Spatially explicit and detailed information on tree species composition is critical for forest management, nature conservation and the assessment of forest ecosystems services. In many countries, forest attributes are monitored regularly through sample-based forest inventories. In combination with satellite imagery, data from such forest inventories have a great potential for developing large area tree species maps. Here, the high temporal and spatial resolution of Sentinel-1 and Sentinel-2 has been useful for extracting vegetation phenology, information that may also be useful for improving forest tree species mapping.

Thus, we generate cloud free time series with 5-day intervals from Sentinel-2 imagery and combine those with monthly Sentinel-1 backscatter composites to map the dominant tree species throughout Germany. Further, we incorporate information on topography, meteorology, and climate to account for environmental gradients in Germany. We use the German NFI as source for training data for a random forest classifier and for validation data to assess map accuracy.

The optical Sentinel-2 data were downloaded, processes and structured in an analysis-ready data cube using with the open-source software FORCE (Framework for operational radiometric correction for environmental monitoring; Frantz, D., 2019, doi.org/10.3390/rs11091124; https://force-eo.readthedocs.io/en/latest/ last accessed: 02. Nov. 2022). The preprocessed Sentinel-1 SAR data was obtained through the CODE-DE platform (Sentinel-1 L3 BS - Monthly Composite, https://code-de.org/en/portfolio/, last accessed: 02. Nov. 2022) and included in the FORCE data cube along with the environmental data.

We used Sentinel-2, Sentinel-1 and environmental data from the two years 2017 and 2018 to generate the map with 11 tree species groups, including two broader deciduous tree classes (ODH: other deciduous trees with high life expectancy, ODL: other deciduous trees with low life expectancy), within the extent of a previously produced tree map: Langner et al. (2022,  doi.org/10.3220/DATA20221205151218).

The reference data for model training and map validation is based on the third German National Forest Inventory which was recorded in 2011/2012 (https://www.bundeswaldinventur.de/en/third-national-forest-inventory, last accessed: 02. Nov. 2022). To exclude inventory plots where tree species composition might have changed in between 2012 and 2018, we used the forest disturbance maps by Senf et al. (2020, doi.org/10.1038/s41893-020-00609-y). The map accuracies were estimated based on a stratified estimator according to Stehman (2014, doi.org/10.1080/01431161.2014.930207).

Map accuracies were estimated for pure species stands in a first step and for the entire map, including mixed species stands, in a second step. The overall accuracy of pure-species stands is 87.07 ± 0.3 % and 75.53 ± 0.07 % for the entire map, including mixed-species stands.

See also publication: Lukas Blickensdörfer, Katja Oehmichen, Dirk Pflugmacher, Birgit Kleinschmit, Patrick Hostert,
National tree species mapping using Sentinel-1/2 time series and German National Forest Inventory data,
Remote Sensing of Environment, Volume 304, 2024,
114069, ISSN 0034-4257, https://doi.org/10.1016/j.rse.2024.114069.

For data access, please follow this link to the OpenAgrar repository: https://doi.org/10.3220/DATA20221214084846

• 2017
• 2018
• Baumarten
• Deutschland
• Wald
• Waldinventur

For information on the map accuracy of the provided dataset, please see the linked table: “Map accuracies to: Dominant Tree Species for Germany (2017/2018)”

To have a look at the map follow the Link

https://atlas.thuenen.de/maps/new?layer=geonode:Dominant_Species_Class&view=True#/