The Influence Of Water And Texture On The NIR Characteristics Of Soil Organic Matter And The Establishment Of Anti-interference Model

The effects of soil texture and soil moisture on the detection of near-infrared soil organic matter(SOM)were analyzed,and the influence of soil moisture on SOM was analyzed.The main research is as follows:1,Contrast analysis of three different sand,loam and clay soil texture of soil,the results show that the absorbance values of different soil texture is different,with sand absorbance value maximum,clay,loam is minimal.The height and width of the absorption peak vary with the texture.2,Using support vector machines SVM to classify different soil texture modeling prediction,the results of SVM classification accuracy of 91.67%,and the soil texture classification forecasting can improve the prediction accuracy of soil organic matter.3,In the collection close emperor mountain in Shanxi Province,the ZhongYang,mesa,NingWu,LouFan,TaiGu 50 soil samples in different parts of the six as the research object,in the laboratory after natural air drying and sieving processing,preparation were 4%,8%,12%and 16%,four kinds of moisture content of soil samples,41 dry soil samples were randomly selected as model samples,predicting samples under different moisture content for(4%,8%,12%,16%and dry soil)of 9 samples of prediction set,the analysis of the effect of soil moisture of near infrared spectrum detection,the results show that the prediction correlation coefficient(R)is 0.569,Prediction(SEP)of 0.835,standard error and the root mean square prediction error(RMSEP)was 0.898,the model of the effect is not ideal,existing in the absorbance spectra of the soil moisture absorption effect,interfere with the accuracy of the model.4,In order to eliminate the influence of moisture,prediction of organic matter,using water sensitive wavelength of 2210 nm,1415 nm and 1929 nm the MDI,moisture correction coefficient of different moisture content using MDI spectrum correction,in order to verify the effect of moisture correction coefficient MDI,choose 4 of the same prediction data comparison and analyses the different moisture content through the moisture correction processing near infrared spectrum curve and the model prediction results.Following conclusions:(1)after moisture coefficient of correction of equivalent dry soil spectral absorbance figure was lower than that without spectrum correction processing of wet soil,but also with the original spectrum close to dry soil sample.(2)using the above dry soil organic matter inversion model,predict the MDI correction after samples of different moisture content prediction,prediction correlation coefficient(R)is 0.783,standard error(SEP)is 0.505,and the root mean square prediction error(RMSEP)of 0.558.It can be seen that the moisture correction coefficient of MDI adjusted forecast sample statistical parameters without calibration process is obviously better than the prediction effect,correlation coefficient R value increased by 0.214,standard error prediction(SEP)value reduced 0.33,the root mean square prediction error is RMSEP value was reduced by 0.34.This suggests that,this study proposed the moisture correction coefficient algorithm can reduce the moisture of soil samples in absorbance spectrum interference,through the establishment of the equivalent dry soil spectrum can eliminate moisture,improve soil organic matter the applicability of the model under different moisture content of soil samples.5,Direct standardization(DS)algorithm is used to eliminate caused by soil moisture prediction model of soil organic matter test conditions lead to transfer problems,select select the same prediction data contrast and analysis through the DS algorithm correction processing of different moisture content near infrared spectrum curve and the model prediction results.Following conclusions:(1)through the DS algorithm adjusted the equivalent dry soil absorbance spectrum of figure was lower than that without the correct processing of the wet earth spectrogram,and with the original spectrum close to dry soil sample.(2)by using the above dry soil organic matter inversion model,predict the DS algorithm adjusted the samples of different moisture content prediction,prediction correlation coefficient(R)is 0.628,standard error(SEP)is 0.683,and the root mean square prediction error(RMSEP)of 0.691.Can be seen that the DS algorithm adjusted the prediction data statistical parameters without calibration process is obviously better than the prediction effect,correlation coefficient R value increased by 0.059,standard error prediction(SEP)value reduced 0.152,the root mean square prediction error is the RMSEP value was reduced by 0.207.This suggests that the DS algorithm can reduce the moisture content of soil samples in absorbance spectrum interference,can eliminate moisture,improve soil organic matter the applicability of the model under different moisture content of soil samples.6,Based on the water correction coefficient method and the DS model transfer algorithm,the model prediction effect of water correction coefficient is relatively good in the prediction model of soil organic matter with different moisture content.