Estimating Plant Nitrogen Accumulation in Winter Wheat Based on Nitrogen Partitioning Model and Hyperspectral Data

Real-time and non-destructive monitoring of nitrogen accumulation is important for accurate nitrogen management and high quality and efficient production of wheat. The problem is most of the existing remote sensing monitoring models for wheat nitrogen accumulation are based on a single growth period and cannot be extended throughout the whole growth period. Therefore, this study comprehensively used hyperspectral data and relative growth stage (RGS) to systematically analyze the dynamic changes of nitrogen allocation between leaf and plant with phenological progress, to construct a nitrogen allocation model, and finally to achieve the prediction of plant nitrogen accumulation (PNA) in the whole growth period. The results showed that (1) the ratio of leaf nitrogen accumulation (LNA) to PNA and RGS showed a significant linear relationship, with a coefficient of determination ($\mathrm{R}^{2}$) of 0.90. (2) The green band normalized difference vegetation index (GNDVI) was highly correlated with LNA, and the linear model based on GNDVI inversion of LNA had high accuracy with $\mathrm{R}^{2}$ and root mean square error (RMSE) of 0.70 and 1.31 g/$\mathrm{m}^{2}$, respectively. (3) The coupled nitrogen partitioning model and LNA inversion model were used to invert the PNA, and the $\mathrm{R}^{2}$ and RMSE of the validation data were 0.72 and 2.98 g/$\mathrm{m}^{2}$, respectively, with a high agreement between the predicted and measured PNA. Overall, this study can accurately estimate the nitrogen accumulation in winter wheat plants at different growth periods and provide a basis for accurate diagnosis and quantitative fertilization of wheat growth.