Published August 26, 2026 | Version v1
Dataset Open

National Tree Species Map of Denmark

Authors/Creators

  • 1. ROR icon DHI
  • 2. ROR icon University of Copenhagen
  • 3. University of copenhagen, Faculty of Science

Description

This dataset provides wall-to-wall maps of dominant tree species across Denmark at 10 m spatial resolution using data from 2020 to 2022. The maps were generated using machine learning models trained on observations from the Danish National Forest Inventory (NFI) and satellite-based remote sensing data, including Sentinel-1, Sentinel-2, and canopy height information. Each pixel is assigned the predicted dominant tree type represented by one of nine categorical classes.
 

Files

  1. DHI_tree_species_map_DK_ForestMask.tif
  2. DHI_tree_species_map_DK_TreeMask.tif
  3. DHI_tree_species_map_DK_NoMask.tif

Description

All files are single-band categorical rasters with the predicted tree type codes (0–8), corresponding to the table below.

DHI_tree_species_map_DK_ForestMask.tif

This file contains predictions only in areas classified as forest in the Danish Digital Forest Map.

DHI_tree_species_map_DK_TreeMask.tif

This file contains predictions only in areas that are classified as trees using the DHI tree cover map dataset. 

No forest mask is applied, and therefore predictions are also available in urban areas and small tree patches. These predictions should be interpreted cautiously because the model was trained primarily on observations collected in forest environments.

DHI_tree_species_map_DK_NoMask.tif

This file contains predictions across Denmark without any filtering. 

Tree type codes

Code English     Danish    
0 Other Broadleaf (OBL) Andet løv (ANL)
1 Fir Ædelgran
2 Beech Bøg
3 Birch Birk
4 Larch Lærk
5 Maple Ahorn
6 Oak Eg
7 Pine Fyr
8 Spruce Gran
255 No data Ingen data

 

Methodology

For both the tree cover and tree type classification, we used a combination of Sentinel-1, Sentinel-2, and canopy height data. Specifically:

Sentinel-1

Multi-temporal Sentinel-1 VV and VH backscatter observations from 2020-2022 were used, together with the difference between VH and VV backscatter.

Sentinel-2

Multi-temporal Sentinel-2 data were used from 2020 to 2022 using all spectral bands except bands 1, 9, and 10.  In addition to the spectral bands, we computed a list of spectral indices.

Canopy height

Canopy height information was derived from national elevation data provided by the Danish Agency for Data Supply and Infrastructure (Dataforsyningen). The dataset includes a digital terrain model (DTM) and a digital surface model (DSM) with a spatial resolution of 0.4 m.

Classification 

For the classification experiments, we used training data from the Danish National Forest Inventory (NFI), provided by the University of Copenhagen (KU), and trained a multi-layer perceptron (MLP) model to predict dominant tree species.

Part of this work is based on a research grant from INNO-CCUS, the Danish government’s CCUS (carbon capture, utilisation and storage) research partnership, supported by Innovation Fond Denmark.

This dataset is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) licence. Users may copy and redistribute the dataset for non-commercial purposes, provided appropriate credit is given to DHI. Modification, adaptation, or creation of derivative works is not permitted.

For access to the full probabilities of the model, or for any other questions, you can contact Lotte Nyborg, ln@dhigroup.com or Alkiviadis Koukos, akou@dhigroup.com.

Files

DHI_tree_species_map_DK_ForestMask.tif

Files (171.3 MB)

Name Size Download all
md5:5e461cb7fe92f0b2910a9672e59b2856
34.8 MB Preview Download
md5:02437e3815cd899fba16aa96d70b63ad
85.6 MB Preview Download
md5:539f664625ad39c99e259530bbdada0f
51.0 MB Preview Download
md5:a2defd91bcb14fb24ff134bddf7a26b4
4.5 kB Preview Download