Measuring, Modeling, and Evaluating the Spatial Properties of Northeast Oregon Forests Using Unmanned Aerial Systems

There is an ever expanding range of applications for the aerial images that unmanned aerial systems can uniquely provide. One such application is the use of high-resolution imagery for stand-level forest inventory. Inventory techniques utilizing unmanned aerial systems could be attractive where conditions demand high-resolution data, or where other aerial imagery sources are cost prohibitive. Here the effectiveness of unmanned aerial systems in this application was tested. Over the summer of 2015, a remote-controlled hexacopter equipped with a micro four thirds camera was flown over multiple 1600 meter-squared forested plots in Eastern Oregon. Additional ground-level validation measurements were collected including stem location, crown radius, and tree height. Agisoft Photoscan was used to construct 3-D point-clouds which then allowed the production of digital surface models of the stands. The first section of this project assesses the accuracy of stem locations derived from segmented imagery. The next section evaluates the accuracy of estimates for tree height, crown radius, and diameter at breast height. In the final section, various spatial metrics such as stand contagion and species mingling were compared with more commonly used metrics to see if significant correlations emerged. The utilized methods did not yield sufficiently accurate estimates for stem location or the various forest biometrics. Yet this work revealed stand density to be a significant influence on model accuracy. Finally, stand density and species diversity were found to be well correlated with the nearest neighbor and species mingling indexes, respectively, potentially supporting a complementary relationship indicating the clustering of various factors within the stand