4Map4Health

Forests have high economic and ecological importance. Increasing causes of forest damages are forest fires and insects, bark beetles in particular, often in combination with or intensified by abiotic stresses such as drought or storm. Tree species information is inaccurate for large forest areas, but is relevant for any commercial use and ecological function of the forest resources. The major research question of this project is how the future multitemporal, multispectral laser scanning data should be processed in order to provide information for environmental sustainability and especially for mapping of the forest health such as bark beetle risk assessment, tree species, and forest fire risk. Common to mapping of forest health, tree species, and forest fire risk, is that data are correlated to moisture of canopies. On the other hand, lidar backscatter is strongly dependent on the moisture and recent studies indicate that it can be derived by use of bispectral airborne lidar. Collection of such data is possible even at country level at a few years interval. The project partner FGI has the world-first multispectral, mobile laser scanner that can be used for such studies complemented with other data sources to support future scanning programs taking all around Europe. The results can be used in early warning systems supporting near-real-time or real-time processing. The innovative prediction framework allows the assessing of bark beetle infestation hazard on forest stand level under particular consideration of drought.

Increasing causes of forest damage are forest fires and insects, often in combination with or exacerbated by abiotic stress factors such as drought or storms. At present, information on tree species is not yet accurate enough. The main research question in 4Map4Health was how future multitemporal, multispectral laser scanning (MS-LiDAR) data can be processed to provide information for for mapping forest health (e.g. bark beetle risk assessment), tree species and forest fire risk. The common problem with mapping forest health, tree species and forest fire risk is, that they are related to canopy moisture. On the other hand, lidar backscatter is highly dependent on moisture, which is why MS-LiDAR devices were investigated. The data was collected using the world's first MS-LiDAR sensor platform from FGI, Finland. For Austria, MS-LiDAR data was collected for two test areas in Lower Austria: 'Dunkelsteiner Wald' and 'Riegersburg', where weather and soil data is also available. We have developed a tree species classification method based on a PointNet++ network and we participated in an international benchmark event within the 4Map4Health project for the classification of tree species based on MS-LiDAR data. The results were published in a joint publication in the ISPRS Journal of Photogrammetry and Remote Sensing. In addition, we conducted an international benchmark campaign as part of the Silvilaser 2021 conference organized by TU Wien. Various companies and institutions were invited to use their devices to capture 3D points for eight selected forest plots. The plots are located in the Vienna Woods. In addition to the remote sensing (RS) data, detailed in-situ data was collected using conventional measuring devices. These data sets serve as a reference for the RS-based data analyses. All data was pre-processed and published on https://doi.org/10.48436/afdjq-ce434. The data is used to develop algorithms and to compare the performance of the different data acquisition devices. Several in-situ measurement devices collect the 3D information in a local coordinate system. In order to compare the data with other in-situ data and other RS data, a common coordinate system is required. To transform the 3D data from the local to the global coordinate system, we have developed a workflow to register different 3D data sources together. First results are published in the IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. A journal article with an extended approach is in preparation. To improve the available bark beetle predisposition models (developed by our national project partner BOKU), we have developed different workflows/methods to derive 3D forest structure information including forest edge information, which is an essential source of information for bark beetle infestation, from LiDAR point clouds. A joint article with the project partner BOKU is in preparation.