Non-Destructive Evaluation of the Leaf Nitrogen Concentration by In-Field Visible/Near-Infrared Spectroscopy in Pear Orchards.

Abstract

Non-destructive and timely determination of leaf nitrogen (N) concentration is urgently needed for N management in pear orchards. A two-year field experiment was conducted in a commercial pear orchard with five N application rates: 0 (N0), 165 (N1), 330 (N2), 660 (N3), and 990 (N4) kg·N·ha -1 . The mid-portion leaves on the year's shoot were selected for the spectral measurement first and then N concentration determination in the laboratory at 50 and 80 days after full bloom (DAB). Three methods of in-field spectral measurement (25° bare fibre under solar conditions, black background attached to plant probe, and white background attached to plant probe) were compared. We also investigated the modelling performances of four chemometric techniques (principal components regression, PCR; partial least squares regression, PLSR; stepwise multiple linear regression, SMLR; and back propagation neural network, BPNN) and three vegetation indices (difference spectral index, normalized difference spectral index, and ratio spectral index). Due to the low correlation of reflectance obtained by the 25° field of view method, all of the modelling was performed on two spectral datasets-both acquired by a plant probe. Results showed that the best modelling and prediction accuracy were found in the model established by PLSR and spectra measured with a black background. The randomly-separated subsets of calibration ( n = 1000) and validation ( n = 420) of this model resulted in high R² values of 0.86 and 0.85, respectively, as well as a low mean relative error (<6%). Furthermore, a higher coefficient of determination between the leaf N concentration and fruit yield was found at 50 DAB samplings in both 2015 (R² = 0.77) and 2014 (R² = 0.59). Thus, the leaf N concentration was suggested to be determined at 50 DAB by visible/near-infrared spectroscopy and the threshold should be 24-27 g/kg.

Plant phenotyping relevance

ナシ葉の窒素濃度という植物形質を、圃場可視/近赤外分光で非破壊推定する測定法の比較・モデル検証が中心であり、植物フェノタイピング手法に該当する。

Code and data availability

The article describes in-field hyperspectral measurements of pear leaves and PLSR/chemometric modeling of leaf nitrogen concentration, but contains no data availability statement, no public deposit of spectra or N-concentration datasets, no author code/scripts, and no trained model checkpoints. Software mentioned (Unsc

No evidence-backed public reproduction asset is currently recorded.