Potential of field hyperspectral imaging as a non destructive method to assess leaf nitrogen content in Wheat

Nitrogen is the most important crop limiting factor, thus plant nitrogen status during plant cycle is a key parameter for crop monitoring. Many new techniques, based on leaf optical properties have been proposed for a non-destructive diagnosis to replace Nitrogen Nutrition Index which is a costly and destructive method. We intend here to study leaf nitrogen concentration accessibility from reflectance (400-1000 nm) spectra of whole plants from a field hyperspectral imaging set-up including difficulties related to variable solar lighting and potential specular reflexion. Firstly, we calibrated a chemo-metrical model between leaf nitrogen concentration and reflectance spectra of flat leaves (R(2)=0.903, SEP = 0.327%DM), which validated the sensor and our reflectance correction process. As a second step, we calibrated a chemometrical model between nitrogen concentration and reflectance spectra of individual leaves from isolated plants grown in pots in greenhouse (R(2) = 0.889, SEP = 0.481%DM) or under field conditions (R(2) = 0.881, SEP = 0.366%DM). Pooling the two datasets provided us a relevant model to predict leaf nitrogen content for the two culture conditions (R(2) = 0.875, SEP = 0.496% DM) suggesting that this technique is promising to assess nitrogen plant parameters with a non destructive method. This tool could be used to follow-up plant nitrogen dynamics criteria or to generate nitrogen spatial cartographies. (C) 2011 Elsevier B.V. All rights reserved.