensing imagery using deep convolutional networks,”Geological Journal, vol. 58, pp. 2254–2267, 6 2023

    work page 2023

  31. [31]

    Deep learning-based framework for monitoring of debris-covered glacier from remotely sensed images,

    A. A. Khan, A. Jamil, D. Hussain, I. Ali, and A. A. Hameed, “Deep learning-based framework for monitoring of debris-covered glacier from remotely sensed images,”Advances in Space Research, vol. 71, pp. 2978–2989, 4 2023

    work page 2023

  32. [32]

    Glacier change in the Poiqu River basin inferred from Landsat data from 1975 to 2010,

    Y . Xiang, Y . Gao, and T. Yao, “Glacier change in the Poiqu River basin inferred from Landsat data from 1975 to 2010,”Quaternary International, vol. 349, pp. 392–401, 10 2014

    work page 1975

  33. [33]

    Acceleration of ice loss across the himalayas over the past 40 years,

    J. M. Maurer, J. M. Schaefer, S. Rupper, and A. Corley, “Acceleration of ice loss across the himalayas over the past 40 years,”Science Advances, vol. 5, no. 6, p. eaav7266, 2019

    work page 2019

  34. [34]

    Earth Observation to Investigate Occurrence, Characteristics and Changes of Glaciers, Glacial Lakes and Rock Glaciers in the Poiqu River Basin (Central Himalaya),

    T. Bolch, T. Yao, A. Bhattacharya, Y . Hu, O. King, L. Liu, J. B. Pronk, P. Rastner, and G. Zhang, “Earth Observation to Investigate Occurrence, Characteristics and Changes of Glaciers, Glacial Lakes and Rock Glaciers in the Poiqu River Basin (Central Himalaya),”Remote Sensing, vol. 14, 4 2022. 15

    work page 2022

  35. [35]

    High Mountain Asian glacier response to climate revealed by multi-temporal satellite observa- tions since the 1960s,

    A. Bhattacharya, T. Bolch, K. Mukherjee, O. King, B. Menounos, V . Kapitsa, N. Neckel, W. Yang, and T. Yao, “High Mountain Asian glacier response to climate revealed by multi-temporal satellite observa- tions since the 1960s,”Nature communications, vol. 12, no. 1, p. 4133, 2021

    work page 2021

  36. [36]

    Reconstructing glacial lake outburst floods in the Poiqu River basin, central Himalaya,

    X. Wang, G. Zhang, G. Veh, A. Sattar, W. Wang, S. K. Allen, T. Bolch, M. Peng, and F. Xu, “Reconstructing glacial lake outburst floods in the Poiqu River basin, central Himalaya,”Geomorphology, vol. 449, 3 2024

    work page 2024

  37. [37]

    Glacial lake evolution and glacier–lake interactions in the Poiqu River basin, central Himalaya, 1964–2017,

    G. Zhang, T. Bolch, S. Allen, A. Linsbauer, W. Chen, and W. Wang, “Glacial lake evolution and glacier–lake interactions in the Poiqu River basin, central Himalaya, 1964–2017,”Journal of Glaciology, vol. 65, no. 251, pp. 347–365, 2019

    work page 1964

  38. [38]

    Potentially dangerous glacial lakes across the Tibetan Plateau revealed using a large- scale automated assessment approach,

    S. K. Allen, G. Zhang, W. Wang, T. Yao, and T. Bolch, “Potentially dangerous glacial lakes across the Tibetan Plateau revealed using a large- scale automated assessment approach,”Science Bulletin, vol. 64, no. 7, pp. 435–445, 2019

    work page 2019

  39. [39]

    Comparing the opportunities of Landsat-TM and Aster data for monitoring a debris-covered glacier in the Italian Alps within the GLIMS project,

    T. Taschner and R. Ranzi, “Comparing the opportunities of Landsat-TM and Aster data for monitoring a debris-covered glacier in the Italian Alps within the GLIMS project,” inIEEE International Geoscience and Remote Sensing Symposium, vol. 2, 2002, pp. 1044–1046

    work page 2002

  40. [40]

    Decadal changes in glacier parameters in the Cordillera Blanca, Peru, derived from remote sensing,

    A. E. Racoviteanu, Y . Arnaud, M. W. Williams, and J. Ordonez, “Decadal changes in glacier parameters in the Cordillera Blanca, Peru, derived from remote sensing,”Journal of Glaciology, vol. 54, no. 186, pp. 499–510, 2008

    work page 2008

  41. [41]

    Remote sensing based assessment of hazards from glacier lake outbursts: a case study in the Swiss Alps,

    C. Huggel, A. K ¨a¨ab, W. Haeberli, P. Teysseire, and F. Paul, “Remote sensing based assessment of hazards from glacier lake outbursts: a case study in the Swiss Alps,”Canadian Geotechnical Journal, vol. 39, no. 2, pp. 316–330, 2002

    work page 2002

  42. [42]

    ASTER ratio indices for supraglacial terrain mapping,

    A. K. Keshri, A. Shukla, and R. P. Gupta, “ASTER ratio indices for supraglacial terrain mapping,”International Journal of Remote Sensing, vol. 30, pp. 519–524, 1 2009

    work page 2009

  43. [43]

    Use of multispectral ASTER images for mapping debris-covered glaciers within the GLIMS project,

    R. Ranzi, G. Grossi, L. Iacovelli, and S. Taschner, “Use of multispectral ASTER images for mapping debris-covered glaciers within the GLIMS project,” inIGARSS 2004. 2004 IEEE International Geoscience and Remote Sensing Symposium, 2004, pp. 1144–1147

    work page 2004

  44. [44]

    Delineation of debris- covered glacier boundaries using optical and thermal remote sensing data,

    A. Shukla, R. P. Gupta, and M. K. Arora, “Delineation of debris- covered glacier boundaries using optical and thermal remote sensing data,”Remote Sensing Letters, vol. 1, pp. 11–17, 3 2010

    work page 2010

  45. [45]

    A new band ratio technique for mapping debris-covered glaciers using Landsat imagery and a digital elevation model,

    H. Alifu, R. Tateishi, and B. Johnson, “A new band ratio technique for mapping debris-covered glaciers using Landsat imagery and a digital elevation model,”International Journal of Remote Sensing, vol. 36, pp. 2063–2075, 4 2015

    work page 2063

  46. [46]

    Combining optical and thermal remote sensing data for mapping debris- covered glaciers (Alamkouh Glaciers, Iran),

    N. Karimi, A. Farokhnia, L. Karimi, M. Eftekhari, and H. Ghalkhani, “Combining optical and thermal remote sensing data for mapping debris- covered glaciers (Alamkouh Glaciers, Iran),”Cold Regions Science and Technology, vol. 71, pp. 73–83, 2 2012

    work page 2012

  47. [47]

    Combining satellite multispectral image data and a digital elevation model for mapping debris-covered glaciers,

    F. Paul, C. Huggel, and A. K ¨a¨ab, “Combining satellite multispectral image data and a digital elevation model for mapping debris-covered glaciers,”Remote Sensing of Environment, vol. 89, pp. 510–518, 2 2004

    work page 2004

  48. [48]

    Novel machine learning method integrating ensemble learning and deep learning for mapping debris-covered glaciers,

    Y . Lu, Z. Zhang, D. Shangguan, and J. Yang, “Novel machine learning method integrating ensemble learning and deep learning for mapping debris-covered glaciers,”Remote Sensing, vol. 13, 7 2021

    work page 2021

  49. [49]

    ITS LIVE Regional Glacier and Ice Sheet Surface Velocities,

    A. S. Gardner, M. A. Fahnestock, and T. A. Scambos, “ITS LIVE Regional Glacier and Ice Sheet Surface Velocities,” 2019

    work page 2019

  50. [50]

    Using L-band SAR coherence to delineate glacier extent,

    D. Atwood, F. Meyer, and A. Arendt, “Using L-band SAR coherence to delineate glacier extent,”Canadian Journal of Remote Sensing, vol. 36, no. sup1, pp. S186–S195, 2010

    work page 2010

  51. [51]

    Compilation of a glacier inventory for the western Himalayas from satellite data: Methods, challenges, and results,

    H. Frey, F. Paul, and T. Strozzi, “Compilation of a glacier inventory for the western Himalayas from satellite data: Methods, challenges, and results,”Remote Sensing of Environment, vol. 124, pp. 832–843, 9 2012

    work page 2012

  52. [52]

    Automated detection of rock glaciers using deep learning and object-based image analysis,

    B. A. Robson, T. Bolch, S. MacDonell, D. H ¨olbling, P. Rastner, and N. Schaffer, “Automated detection of rock glaciers using deep learning and object-based image analysis,”Remote Sensing of Environment, vol. 250, 12 2020

    work page 2020

  53. [53]

    Glacier mapping based on rough set theory in the Manas River watershed,

    L. Yan, J. Wang, X. Hao, and Z. Tang, “Glacier mapping based on rough set theory in the Manas River watershed,”Advances in Space Research, vol. 53, pp. 1071–1080, 4 2014

    work page 2014

  54. [54]

    Mapping of debris-covered glaciers in the Garhwal Himalayas using ASTER DEMs and thermal data,

    R. Bhambri, T. Bolch, and R. K. Chaujar, “Mapping of debris-covered glaciers in the Garhwal Himalayas using ASTER DEMs and thermal data,”International Journal of Remote Sensing, vol. 32, pp. 8095–8119, 2011

    work page 2011

  55. [55]

    A method for trend-based change analysis in Arctic tundra using the 25- year Landsat archive,

    R. Fraser, I. Olthof, M. Carri `ere, A. Deschamps, and D. Pouliot, “A method for trend-based change analysis in Arctic tundra using the 25- year Landsat archive,” inPolar Record, vol. 48, 1 2012, pp. 83–93

    work page 2012

  56. [56]

    Changes over the Last 35 Years in Alaska’s Glaciated Landscape: A Novel Deep Learning Approach to Mapping Glaciers at Fine Temporal Granularity,

    B. M. Roberts-Pierel, P. B. Kirchner, J. B. Kilbride, and R. E. Kennedy, “Changes over the Last 35 Years in Alaska’s Glaciated Landscape: A Novel Deep Learning Approach to Mapping Glaciers at Fine Temporal Granularity,”Remote Sensing, vol. 14, 9 2022

    work page 2022

  57. [57]

    Brief communication: Updated GAMDAM glacier inventory over high-mountain Asia,

    A. Sakai, “Brief communication: Updated GAMDAM glacier inventory over high-mountain Asia,”Cryosphere, vol. 13, pp. 2043–2049, 7 2019

    work page 2043

  58. [58]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    work page 2016

  59. [59]

    U-Net: Convolutional Net- works for Biomedical Image Segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Net- works for Biomedical Image Segmentation,” inMedical Image Comput- ing and Computer-Assisted Intervention – MICCAI 2015, 2015

    work page 2015

  60. [60]

    UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation,

    Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, and J. Liang, “UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation,”IEEE Transactions on Medical Imaging, vol. 39, pp. 1856–1867, 6 2020

    work page 2020

  61. [61]

    Squeeze-and-Excitation Networks,

    J. Hu, L. Shen, and G. Sun, “Squeeze-and-Excitation Networks,” in2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 6 2018, pp. 7132–7141

    work page 2018

  62. [62]

    Recalibrating Fully Convo- lutional Networks With Spatial and Channel ’Squeeze and Excitation’ Blocks,

    A. G. Roy, N. Navab, and C. Wachinger, “Recalibrating Fully Convo- lutional Networks With Spatial and Channel ’Squeeze and Excitation’ Blocks,”IEEE Transactions on Medical Imaging, vol. 38, pp. 540–549, 2 2019

    work page 2019

  63. [63]

    Encoder- decoder with atrous separable convolution for semantic image seg- mentation,

    L.-C. Chen, Y . Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder- decoder with atrous separable convolution for semantic image seg- mentation,” inComputer Vision – ECCV 2018, V . Ferrari, M. Hebert, C. Sminchisescu, and Y . Weiss, Eds. Cham: Springer International Publishing, 2018, pp. 833–851

    work page 2018

  64. [64]

    SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers,

    E. Xie, W. Wang, Z. Yu, A. Anandkumar, J. M. Alvarez, and P. Luo, “SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers,” inAdvances in Neural Information Processing Systems, M. Ranzato, A. Beygelzimer, Y . Dauphin, P. Liang, and J. W. Vaughan, Eds., vol. 34. Curran Associates, Inc., 2021, pp. 12 077– 12 090

    work page 2021

  65. [65]

    Mapping Debris-Covered Glaciers Using High- Resolution Imagery (GF-2) and Deep Learning Algorithms,

    X. Yang, F. Xie, S. Liu, Y . Zhu, J. Fan, H. Zhao, Y . Fu, Y . Duan, R. Fu, and S. Guo, “Mapping Debris-Covered Glaciers Using High- Resolution Imagery (GF-2) and Deep Learning Algorithms,”Remote Sensing, vol. 16, no. 12, 6 2024

    work page 2024

  66. [66]

    Compilation of a glacier inventory for the western Himalayas from satellite data: methods, challenges, and results,

    H. Frey, F. Paul, and T. Strozzi, “Compilation of a glacier inventory for the western Himalayas from satellite data: methods, challenges, and results,”Remote Sensing of Environment, vol. 124, pp. 832–843, 2012

    work page 2012

  67. [67]

    Challenges and recommendations in mapping of glacier parameters from space: results of the 2008 Global Land Ice Measurements from Space (GLIMS) workshop, Boulder, Colorado, USA,

    A. E. Racoviteanu, F. Paul, B. Raup, S. J. S. Khalsa, and R. Armstrong, “Challenges and recommendations in mapping of glacier parameters from space: results of the 2008 Global Land Ice Measurements from Space (GLIMS) workshop, Boulder, Colorado, USA,”Annals of Glaciol- ogy, vol. 50, no. 53, pp. 53–69, 2009

    work page 2008

  68. [68]

    Error sources and guidelines for quality assessment of glacier area, elevation change, and velocity products derived from satellite data in the Glaciers cci project,

    F. Paul, T. Bolch, A. K ¨a¨ab, T. Nagler, C. Nuth, P. Rastner, T. Strozzi et al., “Error sources and guidelines for quality assessment of glacier area, elevation change, and velocity products derived from satellite data in the Glaciers cci project,”Remote Sensing of Environment, vol. 203, pp. 256–275, 2017

    work page 2017