Monitoring leaf nitrogen status with hyperspectral reflectance in wheat

The objectives of this study were to determine the relationships of leaf nitrogen concentration on a leaf dry weight basis (LNC) and leaf nitrogen accumulation per unit soil area (LNA) to ground-based canopy hyperspectral reflectance and derivative parameters, and to establish quantitative models for real-time monitoring of leaf N status with key hyperspectral bands and estimation indices in wheat (Triticum aestivum L.). Three field experiments were conducted with different N application rates and wheat cultivars across three growing seasons, and time-course measurements were taken on canopy hyperspectral reflectance, LNC and leaf dry weights under the various treatments. The results showed that LNC and LNA in wheat increased with increasing nitrogen fertilization rates, and changes in canopy hyperspectral reflectance under varied N rates were all highly significant, with consistent patterns across the different cultivars and years. The sensitive spectral bands occurred mostly within visible light and near infrared regions, and a close correlation existed between red-edge district and LNC or LNA. An integrated linear regression equation of LNC to spectral parameters REIPle and λo well described the dynamic pattern of LNC changes in wheat, giving the determination of coefficients (R2) as 0.831 and 0.834, and the standard errors (SE) as 0.405 and 0.403, respectively. The hyperspectral parameters MSS-SARVI and FD742 were linearly related to LNA, with R2 as 0.861 and 0.873, and SE as 1.11 and 1.06, respectively. When independent data were used to test the derived equations, the R2 values between the measured and estimated LNC from spectral parameters REIPle and mND705 were 0.752 and 0.695, with the average relative errors (RE) as 14.4% and 16.5%, respectively. For spectral parameters FD742 and SDr/SDb, the R2 values between the measured and estimated LNA were 0.872 and 0.828, with RE as 14.1% and 15.2%, respectively. The high fit between the measured and estimated values indicated that the present models based on hyperspectral reflectance could be used for reliable estimation of the leaf N status in wheat plant under different growing conditions.