A review of the application to emergent subfields in viticulture of local reflectance and interactance spectroscopy combined with soft computing and multivariate analysis

Abstract

Spectroscopic techniques have shown great potential due to their quick response, cost-effective, non-destructive and non-invasive nature, and environmental friendliness. These characteristics make this technology very attractive for sustainable industry and research activities, being viticulture industry no exception. Spectroscopic techniques are an appealing alternative for ripeness assessment and harvest date determination as well as for plant variety and clone determination. Numerous recent works have clearly demonstrated that it is highly advantageous to process the high dimensionality spectroscopic data with soft computing or multivariate analysis techniques such as Partial Least Squares, Neural Networks or Support Vector Machines. In this review, focus will be given to two emergent subfields in viticulture where the combination of spectroscopy and soft computing is fundamental: (1) The difficult measurement of enological parameters, namely sugar content, pH and anthocyanin content, in samples containing a small number of grape berries, with the aim of assessing grapes’ ripeness; (2) The multiclass problem of identifying plant varieties and clones. The results of the various works in these subfields will be presented. The present article starts with a brief description of the spectroscopy principles and continues by making an overview of the scientific literature considering the number of berries per sample and the total number of samples in the various works. The use of different varieties, vintages and harvest locations in the same model will also be addressed. Special attention is given to the validation methods employed and to algorithm comparison. Some suggestions are presented in order to facilitate future comparison of published results.