| Sustainable use and protection of the cropland in the black soil region of Northeast China is a major issue directly related to the national food and ecological security.With the increasing pressure of high-intensity human activities on black soil resources,the problem of black soil degradation,as reflected by the widespread soil erosion and barrenness,has become increasingly prominent.The large-scale,high-intensity land development and utilization of the black soil region and the lack of water and soil conservation measures during the reclamation process in the 20 th century have led to the region having the fastest growth rate of soil erosion since the 21 st century.Research on the process of soil erosion in the black soil area is usually carried out with the aid of soil erosion models,but the existing models are inadequate to account for the spatial and temporal dynamics of the soil erodibility parameter.The reason could be attributed to the lack of a fast,efficient and cost-effective method to capture the soil erodibility index.In view of the current severity of soil erosion in the black soil area of Northeast China and the lack of high-resolution soil erodibility data,this study aims to establish a method to effectively quantify soil aggregate stability index,using high-resolution field-and satellite-based hyperspectral remote sensing.The aggregate stability index will then be used as a proxy for soil erodibility,to realize the analysis on the spatial characteristics of soil erodibility in the study area.This study selected three county-level administrative units of Nong’an,Dehui,and Jiutai in the central part of Jilin Province as the study area.Firstly,it adopts methods such as field soil sampling campaign,laboratory tests on elementary soil physical and chemical properties,and soil aggregate stability quantification.Pearson correlation analysis was used to verify whether the significant positive correlation between soil organic carbon(SOC)and soil aggregate stability,which has been widely reported by previous authors,also exists in the black soil area,and establish a linear prediction model for soil aggregate stability,with soil organic carbon as the independent variable;secondly,the study aims to establish a SOC prediction model based on field-and satellite-based hyperspectral imaging technology,and then to adopt linear regression analysis to generate a regional-scale,high-resolution spatial distribution map of soil aggregate stability index,as an indicator of soil erodibility;finally,to carry out regional and field-scale soil erodibility spatial analysis by means of semi-variance analysis,in order to compare the spatial distribution of soil erodibility at different scales and explore their respective control mechanisms.The main conclusions are as follows:(1)The distribution range of SOC content in the study area was mainly between 1%and 2%,with an average SOC content of 1.51%.Using the soil aggregate stability measurement method(LB method)proposed by Le Bissonnais,it was found that the degree of aggregate stability(expressed as the average weight diameter MWD)in the study area was generally low(MWD: 0.2～0.4 mm).Through linear regression analysis,it was found that there was a significant positive correlation between SOC and MWD(Pearson correlation coefficient: 0.88).On this basis,a linear prediction model of soil aggregate stability(MWD=-0.315+0.396×SOC)with SOC as an independent variable was established for the rapid prediction and pixel-based mapping of soil aggregate stability using hyper-spectroscopic techniques.(2)Laboratory visible-near infrared(Vis-NIR)soil spectroscopy was used to verify the feasibility of SOC prediction in the study area.To this end,partial least squares repression(PLSR)was used to establish a SOC prediction model based on laboratory-based soil spectral data.The results show that the PLSR model had a good prediction accuracy(R2=0.74,RMSE=0.22%),which proved the feasibility of using hyperspectral data to predict SOC content in the black soil region.Then,Sentinel-2(S2)satellite imagery was used to develop a methodological framework that is capable of producing pixel-wise prediction of SOC content in the study region.Using a combination of bare soil pixel extraction,multitemporal mosaicking,SOC prediction model development with S2-derived spectra,and uncertainty analysis,the study achieved high-solution SOC map with S2 bare soil spectra as the data source.Compared with the use of single-date S2 images for SOC content prediction,the multi-temporal S2 bare soil composite could provide a more robust SOC prediction model with greater coverage of cropland area and higher accuracy.Compared with previous studies using S2 remote sensing to predict SOC content,the SOC prediction model established in this study reached a higher prediction accuracy in areas other than Europe.This stressed the promising potential of S2 remote sensing for characterizing soil properties in the black soil region.(3)Based on the pixel-level SOC map,together with the linear prediction equation of soil aggregate stability,the soil erodibility index quantification and high-resolution spatial characterization methods were established,and the soil erodibility map for the study area was generated.Through investigating the spatial distribution characteristics of the soil erodibility index at both regional and field scales by means of semi-variance analysis,it was found that at the regional scale,the range of the semi-variogram model was 3 725 m,and the areas with higher soil erodibility values were mainly concentrated in the flat terrain where the Phaeozem and Chernozem soils are found,mostly in Dehui City,while the areas with low soil erodibility were concentrated in the northwest of Nong’an County and the sloping farmland in Jiutai District;at the field scale,contrasting ranges were found from the semi-variograms from 3 selected representative areas with different topographichydrological microenvironments.In particular,the range was relatively small(596～744 m)in areas with substantial variations in topography and hydrologic conditions,while the range became larger(1 234 m)when the gradient of environmental factors was small increases to 1 234 m.This indicates that different controlling factors were responsible for the spatial heterogeneity of soil erodibility at different scales,and at the field scale alone,different topographical conditions could also lead to contrasting spatial structures of soil erodibility.In summary,this study built on the latest research progress of using S2 remote sensing for SOC prediction and mapping.A fast and efficient method for the multi-scale,high-resolution prediction of soil erodibility was developed using satellite-and field-based soil spectroscopy.The research results can not only offer new insights into the evaluation and monitoring of soil erosion in the black soil region of Northeast China,but also provide a theoretical and data basis for the formulation of national black soil protection policies and the precise placement of regional water and soil conservation measures. |
