Research On Estimation Model Of Soil Organic Matter Content In Yellow River Delta Based On Hyperspectral

Soil organic matter(SOM)is an important index to measure soil fertility and an important nutrient source for crop growth.Soil organic matter can not only improve soil fertility but also improve soil physical and chemical properties,which plays an important role in improving the quality of cultivated land.Therefore,accurate and efficient monitoring of soil organic matter content has important practical significance for soil improvement,fertilizer preservation and efficiency improvement.In this paper,taking the reclamation area of Yellow River Delta in Shandong Province as the research area,each soil sample collected in the study area was taken back to the room for air drying and evenly divided into four parts.One was used to measure soil organic matter content,and the other three were made into soil samples with different particle sizes of 2 mm,0.25 mm and 0.15 mm.The Spectral data were measured using ASD(Analytical Spectral Devices)Field Spec 3 surface object spectrometer.Based on the measured original spectral data,Savitzky-Golay smoothing,Savitzky-Golay smooth-first derivative,Savitzky-Golay smooth-multiple scattering correction and other methods were used to preprocess them.Then,the sensitive wavelengths were selected by correlation analysis method,multiple linear stepwise regression method and continuous projection algorithm.Finally,the partial least squares estimation model of organic matter content of soil with different particle sizes was established by using the screened sensitive wavelengths.Through the test of the estimation model,the best estimation model,the best particle size,the best pretreatment method and the best wavelength screening method were selected to provide technical support for the rapid and accurate estimation and soil improvement of topsoil organic matter content in the study area.The main results are as follows:(1)The soil samples were grinded and screened into three different particle sizes of 2 mm,0.25 mm and 0.15 mm.According to the established estimation model,it was found that the particle size had a significant effect on the estimation of soil organic matter content.When the particle size was 0.25mm,the estimation model of soil organic matter content had the best effect.When the particle size was less than 0.25mm,the estimation effect of the model decreased with the decrease of particle size.(2)Savitzky-Golay smoothing,Savitzky-Golay smooth-first derivative and Savitzky-Golay smooth-multiple scattering correction were used to preprocess the original spectra of soil with different particle sizes,which not only reduced the noise influence of the original spectra,but also significantly highlighted the characteristic information of the spectra.Through research,it was found that the effect of the estimation model preprocessed by Savitzky-Golay smoothing spectrum was better than the estimation model preprocessed by Savitzky-Golay smoothing-first derivative and Savitzky-Golay smoothing-multiple scattering correction.(3)Correlation analysis method,stepwise multiple linear regression method and successive projections algorithm were used to screen the sensitive wavelength of soil organic matter,and the influence of different wavelength screening methods on the estimation of soil organic matter content was discussed.By comparing the model estimation effect,it was found that the estimation model based on successive projections algorithm had the best overall effect,followed by stepwise multiple linear regression method and correlation analysis method.Above all,by comparing the different wavelength selection method under 2 mm,0.25 mm and 0.15 mm particle sizes to establish models for predicting the organic matter content of soil,based on successive projection algorithm of particle size of 0.25 mm soil spectra after Savitzky-Golay smoothing pretreatment of partial least squares estimate model prediction precision optimal,the determination coefficient R~2,RMSE and relative root mean square error RPD was0.953,0.351 and 4.029,respectively.The research results shown that it was feasible to divide the soil particle size and use hyperspectral technology to construct a partial least square estimation model to quickly estimate the soil organic matter content of the saline-alkali soil in the Yellow River Delta.