Boundary Aware U-Net for Glacier Segmentation


  • Bibek Aryal The University of Texas at El Paso
  • Katie E. Miles Aberystwyth University
  • Sergio A. Vargas Zesati The University of Texas at El Paso
  • Olac Fuentes The University of Texas at El Paso



Glacier Segmentation, U-Net


Large-scale study of glaciers improves our understanding of global glacier change and is imperative for monitoring the ecological environment, preventing disasters, and studying the effects of global climate change. Glaciers in the Hindu Kush Himalaya (HKH) are particularly interesting as the HKH is one of the world's most sensitive regions for climate change. In this work, we: (1) propose a modified version of the U-Net for large-scale, spatially non-overlapping, clean glacial ice, and debris-covered glacial ice segmentation; (2) introduce a novel self-learning boundary-aware loss to improve debris-covered glacial ice segmentation performance; and (3) propose a feature-wise saliency score to understand the contribution of each feature in the multispectral Landsat 7 imagery for glacier mapping. Our results show that the debris-covered glacial ice segmentation model trained using self-learning boundary-aware loss outperformed the model trained using dice loss. Furthermore, we conclude that red, shortwave infrared, and near-infrared bands have the highest contribution toward debris-covered glacial ice segmentation from Landsat 7 images.


B. Aryal. Glacier Segmentation in Satellite Images for Hindu Kush Himalaya Region. PhD thesis, 2020.

B. Aryal, S. M. Escarzaga, S. A. Vargas Zesati, M. Velez-Reyes, O. Fuentes, and C. Tweedie. Semi-automated semantic segmentation of arctic shorelines using very high-resolution airborne imagery, spectral indices and weakly supervised machine learning approaches. Remote Sensing, 13(22):4572, 2021. doi:10.3390/rs13224572.

S. Bajracharya, S. Maharjan, and F. Shrestha. Clean ice and debris covered glaciers of HKH region. 2011. doi: 10.26066/RDS.31029.

S. R. Bajracharya and B. R. Shrestha. The status of glaciers in the Hindu Kush-Himalayan region. 2011. doi: 10.53055/ICIMOD.551.

S. Baraka, B. Akera, B. Aryal, T. Sherpa, F. Shrestha, A. Ortiz, K. Sankaran, J. M. Lavista Ferres, M. A. Matin, and Y. Bengio. Machine learning for glacier monitoring in the Hindu Kush Himalaya. In NeurIPS 2020 Workshop on Tackling Climate Change with Machine Learning, 2020. URL

R. Bhambri and T. Bolch. Glacier mapping: a review with special reference to the Indian Himalayas. Progress in Physical Geography, 33(5):672–704, 2009. doi: 10.1177/0309133309348112.

M. P. Bishop, J. F. Shroder Jr, and J. L. Ward. SPOT multispectral analysis for producing supraglacial debris-load estimates for Batura glacier, Pakistan. Geocarto International, 10(4):81–90, 1995. doi: 10.1080/ 10106049509354515.

M. P. Bishop, J. F. Shroder Jr, and B. L. Hick- man. SPOT panchromatic imagery and neural networks for information extraction in a complex mountain environment. Geocarto International, 14(2):19–28, 1999. doi: 10.1080/ 10106049908542100.

A. Bokhovkin and E. Burnaev. Boundary loss for remote sensing imagery semantic segmentation. In International Symposium on Neural Networks, pages 388–401. Springer, 2019. doi: 10.1007/978-3-030-22808-8 38.

T. Bolch and U. Kamp. Glacier mapping in high mountains using DEMs, Landsat and ASTER data. In Proceedings of the 8th International Symposium on High Mountain Remote Sensing Cartography. Karl-Franzens-Universitat Graz, 2005.

R. Caruana. Multitask learning. Machine learning, 28(1):41–75, 1997. doi: 10.1023/A: 1007379606734.

K. M. Cuffey and W. S. B. Paterson. The physics of glaciers. Academic Press, 2010. ISBN 9780080919126.

D. Eigen and R. Fergus. Predicting depth, surface normals and semantic labels with a common multi-scale convolutional architecture. In Proceedings of the IEEE International Conference on Computer Vision, pages 2650–2658, 2015.

J. Florath, S. Keller, R. Abarca-del Rio, S. Hinz, G. Staub, and M. Weinmann. Glacier monitoring based on multi-spectral and multi- temporal satellite data: A case study for classification with respect to different snow and ice types. Remote Sensing, 14(4):845, 2022. doi: 10.3390/rs14040845.

Q. He, Z. Zhang, G. Ma, and J. Wu. Glacier identification from Landsat8 OLI imagery us- ing deep U-NET. ISPRS Annals of Photogrammetry, Remote Sensing & Spatial Information Sciences, 5(3), 2020. doi: 10.5194/ isprs-annals-V-3-2020-381-2020.

D. Hendrycks and K. Gimpel. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415, 2016. doi: 10.48550/arXiv. 1606.08415.

W. W. Immerzeel, A. Lutz, M. Andrade, A. Bahl, H. Biemans, T. Bolch, S. Hyde, S. Brumby, B. Davies, A. Elmore, et al. Importance and vulnerability of the world’s water towers. Nature, 577(7790):364–369, 2020. doi: 10.1038/s41586-019-1822-y.

A. Kendall, Y. Gal, and R. Cipolla. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2018.

P. D. Kraaijenbrink, M. Bierkens, A. Lutz, [26] and W. Immerzeel. Impact of a global temperature rise of 1.5 degrees Celsius on Asia’s glaciers. Nature, 549(7671):257–260, 2017. doi: 10.1038/nature23878.

S. Le Moan, A. Mansouri, J. Y. Hardeberg, and Y. Voisin. Saliency for spectral image [27] analysis. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 6(6):2472–2479, 2013. doi: 10.1109/ JSTARS.2013.2257989.

B. Leibe, A. Leonardis, and B. Schiele. Robust [28] object detection with interleaved categorization and segmentation. International Journal of Computer Vision, 77(1):259–289, 2008. doi: 10.1007/s11263-007-0095-3.

W. Luo, Y. Li, R. Urtasun, and R. Zemel. Understanding the effective receptive field in deep convolutional neural networks. Advances [29] in neural information processing systems, 29, 2016. URL

N. Molg, T. Bolch, P. Rastner, T. Strozzi, andF. Paul. A consistent glacier inventory for Karakoram and Pamir derived from Landsat data: distribution of debris cover and mapping challenges. Earth System Science Data, 10(4):1807–1827, 2018.

F. Paul, C. Huggel, and A. K ̈aa ̈b. Combining satellite multispectral image data and a digital elevation model for mapping debris-covered glaciers. Remote sensing of Environment, 89 (4):510–518, 2004.

R. Ranzi, G. Grossi, L. Iacovelli, and S. Taschner. Use of multispectral aster images for mapping debris-covered glaciers within the glims project. In IGARSS 2004. 2004 IEEE International Geoscience and Remote Sensing Symposium, volume 2, pages 1144–1147. IEEE, 2004.

C. Robinson, A. Ortiz, K. Malkin, B. Elias, A. Peng, D. Morris, B. Dilkina, and N. Jojic. Human-machine collaboration for fast land cover mapping. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 2509–2517, 2020. doi: 10.1609/aaai.v34i03.5633.

J. McGlinchy, B. Johnson, B. Muller, M. Joseph, and J. Diaz. Application of UNet fully convolutional neural network to impervious surface segmentation in urban environment from high resolution satellite imagery. In IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium, pages 3915–3918. IEEE, 2019. doi: 10.1109/IGARSS.2019.8900453.

K. E. Miles, B. Hubbard, T. D. Irvine- Fynn, E. S. Miles, D. J. Quincey, and A. V. Rowan. Hydrology of debris-covered glaciers in high mountain Asia. Earth-Science Reviews, 207:103212, 2020. ISSN 0012-8252. doi: 10.1016/j.earscirev.2020.103212. URL

Y. Mohajerani, M. Wood, I. Velicogna, and E. Rignot. Detection of glacier calving margins with convolutional neural networks: A case study. Remote Sensing, 11(1):74, 2019. doi: 10.3390/rs11010074.

O. Ronneberger, P. Fischer, and T. Brox. U-Net: convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention, pages 234– 241. Springer, 2015. ISBN 978-3-319-24574-4.

P. Sermanet, D. Eigen, X. Zhang, M. Mathieu, R. Fergus, and Y. LeCun. Overfeat: integrated recognition, localization and detection using convolutional networks. arXiv preprint arXiv:1312.6229, 2014. doi: 10.48550/arXiv. 1312.6229.

I. A. Shiklomanov et al. Global water resources. Nature and resources, 26(3):34–43, 1990.

K. Simonyan, A. Vedaldi, and A. Zisserman. Deep inside convolutional networks: Visualising image classification models and saliency maps. arXiv preprint arXiv:1312.6034, 2013. doi: 10.48550/arXiv.1312.6034.

S. Tian, Y. Dong, R. Feng, D. Liang, and L. Wang. Mapping mountain glaciers using an improved U-Net model with cSE. International Journal of Digital Earth, 15(1):463–477, 2022. doi: 10.1080/17538947.2022.2036834.

J. Tompson, R. Goroshin, A. Jain, Y. LeCun, and C. Bregler. Efficient object localization using convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 648–656, 2015.

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, 8:83495–83510, 2020. doi: 10.1109/ACCESS. 2020.2991187.

J. Yosinski, J. Clune, A. M. Nguyen, T. J. Fuchs, and H. Lipson. Understanding neural networks through deep visualization. CoRR, abs/1506.06579, 2015. URL http://arxiv. org/abs/1506.06579.

P. Zhang, Y. Ke, Z. Zhang, M. Wang, P. Li, and S. Zhang. Urban land use and land cover classification using novel deep learning models based on high spatial resolution satellite imagery. Sensors, 18(11), 2018. ISSN 1424- 8220. doi: 10.3390/s18113717. URL https: //

M. Zheng, X. Miao, and K. Sankaran. Interactive visualization and representation analysis applied to glacier segmentation. ISPRS International Journal of Geo-Information, 11(8), 2022. ISSN 2220-9964. doi: 10.3390/ijgi11080415. URL