(52d) Evaluating the Accuracy of UAV-Based Measurements through Experiments and Computational Fluid Dynamics Simulations.

Poor air quality is the largest environmental risk for early death, responsible for an estimated 7 million premature deaths each year. In an aim to address air quality concerns, national standards are set for major pollutants including ozone and particulate matter. Enforcing and maintaining air quality standards requires accurate measurements with high spatial and temporal resolution. However, collecting measurements over a wide area is difficult with traditional, stationary monitoring equipment. For this reason, unmanned aerial vehicles (UAVs), also called drones, have become a popular platform for air quality measurements. UAVs are able to measure pollution levels through the atmospheric boundary layer and in locations that are not easily or safely accessible to humans. While many recent studies have demonstrated the usefulness of UAVs in air pollution monitoring, only a few have evaluated the accuracy of the UAV-based measurements by comparing to ground-level or other reference measurements. Among studies that have evaluated measurement accuracy, there is notable variation between the results with coefficients of correlation between UAV and reference measurements as low as 0.30. This work aims to evaluate the accuracy of UAV-based pollution measurements through experiments and computational fluid dynamics (CFD) simulations. We measured vertical profiles of ozone and particulate matter in northern Utah at multiple times throughout the day using a hexarotor UAV. Our data show that in the morning hours when ozone starts to form there is a relatively linear gradient that increases with elevation. By midday, the concentration becomes mostly uniform, and in the evening, ozone dissipates. During the morning and evening periods, we observed notable differences between the ascent and descent of the same flight. To investigate these observations, CFD simulations are used to model the airflow around the UAV body and estimate mixing from the rotors. Our simulations show that differences between ascent and descent are likely due to the fluid interaction with the body of the drone. Surprisingly, our simulations also suggest that the mixing caused by the rotors may help reduce these differences and improve accuracy for vertical measurements across a gradient. This work demonstrates how simulations, coupled with experimental studies, can provide insight into the complex fluid flow around UAVs and improve measurement quality. Future work will include simulations of other measurement scenarios, including sampling from a plume and extending the sensor intake.