Study On Hyperspectral Prediction Of Soil Organic Matter Content Developed By Volcanic Debris

Soil organic matter is a carbonaceous organic compound in soil,which is one of the chemical properties of soil.At present,the traditional soil organic matter content determination methods are potassium dichromate-sulfuric acid digestion,TOC analyzer,elemental analysis.Although the chemical analysis of high accuracy,but the indoor chemical analysis time and effort.In recent years,with the continuous development of hyperspectral remote sensing technology and the in-depth study of mathematical multivariate statistical algorithms,hyperspectral remote sensing has been developed with high spectral resolution and abundant information in soil property prediction.The spectral data of the soil samples obtained by the indoor spectrometer are saved and saved by the established model,and the rapid determination of soil organic matter content can be realized.In this study,the basaltic rocks and the soils developed in the northeastern region were collected and the soil organic matter content was analyzed by chemical analysis.The hyperspectral data of soil samples were obtained by ASD FieldSpec4 spectrometer.By using the method of multiple linear regression,partial least squares regression and principal component regression,two kinds of soils were obtained by smoothing the original spectral reflectance,denoising and then using the continuous removal method,the differential method and the reciprocal logarithm method.The prediction model of the soil organic matter content of the lithology and debris is developed and the prediction model is established according to the correction correlation coefficient,the prediction correlation coefficient,the correction standard error,the forecast standard error,the crossed verification mean square error,the predicted variance and the prediction bias The optimal prediction model of organic matter content in two soils of lithology and volcanic debris was obtained.The results are as follows:In the range of 400?1300 nm,the spectral curves of the developed soils in the basaltic volcanic debris developed a sharp increase,and in the range of 1440?1860 nm increased slowly.In the range of 750?1380 nm the spectral curves of the developed soil were 400?750 nm,and it is gentle in the range of 1420?1880 nm.The spectral response of the organic matter in the basaltic volcanic debris is located at 500 nm and 800 nm.The spectral response of the trachyte debris is at 405 nm,465 nm,575 nm and 1105 nm.The first order differential,second order differential,reflectance reciprocal logarithm,first order differential of reflectance logarithm and second order differential treatment of reflectance logarithm are treated with the corresponding basaltic and trachyte The correlation analysis showed that the correlation of the organic matter content of the volcanic debris was significant.The maximum correlation coefficient of the maximum correlation coefficient is above 0.8,the maximum correlation coefficient of the first order differential is-0.8898,and the rough correlation of the volcanic debris The first order differential of the logarithmic logarithm and the second order differential of the reciprocal logarithm of the reflectance are also above 0.8,and the first order differential of the reciprocal logarithm of the reflectance is the largest The correlation coefficient is-0.9029.In the whole spectrum range,the prediction model of soil organic matter content of two lithologic clastic debris developed by multiple linear regression,partial least squares and principal component regression methods was used to the prediction results.Among them,the optimal prediction model of organic matter content of basaltic volcanic debris development soil is a stepwise linear regression model based on the first order differential of the reciprocal logarithm of spectral reflectance.The number of independent variables is 7,the prediction coefficient Rv2=0.9720,the predicted mean square error RMSEP = 2.0590,sig =-0.003<0.01.The optimal prediction model of soil organic matter content in the developed soil is characterized by the partial least squares regression model based on the first order differential of the reciprocal logarithm of spectral reflectance.The model independent variable number Pc=5,modeling correlation coefficient Rc = 0.9872,decision correlation coefficient Rc2=0.9745,modeling root mean square error RMSEC = 0.4821,standard error of calibration SEC = 0.4906,prediction decision coefficient Rv2 = 0.9702,root mean square error of prediction RMSEP =0.9563,standard error of prediction SEP = 0.9711,prediction bias difference Bias =-0.0637.