At present there exist only calorimetric techniques to measure the ice fraction, which are far too laborious and time consuming for online applications.  Using NIR, in the wavelength range from 760nm to 1040nm, it is possible to detect shifts that occur in the second order water peak as the ice fraction level increases. By using non contact scanning it is possible to cope with the equilibrium state of the salmon, which is sensitive to external temperature changes when handled.

Norwegian Atlantic Salmon (Salmo salar), originating from a fish farm in mid-Norway, were used for the experiment. The fish was filleted and packed by Lerøy Midnor, Hitra, 5 to 7 days before superchilling and stored on ice. Superchilling was performed in a wind tunnel with an air velocity of 2 m/s, a temperature of -25 °C and at different time intervals in order to reach ice levels of 5, 10, 20 and 30 % in the salmon fillets.  NIR measurements were performed on superchilled samples immediately after using the non-line imaging scanner from QVision. The ice fraction level in the salmon fillets was determined using calorimetric methods. The level of ice in the product was estimated based on the temperature change as it melted and information about the product’s thermal and physical properties. A Partial Least Squares Model was built and with an estimated root mean square error (RMSEP) of the ice fraction model was 2.5%, and a correlation of 0.96.

Shift in water peak due to change in ice fraction level.