Remote detection of canopy leaf nitrogen concentration in winter wheat by using water resistance vegetation indices from in-situ hyperspectral data

Abstract Non-destructive monitoring of wheat nitrogen (N) status is essential for precision N management during wheat production. In this study, the quantitative correlation between leaf N concentration (LNC) and ground-based canopy hyperspectral reflectance in winter wheat was investigated. Field experiments were conducted for four years at different locations (Xinyang, Zhengzhou, and Shangshui) in China. Different N application rates, planting density, growth stages, and wheat cultivars were used. We developed a novel index (water resistance N index [WRNI]) that integrated the advantages of an index that minimizes water effects and an index sensitive to LNC. Data showed that the proposed combined index (WRNI), the ratio of the normalized difference red-edge index (NDRE) and floating-position water band index (FWBI) was both sensitive to LNC and resistant to variations in leaf water. Then, we optimized the bands of NDRE/FWBI to create an integrated narrow-band vegetable index ( ( R 735 - R 720 ) × R 900 R min ( 930 - 980 ) × ( R 735 + R 720 ) ) to trace the dynamic changes in LNC in winter wheat. Our novel index and 15 selected common indices were tested for stability across growth stages, locations, years, treatments, cultivars, and plant types in estimating LNC in winter wheat. Six of the 16 previously determined indices performed well, and R 705 /(R 717  + R 491 ) and mND 705 both showed the highest coefficients of determination (R 2  = 0.832 and 0.818, respectively) and the lowest root mean square error (RMSE = 0.401 and 0.417, respectively). When compared with the optimized common indices, the novel index WRNI was most closely correlated with LNC, and the corresponding linear equation yielded R 2  = 0.843 and RMSE = 0.382 across the whole 16 datasets; this further indicated a superior trace for LNC changes under heterogeneous field conditions. These models can accurately estimate LNC in winter wheat, and the novel index WRNI is promising for detecting LNC on a regional scale in heterogeneous fields under variable climatic conditions.