Articles | Volume 11, issue 2
https://doi.org/10.5194/soil-11-833-2025
© 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/soil-11-833-2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Original research article
| 
15 Oct 2025
Original research article |
| 15 Oct 2025
| 15 Oct 2025
Quantifying spatial uncertainty to improve soil predictions in data-sparse regions
Department of Geoscience, University of Tübingen, Rümelinstraße 19–23, 72070 Tübingen, Baden-Württemberg, Germany
Cluster of Excellence Machine Learning: New Perspectives for Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Tübingen AI Center, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Katharina Eggensperger
Department of Computer Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Cluster of Excellence Machine Learning: New Perspectives for Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Frank Schneider
Department of Computer Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Tübingen AI Center, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Michael Blaschek
Department 9: State Authority for Geology, Mineral Resources and Mining (LGRB), Regional Council Freiburg, Albertstraße 5, 79104 Freiburg im Breisgau, Baden-Württemberg, Germany
Philipp Hennig
Department of Computer Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Cluster of Excellence Machine Learning: New Perspectives for Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Tübingen AI Center, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Thomas Scholten
Department of Geoscience, University of Tübingen, Rümelinstraße 19–23, 72070 Tübingen, Baden-Württemberg, Germany
Cluster of Excellence Machine Learning: New Perspectives for Science, University of Tübingen, Maria-von-Linden-Straße 6, 72076 Tübingen, Baden-Württemberg, Germany
Related authors
No articles found.
Rabindra Adhikari, Jan Huber, Amrit Maharjan, Madhavi Parajuli, Corinna Gall, Steffen Seitz, Chandra Kanta Subedi, Niels Schwab, Ram Prasad Chaudhary, Udo Schickhoff, Jürgen Böhner, Yvonne Oelmann, and Thomas Scholten
EGUsphere, https://doi.org/10.5194/egusphere-2026-3815, https://doi.org/10.5194/egusphere-2026-3815, 2026
This preprint is open for discussion and under review for SOIL (SOIL).
Short summary
Short summary
Mountain treeline ecosystems are highly sensitive to climate change, yet the role of soil microbes remains poorly understood. We investigated microbial biomass and soil nutrients across a near-natural Himalayan treeline ecotone in Nepal. Our results show that soil horizon development and altitudinal-vegetation zonation are stronger controls on microbial biomass than bulk soil nutrient concentrations, providing new insights into nutrient cycling in mountain soils.
Madhavi Parajuli, Rabindra Adhikari, Xiaoxia Yang, Luisa Brosch, Nadja Fürst, Thomas Scholten, Steffen Seitz, and Corinna Gall
EGUsphere, https://doi.org/10.5194/egusphere-2026-4029, https://doi.org/10.5194/egusphere-2026-4029, 2026
This preprint is open for discussion and under review for Biogeosciences (BG).
Short summary
Short summary
We studied how moss growing on forest soils influences long-term carbon storage in two temperate forests in Germany. By separating soil into different organic matter components, we found that moss can increase the amount of carbon held in the most stable form, especially in coniferous forests. These findings show that moss plays an important role in helping forests store carbon and should be considered in forest carbon assessments and climate change research.
Mathias Bellat, Mjahid Zebari, Benjamin Glissmann, Tobias Rentschler, Paola Sconzo, Nafiseh Kakhani, Ruhollah Taghizadeh-Mehrjardi, Pegah Kohsravani, Bekas Brifkany, Peter Pfälzner, and Thomas Scholten
Earth Syst. Sci. Data, 18, 2507–2548, https://doi.org/10.5194/essd-18-2507-2026, https://doi.org/10.5194/essd-18-2507-2026, 2026
Short summary
Short summary
This dataset presents the first soil maps for the region produced using digital mapping techniques. It includes predictions for ten major physical and chemical soil properties at various depths, plus a map of total soil depth. For each property, we selected the most accurate models and key environmental drivers. In Southwestern Asia and many arid or semi-arid regions, detailed soil data are often missing. This dataset fills that gap, supporting agriculture, research, planning, and local policy.
Kay D. Seufferheld, Pedro V. G. Batista, Hadi Shokati, Thomas Scholten, and Peter Fiener
SOIL, 12, 301–319, https://doi.org/10.5194/soil-12-301-2026, https://doi.org/10.5194/soil-12-301-2026, 2026
Short summary
Short summary
Soil erosion threatens global food security, yet modeling soil conservation remains challenging. We evaluated WaTEM/SEDEM (Water and Tillage Erosion Model/Sediment Delivery Model) in six highly instrumented micro-scale watersheds optimised for soil conservation using a GLUE (Generalized Likelihood Uncertainty Estimation) framework. The model captured the magnitude of very low sediment yields but showed limited accuracy for annual steps. However, it performed well over eight-year timeframes and larger spatial scales, demonstrating its suitability for strategic, long-term soil conservation planning.
Hadi Shokati, Kay D. Seufferheld, Peter Fiener, and Thomas Scholten
Hydrol. Earth Syst. Sci., 30, 743–756, https://doi.org/10.5194/hess-30-743-2026, https://doi.org/10.5194/hess-30-743-2026, 2026
Short summary
Short summary
Floods threaten lives and property and require rapid mapping. We compared two artificial intelligence approaches on aerial imagery: a fine‑tuned Segment Anything Model (SAM) guided by point or bounding box prompts, and a U‑Net network with ResNet‑50 and ResNet‑101 backbones. The point‑based SAM was the most accurate with precise boundaries. Faster and more reliable flood maps help rescue teams, insurers, and planners to act quickly.
Fedor Scholz, Manuel Traub, Christiane Zarfl, Thomas Scholten, and Martin V. Butz
Hydrol. Earth Syst. Sci., 29, 6257–6283, https://doi.org/10.5194/hess-29-6257-2025, https://doi.org/10.5194/hess-29-6257-2025, 2025
Short summary
Short summary
We present a neural network model that estimates river discharge based on gridded elevation, precipitation, and solar radiation. Some instances of our model produce more accurate forecasts than the European Flood Awareness System (EFAS) when simulating discharge with lead times of 50 days on the Neckar river network in Germany. It consists of multiple components that are designed to model distinct sub-processes. We show that this makes the model behave in a more physically realistic way.
Corinna Gall, Silvana Oldenburg, Martin Nebel, Thomas Scholten, and Steffen Seitz
SOIL, 11, 199–212, https://doi.org/10.5194/soil-11-199-2025, https://doi.org/10.5194/soil-11-199-2025, 2025
Short summary
Short summary
Soil erosion is a major issue in vineyards due to often steep slopes and fallow interlines. While cover crops are typically used for erosion control, moss restoration has not yet been explored. In this study, moss restoration reduced surface runoff by 71.4 % and sediment discharge by 75.8 % compared with bare soil, similar to cover crops. Mosses could serve as ground cover where mowing is impractical, potentially reducing herbicide use in viticulture, although further research is needed.
Wanjun Zhang, Thomas Scholten, Steffen Seitz, Qianmei Zhang, Guowei Chu, Linhua Wang, Xin Xiong, and Juxiu Liu
Hydrol. Earth Syst. Sci., 28, 3837–3854, https://doi.org/10.5194/hess-28-3837-2024, https://doi.org/10.5194/hess-28-3837-2024, 2024
Short summary
Short summary
Rainfall input generally controls soil water and plant growth. We focus on rainfall redistribution in succession sequence forests over 22 years. Some changes in rainwater volume and chemistry in the throughfall and stemflow and drivers were investigated. Results show that shifted open rainfall over time and forest factors induced remarkable variability in throughfall and stemflow, which potentially makes forecasting future changes in water resources in the forest ecosystems more difficult.
Nicolás Riveras-Muñoz, Steffen Seitz, Kristina Witzgall, Victoria Rodríguez, Peter Kühn, Carsten W. Mueller, Rómulo Oses, Oscar Seguel, Dirk Wagner, and Thomas Scholten
SOIL, 8, 717–731, https://doi.org/10.5194/soil-8-717-2022, https://doi.org/10.5194/soil-8-717-2022, 2022
Short summary
Short summary
Biological soil crusts (biocrusts) stabilize the soil surface mainly in arid regions but are also present in Mediterranean and humid climates. We studied this stabilizing effect through wet and dry sieving along a large climatic gradient in Chile and found that the stabilization of soil aggregates persists in all climates, but their role is masked and reserved for a limited number of size fractions under humid conditions by higher vegetation and organic matter contents in the topsoil.
Corinna Gall, Martin Nebel, Dietmar Quandt, Thomas Scholten, and Steffen Seitz
Biogeosciences, 19, 3225–3245, https://doi.org/10.5194/bg-19-3225-2022, https://doi.org/10.5194/bg-19-3225-2022, 2022
Short summary
Short summary
Soil erosion is one of the most serious environmental challenges of our time, which also applies to forests when forest soil is disturbed. Biological soil crusts (biocrusts) can play a key role as erosion control. In this study, we combined soil erosion measurements with vegetation surveys in disturbed forest areas. We found that soil erosion was reduced primarily by pioneer bryophyte-dominated biocrusts and that bryophytes contributed more to soil erosion mitigation than vascular plants.
Cited articles
Abbaszadeh Afshar, F., Ayoubi, S., and Jafari, A.: The extrapolation of soil great groups using multinomial logistic regression at regional scale in arid regions of Iran, Geoderma, 315, 36–48, https://doi.org/10.1016/j.geoderma.2017.11.030, 2018. a
Abdar, M., Pourpanah, F., Hussain, S., Rezazadegan, D., Liu, L., Ghavamzadeh, M., Fieguth, P., Cao, X., Khosravi, A., Acharya, U. R., Makarenkov, V., and Nahavandi, S.: A review of uncertainty quantification in deep learning: Techniques, applications and challenges, Inform. Fusion, 76, 243–297, https://doi.org/10.1016/j.inffus.2021.05.008, 2021. a
Amelung, W., Blume, H.-P., Fleige, H., Horn, R., Kandeler, E., Kögel-Knabner, I., Kretzschmar, R., Stahr, K., and Wilke, B.-M.: Scheffer/Schachtschabel Lehrbuch der Bodenkunde, Springer-Verlag, https://doi.org/10.1007/978-3-662-55871-3, 2018. a
Bao, Y., Yao, F., Meng, X., Wang, J., Liu, H., Wang, Y., Liu, Q., Zhang, J., and Mouazen, A. M.: A fine digital soil mapping by integrating remote sensing-based process model and deep learning method in Northeast China, Soil Till. Res., 238, 106010, https://doi.org/10.1016/j.still.2024.106010, 2024. a
Bauer, J., Rohdenburg, H., and Bork, H.: Ein digitales Reliefmodell als Vorraussetzung für ein deterministisches Modell der Wasser-und Stoff-Flüsse, Landschaftsgenese und Landschaftsökologie, 10–15, 1985. a
Behrens, T., Förster, H., Scholten, T., Steinrücken, U., Spies, E.-D., and Goldschmitt, M.: Digital soil mapping using artificial neural networks, J. Plant Nutr. Soil Sc., 168, 21–33, https://doi.org/10.1002/jpln.200421414, 2005. a, b
Belkadi, W. H. and Drias, Y.: Advancements in Digital Soil Mapping: From Data Acquisition to Uncertainty Estimation – A Comprehensive Review, in: Artificial Intelligence Doctoral Symposium, edited by Drias, H., Yalaoui, F., and Hadjali, A., Springer Nature Singapore, Singapore, ISBN 978-981-99-4484-2, 162–177, https://doi.org/10.1007/978-981-99-4484-2_13, 2023. a
Bock, M., Böhner, J., Conrad, O., Köthe, R., and Ringeler, A.: XV. Methods for creating functional soil databases and applying digital soil mapping with SAGA GIS, JRC Scientific and technical Reports, Office for Official Publications of the European Communities, Luxemburg, https://www.researchgate.net/publication/267706527_XV_Methods_for_creating_Functional_Soil_Databases_and_applying_Digital_Soil_Mapping_with_SAGA_GIS/references (last access: 21 September 2025), 2007. a
Bodaghabad, M. B., Martínez-Casasnovas, J., Salehi, M. H., Mohammadi, J., Borujeni, I. E., Toomanian, N., and Gandomkar, A.: Digital soil mapping using artificial neural networks and terrain-related attributes, Pedosphere, 25, 580–591, 2015. a
Bohn, M. P. and Miller, B. A.: Locally enhanced digital soil mapping in support of a bottom-up approach is more accurate than conventional soil mapping and top-down digital soil mapping, Geoderma, 442, 116781, https://doi.org/10.1016/j.geoderma.2024.116781, 2024. a
Böhner, J. and Köthe, R.: Bodenregionalisierung und Prozessmodellierung: Instrumente für den Bodenschutz, Petermann. Geogr. Mitt., 147, 72–82, 2003. a
Boruvka, L. and Penizek, V.: Chapter 30 A Test of an Artificial Neural Network Allocation Procedure using the Czech Soil Survey of Agricultural Land Data, in: Digital Soil Mapping, edited by: Lagacherie, P., McBratney, A., and Voltz, M., vol. 31, Developments in Soil Science, Elsevier, 415–424, https://doi.org/10.1016/S0166-2481(06)31030-6, 2006. a
Breiman, L.: Statistical modeling: The two cultures (with comments and a rejoinder by the author), Stat. Sci., 16, 199–231, https://doi.org/10.1214/ss/1009213726, 2001. a
Bridle, J. S.: Probabilistic Interpretation of Feedforward Classification Network Outputs, with Relationships to Statistical Pattern Recognition, in: Neurocomputing, edited by: Soulié, F. F. and Hérault, J., Springer Berlin Heidelberg, Berlin, Heidelberg, 227–236, ISBN 978-3-642-76153-9, https://doi.org/10.1007/978-3-642-76153-9_28, 1990. a
Broeg, T., Blaschek, M., Seitz, S., Taghizadeh-Mehrjardi, R., Zepp, S., and Scholten, T.: Transferability of Covariates to Predict Soil Organic Carbon in Cropland Soils, Remote Sensing, 15, https://doi.org/10.3390/rs15040876, 2023. a
Brungard, C. W., Boettinger, J. L., Duniway, M. C., Wills, S. A., and Edwards, T. C.: Machine learning for predicting soil classes in three semi-arid landscapes, Geoderma, 239–240, 68–83, https://doi.org/10.1016/j.geoderma.2014.09.019, 2015. a
Bui, E. N. and Moran, C. J.: A strategy to fill gaps in soil survey over large spatial extents: an example from the Murray–Darling basin of Australia, Geoderma, 111, 21–44, https://doi.org/10.1016/S0016-7061(02)00238-0, 2003. a
Carvalho Monteiro, M. E., Avalos, F. P., Procópio Pelegrino, M. H., Brito Vilela, R., Weimar Acerbi Júnior, F., Bueno, I. T., Li, N., Godinho Silva, S. H., Giasson, E., Curi, N., and Duarte de Menezes, M.: Digital mapping of soil classes in Southeast Brazil: environmental covariate selection, accuracy, and uncertainty, J. S. Am. Earth Sci., 132, 104640, https://doi.org/10.1016/j.jsames.2023.104640, 2023. a
Chen, S., Arrouays, D., Leatitia Mulder, V., Poggio, L., Minasny, B., Roudier, P., Libohova, Z., Lagacherie, P., Shi, Z., Hannam, J., Meersmans, J., de Forges, A. C. R., and Walter, C.: Digital mapping of GlobalSoilMap soil properties at a broad scale: A review, Geoderma, 409, 115567, https://doi.org/10.1016/j.geoderma.2021.115567, 2022. a
Coelho, F. F., Giasson, E., Campos, A. R., de Oliveira e Silva, R. G. P., and Costa, J. J. F.: Geographic object-based image analysis and artificial neural networks for digital soil mapping, CATENA, 206, 105568, https://doi.org/10.1016/j.catena.2021.105568, 2021. a
Dash, P. K., Panigrahi, N., and Mishra, A.: Identifying opportunities to improve digital soil mapping in India: A systematic review, Geoderma Regional, 28, https://doi.org/10.1016/j.geodrs.2021.e00478, 2022. a
Daxberger, E. A., Kristiadi, A., Immer, A., Eschenhagen, R., Bauer, M., and Hennig, P.: Laplace Redux – Effortless Bayesian Deep Learning, CoRR, arxiv, https://doi.org/10.48550/arXiv.2106.14806, 2021. a, b
de Arruda, G. P., Dematte, J. A. M., Chagas, C. d. S., Fiorio, P. R., Souza, A. B. e., and Fongaro, C. T.: Digital soil mapping using reference area and artificial neural networks, Sci. Agr., 73, 266–273, https://doi.org/10.1590/0103-9016-2015-0131, 2016. a
Eckelmann, W., Sponagel, H., Grottenthaler, W., Hartmann, K.-J., Hartwich, R., Janetzko, P., Joisten, H., Kühn, D., Sabel, K.-J., and Traidl, R. (Ad-hoc-Arbeitsgruppe Boden) (Eds.): Bodenkundliche Kartieranleitung, 5th edn., Schweizerbart'sche Verlagsbuchhandlung, ISBN 978-3-510-95920-4, 2005. a
Freeman, T. G.: Calculating catchment area with divergent flow based on a regular grid, Comput. Geosci., 17, 413–422, https://doi.org/10.1016/0098-3004(91)90048-I, 1991. 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
Glorot, X., Bordes, A., and Bengio, Y.: Deep sparse rectifier neural networks, JMLR Workshop and Conference Proceedings, 315–323, https://proceedings.mlr.press/v15/glorot11a.html (last access: 21 September 2025), 2011. a
Grinand, C., Arrouays, D., Laroche, B., and Martin, M. P.: Extrapolating regional soil landscapes from an existing soil map: Sampling intensity, validation procedures, and integration of spatial context, Geoderma, 143, 180–190, https://doi.org/10.1016/j.geoderma.2007.11.004, 2008. a, b
Grunwald, S., Thompson, J., and Boettinger, J.: Digital soil mapping and modeling at continental scales: Finding solutions for global issues, Soil Sci. Soc. Am. J., 75, 1201–1213, https://doi.org/10.2136/sssaj2011.0025, 2011. a
Guo, C., Pleiss, G., Sun, Y., and Weinberger, K. Q.: On Calibration of Modern Neural Networks, CoRR, arXiv, https://doi.org/10.48550/arXiv.1706.04599, 2017. a, b
Haykin, S.: Neural Networks: A Comprehensive Foundation, Prentice Hall PTR, ISBN 978-0023527616, 1994. a
Heerdegen, R. G. and Beran, M. A.: Quantifying source areas through land surface curvature and shape, J. Hydrol., 57, 359–373, https://doi.org/10.1016/0022-1694(82)90155-X, 1982. a
Hein, M., Andriushchenko, M., and Bitterwolf, J.: Why relu networks yield high-confidence predictions far away from the training data and how to mitigate the problem, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 41–50, https://doi.org/10.1109/CVPR.2019.00013, 2019. a, b, c
Hengl, T., Mendes de Jesus, J., Heuvelink, G. B. M., Ruiperez Gonzalez, M., Kilibarda, M., Blagotić, A., Shangguan, W., Wright, M. N., Geng, X., Bauer-Marschallinger, B., Guevara, M. A., Vargas, R., MacMillan, R. A., Batjes, N. H., Leenaars, J. G. B., Ribeiro, E., Wheeler, I., Mantel, S., and Kempen, B.: SoilGrids250m: Global gridded soil information based on machine learning, PLoS one, 12, e0169748, https://doi.org/10.1371/journal.pone.0169748, 2017. a
Heung, B., Ho, H. C., Zhang, J., Knudby, A., Bulmer, C. E., and Schmidt, M. G.: An overview and comparison of machine-learning techniques for classification purposes in digital soil mapping, Geoderma, 265, 62–77, https://doi.org/10.1016/j.geoderma.2015.11.014, 2016. a
Heuvelink, G. and Webster, R.: Modelling soil variation: past, present, and future, Geoderma, 100, 269–301, https://doi.org/10.1016/S0016-7061(01)00025-8, 2001. a, b
Hewitt, A.: Predictive modelling in soil survey, Soils and Fertilizers, 56, 305–314, 1993. a
Hounkpatin, K. O., Schmidt, K., Stumpf, F., Forkuor, G., Behrens, T., Scholten, T., Amelung, W., and Welp, G.: Predicting reference soil groups using legacy data: A data pruning and Random Forest approach for tropical environment (Dano catchment, Burkina Faso), Sci. Rep., 8, 1–16, https://doi.org/10.1038/s41598-018-28244-w, 2018. a
Huang, Y.-C., Padarian, J., Minasny, B., and McBratney, A. B.: Using Monte Carlo conformal prediction to evaluate the uncertainty of deep-learning soil spectral models, SOIL, 11, 553–563, https://doi.org/10.5194/soil-11-553-2025, 2025. a
IUSS Working Group: World Reference Base for Soil Resources. International soil classification system for naming soils and creating legends for soil maps, 4th edn., International Union of Soil Sciences (IUSS), https://www.isric.org/sites/default/files/WRB_fourth_edition_2022-12-18.pdf (last access: 21 September 2025), 2022. a
Jenny, H.: Factors of soil formation, 281 pp., New York, 1941. a
Johnson, J. M. and Khoshgoftaar, T. M.: Survey on deep learning with class imbalance, Journal of Big Data, 6, 1–54, https://doi.org/10.1186/s40537-019-0192-5, 2019. a
Kasiviswanathan, K., Sudheer, K., and He, J.: Probabilistic and ensemble simulation approaches for input uncertainty quantification of artificial neural network hydrological models, Hydrolog. Sci. J., 63, 101–113, https://doi.org/10.1080/02626667.2017.1393686, 2018. a
Kebonye, N. M., Agyeman, P. C., and Biney, J. K.: Optimized modelling of countrywide soil organic carbon levels via an interpretable decision tree, Smart Agricultural Technology, 3, 100106, https://doi.org/10.1016/j.atech.2022.100106, 2023. a
Köthe, R. and Bock, M.: Development and use in practice of SAGA modules for high quality analysis of geodata, Free And Open GIS-SAGA-GIS, 115, 85–96, 2006. a
Kotzé, J. and van Tol, J.: Extrapolation of Digital Soil Mapping Approaches for Soil Organic Carbon Stock Predictions in an Afromontane Environment, Land, 12, https://doi.org/10.3390/land12030520, 2023. a
Lagacherie, P., Legros, J., and Burfough, P.: A soil survey procedure using the knowledge of soil pattern established on a previously mapped reference area, Geoderma, 65, 283–301, https://doi.org/10.1016/0016-7061(94)00040-H, 1995. a
Lemercier, B., Lacoste, M., Loum, M., and Walter, C.: Extrapolation at regional scale of local soil knowledge using boosted classification trees: A two-step approach, Geoderma, 171–172, 75–84, https://doi.org/10.1016/j.geoderma.2011.03.010, 2012. a
Li, Y., Yan, S., and Gong, J.: Quantifying uncertainty in soil moisture retrieval using a Bayesian neural network framework, Comput. Electron. Agr., 215, 108414, https://doi.org/10.1016/j.compag.2023.108414, 2023. a
Ludwig, M., Moreno-Martinez, A., Hölzel, N., Pebesma, E., and Meyer, H.: Assessing and improving the transferability of current global spatial prediction models, Global Ecol. Biogeogr., 32, 356–368, https://doi.org/10.1111/geb.13635, 2023. a
McBratney, A., Mendonça Santos, M., and Minasny, B.: On Digital Soil Mapping, Geoderma, 117, 3–52, https://doi.org/10.1016/S0016-7061(03)00223-4, 2003. a, b
Meyer, H. and Pebesma, E.: Machine learning-based global maps of ecological variables and the challenge of assessing them, Nat. Commun., 13, 2208, https://doi.org/10.1038/s41467-022-29838-9, 2022. a
Minasny, B. and McBratney, A.: Digital soil mapping: A brief history and some lessons, Geoderma, 264, 301–311, https://doi.org/10.1016/j.geoderma.2015.07.017, 2016. a
Nair, V. and Hinton, G. E.: Rectified linear units improve restricted boltzmann machines, Proceedings of the 27th International Conference on International Conference on Machine Learning (ICML), 807–814, https://icml.cc/2010/papers/432.pdf (last access: 21 September 2025), 2010. a
Nenkam, A. M., Wadoux, A. M. J.-C., Minasny, B., McBratney, A. B., Traore, P. C. S., Falconnier, G. N., and Whitbread, A. M.: Using homosoils for quantitative extrapolation of soil mapping models, Eur. J. Soil Sci., 73, e13285, https://doi.org/10.1111/ejss.13285, 2022. a
Neyestani, M., Sarmadian, F., Jafari, A., Keshavarzi, A., and Sharififar, A.: Digital mapping of soil classes using spatial extrapolation with imbalanced data, Geoderma Regional, 26, e00422, https://doi.org/10.1016/j.geodrs.2021.e00422, 2021. a, b
Nguyen, A., Yosinski, J., and Clune, J.: Deep neural networks are easily fooled: High confidence predictions for unrecognizable images, Proceedings of the IEEE conference on computer vision and pattern recognition, arXiv, 427–436, https://doi.org/10.48550/arXiv.1412.1897, 2015. a
Padarian, J., McBratney, A. B., and Minasny, B.: Game theory interpretation of digital soil mapping convolutional neural networks, SOIL, 6, 389–397, https://doi.org/10.5194/soil-6-389-2020, 2020a. a
Padarian, J., Minasny, B., and McBratney, A. B.: Machine learning and soil sciences: a review aided by machine learning tools, SOIL, 6, 35–52, https://doi.org/10.5194/soil-6-35-2020, 2020b. a
Padarian, J., Minasny, B., and McBratney, A.: Assessing the uncertainty of deep learning soil spectral models using Monte Carlo dropout, Geoderma, 425, 116063, https://doi.org/10.1016/j.geoderma.2022.116063, 2022. a
Probst, P., Boulesteix, A.-L., and Bischl, B.: Tunability: Importance of Hyperparameters of Machine Learning Algorithms, J. Mach. Learn. Res., 20, 1–32, http://jmlr.org/papers/v20/18-444.html (last access: 21 September 2025), 2019. a
Rentschler, T., Bartelheim, M., Behrens, T., Behrens, T., Díaz-Zorita Bonilla, M., Teuber, S., Scholten, T., and Schmidt, K.: Contextual spatial modelling in the horizontal and vertical domains, Sci. Rep., 12, 9496, https://doi.org/10.1038/s41598-022-13514-5, 2022. a
Saygın, F., Aksoy, H., Alaboz, P., and Dengiz, O.: Different approaches to estimating soil properties for digital soil map integrated with machine learning and remote sensing techniques in a sub-humid ecosystem, Environ. Monit. Assess., 195, 1061, https://doi.org/10.1007/s10661-023-11681-0, 2023. a
Schmidhuber, J.: Deep learning in neural networks: An overview, Neural Networks, 61, 85–117, https://doi.org/10.1016/j.neunet.2014.09.003, 2015. a
Schmidinger, J. and Heuvelink, G. B.: Validation of uncertainty predictions in digital soil mapping, Geoderma, 437, 116585, https://doi.org/10.1016/j.geoderma.2023.116585, 2023. a, b
Schütze, H., Manning, C. D., and Raghavan, P.: Introduction to information retrieval, Cambridge University Press, ISBN 978-0-521-86571-5, 2008. a
Scull, P., Franklin, J., Chadwick, O. A., and McArthur, D.: Predictive soil mapping: a review, Prog. Phys. Geog., 27, 171–197, https://doi.org/10.1191/0309133303pp366ra, 2003. a
Scull, P., Franklin, J., and Chadwick, O.: The application of classification tree analysis to soil type prediction in a desert landscape, Ecol. Model., 181, 1–15, https://doi.org/10.1016/j.ecolmodel.2004.06.036, 2005. a
Silveira, C. T., Oka-Fiori, C., Santos, L. J. C., Sirtoli, A. E., Silva, C. R., and Botelho, M. F.: Soil prediction using artificial neural networks and topographic attributes, Geoderma, 195–196, 165–172, https://doi.org/10.1016/j.geoderma.2012.11.016, 2013. a
Stumpf, F., Schmidt, K., Goebes, P., Behrens, T., Schönbrodt-Stitt, S., Wadoux, A., Xiang, W., and Scholten, T.: Uncertainty-guided sampling to improve digital soil maps, CATENA, 153, 30–38, https://doi.org/10.1016/j.catena.2017.01.033, 2017. a
Taghizadeh-Mehrjardi, R., Sheikhpour, R., Zeraatpisheh, M., Amirian-Chakan, A., Toomanian, N., Kerry, R., and Scholten, T.: Semi-supervised learning for the spatial extrapolation of soil information, Geoderma, 426, 116094, https://doi.org/10.1016/j.geoderma.2022.116094, 2022. a
Taghizadeh-Mehrjardi, R., Nabiollahi, K., Kebonye, N. M., Kakhani, N., Ghebleh-Goydaragh, M., Heung, B., Amirian-Chakan, A., Hossaini, S. M. T., and Scholten, T.: High-performance soil class delineation via UMAP coupled with machine learning in Kurdistan Province, Iran, Geoderma Regional, 36, e00754, https://doi.org/10.1016/j.geodrs.2024.e00754, 2024. a
van der Westhuizen, S., Heuvelink, G. B., and Hofmeyr, D. P.: Multivariate random forest for digital soil mapping, Geoderma, 431, 116365, https://doi.org/10.1016/j.geoderma.2023.116365, 2023. a
Wadoux, A. M.-C. and Molnar, C.: Beyond prediction: methods for interpreting complex models of soil variation, Geoderma, 422, 115953, https://doi.org/10.1016/j.geoderma.2022.115953, 2022. a
Wadoux, A. M.-C., Minasny, B., and McBratney, A. B.: Machine learning for digital soil mapping: Applications, challenges and suggested solutions, Earth-Sci. Rev., 210, 103359, https://doi.org/10.1016/j.earscirev.2020.103359, 2020. a, b, c, d
Wang, L. and Liu, H.: An efficient method for identifying and filling surface depressions in digital elevation models for hydrologic analysis and modelling, Int. J. Geogr. Inf. Sci., 20, 193–213, https://doi.org/10.1080/13658810500433453, 2006. a
Warrick, A. W.: Soil physics companion, CRC press, https://doi.org/10.1201/9781420041651, 2001. a
Wiechmann, H.: Die bodensystematische Kennzeichnung von Auenböden, Stoffhaushalt von Auenökosystemen: Böden und Hydrologie, Schadstoffe, Bewertungen, 19–25, https://doi.org/10.1007/978-3-642-59744-2_2, 2000. a
Zeraatpisheh, M., Jafari, A., Bagheri Bodaghabadi, M., Ayoubi, S., Taghizadeh-Mehrjardi, R., Toomanian, N., Kerry, R., and Xu, M.: Conventional and digital soil mapping in Iran: Past, present, and future, CATENA, 188, 104424, https://doi.org/10.1016/j.catena.2019.104424, 2020. a
Zhang, W., Gu, X., Tang, L., Yin, Y., Liu, D., and Zhang, Y.: Application of machine learning, deep learning and optimization algorithms in geoengineering and geoscience: Comprehensive review and future challenge, Gondwana Res., 109, 1–17, https://doi.org/10.1016/j.gr.2022.03.015, 2022. a, b
Zhu, A.-X.: Mapping soil landscape as spatial continua: The Neural Network Approach, Water Resour. Res., 36, 663–677, https://doi.org/10.1029/1999WR900315, 2000. a, b
Short summary
We developed an uncertainty method to show where machine learning (ML) models predicting soil units are most reliable, especially for transfer tasks. The model was able to correctly predict soil patterns, especially along rivers, in a new but similar region without retraining. It was too confident about common soil types, showing the need for balanced data. This helps improve soil maps and guides better planning for future data collection, saving time and resources while showing uncertainty.
We developed an uncertainty method to show where machine learning (ML) models predicting soil...
