HYPERSPECTRAL PREDICTION OF SOIL ORGANICMATTER IN WINTER WHEAT FIELD BASED ONCONTINUUM-REMOVED METHOD

摘要

In present study, the correlation relationship be- tween soil organic matter (SOM) and the raw spectra, as well as the continuum-removed (CR) spectra were analyzed and the partial least squares regression (PLSR) models of SOM based on two types of spec- tra were also established to improve the predictive accuracy and efficiency of soil organic matter (SOM) by using the spectroscopy technology. The results showed that there was a higher correlation coeffi- cient for the CR spectra than the raw spectra. More- over, the model performance for predicting the SOM based on the CR pre-treatment was superior to the raw spectra with R2-0.845. RMSE-0.253 and RPD-2.536. It implied that CR method could high- light the SOM spectral information and improve the predictive accuracy of SOM model. The study wouldsupply some methods and references for monitoringSOM and other soil attributes by using the hyper- spectral technology.