The aim of the project is the drone-based derivation of forest parameters relevant to forestry for the implementation of digital forest inventories. Conventional forest inventories are usually conducted manually and sample-based, with the assumption that these samples are representative for the entire forest stand. This assumption is subject to large uncertainties with respect to the heterogeneity of deciduous and mixed forest stands in particular. In comparison, the digital inventory methods to be developed in this project promise a higher accuracy as well as a significantly higher efficiency, since they enable an area-wide coverage of forest areas and require no time-consuming inspection of forest stands.
In the course of the project, drones will be used for practical, fast and cost-effective data acquisition. Using the products derived from these datasets (3D point clouds, orthomosaics and height models), procedures will be developed that enable user-friendly, accurate and largely automated extraction of forest parameters. The main focus lies on the extraction of single tree parameters such as diameter at breast height, tree height, coarse wood debris, tree stem position and individual tree crown delineation. Rule-based algorithms and machine learning methods will be developed, using both point- and raster-based approaches. In addition, the combined use of data sets with different drone flight configurations (e.g., acquisition time and camera viewing angle) will be investigated.
Potential users and customers include forest owners, the timber industry, forest managers, national parks, forest scientists, climate researchers, forestry institutions, nature and environmental associations and civil society in general.
Profile cross-section of a UAV point cloud. The upper section shows the point cloud under leaf-off conditions (winter), the middle section the point cloud under leaf-on conditions (summer) and the lower section the combination of both data sets.
