Predictive mapping of soil organic carbon in wet cultivated lands using classification-tree based models: the case study of Denmark

Estimating divergent forest carbon stocks and sinks via a knife set approach

Forest carbon stocks and sinks (CSSs) have been widely estimated using climate classification tables and linear regression (LR) models with common independent variables (IVs) such as the average diameter at breast height (DBH) of stems and root shoot ratio. However, this approach is relatively ineffective when the explanatory power of IVs is lower than that of unobservable variables. Various environmental and anthropogenic factors affect target variables that cause the correlation between them to be chaotic. Here, we designed a knife set (KS) approach combining LR models and the wandering through random forests (WTF) algorithm and applied it in a specific case of Phyllostachys edulis (Carrière) J. Houz. (P. edulis) forests, which have an irregular relationship between their belowground carbon (BGC) stocks and average DBH. We then validated the KS approach performed by cluster computing to estimate the aboveground carbon (AGC) and BGC stocks and the total net primary production (TNPP). The estimated CSSs were compared to the benchmark of the methodology that applied Tier 1 in the Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories via 10-fold cross validation, and the KS approach significantly increased precision and accuracy of estimations. Our approach provides general insights to accurately estimate forest CSSs relying on evidence-based field data, even if some target variables are divergent in specific forest types. We also pointed out the reason why current fancy models containing machine learning (ML) or deep learning algorithms are not effective in predicting the target variables of certain chaotic systems is perhaps that the total explanatory power of observable variables is less than that of the total unobservable variables. Quantifying unobservable variables into observable variables is a linchpin of future works related to chaotic system estimation.

National wetland mapping using remote-sensing-derived environmental variables, archive field data, and artificial intelligence

While wetland ecosystem services are widely recognized, the lack of fine-scale national inventories prevents successful implementation of conservation policies. Wetlands are difficult to map due to their complex fine-grained spatial pattern and fuzzy boundaries. However, the increasing amount of open high-spatial-resolution remote sensing data and accurately georeferenced field data archives, as well as progress in artificial intelligence (AI), provide opportunities for fine-scale national wetland mapping. The objective of this study was to map wetlands over mainland France (ca. 550,000 km²) by applying AI to environmental variables derived from remote sensing and archive field data. A random forest model was calibrated using spatial cross-validation according to the precision-recall area under the curve (PR-AUC) index using ca. 135,000 soil or flora plots from archive databases, as well as 5 m topographical variables derived from an airborne DTM and a geological map. The model was validated using an experimentally designed sampling strategy with ca. 3000 plots collected during a ground survey in 2021 along non-wetland/wetland transects. Map accuracy was then compared to those of nine existing wetland maps with global, European, or national coverage. The model-derived suitability map (PR-AUC 0.76) highlights the gradual boundaries and fine-grained pattern of wetlands. The binary map is significantly more accurate (F1-score 0.75, overall accuracy 0.67) than existing wetland maps. The approach and end-results are of important value for spatial planning and environmental management since the high-resolution suitability and binary maps enable more targeted conservation measures to support biodiversity conservation, water resources maintenance, and carbon storage.

Mapping Soil Organic Carbon in Low-Relief Farmlands Based on Stratified Heterogeneous Relationship

Accurate mapping of farmland soil organic carbon (SOC) provides valuable information for evaluating soil quality and guiding agricultural management. The integration of natural factors, agricultural activities, and landscape patterns may well fit the high spatial variation of SOC in low-relief farmlands. However, commonly used prediction methods are global models, ignoring the stratified heterogeneous relationship between SOC and environmental variables and failing to reveal the determinants of SOC in different subregions. Using 242 topsoil samples collected from Jianghan Plain, China, this study explored the stratified heterogeneous relationship between SOC and natural factors, agricultural activities, and landscape metrics, determined the dominant factors of SOC in each stratum, and predicted the spatial distribution of SOC using the Cubist model. Ordinary kriging, stepwise linear regression (SLR), and random forest (RF) were used as references. SLR and RF results showed that land use types, multiple cropping index, straw return, and percentage of water bodies are global dominant factors of SOC. Cubist results exhibited that the dominant factors of SOC vary in different cropping systems. Compared with the SOC of paddy fields, the SOC of irrigated land was more affected by irrigation-related factors. The effect of straw return on SOC was diverse under different cropping intensities. The Cubist model outperformed the other models in explaining SOC variation and SOC mapping (fitting R2 = 0.370 and predicted R2 = 0.474). These results highlight the importance of exploring the stratified heterogeneous relationship between SOC and covariates, and this knowledge provides a scientific basis for farmland zoning management. The Cubist model, integrating natural factors, agricultural activities, and landscape metrics, is effective in explaining SOC variation and mapping SOC in low-relief farmlands.

Modelling soil organic carbon using vegetation indices across large catchments in eastern Australia

Soil organic carbon (SOC) is an important soil component. However, examining SOC at the large catchment scale is difficult due to the intensive labour requirements. This study examines SOC distribution at large (>500 km²) catchment scales using field-sampled SOC data and remote sensed vegetation indices located in eastern Australia (Krui River catchment - 562 km²; Merriwa River catchment - 808 km²) on grazing land-use basalt soil. The SOC data obtained was compared to digital elevation model (DEM) derived elevation and insolation data, as well as Normalised Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI) values corresponding to each sample site. These indices were obtained from the MODIS sensor (Terra/Aqua) and Landsat series satellites. Vegetation Indices (VI) captured immediately prior to sampling demonstrated a poor correlation with SOC. The use of multiple, aggregated, prior VI data sets provided a good match with SOC. The strongest match occurred for Landsat 8 EVI, indicating that VIs with higher spatial and spectral resolution, which can account for atmospheric interference, have the potential to produce more accurate SOC mapping (Krui samples in 2006, R² = 0.31, P < 0.01; Krui sampled in 2014, R² = 0.41, P < 0.01; Merriwa samples in 2015, R² = 0.37, P < 0.01). A sensitivity test for both remote sensing platforms demonstrated that the findings were robust. The results demonstrate that VIs are a reliable surrogate for historical vegetation growth in pasture dominated landscapes and therefore soil carbon inputs allowing for mapping of SOC across large catchment scales. Both Landsat and MODIS produced similar results and demonstrate that SOC can be reliably predicted at the large catchment scale and for different catchments in this environment with RMSE range of 0.79 to 1.06. The method and data can be applied globally and provides a new method for environmental assessment.

Mapping Regional Soil Organic Matter Based on Sentinel-2A and MODIS Imagery Using Machine Learning Algorithms and Google Earth Engine

Many studies have attempted to predict soil organic matter (SOM), whereas mapping high-precision and high-resolution SOM maps remains a challenge due to the difficulty of selecting appropriate satellite data sources and prediction algorithms. This study aimed to investigate the influence of different remotely sensed images and machine learning algorithms on SOM prediction. We constructed two comparative experiments, i.e., full-band and common-band variable datasets of Sentinel-2A and MODIS images using Google Earth Engine (GEE). The predictive performances of random forest (RF), artificial neural network (ANN), and support vector regression (SVR) algorithms were evaluated, and the SOM map was generated for the Songnen Plain. Results showed that the model based on the full-band Sentinel-2A dataset achieved the best performance. The application of Sentinel-2A data resulted in mean relative improvements (RIs) of 7.67% and 5.87%, respectively. The RF achieved a lower root mean squared error (RMSE = 0.68%) and a higher coefficient of determination (R2 = 0.67) in all of the predicted scenarios than ANN and SVR. The resultant SOM map accurately characterized the SOM spatial distribution. Therefore, the Sentinel-2A data have obvious advantages over MODIS due to their higher spectral and spatial resolutions, and the combination of the RF algorithm and GEE is an effective approach to SOM mapping.

Using Machine Learning Algorithms to Estimate Soil Organic Carbon Variability with Environmental Variables and Soil Nutrient Indicators in an Alluvial Soil

Soil organic carbon (SOC) is an important indicator of soil quality and directly determines soil fertility. Hence, understanding its spatial distribution and controlling factors is necessary for efficient and sustainable soil nutrient management. In this study, machine learning algorithms including artificial neural network (ANN), support vector machine (SVM), cubist regression, random forests (RF), and multiple linear regression (MLR) were chosen for advancing the prediction of SOC. A total of sixty (n = 60) soil samples were collected within the research area at 30 cm soil depth and measured for SOC content using the Walkley–Black method. From these samples, 80% were used for model training and 21 auxiliary data were included as predictors. The predictors include effective cation exchange capacity (ECEC), base saturation (BS), calcium to magnesium ratio (Ca_Mg), potassium to magnesium ratio (K_Mg), potassium to calcium ratio (K_Ca), elevation, plan curvature, total catchment area, channel network base level, topographic wetness index, clay index, iron index, normalized difference build-up index (NDBI), ratio vegetation index (RVI), soil adjusted vegetation index (SAVI), normalized difference vegetation index (NDVI), normalized difference moisture index (NDMI) and land surface temperature (LST). Mean absolute error (MAE), root-mean-square error (RMSE) and R2 were used to determine the model performance. The result showed the mean SOC to be 1.62% with a coefficient of variation (CV) of 47%. The best performing model was RF (R2 = 0.68) followed by the cubist model (R2 = 0.51), SVM (R2 = 0.36), ANN (R2 = 0.36) and MLR (R2 = 0.17). The soil nutrient indicators, topographic wetness index and total catchment area were considered an indicator for spatial prediction of SOC in flat homogenous topography. Future studies should include other auxiliary predictors (e.g., soil physical and chemical properties, and lithological data) as well as cover a broader range of soil types to improve model performance.

High-resolution digital mapping of soil organic carbon and soil total nitrogen using DEM derivatives, Sentinel-1 and Sentinel-2 data based on machine learning algorithms

Soil organic carbon (SOC) and soil total nitrogen (STN) are important indicators of soil health and play a key role in the global carbon and nitrogen cycles. High-resolution radar Sentinel-1 and multispectral Sentinel-2 images have the potential to investigate soil spatial distribution information over a large area, although Sentinel-1 and Sentinel-2 data have rarely been combined to map either SOC or STN content. In this study, we applied machine learning techniques to map both SOC and STN content in the southern part of Central Europe using digital elevation model (DEM) derivatives, multi-temporal Sentinel-1 and Sentinel-2 data, and evaluated the potential of different remote sensing sensors (Sentinel-1 and Sentinel-2) to predict SOC and STN content. Four machine-learners including random forest (RF), boosted regression trees (BRT), support vector machine (SVM) and Bagged CART were used to construct predictive models of SOC and STN contents based on 179 soil samples and different combinations of environmental covariates. The performance of these models was evaluated based on a 10-fold cross-validation method by three statistical indicators. Overall, the BRT model performed better than RF, SVM and Bagged CART, and these models yielded similar spatial distribution patterns of SOC and STN. Our results showed that multi-source sensor methods provided more accurate predictions of SOC and STN contents than individual sensors. The application of radar Sentinel-1 and multispectral Sentinel-2 images proved useful for predicting SOC and STN. A combination of Sentinel-1/2-derived predictors and DEM derivatives yielded the highest prediction accuracy. The prediction accuracy changed with and without the Sentinel-1/2-derived predictors, with the R² for estimating both SOC and STN content using the BRT model increasing by 12.8% and 18.8%, respectively. Topographic variables were the main explanatory variables for SOC and STN predictions, where elevation was assigned as the variable with the most importance by the models. The results of this study illustrate the potential of free high-resolution radar Sentinel-1 and multispectral Sentinel-2 data as input when developing SOC and STN prediction models.

The Spatial Prediction of Soil Texture Fractions in Arid Regions of Iran

To predict the soil texture fractions, 115 profiles were identified based on the Latin hypercube sampling technique, the horizons were sampled, and the clay, sand, and silt contents (in percentages) of soil samples were measured. Then equal-area quadratic spline depth functions were used to derive clay, sand, and silt contents at five standard soil depths (0–5, 5–15, 15–30, 30–60, and 60–100 cm). Auxiliary variables used in this study include the terrain attributes (derived from a digital elevation model), Landsat 8 image data (acquired in 2015), geomorphological map, and spectrometric data (laboratory data). Artificial neural network (ANN), regression tree (RT), and neuro-fuzzy (ANFIS) models were used to make a correlation between soil data (clay, sand, and silt) and auxiliary variables. The results of this study showed that the ANFIS model was more accurate in the prediction of the three parameters of clay, silt, and sand than ANN and RT. Moreover, the ability of ANFIS model to estimate the soil texture fractions in the surface layers was higher than the lower layers. The mean coefficient of determination (R2) values calculated by 10-fold cross validation suggested the higher prediction performance in the upper depth intervals and higher prediction error in the lower depth intervals (e.g., R2 = 0.91, concordance correlation coefficient (CCC) = 0.90, RMSE = 4.00 g kg−1 for sand of 0–5 cm depth, and R2 = 0.68, CCC = 0.60, RMSE = 8.03 g kg−1 for 60–100 cm depth). The results also showed that the most important auxiliary variables are spectrometric data, multi-resolution, valley-bottom flatness index and wetness index. Overall, it is recommended to use ANFIS models for the digital mapping of soil texture fractions in other arid regions of Iran.

Prediction of Soil Organic Carbon based on Landsat 8 Monthly NDVI Data for the Jianghan Plain in Hubei Province, China

High-precision maps of soil organic carbon (SOC) are beneficial for managing soil fertility and understanding the global carbon cycle. Digital soil mapping plays an important role in efficiently obtaining the spatial distribution of SOC, which contributes to precision agriculture. However, traditional soil-forming factors (i.e., terrain or climatic factors) have weak variability in low-relief areas, such as plains, and cannot reflect the spatial variation of soil attributes. Meanwhile, vegetation cover hinders the acquisition of the direct information of farmland soil. Thus, useful environmental variables should be utilized for SOC prediction and the digital mapping of such areas. SOC has an important effect on crop growth status, and remote sensing data can record the apparent spectral characteristics of crops. The normalized difference vegetation index (NDVI) is an important index reflecting crop growth and biomass. This study used NDVI time series data rather than traditional soil-forming factors to map SOC. Honghu City, located in the middle of the Jianghan Plain, was selected as the study region, and the NDVI time series data extracted from Landsat 8 were used as the auxiliary variables. SOC maps were estimated through stepwise linear regression (SLR), partial least squares regression (PLSR), support vector machine (SVM), and artificial neural network (ANN). Ordinary kriging (OK) was used as the reference model, while root mean square error of prediction (RMSEP) and coefficient of determination of prediction (R2P) were used to evaluate the model performance. Results showed that SOC had a significant positive correlation in July and August (0.17, 0.29) and a significant negative correlation in January, April, and December (−0.23, −0.27, and −0.23) with NDVI time series data. The best model for SOC prediction was generated by ANN, with the lowest RMSEP of 3.718 and highest R2P of 0.391, followed by SVM (RMSEP = 3.753, R2P = 0.361) and PLSR (RMSEP = 4.087, R2P = 0.283). The SLR model was the worst model, with the lowest R2P of 0.281 and highest RMSEP of 3.930. ANN and SVM were better than OK (RMSEP = 3.727, R2P = 0.372), whereas PLSR and SLR were worse than OK. Moreover, the prediction results using single-data NDVI or short time series NDVI showed low accuracy. The effect of the terrain factor on SOC prediction represented unsatisfactory results. All these results indicated that the NDVI time series data can be used for SOC mapping in plain areas and that the ANN model can maximally extract additional associated information between NDVI time series data and SOC. This study presented an effective method to overcome the selection of auxiliary variables for digital soil mapping in plain areas when the soil was covered with vegetation. This finding indicated that the time series characteristics of NDVI were conducive for predicting SOC in plains.

Identifying localized and scale-specific multivariate controls of soil organic matter variations using multiple wavelet coherence

Environmental factors have shown localized and scale-dependent controls over soil organic matter (SOM) distribution in the landscape. Previous studies have explored the relationships between SOM and individual controlling factors; however, few studies have indicated the combined control from multiple environmental factors. In this study, we compared the localized and scale-dependent univariate and multivariate controls of SOM along two long transects (northeast, NE transect and north, N transect) from China. Bivariate wavelet coherence (BWC) between SOM and individual factors and multiple wavelet coherence (MWC) between SOM and factor combinations were calculated. Average wavelet coherence (AWC) and percent area of significant coherence (PASC) were used to assess the relative dominance of individual and a combination of factors to explain SOM variations at different scales and locations. The results showed that (in BWC analysis) mean annual temperature (MAT) with the largest AWC (0.39) and PASC (16.23%) was the dominant factor in explaining SOM variations along the NE transect. The topographic wetness index (TWI) was the dominant factor (AWC = 0.39 and PASC = 20.80%) along the N transect. MWC identified the combination of Slope, net primary production (NPP) and mean annual precipitation (MAP) as the most important combination in explaining SOM variations along the NE transect with a significant increase in AWC and PASC at different scales and locations (e.g. AWC = 0.91 and PASC = 58.03% at all scales). The combination of TWI, NPP and normalized difference vegetation index (NDVI) was the most influential along the N transect (AWC = 0.83 and PASC = 32.68% at all scales). The results indicated that the combined controls of environmental factors on SOM variations at different scales and locations in a large area can be identified by MWC. This is promising for a better understanding of the multivariate controls in SOM variations at larger spatial scales and may improve the capability of digital soil mapping.

Spatio-temporal topsoil organic carbon mapping of a semi-arid Mediterranean region: The role of land use, soil texture, topographic indices and the influence of remote sensing data to modelling

SOC is the most important indicator of soil fertility and monitoring its space-time changes is a prerequisite to establish strategies to reduce soil loss and preserve its quality. Here we modelled the topsoil (0-0.3m) SOC concentration of the cultivated area of Sicily in 1993 and 2008. Sicily is an extremely variable region with a high number of ecosystems, soils, and microclimates. We studied the role of time and land use in the modelling of SOC, and assessed the role of remote sensing (RS) covariates in the boosted regression trees modelling. The models obtained showed a high pseudo-R² (0.63-0.69) and low uncertainty (s.d.<0.76gCkg^-1 with RS, and <1.25gCkg^-1 without RS). These outputs allowed depicting a time variation of SOC at 1arcsec. SOC estimation strongly depended on the soil texture, land use, rainfall and topographic indices related to erosion and deposition. RS indices captured one fifth of the total variance explained, slightly changed the ranking of variance explained by the non-RS predictors, and reduced the variability of the model replicates. During the study period, SOC decreased in the areas with relatively high initial SOC, and increased in the area with high temperature and low rainfall, dominated by arables. This was likely due to the compulsory application of some Good Agricultural and Environmental practices. These results confirm that the importance of texture and land use in short-term SOC variation is comparable to climate. The present results call for agronomic and policy intervention at the district level to maintain fertility and yield potential. In addition, the present results suggest that the application of RS covariates enhanced the modelling performance.

Spatial 3D distribution of soil organic carbon under different land use types

Soil organic carbon (SOC) has been assessed in three dimension (3D) in several studies, but little is known about the combined effects of land use and soil depth on SOC stocks in semi-arid areas. This paper investigates the 3D distribution of SOC to a depth of 1 m in a 4600-ha area in southeastern Iran with different land uses under the irrigated farming (IF), dry farming (DF), orchards (Or), range plants on the Gachsaran formation (RaG), and range plants on a quaternary formation (RaQ). Predictions were made using the artificial neural networks (ANNs), regression trees (RTs), and spline functions with auxiliary covariates derived from a digital elevation model (DEM), the Landsat 8 imagery, and land use types. Correlation analysis showed that the main predictors for SOC in the topsoil were covariates derived from the imagery; however, for the lower depths, covariates derived from both the DEM and imagery were important. ANNs showed more efficiency than did RTs in predicting SOC. The results showed that 3D distribution of SOC was significantly affected by land use types. SOC stocks of soils under Or and IF were significantly higher than those under DF, RaG, and RaQ. The SOC below 30 cm accounted for about 59% of the total soil stock. Results showed that depth functions combined with digital soil mapping techniques provide a promising approach to evaluate 3D SOC distribution under different land uses in semi-arid regions and could be used to assess changes in time to determine appropriate management strategies.

Ultrahigh Dimensional Variable Selection for Interpolation of Point Referenced Spatial Data: A Digital Soil Mapping Case Study

Modern soil mapping is characterised by the need to interpolate point referenced (geostatistical) observations and the availability of large numbers of environmental characteristics for consideration as covariates to aid this interpolation. Modelling tasks of this nature also occur in other fields such as biogeography and environmental science. This analysis employs the Least Angle Regression (LAR) algorithm for fitting Least Absolute Shrinkage and Selection Operator (LASSO) penalized Multiple Linear Regressions models. This analysis demonstrates the efficiency of the LAR algorithm at selecting covariates to aid the interpolation of geostatistical soil carbon observations. Where an exhaustive search of the models that could be constructed from 800 potential covariate terms and 60 observations would be prohibitively demanding, LASSO variable selection is accomplished with trivial computational investment.

The effectiveness of digital soil mapping to predict soil properties over low-relief areas

This study investigates the ability of different digital soil mapping (DSM) approaches to predict some of physical and chemical topsoil properties in the Shahrekord plain of Chaharmahal-Va-Bakhtiari province, Iran. According to a semi-detailed soil survey, 120 soil samples were collected from 0 to 30 cm depth with approximate distance of 750 m. Particle size distribution, coarse fragments (CFs), electrical conductivity (EC), pH, organic carbon (OC), and calcium carbonate equivalent (CCE) were determined. Four machine learning techniques, namely, artificial neural networks (ANNs), boosted regression tree (BRT), generalized linear model (GLM), and multiple linear regression (MLR), were used to identify the relationship between soil properties and auxiliary information (terrain attributes, remote sensing indices, geology map, existing soil map, and geomorphology map). Root-mean-square error (RMSE) and mean error (ME) were considered to determine the performance of the models. Among the studied models, GLM showed the highest performance to predict pH, EC, clay, silt, sand, and CCE, whereas the best model is not necessarily able to make accurate estimation. According to RMSE%, DSM has a good efficiency to predict soil properties with low and moderate variabilities. Terrain attributes were the main predictors among different studied auxiliary information. The accuracy of the estimations with more observations is recommended to give a better understanding about the performance of DSM approach over low-relief areas.

Predictive Mapping of Topsoil Organic Carbon in an Alpine Environment Aided by Landsat TM

The objective of this study was to examine the reflectance of Landsat TM imagery for mapping soil organic Carbon (SOC) content in an Alpine environment. The studied area (ca. 3*10⁴ km²) is the upper reaches of the Heihe River at the northeast edge of the Tibetan plateau, China. A set (105) of topsoil samples were analyzed for SOC. Boosted regression tree (BRT) models using Landsat TM imagery were built to predict SOC content, alone or with topography and climate covariates (temperature and precipitation). The best model, combining all covariates, was only marginally better than using only imagery. Imagery alone was sufficient to build a reasonable model; this was a bit better than only using topography and climate covariates. The Lin’s concordance correlation coefficient values of the imagery only model and the full model are very close, larger than the topography and climate variables based model. In the full model, SOC was mainly explained by Landsat TM imagery (65% relative importance), followed by climate variables (20%) and topography (15% of relative importance). The good results from imagery are likely due to (1) the strong dependence of SOC on native vegetation intensity in this Alpine environment; (2) the strong correlation in this environment between imagery and environmental covariables, especially elevation (corresponding to temperature), precipitation, and slope aspect. We conclude that multispectral satellite data from Landsat TM images may be used to predict topsoil SOC with reasonable accuracy in Alpine regions, and perhaps other regions covered with natural vegetation, and that adding topography and climate covariables to the satellite data can improve the predictive accuracy.

Digital mapping of soil organic carbon contents and stocks in Denmark

Estimation of carbon contents and stocks are important for carbon sequestration, greenhouse gas emissions and national carbon balance inventories. For Denmark, we modeled the vertical distribution of soil organic carbon (SOC) and bulk density, and mapped its spatial distribution at five standard soil depth intervals (0-5, 5-15, 15-30, 30-60 and 60-100 cm) using 18 environmental variables as predictors. SOC distribution was influenced by precipitation, land use, soil type, wetland, elevation, wetness index, and multi-resolution index of valley bottom flatness. The highest average SOC content of 20 g kg(-1) was reported for 0-5 cm soil, whereas there was on average 2.2 g SOC kg(-1) at 60-100 cm depth. For SOC and bulk density prediction precision decreased with soil depth, and a standard error of 2.8 g kg(-1) was found at 60-100 cm soil depth. Average SOC stock for 0-30 cm was 72 t ha(-1) and in the top 1 m there was 120 t SOC ha(-1). In total, the soils stored approximately 570 Tg C within the top 1 m. The soils under agriculture had the highest amount of carbon (444 Tg) followed by forest and semi-natural vegetation that contributed 11% of the total SOC stock. More than 60% of the total SOC stock was present in Podzols and Luvisols. Compared to previous estimates, our approach is more reliable as we adopted a robust quantification technique and mapped the spatial distribution of SOC stock and prediction uncertainty. The estimation was validated using common statistical indices and the data and high-resolution maps could be used for future soil carbon assessment and inventories.