Study On Hyperspectral Characteristics Of Chlorophyll And Total Nitrogen Content Of Camellia Oleifera And Rice In Red Soil Area

As the largest soil type area in China,red soil region is an important food and oil crop production area of camellia oil and rice in south China.In this region,soil fertility is low,soil compaction is easy,and soil erosion is serious.How to develop modern agriculture in the fragile ecological environment of red soil region to drive the development of local economy is a major challenge.The appearance of hyperspectral quantitative remote sensing technology provides a new way to realize the development of modern agriculture.In this study,two green plants Camellia Oleifera Abel and Oryza sativa in the red soil area of Jiangxi province were selected as the research objects.Through obtaining the canopy and leaf hyperspectral data of Camellia oleifera and rice,combined with the chlorophyll and total nitrogen contents of Camellia oleifera Abel and rice measured in the laboratory,Extracted the high spectral bands,the spectral index and spectrum characteristic parameters with different high spectrum parameters,such as correlation analysis,so as to obtain different instructions oil-tea and chlorophyll content,total nitrogen content change of rice high sensitive spectral parameters,in different high spectral parameters as the model input vector to establish different estimation model of chlorophyll,total nitrogen,The hyperspectral characteristics of camellia oleifera and rice were compared comprehensively,and the differences and similarities of the best estimation model and the best input model for chlorophyll content and total nitrogen content of camellia oleifera and rice were clarified.The main conclusions are as follows:(1)Camellia and rice leaves,crown height spectrum with the change of chlorophyll content,total nitrogen content are mainly concentrated in the visible light response of the band-near infrared area(400～1350 nm),and the change rule of are the same,in the visible light region,with the increase of chlorophyll content,total nitrogen content,reduce spectral reflectance,showed opposite regularity in near infrared region.(2)In the visible region,the leaf reflectance of camellia sinensis with different chlorophyll and total nitrogen contents was lower than that of rice,and in the near infrared region,the canopy reflectance of camellia sinensis with different chlorophyll and total nitrogen contents was higher than that of rice.With the change of chlorophyll content and total nitrogen content,the spectral reflectance of camellia oleifera leaves and canopy varied greatly in the red edge and near infrared region,and the spectral variation of rice leaves and canopy varied greatly in the whole band(400～1350nm).(3)MCARI,TCARI,OSAVI,Bm ND705,VARI,VOG2,VOG1,Db,Dy,Dr,λr,SDb,SDr,SRg and SRr were common hyperspectral parameters sensitive to changes in leaf chlorophyll content of camelia oil and rice.NDVI,SAVI,OSAVI,Db and SDb were common hyperspectral parameters sensitive to canopy chlorophyll content changes of camellia oleifera and rice.MCARI,TCARI,VOG2,VOG1,Db,SDb and SDy were common hyperspectral parameters sensitive to the change of leaf total nitrogen content in camoil tea and rice.NPCI,VOG2 and VOG1 were common hyperspectral parameters sensitive to canopy total nitrogen content of camellia oleifera and rice.(4)The RFR model based on hyperspectral characteristic parameters was the best model to estimate the chlorophyll content of camellia oil leaves,The modeling R~2,RMSE and RPD were 0.89,0.03 and 3.03,respectively,The validation R~2v,RMSEv and RPDv were 0.68,0.05 and 1.78,respectively.The RFR model established by hyperspectral characteristic parameters is the best model to estimate the total nitrogen content of camellia oil leaves,The modeling R~2is 0.79,RMSE is 0.72,and RPD is 2.18,The validation R~2v,RMSEv and RPDv are 0.62,1.03,and 1.63,respectively.The PLSR model based on the full spectrum has a slightly lower accuracy and can also better estimate the total nitrogen content in camellia oleifera leaves The modeling R~2,RMSE and RPD were 0.75,0.78 and2.02,respectively,and the validation R~2v,RMSEv and RPDv were 0.62,1.03 and 1.63,respectively.(5)Among the estimation models of chlorophyll and total nitrogen contents in rice at different growth stages,the estimation models of chlorophyll and total nitrogen contents in rice leaves and canopy at elongation and heading stages had the highest accuracy,and the SVM model based on full spectrum(400～1350 nm)had the best estimation effect.