首页|Predicting Environmental Covariates of Soil Organic Matter at Sub-Regional Scale for Sustainable Agricultural Development in Southeast Nigeria
Predicting Environmental Covariates of Soil Organic Matter at Sub-Regional Scale for Sustainable Agricultural Development in Southeast Nigeria
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Hard
Soil organic matter is an important indicator of soil health. It is a constituent of the ecological system that is vital to agricultural development and understanding of the global carbon cycle. The study used random forest regression, a machine learning algorithm, to identify relevant predictors of soil organic matter through the integration of field and Sentinel-2 derived vegetation indices and a selected reanalysis of climate data with topography. Three landcover types were purposefully delineated, and 72 soil samples were collected at a soil depth of 20 cm across the entire Cross River State, Nigeria. The samples were labeled and taken to the laboratory, where standard procedures were used in extracting the SOM. 80% of the point data sets were used in model calibration, while 20% were used to validate the model. Model analysis revealed that environmental covariates of SOM (topography, rainfall, maximum air temperature, OSAVI, EVI, and NDVI) produced high prediction accuracy with lower uncertainty. The maximum plot SOM was estimated to be 7.20% with overall mean values of 2.61. The test data sets yielded a model accuracy of 0.85, an RMSE of 36.7, a relRMSE of 34.3%, and a bias of 3.7 t/ha. Based on this, the paper argues that the identified environmental covariates can be optimized for the effective management of SOM for sustained agricultural development. This is pertinent in areas with highly weathered soils characterized by low nutrients and poor crop yields. The SOM map of this study can be used as a baseline for subsequent monitoring and management of SOM in the study area.
agricultural developmentenvironmental covariatesCross River StateSOMrandom forestCARBON SEQUESTRATIONFOREST
NETL
NSTL
Hard
