Quantifying Aerial Concentrations of Maize Pollen in the Atmospheric Surface Layer Using Remote-Piloted Airplanes and Lagrangian Stochastic Modeling

Abstract

Abstract The extensive adoption of genetically modified crops has led to a need to understand better the dispersal of pollen in the atmosphere because of the potential for unwanted movement of genetic traits via pollen flow in the environment. The aerial dispersal of maize pollen was studied by comparing the results of a Lagrangian stochastic (LS) model with pollen concentration measurements made over cornfields using a combination of tower-based rotorod samplers and airborne radio-controlled remote-piloted vehicles (RPVs) outfitted with remotely operated pollen samplers. The comparison between model and measurements was conducted in two steps. In the first step, the LS model was used in combination with the rotorod samplers to estimate the pollen release rate Q for each sampling period. In the second step, a modeled value for the concentration Cmodel, corresponding to each RPV measured value Cmeasure, was calculated by simulating the RPV flight path through the LS model pollen plume corresponding to the atmospheric conditions, field geometry, wind direction, and source strength. The geometric mean and geometric standard deviation of the ratio Cmodel/Cmeasure over all of the sampling periods, except those determined to be upwind of the field, were 1.42 and 4.53, respectively, and the lognormal distribution corresponding to these values was found to fit closely the PDF of Cmodel/Cmeasure. Model output was sensitive to the turbulence parameters, with a factor-of-100 difference in the average value of Cmodel over the range of values encountered during the experiment. In comparison with this large potential variability, it is concluded that the average factor of 1.4 between Cmodel and Cmeasure found here indicates that the LS model is capable of accurately predicting, on average, concentrations over a range of atmospheric conditions.