Hyperspectral Monitoring Of Siol Organic Carbon And Nitrogen In Millet Field The Stufy

Soil carbon and nitrogen,as one of the most important nutrients in millet field,is an important index to measure soil fertility characteristics,and also an essential element for millet growth,yield and quality formation.The traditional methods of soil carbon and nitrogen determination are mainly chemical analysis in laboratory.Although the results are reliable and accurate,there are some problems such as time-consuming,complicated operation,high cost and environmental pollution.As a new modern science and technology,hyperspectral analysis technology has the characteristics of fast,low cost,environmental protection,non-destructive and repeatable.It can obtain field data information in time and meet the current development needs of "precision agriculture".In view of this,this experiment takes the Millet Field Soil of Shanxi Agricultural University experimental station in Taigu County as the research object,studies the response mechanism of soil and hyperspectral,extracts and determines the hyperspectral characteristic information of soil particulate carbon and nitrogen,and comparatively analyses the influence of hyperspectral pretreatment methods and model construction on the overall performance of soil particulate organic carbon and nitrogen model,and determines the best spectral pre-monitoring of soil carbon and nitrogen.Finally,the accurate spectral monitoring of soil carbon and nitrogen content is realized.The main conclusions are as follows:(1)Different spectral pretreatment methods have different effects on the correlation between spectra and particulate organic carbon and nitrogen content.Soil particulate organic carbon and nitrogen content is negatively correlated with the original spectral reflectance,but the correlation between them will change greatly after different transformation treatments,and the correlation has been significantly improved.After R'transform treatment,the correlation coefficient between spectral reflectance and soil particulate organic carbon and nitrogen content was 0.613 and 0.493,respectively.(2)There are some differences in the characteristic bands of particulate organic carbon and nitrogen under different transform spectra.In this paper,we use stepwise multiple linear regression to extract the sensitive bands of the original spectra of particulate organic carbon,which are mainly located in the region of 400-420 nm;(1/R)'sensitive bands are mainly located near 866 and 1650-1690 nm;R'sensitive bands are mainly located in the bands of 800-840 and 2308 nm;and R' sensitive bands are mainly located near 583 nm and 1600-1620 nm.The original spectral sensitive band of particulate nitrogen is mainly located in the400-450 nm region;(1/R)'sensitive band is located at 2200-2230 nm;R'sensitive band is mainly located at789-830 nm;and R' spectral sensitive band is 1999 nm.(3)In this study,PLS-SMLR was used to construct the characteristic band model of soil particulate organic carbon and nitrogen content,and the models were compared under different pretreatment transformations.Among them,the model with R'transform is the best and achieves better prediction results.The model of particulate organic carbon content was the best under R'transform treatment(R2C= 0.71,R2V= 0.66),and the model of nitrogen content was the best under R'transform treatment(R2C= 0.93,R2V=0.52).The practical application potential of the comprehensive model,combined with PLS-SMLR and R'transform spectroscopy,can be considered to estimate the carbon and nitrogen content of particulate organic carbon and nitrogen model.The soil granule organic carbon and nitrogen model established in this study has good prediction ability,and can accurately and quantitatively estimate soil particulate organic carbon and nitrogen.