Methods for estimating leaf nitrogen concentration of winter oilseed rape (Brassica napus L.) using in situ leaf spectroscopy

⿢Four spectral methods for estimating leaf nitrogen concentration were developed.⿢A partial least square model was used to evaluate the performance of these methods.⿢The first derivative reflectance was found the most accurate method.⿢Effective wavelengths were identified to reduce dimensionality in spectral analysis.

Accurate and nondestructive assessment of leaf nitrogen (N) nutritional status is important for site-specific N management in winter oilseed rape production. To develop a method for determining leaf N concentration (LNC) in oilseed rape, a field experiment with different N fertilizer levels was conducted in two successive years by measuring leaf spectral reflectance (400⿿1300 nm) and LNC at varying developmental stages. A partial least square (PLS) regression analysis was performed with four spectral methods: (i) the raw spectral reflectance (R), (ii) inverse-log reflectance data (log(1/R)), (iii) continuum removal (CR) method and (iv) first derivative reflectance (FDR). The results indicated that LNC and leaf reflectance significantly varied with the levels of N fertilization, and a good correlation was observed for all the spectral methods. Using a calibration dataset, the best results were obtained with the FDR-PLS method, which yielded the highest coefficient of determination (r2cal) of 0.963, the ratio prediction to deviation (RPD cal) of 5.207, and the lowest root mean square error (RMSE cal) of 0.294. Tests with the independent validation dataset also showed that the FDR-PLS method could well predict LNC in oilseed rape, with the values of r2val, RPD val, and RMSE val being 0.966, 5.488 and 0.276, respectively. The variable importance in projection (VIP) scores resulting from this PLS regression analysis were used to determine the effective wavelengths and reduce the dimensionality of the spectral reflectance data. The newly-developed FDR-PLS model using the effective wavelengths (432, 467, 519, 614, 772, 912 and 1072 nm) performed well in LNC prediction with r2val = 0.884, RPD val = 2.971 and RMSE val = 0.508. The overall results indicate that the LNC of winter oilseed rape could be reliably estimated with the in situ developed FDR-PLS method in this study.