Nitrogen content inversion of wheat canopy leaf based on ground spectral reflectance data

Abstract

Canopy nitrogen content in wheat is a key indicator of wheat grain yield and quality. When using remote sensing technology to predict wheat canopy nitrogen content, a hyperspectral mode with high adaptability and high accuracy is needed to improve the inversion efficiency. We developed a new three-band spectral vegetation index (NEW-NDRE) by combining a two-band spectral index NDRE and the spectral reflectance at 550 nm based on field data collected from different sites, years, with different varieties and nitrogen levels and at multiple growth stages. The NEW-NDRE was compared with 11 traditional spectral vegetation indices in terms of wheat canopy nitrogen content inversion. NEW-NDRE and three traditional indices (NDRE, NDDA and RI-1dB) all closely correlated with wheat canopy nitrogen content. NEW-NDRE displayed the highest correlation with wheat canopy nitrogen content at early grain filling stage, with a coefficient (R2) of 0.9 and a root mean squared error (RMSE) of 0.4. The inversion model developed with the NEW-NDRE was validated with an independent dataset. The relative error (RE) of the model was 9.3%, which was significantly lower than that of NDRE, NDDA and RI-1dB. Generally, NEW-NDRE is a more robust index for wheat canopy nitrogen content inversion than traditional indices through eliminating environmental limitation, and it could be used as a new tool for precise fertilizer application.