Identifying Ideal Wine Growing Areas in New Zealand
Predictive Modelling Approach
The aim of the projects was to generate predictive models to identify areas that are ideal for growing good wine producing grapes using existing vineyards as a guide. These models were generated using our spatial data modelling techniques.
The modelling techniques were used to prioritise potential locations for the establishment of new vineyards in New Zealand and to prioritise grape varieties for specific sites. Other industries can also benefit from the techniques such as fruit growers (e.g. apples, cherries, kiwifruit), and even growers of exotic produce (tea, coffee, spices).
The weights of evidence technique has been used in our studies to integrate and interpret available data for New Zealand in general and the Hawke’s Bay Region specifically, to identify optimum places to grow grapes for the production of wine.
The New Zealand Wine Model
The New Zealand winegrowing industry is spread over several regions throughout the country. Over 40 % of the national grapevine-covered area is in the Marlborough region, with 24 % in Hawke's Bay and 14 % in the Gisborne region. Other grape producing areas, including the Wairarapa, Nelson, Canterbury and Otago regions, together represent 13 % of the total New Zealand wine-growing area.
The New Zealand winegrowing industry has changed significantly over the past decade. Both the total grapevine area and the wine producing area have increased dramatically as the export of New Zealand wine has increased by 20 % per annum for the last two decades.
We have developed a first-pass scoping study to identify potential new sites for growing grapes. A number of features important for grape growth and the success of a vinyard have been identified and tested in this project, namely: climatic factors, soil properties, location features and social factors.
The Hawke's Bay Wine Model
The aim of the Hawkes Bay Wine Model was to generate a predictive model that identified optimal areas to grow grapes for the production of wine in one of the regions identified by the larger-scale New Zealand Model as ideal.
The Hawke’s Bay Wine Model was a pilot study which used only the coarse scale data that was available at the time. The weights of evidence modelling method requires a set of training data to test the correlation of the predictive maps generated. A selection of vineyards already established in the Hawke’s Bay region was used for this purpose. The study area for the model covered Central Hawke’s Bay, Hastings and Napier City; a large area that contains both excellent and poor grape growing areas.
For each model a suitability map was developed using the weights of evidence technique. This included the themes that had the best regional coverage and a good spatial correlation with the known award-winning vineyards. The suitability maps shown here clearly highlight the most favourable wine growing areas in New Zealand (above) and in the Hawkes Bay region (below); the highly suitable areas for grape growing are shown in red. The models were successful at the scale of the data used, as indicated by their finding other vineyards that were not included as training data.