Hyperspectral Characteristics Of Organic Matter During The Aerobic Compost Of Agricultural Waste

The resource utilization of rice straw and other agricultural waste is mainly approached by aerobic composting,and the evaluation and monitoring of maturity is an important link for its compost productization.At present,conventional methods to judge the compost maturity degree by temperature,color or experience are difficult to accurately reflect the compost maturity process and degree of compost maturity under different conditions,such as the type of raw materials,proportion and addition of fermentation agent.Hyperspectral technology,with its high spectrum resolution and nanometer continuous spectral information acquisition ability,can effectively obtain the subtle spectral characteristics of maturity indexes,thus to provide a possibility for rapidly evaluating compost maturity.In this paper,we tracked the two modes of rice straw aerobic compost experiment:the open-type and semi-closed,and applied the hyperspectral technique and chemometrics to analyze the hyperspectral characteristics of organic matter(OM),total nitrogen(TN)and carbon-nitrogen ratio(C/N)that are the maturity indexes in aerobic composting.With detaily analysis of the effects of 13 kinds of spectral transformations on the spectral characteristics of OM,TN and C/N in different wavelength range,we used stepwise regression(SR),competitive adaptive reweighted sampling(CARS),successive projections algorithm(SPA),and synergy interval partial least squares(siPLS)to analyze and extracte the sensitive wavelengths of OM,TN and C/N,then established the spectral inversion model of OM,TN and C/N,respectively,in the process of aerobic compost based on the partial least squares regression(PLSR)with full-band and sensitive wavelengths.The main conclusions are as follows:1)450nm~1000nm spectral reflectance can indicate the changes of OM content,TN content and C/N.In the process of aerobic compost,OM content,TN content and C/N decrease gradually and the average spectral reflectance of 450nm~1000nm present a negative correlation with OM content,TN content and C/N,respectively.The spectral response characteristic of compost substrate is stable in 450nm~1650nm and not affected by the compost mode.2)Spectral transformation can improve the correlation of compost substrate spectrum with OM content,TN content and C/N to some degree in rice straw aerobic compost,which is conducive to the extraction of OM and TN sensitive wavelengths.This study finds that best spectral transformation for OM and TN is linear baseline correction(LBC)and multiplicative scatter correction(MSC),respectively.The best way to analyze C/N is using raw reflectance spectra(R).3)The extraction of sensitive wavelengths can reduce the number of spectral variables so as to simplify the inversion model.In the process of two aerobic compost experiments,the number and position of sensitive wavelengths of OM,TN and C/N in 450nm~1650nm are relatively stable.This study finds that the best extraction method for OM sensitive wavelengths is siPLS,and the results are concentrated in 1440.21nm~1645.19nm.The best method for TN is SPA,and the results are 970.73nm,1014.99nm and 1016.63nm.The best method for C/N is SPA,and the results are 454.21nm and 1468.32nm.4)In the open-type aerobic compost experiment,the best precision model of OM is LBC-siPLS-PLSR with R_p~2,RMSEP and RPD of 0.746,4.061 and 2.02,respectively.The best precision model of TN is MSC-SPA-PLSR with R_p~2,RMSEP and RPD of 0.650,0.205 and 1.72,respectively.The best precision of C/N is R-SPA-PLSR with R_p~2,RMSEP and RPD of 0.706,1.110 and 2.07,respectively.The reliability of these modeling methods is verified in the semi-closed compost experiment,and the precision of each index model is better than that of the original full-band PLSR model,indicating that the above models have good robustness.In conclusion,research on the quantitative relationshion between OM content,TN content,C/N and their spectral characteristics under different modes of rice straw aerobic compost shows:combining near infrared hyperspectral technique with chemometrics is an important and effective means to quantitatively detect aerobic compost OM content,TN content and C/N,and these results can provide scientific basis and data reference for accurately monitoring the process of straw fermentation,reducing the quality control cost of compost and evaluating the fertility of products,which is important scientific significance for intelligent monitoring of rice straw compost maturity by hyperspectral analysis technology.