Monitoring of leaf nitrogen content of winter wheat using multi-angle hyperspectral data

Hyperspectral technology, which has been used in rapid and non-destructive monitoring of crop nitrogen status, is of great significance for the nitrogen fertilization management in modern agriculture. However, most researches collect hyperspectral data from the vertical angle, always leading to the inaccurate estimation of crop nitrogen. Studies have found that using multi-angle spectral data could improve the accuracy in estimating crop leaf nitrogen content (LNC). In this study, the LNC at different leaf positions of winter wheat under five nitrogen treatments was measured and the multi-angle spectral reflectance of leaves were collected. The results showed that the top third leaf were the most sensitive to nitrogen application rate. Correlation analysis showed that the correlation between LNC and spectral reflectance obtained at 0 degrees was the highest, followed by that between LNC and spectral reflectance obtained at 10 degrees, 30 degrees, 40 degrees, and 20 degrees. Moreover, a model based on the multi-angle composite vegetation index (MACVI) for LNC estimation was constructed through combining the difference vegetable index (DVI), the normalized difference vegetable index (NDVI), and the ratio vegetable index (RVI) and using the spectral data obtained from multiple leaf inclination angles, finding that this model could improve the estimation accuracy. The accuracy of the model based on the spectral reflectance obtained at 0 degrees, 10 degrees, and 20 degrees was higher than the others, and the coefficient of determination (R (2)) for the MACVI(D,R)-based model was the highest. The MACVI-based model proposed in this study could effectively improve the estimation accuracy of winter wheat nitrogen content, and provide scientific guidance for the nitrogen fertilization management in winter wheat cultivation.