Development and Evaluation of a Weather-based Epidemiological Model for the Prediction of Brown Patch in Creeping Bentgrass

Show full item record

Title: Development and Evaluation of a Weather-based Epidemiological Model for the Prediction of Brown Patch in Creeping Bentgrass
Author: Palmieri, Richard
Advisors: Gary Lackmann, Committee Member
Lane Tredway, Committee Co-Chair
Dev Niyogi, Committee Chair
Abstract: The turfgrass industry is one of the largest agricultural industries in the United States. The nature of turfgrass favors the long-term buildup of pests, including many fungi that cause disease (including Rhizoctonia solani, the causal agent of the brown patch turfgrass disease). To combat these diseases, turfgrass managers must frequently apply expensive, harmful fungicides to their turf. Previous studies have shown that fungal infections of turfgrass are related to meteorological conditions, and that weather-based epidemiological models can be used to help turfgrass managers reduce the number of fungicide applications they are required to make, while maintaining the aesthetic quality of their turf. However, these previously-developed epidemiological models had not been tested in North Carolina, and their efficacy in this state was in question. Therefore, this study was undertaken to determine if the Schumann and Fidanza epidemiological models for brown patch could be used in North Carolina. Disease observations and weather data collected over the summers of 2003 and 2004 at the NC State Faculty Club Turfgrass Field Laboratory were combined, and resulted in the conclusion that both the Schumann and the Fidanza epidemiological models met only with very limited success in predicting brown patch outbreaks. However, several other conclusions were reached by the end of the study, including: The most accurate method for measuring brown patch activity is uncertain, as a smaller data set of disease incidence measures disagreed frequently with the once-daily observations of brown patch activity. Weather data from regional-scale observational networks, and from operational numerical weather prediction (NWP) models, can likely be used as a proxy for on-site weather measurements. When weather data from the Raleigh-Durham International Airport, and from the operational run of the then National Centers for Environmental Prediction Eta model were used as inputs into the epidemiological models above, the results varied little as compared to the results using weather data from an on-site observation station. After determining that no existing epidemiological model provided accurate predictions of brown patch (assuming that the disease activity observations were sufficient; an assumption in question), an effort was made to develop a new epidemiological model was undertaken. Several statistical methods were used, including an autoregressive model and a logistic regression model, but neither was able to accurately explain brown patch activity. A process-based epidemiological model was also developed, under the assumption that hot, humid conditions are a necessary condition for the development of brown patch disease. While this model did not lead to an accurate predictive index, it was the most promising effort, typically hampered by high false alarm ratios. This leads to the conclusion that an investigation into those meteorological conditions that are unfavorable for disease development, when all other factors appear favorable, is likely in order. Finally, a NWP model sensitivity study was undertaken using the Pennsylvania State University-National Center for Atmospheric Research Mesoscale Model (MM5). Two cases were chosen to represent two very different summertime regimes for North Carolina, and for each case, two MM5 runs were performed; one using a 5-layer soil model, the other using the NOAH land-surface model (LSM). It was found that large variability can exist between model runs based on the land-surface parameterizations used (especially in convective regimes), and that in this study, these differences can be of a greater magnitude than those differences seen using varying sources of weather data as inputs for the epidemiological models.
Date: 2005-09-06
Degree: MS
Discipline: Marine, Earth and Atmospheric Sciences

Files in this item

Files Size Format View
etd.pdf 5.467Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record