Pan-European Forest Maps Produced with a Combination of Earth Observation Data and National Forest Inventory Plots

The dataset includes Pan-European maps of timber volume (Vol), above-ground biomass (AGB), and deciduous-coniferous proportion (DCP) with a pixel size of 10 x 10 m for the reference year 2020. In addition, a measure of prediction uncertainty is provided for each pixel. The maps have been created using a combination of a Sentinel-2 mosaic, Copernicus layers, and National Forest Inventory (NFI) data.
The mapping was done with the k-Nearest Neighbour (kNN, k=7) approach with harmonized data of species-specific Vol and AGB from 14 NFIs consisting of approximately 151 000 field plots across Europe. The maps cover 40 European countries, forming a continuous coverage of the western part of the European continent.
A sample of 1/3 of NFI plots was left out for validation, whereas 2/3 of the plots were used for mapping. Maps were created independently for 13 multi-country processing areas. Root-meansquared-errors (RMSEs) for AGB ranged from 53 % in the Nordic processing area to 73 % in the South-Eastern area. The maps are on average nearly unbiased on European level (1.0 % of the mean AGB), but show significant overestimation for small biomass values (53% bias for forests with AGB less than 150 t/ha) and underestimation for high biomass values (-55% bias for forests with AGB higher than 500 t/ha).
The created maps are the first of their kind as they are utilizing a large number of harmonized NFI plot observations and consistent remote sensing data for high-resolution forest attribute mapping. While the published maps can be useful for visualization and other purposes, they are primarily meant as auxiliary information in model-assisted estimation where model-related biases can be mitigated, and field-based estimates improved. Therefore, additional calibration procedures were not applied, and especially high Vol and AGB values tend to be underestimated. We therefore discourage from summarizing map values (pixel counting) over areas in interest, as this may inadvertently result in biased estimates.