DETERMINATION OF SIGNIFICANT WAVELENGTHS FOR NITROGEN PREDICTION IN CITRUS LEAVES USING STEPWISE MULTIPLE LINEAR REGRESSION

Hyperspectral data measurements are a non-destructive and lesser time-consuming method for the determination of leaf nitrogen (N) concentration, which is highly needed for the appropriate management of citrus crops. We measured the spectrum of 32 different orange trees using a field spectroradiometer and calculate their reflectance and, later, absorbance levels. We proceeded to assemble these 32 spectra data, and determined the correlation between the N content in leaves and wavelengths measured values using the Stepwise Multiple Linear Regression (SMLR) method. As results, the blue (423 nm) and red (669 nm) wavelengths are better correlated with the N concentration on citrus leaves, with an R2 equal to 0.9608 and 0.9663, respectively.