摘要
Prediction and accurate digital soil mapping (DSM) of soil organic carbon (SOC) at a local scale is a key factor for any agro-ecological modelling. This study aims to use remote sensing and terrain derivatives to provide a reliable method for SOC prediction. An advanced geostatistical-based empirical Bayesian Kriging regression (EBKR) method was used and performance was compared with the artificial neural network (ANN) and hybrid ANN, i.e. ANN-OK (ordinary kriging) and ANN-CK (cokriging). The result showed that the hybrid ANN model performs better than ANN, whereas the EBKR method outperforms all other methods with the highestR(2)of 0.936. The DSM map shows that the highest SOC concentration was found in easternmost part of the study area with grass and agricultural land. This work shows the robustness of the EBKR prediction method over other techniques. The study will also aid the policymakers in adopting sustainable land use management.
学科与关键词
- A Comparative Assessment of Geostatistical, Machine Learning, and Hybrid Approaches for Mapping Topsoil Organic Carbon ContentChen, Lin; Ren, Chunying*; Li, Lin; Wang, Yeqiao; Zhang, Bai; Wang, Zongming; Li, LinfengISPRS International Journal of Geo-Information, 2019, 8(4): 174.被引 56
- Digital soil mapping algorithms and covariates for soil organic carbon mapping and their implications: A reviewLamichhane, Sushil*; Kumar, Lalit; Wilson, BrianGeoderma, 2019, 352: 395-413.被引 265
- Prediction of Soil Organic Carbon based on Landsat 8 Monthly NDVI Data for the Jianghan Plain in Hubei Province, ChinaZhang, Yangchengsi; Guo, Long*; Chen, Yiyun; Shi, Tiezhu; Luo, Mei; Ju, QingLan; Zhang, Haitao; Wang, ShanqinRemote Sensing, 2019, 11(14): 1683.被引 81
