NDSU Agriculture

Small Grain Disease Forecasting Models

Wet period prediction

The length of time that moisture can be found on a plant is a critical measurement for crop disease development but one that is difficult to measure with present technology.  Researchers at NDSU have created models that use a form of artificial intelligence to predict the presence of free moisture in a crop. The moisture prediction is based on a pattern fitting analysis called neural networks. Tests showed that this method outperformed other available models; it correctly predicted moisture occurrence 92% of the time; and, when moisture did occur, the model most often could predict the accumulated length of time within 2 hours.

Note that we are trying to predict wetness on the flag leaf and (when it emerges) the head of small grains.  Longer moisture period typically occur on lower leaves.   Also, other crops with a different plant stature may vary in their length of wet period.

The hours of wetness are used as input for disease prediction by two neural network models, one predicts tan spot and the other Stagonospora (Septoria) blotch.  This is the first time that neural network models have been applied in plant disease management and their performance in the real world is undergoing intense examination.

Small Grain Disease Forcasting Home
Plant Pathology Department,
North Dakota State University, 306 Walster Hall, Fargo, ND 58105-5012
Web Site:  http://www.ag.ndsu.nodak.edu   Email:  forecast@ndsuext.nodak.edu