Published October 9, 2026 | Version v1.0
Dataset Restricted

National-scale mapping of tree species composition in Switzerland from high-resolution aerial imagery and multimodal data

  • 1. ROR icon ETH Zurich
  • 2. ROR icon Swiss Federal Institute for Forest, Snow and Landscape Research

Description

This dataset contains the national-scale tree species composition maps of Switzerland described in Zhou et al. (submitted to Remote Sensing of Environment). For every forest pixel of a 30 m grid, the maps give the predicted basal-area share of the ten main tree species of the Swiss National Forest Inventory (NFI) and a residual Other class.

Method in brief

A multimodal, multi-task deep-learning model was trained on the basal-area composition of 3,992 Swiss NFI sites. It combines SWISSIMAGE aerial RGB orthophotos, swissSURFACE3D airborne laser scanning, annual AlphaEarth satellite embeddings and CHELSA BIOCLIM+ climate variables. Each pixel uses the most recent leaf-on SWISSIMAGE acquisition available for its area (2018–2025), and the matching annual AlphaEarth embedding; this reference year is provided as a separate layer.

Files

All files are Cloud-Optimized GeoTIFFs on the same grid: EPSG:2056 (CH1903+ / LV95), 30 m resolution, 11,586 × 7,355 pixels. Non-forest pixels are NoData.

File Content Values NoData
species_share_01_spruce.tif … species_share_11_other.tif Predicted basal-area share of each class: 01 Spruce (Picea spp.), 02 Fir (Abies alba), 03 Pine (Pinus spp. excl. P. cembra), 04 Larch (Larix spp.), 05 Arolla pine (Pinus cembra), 06 Beech (Fagus sylvatica), 07 Maple (Acer spp.), 08 Ash (Fraxinus spp.), 09 Oak (Quercus spp.), 10 Chestnut (Castanea sativa), 11 Other (all remaining taxa) 0–1, float32 −9999
species_argmax.tif Dominant class: the class with the largest predicted basal-area share 1–11 (class codes as above), uint8, with colour table 255
leaf_type_share_01_broadleaved.tif, leaf_type_share_02_coniferous.tif Predicted broadleaved and coniferous share of total basal area (the two sum to 1) 0–1, float32 −9999
inference_year.tif Reference year: year of the leaf-on SWISSIMAGE acquisition (and AlphaEarth embedding) used for inference 2018–2025, uint16, with colour table 0

Citation

Please cite the associated article when using these data: (Not decided)

Files

Restricted

The record is publicly accessible, but files are restricted. Log in to check if you have access.

Additional details

Related works

Funding

Swiss National Science Foundation
Nowcasting, forecasting, upscaling - Novel avenues for forest vitality monitoring and anticipating forest dynamics (UPSCALE) 216562