LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping

Digital soil mapping (DSM) relies on a broad pool of statistical methods, yet determining the optimal method for a given context remains challenging and contentious. Benchmarking studies on multiple datasets are needed to reveal strengths and limitations of commonly used methods. Existing DSM studie...

Full description

Saved in:
Bibliographic Details
Main Authors: Jonas Schmidinger, Sebastian Vogel, Viacheslav Barkov, Anh-Duy Pham, Robin Gebbers, Hamed Tavakoli, Jose Correa, Tiago R. Tavares, Patrick Filippi, Edward J. Jones, Vojtech Lukas, Eric Boenecke, Joerg Ruehlmann, Ingmar Schroeter, Eckart Kramer, Stefan Paetzold, Masakazu Kodaira, Alexandre M.J.-C. Wadoux, Luca Bragazza, Konrad Metzger, Jingyi Huang, Domingos S.M. Valente, Jose L. Safanelli, Eduardo L. Bottega, Ricardo S.D. Dalmolin, Csilla Farkas, Alexander Steiger, Taciara Z. Horst, Leonardo Ramirez-Lopez, Thomas Scholten, Felix Stumpf, Pablo Rosso, Marcelo M. Costa, Rodrigo S. Zandonadi, Johanna Wetterlind, Martin Atzmueller
Format: Article
Language:English
Published: Elsevier 2025-07-01
Series:Geoderma
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S0016706125001752
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849687978022010880
author Jonas Schmidinger
Sebastian Vogel
Viacheslav Barkov
Anh-Duy Pham
Robin Gebbers
Hamed Tavakoli
Jose Correa
Tiago R. Tavares
Patrick Filippi
Edward J. Jones
Vojtech Lukas
Eric Boenecke
Joerg Ruehlmann
Ingmar Schroeter
Eckart Kramer
Stefan Paetzold
Masakazu Kodaira
Alexandre M.J.-C. Wadoux
Luca Bragazza
Konrad Metzger
Jingyi Huang
Domingos S.M. Valente
Jose L. Safanelli
Eduardo L. Bottega
Ricardo S.D. Dalmolin
Csilla Farkas
Alexander Steiger
Taciara Z. Horst
Leonardo Ramirez-Lopez
Thomas Scholten
Felix Stumpf
Pablo Rosso
Marcelo M. Costa
Rodrigo S. Zandonadi
Johanna Wetterlind
Martin Atzmueller
author_facet Jonas Schmidinger
Sebastian Vogel
Viacheslav Barkov
Anh-Duy Pham
Robin Gebbers
Hamed Tavakoli
Jose Correa
Tiago R. Tavares
Patrick Filippi
Edward J. Jones
Vojtech Lukas
Eric Boenecke
Joerg Ruehlmann
Ingmar Schroeter
Eckart Kramer
Stefan Paetzold
Masakazu Kodaira
Alexandre M.J.-C. Wadoux
Luca Bragazza
Konrad Metzger
Jingyi Huang
Domingos S.M. Valente
Jose L. Safanelli
Eduardo L. Bottega
Ricardo S.D. Dalmolin
Csilla Farkas
Alexander Steiger
Taciara Z. Horst
Leonardo Ramirez-Lopez
Thomas Scholten
Felix Stumpf
Pablo Rosso
Marcelo M. Costa
Rodrigo S. Zandonadi
Johanna Wetterlind
Martin Atzmueller
author_sort Jonas Schmidinger
collection DOAJ
description Digital soil mapping (DSM) relies on a broad pool of statistical methods, yet determining the optimal method for a given context remains challenging and contentious. Benchmarking studies on multiple datasets are needed to reveal strengths and limitations of commonly used methods. Existing DSM studies usually rely on a single dataset with restricted access, leading to incomplete and potentially misleading conclusions. To address these issues, we introduce an open-access dataset collection called Precision Liming Soil Datasets (LimeSoDa). LimeSoDa consists of 31 field- and farm-scale datasets from various countries. Each dataset has three target soil properties: (1) soil organic matter or soil organic carbon, (2) clay content and (3) pH, alongside a set of features. Features are dataset-specific and were obtained by optical spectroscopy, proximal- and remote soil sensing. All datasets were aligned to a tabular format and are ready-to-use for modeling. We demonstrated the use of LimeSoDa for benchmarking by comparing the predictive performance of four learning algorithms across all datasets. This comparison included multiple linear regression (MLR), support vector regression (SVR), categorical boosting (CatBoost) and random forest (RF). The results showed that although no single algorithm was universally superior, certain algorithms performed better in specific contexts. MLR and SVR performed better on high-dimensional spectral datasets, likely due to better compatibility with principal components. In contrast, CatBoost and RF exhibited considerably better performances when applied to datasets with a moderate number (<20) of features. These benchmarking results illustrate that the performance of statistical methods can be highly context-dependent. LimeSoDa therefore provides an important resource for improving the development and evaluation of statistical methods in DSM.
format Article
id doaj-art-d9049181b7e0408fb32fcade915178a7
institution DOAJ
issn 1872-6259
language English
publishDate 2025-07-01
publisher Elsevier
record_format Article
series Geoderma
spelling doaj-art-d9049181b7e0408fb32fcade915178a72025-08-20T03:22:11ZengElsevierGeoderma1872-62592025-07-0145911733710.1016/j.geoderma.2025.117337LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mappingJonas Schmidinger0Sebastian Vogel1Viacheslav Barkov2Anh-Duy Pham3Robin Gebbers4Hamed Tavakoli5Jose Correa6Tiago R. Tavares7Patrick Filippi8Edward J. Jones9Vojtech Lukas10Eric Boenecke11Joerg Ruehlmann12Ingmar Schroeter13Eckart Kramer14Stefan Paetzold15Masakazu Kodaira16Alexandre M.J.-C. Wadoux17Luca Bragazza18Konrad Metzger19Jingyi Huang20Domingos S.M. Valente21Jose L. Safanelli22Eduardo L. Bottega23Ricardo S.D. Dalmolin24Csilla Farkas25Alexander Steiger26Taciara Z. Horst27Leonardo Ramirez-Lopez28Thomas Scholten29Felix Stumpf30Pablo Rosso31Marcelo M. Costa32Rodrigo S. Zandonadi33Johanna Wetterlind34Martin Atzmueller35Osnabrück University, Joint Lab Artificial Intelligence and Data Science, Osnabrück, Germany; Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, Germany; Corresponding author.Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, GermanyOsnabrück University, Joint Lab Artificial Intelligence and Data Science, Osnabrück, Germany; Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, GermanyOsnabrück University, Joint Lab Artificial Intelligence and Data Science, Osnabrück, Germany; Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, GermanyLeibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, GermanyLeibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, GermanyLeibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Department of Agromechatronics, Potsdam, GermanyUniversity of São Paulo (USP), Center of Nuclear Energy in Agriculture (CENA), Piracicaba, BrazilThe University of Sydney, Sydney Institute of Agriculture, Sydney, AustraliaThe University of Sydney, Sydney Institute of Agriculture, Sydney, AustraliaMendel University in Brno, Department of Agrosystems and Bioclimatology, Brno, Czech RepublicLeibniz Institute of Vegetable and Ornamental Crops, Next Generation Horticultural Systems, Grossbeeren, GermanyLeibniz Institute of Vegetable and Ornamental Crops, Next Generation Horticultural Systems, Grossbeeren, GermanyEberswalde University for Sustainable Development, Landscape Management and Nature Conservation, Eberswalde, GermanyEberswalde University for Sustainable Development, Landscape Management and Nature Conservation, Eberswalde, GermanyUniversity of Bonn, Institute of Crop Science and Resource Conservation (INRES)—Soil Science and Soil Ecology, Bonn, GermanyTokyo University of Agriculture and Technology, Institute of Agriculture, Tokyo, JapanLISAH, Univ. Montpellier, AgroParisTech, INRAE, IRD, L’Institut Agro, Montpellier, FranceAgroscope, Field-Crop Systems and Plant Nutrition, Nyon, SwitzerlandAgroscope, Field-Crop Systems and Plant Nutrition, Nyon, SwitzerlandUniversity of Wisconsin-Madison, Department of Soil Science, Madison, USAFederal University of Viçosa, Department of Agricultural Engineering, Viçosa, BrazilWoodwell Climate Research Center, Falmouth, USAFederal University of Santa Maria (UFSM), Academic Coordination, Santa Maria, BrazilFederal University of Santa Maria (UFSM), Soil Department, Santa Maria, BrazilNorwegian Institute of Bioeconomy Research (NIBIO), Division of Environment and Natural Resources, Aas, NorwayUniversity of Rostock, Chair of Geodesy and Geoinformatics, Rostock, GermanyFederal Technological University of Paraná, Dois Vizinhos, BrazilBÜCHI Labortechnik AG, Data Science Department, Flawil, Switzerland; Imperial College London, Imperial College Business School, London, UKUniversity of Tübingen, Department of Geosciences, Tübingen, Germany; University of Tübingen, DFG Cluster of Excellence ‘Machine Learning for Science’, GermanyBern University of Applied Sciences, Competence Center for Soils, Zollikofen, SwitzerlandLeibniz Centre for Agricultural Landscape Research (ZALF), Simulation and Data Science, Müncheberg, GermanyFederal University of Jataí, Institute of Agricultural Sciences, Jatai, BrazilFederal University of Mato Grosso, Instute of Agricultural and Environmental Scinces, Sinop, BrazilSwedish University of Agricultural Sciences (SLU), Department of Soil and Environment, Skara, SwedenOsnabrück University, Joint Lab Artificial Intelligence and Data Science, Osnabrück, Germany; German Research Center for Artificial Intelligence (DFKI), Research Department Plan-Based Robot Control, Osnabrück, GermanyDigital soil mapping (DSM) relies on a broad pool of statistical methods, yet determining the optimal method for a given context remains challenging and contentious. Benchmarking studies on multiple datasets are needed to reveal strengths and limitations of commonly used methods. Existing DSM studies usually rely on a single dataset with restricted access, leading to incomplete and potentially misleading conclusions. To address these issues, we introduce an open-access dataset collection called Precision Liming Soil Datasets (LimeSoDa). LimeSoDa consists of 31 field- and farm-scale datasets from various countries. Each dataset has three target soil properties: (1) soil organic matter or soil organic carbon, (2) clay content and (3) pH, alongside a set of features. Features are dataset-specific and were obtained by optical spectroscopy, proximal- and remote soil sensing. All datasets were aligned to a tabular format and are ready-to-use for modeling. We demonstrated the use of LimeSoDa for benchmarking by comparing the predictive performance of four learning algorithms across all datasets. This comparison included multiple linear regression (MLR), support vector regression (SVR), categorical boosting (CatBoost) and random forest (RF). The results showed that although no single algorithm was universally superior, certain algorithms performed better in specific contexts. MLR and SVR performed better on high-dimensional spectral datasets, likely due to better compatibility with principal components. In contrast, CatBoost and RF exhibited considerably better performances when applied to datasets with a moderate number (<20) of features. These benchmarking results illustrate that the performance of statistical methods can be highly context-dependent. LimeSoDa therefore provides an important resource for improving the development and evaluation of statistical methods in DSM.http://www.sciencedirect.com/science/article/pii/S0016706125001752Machine learningBenchmarkingOpen dataDataset collectionDigital soil mappingPedometrics
spellingShingle Jonas Schmidinger
Sebastian Vogel
Viacheslav Barkov
Anh-Duy Pham
Robin Gebbers
Hamed Tavakoli
Jose Correa
Tiago R. Tavares
Patrick Filippi
Edward J. Jones
Vojtech Lukas
Eric Boenecke
Joerg Ruehlmann
Ingmar Schroeter
Eckart Kramer
Stefan Paetzold
Masakazu Kodaira
Alexandre M.J.-C. Wadoux
Luca Bragazza
Konrad Metzger
Jingyi Huang
Domingos S.M. Valente
Jose L. Safanelli
Eduardo L. Bottega
Ricardo S.D. Dalmolin
Csilla Farkas
Alexander Steiger
Taciara Z. Horst
Leonardo Ramirez-Lopez
Thomas Scholten
Felix Stumpf
Pablo Rosso
Marcelo M. Costa
Rodrigo S. Zandonadi
Johanna Wetterlind
Martin Atzmueller
LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping
Geoderma
Machine learning
Benchmarking
Open data
Dataset collection
Digital soil mapping
Pedometrics
title LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping
title_full LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping
title_fullStr LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping
title_full_unstemmed LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping
title_short LimeSoDa: A dataset collection for benchmarking of machine learning regressors in digital soil mapping
title_sort limesoda a dataset collection for benchmarking of machine learning regressors in digital soil mapping
topic Machine learning
Benchmarking
Open data
Dataset collection
Digital soil mapping
Pedometrics
url http://www.sciencedirect.com/science/article/pii/S0016706125001752
work_keys_str_mv AT jonasschmidinger limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT sebastianvogel limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT viacheslavbarkov limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT anhduypham limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT robingebbers limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT hamedtavakoli limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT josecorrea limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT tiagortavares limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT patrickfilippi limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT edwardjjones limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT vojtechlukas limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT ericboenecke limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT joergruehlmann limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT ingmarschroeter limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT eckartkramer limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT stefanpaetzold limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT masakazukodaira limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT alexandremjcwadoux limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT lucabragazza limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT konradmetzger limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT jingyihuang limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT domingossmvalente limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT joselsafanelli limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT eduardolbottega limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT ricardosddalmolin limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT csillafarkas limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT alexandersteiger limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT taciarazhorst limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT leonardoramirezlopez limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT thomasscholten limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT felixstumpf limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT pablorosso limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT marcelomcosta limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT rodrigoszandonadi limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT johannawetterlind limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping
AT martinatzmueller limesodaadatasetcollectionforbenchmarkingofmachinelearningregressorsindigitalsoilmapping