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.

Digital soil mapping (DSM) involves acquisition of field soil observations and matching them with environmental variables that can explain the distribution of soils. The harmonization of these data sets, through computer-based methods, are increasingly being found to be as reliable as traditional soil mapping practices, but without the prohibitive costs. Therefore, the present research developed decision tree models for spatial prediction of soil classes in a 720 km² area located in an arid region of central Iran, where traditional soil survey methods are difficult to undertake. Using the conditioned Latin hypercube sampling method, the locations of 187 soil profiles were selected, which were then described, sampled, analyzed, and allocated to six Great Groups according to the USDA Soil Taxonomy system. Auxiliary data representing the soil forming factors were derived from a digital elevation model (DEM), Landsat 7 ETM⁺ images, and a map of geomorphology. The accuracy of the decision tree models was evaluated using overall, user, and producer accuracy based on an independent validation data set. Our results showed some auxiliary variables had more influence on the prediction of soil classes which included: topographic wetness index, geomorphological map, multiresolution index of valley bottom flatness, elevation, and principal components of Landsat 7 ETM⁺ images. Furthermore, the results have confirmed the DSM model successfully predicted Great Groups with overall accuracy up to 67.5%. Our results suggest that the developed methodology could be used to predict soil classes in the arid region of Iran.

This study aimed to evaluate the power of various vegetation indices for sugarcane yield modelling in Shoeibeyeh area in Khuzestan province of Iran. Seven indices were extracted from satellite images and were then converted to seven days' time-series via interpolation. To eliminate noise from the time-series data, all of them were reconstructed using the Savitzky-Golay algorithm. Thus seven different time-series of vegetation indices were obtained. The growth profile was drawn via averaging of NDVI time-series data and was divided into three growth intervals. Then the accumulative values of vegetation indices related to first and second periods of growth (from 2004 to 2016 extracted from time-series data) were evaluated by simple linear regression models against the average observed yields efficiency. The result showed the accumulative IAVI (γ = 1.4) vegetation index relative to first period of growth with R² = 0.66 and RMSE = 3.78 ton/ha and the accumulative NDI vegetation index relative to second period of growth with R² = 0.66 and RMSE = 3.79 ton/ha and the accumulative NDI vegetation index relative to sum of the first and the second growth periods with R² = 0.78 and RMSE = 3.09 ton/ha had good agreement with sugarcane stem yield efficiency at the middle of growth and before harvesting season.

— The influence of biological and physicochemical soil properties on the variations in soil organic and inorganic carbon (OC and IC) contents at the soil surface was studied. Three-dimensional...

Soil organic carbon (SOC) is a key indicator of soil health, stability, and security, and a major component of global carbon sequestration. Accurate SOC modeling is vital for environmental management and climate change mitigation. Despite machine learning advances, spatial autocorrelation remains a challenge in SOC prediction. This study assesses its impact by comparing four Random Forest (RF) models at 0–30 cm soil depth: (1) classic RF with non-spatial variables and random cross-validation, (2) classic RF with spatial and non-spatial variables and random cross-validation, (3) classic RF with non-spatial variables and spatial cross-validation, and (4) spatial RF with non-spatial variables and random cross-validation. Using 281 soil samples collected over six years in Abyek, Iran, the study integrates spatial and non-spatial predictors, including topographic, climatic, and vegetation indices. Scenario 4 yields the highest predictive accuracy (R2 = 0.86, RMSE = 0.11) and eliminates spatial autocorrelation in residuals (Moran’s I = -0.01, p = 0.82). In contrast, models lacking spatial components (e.g., Scenario 1) show residual clustering and biased predictions (Moran’s I > 0.15, p < 0.001). These results highlight the importance of incorporating spatial dependencies and proper validation to enhance SOC prediction. The study informs digital soil mapping, carbon stock assessment, and land management for sustainable agriculture.