Data Mining Techniques for Modeling the Influence of Weather Conditions

Influences of daily weather extremes, such as maximum/ minimum temperatures, humidity, and precipitation, are observable in perennial crop phonology that in turn determines the annual crop yield in quality and quantity. In viticulture, grapevine phonology determines the quality of vintage produced from the grapes apart from the best effects by winemaker. Following a brief review of current literature in this research domain, the paper describes a data mining approach being developed to data association modeling to depict dependency relationships between daily weather extremes, grapevine phonology and yield indicators using data from a vineyard in northern New Zealand and daily weather extremes logged at a nearby meteorology station. An artificial neural network algorithm was used to classify the data associations and the chisquare test was used to establish the degree of dependence between the related variable values. The initial results of the approach to daily maximum weather conditions show potential. Keywords-Artificial neural network algorithm was,