Maqbool, R., Percival, D. C., Adl, M. S., Zaman, Q. U. and Buszard, D. 2012. In situ estimation of foliar nitrogen in wild blueberry using reflectance spectra. Can. J. Plant Sci. 92: 1155-1161. Remote sensing techniques have the potential to serve as an important nutrient management tool in wild blueberry. The potential of visible (VIS), near infrared (NIR) and shortwave infrared (SWIR) spectroscopy was evaluated during 2006 (sprout/vegetative phase of production) to estimate foliar nitrogen (N). Canopy reflectance measurements were taken from two nutrient management experimental sites located in Nova Scotia (NS) and New Brunswick (NB). Partial least squares regression (PLSR) estimated foliar N, giving the coefficients of determination (R2) values ranging from 0.69 to 0.85, and root mean square errors of cross validation (RMSECV) from 0.16% (±8.29% of mean) to 0.24% (±12.43% of mean) for different spectral ranges used in this study. The green peak region located in the VIS region best estimated foliar N. The tested spectral ranges differed in their predictive ability, but generally followed the biochemical basis. Variable importance in projection scores (VIP), regression vector coefficients and PLSR loading weights (LWs) plots highlight the importance of wavebands (∼550 nm, ∼610 nm, 1510 nm, ∼1690 nm, ∼1730 nm, ∼1980 nm and ∼2030 nm) for in situ foliar N estimations. Thus, it was concluded that reflectance spectra may be used to estimate and ultimately map foliar N in wild blueberry production. The results illustrated the ability of multivariate techniques, such as PLSR to explore hyperspectral data and estimate leaf tissue nutrient content.
partial least squares regression
régression des moindres carrés partiels
Spectre de réflectance