LATE BLIGHT PREDICTION AT THE GROWER LEVEL
University of Maine, Presque Island, USA
Background and objectives
Personal computers can easily handle the data analyses required for late blight prediction. Good quality automated weather stations are available and reasonably priced. Automated weather stations connected to grower-owned and -operated computers can do a credible job of weather data collection. The final need is for computer software to analyse the weather data collected for late blight prediction. The computer program NoBlight was developed specifically for this purpose. Severity values  and spray interval recommendations are determined using current on-site weather data.
The initial network consisted of seven weather stations covering 25,000 ha of potatoes grown on nearly 475 farms. The weather stations and the late bight prediction software were on site and operated by the growers. Twice weekly during the growing season, information was electronically sent to a central location. The information was interpreted and put onto a Late Blight Hotline. This was a voice-mail system operating on a toll-free telephone line available on a 24-hour basis.
Results and conclusions
The interpretation of the weather data and the skill of the forecaster are still critical factors in late blight prediction. With practice, growers can interpret weather data and become better forecasters. The more information available to the grower, the better the decision becomes. A decentralized weather network and late blight prediction network actively involves the producers in their own decisions. On-site late blight prediction by the growers is proving a successful approach and is increasing in Maine.