TY  - JOUR
AU  - Witjes, Martijn
AU  - Parente, Leandro
AU  - J van Diemen, Chris
AU  - Hengl, Tomislav
AU  - Landa, Martin
AU  - Brodský, Lukáš
AU  - Halounova, Lena
AU  - Križan, Josip
AU  - Antonić, Luka
AU  - Ilie, Codrina Maria
AU  - Craciunescu, Vasile
AU  - Kilibarda, Milan
AU  - Antonijević, Ognjen
AU  - Glušica, Luka
PY  - 2022
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/3343
AB  - A spatiotemporal machine learning framework for automated prediction and analysis of long-term Land Use/Land Cover dynamics is presented. The framework includes: (1) harmonization and preprocessing of spatial and spatiotemporal input datasets (GLAD Landsat, NPP/VIIRS) including five million harmonized LUCAS and CORINE Land Cover-derived training samples, (2) model building based on spatial k-fold cross-validation and hyper-parameter optimization, (3) prediction of the most probable class, class probabilities and model variance of predicted probabilities per pixel, (4) LULC change analysis on time-series of produced maps. The spatiotemporal ensemble model consists of a random forest, gradient boosted tree classifier, and an artificial neural network, with a logistic regressor as meta-learner. The results show that the most important variables for mapping LULC in Europe are: seasonal aggregates of Landsat green and near-infrared bands, multiple Landsat-derived spectral indices, long-term surface water probability, and elevation. Spatial cross-validation of the model indicates consistent performance across multiple years with overall accuracy (a weighted F1-score) of 0.49, 0.63, and 0.83 when predicting 43 (level-3), 14 (level-2), and five classes (level-1). Additional experiments show that spatiotemporal models generalize better to unknown years, outperforming single-year models on known-year classification by 2.7% and unknown-year classification by 3.5%. Results of the accuracy assessment using 48,365 independent test samples shows 87% match with the validation points. Results of time-series analysis (time-series of LULC probabilities and NDVI images) suggest forest loss in large parts of Sweden, the Alps, and Scotland. Positive and negative trends in NDVI in general match the land degradation and land restoration classes, with “urbanization” showing the most negative NDVI trend. An advantage of using spatiotemporal ML is that the fitted model can be used to predict LULC in years that were not included in its training dataset, allowing generalization to past and future periods, e.g. to predict LULC for years prior to 2000 and beyond 2020. The generated LULC time-series data stack (ODSE-LULC), including the training points, is publicly available via the ODSE Viewer. Functions used to prepare data and run modeling are available via the eumap library for Python.
T2  - PeerJ
T1  - A spatiotemporal ensemble machine learning framework for generating land use/land cover time-series maps for Europe (2000–2019) based on LUCAS, CORINE and GLAD Landsat
IS  - 10
SP  - e13573
DO  - doi.org/10.7717/peerj.13573
ER  - 

@article{
author = "Witjes, Martijn and Parente, Leandro and J van Diemen, Chris and Hengl, Tomislav and Landa, Martin and Brodský, Lukáš and Halounova, Lena and Križan, Josip and Antonić, Luka and Ilie, Codrina Maria and Craciunescu, Vasile and Kilibarda, Milan and Antonijević, Ognjen and Glušica, Luka",
year = "2022",
abstract = "A spatiotemporal machine learning framework for automated prediction and analysis of long-term Land Use/Land Cover dynamics is presented. The framework includes: (1) harmonization and preprocessing of spatial and spatiotemporal input datasets (GLAD Landsat, NPP/VIIRS) including five million harmonized LUCAS and CORINE Land Cover-derived training samples, (2) model building based on spatial k-fold cross-validation and hyper-parameter optimization, (3) prediction of the most probable class, class probabilities and model variance of predicted probabilities per pixel, (4) LULC change analysis on time-series of produced maps. The spatiotemporal ensemble model consists of a random forest, gradient boosted tree classifier, and an artificial neural network, with a logistic regressor as meta-learner. The results show that the most important variables for mapping LULC in Europe are: seasonal aggregates of Landsat green and near-infrared bands, multiple Landsat-derived spectral indices, long-term surface water probability, and elevation. Spatial cross-validation of the model indicates consistent performance across multiple years with overall accuracy (a weighted F1-score) of 0.49, 0.63, and 0.83 when predicting 43 (level-3), 14 (level-2), and five classes (level-1). Additional experiments show that spatiotemporal models generalize better to unknown years, outperforming single-year models on known-year classification by 2.7% and unknown-year classification by 3.5%. Results of the accuracy assessment using 48,365 independent test samples shows 87% match with the validation points. Results of time-series analysis (time-series of LULC probabilities and NDVI images) suggest forest loss in large parts of Sweden, the Alps, and Scotland. Positive and negative trends in NDVI in general match the land degradation and land restoration classes, with “urbanization” showing the most negative NDVI trend. An advantage of using spatiotemporal ML is that the fitted model can be used to predict LULC in years that were not included in its training dataset, allowing generalization to past and future periods, e.g. to predict LULC for years prior to 2000 and beyond 2020. The generated LULC time-series data stack (ODSE-LULC), including the training points, is publicly available via the ODSE Viewer. Functions used to prepare data and run modeling are available via the eumap library for Python.",
journal = "PeerJ",
title = "A spatiotemporal ensemble machine learning framework for generating land use/land cover time-series maps for Europe (2000–2019) based on LUCAS, CORINE and GLAD Landsat",
number = "10",
pages = "e13573",
doi = "doi.org/10.7717/peerj.13573"
}

Witjes, M., Parente, L., J van Diemen, C., Hengl, T., Landa, M., Brodský, L., Halounova, L., Križan, J., Antonić, L., Ilie, C. M., Craciunescu, V., Kilibarda, M., Antonijević, O.,& Glušica, L.. (2022). A spatiotemporal ensemble machine learning framework for generating land use/land cover time-series maps for Europe (2000–2019) based on LUCAS, CORINE and GLAD Landsat. in PeerJ(10), e13573.
https://doi.org/doi.org/10.7717/peerj.13573

Witjes M, Parente L, J van Diemen C, Hengl T, Landa M, Brodský L, Halounova L, Križan J, Antonić L, Ilie CM, Craciunescu V, Kilibarda M, Antonijević O, Glušica L. A spatiotemporal ensemble machine learning framework for generating land use/land cover time-series maps for Europe (2000–2019) based on LUCAS, CORINE and GLAD Landsat. in PeerJ. 2022;(10):e13573.
doi:doi.org/10.7717/peerj.13573 .

Witjes, Martijn, Parente, Leandro, J van Diemen, Chris, Hengl, Tomislav, Landa, Martin, Brodský, Lukáš, Halounova, Lena, Križan, Josip, Antonić, Luka, Ilie, Codrina Maria, Craciunescu, Vasile, Kilibarda, Milan, Antonijević, Ognjen, Glušica, Luka, "A spatiotemporal ensemble machine learning framework for generating land use/land cover time-series maps for Europe (2000–2019) based on LUCAS, CORINE and GLAD Landsat" in PeerJ, no. 10 (2022):e13573,
https://doi.org/doi.org/10.7717/peerj.13573 . .

TY  - JOUR
AU  - Hengl, Tomislav
AU  - Miller, Matthew A. E.
AU  - Križan, Josip
AU  - Shepherd, Keith D.
AU  - Sila, Andrew
AU  - Kilibarda, Milan
AU  - Antonijević, Ognjen
AU  - Glušica, Luka
AU  - Dobermann, Achim
AU  - Haefele, Stephan M.
AU  - McGrath, Steve P.
AU  - Acquah, Gifty E.
AU  - Collinson, Jamie
AU  - Parente, Leandro
AU  - Sheykhmousa, Mohammadreza
AU  - Saito, Kazuki
AU  - Johnson, Jean‑Martial
AU  - Chamberlin, Jordan
AU  - Silatsa, Francis B. T.
AU  - Yemefack, Martin
AU  - Wendt, John
AU  - MacMillan, Robert A.
AU  - Wheeler, Ichsani
AU  - Crouch, Jonathan
PY  - 2021
UR  - https://www.nature.com/articles/s41598-021-85639-y
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/2362
AB  - Soil property and class maps for the continent of Africa were so far only available at very generalised scales, with many countries not mapped at all. Thanks to an increasing quantity and availability of soil samples collected at field point locations by various government and/or NGO funded projects, it is now possible to produce detailed pan-African maps of soil nutrients, including micro-nutrients at fine spatial resolutions. In this paper we describe production of a 30 m resolution Soil Information System of the African continent using, to date, the most comprehensive compilation of soil samples (N≈150,000) and Earth Observation data. We produced predictions for soil pH, organic carbon (C) and total nitrogen (N), total carbon, effective Cation Exchange Capacity (eCEC), extractable—phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), sulfur (S), sodium (Na), iron (Fe), zinc (Zn)—silt, clay and sand, stone content, bulk density and depth to bedrock, at three depths (0, 20 and 50 cm) and using 2-scale 3D Ensemble Machine Learning framework implemented in the mlr (Machine Learning in R) package. As covariate layers we used 250 m resolution (MODIS, PROBA-V and SM2RAIN products), and 30 m resolution (Sentinel-2, Landsat and DTM derivatives) images. Our fivefold spatial Cross-Validation results showed varying accuracy levels ranging from the best performing soil pH (CCC = 0.900) to more poorly predictable extractable phosphorus (CCC = 0.654) and sulphur (CCC = 0.708) and depth to bedrock. Sentinel-2 bands SWIR (B11, B12), NIR (B09, B8A), Landsat SWIR bands, and vertical depth derived from 30 m resolution DTM, were the overall most important 30 m resolution covariates. Climatic data images—SM2RAIN, bioclimatic variables and MODIS Land Surface Temperature—however, remained as the overall most important variables for predicting soil chemical variables at continental scale. This publicly available 30-m Soil Information System of Africa aims at supporting numerous applications, including soil and fertilizer policies and investments, agronomic advice to close yield gaps, environmental programs, or targeting of nutrition interventions.
T2  - Scientific Reports
T1  - African soil properties and nutrients mapped at 30 m spatial resolution using two‑scale ensemble machine learning
IS  - 11
SP  - 6130 (2021)
DO  - https://doi.org/10.1038/s41598-021-85639-y
ER  - 

@article{
author = "Hengl, Tomislav and Miller, Matthew A. E. and Križan, Josip and Shepherd, Keith D. and Sila, Andrew and Kilibarda, Milan and Antonijević, Ognjen and Glušica, Luka and Dobermann, Achim and Haefele, Stephan M. and McGrath, Steve P. and Acquah, Gifty E. and Collinson, Jamie and Parente, Leandro and Sheykhmousa, Mohammadreza and Saito, Kazuki and Johnson, Jean‑Martial and Chamberlin, Jordan and Silatsa, Francis B. T. and Yemefack, Martin and Wendt, John and MacMillan, Robert A. and Wheeler, Ichsani and Crouch, Jonathan",
year = "2021",
abstract = "Soil property and class maps for the continent of Africa were so far only available at very generalised scales, with many countries not mapped at all. Thanks to an increasing quantity and availability of soil samples collected at field point locations by various government and/or NGO funded projects, it is now possible to produce detailed pan-African maps of soil nutrients, including micro-nutrients at fine spatial resolutions. In this paper we describe production of a 30 m resolution Soil Information System of the African continent using, to date, the most comprehensive compilation of soil samples (N≈150,000) and Earth Observation data. We produced predictions for soil pH, organic carbon (C) and total nitrogen (N), total carbon, effective Cation Exchange Capacity (eCEC), extractable—phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), sulfur (S), sodium (Na), iron (Fe), zinc (Zn)—silt, clay and sand, stone content, bulk density and depth to bedrock, at three depths (0, 20 and 50 cm) and using 2-scale 3D Ensemble Machine Learning framework implemented in the mlr (Machine Learning in R) package. As covariate layers we used 250 m resolution (MODIS, PROBA-V and SM2RAIN products), and 30 m resolution (Sentinel-2, Landsat and DTM derivatives) images. Our fivefold spatial Cross-Validation results showed varying accuracy levels ranging from the best performing soil pH (CCC = 0.900) to more poorly predictable extractable phosphorus (CCC = 0.654) and sulphur (CCC = 0.708) and depth to bedrock. Sentinel-2 bands SWIR (B11, B12), NIR (B09, B8A), Landsat SWIR bands, and vertical depth derived from 30 m resolution DTM, were the overall most important 30 m resolution covariates. Climatic data images—SM2RAIN, bioclimatic variables and MODIS Land Surface Temperature—however, remained as the overall most important variables for predicting soil chemical variables at continental scale. This publicly available 30-m Soil Information System of Africa aims at supporting numerous applications, including soil and fertilizer policies and investments, agronomic advice to close yield gaps, environmental programs, or targeting of nutrition interventions.",
journal = "Scientific Reports",
title = "African soil properties and nutrients mapped at 30 m spatial resolution using two‑scale ensemble machine learning",
number = "11",
pages = "6130 (2021)",
doi = "https://doi.org/10.1038/s41598-021-85639-y"
}

Hengl, T., Miller, M. A. E., Križan, J., Shepherd, K. D., Sila, A., Kilibarda, M., Antonijević, O., Glušica, L., Dobermann, A., Haefele, S. M., McGrath, S. P., Acquah, G. E., Collinson, J., Parente, L., Sheykhmousa, M., Saito, K., Johnson, J., Chamberlin, J., Silatsa, F. B. T., Yemefack, M., Wendt, J., MacMillan, R. A., Wheeler, I.,& Crouch, J.. (2021). African soil properties and nutrients mapped at 30 m spatial resolution using two‑scale ensemble machine learning. in Scientific Reports(11), 6130 (2021).
https://doi.org/https://doi.org/10.1038/s41598-021-85639-y

Hengl T, Miller MAE, Križan J, Shepherd KD, Sila A, Kilibarda M, Antonijević O, Glušica L, Dobermann A, Haefele SM, McGrath SP, Acquah GE, Collinson J, Parente L, Sheykhmousa M, Saito K, Johnson J, Chamberlin J, Silatsa FBT, Yemefack M, Wendt J, MacMillan RA, Wheeler I, Crouch J. African soil properties and nutrients mapped at 30 m spatial resolution using two‑scale ensemble machine learning. in Scientific Reports. 2021;(11):6130 (2021).
doi:https://doi.org/10.1038/s41598-021-85639-y .

Hengl, Tomislav, Miller, Matthew A. E., Križan, Josip, Shepherd, Keith D., Sila, Andrew, Kilibarda, Milan, Antonijević, Ognjen, Glušica, Luka, Dobermann, Achim, Haefele, Stephan M., McGrath, Steve P., Acquah, Gifty E., Collinson, Jamie, Parente, Leandro, Sheykhmousa, Mohammadreza, Saito, Kazuki, Johnson, Jean‑Martial, Chamberlin, Jordan, Silatsa, Francis B. T., Yemefack, Martin, Wendt, John, MacMillan, Robert A., Wheeler, Ichsani, Crouch, Jonathan, "African soil properties and nutrients mapped at 30 m spatial resolution using two‑scale ensemble machine learning" in Scientific Reports, no. 11 (2021):6130 (2021),
https://doi.org/https://doi.org/10.1038/s41598-021-85639-y . .

TY  - JOUR
AU  - Pejović, Milutin
AU  - Nikolić, Mladen
AU  - Heuvelink, Gerard B. M.
AU  - Hengl, Tomislav
AU  - Kilibarda, Milan
AU  - Bajat, Branislav
PY  - 2018
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/943
AB  - An approach for using lasso (Least Absolute Shrinkage and Selection Operator) regression in creating sparse 3D models of soil properties for spatial prediction at multiple depths is presented. Modeling soil properties in 3D benefits from interactions of spatial predictors with soil depth and its polynomial expansion, which yields a large number of model variables (and corresponding model parameters). Lasso is able to perform variable selection, hence reducing the number of model parameters and making the model more easily interpretable. This also prevents overfitting, which makes the model more accurate. The presented approach was tested using four variable selection approaches - none, stepwise, lasso and hierarchical lasso, on four kinds of models - standard linear model, linear model with polynomial expansion of depth, linear model with interactions of covariates with depth and linear model with interactions of covariates with depth and its polynomial expansion. This framework was used to predict Soil Organic Carbon (SOC) in three contrasting study areas: Bor (Serbia), Edgeroi (Australia) and the Netherlands. Results show that lasso yields substantial improvements in accuracy over standard and stepwise regression - up to 50 % of total variance. It yields models which contain up to five times less nonzero parameters than the full models and that are usually more sparse than models obtained by stepwise regression, up to three times. Extension of the standard linear model by including interactions typically improves the accuracy of models produced by lasso, but is detrimental to standard and stepwise regression. Regarding computation time, it was demonstrated that lasso is several orders of magnitude more efficient than stepwise regression for models with tens or hundreds of variables (including interactions). Proper model evaluation is emphasized. Considering the fact that lasso requires meta-parameter tuning, standard cross-validation does not suffice for adequate model evaluation, hence a nested cross-validation was employed. The presented approach is implemented as publicly available sparsereg3D R package.
PB  - Elsevier Ltd
T2  - Computers & Geosciences
T1  - Sparse regression interaction models for spatial prediction of soil properties in 3D
EP  - 13
SP  - 1
VL  - 118
DO  - 10.1016/j.cageo.2018.05.008
ER  - 

@article{
author = "Pejović, Milutin and Nikolić, Mladen and Heuvelink, Gerard B. M. and Hengl, Tomislav and Kilibarda, Milan and Bajat, Branislav",
year = "2018",
abstract = "An approach for using lasso (Least Absolute Shrinkage and Selection Operator) regression in creating sparse 3D models of soil properties for spatial prediction at multiple depths is presented. Modeling soil properties in 3D benefits from interactions of spatial predictors with soil depth and its polynomial expansion, which yields a large number of model variables (and corresponding model parameters). Lasso is able to perform variable selection, hence reducing the number of model parameters and making the model more easily interpretable. This also prevents overfitting, which makes the model more accurate. The presented approach was tested using four variable selection approaches - none, stepwise, lasso and hierarchical lasso, on four kinds of models - standard linear model, linear model with polynomial expansion of depth, linear model with interactions of covariates with depth and linear model with interactions of covariates with depth and its polynomial expansion. This framework was used to predict Soil Organic Carbon (SOC) in three contrasting study areas: Bor (Serbia), Edgeroi (Australia) and the Netherlands. Results show that lasso yields substantial improvements in accuracy over standard and stepwise regression - up to 50 % of total variance. It yields models which contain up to five times less nonzero parameters than the full models and that are usually more sparse than models obtained by stepwise regression, up to three times. Extension of the standard linear model by including interactions typically improves the accuracy of models produced by lasso, but is detrimental to standard and stepwise regression. Regarding computation time, it was demonstrated that lasso is several orders of magnitude more efficient than stepwise regression for models with tens or hundreds of variables (including interactions). Proper model evaluation is emphasized. Considering the fact that lasso requires meta-parameter tuning, standard cross-validation does not suffice for adequate model evaluation, hence a nested cross-validation was employed. The presented approach is implemented as publicly available sparsereg3D R package.",
publisher = "Elsevier Ltd",
journal = "Computers & Geosciences",
title = "Sparse regression interaction models for spatial prediction of soil properties in 3D",
pages = "13-1",
volume = "118",
doi = "10.1016/j.cageo.2018.05.008"
}

Pejović, M., Nikolić, M., Heuvelink, G. B. M., Hengl, T., Kilibarda, M.,& Bajat, B.. (2018). Sparse regression interaction models for spatial prediction of soil properties in 3D. in Computers & Geosciences
Elsevier Ltd., 118, 1-13.
https://doi.org/10.1016/j.cageo.2018.05.008

Pejović M, Nikolić M, Heuvelink GBM, Hengl T, Kilibarda M, Bajat B. Sparse regression interaction models for spatial prediction of soil properties in 3D. in Computers & Geosciences. 2018;118:1-13.
doi:10.1016/j.cageo.2018.05.008 .

Pejović, Milutin, Nikolić, Mladen, Heuvelink, Gerard B. M., Hengl, Tomislav, Kilibarda, Milan, Bajat, Branislav, "Sparse regression interaction models for spatial prediction of soil properties in 3D" in Computers & Geosciences, 118 (2018):1-13,
https://doi.org/10.1016/j.cageo.2018.05.008 . .

TY  - JOUR
AU  - Hengl, Tomislav
AU  - de Jesus, Jorge Mendes
AU  - Heuvelink, Gerard B. M.
AU  - Gonzalez, Maria Ruiperez
AU  - Kilibarda, Milan
AU  - Blagotić, Aleksandar
AU  - Shangguan, Wei
AU  - Wright, Marvin N.
AU  - Geng, Xiaoyuan
AU  - Bauer-Marschallinger, Bernhard
AU  - Guevara, Mario Antonio
AU  - Vargas, Rodrigo
AU  - MacMillan, Robert A.
AU  - Batjes, Niels H.
AU  - Leenaars, Johan G. B.
AU  - Ribeiro, Eloi
AU  - Wheeler, Ichsani
AU  - Mantel, Stephan
AU  - Kempen, Bas
PY  - 2017
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/885
AB  - This paper describes the technical development and accuracy assessment of the most recent and improved version of the SoilGrids system at 250m resolution (June 2016 update). SoilGrids provides global predictions for standard numeric soil properties (organic carbon, bulk density, Cation Exchange Capacity (CEC), pH, soil texture fractions and coarse fragments) at seven standard depths (0, 5, 15, 30, 60, 100 and 200 cm), in addition to predictions of depth to bedrock and distribution of soil classes based on the World Reference Base (WRB) and USDA classification systems (ca. 280 raster layers in total). Predictions were based on ca. 150,000 soil profiles used for training and a stack of 158 remote sensing-based soil covariates (primarily derived from MODIS land products, SRTM DEM derivatives, climatic images and global landform and lithology maps), which were used to fit an ensemble of machine learning methods D random forest and gradient boosting and/or multinomial logistic regression D as implemented in the R packages ranger, xgboost, nnet and caret. The results of 10 -fold cross-validation show that the ensemble models explain between 56% (coarse fragments) and 83% (pH) of variation with an overall average of 61%. Improvements in the relative accuracy considering the amount of variation explained, in comparison to the previous version of SoilGrids at 1 km spatial resolution, range from 60 to 230%. Improvements can be attributed to: (1) the use of machine learning instead of linear regression, (2) to considerable investments in preparing finer resolution covariate layers and (3) to insertion of additional soil profiles. Further development of SoilGrids could include refinement of methods to incorporate input uncertainties and derivation of posterior probability distributions (per pixel), and further automation of spatial modeling so that soil maps can be generated for potentially hundreds of soil variables. Another area of future research is the development of methods for multiscale merging of SoilGrids predictions with local and/or national gridded soil products (e.g. up to 50 m spatial resolution) so that increasingly more accurate, complete and consistent global soil information can be produced. SoilGrids are available under the Open Data Base License.
PB  - Public Library of Science
T2  - PLOS One
T1  - SoilGrids250m: Global gridded soil information based on machine learning
IS  - 2
VL  - 12
DO  - 10.1371/journal.pone.0169748
ER  - 

@article{
author = "Hengl, Tomislav and de Jesus, Jorge Mendes and Heuvelink, Gerard B. M. and Gonzalez, Maria Ruiperez and Kilibarda, Milan and Blagotić, Aleksandar and Shangguan, Wei and Wright, Marvin N. and Geng, Xiaoyuan and Bauer-Marschallinger, Bernhard and Guevara, Mario Antonio and Vargas, Rodrigo and MacMillan, Robert A. and Batjes, Niels H. and Leenaars, Johan G. B. and Ribeiro, Eloi and Wheeler, Ichsani and Mantel, Stephan and Kempen, Bas",
year = "2017",
abstract = "This paper describes the technical development and accuracy assessment of the most recent and improved version of the SoilGrids system at 250m resolution (June 2016 update). SoilGrids provides global predictions for standard numeric soil properties (organic carbon, bulk density, Cation Exchange Capacity (CEC), pH, soil texture fractions and coarse fragments) at seven standard depths (0, 5, 15, 30, 60, 100 and 200 cm), in addition to predictions of depth to bedrock and distribution of soil classes based on the World Reference Base (WRB) and USDA classification systems (ca. 280 raster layers in total). Predictions were based on ca. 150,000 soil profiles used for training and a stack of 158 remote sensing-based soil covariates (primarily derived from MODIS land products, SRTM DEM derivatives, climatic images and global landform and lithology maps), which were used to fit an ensemble of machine learning methods D random forest and gradient boosting and/or multinomial logistic regression D as implemented in the R packages ranger, xgboost, nnet and caret. The results of 10 -fold cross-validation show that the ensemble models explain between 56% (coarse fragments) and 83% (pH) of variation with an overall average of 61%. Improvements in the relative accuracy considering the amount of variation explained, in comparison to the previous version of SoilGrids at 1 km spatial resolution, range from 60 to 230%. Improvements can be attributed to: (1) the use of machine learning instead of linear regression, (2) to considerable investments in preparing finer resolution covariate layers and (3) to insertion of additional soil profiles. Further development of SoilGrids could include refinement of methods to incorporate input uncertainties and derivation of posterior probability distributions (per pixel), and further automation of spatial modeling so that soil maps can be generated for potentially hundreds of soil variables. Another area of future research is the development of methods for multiscale merging of SoilGrids predictions with local and/or national gridded soil products (e.g. up to 50 m spatial resolution) so that increasingly more accurate, complete and consistent global soil information can be produced. SoilGrids are available under the Open Data Base License.",
publisher = "Public Library of Science",
journal = "PLOS One",
title = "SoilGrids250m: Global gridded soil information based on machine learning",
number = "2",
volume = "12",
doi = "10.1371/journal.pone.0169748"
}

Hengl, T., de Jesus, J. M., Heuvelink, G. B. M., Gonzalez, M. R., 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.,& Kempen, B.. (2017). SoilGrids250m: Global gridded soil information based on machine learning. in PLOS One
Public Library of Science., 12(2).
https://doi.org/10.1371/journal.pone.0169748

Hengl T, de Jesus JM, Heuvelink GBM, Gonzalez MR, Kilibarda M, Blagotić A, Shangguan W, Wright MN, Geng X, Bauer-Marschallinger B, Guevara MA, Vargas R, MacMillan RA, Batjes NH, Leenaars JGB, Ribeiro E, Wheeler I, Mantel S, Kempen B. SoilGrids250m: Global gridded soil information based on machine learning. in PLOS One. 2017;12(2).
doi:10.1371/journal.pone.0169748 .

Hengl, Tomislav, de Jesus, Jorge Mendes, Heuvelink, Gerard B. M., Gonzalez, Maria Ruiperez, Kilibarda, Milan, Blagotić, Aleksandar, Shangguan, Wei, Wright, Marvin N., Geng, Xiaoyuan, Bauer-Marschallinger, Bernhard, Guevara, Mario Antonio, Vargas, Rodrigo, MacMillan, Robert A., Batjes, Niels H., Leenaars, Johan G. B., Ribeiro, Eloi, Wheeler, Ichsani, Mantel, Stephan, Kempen, Bas, "SoilGrids250m: Global gridded soil information based on machine learning" in PLOS One, 12, no. 2 (2017),
https://doi.org/10.1371/journal.pone.0169748 . .

TY  - JOUR
AU  - Kilibarda, Milan
AU  - Tadić-Percec, Melita
AU  - Hengl, Tomislav
AU  - Luković, Jelena
AU  - Bajat, Branislav
PY  - 2015
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/695
AB  - This article highlights the results of an assessment of representation and usability of global temperature station data for global spatio-temporal analysis. Datasets from the Global Surface Summary of Day (GSOD) and the European Climate Assessment & Dataset (ECA&D) were merged and consisted of 10,695 global stations for the year 2011. Three aspects of data quality were considered: (a) representation in the geographical domain, (b) representation in the feature space (based on the MaxEnt method), and
PB  - Elsevier
T2  - Spatial Statistics
T1  - Global geographic and feature space coverage of temperature data in the context of spatio-temporal interpolation
EP  - 38
SP  - 22
VL  - 14
DO  - 10.1016/j.spasta.2015.04.005
ER  - 

@article{
author = "Kilibarda, Milan and Tadić-Percec, Melita and Hengl, Tomislav and Luković, Jelena and Bajat, Branislav",
year = "2015",
abstract = "This article highlights the results of an assessment of representation and usability of global temperature station data for global spatio-temporal analysis. Datasets from the Global Surface Summary of Day (GSOD) and the European Climate Assessment & Dataset (ECA&D) were merged and consisted of 10,695 global stations for the year 2011. Three aspects of data quality were considered: (a) representation in the geographical domain, (b) representation in the feature space (based on the MaxEnt method), and",
publisher = "Elsevier",
journal = "Spatial Statistics",
title = "Global geographic and feature space coverage of temperature data in the context of spatio-temporal interpolation",
pages = "38-22",
volume = "14",
doi = "10.1016/j.spasta.2015.04.005"
}

Kilibarda, M., Tadić-Percec, M., Hengl, T., Luković, J.,& Bajat, B.. (2015). Global geographic and feature space coverage of temperature data in the context of spatio-temporal interpolation. in Spatial Statistics
Elsevier., 14, 22-38.
https://doi.org/10.1016/j.spasta.2015.04.005

Kilibarda M, Tadić-Percec M, Hengl T, Luković J, Bajat B. Global geographic and feature space coverage of temperature data in the context of spatio-temporal interpolation. in Spatial Statistics. 2015;14:22-38.
doi:10.1016/j.spasta.2015.04.005 .

Kilibarda, Milan, Tadić-Percec, Melita, Hengl, Tomislav, Luković, Jelena, Bajat, Branislav, "Global geographic and feature space coverage of temperature data in the context of spatio-temporal interpolation" in Spatial Statistics, 14 (2015):22-38,
https://doi.org/10.1016/j.spasta.2015.04.005 . .

TY  - CONF
AU  - Kilibarda, Milan
AU  - Bajat, Branislav
AU  - Hengl, Tomislav
AU  - Pejović, Milutin
PY  - 2014
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/1130
PB  - Department of Geodesy and Geoinformation, Vienna University of Technology, Vienna
C3  - Extended Abstract Proceedings of the GIScience 2014, 23-26 September, Vienna, Austria
T1  - Meteo: package for automated meteorological spatiotemporal mapping
EP  - 327
SP  - 323
UR  - https://hdl.handle.net/21.15107/rcub_grafar_1130
ER  - 

@conference{
author = "Kilibarda, Milan and Bajat, Branislav and Hengl, Tomislav and Pejović, Milutin",
year = "2014",
publisher = "Department of Geodesy and Geoinformation, Vienna University of Technology, Vienna",
journal = "Extended Abstract Proceedings of the GIScience 2014, 23-26 September, Vienna, Austria",
title = "Meteo: package for automated meteorological spatiotemporal mapping",
pages = "327-323",
url = "https://hdl.handle.net/21.15107/rcub_grafar_1130"
}

Kilibarda, M., Bajat, B., Hengl, T.,& Pejović, M.. (2014). Meteo: package for automated meteorological spatiotemporal mapping. in Extended Abstract Proceedings of the GIScience 2014, 23-26 September, Vienna, Austria
Department of Geodesy and Geoinformation, Vienna University of Technology, Vienna., 323-327.
https://hdl.handle.net/21.15107/rcub_grafar_1130

Kilibarda M, Bajat B, Hengl T, Pejović M. Meteo: package for automated meteorological spatiotemporal mapping. in Extended Abstract Proceedings of the GIScience 2014, 23-26 September, Vienna, Austria. 2014;:323-327.
https://hdl.handle.net/21.15107/rcub_grafar_1130 .

Kilibarda, Milan, Bajat, Branislav, Hengl, Tomislav, Pejović, Milutin, "Meteo: package for automated meteorological spatiotemporal mapping" in Extended Abstract Proceedings of the GIScience 2014, 23-26 September, Vienna, Austria (2014):323-327,
https://hdl.handle.net/21.15107/rcub_grafar_1130 .

TY  - JOUR
AU  - Kilibarda, Milan
AU  - Hengl, Tomislav
AU  - Heuvelink, Gerard B. M.
AU  - Graeler, Benedikt
AU  - Pebesma, Edzer
AU  - Tadić-Percec, Melita
AU  - Bajat, Branislav
PY  - 2014
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/639
AB  - Combined Global Surface Summary of Day and European Climate Assessment and Dataset daily meteorological data sets (around 9000 stations) were used to build spatio-temporal geostatistical models and predict daily air temperature at ground resolution of 1km for the global land mass. Predictions in space and time were made for the mean, maximum, and minimum temperatures using spatio-temporal regression-kriging with a time series of Moderate Resolution Imaging Spectroradiometer (MODIS) 8 day images, topographic layers (digital elevation model and topographic wetness index), and a geometric temperature trend as covariates. The accuracy of predicting daily temperatures was assessed using leave-one-out cross validation. To account for geographical point clustering of station data and get a more representative cross-validation accuracy, predicted values were aggregated to blocks of land of size 500x500km. Results show that the average accuracy for predicting mean, maximum, and minimum daily temperatures is root-mean-square error (RMSE) =2 degrees C for areas densely covered with stations and between 2 degrees C and 4 degrees C for areas with lower station density. The lowest prediction accuracy was observed at high altitudes (>1000m) and in Antarctica with an RMSE around 6 degrees C. The model and predictions were built for the year 2011 only, but the same methodology could be extended for the whole range of the MODIS land surface temperature images (2001 to today), i.e., to produce global archives of daily temperatures (a next-generation repository) and to feed various global environmental models. Key Points  Global spatio-temporal regression-kriging daily temperature interpolation   Fitting of global spatio-temporal models for the mean, maximum, and minimum temperatures   Time series of MODIS 8 day images as explanatory variables in regression part
PB  - Wiley-Blackwell
T2  - Journal of Geophysical Research-Atmospheres
T1  - Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution
EP  - 2313
IS  - 5
SP  - 2294
VL  - 119
DO  - 10.1002/2013JD020803
ER  - 

@article{
author = "Kilibarda, Milan and Hengl, Tomislav and Heuvelink, Gerard B. M. and Graeler, Benedikt and Pebesma, Edzer and Tadić-Percec, Melita and Bajat, Branislav",
year = "2014",
abstract = "Combined Global Surface Summary of Day and European Climate Assessment and Dataset daily meteorological data sets (around 9000 stations) were used to build spatio-temporal geostatistical models and predict daily air temperature at ground resolution of 1km for the global land mass. Predictions in space and time were made for the mean, maximum, and minimum temperatures using spatio-temporal regression-kriging with a time series of Moderate Resolution Imaging Spectroradiometer (MODIS) 8 day images, topographic layers (digital elevation model and topographic wetness index), and a geometric temperature trend as covariates. The accuracy of predicting daily temperatures was assessed using leave-one-out cross validation. To account for geographical point clustering of station data and get a more representative cross-validation accuracy, predicted values were aggregated to blocks of land of size 500x500km. Results show that the average accuracy for predicting mean, maximum, and minimum daily temperatures is root-mean-square error (RMSE) =2 degrees C for areas densely covered with stations and between 2 degrees C and 4 degrees C for areas with lower station density. The lowest prediction accuracy was observed at high altitudes (>1000m) and in Antarctica with an RMSE around 6 degrees C. The model and predictions were built for the year 2011 only, but the same methodology could be extended for the whole range of the MODIS land surface temperature images (2001 to today), i.e., to produce global archives of daily temperatures (a next-generation repository) and to feed various global environmental models. Key Points  Global spatio-temporal regression-kriging daily temperature interpolation   Fitting of global spatio-temporal models for the mean, maximum, and minimum temperatures   Time series of MODIS 8 day images as explanatory variables in regression part",
publisher = "Wiley-Blackwell",
journal = "Journal of Geophysical Research-Atmospheres",
title = "Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution",
pages = "2313-2294",
number = "5",
volume = "119",
doi = "10.1002/2013JD020803"
}

Kilibarda, M., Hengl, T., Heuvelink, G. B. M., Graeler, B., Pebesma, E., Tadić-Percec, M.,& Bajat, B.. (2014). Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution. in Journal of Geophysical Research-Atmospheres
Wiley-Blackwell., 119(5), 2294-2313.
https://doi.org/10.1002/2013JD020803

Kilibarda M, Hengl T, Heuvelink GBM, Graeler B, Pebesma E, Tadić-Percec M, Bajat B. Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution. in Journal of Geophysical Research-Atmospheres. 2014;119(5):2294-2313.
doi:10.1002/2013JD020803 .

Kilibarda, Milan, Hengl, Tomislav, Heuvelink, Gerard B. M., Graeler, Benedikt, Pebesma, Edzer, Tadić-Percec, Melita, Bajat, Branislav, "Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution" in Journal of Geophysical Research-Atmospheres, 119, no. 5 (2014):2294-2313,
https://doi.org/10.1002/2013JD020803 . .

TY  - JOUR
AU  - Bajat, Branislav
AU  - Hengl, Tomislav
AU  - Kilibarda, Milan
AU  - Krunić, Nikola
PY  - 2011
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/416
AB  - Niche analysis methods developed within the biogeography community are routinely used for species distribution modeling of wildlife and endangered species. So far, such techniques have not been used to explain distribution of people in an area, nor to assess spatio-temporal dynamics of human populations. In this paper, the MaxEnt approach to species distribution modeling and publicly available gridded predictors were used to analyze the population dynamics in Southern Serbia (South Pomoravlje Region) for the period 1961-2027. Population values from the census administrative units were first downscaled to 200 m grid using a detailed map of populated places and dasymetric interpolation. In the second step, a point pattern representing the whole population (468,500 inhabitants in 2002) was simulated using the R package spatstat. MaxEnt was then used to derive habitat suitability index (HSI) as a function of gridded predictors: distance to roads, elevation, slope, topographic wetness index, enhanced vegetation index and land cover classes. HSI and environmental predictors were further used to explain spatial patterns in the population change index (PCI) through regression modeling. The results show that inhabiting preference for year 1961 is mainly a function of topography (TWI, elevation). The HSI for year 2027 shows that large portions of remote areas are becoming less preferred for inhabiting. The results of cross-validation in MaxEnt show that distribution of population is distinctly controlled by environmental factors (AUC  gt 0.84). Population decrease is particularly significant in areas >25 km distant from the main road network. The results of regression analysis show that 40% of variability in the PCI values can be explained with these environmental maps, distance to roads and urban areas being the main drivers of migration process. This approach allows precise mapping of demographic patterns that otherwise would not be visible from the census data alone.
T2  - Computers Environment and Urban Systems
T1  - Mapping population change index in Southern Serbia (1961-2027) as a function of environmental factors
EP  - 44
IS  - 1
SP  - 35
VL  - 35
DO  - 10.1016/j.compenvurbsys.2010.09.005
ER  - 

@article{
author = "Bajat, Branislav and Hengl, Tomislav and Kilibarda, Milan and Krunić, Nikola",
year = "2011",
abstract = "Niche analysis methods developed within the biogeography community are routinely used for species distribution modeling of wildlife and endangered species. So far, such techniques have not been used to explain distribution of people in an area, nor to assess spatio-temporal dynamics of human populations. In this paper, the MaxEnt approach to species distribution modeling and publicly available gridded predictors were used to analyze the population dynamics in Southern Serbia (South Pomoravlje Region) for the period 1961-2027. Population values from the census administrative units were first downscaled to 200 m grid using a detailed map of populated places and dasymetric interpolation. In the second step, a point pattern representing the whole population (468,500 inhabitants in 2002) was simulated using the R package spatstat. MaxEnt was then used to derive habitat suitability index (HSI) as a function of gridded predictors: distance to roads, elevation, slope, topographic wetness index, enhanced vegetation index and land cover classes. HSI and environmental predictors were further used to explain spatial patterns in the population change index (PCI) through regression modeling. The results show that inhabiting preference for year 1961 is mainly a function of topography (TWI, elevation). The HSI for year 2027 shows that large portions of remote areas are becoming less preferred for inhabiting. The results of cross-validation in MaxEnt show that distribution of population is distinctly controlled by environmental factors (AUC  gt 0.84). Population decrease is particularly significant in areas >25 km distant from the main road network. The results of regression analysis show that 40% of variability in the PCI values can be explained with these environmental maps, distance to roads and urban areas being the main drivers of migration process. This approach allows precise mapping of demographic patterns that otherwise would not be visible from the census data alone.",
journal = "Computers Environment and Urban Systems",
title = "Mapping population change index in Southern Serbia (1961-2027) as a function of environmental factors",
pages = "44-35",
number = "1",
volume = "35",
doi = "10.1016/j.compenvurbsys.2010.09.005"
}

Bajat, B., Hengl, T., Kilibarda, M.,& Krunić, N.. (2011). Mapping population change index in Southern Serbia (1961-2027) as a function of environmental factors. in Computers Environment and Urban Systems, 35(1), 35-44.
https://doi.org/10.1016/j.compenvurbsys.2010.09.005

Bajat B, Hengl T, Kilibarda M, Krunić N. Mapping population change index in Southern Serbia (1961-2027) as a function of environmental factors. in Computers Environment and Urban Systems. 2011;35(1):35-44.
doi:10.1016/j.compenvurbsys.2010.09.005 .

Bajat, Branislav, Hengl, Tomislav, Kilibarda, Milan, Krunić, Nikola, "Mapping population change index in Southern Serbia (1961-2027) as a function of environmental factors" in Computers Environment and Urban Systems, 35, no. 1 (2011):35-44,
https://doi.org/10.1016/j.compenvurbsys.2010.09.005 . .

TY  - JOUR
AU  - Hengl, Tomislav
AU  - Bajat, Branislav
AU  - Blagojević, Dragan
AU  - Reuter, Hannes I.
PY  - 2008
UR  - https://grafar.grf.bg.ac.rs/handle/123456789/201
AB  - This paper recommends computational procedures for employing auxiliary maps, such as maps of drainage patterns, land cover and remote-sensing-based indices, directly in the geostatistical modeling of topography. The methodology is based on the regression-kriging technique, as implemented in the R package gstat. The computational procedures are illustrated using a case study in the south-west part of Serbia. Two point data sets were used for geostatistical modeling: (I) 2051 elevation points were used to generate DEMs and (2) an independent error assessment data set (1020 points) was used to assess errors in the topo-DEM and the SRTM-DEM. Four auxiliary maps were used to improve generation of DEMs from point data: (1) distance to streams, (2) terrain complexity measured by standard deviation filter, (3) analytical hillshading map and (4) NDVI map derived from the Landsat image. The auxiliary predictors were significantly correlated with elevations (adj.R-2 = 0.20) and DEM errors (adj.R-2 = 0.27). By including auxiliary maps in the geostatistical modeling of topography, realizations of DEMs can be generated that represent geomorphology of a terrain more accurately. In addition, downscaling of a coarse 3 arcsec SRTM DEM using auxiliary maps and regression-kriging is demonstrated using the same case study. A methodological advantage of regression-kriging, compared to splines, is the possibility to automate the data processing and incorporate multiple auxiliary predictors. The remaining open issues are computational efficiency, application of local regression-kriging algorithms and preparation of suitable auxiliary data layers for such analyses.
T2  - Computers & Geosciences
T1  - Geostatistical modeling of topography using auxiliary maps
EP  - 1899
IS  - 12
SP  - 1886
VL  - 34
DO  - 10.1016/j.cageo.2008.01.005
ER  - 

@article{
author = "Hengl, Tomislav and Bajat, Branislav and Blagojević, Dragan and Reuter, Hannes I.",
year = "2008",
abstract = "This paper recommends computational procedures for employing auxiliary maps, such as maps of drainage patterns, land cover and remote-sensing-based indices, directly in the geostatistical modeling of topography. The methodology is based on the regression-kriging technique, as implemented in the R package gstat. The computational procedures are illustrated using a case study in the south-west part of Serbia. Two point data sets were used for geostatistical modeling: (I) 2051 elevation points were used to generate DEMs and (2) an independent error assessment data set (1020 points) was used to assess errors in the topo-DEM and the SRTM-DEM. Four auxiliary maps were used to improve generation of DEMs from point data: (1) distance to streams, (2) terrain complexity measured by standard deviation filter, (3) analytical hillshading map and (4) NDVI map derived from the Landsat image. The auxiliary predictors were significantly correlated with elevations (adj.R-2 = 0.20) and DEM errors (adj.R-2 = 0.27). By including auxiliary maps in the geostatistical modeling of topography, realizations of DEMs can be generated that represent geomorphology of a terrain more accurately. In addition, downscaling of a coarse 3 arcsec SRTM DEM using auxiliary maps and regression-kriging is demonstrated using the same case study. A methodological advantage of regression-kriging, compared to splines, is the possibility to automate the data processing and incorporate multiple auxiliary predictors. The remaining open issues are computational efficiency, application of local regression-kriging algorithms and preparation of suitable auxiliary data layers for such analyses.",
journal = "Computers & Geosciences",
title = "Geostatistical modeling of topography using auxiliary maps",
pages = "1899-1886",
number = "12",
volume = "34",
doi = "10.1016/j.cageo.2008.01.005"
}

Hengl, T., Bajat, B., Blagojević, D.,& Reuter, H. I.. (2008). Geostatistical modeling of topography using auxiliary maps. in Computers & Geosciences, 34(12), 1886-1899.
https://doi.org/10.1016/j.cageo.2008.01.005

Hengl T, Bajat B, Blagojević D, Reuter HI. Geostatistical modeling of topography using auxiliary maps. in Computers & Geosciences. 2008;34(12):1886-1899.
doi:10.1016/j.cageo.2008.01.005 .

Hengl, Tomislav, Bajat, Branislav, Blagojević, Dragan, Reuter, Hannes I., "Geostatistical modeling of topography using auxiliary maps" in Computers & Geosciences, 34, no. 12 (2008):1886-1899,
https://doi.org/10.1016/j.cageo.2008.01.005 . .