A bayesian network model for yellow rust forecasting in winter wheat

Abstract

Yellow rust (YR) is one of the most destructive diseases of wheat. We introduced the Bayesian network analysis as a core method and develop a large-scale YR forecasting model based on several important meteorological variables that associate with disease occurrence. To guarantee an effective model calibration and validation, we used multiple years (2010–2012) of meteorological data and the ground survey data in Gansu Province where the YR intimidated most severely in China. The validation results showed that the disease forecasting model is able to produce a reasonable risk map to indicate the disease pressure across the region. In addition, the temporal dispersal of YR can also be delineated by the model. Through a comparison with some classic methods, the Bayesian network outperformed BP neutral network and FLDA in accuracy, which thereby suggested a great potential of Bayesian network in disease forecasting at a regional scale.