Topological Simplification and Machine Learning for Real-time Prediction of Off-target Pesticide Spray Drift

Pesticide drift is a phenomenon with a variety of very significant environmental impacts. In addition to immediate economic impacts of lost crops or reduced yields and the social impact of conflict between neighboring land utilizers, off-target drift can result in herbicide resistant weeds, human health effects, and can damage sensitive habitats. Wind speed and direction is a predominant contributing factor to spray drift at the time of application. Our technology will rapidly deliver a hyper-local forecast of wind velocity and expected drift via a web?delivered application. We will utilize mathematically motivated simplification algorithms to make feasible a rapid and inexpensive forecast, and we will use weather forecast products with national coverage to facilitate applicability in multiple regions. This will enable farmers and applicators to better plan pesticide applications, reduce the incidence of off-target drift, and raise awareness of drift potential. The global agricultural biotechnology market (itself a subset of the larger agricultural market) was $50.5 billion in 2019 and is growing. According to the US EPA CPARD, there were more than 1.1 million licensed applicators in the US in 2020. We anticipate end users of our product to include at least pesticide applicators and extension agents, however we expect that regulators and larger agrochemical manufacturers will also derive value from our technology. Based on our interviews, applicators tend to rely on ad hoc assessments in the field on weather conditions and nearby weather station reports. In North Carolina, the NC ECONet Spray Conditions Tool has achieved considerable engagement, indicating a desire for more informative assessments of pesticide drift risk. Our product will advance the state of the art by delivering a forecast tailored for spray drift risk assessment that accounts for local topography when resampling wind forecasts and other weather data to the exact application location. Reductions in off-target pesticide drift as a result of utilizing our technology will result in ample environmental quality returns, including slowing evolution of pesticide resistance, protecting sensitive habitats, and mitigating risks to human health from chemical exposures.