Diagnosis Of Nitrogen Nutrition In Winter Oilseed Rape Based On Hyperspectral Reflectance Data

Winter oilseed rape is a very important oil crop in our country,and has the characteristics of easy planting,easy management and wide adaptability.Nowadays it has become the fourth most important crop after rice,wheat and corn.Nitrogen(N)is the key nutrient element for crop growth.The amount of N fertilizer applied to the field has a great impact on the quality and yield of winter oilseed rape.Thus accurate and rapid diagnosis of crop N status is important to ensure grain yield and quality while minimizing environmental damages caused by the excessive application of N fertilizers.The diagnosis of crop nitrogen nutrition based on the hyperspectral remote sensing method is a hot research topic since the 1970 s.With the fast development of hyperspectral remote sensing technology,high spectral resolution has been greatly improved in recent years.Information of continuous narrow spectral bands in the visible light and near infrared,intermediate infrared and thermal infrared wavelength range are gained through hyperspectral remote sensing,which are useful to detect crop nitrogen status.The paper aims to establish the model based on remote sensing technology to accurately diagnosse N status in winter oilseed rape.The nitrogen nutrition index(NNI)can directly determine whether winter rape plant nitrogen concentration to meet the current need to grow.Through establishing the inversion model of winter rape canopy spectral reflectance with NNI,the purpose of estimating winter rapeseed crop nitrogen nutrition status can be realized by using remote sensing technology.The research conducted the empirical method and mechanistic method to estimate NNI.The partial least-squares regression(PLSR)method was used to directly establish the model for winter rape canopy spectral reflectance after the logarithmic transformation to predict NNI in the direct method of estimating NNI.The indirect method estimated two parameters,which were plant nitrogen concentration(PNC)and critical nitrogen concentration(Nc),separately based on hyperspectral reflectance data.We analyzed the physiological and biochemical parameters sensitive to PNC,leaf area index(LAI)and dry matter(DM),using hyperspectral vegetation indices(VIs)based on the mechanism of inversting PNC,LAI and DM.We found that VIs that are effective to estimate of PNC needed to be sensitive to chlorophyll content(Chl)and carotenoid content(Car),particularly sensitive to Chl,and not sensitive to the changes of LAI and DM.Further analysis of the relationships between VIs and PNC showed that the anthocyanin reflectance index 1(ARI-1)provided the most accurate estimates of PNC.We estimated DM based on the logarithmic relationship between LAI and DM,in which LAI was estimated with the green index 1(GM-1).DM was then used to calculate Nc,and NNI was the ratio of PNC estimates to Nc estimates.Compared with the mechanistic method,the empirical method,especially PLSR with the logarithmic transformation of spectra,provided more accurate NNI estimates.Although the mechanistic method involved more parameters,and several estimations are needed to compute NNI.The mechanistic method contained more comprehensive physiological information,and had a stronger interpret ability.Results from this study provided strong evidence of estimating NNI from hyperspectral remotely sensed data for diagnosing crop N status and facilitating N fertilizer management.