Abstract:

Grain protein is an important index indicating wheat quality status, and nondestructive and quick assessments of grain protein accumulationdynamics is necessary for cultural regulation and quality classification in wheat (Triticum aestivum L.) production. The objectives of this study were to determine the relationships between plant nitrogen nutrition status, grain protein accumulation, and canopy reflectance spectra in winter wheat, therefore, to derive regression equations for monitoring grain protein accumulation with canopy hyper-spectral remote sensing. Three types of cultivars, i.e., high protein content (Xuzhou 26 and Yumai 34), medium protein content (Yangmai 10, Yangmai 12, and Huaimai 20), and low protein content (Ningmai 9) were used in three field experiments under different nitrogen levels in the growing seasons of 2003–2006. Time-course measurements were taken on canopy hyperspectral reflectance, plant weight, nitrogen content and grain protein accumulation (GPA) during the experimental periods. The results showed that the cumulative value of canopy nitrogen nutrition index (CNNI) from anthesis to specific day were highly correlated with grain protein accumulation at corresponding day across grain filling with the best predictor of plant N accumulation (PNA). According to the regression analyses between vegetation indices and canopy nitrogen nutrition index,several key spectral parameters could accurately estimate the changes in plant N status across different growth stages, nitrogen levels, and growing seasons with the same spectral parameters for each wheat cultivar. According to the technical route of key spectral parameters-canopy N nutrition index-grain protein accumulation, estimating models on grain protein accumulation were constructed on the basis of canopy hyper-spectral parameters by linking the above two models with canopy N nutrition index as intersection in wheat. Tests with other independent dataset showed that the key spectral index SDr/SDb on the basis of the technical route of SDr/SDb-PNA-GNA could be used to predict grain protein accumulation from 7d after anthesis to maturity, with the predictive precision (R²) of 0.954 and the relative error (RE) of 13.1%, respectively. It can be concluded that dynamic change of grain protein accumulation in wheat could be monitored directly with key vegetation spectral index.

Key words: Wheat（Triticumaestivum L.), Canopy spectral information, Grain protein, Accumulation dynamics, Monitoring model