Digital mapping of soil organic carbon using ensemble learning model in Mollisols of Hyrcanian forests, northern Iran

Tajik, Samaneh; Ayoubi, Shamsollah; Zeraatpisheh, Mojtaba

doi:10.1016/j.geodrs.2020.e00256

Cited by 81 publications

(23 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We emphasize that the stacking strategy is more accurate than any of its individual models if the individual models are accurate and diverse. The success of the stacking method would generally be linked to two facts: (1) the training data does not always provide enough information to select a single accurate model, and (2) the learning processes of the individual model may be imperfect [40,68]. Those are the reasons why stacking never did worse than selecting the individual models in our case study.…”

Section: Performances Of ML Models In Two Different Climatic Regionsmentioning

confidence: 91%

“…Nevertheless, stacking often performs better than all individual models, especially when combined with rescanning the original covariate space [39]. For instance, Tajik et al [40], Zhou et al [41], and Chen et al [42] recently evaluated the efficacy of the ensemble models-by averaging the model predictions-to predict the spatial variation of soil properties in Iran, China, and France, respectively.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Improving the Spatial Prediction of Soil Organic Carbon Content in Two Contrasting Climatic Regions by Stacking Machine Learning Models and Rescanning Covariate Space

et al. 2020

Self Cite

View full text Add to dashboard Cite

Understanding the spatial distribution of soil organic carbon (SOC) content over different climatic regions will enhance our knowledge of carbon gains and losses due to climatic change. However, little is known about the SOC content in the contrasting arid and sub-humid regions of Iran, whose complex SOC–landscape relationships pose a challenge to spatial analysis. Machine learning (ML) models with a digital soil mapping framework can solve such complex relationships. Current research focusses on ensemble ML models to increase the accuracy of prediction. The usual ensemble method is boosting or weighted averaging. This study proposes a novel ensemble technique: the stacking of multiple ML models through a meta-learning model. In addition, we tested the ensemble through rescanning the covariate space to maximize the prediction accuracy. We first applied six state-of-the-art ML models (i.e., Cubist, random forests (RF), extreme gradient boosting (XGBoost), classical artificial neural network models (ANN), neural network ensemble based on model averaging (AvNNet), and deep learning neural networks (DNN)) to predict and map the spatial distribution of SOC content at six soil depth intervals for both regions. In addition, the stacking of multiple ML models through a meta-learning model with/without rescanning the covariate space were tested and applied to maximize the prediction accuracy. Out of six ML models, the DNN resulted in the best modeling accuracies, followed by RF, XGBoost, AvNNet, ANN, and Cubist. Importantly, the stacking of models indicated a significant improvement in the prediction of SOC content, especially when combined with rescanning the covariate space. For instance, the RMSE values for SOC content prediction of the upper 0–5 cm of the soil profiles of the arid site and the sub-humid site by the proposed stacking approaches were 17% and 9% respectively, less than that obtained by the DNN models—the best individual model. This indicates that rescanning the original covariate space by a meta-learning model can extract more information and improve the SOC content prediction accuracy. Overall, our results suggest that the stacking of diverse sets of models could be used to more accurately estimate the spatial distribution of SOC content in different climatic regions.

show abstract

Section: Performances Of ML Models In Two Different Climatic Regionsmentioning

confidence: 91%

Section: Introductionmentioning

confidence: 99%

Improving the Spatial Prediction of Soil Organic Carbon Content in Two Contrasting Climatic Regions by Stacking Machine Learning Models and Rescanning Covariate Space

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Meanwhile, topographic factors affect the transposition and redistribution of soil nutrient contents [13], and their effects on soil nutrients vary at different scales [14,15]. A number of prediction methods for soil properties based on their relationships with topographic factors are proposed [16], including the ensemble learning model [17], extended models [18], regression model [19], and so on. Nevertheless, these methods do not consider their multiscale relationships between soil nutrients and the covariates.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Soil Nutrients Based on Topographic Factors and Remote Sensing Index in a Coal Mining Area, China

et al. 2020

View full text Add to dashboard Cite

1) Background: Coal mining operations caused severe land subsidence and altered the distributions of soil nutrients that influenced by multiple environmental factors at different scales. However, the prediction performances for soil nutrients based on their scale-specific relationships with influencing factors remains undefined in the coal mining area. The objective of this study was to establish prediction models of soil nutrients based on their scale-specific relationships with influencing factors in a coal mining area. (2) Methods: Soil samples were collected based on a 1 × 1 km regular grid, and contents of soil organic matter, soil available nitrogen, soil available phosphorus, and soil available potassium were measured. The scale components of soil nutrients and the influencing factors collected from remote sensing and topographic factors were decomposed by two-dimensional empirical mode decomposition (2D-EMD), and the predictions for soil nutrients were established using the methods of multiple linear stepwise regression or partial least squares regression based on original samples (MLSR Ori or PLSR Ori ), partial least squares regression based on bi-dimensional intrinsic mode function (PLSR BIMF ), and the combined method of 2D-EMD, PLSR, and MLSR (2D-EMD PM ).(3) Results: The correlation types and correlation coefficients between soil nutrients and influencing factors were scale-dependent. The variances of soil nutrients at smaller scale were stochastic and non-significantly correlated with influencing factors, while their variances at the larger scales were stable. The prediction performances in the coal mining area were better than those in the non-coal mining area, and 2D-EMD PM had the most stable performance. (4) Conclusions: The scale-dependent predictions can be used for soil nutrients in the coal mining areas.

show abstract

“…The Digital Soil Mapping (DSM) is an easy and feasible approach to improve the understanding of soil attributes and properties. The DSM has been applied to predict soil classes (Silva et al, 2016;Triantafilis et al, 2009;, morphology (Demattê, 2016;, parent materials , and attributes, such as clay (Loiseau et al, 2019), SOM , pH , among others , contributing to the improvement of management practices Tajik et al, 2020). The DSM basis was formalised in the scorpan model by and it takes into account the model of soil formation established by .…”

Section: Introductionmentioning

confidence: 99%

“…The knowledge generated from the stochastic, deterministic, and hybrid models has uncovered quantitative patterns of soil attributes and formation factors. These models comprise the Bayesian Regularised Neural Network (Poggio et al, 2016;Tien Bui et al, 2012), Generalised Linear Model (McKenzie andAustin, 1993;Tajik et al, 2020), Random Forest (Castro-Franco et al, 2018, Cubist (Bonfatti et al, 2016;, Regression Kriging , and among other untested algorithms in soil science. Machine learning algorithms classify instances of unknown identity using samples of known targets (Cracknell, 2007).…”

Section: Introductionmentioning

confidence: 99%

Geotechnologies applied in digital soil mapping

Mendes¹

View full text Add to dashboard Cite

Geotechnologies applied in digital soil mappingThe civilisation lives in a world of maps and soil maps are vital at regional and farm levels to achieve best management agricultural practices. Soil is the substrate for plant growth and vital to the fulfilment of the food demand. However, the cartographic scale of those soil maps, which for the best management agricultural practice (BMAP) have to be the most detailed as possible and they are scarce. The Digital Soil Mapping (DSM) became the easiest and feasible approach to achieve such demand. Despite previous studies have tried to better characterise soil depths, there is space for improvements on its dynamics and mapping. Looking at this goal, Remote Sensing (RS) technologies have proven to be a great power on this task. Nevertheless, some aspects of that approach still need to be tested using another hybrid, stochastic, and deterministic models for the predictions of magnetic susceptibility (MS) and soil attributes at surface and subsurface. Therefore, chapter 1 presents the evaluation of nine machine learning algorithms (MLAs) to predict the free iron content at the soil surface (e.g. 0 -20 cm) using the DSM framework. Based on the best performance of those nine MLAs, we selected five MLAs. Chapter 2 shows the use of those five MLAs with usual and new environmental variables (e.g. DEM, drainage network, and soil spectroscopy) to predict the MS and soil attributes up to 100 cm depth. Attempts on quantifying soil mineral consist of having an observation measured using traditional laboratory soil analysis. However, developments in interpreting and analysing the visible and near-infrared (VNIR) diffuse reflectance have allowed quantifying some soil minerals. In chapter 3, it implements a novel framework using VNIR spectroscopy to quantify the main soil minerals and evaluates the application of digital soil mapping framework to spatialise those soil minerals. Last but not least, the chapter 4 presents the novelty of using all predict soil components as predictors of the soil mapping units in the region of Piracicaba-SP at farm scale (1:20,000), generating the first detailed digital soil map of the region. Additionally in this chapter, it was created the digital yield environmental map for sugarcane production. Thus, this thesis presents a new integrative framework to achieve detail soil maps for the BMAP and serves as a guide for future soil surveys across the world.

show abstract

Digital mapping of soil organic carbon using ensemble learning model in Mollisols of Hyrcanian forests, northern Iran

Cited by 81 publications

References 42 publications

Improving the Spatial Prediction of Soil Organic Carbon Content in Two Contrasting Climatic Regions by Stacking Machine Learning Models and Rescanning Covariate Space

Improving the Spatial Prediction of Soil Organic Carbon Content in Two Contrasting Climatic Regions by Stacking Machine Learning Models and Rescanning Covariate Space

Prediction of Soil Nutrients Based on Topographic Factors and Remote Sensing Index in a Coal Mining Area, China

Geotechnologies applied in digital soil mapping

Contact Info

Product

Resources

About