Prediction of N, K and Mg on Leaves of Four Citrus Rootstocks Seedlings Using Hyper Spectral Data

Abstract

Sufficient application of nutrients is one of the most important factors in the development of seedlings from quality citrus root stocks. Some of the key guides for preparing citrus fertilizer programs is by tracking the nutrient content of plants directly. This includes, however, an examination of a large number of leaf samples using costly and time-chemical techniques. It has been shown over the last 10 years that it is possible to quantitatively estimate such nutrient elements in citrus leaves using the spectral reflectance values obtained using hyperspectral spectroscopy. This technique is quick, non-destructive, cost-effective and eco- friendly. Therefore, estimating nitrogen, potassium, and magnesium in seedling leaves of citrus rootstocks by this approach would be useful in determining the seedlings ' mineral status. In this research,168 leaf samples from four citrus rootstocks seedlings (Volkamer lemon, Sour orange, Trifoliate orange and Balady Lime) were used to conduct three vegetation indices; normalized difference vegetation index (NDVI), normalized difference nitrogen index (NDNI) and modified chlorophyll absorption ratio index (MCARI) and subsequent nutrient estimates for N, K, and Mg concentration. Simple regression models and chemical analysis were used to produce the best model of estimation to predict the values of the three components. A high correlation coefficient (R2) was verified in the estimate of N (R2=0.982) with the lowest root mean square error (RMSE=0.0472) and k (R2=0.983) with the lowest root mean square error (RMSE=0.0491) and the lowest root mean square error (RMSE=0.0062) was also verified.