Evaluating the feasibility and potential of unmanned aerial vehicles to monitor implementation of forestry best management practices in the coastal plain of the southeastern United States

Compared to undisturbed areas, timber harvest activities, including road construction, skid trails, and stream crossings, exhibit the potential for causing erosion and sedimentation if forestry best management practices (BMPs) are not implemented. Forestry BMP monitoring serves as a tool to evaluate BMP implementation status. Conventional on-the-ground BMP monitoring has contributed substantially to assessing BMP implementation. However, conventional on-the-ground surveys can be time consuming. Unmanned aerial vehicles (UAVs) have been rapidly emerging as a new tool for local-scale monitoring. In this study, we evaluated the feasibility and potential of UAVs for monitoring forestry BMPs in the coastal plain of southeastern United States. We compared the performance of UAV-live feed surveys and UAV-created map surveys with conventional on-the-ground surveys across major BMP categories on 30 study sites using BMP guidelines and implementation survey questions from Alabama, Georgia, and Florida for the sites specific to that state (10 sites in each state). The UAV-live feed surveys involved real-time assessments conducted using the live feed function, enabling on-the-spot evaluations. Whereas the UAV-created map surveys entailed autonomous UAV flight to collect imagery specifically for generating maps and facilitating subsequent assessments. We found that using a UAV for monitoring BMPs efficiently provided an overview of a timber harvest area from above. UAV-live feed surveys were as effective as conventional on-the-ground surveys across all major BMP categories. Specifically, the Cramer's V correlation between the responses from UAV-live feed and conventional on-the-ground surveys for all BMP categories, was 0.95 (p < 0.0001) for detecting implemented BMPs and 0.86 (p < 0.0001) for detecting when BMPs were not implemented but needed to be. UAV-created map survey results vs. conventional on-the-ground surveys were not as effective (V of 0.79 for detecting implemented BMPs and 0.46 for detecting BMPs not implemented but needed). Waste and stream crossing BMP implementation were the most difficult to verify using the map surveys, which was not an issue with the live feed surveys due to the capability to evaluate those BMPs at different camera angles. While UAV-created map surveys may be less effective, they provide an observation record that includes a map of the site. This project also developed a standardized framework for using UAVs to monitor forestry BMPs, including instructions and procedures for conducting UAV-live feed and map surveys. In conclusion, a UAV using the live feed function in the field is a feasible option to monitor forestry BMPs.