| Soil is the largest carbon reservoir in terrestrial ecosystems and is therefore of great interest in studies of global carbon cycles and balances and climate change.Soil Organic Carbon(SOC),as an important attribute of soil,is an important indicator for assessing soil fertility and quality,affecting various aspects such as agricultural production,climate change,land degradation control and ecosystem service functions.The spatial distribution of SOC and its stock are the basis for studying its role in the global carbon cycle and for integrating SOC changes into environmental process modelling.this research takes the Jianghan Plain as a case study to map the soil organic carbon stock(SOCstock)at 1 m depth with high spatial resolution.Within the framework of Digital Soil Mapping(DSM),this paper uses historical soil data from the Jianghan Plain,combined with high-precision environmental covariate data,to establish a soil-landscape prediction model using the random forest method,to map SOC content,SOC density and their uncertainty distribution at different depths,and to analyse their spatial distribution characteristics;to analyse and compare the differences in SOC content and density under different soil types and land use practices;and to calculate the 1 m deep SOCstock.The main research contents and results are as follows.(1)A soil-landscape prediction model based on the "Scorpan" function was developed based on 66 historical sample points with topographic,climatic,biological,soil and land use covariates extracted at 90 m resolution.The results showed that the R2 of the modelling set was 0.84 and the RMSE was 1.88:after 50 random sample returns,the validated R2reached 0.59 and the RMSE reached 2.61.The model accuracy of the surface layer(0-30 cm)reached 0.60 and the RMSE was 2.62;the model accuracy of the middle layer(30-60 cm)and The accuracy of the model for the middle layer(30-60 cm)and the model for the deep layer(60-100 cm)were lower,with RMSEs of 3.70 and 4.93,respectively,indicating that the prediction ability of the model for the surface layer was higher than that for the deep layer,and more means should be explored in future studies to improve the prediction accuracy of the lower layer SOC.(2)The SOC content and density at vertical depth basically showed a decreasing trend with increasing depth.The surface SOC content was up to 16.81 g kg-1 and the density up to 6.41 kg m-2.The spatial distribution of SOC content and density decreases from southeast to northwest at different depths,with the most obvious trend in the middle layer(30-60 cm).The uncertainty of SOC content mapping showed an opposite trend,with the surface layer showing the least uncertainty and the middle layer showing the greatest uncertainty.(3)SOC content and density were positively correlated.The average SOC content and density relationships between different soil types in the Jianghan Plain are:1)yellow-brown soil>red soil>rice soil>swampy soil>tide soil>yellow-brown soil in the surface layer;2)swampy soil>red soil>rice soil>yellow-brown soil>tide soil>yellow-brown soil in the middle layer;3)yellow-brown soil>rice soil>red soil>yellow-brown soil>swampy soil>tide soil in the deep layer.SOC content and density vary depending on the physical and chemical properties of the soil,texture structure,intensity of human disturbance and cropping structure.(4)Using the DSM technique,we obtained SOC content and density maps at 1 m depth at 90 m resolution,and the SOC stock and content and density at different depths showed the same trend as the surface layer(709.13×105t)>(608.2×105 t)>the middle layer(475.53×105 t),i.e.the total SOC stock at 1 m depth in the Jianghan Plain was obtained as 1792.86 × 105 t.This study has identified a baseline of SOC stock in the study area,which will provide a scientific reference basis for later soil quality assessment and agricultural policy formulation. |
PDF Full Text Request