Pendugaan komposisi kimia biji nyamplung (Calophyllum inophyllum L.) secara non-destruktif dengan metode Near Infrared (NIR)

Abstract

The objective of this study was to apply NIR method to predict chemical composition of Calophyllum inophyllum L. seeds accurately (moisture, fat, acid number, free fatty acid contents). The reflectance of seeds at the wavelength ranges from 1000 to 2500 nm (4000-10000 cm-1 at intervals of 4 cm-1) were measured by NIRFlex Solids Petri Apparatus. Calibration method which used in this study are principal component regression (PCR) and partial least squares (PLS). Data treatment on the reflectance and absorbance spectrum are used that is: smooth average 3 points, second derivative Savitzky-Golay 9 points, normalization 0-1, combination both of smooth average 3 points and second derivative Savitzky-Golay 9 points, and combination of all data treatment. A number of 70 Calophyllum inophyllum L. seeds were used as samples. Samples were divided into two parts: ± 45 samples (2/3 of total samples) for developing calibration equation and ± 25 samples (1/3 of total samples) for performing validation. NIR data analysis shows that PLS method with NIR reflectance data and the combination both of smooth average 3 points and second derivative Savitzky-Golay 9 points is the best method of calibration and data treatment to predicting moisture contents of Calophyllum inophyllum L. seeds was also prediction with standard error of prediction (SEP) of 0.45% and coefficient of variability (CV) of 0.81%. Prediction of fat contents of Calophyllum inophyllum L. seeds best obtained with the PLS method, the reflectance data, and the combination both of smooth average 3 points and second derivative Savitzky-Golay 9 points was also prediction with standard error of prediction (SEP) of 0.37% and coefficient of variability (CV) of 0.62%. Prediction of acid number contents of Calophyllum inophyllum L. seeds best obtained with the PLS method, the absorbance data, and the second derivative Savitzky-Golay 9 points data treatment was also prediction with standard error of prediction (SEP) of 0.04% and coefficient of variability (CV) of 0.09%. And prediction of free fatty acid contents of Calophyllum inophyllum L. seeds best obtained with the PLS method, the reflectance data, and the combination was all data treatments was also prediction with standard error of prediction (SEP) of 0.04% and coefficient of variability (CV) of 0.18%.