Concept: To provide scientifically based information for use by grape growers and wine producers in their ongoing endeavour to improve crop production, product quality and yield through the synthesis of environmental and human sensory perception data.

The research relates to grape growing and wine production. This research is intended to be a long-term project observing climate, atmosphere, plant and soil data in vineyards, combined with data concerning wine quality gathered and analysed using a both expert and casual opinions. The methods used in the research are contemporary mathematical and software engineering methods and include novel techniques for how to select the ‘rough data’, how best to classify and model ‘fuzzy data’ and how best to synthesise precise or exact (quantitative) data, such as that from climatic and environmental attributes with other imprecise or inexact (qualitative) values. These qualitative values can be considered as fuzzy data which includes human sensory responses that form personal opinions based on smell, taste and other associated variables that form a spectrum of qualitative values. In the domain of wine tasting, these sensory variables are generally referred to as the ‘structure’ and ‘complexity’ of the finished product.

Arising from the proposition that wine is generally regarded as being of better quality in some production years than others, the mathematics and software engineering methods mentioned above are combined in this research with data processing techniques from the fields of spatial information processing, environmental and bio-ecological modelling, to develop a system that integrates all available data and synthesises them to produce scenarios that can be used to modify growing and production methods in order to achieve the best possible wine in any given year.

Land-Satellite imagery will be used to provide a sequence of terrain maps showing changing weather patterns over time. Superimposed on these will be environmental data collected in real time from sample locations in Chile, Japan, Uruguay, Argentina, the USA and New Zealand to illustrate the atmospheric variability of the individual regions, which will provide a visual comparison in addition to the numeric correlations resulting from analysis of the data collected by telemetry devices feeding in real time to the GIS database, which can be interrogated for numerous cross-correlation purposes.

In some cases the data received will be in fuzzy form. Using contemporary neural network modelling software, these results will be collated and depicted in such a manner that scenarios for considering the ‘best year’ proposition can be appreciated and in conjunction with other factors, lead to the modification and optimisation of both growing and production methods. These neural networks will be analysed statistically to extract knowledge as to which variables have most effect on desired outcomes.

Our industry partners include:

• AUT Radio Telescope, Warkworth, NZ
• Awarua Vineyard, Hawkes Bay, NZ
• Cable Bay Winery, Waiheke Island, NZ
• Kumeu River Winery, Kumeu, NZ
• Mahurangi River Winery, Warkworth, NZ
• Casa Donoso Winery, Valle del Maule, Chile
• Campus Los Niches, UCM, Chile
• Campus San Miguel, UCM, Chile
• Forestal El Colorado, UCM, Chile
• Forestal, Costa Azul, UCM, Chile
• Fundo Hugo Casanova Winery, Chile
• Fundo Santa Elisa, Chile
• Los Reyunos, UTN, Argentina
• San Rafael, UTN, Argentina
• Viña Casa Bianchi, Argentina
• La Agricola Jackson Montevideo, Uruguay
Ajimu Vineyard, Beppu, Japan
• Ajimu Winery, Beppu, Japan
• Fallbrook Winery, South California, USA

The figure below illustrates the principal components, their interactions and data flow through the concept intended for prototype implementation.

Please see some sample research results below:

Sallis, P., Jarur, M., and Trujillo, M.(2009). Frost prediction characteristics and classification using computational neural networks. In Australian Journal of Intelligent Information Processing Systems (AJIIPS) volume 10.1, 2008 (ISSN 1321-2133) pp50-58. Also published in M. Kppen et al. (Eds.): ICONIP 2008, Part I, LNCS 5506, 2009. Springer-Verlag Berlin Heidelberg 2009. pp. 1211-1220: p1213.

Sallis, P., Jarur, M., Trujillo, M., and Ghobakhlou, A. (2009). Frost prediction using a combinational model of supervised and unsupervised neural networks for crop management in vineyards. In B. Anderssen et al. (eds) /18th IMACS World Congress - MODSIM09 International Congress on Modelling and Simulation/, 13-17 July 2009, Cairns, Australia. ISBN: 978-0-9758400-7 -8. pp. 789-795: p791

Sallis, P., Jarur, M., Trujillo, M., and Ghobakhlou, A. (2009). Frost prediction using a combinational model of supervised and unsupervised neural networks for crop management in vineyards. In B. Anderssen et al. (eds) /18th IMACS World Congress - MODSIM09 International Congress on Modelling and Simulation/, 13-17 July 2009, Cairns, Australia. ISBN: 978-0-9758400-7 -8. pp. 789-795: p792&3.

http://lipas.uwasa.fi/stes/step96/step96/lagus/

http://www.artisanvineyards.com/About/VintageChart.aspx
http://www.landware.com/wineguide/ppc/tour/images/Vintages.gif

Shanmuganathan, S.,Sallis, P., and Narayanan, A. (2009). Unsupervised artificial neural nets for modelling the effects of climate change on New Zealand grape wines. In B. Anderssen et al. (eds) /18th IMACS World Congress - MODSIM09 International Congress on Modelling and Simulation/, 13-17 July 2009, Cairns, Australia. ISBN: 978-0-9758400-7-8. pp. 803-809: p807