The effect of water and variety on nitrogen sensing of citrus leaf

Two citrus leaf sample sets of Valencia and Hamlin varieties years were collected in 2002 and 2003 and used to study relationships between spectral reflectance and nitrogen concentration of the samples. Correlation coefficient spectrum, partial least squares (PLS) regression, stepwise multiple linear regression (SMLR) were used to study water content and variety effects on nitrogen sensing. It was found that using wet basis nitrogen concentration of fresh leaf instead of dry basis nitrogen concentration of dried leaf based on weight could produce more stable and better calibration model for nitrogen prediction. The SMLR model for the 2002 year data based on nitrogen concentration of wet samples has R square of 0.906 for validation dataset where only 5 wavelengths were used. There is no strong evidence to indicate that different calibration models need to be developed for different citrus variety to predict nitrogen concentration of unknown samples.