Effect Of Water Content On Soil Organic Matter Of Near Infrared Spectrum Detection

The main objective of the study was to analyze the influence to moisture on soil organic matter by using near infrared spectroscopy.54soil samples were collected in taigu and guandishan in shanxi. According to soil moisture content of preparation method maked up the different gradient of soil moisture content to54soil samples. Then different gradients of soil moisture were detected by the FieldSpec3of ASD spectrometers. Simultaneously standard testing methods was used to measure the soil organic matter content. Carrying out the following work.(1) The NIR spectrum of a one-dimensional characteristic bands and two-dimensional correlation were analyzed to different soil moisture content. The results showed that,within region350-2500nm, synchronous two-dimensional correlation spectra of54species relatively similar. Near2210nm and1929nm appear stronger the automatic peak with a weak peak near1415nm. According to the intensity of the three autocorrelation peak,1929nm at the corresponding functional groups is most sensitive to moisture, followed at2210nm place, at least sensitive to1415nm.(2)For soil moisture content, a (Partial Least Squares, PLS) model was available, using1365nm-1465nm、1879nm-1979nm、2160nm-2260nm wavelengths, fitted correlation coefficient and Root-Mean-Square Error of Fitted value is0.9587and0.0177respectively, and predicted correlation coefficient and Root-Mean-Square Error of Prediction (RMEP) is0.9443and0.0207respectively. Thus confirms the previous conclusion.(3)First, for dry soil, model of soil organic matter content was developed based on PLS using the spectral range from600nm to2500nm, then analyzed the result. Model of soil organic matter content was developed based on PLS by gradually removing unwanted and interfered bonds. Finally, the best bonds(700nm-800nm、999nm-1879nm、979nm-2500nm wavelengths) were got to develop model. fitted correlation coefficient is0.9764, Root-Mean-Square Error of Fitted value is0.0214,predicted correlation coefficient is0.9335, RMEP is0.588.(4)For soil samples of different moisture content, model of soil organic matter content was developed based on PLS by appling got the conclusions. Finally, the best bonds (999nm-1879nm、1980nm-2400nm bonds)were got to develop model. fitted correlation of coefficient is0.9605, Root-Mean-Square Error of Fitted value is0.0276,predicted correlation of coefficient is0.9501, RMEP is0.0640.(5)The results showed that, to700nm-800nm band,in modeling soil organic matter, is an important degree band. This band is not removed in modeling soil organic matter.(6)The results showed that, to1929nm band near, whether dry soil to modeling soil organic matter or different soil moisture content to modeling soil organic matter,,is interference band.