Skip to main content

Forests mapped from space

The Thünen Institute of Forest Ecosystems publishes for the first time national forest maps of Germany based on current satellite imagery and terrestrial data from the 2012 National Forest Inventory.

Kartenausschnitte zum bestockten Holzboden (links; grüne Flächen) und zu den dominierenden Baumarten (rechts) mit Übersicht der kartierten Baumartengruppen (rechts; bunte Flächen)..
© Thünen-Institut/WO

Kartenausschnitte zum bestockten Holzboden (links) und zu den dominierenden Baumarten (rechts) mit Übersicht der kartierten Baumartengruppen. Die Klassen ALH und ALN fassen die nicht einzeln aufgeführten Laubbaumarten mit hoher oder niedriger Lebenserwartung zusammen.

The Thünen Institute now offers interactive forest maps through the Thünen Atlas (atlas.thuenen.de), which provide a nationwide overview of the stocked forest area – i.e. the area on which trees grow – and the dominant tree species. They are not limited to being important sources of information for public authorities, policy, NGOs and associations, but can also be used for further forest and environmental research and education.

The digital maps have a resolution of 10 m x 10 m. The map of dominant tree species is based on data from the Sentinel satellites. These satellites are part of the Copernicus Program form the European Space Agency (ESA) and are used for global environmental monitoring, for example to record changes in land use, to monitor water bodies and coasts, and to collect satellite-based climate data. The nationwide consistent data sample from the 2012 National Forest Inventory (www.bundeswaldinventur.de/en) and the 2017 Carbon Inventory were used for tree species classification and for the comprehensive statistical validation of the maps.

Satellite data thus provide valuable complementary information for monitoring current forest state and forest development. However, the level of detail on forest ecosystems required by policymakers and society goes beyond what is observable from the satellite. The main source of information on our forests will therefore continue to be terrestrial inventories such as the National Forest Inventory, the Carbon Inventory and the National Forest Soil Inventory.
 

Stocked forest area map 2018
(https://atlas.thuenen.de/layers/geonode:fnews_holzbodenmaske_2018_32632)

The 2018 stocked forest area map represents the area covered by forest trees in 2018. It provides the framework for further forest related remote sensing analyses like tree species or biomass mapping, but also for identifying future changes in stocked forest area. For this purpose, it is planned to update the "stocked forest area map" at regular intervals, provided there is respective funding.

A consistent definition of forest area is the basis for remote sensing-based forest area mapping. The Copernicus Earth Observation program builds on the international FAO forest definition. In contrast, the map produced at the Thünen Institute of Forest Ecosystems uses the forest definition by the German National Forest Inventory, which has been optimized on the national level. The map is validated with data from the 2017 Carbon Inventory and has an overall accuracy of greater 95%.
 

Dominant tree species map (https://atlas.thuenen.de/layers/geonode:Dominant_Species_Class)

The map of dominant tree species, developed by the Thünen Institute of Forest Ecosystems in cooperation with the Humboldt University Berlin and the Technical University Berlin, covers the entire forest area of Germany for the reference period 2017/2018. For example, this enables the analysis of tree species-specific changes in the years after 2018, a time period characterized by forest disturbance. Likewise, the tree species map can be an important input for the analysis of forest resilience to the changing climate, i.e. the question of how certain tree species or forest compositions cope with climate change.

To produce the tree species map, data from the National Forest Inventory were combined with Sentinel-2 and Sentinel-1 satellite data from the Copernicus-Program and processed using machine learning methods. Regional growth conditions were accounted for by weather and climate data from the German Meteorological Service and by including a nationwide elevation model.

The extensive data of the National Forest Inventory served as reference data for the mapping of tree species. The statistical validation showed that the most common tree species such as spruce, pine, beech and oak can be well separated. However, it also points towards areas of further research, for example the mapping of very young stands or mixed forests with many different tree species over small areas. Nevertheless, the map now published with 11 tree species groups is a central step in the development of area-wide, forest-related information that will enable many follow-up analyses at the national level.

[Translate to English:]

[Translate to English:]

Karte dominierender Baumarten

Die Karte der dominierenden Baumarten im Wald, die das Thünen-Institut für Waldökosysteme in Kooperation mit der Humboldt-Universität zu Berlin und der Technischen Universität Berlin entwickelt hat, deckt die gesamte Waldfläche Deutschlands für den Referenzzeitraum 2017/2018 ab. Dadurch wird es zum Beispiel möglich, baumartenspezifische Veränderungen des Waldes in den von Waldschäden geprägten Jahren nach 2018 zu analysieren. Ebenso kann die Baumartenkarte eine wichtige Eingangsgröße für eine Resilienz-Analyse des Waldes mit Blick auf das sich verändernde Klima darstellen – also die Frage, wie bestimmte Baumarten oder Baumbestände mit den Klimaänderungen zurechtkommen.

Für die Kartenerstellung wurden die Daten der Bundeswaldinventur mit Sentinel-2- und Sentinel-1-Satellitendaten des Copernicus-Programms kombiniert und mit Verfahren des maschinellen Lernens verarbeitet. Regionale Wuchsbedingungen wurden durch Wetter- und Klimadaten des Deutschen Wetterdienstes und durch ein deutschlandweites Höhenmodell berücksichtigt.

Als Referenzdaten zur Erstellung der Baumartenkarte dienten die umfangreichen Daten der Bundeswaldinventur. Die statistische Überprüfung zeigte, dass sich die häufigsten Baumarten wie Fichte, Kiefer, Buche und Eichen gut voneinander trennen lassen. Es zeigte sich aber auch, dass noch Forschungs- und Optimierungsbedarf besteht bei sehr jungen Beständen oder bei Mischwäldern mit vielen unterschiedlichen Baumarten auf kleinem Raum. Dennoch ist die jetzt veröffentlichte Karte mit 11 Baumartengruppen ein zentraler Schritt in der Entwicklung von flächendeckenden, waldbezogenen Informationen, die viele Folgeanalysen auf nationaler Ebene ermöglicht.

Contact:

Institute of Forest Ecosystems
Head of working group Inventory (Forest resources and climate protection), contact person National Forest Inventory
Scroll to top