Development and evaluation of a forecasting system for fungal disease in turfgrass

Abstract

A forecasting system for fungal infection of turfgrass using weather-based empirical indices (the ‘Fidanza’ and ‘Schumann’ models) was developed and evaluated for its ability to predict the occurrence of brown patch (Rhizoctonia blight) infection episodes at an experimental site in southeastern USA. Disease observations took place at the Turfgrass Field Laboratory in Raleigh, North Carolina between 8 June and 17 August 2003. Three meteorological data sources were used to generate disease risk indices using the empirical models: an on-site observing station, an observing station at a nearby airport, and the US National Weather Service's operational Eta weather forecast model. Visual observations of brown patch activity were conducted in the field and used to evaluate the accuracy of the disease prediction models. Results indicate that the Fidanza and Schumann models correctly predicted brown patch activity on 48% and 30% of the days on which disease occurred, respectively. A diagnosis of the model performance of these disease indices was undertaken. Results are dependent on occurrence of high temperatures and rainfall and independent of the source of the meteorological information (on-site, airport and the Eta model); therefore, regional meteorological information can be effectively applied to develop turfgrass disease forecasting systems. Ongoing efforts are directed towards developing new disease indices and modifying existing indices before an operational disease forecasting system can be implemented.