|
|
Le modèle a donné de bonnes prédictions pour le pH, la CE et les macroéléments. Les prédictions étaient même excellentes pour tous les macroéléments (R2 > 0,91, RPD > 3,19) sauf le Na+ (R2 = 0,81, RPD = 2,26).
|
|
|
The increasing popularity of hydroponic culture calls for rapid, inexpensive, environmental-friendly, and simultaneous assessment of multiple elements in large numbers of solution samples. The objective of this study was to assess the potential of Fourier transform transmission spectra in the mid-infrared range for determination of solution nutrients in hydroponic samples for greenhouse production systems. Partial least squares regression models were developed using 90 greenhouse solution samples for the prediction of pH, EC, and the concentrations of 8 macro- and 6 trace-nutrients. Several properties in the hydroponic solutions were closely related to each other. Good calibration models were obtained for pH, EC, and macroelements (determination coefficient, R2 = 0.80-0.97, and residual prediction deviation, RPD = 2.19-9.63). Among them, the calibration models for Mg2+, Ca2+, NO3-, and P (PO43-, HPO42-, H2PO4-) were excellent with R2 = 0.91-0.97 and RPD = 3.16-9.63. The model predictions were good for pH, EC, and macroelements, and predictions were excellent for the macroelements (R2 > 0.91, RPD > 3.19) except for Na+ (R2 = 0.81, RPD = 2.26). Model calibration was good for trace-elements with R2 = 0.76-0.85 and RPD = 2.05-4.48. The model predictions for Cu2+, Zn2+, Fe (Fe2+, Fe3+) and B (BO32-, B4O72-) were good with R2 > 0.84 and RPD > 2.52, while predictions for MoO42- and Mn2+ (extremely low concentration elements) were classified as reasonable. Although the reasons for the successful predictions of the metallic elements were not clear, our results confirmed that metal nutrients could be quantified using mid-infrared spectroscopic technique. There is a great opportunity to determine and monitor real-time nutrient solutions in hydroponic systems using mid-infrared transmission spectra.
|