Human visual perception of haze in white wine and model solutions and relationships with instrumental turbidity and imaging models

Abstract

White wines are liable to develop haze during storage and transport due to residual protein in the finished wine. Since haze in white wines is considered unacceptable, winemakers attempt to reduce the protein content prior to bottling through the use of fining agents such as bentonite. The use of bentonite is not without economic cost with the estimated annual impact to the New Zealand wine industry in the region of NZD 35 million. The optimum quantity of bentonite to be used is normally determined using a heat test in conjunction with empirical fining trials. During these tests the winemaker is required to conduct various haze comparisons. Since these assessments are difficult to perform visually, researchers recommend the use of suitable nephelometric instrumentation to ensure consistency and precision. Despite these recommendations, most winemakers have not purchased the necessary equipment; it is thought that cost is the primary deterrent. The overall goal of this project was to explore the relationships between nephelometry, digital imaging and human perception of haze in terms of thresholds and intensity. In addition, the feasibility of using digital imaging devices to make nephelometric measurements was to be investigated. A range of achromatic and yellow turbid suspensions were prepared by suspending synthetic polymer microspheres of diameter 0.25 μm in aqueous solutions. Base wine stocks exhibiting high turbidity were prepared by heating commercial grade Chardonnay wine at 90˚ for two hours. A range of wine samples of varying turbidities were then prepared by dilution with untreated wine. Polymer microsphere suspensions and wine samples of varying visual haze were used in a series of sensory and instrumentation experiments. The sensory experiments examined haze detection thresholds and intensity scaling in human subjects. All evaluations took place in purpose designed sensory booths utilising overhead light emitting diode (LED) illumination. The booth backgrounds and sides were modified to present dark or light surrounds by hanging black or white linen on the walls. Samples were presented to human subjects in standard ISO tasting glasses and the subjects were permitted to manipulate the glass freely whilst they evaluated the samples. Turbidity measurements were made with a Hach 2100P nephelometric device. Image data was captured using a modified Canon A2300 digital still camera using 90˚ viewing geometries. Haze threshold determinations were made for achromatic and yellow microsphere suspensions using the Ascending Method of Limits (AML) and 3-Alternate Forced Choice (3AFC) methods. The lowest thresholds were found when achromatic suspensions were presented in a dark surround where individual thresholds ranged from 0.21 Nephelometric Turbidity Units (NTU) to 1.07 NTU. The highest thresholds were found when yellow suspensions were presented in a light surround where individual thresholds ranged from 1.89 NTU to 25.16 NTU. Heat treated Chardonnay wine samples were also evaluated in a dark surround where individual thresholds ranged from 0.52 NTU to 1.39 NTU. Visual haze intensities in achromatic and yellow microsphere suspensions were assessed by human subjects in dark and light surrounds using modulus Magnitude Estimation (ME). Sample luminance was determined by digitally photographing the samples under the same conditions employed for the ME evaluations. The perceived haze intensity with respect to luminance was found to follow generalised power or logarithmic functions similar to psycho-physical models commonly proposed for the response of the human visual system to brightness and lightness. The data exhibited a bi-segment nature indicative of surround and/or planar contrast induction. Similar responses were found for heat treated Chardonnay wine samples using turbidity as the independent variable. Turbidity values were measured for a range achromatic and yellow microsphere suspension samples. Luminance values were derived from digital images taken of ISO tasting glasses containing the same microsphere suspensions. The turbidity and luminance values were found to be linearly related below 50 NTU (dark surround, R2=0. 9978; light surround, R2=0. 9813). A subsequent experiment examined turbidity values and luminance data in achromatic and yellow microsphere suspension samples over a low range (< 6 NTU) by imaging the sample surface directly rather than through the glass receptacle. Turbidity vales and luminance data were again found to be linearly related (R2=0.9628). Measurements from the Hach 2100P device and a luminance based measurement model were found to be mean and median equivalent for the low range experimental data.