Visualisation techniques for modelling environmental influences in viticulture

Abstract

The quality of wine is in its making but significantly also depends on the grape ripening conditions, such as environmental, weather and growing factors, as well as grape varietal characteristics. Some centuries-old Mediterranean grape growing tradition has helped the winemaking industry pool resources in an effort to gain further understanding of the interrelationships between genetic (or "cultivars") and environmental factors, such as climate and soil ( or "terrior"), The industry seems to be successful in using historic, in some cases, years old data to improve cultivation practices or "vintages" with an ultimate aim of producing finer wine. This paper outlines past research on enhancing data depiction methodologies generally used to represent environmental influence factors in viticulture along with some recent approaches, with special emphasis on data clustering techniques applied to analysing multivariate data sets using standard pixel representations and also some dependency relationship visualisation methodologies based on self-organising map (SOM) techniques. Literature reviewed reveals the limited research in the visualisation of environmental data in viticulture. The paper then elaborates upon a system design with some approaches being investigated for the implementation of novel visualisation techniques that could facilitate knowledge extraction from data on environmental influence factors in viticulture. The two main sets of approaches being referred to as neural net (NN) and non NN. The NN approach encompasses various visualisation techniques that utilises computational neural networks, such as SOM. Those in the non NN category are novel ways that are in complimentary to those in the NN with capabilities to overcome the limitations of the former, by creating a synergetic effect. An example of a possible non NN visualisation could be MIR-max and other algorithmic techniques. The system proposed herein attempts to integrate the two broad techniques into one system to enhance the visualization of environmental influence factors in viticulture especially for use in assessing any environmental impact with novel data visualisation techniques for decision making and in determining "what makes a good year for wine". The system would facilitate visualisation techniques over a distributed system, such as the Internet, and provide functions with user interface useful to both experts and novices in viticulture.