Prediction of water content in Lintong green bean coffee using FT-NIRS and PLS method

Abstract

Water content is one of the coffee compositions that affects the coffee flavour, texture and appearance. Measurement of chemical composition is usually conducted using the chemical method which is known to be expensive and time consuming. Nowadays, Fourier Transform Near Infrared Spectroscopy (FT-NIRS) method is developed in measuring the chemical content of coffee in a short time. The purpose of this study was to predict the water content of Lintong green bean using FT-NIRS and PLS method. The spectra measurement was conducted in the wavelength of Near Infrared of 1000-2500 nm, then followed by the water content measurement using chemical methods. Several data pretreatment was used, namely first derivative, second derivative, multiple scatter correction (MSC), standard normal variate (SNV), first derivative + MSC and second derivative + MSC. Result showed that the best calibration model was obtained using MSC with 7 factors of PLS, indicated by r = 0.860; SEC = 0.12; SEP = 0.14; CV = 1.14; and RPD = 2.09.