244
Evaluation of peanut disease development forecasting

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 18 January 2010
Exhibit Hall B2 (GWCC)
John McGuire, North Carolina State University, Raleigh, NC; and M. Brooks, A. Sims, B. Shew, and R. Boyles

Handout (201.8 kB)

North Carolina ranks fourth in the nation for peanut production, which was worth $90 million in 2008. Disease prevention and mitigation are critical for this industry, which can realize crop losses of up to 80% from Sclerotinia blight and 50% by peanut leaf spot. Previous work developed algorithms and automated, location-specific advisories using data from the North Carolina Environment and Climate Observing Network (ECONet). Growers have access to the advisories via their county extension agent. Hourly weather observations from previous days are used to estimate risk. By helping growers identify the favorable times for disease development and fungicide application, time and money can be saved and crop yields can be increased. In the 2005 growing season, it is estimated that two sprays were saved across the state, which equates to USD $2.2 million in potential savings.

Efforts are now underway to operationally forecast favorable conditions for Sclerotinia and leaf spot during the upcoming 72 hours. These forecasts are based on operational WRF and MM5 simulations produced by the State Climate Office of North Carolina and NWS's National Digital Forecast Database (NDFD). Surface observations in the peanut production regions of eastern North Carolina are compared with these operational forecasts during the 2008 growing season (1 May 15 Oct). The WRF model utilized two domains (15km and 5km) and the MM5 had a resolution of 12km, all three centered over North Carolina, using similar physics packages. The NDFD used the 5km CONUS grid. The results suggested that the MM5 model is the least accurate for predicting leaf spot favorable hours. NDFD also compared poorly with observations, and in a few instances had higher errors as compared to MM5. The WRF model generally had the lowest error both on a monthly and overall basis in comparison to MM5 and NDFD.