Articles | Volume 10, issue 2
https://doi.org/10.5194/soil-10-679-2024
https://doi.org/10.5194/soil-10-679-2024
Original research article
 | 
30 Sep 2024
Original research article |  | 30 Sep 2024

Insights into the prediction uncertainty of machine-learning-based digital soil mapping through a local attribution approach

Jeremy Rohmer, Stephane Belbeze, and Dominique Guyonnet

Related authors

Extending the range and reach of physically-based Greenland ice sheet sea-level projections
Heiko Goelzer, Constantijn J. Berends, Fredrik Boberg, Gael Durand, Tamsin Edwards, Xavier Fettweis, Fabien Gillet-Chaulet, Quentin Glaude, Philippe Huybrechts, Sébastien Le clec'h, Ruth Mottram, Brice Noël, Martin Olesen, Charlotte Rahlves, Jeremy Rohmer, Michiel van den Broeke, and Roderik S. W. van de Wal
EGUsphere, https://doi.org/10.5194/egusphere-2025-3098,https://doi.org/10.5194/egusphere-2025-3098, 2025
This preprint is open for discussion and under review for The Cryosphere (TC).
Short summary
Investigating metamodeling capability to predict sea levels and marine flooding maps for early-warning systems: application on the Arcachon Lagoon (France)
Sophie Lecacheux, Jeremy Rohmer, Eva Membrado, Rodrigo Pedreros, Andrea Filippini, Déborah Idier, Servane Gueben-Vénière, Denis Paradis, Alice Dalphinet, and David Ayache
EGUsphere, https://doi.org/10.5194/egusphere-2024-3615,https://doi.org/10.5194/egusphere-2024-3615, 2025
This preprint is open for discussion and under review for Natural Hazards and Earth System Sciences (NHESS).
Short summary
Assessing atoll island future habitability in the context of climate change using Bayesian networks
Mirna Badillo-Interiano, Jérémy Rohmer, Gonéri Le Cozannet, and Virginie Duvat
EGUsphere, https://doi.org/10.5194/egusphere-2024-3884,https://doi.org/10.5194/egusphere-2024-3884, 2025
Short summary
Drawing lessons for multi-model ensemble design from emulator experiments: application to future sea level contribution of the Greenland ice sheet
Jeremy Rohmer, Heiko Goelzer, Tamsin Edwards, Goneri Le Cozannet, and Gael Durand
EGUsphere, https://doi.org/10.5194/egusphere-2025-52,https://doi.org/10.5194/egusphere-2025-52, 2025
Short summary
Improving interpretation of sea-level projections through a machine-learning-based local explanation approach
Jeremy Rohmer, Remi Thieblemont, Goneri Le Cozannet, Heiko Goelzer, and Gael Durand
The Cryosphere, 16, 4637–4657, https://doi.org/10.5194/tc-16-4637-2022,https://doi.org/10.5194/tc-16-4637-2022, 2022
Short summary

Cited articles

Aas, K., Jullum, M., and Løland, A.: Explaining individual predictions when features are dependent: More accurate approximations to Shapley values, Artif. Intell., 298, 103502, https://doi.org/10.1016/j.artint.2021.103502, 2021. 
Abdar, M., Pourpanah, F., Hussain, S., Rezazadegan, D., Liu, L., Ghavamzadeh, M., Fieguth, P., Cao, X., Khosravi, A., Rajendra Acharya, U., Makarenkov, V., and Nahavandi, S.: A review of uncertainty quantification in deep learning: Techniques, applications and challenges, Inform. Fusion, 76, 243–297, 2021. 
Adhikari, K. and Hartemink, A. E.: Linking soils to ecosystem services – A global review, Geoderma, 262, 101–111, 2016. 
Arrouays, D., McBratney, A., Bouma, J., Libohova, Z., Richerde-Forges, A. C., Morgan, C. L. S., Roudier, P., Poggio, L., and Mulder, V. L.: Impressions of digital soil maps: The good, the not so good, and making them ever better, Geoderma Regional, 20, e00255, https://doi.org/10.1016/j.geodrs.2020.e00255, 2020. 
Behrens, T., Schmidt, K., Viscarra Rossel, R. A., Gries, P., Scholten, T., and MacMillan, R. A.: Spatial modelling with Euclidean distance fields and machine learning, Eur. J. Soil Sci., 69, 757–770, 2018. 
Download
Short summary
Machine learning (ML) models have become key ingredients for digital soil mapping. To explain why the ML model is reliable, we apply a popular method from explainable artificial intelligence to the uncertainty prediction, with an application to the mapping of hydrocarbon pollutants on urban soil. We show the benefit of a joint analysis of the influence on the best estimate and the uncertainty to improve communication with end users and support decisions regarding covariates’ characterisation.
Share