Research On Remote Sensing Inversion Model Of Soil Organic Carbon Based On Landsat8

Soil carbon content and carbon change have a significant impact on global climate change.The change of soil organic carbon is affected by many factors.The complex feedback mechanism between the carbon cycle process and carbon flux determines that its estimation needs to go through a complicated calculation process.The advantages of remote sensing technology provide convenience for the indirect estimation of soil organic carbon.In order to accurately estimate the content of soil organic carbon,understand its regional spatial distribution,and select the optimal organic carbon prediction model,this paper is based on Landsat8 remote sensing images and takes the Changhe River Basin as the research area,Using the normalized vegetation index NDVI(Normalized Vegetation Index),ratio vegetation index RS(Simple Ratio Index),combined with relevant meteorological and vegetation data,etc.,using CASA(Carnegie-Ames-Stanford approach)combining remote sensing and carbon cycle models The model inverted the soil organic carbon in the study area in 2015,and compared with the measured organic carbon to obtain the soil organic carbon inversion accuracy based on the model;at the same time,the DNDC(Denitrification-Decomposition)model was used for the area in 2015.The organic carbon of three different crop planting systems was simulated,and the organic carbon simulation value based on the crop model DNDC was obtained.Finally,based on the set Kalman filter assimilation algorithm,the data prediction of the two models was assimilated,and the different prediction accuracy was compared.Excellent prediction model.The main research conclusions are as follows:(1)The spatial distribution pattern of soil organic carbon inversion shows that: the organic carbon content in the study area is about 99% of the total content in the range of 0-20 g/kg;the organic carbon content in the edge hilly area is lower than that of Changhe The planting areas on both sides;the implementation of land reclamation and ecological reconstruction measures after mining has brought the organic carbon content in parts of the central region to 20-30 g/kg;(2)The total average accuracy of organic carbon prediction based on remote sensing has reached 83.4%,out of mold The average prediction accuracy of the verification points is 86.6%,and the overall average accuracy can well predict the soil organic carbon in the study area;the measured average and inversion average are 10.266 and 8.951,respectively,and the inversion value is slightly lower than the actual measurement;(3)DNDC The average simulation accuracy of the simulated organic carbon value and the measured value is 82.7%,the relative error is 17.3%,and the prediction accuracy is better;compared with the measured value,the simulated average and maximum and minimum values of DNDC are higher than the measured value;(4)Comparison of the accuracy of the CASA model and the crop model DNDC found that the carbon cycle model based on remote sensing has higher organic carbon prediction accuracy.The assimilation accuracy of the two models after assimilation reached 85.8%.The accuracy evaluation result is assimilation algorithm>CASA model>DNDC model.