ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

Mining Derived Weather Forecast Variables for Crop Disease Risk Prediction

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  7th World Congress on Computers in Agriculture Conference Proceedings, 22-24 June 2009, Reno, Nevada (electronic only)  711P0409e.(doi:10.13031/2013.29062)
Authors:   Jason A Smith, Kathleen M Baker, Robert Trenary
Keywords:   Data mining, artificial neural networks, plant pathology, forecasting

Changing climate conditions have increased the demand for agricultural decision support systems to include true forecasts. Forecasting for crop disease risk can assist growers in reducing the costs and environmental impacts of pesticide use. This paper describes the development of techniques to establish an appropriate variable set from publicly available National Weather Service forecast data for use in crop disease risk forecast models. Our team has been working for several years on artificial neural network (ANN) models to predict disease risk, with potato late blight as our test case. Since the ANN uses a wide range of derived, measured and implied data values, selecting appropriate data is a challenge in its own right. To help gather and dissect data used by the ANN, various tools and methods have been developed to quickly test different ANN models. Automation of processes such as variable selection and derivation through an object oriented framework has the potential to measurably improve the accuracy of the ANN model, and allow for easier implementation of similar models for a variety of crop diseases.

(Download PDF)    (Export to EndNotes)