Increasing regulation of spray drift has led to the development of real-time drift monitoring systems that present drift potential to applicators so that drift reduction spraying techniques can be implemented on an as-needed basis. The central component in each of these state-of-the-art systems is a drift prediction model. A real-time drift monitoring system was developed using look-up tables produced from simulations of a random-walk model (FLUENT via DRIFTSIM). The predictive accuracy of this system, evaluated as the difference between predicted drift and in-field measured drift, was compared to alternative prediction models to determine the suitability of random-walk models for real-time drift prediction. DRIFTSIM was found to produce a significantly more accurate representation of real-time predicted drift when compared to four of the six alternative models tested. No significant difference in predictive accuracy was found when comparing DRIFTSIM to the two other models. When compared to alternative models at incremented distances downwind from the point of spraying, DRIFTSIM’s predictions were found to be overall more accurate up to 10 m from the boom edge; however, three alternative models provided more accurate predictions for long-distance drift (20 to 50 m from the boom). These results suggest the potential of using DRIFTSIM in future real-time drift monitoring for increased accuracy and performance. However, additional development is needed to improve far-field (>10 m downwind of an application) drift prediction accuracy.
Available at: http://works.bepress.com/mark_hanna/30/