Hyperspectral Response And Quantitative Monitor On Soil Particulate Organic Carbon In Winter Wheat Field

Winter wheat is an important food crop in China.Particulate organic carbon(POC)is one of the determinants of soil fertility,which can affect the soil quality of farmland and affect the yield of winter wheat crop.The accurate monitor of its content is another scientific research.Hyperspectral technology has specific function of high spectral resolution,rapid,non-destructive and accurate acquisition of target properties,and has become an effective means to obtain the organic carbon content of soil particles.In this experiment,the effects of tillage and fertilization on soil particulate organic carbon content and hyperspectral characteristics of different soil particulate organic carbon levels were investigated.The influence of different spectral pretreatment methods and modeling methods on POC hyperspectral estimation results were explored.To achieve accurate hyperspectral estimation of wheat field POC,the main conclusions are as follows:(1)With the advancement of wheat growth period,the content of POC in soil particulate organic matter firstly increased and then decreased,in 0-10 cm,10-20 cm soil layer,compared with traditional tillage(CT),no-till(NT)is more conducive to the increase of POC content in soil particulate organic carbon,and the highest content of POC is treated by no-tillage compound fertilizer(NT-CF)in the middle of filling.In the same growth period,the POC content gradually decreases as the depth of the soil increases.(2)The range of 450-600 nm is the main spectral response region of particulate organic carbon.The POC content level is inversely proportional to the spectral reflectance.The decrease of POC content leads to the increase of soil spectral reflectivity.In a sense,the soil spectral reflectance values or curve trend characteristics can indicate the level of POC content.(3)Different spectral pretreatment methods have different effects on the correlation between spectra and POC,and can improve the correlation between soil spectrum and POC content in some bands to varying degrees,among which,first-order differential(1-Der)and The second-order differential(2-Der)preprocessing method has the largest increase,especially the 2-Der transform spectrum has the highest correlation with POC,the correlation coefficient is 0.39.(4)The spectral pretreatment method has different effects on the monitoring model of POC.Except for multi-scattering correction(MSC),other pretreatments have improved the accuracy of POC model.The2-Der transform spectrum has significantly improved the soil particulate organic carbon.The accuracy of the partial least squares(PLSR)monitoring model(R2c=0.73,R2v=0.78),the model performed best.In this study,compared with the PLSR modeling method,the multiple linear regression(SPA-MLR)relatively reduces the accuracy of the POC correction model.The MLR model with 1-Der pre-processed spectra is the best.The POC model constructed in this study has a good predictive ability and can achieve accurate quantitative estimation of POC.