Article
Environmental Sciences
Qikai Lu, Shuang Tian, Lifei Wei
Summary: Soil pH and carbonates are important indicators of soil chemistry and fertility. This research mapped their spatial distribution in Europe using multi-source environmental variables and machine learning approaches. The results show the importance of MODIS products and climatic variables in predicting soil pH and CaCO3.
SCIENCE OF THE TOTAL ENVIRONMENT
(2023)
Article
Soil Science
Chenconghai Yang, Lin Yang, Lei Zhang, Chenghu Zhou
Summary: This study compares the mapping of soil organic matter (SOM) using the INLA-SPDE model with RS-based soil moisture indices and FTD variables in different vegetation-covered areas of Anhui Province, China. The results show that the optimal combination of RS-based soil moisture indices and FTD variables improves the mapping accuracy of SOM, and the INLA-SPDE model has a higher prediction accuracy than the Random Forest model.
Article
Environmental Sciences
Zhiyuan Tian, Feng Liu, Yin Liang, Xuchao Zhu
Summary: The study mapped soil erodibility (K factor) using a random forest (RF) model and environmental variables, identifying relief, climate, land surface temperature, and vegetation indexes as important predictors. By optimizing the model and generating a digital map, the spatial resolution of the K factor was significantly improved, demonstrating the effectiveness of the mapping approach.
INTERNATIONAL SOIL AND WATER CONSERVATION RESEARCH
(2022)
Article
Remote Sensing
Nai-Qing Fan, Fang-He Zhao, Liang-Jun Zhu, Cheng-Zhi Qin, A-Xing Zhu
Summary: The effective use of environmental covariates is crucial for digital soil mapping. Traditional approaches ignore the varying applicability of covariates in characterizing soil-environment relationships. This study proposes an adaptive method that quantifies covariate applicability based on terrain conditions and integrates it into the iPSM method, outperforming other methods in predicting soil organic matter content.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Soil Science
Xingchen Fan, Naiqing Fan, Cheng-Zhi Qin, Fang-He Zhao, Liang-Jun Zhu, A-Xing Zhu
Summary: Digital soil mapping based on environmental similarity is a useful method for predicting soil properties at unvisited locations. However, the current method lacks consideration of the First Law of Geography and spatial variation of sample distribution. In this paper, a new large-area DSM method is proposed, which considers the spatial distance to samples by using adaptive distance-decay parameter values. Evaluation results show that this method achieves higher accuracy compared to other environmental-similarity-based DSM methods and a large-area DSM method not based on environmental similarity.
Article
Biodiversity Conservation
Wei-chun Zhang, He-shuang Wan, Ming-hou Zhou, Wei Wu, Hong -bin Liu
Summary: This study used machine learning models to predict the spatial distribution of soil organic carbon (SOC) and soil total carbon content (STC), and quantified the contribution of environmental factors to the variability of SOC and STC. The results showed that the Random Forest plus residuals Kriging (RFRK) model performed best in prediction and uncertainty estimation, while the eXtreme Gradient Boosting (XGBoost) model performed well in uncertainty estimation. Land use types, mean annual Normalized Difference Vegetation Index, and elevation were identified as the top three important indicators in determining the spatial variability of SOC and STC.
ECOLOGICAL INDICATORS
(2022)
Article
Soil Science
Christopher Blackford, Brandon Heung, Kara L. Webster
Summary: Digital soil mapping combines soil plot data with environmental datasets to model variation in soil properties across a landscape. Optimizing site selection and sampling intensity can improve the accuracy of digital soil maps and reduce field costs. Generating preliminary uncertainty maps can guide subsequent field soil sample collections to significantly improve model performance. This study demonstrates the effectiveness of uncertainty maps in guiding sampling by simulating a multiyear soil sampling campaign and testing different uncertainty metrics and sampling intensity levels.
Article
Geosciences, Multidisciplinary
Mojtaba Zeraatpisheh, Gillian L. Galford, Alissa White, Adam Noel, Heather Darby, Carol Adair
Summary: This study evaluated the impact of spatial resolution of environmental variables on the prediction of soil organic carbon stocks. The random forest algorithm outperformed other algorithms in estimating soil organic carbon stocks at different spatial resolutions. Furthermore, soil maps and geology/landform maps had a significant influence on the predictive results. The results of this study highlight the importance of data sources, model type, and the combination of environmental variables in predicting soil organic carbon stocks.
Article
Environmental Sciences
Yiming Xu, Bin Li, Junhong Bai, Guangliang Zhang, Xin Wang, Scot E. Smith, Shudong Du
Summary: This research examined the effects of environmental variables with seasonal variations on digital soil mapping (DSM) in coastal wetlands. Machine learning methods were used to establish multiple prediction models of soil organic carbon (SOC) based on multi-temporal data. The results showed significant variations in the relationships between SOC and environmental variables in different months. The environmental variables in the wet season had stronger correlations and higher importance scores with SOC compared to the dry season. Furthermore, the prediction models in the wet season and April had stronger performance than those in the dry season. This study highlights the importance of considering seasonal variations in the establishment of a DSM model in coastal wetlands.
LAND DEGRADATION & DEVELOPMENT
(2022)
Article
Geosciences, Multidisciplinary
Sabrina C. Y. Ip, Alfrendo Satyanaga, Harianto Rahardjo
Summary: Soil shear strength is a critical parameter in slope stability and its spatial variation is complex. The study shows that Random Forest model has higher accuracy and spatial heterogeneity in predicting soil shear strength, and it is more sensitive to sample size.
Article
Soil Science
Ren-Min Yang, Li-An Liu, Xin Zhang, Ri-Xing He, Chang-Ming Zhu, Zhong-Qi Zhang, Jian-Guo Li
Summary: The study found that incorporating temporal variables into soil organic carbon prediction models can improve their accuracy. Utilizing temporal changes in environmental factors can better explain the spatiotemporal variability of soil organic carbon, which is crucial for future predictions of soil organic carbon changes.
Article
Environmental Studies
Roomesh Kumar Jena, Pravash Chandra Moharana, Subramanian Dharumarajan, Gulshan Kumar Sharma, Prasenjit Ray, Partha Deb Roy, Dibakar Ghosh, Bachaspati Das, Amnah Mohammed Alsuhaibani, Ahmed Gaber, Akbar Hossain
Summary: This study used a random forest model to predict soil particle-size fractions in the Ri-Bhoi district of Meghalaya state, India. The results show that the prediction accuracy of sand, silt, and clay varies at different depth ranges. Channel network base level and LS-Factor were found to be the most crucial variables for sand and silt prediction, while Min Temperature of Coldest Month (BIO6) was discovered for clay prediction. Additional research is needed to explore novel methodologies for extensive digital soil mapping.
Article
Environmental Sciences
Azadeh Katebikord, Seyed Hamidreza Sadeghi, Vijay P. Singh
Summary: The relationship between soil organic carbon (SOC) and environmental parameters was investigated in the Galazchai Watershed, Iran. The study analyzed the correlation between SOC amounts and remote sensing indices, topographic variables, and soil texture. The results showed that none of the combinations of variables accurately estimated SOC, but specific remote sensing indices and geographically weighted regression methods performed better. Future studies should consider more uniform and denser sampling and explore alternative methods to investigate the relationship between variables.
ENVIRONMENTAL MONITORING AND ASSESSMENT
(2022)
Article
Soil Science
Mojtaba Zeraatpisheh, Shamsollah Ayoubi, Zahra Mirbagheri, Mohammad Reza Mosaddeghi, Ming Xu
Summary: This study spatially predicted soil aggregate stability indices and SOC in various aggregate sizes across the landscape using digital soil mapping and machine learning models. The random forest model performed best for MWD, GMD, and WSA, while kNN and SVM models showed the best prediction for SOC in different aggregate fractions. The ensemble model increased prediction accuracies for all soil targets, highlighting the importance of machine learning-based models for land use planning and decision making.
Article
Geosciences, Multidisciplinary
Gerard B. M. Heuvelink, Richard Webster
Summary: Pedologists traditionally mapped soil by drawing boundaries, but the introduction of geostatistics and ordinary kriging in the 1980s revolutionized soil mapping. Machine learning techniques have also been adopted, but they lack transparency and spatial correlation considerations. Spatial statisticians and pedometricians have important roles in incorporating uncertainty and communicating it to end users.
SPATIAL STATISTICS
(2022)