Predictive model for stored rice qualities using near-infrared spectroscopy

Abstract

The objectives of this research are to study aging of high and low amylose rice during storage at different temperatures and explain the changes by kinetic model and to develop the predictive models between rice properties and NIR spectra in the diffuse-reflectance mode during storage at different temperatures. The samples used in this research included three varieties of low amylose rice, i.e. Khao Dawk Mali 105, Pathumthani 1 and Rice Department (RD) 45 and three varieties of high amylose rice, i.e. RD 47, Chai Nat 1 and Phitsanulok 2. Paddy rice in plastic woven sacks was stored at ambient temperature (30 ºC±2 ºC) and chilled temperature (8 ºC ±2 ºC) for 9 and 18 months, respectively. The qualities determined during storage were paddy property, milling quality, cooking qualities, cooked rice properties and rice flour properties. Some qualities of all six rice varieties; namely fat content, fiber content, ash content, carbohydrate content, enthalpy of gelatinization of rice flour, whiteness index, springiness and cohesiveness of cooked rice, varied in a narrow range. However, low amylose rice varieties had lower protein content, amylose content, pasting temperature, setback, melting enthalpy of amylose/lipid complex of rice flour, head rice yield, solid loss, hardness, gumminess and chewiness of cooked rice but higher water uptake, peak viscosity, breakdown, onset temperature, peak temperature, conclusion temperature and gelatinization temperature range of rice flour as well as adhesiveness of cooked rice. Aging led to an increase in head rice yield, minimum cooking time, water uptake, volume expansion ratio, hardness, cohesiveness, springiness, gumminess and chewiness of cooked rice and pasting temperature of rice flour and a decrease in whiteness index of rice grain, solid loss, peak viscosity and breakdown of rice flour and adhesiveness of cooked rice. Thermal properties and MW distribution pattern of rice flour protein slightly changed during storage. The rate of changes for all rice varieties stored at 8 ºC was lower than that at 30 ºC. Principal Component Analysis (PCA) of all observed variables classified the samples into three groups; low amylose rice (Khao Dawk Mali 105, Pathumthani 1 and RD 45, high amylose rice (RD 47 and Phitsanulok 2), and high amylose Chai Nat 1. The predictive models between rice properties and NIR spectra in the diffuse-reflectance mode were produced using partial least square (PLS) regression. Only 14 parameters, i.e. head rice yield, minimum cooking time, solid loss, water uptake, volume expansion ratio, pasting temperature, peak viscosity, breakdown and setback of rice flour, hardness and adhesiveness of cooked rice, onset temperature, peak temperature and conclusion temperature of rice flour, could be used to develop good prediction models (R2 > 0.7). The first-order fractional conversion kinetic model reasonably explained the changes of nine variables, i.e. solid loss, water uptake and volume expansion ratio of cooked rice, pasting temperature, peak viscosity, breakdown and setback of rice flour, hardness and adhesiveness of cooked rice, during aging at 30 oC (R2 ≥ 0.7). Changes in rice qualities during aging at 8 oC were marginal, thus could not be explained by kinetic models.