A drone is also called a UAV ("Unmanned Aerial Vehicle") or UAS ("Unmanned Aerial System") according to its specification. The FVA has started to investigate the possible applications of this technique for forestry research. For this purpose, the FVA has a DJI Matrice 200 at its own disposal. The project “Walddrohnen” (forest drones) evaluates the usability of UAVs for forest applications and implements a department-independent workflow. The processes can expand into forest inventory, forest growth, forest health, forest production, wildlife ecology and forest protection.
The benefit of drone technology is defined by the platform, the sensors, the flight parameters and the analysis options. The construction of the platform, e.g. fixed wing, copter or a combination of both, determines the ground coverage, the payload, the starting and landing capability, the flight time, flight stability and flight quality.
Different sensors can be mounted on drones. The data collected by UAVs have a very high spatial resolution, between one and a few centimeters.
- Multispectral- and hyperspectral cameras register different wavelengths and can indicate tree condition by characteristics of the tree crown reflection.
- Cameras, which operate in the visible spectrum, can be used for detailed assessment of forest structure or single tree properties.
- A laser-sensor on a UAV provides very high point densities and therefore a detailed image of the stand and crown structure as well as the ground surface structure.
- Thermal imaging cameras register the different thermal reflections and can probably be useful for ecological topics.
The quality of analysis options is influenced by the sensors for positioning and orientation of the platform during data capture. In general: The more precise the data, the better the subsequent analysis can be and the less ground control points are necessary. Additionally, the flight parameters flying height above the forest stand and image overlap determine the usability of drone-based data.
The current weather conditions are crucial for the quality of data capture. In order to prevent tree crowns from moving, absolute calm is required. For passive optical recording, irregular clouds and ground-level fog has to be excluded. When using laser sensors, rain, fog or snowfall lead to disturbing data noise. A drone flight can capture several hundreds of images, which then must be stitched in to a single mosaic or must be prepared for an analysis of the stand situation or for single tree geometry.
In 2018 several test areas were recorded. Based on these data, the initial calculations for assessing 3D information of the areas have been completed. The aim of these evaluations was to assess the influence of different flight parameters. First steps in standardizing and facilitating the workflow of importing drone data into the FVA data infrastructure have been developed. The influence of flight parameters on the accuracy of tree height measurements derived from drone data was investigated within a test area. In the upcoming year the investigations will be expanded to other areas and workflows will be further developed. Furthermore, different algorithms and software will be evaluated according to their suitability for FVAs purposes.
How accurate are derived tree heights, crown radii and crown base heights using different sensor systems?
This question is addressed by the project: "Further development of statistical timber volume forecasting tools for differentiation of roundwood assortments and product quality (Pro-Qual-Tools)". The project is funded by the Federal Ministry of Food and Agriculture. Beside drone-based data with laser sensor and camera, gyrocopter-based data and aerial images from state survey flights have been included in the investigation. The study site is located in an even-aged 50-year-old Douglas fir stand.
Overall, the results for deriving tree height and crown radius were promising with regards to the accuracy achieved. The differences between the results of laser and photogrammetric data were smaller than expected. However, the photogrammetric data indicated that a high point density is an important factor for successfully deriving single tree attributes. Consequently, the best results were achieved with drone data.
Deriving crown base height at the single tree level did not turn out to be applicable. Even the very high-resolution laser data failed in deriving crown base height. An important consideration here is the definition of the crown base height, which can be either the first green or first dead branch. Both can be very small and are therefore not recorded by the sensor system or detected in the automatic algorithms.
Altogether, results show that drone technology is a tool for small scale assessment of forest and tree properties and can complement the existing remote sensing techniques. However, more research and development is necessary to exploit the potential of the drone technique for practical applications.