Assessing the accuracy of vegetative roughness estimates using unmanned aerial vehicles [UAVs

Abstract

Cost-effective UAV (Unmanned Aerial Vehicle) technologies were utilized to map both submerged and emergent aquatic vegetation in natural rivers. This study was undertaken along reaches characterized by vegetative conditions ranging from homogeneously distributed to strongly heterogeneous and anisotropic. Spatial extent of vegetation was identified using both manual and automated image post-processing methods. For the study reaches assessed here, if 13,000 pixels/m2 image resolution is maintained (which mostly depends on flight elevation and camera resolution) aquatic vegetation can be detected accurately. The methods presented here can be used to inventory aquatic vegetation in under one hour, as opposed to field data collection methods which would require approximately 9.5 h to achieve a comparable level of spatial resolution. Results were also applied to re-evaluate the accuracy of flow resistance formulations based on plants spatial distribution found in literature. It was found that there can be up to a 20% difference if vegetation is inventoried at the cross-sectional scale rather than at the planform scale, however for dense vegetation cover this difference is much lower.