Digital mapping of cultivated land soil organic matter in hill-mountain and plain regions
JOURNAL OF SOILS AND SEDIMENTS(2023)
Abstract
Purpose Spatial prediction of soil organic matter (SOM) in cultivated land is crucial for evaluating soil productivity and its role in terrestrial carbon cycling. Cultivated soils in mountainous regions are commonly scattered on the footslope whereas those in the plain regions are continuously planar distributed; hence, they are quite different in the degree of variation in soil-forming factors and thereby the soil properties including SOM. Materials and methods In this study, we used the digital soil mapping approach (DSM) to predict SOM (0-20 cm) in cultivated soils in a hill-mountain region, Longshan County (LS), and a plain-platform region, Nanxian County (NX), which are both located at the same latitude in Southern China. By using 6746 and 9571 soil sampling points for LS and NX, respectively, together with 33 environmental covariates, the optimal spatial interpolation models and the reasonable sample strategy were carefully discussed. Results and discussion Descriptive statistical results showed that SOM in LS and NX were both moderate variations (coefficient variation, 0.34) and were approximately normal distribution. SOM in NX was strongly spatially dependent while SOM in LS was a moderate spatial dependence. The conditional Latin hypercube sampling (cLHS) was more appropriate compared with the Simple Random Sampling (SRS) as the sampling strategy. The optimal model for predicting cultivated land SOM was the Random Forest (RF) model for both LS and NX. The prediction accuracy was positively correlated with the sampling density. Specifically, to obtain a high prediction accuracy, the reasonable sampling density for SOM in LS should be controlled at >= 4.0 per km(2), higher than that in NX (>= 2.0 per km(2)). Conclusions The combination of cLHS and the RF model probably is the best choice for cultivated land SOM spatial prediction in different terrains. Therefore, our results provide a basis for future DSM of SOM in similar regions and help optimize soil sampling density.
MoreTranslated text
Key words
Soil organic matter,Spatial prediction,Digital soil mapping,Sampling density optimization,Random forest
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined