34315

Evaluation of fertilizer coating efficiency by NIRS and NIR hyperspectral imaging

Bernard LECLER 1 ; Philippe Vermeulen 1 ; Olivier MINET 1 ; Pierre Dardenne 1 ; Vincent Baeten 2 ; Juan Antonio Fernandez-Pierna 1
Favorite this paper

Introduction
Nowadays private industry launches new alternative forms of fertilizer on the market. One of them consists to apply a coating with humic and fulvic acids (HFA) on Diammonium Phosphate (DAP) fertilizer using a sticker. Rapid and non-invasive methods are needed to assess the quantity of HFA and the homogeneity of the coating. In this work, NIR spectroscopy and NIR hyperspectral imaging have been assessed for this purpose.
Experimental design
To measure the HFA concentration, a classical NIR calibration was built using 10 samples ground with a known HFA concentration. Samples covering the range 1.5 to 2.4 % and wetted with sticker were prepared in the laboratory.
To assess the homogeneity of the coating, 8 samples of DAP coated with sticker and 1, 2, 4, 6, 8 or 10% of HFA were prepared and measured with a NIR hyperspectral line scan imaging system combined with a conveyor belt (BurgerMetrics).
Results and discussion
The NIR calibration based on a MLR algorithm was validated with 6 independent samples covering the range 1.65 to 2.25% in HFA by showing a RMSEP value quite similar to the SEC and a RPD closed to 5. This method requires the grinding of the fertilizer.
To assess the efficiency and the homogeneity of the coating, the percentage of sticker and HFA had to be quantified particle by particle. All images consisted of lines of 320 pixels acquired according to the conveyor belt speed. Partial Least Squares Discriminant Analysis (PLSDA) was used as classification method for the construction of the discrimination model. It was applied to all the individual pixels in the images of the fertilizer samples in order to isolate and quantify the number of pixels detected as pure fertilizer (DAP), sticker and organic matter (HFA).
Conclusion
This study shows the potential of vibrational spectroscopy (classical NIR and NIR hyperspectral imaging) to propose solutions for the study of the coating efficiency of fertilizer. The quantity of organic matter coated on the fertilizer and the homogeneity of the coating can be assessed by this methodology.