A Work Flow and Evaluation of Using Unmanned Aerial Systems for Deriving Forest Stand Characteristics in Mixed Hardwoods of West Virginia

Forest inventory information is a principle driver for forest management decisions. Information gathered through these inventories provides a summary of the condition of forested stands. The method by which remote sensing aids land managers is changing rapidly. Imagery produced from unmanned aerial systems (UAS) offer high temporal and spatial resolutions and have added another approach to small-scale forest management. UAS imagery is less expensive and easier to coordinate to meet project needs compared to traditional manned aerial imagery. This study focused on producing an efficient and approachable work flow for producing forest stand board foot volume estimates from UAS imagery in mixed hardwood stands of West Virginia. A supplementary aim of this project was to evaluate which season was best to collect imagery for forest inventory. True color imagery was collected with a DJI Phantom 3 Professional UAS and was processed in Agisoft Photoscan Professional. Automated segmentation was performed with Trimble eCognition Developer\u27s multi-resolution segmentation function with manual optimization of parameters through an iterative process. Individual tree volume metrics were derived from field data relationships and volume estimates were processed in EZ CRUZ forest inventory software. The software, at best, correctly segmented 43% of the individual tree crowns. No correlation between season of imagery acquisition and quality of segmentation was shown. Volume and other stand characteristics were not accurately estimated and were faulted by poor segmentation. However, the imagery was able to capture gaps consistently and the high resolution imagery was able to provide a visualization of forest health. Difficulties, successes and time required for these procedures were thoroughly noted