Chlorophyll analysis by NIR

In Canada, grains are marketed according to grades. Immature green seed count is a rapid method to assign a grade in canola while estimating chlorophyll level. However, there is a poor linear relationship between the green seed counts and the real chlorophyll contents of the samples. Chlorophyll is a negative factor for canola quality and has to be removed from the oil during the refining stage of the processing. The purpose of the project was to assess if modified near-infrared spectroscopy could be used at grain delivery points to measure chlorophyll content of canola samples, the ultimate goal of the project being to use chlorophyll measurement as a grading factor.
Seven Dickey-john Instalab 600 instruments were modified to include two filters in the visible range 674 and 696 nm and one in the NIR range 2310 nm based on the work of Tkachuk & Kuzina (1982). Five NIR instruments were placed at five industry locations, three primary grain elevators and two port grain terminals. Two other instruments were also used at the Canadian Grain Commission. Railcars containing canola were sub-sampled and analyzed by the Dickey-john at the three primary elevators upon leaving and at two Vancouver terminal elevators upon delivery. Both primary and terminal sub-samples were sent to CGC Winnipeg where they were also analyzed by the two CGC Dickey-john instruments.
Statistical analyses were run to establish repeatability and reproducibility. Results showed that even if there was a statistical difference between the NIR results obtained at the primary and the terminal elevators (paired t-test, Pr < 0.001) the difference was negligible - mean difference 0.68 mg/kg. No statistical difference between the terminal and primary sub-sample results were observed using the two instruments in Winnipeg suggesting that sampling had no effect of the chlorophyll measurement. Some samples were selected and analyzed by the chlorophyll reference method (ISO 10519:1997) to assess the accuracy; the obtained SEPs and RPDs suggested that the models were very good. The results suggested that the tested NIR could be used to accurately measure chlorophyll in canola samples.