Riparian vegetation restoration efforts demand cost effective, accurate, and replicable impact assessments. In this thesis a method is presented using an Unmanned Aerial Vehicle (UAV) equipped with a GoPro digital camera to collect photogrammetric data of a 2.02-acre riparian restoration. A three-dimensional point cloud was created from the photos using Structure from Motion (SfM) techniques. The point cloud was analyzed and compared to traditional, ground-based monitoring techniques. Ground truth data collected using the status-quo approach was collected on 6.3% of the study site and averaged across the entire site to report stem heights in stems/acre in three height classes, 0-3 feet, 3-7 feet, and greater than 7 feet. The project site was divided into four analysis sections, one for derivation of parameters used in the UAV data analysis, and the remaining three sections reserved for method validation. The most conservative of several methods tested comparing the ground truth data to the UAV generated data produced an overall error of 21.6% and indicated an r2 value of 0.98. A Bland Altman analysis indicated a 99% probability that the UAV stems/plot result will be within 159 stems/plot of the ground truth data. The ground truth data is reported with an 80% confidence interval of +/- 844 stems/plot, thus the UAV was able to estimate stems well within this confidence interval. Further research is required to validate this method longitudinally at this same site and across varying ecologies. These results suggest that UAV derived environmental impact assessments at riparian restoration sites may offer competitive performance and value.
展开▼
