Towards Cyberinfrastructure for Multi-scale Crop Disease Early Warning Systems

摘要

Access to cyberinfrastructure is critical for regional forecasting of specific economically important crop diseases. We present our team's initial steps to create, implement, and validate a multiscale, multi-regional crop disease forecasting system funded by the USDA. Rapid synoptic and mesoscale crop disease forecasting, especially for emergency decision-making, requires that the model workflow relies on integration of real-time data services from multiple sources, and is executed over a pool of high performance computing resources. Spatially explicit weather forecast models runs are initiated through the Linked Environments for Atmospheric Discovery (LEAD) portal. Hourly forecast variables are extracted from LEAD workflow outputs and are used as inputs to crop disease forecast models implemented as ArcGIS workflows. Workflows include standard models for potato late blight and leaf spot of peanut as well as newly developed models for Fusarium head blight of barley. Resulting crop specific forecasts will inform farm management strategies with the goals of increasing product quality, limiting expenditures, and reducing the amount of chemical released to the environment. Initial results from the 2008 growing season are highly accurate and support continued development of such systems.