2026 Finnish Landuse Import
Hello mappers! I write in English to accomodate any non-Finnish readers. I am proposing to import the forest stand register (metsävarakuviot) and agricultural parcel register (kasvulohkot) datasets, sourced from Metsäkeskus (Finnish Forest Centre) and Ruokavirasto (Finnish Food Authority), into OSM. Currently Finland’s forests have been sparsely mapped, with the majority of the existing forest area being from the low-detail raster-based Corine 2006 landuse import.
Documentation
This is the wiki page for my import:
These are the source datasets’ websites:
The direct data downloads are available here for the forest datasets and here for the farmland parcel dataset.
As a sample, here’s the prepared OSM data file for the 12km × 12km map sheet tile M4434R near Mikkeli:
License
Both datasets are distributed under CC BY 4.0. ODbL-compliance confirmation from the providers is pending: waiver request emails have been sent to both, and Ruokavirasto has verbally confirmed. Metsäkeskus is awaiting a written reply. The status is tracked on the wiki page.
Abstract
This is a one-time, semi-automated import of two official Finnish registers, delivered one 12km × 12km TM35 map sheet at a time.
- Information about the dataset: The forest stand register (Metsäkeskus) is the national register of forest stands: soil type, stand development class, tree species, mire and wetland classes, and water bodies. The agricultural parcel register (Ruokavirasto) is the national subsidy-register layer of farmland parcels: crop codes, organic indicators and official parcel references. Both are authoritative government registers published as open data with CC BY 4.0 licenses. Neither dataset comprehensively covers all forest or farmland areas.
- Relationship to the dataset: I have no relationship with either agency.
- What the dataset contains: Area landcover only. Productive forest, clearcuts, low-productivity and open land (scrub, heath, bare rock), mires, farmland with crops, plus the landuse overrides for semi-natural grassland, orchards and plant nurseries. Roads, buildings, water and other linear features already in OSM are not touched.
- Size: Resulting OSM file sizes are hard to estimate before full preprocessing. The dataset contains approximately 11.9 million forest stands and 985,000 farmland parcels nationwide.
- Method: Offline preprocessing in a Python pipeline (GeoPandas, Shapely, Mapshaper, pyogrio): mapshaper cleanup and simplification (6.0m
keep-shapesinterval, in EPSG:3067), centroid-based assignment to TM35 sheets (importable polygons may extend outside assigned tile to avoid having to clip them), tag generation from the mapping spreadsheets, a 3m combined snapping pass between the two datasets, dissolving of adjacent features with identical tags, and PBF export. Each sheet is then intended to be reviewed manually in JOSM with a conflation plugin that I’m currently developing and uploaded with dedicated import accounts. The script and data repositories will be published later. - Tag translation: The mapping is defined in two human-readable spreadsheets that the pipeline reads;
scripts/tag_mapping_farm.xlsx(farmland, one row per crop code) andscripts/tag_mapping_forest.xlsx(forest landcover, development-class and species rules). The tag spreadsheets are mirrored in full on the wiki page.
| OSM tags | Source |
|---|---|
landuse=farmland |
Default for Ruokavirasto farmland parcels where the crop code doesn’t indicate another landuse=* value. |
landuse=meadow, landuse=plant_nursery, landuse=orchard |
Ruokavirasto farmland parcels with appropriate crop codes. |
organic=yes |
Organic farming flag in farmland parcel. |
landuse=forest, leaf_type=*, leaf_cycle=*, genus=*, species=* |
Forest stand and tree-species classes. |
man_made=clearcut |
Recent clearcuts (development class A0 in dataset). |
natural=scrub, natural=heath, natural=bare_rock |
Low-productivity and open stands inferred from stand data values. |
natural=wetland, wetland=* |
Mire classes (swamp / bog / fen) inferred from stand data values. |
source, source:date |
Metadata on every imported object. |
- Conflation: Each map sheet is conflated manually in JOSM: overlaps between the import layer and existing OSM landuse are reviewed one by one and decided by hand (keep / replace / merge / add / flag), with undo. There is no automatic acceptance. The import is limited to area features.
The repository containing the preprocessing data and scripts, and the conflation plugin’s repository, will be released soon.
The review thread will run for at least 14 days per the Import/Guidelines. Feedback is welcome here and on the wiki page.