An Ensemble Approach In Modeling Atmospheric Dispersion Of Pesticide Spray Drift In Yuma County: Results And Next Steps
Presenter: Lauren Knickrehm1
Co-Author(s): Sunyi Yuana, Chayan Roychoudhurya,and Melissa Furlongb
Advisor(s): Dr. Ave Arellano
1Department of Hydrology and Atmospheric Sciences, University of Arizona
Panapto Presentation Video
Poster PDF
Poster Session 2
Pesticide vapor and particles from spray drift have the potential to pose a risk to public health. We completed an initial investigation of pesticide drift transport and deposition from agricultural application sites in Yuma County, Arizona. Nine application sites were selected with two downwind communities as potential receptors. This study area was placed within a 50x50 domain centered in Yuma County. We used dispersion model CALPUFF with meteorology from high-resolution Weather Research and Forecasting (WRF) ensemble forecasts to investigate spatiotemporal distribution of pesticide abundance due to spray drift. Nine pesticide applications in September and October 2011 were retroactively simulated for 48-hours to assess abundance of active ingredient lambda-cyhalothrin. To represent uncertainties in key model parameters, we conducted approximately 21 WRF/CALPUFF manual simulations with varying emission, chemical lifetime, deposition rate, application height, and meteorology input values. Our results show 99.4% of vapor mass advects out of the domain and 99.5% of particle mass deposits within the domain. Dispersion of the vapor plumes depend on prevailing wind speed and direction. The ensemble mean vapor concentrations of the 48-hour average range from 0.2 ng/m3 to 200 ng/m3, and the peak is up to 1000 ng/m3 (near application sites) and 50 ng/m3 (near one community). Unlike vapor, the particles deposit mainly near the application sites, although some application sites were relatively close to one of the communities. The study is being expanded to automate the ensemble, simulate a greater number of application sites, evaluate more active ingredients, and include field measurements for model validation.