FT-NIR Spectroscopic Analysis of Nitrogen in Cotton Leaves

Abstract

Near-infrared spectroscopy was evaluated as a means to quantify the nitrogen content in fresh cotton leaves (Gossypium hirsutum L. var. Delta Pine 90) subjected to a factorial design experiment of varying nitrogen and water applications. Absorbance spectra were collected in the 10 000–4000 cm⁻¹ (1000–2500 nm) region from fresh cotton leaves over a two month portion of the growing season. Total nitrogen content was quantified by a wet chemistry Kjeldahl method for validation purposes. Partial least-squares regression analysis, using an automated grid search method, selected the spectral region 6041 to 5651 cm⁻¹ (1650–1770 nm) for analysis based on having the lowest standard error of prediction of total nitrogen content. This region includes protein spectral features. Nitrogen predictions resulted in a correlation coefficient of 0.83, and a standard error of prediction of 0.29% for nitrogen levels ranging from 3.1 to 5.2% total nitrogen. This approach has promise for providing rapid plant chemical analyses for cotton crop fertilization management purposes.