Scientists at the University of Copenhagen and the nonprofit organization CTrees have produced the first map of Europe to measure tree biomass both inside and outside forests – such as in hedgerows, gardens, parks, urban areas, grasslands, and agricultural lands.

Published in Science Advances, the map provides a comprehensive level of detail for Europe that was previously only possible in more narrow geographies through airborne light detection and ranging (LiDAR) surveys, which are costly and therefore rarely conducted at a national level. Comparing the national carbon stocks with forest inventories, the overall difference between the new map and national inventory data is only 7.6%, which is considerably less than in previous satellite based biomass maps.

The paper describes an innovative approach to measuring canopy cover, tree heights, and biomass across the continent using deep learning algorithms and satellite data. The map product is based on three-meter resolution data from the PlanetScope satellite constellation, provided by Planet Labs PBC via an academic research license. Reference data on canopy heights from existing airborne LiDAR campaigns was used to train the algorithm.

The study reveals the extent of trees outside forests and their contribution to national biomass and carbon stocks for every country in Europe. While Europe’s forests contain an aboveground biomass about forty times that of trees outside forests, the proportion varies by country. Countries with the highest share of biomass in trees outside forests are Ireland (16.5%), the United Kingdom (14.8%), the Netherlands (12.2%), and Denmark (9.7%).

Governments and scientists can use tree-level data to develop more accurate national inventories and carbon budgets, monitor local and national tree resources, and verify the effectiveness of carbon credit programs and other natural climate solutions.

Dr. Martin Brandt, professor of geography at the University of Copenhagen and a co-author of the paper said, “Governments need accurate measurements of their wood resources to effectively manage timber, habitat, and carbon stocks. But until now, the United Kingdom was the only country in Europe with a full national inventory of trees outside forests. By using low-cost satellite data and deep learning techniques, we can now map trees outside forests in every country with a level of detail that was only possible before through expensive LiDAR surveys.”

The methodology overcomes significant challenges for monitoring trees both inside and outside forests at a reasonable cost. Publicly available satellite data is often not at a high enough resolution to identify non-forest trees or distinguish between forest and non-forest areas. And high-quality aerial imagery that detects canopy height – like LiDAR scans captured from planes – is expensive to produce.

Siyu Liu, a Ph.D. fellow at the University of Copenhagen, and lead author of the paper, said, “Most countries creating national inventories don’t have data for trees outside forests, despite their important roles in providing habitat, reducing flood risk, and storing carbon. Our new method brings together the latest satellite tech with deep learning to show a way for all countries to have access to comprehensive tree-level data at a relatively low cost.”

Co-authors of the paper include CTrees scientists and advisors Jerome Chave, Florian Reiner, Philippe Ciais, Sizhuo Li, Maurice Mugabowindekwe, Sassan Saatchi, and Rasmus Fensholt. The study adds to a growing set of groundbreaking tree-level maps published by CTrees scientists, including for continental Africa earlier this year. CTrees’ ambition is to produce a global tree-level map by the end of 2024, and make it available to governments and organizations seeking to implement programs, policies, and investments to protect their trees and forests.