Articles | Volume 18, issue 5
https://doi.org/10.5194/gmd-18-1829-2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/gmd-18-1829-2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Development and technical paper
| 
17 Mar 2025
Development and technical paper |
| 17 Mar 2025
| 17 Mar 2025
Towards deep-learning solutions for classification of automated snow height measurements (CleanSnow v1.0.2)
Jan Svoboda
CORRESPONDING AUTHOR
WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
Marc Ruesch
WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
David Liechti
WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
Corinne Jones
Swiss Data Science Center, ETH Zürich and EPFL, Zurich, Switzerland
Michele Volpi
Swiss Data Science Center, ETH Zürich and EPFL, Zurich, Switzerland
Michael Zehnder
WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
Institute of Integrative Biology, ETH Zürich, Zurich, Switzerland
Jürg Schweizer
WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland
Related authors
No articles found.
Grégoire Bobillier, Bertil Trottet, Bastian Bergfeld, Ron Simenhois, Alec van Herwijnen, Jürg Schweizer, and Johan Gaume
Nat. Hazards Earth Syst. Sci., 25, 2215–2223, https://doi.org/10.5194/nhess-25-2215-2025, https://doi.org/10.5194/nhess-25-2215-2025, 2025
Short summary
Short summary
Our study investigates the initiation of snow slab avalanches. Combining experimental data with numerical simulations, we show that on gentle slopes, cracks form and propagate due to compressive fractures within a weak layer. On steeper slopes, crack velocity can increase dramatically after approximately 5 m due to a fracture mode transition from compression to shear. Understanding these dynamics provides a crucial missing piece in the puzzle of dry-snow slab avalanche formation.
Jakob Boyd Pernov, William H. Aeberhard, Michele Volpi, Eliza Harris, Benjamin Hohermuth, Sakiko Ishino, Ragnhild B. Skeie, Stephan Henne, Ulas Im, Patricia K. Quinn, Lucia M. Upchurch, and Julia Schmale
Atmos. Chem. Phys., 25, 6497–6537, https://doi.org/10.5194/acp-25-6497-2025, https://doi.org/10.5194/acp-25-6497-2025, 2025
Short summary
Short summary
Particulate methanesulfonic acid (MSAp) is vital for the Arctic climate system. Numerical models struggle to reproduce the MSAp seasonal cycle. We evaluate three numerical models and one reanalysis product’s ability to simulate MSAp. We develop data-driven models for MSAp at four Arctic stations. The data-driven models outperform the numerical models and reanalysis product and identified precursor source-, chemical-processing-, and removal-related features as being important for modeling MSAp.
Cristina Pérez-Guillén, Frank Techel, Michele Volpi, and Alec van Herwijnen
Nat. Hazards Earth Syst. Sci., 25, 1331–1351, https://doi.org/10.5194/nhess-25-1331-2025, https://doi.org/10.5194/nhess-25-1331-2025, 2025
Short summary
Short summary
This study assesses the performance and explainability of a random-forest classifier for predicting dry-snow avalanche danger levels during initial live testing. The model achieved ∼ 70 % agreement with human forecasts, performing equally well in nowcast and forecast modes, while capturing the temporal dynamics of avalanche forecasting. The explainability approach enhances the transparency of the model's decision-making process, providing a valuable tool for operational avalanche forecasting.
Amelie Fees, Michael Lombardo, Alec van Herwijnen, Peter Lehmann, and Jürg Schweizer
The Cryosphere, 19, 1453–1468, https://doi.org/10.5194/tc-19-1453-2025, https://doi.org/10.5194/tc-19-1453-2025, 2025
Short summary
Short summary
Glide-snow avalanches release at the soil–snow interface due to a loss of friction, which is suspected to be linked to interfacial water. The importance of the interfacial water was investigated with a spatio-temporal monitoring setup for soil and local snow on an avalanche-prone slope. Seven glide-snow avalanches were released on the monitoring grid (winter seasons 2021/22 to 2023/24) and provided insights into the source, quantity, and spatial distribution of interfacial water before avalanche release.
Michael Lombardo, Amelie Fees, Anders Kaestner, Alec van Herwijnen, Jürg Schweizer, and Peter Lehmann
EGUsphere, https://doi.org/10.5194/egusphere-2025-304, https://doi.org/10.5194/egusphere-2025-304, 2025
Short summary
Short summary
Water flow in snow is important for many applications including snow hydrology and avalanche forecasting. This work investigated the role of capillary forces at the soil-snow interface during capillary rise experiments using neutron radiography. The results showed that the properties of both the snow and the transitional layer below the snow affected the water flow. This work will allow for better representations of water flow across the soil-snow interface in snowpack models.
Stephanie Mayer, Martin Hendrick, Adrien Michel, Bettina Richter, Jürg Schweizer, Heini Wernli, and Alec van Herwijnen
The Cryosphere, 18, 5495–5517, https://doi.org/10.5194/tc-18-5495-2024, https://doi.org/10.5194/tc-18-5495-2024, 2024
Short summary
Short summary
Understanding the impact of climate change on snow avalanche activity is crucial for safeguarding lives and infrastructure. Here, we project changes in avalanche activity in the Swiss Alps throughout the 21st century. Our findings reveal elevation-dependent patterns of change, indicating a decrease in dry-snow avalanches alongside an increase in wet-snow avalanches at elevations above the current treeline. These results underscore the necessity to revisit measures for avalanche risk mitigation.
Alessandro Maissen, Frank Techel, and Michele Volpi
Geosci. Model Dev., 17, 7569–7593, https://doi.org/10.5194/gmd-17-7569-2024, https://doi.org/10.5194/gmd-17-7569-2024, 2024
Short summary
Short summary
By harnessing AI models, this work enables processing large amounts of data, including weather conditions, snowpack characteristics, and historical avalanche data, to predict human-like avalanche forecasts in Switzerland. Our proposed model can significantly assist avalanche forecasters in their decision-making process, thereby facilitating more efficient and accurate predictions crucial for ensuring safety in Switzerland's avalanche-prone regions.
Amelie Fees, Alec van Herwijnen, Michael Lombardo, Jürg Schweizer, and Peter Lehmann
Nat. Hazards Earth Syst. Sci., 24, 3387–3400, https://doi.org/10.5194/nhess-24-3387-2024, https://doi.org/10.5194/nhess-24-3387-2024, 2024
Short summary
Short summary
Glide-snow avalanches release at the ground–snow interface, and their release process is poorly understood. To investigate the influence of spatial variability (snowpack and basal friction) on avalanche release, we developed a 3D, mechanical, threshold-based model that reproduces an observed release area distribution. A sensitivity analysis showed that the distribution was mostly influenced by the basal friction uniformity, while the variations in snowpack properties had little influence.
Andri Simeon, Cristina Pérez-Guillén, Michele Volpi, Christine Seupel, and Alec van Herwijnen
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2024-76, https://doi.org/10.5194/gmd-2024-76, 2024
Revised manuscript under review for GMD
Short summary
Short summary
Avalanche seismic detection systems are key for forecasting, but distinguishing avalanches from other seismic sources remains challenging. We propose novel autoencoder models to automatically extract features and compare them with standard seismic attributes. These features are then used to classify avalanches and noise events. The autoencoder feature classifiers have the highest sensitivity to detect avalanches, while the standard seismic classifier performs better overall.
Stephanie Mayer, Frank Techel, Jürg Schweizer, and Alec van Herwijnen
Nat. Hazards Earth Syst. Sci., 23, 3445–3465, https://doi.org/10.5194/nhess-23-3445-2023, https://doi.org/10.5194/nhess-23-3445-2023, 2023
Short summary
Short summary
We present statistical models to estimate the probability for natural dry-snow avalanche release and avalanche size based on the simulated layering of the snowpack. The benefit of these models is demonstrated in comparison with benchmark models based on the amount of new snow. From the validation with data sets of quality-controlled avalanche observations and danger levels, we conclude that these models may be valuable tools to support forecasting natural dry-snow avalanche activity.
Bastian Bergfeld, Alec van Herwijnen, Grégoire Bobillier, Philipp L. Rosendahl, Philipp Weißgraeber, Valentin Adam, Jürg Dual, and Jürg Schweizer
Nat. Hazards Earth Syst. Sci., 23, 293–315, https://doi.org/10.5194/nhess-23-293-2023, https://doi.org/10.5194/nhess-23-293-2023, 2023
Short summary
Short summary
For a slab avalanche to release, the snowpack must facilitate crack propagation over large distances. Field measurements on crack propagation at this scale are very scarce. We performed a series of experiments, up to 10 m long, over a period of 10 weeks. Beside the temporal evolution of the mechanical properties of the snowpack, we found that crack speeds were highest for tests resulting in full propagation. Based on these findings, an index for self-sustained crack propagation is proposed.
Stephanie Mayer, Alec van Herwijnen, Frank Techel, and Jürg Schweizer
The Cryosphere, 16, 4593–4615, https://doi.org/10.5194/tc-16-4593-2022, https://doi.org/10.5194/tc-16-4593-2022, 2022
Short summary
Short summary
Information on snow instability is crucial for avalanche forecasting. We introduce a novel machine-learning-based method to assess snow instability from snow stratigraphy simulated with the snow cover model SNOWPACK. To develop the model, we compared observed and simulated snow profiles. Our model provides a probability of instability for every layer of a simulated snow profile, which allows detection of the weakest layer and assessment of its degree of instability with one single index.
Cristina Pérez-Guillén, Frank Techel, Martin Hendrick, Michele Volpi, Alec van Herwijnen, Tasko Olevski, Guillaume Obozinski, Fernando Pérez-Cruz, and Jürg Schweizer
Nat. Hazards Earth Syst. Sci., 22, 2031–2056, https://doi.org/10.5194/nhess-22-2031-2022, https://doi.org/10.5194/nhess-22-2031-2022, 2022
Short summary
Short summary
A fully data-driven approach to predicting the danger level for dry-snow avalanche conditions in Switzerland was developed. Two classifiers were trained using a large database of meteorological data, snow cover simulations, and danger levels. The models performed well throughout the Swiss Alps, reaching a performance similar to the current experience-based avalanche forecasts. This approach shows the potential to be a valuable supplementary decision support tool for assessing avalanche hazard.
Achille Capelli, Franziska Koch, Patrick Henkel, Markus Lamm, Florian Appel, Christoph Marty, and Jürg Schweizer
The Cryosphere, 16, 505–531, https://doi.org/10.5194/tc-16-505-2022, https://doi.org/10.5194/tc-16-505-2022, 2022
Short summary
Short summary
Snow occurrence, snow amount, snow density and liquid water content (LWC) can vary considerably with climatic conditions and elevation. We show that low-cost Global Navigation Satellite System (GNSS) sensors as GPS can be used for reliably measuring the amount of water stored in the snowpack or snow water equivalent (SWE), snow depth and the LWC under a broad range of climatic conditions met at different elevations in the Swiss Alps.
Sebastian Landwehr, Michele Volpi, F. Alexander Haumann, Charlotte M. Robinson, Iris Thurnherr, Valerio Ferracci, Andrea Baccarini, Jenny Thomas, Irina Gorodetskaya, Christian Tatzelt, Silvia Henning, Rob L. Modini, Heather J. Forrer, Yajuan Lin, Nicolas Cassar, Rafel Simó, Christel Hassler, Alireza Moallemi, Sarah E. Fawcett, Neil Harris, Ruth Airs, Marzieh H. Derkani, Alberto Alberello, Alessandro Toffoli, Gang Chen, Pablo Rodríguez-Ros, Marina Zamanillo, Pau Cortés-Greus, Lei Xue, Conor G. Bolas, Katherine C. Leonard, Fernando Perez-Cruz, David Walton, and Julia Schmale
Earth Syst. Dynam., 12, 1295–1369, https://doi.org/10.5194/esd-12-1295-2021, https://doi.org/10.5194/esd-12-1295-2021, 2021
Short summary
Short summary
The Antarctic Circumnavigation Expedition surveyed a large number of variables describing the dynamic state of ocean and atmosphere, freshwater cycle, atmospheric chemistry, ocean biogeochemistry, and microbiology in the Southern Ocean. To reduce the dimensionality of the dataset, we apply a sparse principal component analysis and identify temporal patterns from diurnal to seasonal cycles, as well as geographical gradients and
hotspotsof interaction. Code and data are open access.
Bastian Bergfeld, Alec van Herwijnen, Benjamin Reuter, Grégoire Bobillier, Jürg Dual, and Jürg Schweizer
The Cryosphere, 15, 3539–3553, https://doi.org/10.5194/tc-15-3539-2021, https://doi.org/10.5194/tc-15-3539-2021, 2021
Short summary
Short summary
The modern picture of the snow slab avalanche release process involves a
dynamic crack propagation phasein which a whole slope becomes detached. The present work contains the first field methodology which provides the temporal and spatial resolution necessary to study this phase. We demonstrate the versatile capabilities and accuracy of our method by revealing intricate dynamics and present how to determine relevant characteristics of crack propagation such as crack speed.
Jürg Schweizer, Christoph Mitterer, Benjamin Reuter, and Frank Techel
The Cryosphere, 15, 3293–3315, https://doi.org/10.5194/tc-15-3293-2021, https://doi.org/10.5194/tc-15-3293-2021, 2021
Short summary
Short summary
Snow avalanches threaten people and infrastructure in snow-covered mountain regions. To mitigate the effects of avalanches, warnings are issued by public forecasting services. Presently, the five danger levels are described in qualitative terms. We aim to characterize the avalanche danger levels based on expert field observations of snow instability. Our findings contribute to an evidence-based description of danger levels and to improve consistency and accuracy of avalanche forecasts.
Bettina Richter, Alec van Herwijnen, Mathias W. Rotach, and Jürg Schweizer
Nat. Hazards Earth Syst. Sci., 20, 2873–2888, https://doi.org/10.5194/nhess-20-2873-2020, https://doi.org/10.5194/nhess-20-2873-2020, 2020
Short summary
Short summary
We investigated the sensitivity of modeled snow instability to uncertainties in meteorological input, typically found in complex terrain. The formation of the weak layer was very robust due to the long dry period, indicated by a widespread avalanche problem. Once a weak layer has formed, precipitation mostly determined slab and weak layer properties and hence snow instability. When spatially assessing snow instability for avalanche forecasting, accurate precipitation patterns have to be known.
Frank Techel, Karsten Müller, and Jürg Schweizer
The Cryosphere, 14, 3503–3521, https://doi.org/10.5194/tc-14-3503-2020, https://doi.org/10.5194/tc-14-3503-2020, 2020
Short summary
Short summary
Exploring a large data set of snow stability tests and avalanche observations, we quantitatively describe the three key elements that characterize avalanche danger: snowpack stability, the frequency distribution of snowpack stability, and avalanche size. The findings will aid in refining the definitions of the avalanche danger scale and in fostering its consistent usage.
Cited articles
Avanzi, F., De Michele, C., Ghezzi, A., Jommi, C., and Pepe, M.: A processing-modeling routine to use SNOTEL hourly data in snowpack dynamic models, Adv. Water Resour., 73, 16–29, https://doi.org/10.1016/j.advwatres.2014.06.011, 2014. a
Bavay, M. and Egger, T.: MeteoIO 2.4.2: a preprocessing library for meteorological data, Geosci. Model Dev., 7, 3135–3151, https://doi.org/10.5194/gmd-7-3135-2014, 2014. a, b
Blandini, G., Avanzi, F., Gabellani, S., Ponziani, D., Stevenin, H., Ratto, S., Ferraris, L., and Viglione, A.: A random forest approach to quality-checking automatic snow-depth sensor measurements, The Cryosphere, 17, 5317–5333, https://doi.org/10.5194/tc-17-5317-2023, 2023. a
Breiman, L.: Random Forests, Mach. Learn., 45, 5–32, https://doi.org/10.1023/A:1010950718922, 2001. a
Cho, K., van Merriënboer, B., Bahdanau, D., and Bengio, Y.: On the Properties of Neural Machine Translation: Encoder–Decoder Approaches, in: Proceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation, Doha, Qatar, 103–111, https://doi.org/10.3115/v1/W14-4012, 2014. a
Cybenko, G. V.: Approximation by superpositions of a sigmoidal function, Math. Control Signal., 2, 303–314, https://doi.org/10.1007/BF02551274, 1989. a
Domine, F.: Physical properties of snow, in: Encyclopedia of Snow, Ice and Glaciers, edited by: Singh, V. P., Singh, P., and Haritashya, U. K., Springer Netherlands, Dordrecht, https://doi.org/10.1007/978-90-481-2642-2_422, 859–863, 2011. a
Egan, J. P.: Signal detection theory and ROC analysis, Series in Cognition and Perception, Academic Press, New York, NY, ISBN 978-0122328503 1975. a
Fiebrich, C., Morgan, C., McCombs, A., Hall, P., and Mcpherson, R.: Quality assurance procedures for mesoscale meteorological data, J. Atmos. Ocean. Tech., 27, 1565–1582, https://doi.org/10.1175/2010JTECHA1433.1, 2010. a
Flanner, M., Shell, K., Barlage, M., Perovich, D., and Tschudi, M.: Radiative forcing and albedo feedback from the Northern Hemisphere cryosphere between 1979 and 2008, Nat. Geosci., 4, 151–155, https://doi.org/10.1038/NGEO1062, 2011. a
Fontana, F., Rixen, C., Jonas, T., Aberegg, G., and Wunderle, S.: Alpine grassland phenology as seen in AVHRR, VEGETATION, and MODIS NDVI time series – a comparison with in situ measurements, Sensors, 8, 2833–2853, https://doi.org/10.3390/s8042833, 2008. a
Fukushima, K.: Visual Feature Extraction by a Multilayered Network of Analog Threshold Elements, IEEE T. Syst. Sci. Cyb., 5, 322–333, https://doi.org/10.1109/TSSC.1969.300225, 1969. a
Fukushima, K.: Neocognitron: A hierarchical neural network capable of visual pattern recognition, Neural Networks, 1, 119–130, https://doi.org/10.1016/0893-6080(88)90014-7, 1988. a
Gini, C.: Measurement of Inequality of Incomes, The Economic Journal, 31, 124–126, https://doi.org/10.2307/2223319, 1921. a
Goehry, B., Yan, H., Goude, Y., Massart, P., and Poggi, J.-M.: Random forests for time series, REVSTAT-Stat. J., 21, 283–302, https://doi.org/10.57805/revstat.v21i2.400, 2023. a
Good, I. J.: Rational Decisions, J. Roy. Stat. Soc. B Met., 14, 107–114, https://doi.org/10.1111/j.2517-6161.1952.tb00104.x, 1952. a
Goodfellow, I., Bengio, Y., and Courville, A.: Deep Learning, MIT Press, ISBN 9780262035613, 2016. a
He, Y. and Zhao, J.: Temporal convolutional networks for anomaly detection in time series, J. Phys. C, 1213, 042050, https://doi.org/10.1088/1742-6596/1213/4/042050, 2019. a
Herla, F., Haegeli, P., Horton, S., and Mair, P.: A large-scale validation of snowpack simulations in support of avalanche forecasting focusing on critical layers, Nat. Hazards Earth Syst. Sci., 24, 2727–2756, https://doi.org/10.5194/nhess-24-2727-2024, 2024. a
Hewage, P., Behera, A., Trovati, M., Pereira, E., Ghahremani, M., Palmieri, F., and Liu, Y.: Temporal convolutional neural (TCN) network for an effective weather forecasting using time-series data from the local weather station, Soft Computing, 24, 16453–16482, https://doi.org/10.1007/s00500-020-04954-0, 2020. a
Hochreiter, S. and Schmidhuber, J.: Long Short-term Memory, Neural Computation, 9, 1735–1780, https://doi.org/10.1162/neco.1997.9.8.1735, 1997. a, b
Hornik, K., Stinchcombe, M., and White, H.: Multilayer feedforward networks are universal approximators, Neural Networks, 2, 359–366, https://doi.org/10.1016/0893-6080(89)90020-8, 1989. a
Inouye, D. W.: Climate change and phenology, WIREs Climate Change, 13, e764, https://doi.org/10.1002/wcc.764, 2022. a
Ioffe, S. and Szegedy, C.: Batch normalization: accelerating deep network training by reducing internal covariate shift, in: Proceedings of the 32nd International Conference on Machine Learning, Lille, France, 6–11 July 2015, vol. 37, 448–456, http://dblp.uni-trier.de/db/conf/icml/icml2015.html#IoffeS15 (last access: 2 January 2025), 2015. a
Jerome, D., Petry, W., Mooney, K., and Iler, A.: Snowmelt timing acts independently and in conjunction with temperature accumulation to drive subalpine plant phenology, Glob. Change Biol., 27, 5054–5069, https://doi.org/10.1111/gcb.15803, 2021. a
Jonas, T., Rixen, C., Sturm, M., and Stoeckli, V.: How alpine plant growth is linked to snow cover and climate variability, J. Geophys. Res.-Biogeosci., 113, G03013, https://doi.org/10.1029/2007JG000680, 2008. a, b
Jonas, T., Marty, C., and Magnusson, J.: Estimating the snow water equivalent from snow depth measurements in the Swiss Alps, J. Hydrol., 378, 161–167, https://doi.org/10.1016/j.jhydrol.2009.09.021, 2009. a
Kendall, M. G. and Stuart, A.: The advanced theory of statistics. Volume 3: Design and analysis, and time-series, in: Griffin's Statistical Monographs and Courses, vol. 3, Charles Griffin & Co. Ltd., London, ISBN 9780852640692, 1966. a
Kleene, S. C.: Representation of Events in Nerve Nets and Finite Automata, RAND Corporation, Santa Monica, CA, https://doi.org/10.1515/9781400882618-002, 1951. a
Kong, D., McVicar, T., Mingzhong, X., Zhang, Y., Peña-Arancibia, J., Filippa, G., Xie, Y., and Xihui, G.: phenofit: An R package for extracting vegetation phenology from time series remote sensing, Methods Ecol. Evol., 13, 1508–1527, https://doi.org/10.1111/2041-210X.13870, 2022. a
Lam, R., Sanchez-Gonzalez, A., Willson, M., Wirnsberger, P., Fortunato, M., Alet, F., Ravuri, S., Ewalds, T., Eaton-Rosen, Z., Hu, W., Merose, A., Hoyer, S., Holland, G., Vinyals, O., Stott, J., Pritzel, A., Mohamed, S., and Battaglia, P.: Learning skillful medium-range global weather forecasting, Science, 382, 1416–1421, https://doi.org/10.1126/science.adi2336, 2023. a
Lea, C., Vidal, R., Reiter, A., and Hager, G. D.: Temporal Convolutional Networks: A Unified Approach to Action Segmentation, in: Computer Vision – ECCV 2016 Workshops. ECCV 2016. Lecture Notes in Computer Science, edited by: Hua, G. and Jégou, H., vol. 9915, Springer, Cham, https://doi.org/10.1007/978-3-319-49409-8_7, 2016. a, b
Lecun, Y., Bottou, L., Bengio, Y., and Haffner, P.: Gradient-based learning applied to document recognition, P. IEEE, 86, 2278–2324, https://doi.org/10.1109/5.726791, 1998. a
Lehning, M., Bartelt, P., Brown, B., Russi, T., Stöckli, U., and Zimmerli, M.: SNOWPACK model calculations for avalanche warning based upon a network of weather and snow stations, Cold Reg. Sci. Technol., 30, 145–157, https://doi.org/10.1016/S0165-232X(99)00022-1, 1999. a, b, c
Liechti, D. and Schweizer, J.: The Swiss network of automated snow and weather stations for avalanche forecasting – success factors to its robustness and longevity, in: Proceedings of the International Snow Science Workshop, ISSW International Snow Science Workshop, Tromso, Norway, 23–27 September 2024, 1174–1179, https://www.dora.lib4ri.ch/wsl/islandora/object/wsl%3A37841 (last access: 24 September 2024), 2024. a
Lin, T., Goyal, P., Girshick, R. B., He, K., and Dollár, P.: Focal Loss for Dense Object Detection, in: Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy, 22–29 October 2017, 2999–3007, IEEE Computer Society, https://openaccess.thecvf.com/content_ICCV_2017/papers/Lin_Focal_Loss_for_ICCV_2017_paper.pdf (last access: 2 January 2025), 2017. a
Loshchilov, I. and Hutter, F.: Decoupled Weight Decay Regularization, in: Proceedings of the International Conference on Learning Representations, New Orleans, LA, USA, 6–9 May 2019, https://openreview.net/forum?id=Bkg6RiCqY7 (last access: 19 May 2023), 2019. a
Luković, M., Zweifel, R., Thiry, G., Zhang, C., and Schubert, M.: Reconstructing radial stem size changes of trees with machine learning, J. R. Soc. Interface, 19, 20220349, https://doi.org/10.1098/rsif.2022.0349, 2022. a
Matiu, M., Crespi, A., Bertoldi, G., Carmagnola, C. M., Marty, C., Morin, S., Schöner, W., Cat Berro, D., Chiogna, G., De Gregorio, L., Kotlarski, S., Majone, B., Resch, G., Terzago, S., Valt, M., Beozzo, W., Cianfarra, P., Gouttevin, I., Marcolini, G., Notarnicola, C., Petitta, M., Scherrer, S. C., Strasser, U., Winkler, M., Zebisch, M., Cicogna, A., Cremonini, R., Debernardi, A., Faletto, M., Gaddo, M., Giovannini, L., Mercalli, L., Soubeyroux, J.-M., Sušnik, A., Trenti, A., Urbani, S., and Weilguni, V.: Observed snow depth trends in the European Alps: 1971 to 2019, The Cryosphere, 15, 1343–1382, https://doi.org/10.5194/tc-15-1343-2021, 2021. a
McCulloch, W. S. and Pitts, W.: A logical calculus of the ideas immanent in nervous activity, B. Math. Biophys., 5, 115–133, https://doi.org/10.1007/bf02478259, 1943. a
Morin, S., Horton, S., Techel, F., Bavay, M., Coléou, C., Fierz, C., Gobiet, A., Hagenmuller, P., Lafaysse, M., Ližar, M., Mitterer, C., Monti, F., Müller, K., Olefs, M., Snook, J. S., van Herwijnen, A., and Vionnet, V.: Application of physical snowpack models in support of operational avalanche hazard forecasting: a status report on current implementations and prospects for the future, Cold Reg. Sci. Technol., 170, 102910, https://doi.org/10.1016/j.coldregions.2019.102910, 2020. a
Mott, R., Winstral, A., Cluzet, B., Helbig, N., Magnusson, J., Mazzotti, G., Quéno, L., Schirmer, M., Webster, C., and Jonas, T.: Operational snow-hydrological modeling for Switzerland, Front. Earth Sci., 11, 1228158, https://doi.org/10.3389/feart.2023.1228158, 2023. a
Nair, V. and Hinton, G. E.: Rectified linear units improve restricted boltzmann machines, in: Proceedings of the 27th International Conference on International Conference on Machine Learning, Haifa, Israel, ICML’10, 807–814, Omnipress, Madison, WI, USA, ISBN 9781605589077, 2010 a
Pascanu, R., Mikolov, T., and Bengio, Y.: On the difficulty of training recurrent neural networks, in: Proceedings of the International Conference on Machine Learning, vol. 28 of JMLR Workshop and Conference Proceedings, Atlanta, Georgia, USA, 17–19 June 2013, 1310–1318, https://proceedings.mlr.press/v28/pascanu13.html (last access: 16 June 2013), 2013. a
Pearson, R. K.: Data cleaning for dynamic modeling and control, 1999 European Control Conference (ECC), Karlsruhe, Germany, 31 August–3 September 1999, 1999 2584–2589, https://doi.org/10.23919/ECC.1999.7099714, 1999. a
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., and Duchesnay, E.: Scikit-learn: machine learning in Python, J. Mach. Learn. Res., 12, 2825–2830, 2011. a
Pelletier, C., Webb, G. I., and Petitjean, F.: Temporal convolutional neural network for the classification of satellite image time series, Remote Sens.-Basel, 11, 523, https://doi.org/10.3390/rs11050523, 2019. a
Pérez-Guillén, C., Techel, F., Hendrick, M., Volpi, M., van Herwijnen, A., Olevski, T., Obozinski, G., Pérez-Cruz, F., and Schweizer, J.: Data-driven automated predictions of the avalanche danger level for dry-snow conditions in Switzerland, Nat. Hazards Earth Syst. Sci., 22, 2031–2056, https://doi.org/10.5194/nhess-22-2031-2022, 2022. a
Raissi, M., Perdikaris, P., and Karniadakis, G.: Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, J. Comput. Phys., 378, 686–707, https://doi.org/10.1016/j.jcp.2018.10.045, 2019. a
Robinson, D.: Evaluation of the collection, archiving and publication of daily snow data in the United States, Phys. Geogr., 10, 120–130, https://doi.org/10.1080/02723646.1989.10642372, 1989. a
Rosenblatt, F.: The perceptron: a probabilistic model for information storage and organization in the brain, Psychol. Rev., 65, 386–408, https://doi.org/10.1037/h0042519, 1958. a
Ryan, W. A., Doesken, N. J., and Fassnacht, S. R.: Evaluation of ultrasonic snow depth sensors for US snow measurements, J. Atmos. Ocean. Tech., 25, 667–684, https://doi.org/10.1175/2007JTECHA947.1, 2008. a
Svoboda, J., Ruesch, M., Liechti, D., Jones, C., Volpi, M., Zehnder, M., and Schweizer, J.: Snow Height Classification Dataset, Zenodo [data set], https://doi.org/10.5281/zenodo.13324736, 2024. a
Svoboda, J., Ruesch, M., Liechti, D., Jones, C., Volpi, M., Zehnder, M., and Schweizer, J.: Towards deep learning solutions for classification of automated snow height measurements, GitLab [code], https://gitlabext.wsl.ch/jan.svoboda/snow-height-classification (last access: 2 January 2025), 2025a. a
Svoboda, J., Ruesch, M., Liechti, D., Jones, C., Volpi, M., Zehnder, M., and Schweizer, J.: Towards deep learning solutions for classification of automated snow height measurements (CleanSnow v1.0.2), Zenodo [code], https://doi.org/10.5281/zenodo.14587841, 2025b. a
Tilg, A.-M., Marty, C., and Klein, G.: An automatic algorithm for validating snow depth measurements of IMIS stations (Abstract), 13th Swiss Geoscience Meeting, Basel, Switzerland, 20–21 November 2015, p. 339, https://geoscience-meeting.ch/sgm2015_archived/wp-content/uploads/Abstract_Volume_SGM_2015.pdf (last access: 15 October 2015), 2015. a, b, c
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., and Polosukhin, I.: Attention is all you need, in: Advances in Neural Information Processing Systems, 5998–6008, https://papers.neurips.cc/paper/7181-attention-is-all-you-need.pdf (last access: 2 January 2025), 2017. a, b
Vaughan, A., Tebbutt, W., Hosking, J. S., and Turner, R. E.: Convolutional conditional neural processes for local climate downscaling, Geosci. Model Dev., 15, 251–268, https://doi.org/10.5194/gmd-15-251-2022, 2022. a
Vitasse, Y., Rebetez, M., Filippa, G., Cremonese, E., Klein, G., and Rixen, C.: “Hearing' alpine plants growing after snowmelt: ultrasonic snow sensors provide long-term series of alpine plant phenology, Int. J. Biometeorol., 61, 349–361, https://doi.org/10.1007/s00484-016-1216-x, 2017. a
Waibel, A., Hanazawa, T., Hinton, G., Shikano, K., and Lang, K.: Phoneme recognition using time-delay neural networks, IEEE T. Acoust. Speech, 37, 328–339, https://doi.org/10.1109/29.21701, 1989. a
Wan, R., Mei, S., Wang, J., Liu, M., and Yang, F.: Multivariate temporal convolutional network: a deep neural networks approach for multivariate time series forecasting, Electronics, 8, 876, https://doi.org/10.3390/electronics8080876, 2019. a
Weng, J., Ahuja, N., and Huang, T.: Learning recognition and segmentation of 3-D objects from 2-D images, in: 1993 (4th) International Conference on Computer Vision, Berlin, Germany, 11–14 May 1993, 121–128, https://doi.org/10.1109/ICCV.1993.378228, 1993. a
Willibald, F., Kotlarski, S., Ebner, P. P., Bavay, M., Marty, C., Trentini, F. V., Ludwig, R., and Grêt-Regamey, A.: Vulnerability of ski tourism towards internal climate variability and climate change in the Swiss Alps, Sci. Total Environ., 784, 147054, https://doi.org/10.1016/j.scitotenv.2021.147054, 2021. a
Ying, X.: An overview of overfitting and its solutions, J. Phys. C, 1168, 022022, https://doi.org/10.1088/1742-6596/1168/2/022022, 2019. a
Short summary
Accurately measuring snow height is key for modeling approaches in climate science, snow hydrology, and avalanche forecasting. Erroneous snow height measurements often occur when snow height is low or changes, for instance during snowfall in summer. We prepare a new benchmark dataset with annotated snow height data and demonstrate how to improve the measurement quality using modern deep-learning approaches. Our approach can be easily implemented in a data pipeline for snow modeling.
Accurately measuring snow height is key for modeling approaches in climate science, snow...
