Study On Spatial Pattern Of Surface Soil Organic Carbon In Hilly Areas Based On Remote Sensing And Ground Support

To make a quick and easy prediction of soil organic carbon spatial distribution,this paper uses 289 sampling points of soil organic carbon(0 ~ 20 cm)collected from Mingshan District in Ya'an city,combining various brands of Landsat 8 remote sensing images,introducing the ground cofactors,and jointly establishing a prediction model of organic carbon content,and adopting 40 unmodeled sample inspection model accuracy,according to the root mean square error and the average deviation,standard deviation from the mean and standardized mean deviation,carrying different forecast model on the contrast and analysis,selecting the optimal model,while making space inversion in the last step to forecast spatial distribution of soil organic carbon in Mingshan District,The research results indicate that:(1)Soil organic carbon is significantly correlated with remote sensing band 1,3,4,7,NDVI,altitude,slope and topographic humidity index.Among the cofactors,organic carbon content and plane curvature were positively correlated with the topographic humidity index,but negatively correlated with elevation,slope,slope direction,section curvature,confluence dynamic index and sediment transport capacity.Soil organic carbon content was significantly positively correlated with the topographic humidity index(P<0.01).It has a significant negative correlation with slope and altitude(P<0.01),and vegetation and topographic factors significantly affected the spatial distribution of soil organic carbon.(2)The prediction model adopts the remote sensing image together with single-band and multi-band separate modeling methods and the ground factor common modeling method,the established prediction model can predict the soil organic carbon content in the study area.The prediction models of soil organic carbon content in remote sensing spectral bands are all up to the extremely significant level,indicating that the prediction model of soil organic carbon can be established only by remote sensing spectral bands.With the introduction of ground cofactor,the determination coefficient R2 of the organic carbon content prediction model was increased from 0.127 to 0.575.Among them,the prediction model of band 1,band 3,band 4,band 7 and NDVI,elevation,slope and topographic humidity index jointly established has a good performance,indicating that the method combined with ground cofactor can improve the accuracy of the prediction model.Therefore,it is necessary to consider the ground auxiliary data when using remote sensing to study the prediction model of organic carbon content in the area with complex terrain.(3)Spatial inversion results show that the overall trend of soil organic carbon spatial distribution in the Mingshan District is gradually increasing from northwest to southeast,which is basically consistent with DEM trend and NDVI distribution trend in the study area.The results indicated that Landsat 8 OLI image spectral information and ground cofactor could be used to better reflect the distribution pattern of soil organic carbon,and the results provided us with theoretical basis and method reference for the prediction of soil organic carbon content and spatial distribution in hilly areas.