iBIOM-EU Database

The iBIOM-EU Database is a harmonised geospatial database for Europe, developed to serve the integrated modelling work of the Biodiversity and Natural Resources (BNR) programme at IIASA. It brings together land use and management, and other biophysical data on forestry, agriculture, water, and biodiversity, into consistent spatial layers that can be used directly across BNR’s modelling ecosystem, including GLOBIOM, EPIC, G4M, CWatM, and iBIS.

The database is under active development. This page summarises what is currently available, how to access it, and what is on the near-term roadmap.

The iBIOM metadatabase is the live register of what exists across models, datasets, and variables. This site is the written explanation of what those things are.

Dataset README convention

Every iBIOM dataset ships with a metadata file named README_<dataset_id>.yaml alongside <dataset_id>.tif. Open the README before using the file. It records the title and variable, units stated explicitly, temporal coverage and what the year actually refers to, spatial coverage and extent, resolution, CRS, licence, source, format, band list with frozen order where the dataset is multi-band, processing level, quality status, related datasets, and known limitations.

The following is a worked example from the production database:

# Dataset Metadata
ID: ibiom_LUM_europe_100m_2018
Title: LAMASUS Land Use Management 100m - European Coverage 2018
Variable: land use management class
Units: categorical
Date: 2018-01-01
Version: 1.0.0
DOI: 10.5281/zenodo.21787459
Citation: "See, L. et al (2025). LAMASUS Land Use Management 100m - European Coverage 2018. iBIOM-EU Database, IIASA, Laxenburg, Austria. DOI: 10.5281/zenodo.21787459"
Temporal coverage: [2000, 2010, 2018]
Temporal resolution: Discrete annual snapshots
Reference period: Annual snapshot for 2018
Spatial coverage: [-55.5, 24.5, 61.0, 72.1]
Spatial extent EPSG:3035: [918900, 916700, 6552100, 5440900]
Resolution: 100m
CRS: EPSG:3035 (ETRS89-LAEA Europe)
License: IIASA Internal Use - Contact see@iiasa.ac.at
Source: LAMASUS Project
Keywords: land use, land management, management intensity, agriculture, forestry, cropland, grassland, Europe
Type: raster
Format: GeoTIFF
Assets:
  - file: LUM_fit_with_energy_levels_and_new_FM_2018.tif
    type: image/tiff; application=geotiff
    roles: [data]
    title: Original GeoTIFF
    size_gb: 0.23
  - file: ibiom_LUM_europe_100m_2018.tif
    type: image/tiff; application=geotiff; profile=cloud-optimized
    roles: [data, derived]
    title: Cloud-Optimized GeoTIFF (COG)
    size_gb: 0.15
Description: High-resolution (100m) land use and management intensity classification for Europe. Baseline foundation data for iBIOM-EU integrated database. Includes cropland, permanent crops, grassland, and forest with management intensity levels.
Author: Linda See (LAMASUS Project)
Contact: see@iiasa.ac.at
Database integration: Sara Pruckner
Processing level: L3 - Harmonized spatial product
Creation date: 2025-04-29
Quality status: Production quality - validated
Related datasets:
  other_years: [ibiom_LUM_europe_100m_2000, ibiom_LUM_europe_100m_2010]
  integrated_with: ibiom_forest_biomass_2010

The iBIOM spatial framework

Every 100 m layer in iBIOM sits on one grid. This is not a convention of convenience: two layers can both be in EPSG:3035 at 100 m and still be half a pixel apart, and a half-pixel offset is invisible on a map and fatal to a pixel-wise overlay. Layers are therefore warped onto the reference grid itself rather than merely reprojected to the same CRS.

The reference grid also carries the model unit systems. Rather than every consumer reprojecting polygons on the fly, GLOBIOM Simulation Units, GLOBIOM Land Units and the G4M half-degree cells are each rasterised onto the same pixels. Summing any aligned iBIOM layer by one of those rasters gives model unit totals with no reprojection, and because all systems are derived from the same pixels, totals agree across them.

Note

The grid extent quoted throughout, WGS84 [-55.514, 24.497, 60.997, 72.125] and EPSG:3035 [918900, 916700, 6552100, 5440900], is the envelope of the EEA reference grid, not the extent of the data. Data exist only on land and water pixels covered by LUM. The western bound in particular is an artefact of the grid definition and should not be read as coverage.

The 100 m reference grid

The EEA reference grid, which is the INSPIRE grid for geographical grid systems. EEA cell codes are the official iBIOM cell identifiers, a decision taken in September 2026, so any 100 m iBIOM layer can be joined, tabulated or exported by cell ID without reprojection.

  • Dataset ID: ibiom_eea_grid_cellid_europe_100m

  • Resolution: 100 m; one pixel is one hectare

  • Coordinate system: EPSG:3035 (ETRS89-LAEA Europe)

  • Grid: 56 332 columns x 45 242 rows, identical extent, transform and pixel grid to LUM v5

  • Format: Cloud-Optimized GeoTIFF, Int64, DEFLATE with predictor 2, 512 x 512 tiles, BigTIFF

  • Calibrated against: not calibrated. A definitional layer, derived from the EEA and INSPIRE grid naming rule

Cell identifiers follow the EEA code 100mE{E}N{N}, where E and N are the easting and northing of the cell’s lower-left corner in EPSG:3035 divided by 100. The raster stores them as integers, id = E * 1000000 + N, so that 100mE9189N54408 becomes 9189054408. Parent cells follow the same rule at coarser units: 1kmE{E // 10}N{N // 10} and 10kmE{E // 100}N{N // 100}.

Known limitations

The layer adds no information beyond grid position; identifiers can be computed from any aligned pixel. It exists so that a GIS user can do look-ups and table joins without writing code.

Overview levels carry the identifier of a single 100 m child cell rather than a coarser cell code. Use overviews for display only.

Int64 requires GDAL 3.5 or later. Recent QGIS, rasterio and terra read it.

GLOBIOM Simulation Units

Each 100 m pixel holds the GLOBIOM SimU it belongs to. This is the master spatial link between iBIOM layers and GLOBIOM.

  • Dataset ID: ibiom_zone_globiom_simu_100m

  • Variable: SimUID of the NUTS2-split SimU set. Always name the column globiom_simu_id; G4M uses different numbering

  • Format: Cloud-Optimized GeoTIFF, Int32, nodata -1

  • Rule: a pixel belongs to the SimU polygon containing the pixel centre

  • Mask: pixels that are nodata in LUM v5 2018 are nodata here

  • Calibrated against: not calibrated. A definitional layer; the SimU set is the GLOBIOM team’s, split by NUTS2 for iBIOM

  • Quality: 15 074 source SimUs, 15 017 with pixels. Pixel area matches polygon area to 0.0014 per cent before masking. Every SimU lies in exactly one GLOBIOM Land Unit

Known limitations

SimUs include water. The NUTS2 update first extended the original SimUs to the full LUM grid, because the original coastline had large gaps around lakes and islands, and then split them by NUTS2. Gap filling added inland water and some sea: 10.92 million hectares of the 436.92 million hectare SimU area is LUM water, and a few SimUs are mostly water. Weight by land area, not total area. The crosswalk tables provide land_area_ha for this purpose.

AllColRow is a label, not a location. For SimUs flagged changed, the AllColRow field, and the inherited altitude, slope and soil classes, describe only the original part of the SimU. The gap-filled part can lie in a neighbouring half-degree cell: 5.04 million hectares, 1.15 per cent of SimU area overall, concentrated in 969 SimUs and worst in FI1D, FI1C, SE23, SE31, UKM6, SE12, ES12, HR03 and NL23. Use the area crosswalk to move between SimUs and G4M cells, never AllColRow.

Outermost regions are outside the grid. 57 SimUs, 51 in ES70 Canarias and 6 in PT30 Madeira, totalling 824 854 hectares, have no pixels and no iBIOM data. This is the same boundary that excludes those regions from the forest area reference target described under Land Use and Management above.

GLOBIOM Land Units

Land Units are aggregations of SimUs. This layer allows iBIOM data to be summarised directly at LU level and confirms the SimU to LU nesting.

  • Dataset ID: ibiom_zone_globiom_lu_100m

  • Variable: LU_ID, stored as integer codes 1 to 911 because the source identifier is text. Join on globiom_lu_code inside iBIOM tables and report globiom_lu_id to modellers; the lookup is in the accompanying idmap table

  • Format: Cloud-Optimized GeoTIFF, Int32, nodata -1

  • Calibrated against: not calibrated. A definitional layer; the LU set is the GLOBIOM team’s, split by NUTS2 for iBIOM

  • Quality: 911 source LUs, 904 with pixels. Zero SimUs are split across LUs

Known limitations

Like SimUs, LUs include gap-filled inland water; use land area for weights. The same 824 854 hectares in Canarias and Madeira fall outside the grid, affecting 7 LUs.

G4M half-degree cells

Each 100 m pixel holds the G4M half-degree cell whose area contains the pixel centre. This links iBIOM layers to G4M without on-the-fly reprojection.

  • Dataset ID: ibiom_zone_g4m_simu_100m

  • Variable: simu_id of the G4M half-degree grid. Always name the column g4m_simu_id; these are not GLOBIOM SimU identifiers

  • Format: Cloud-Optimized GeoTIFF, Int32, nodata -1

  • Rule: the pixel centre is transformed to WGS84 and assigned to the half-degree cell containing it, with cell edges on multiples of 0.5 degrees

  • Calibrated against: not calibrated. A definitional layer; the grid is the G4M team’s

  • Quality: 167 810 global source cells, 3 024 with pixels in the European extent. Total area equals the GLOBIOM SimU layer

Known limitations

The G4M grid is global and numbers sea cells, so the layer is masked with LUM v5 2018. Without that mask every pixel, sea included, would carry a G4M identifier.

GLOBIOM SimUs and G4M cells are kept as separate spatial unit systems, a decision taken in September 2026, rather than reshaping SimUs onto the half-degree grid. Both aggregate the same 100 m pixels, so totals stay consistent. The official link between them is the area crosswalk, not the AllColRow field, which maps one to one onto G4M cells for 99.5 per cent of values but is unreliable for the changed SimUs described above.

Spatial unit crosswalks

Lookup tables linking the spatial unit systems, all derived from the same 100 m pixels so that totals agree across systems. Use them to aggregate iBIOM data to model units and to move values between GLOBIOM SimUs, GLOBIOM LUs and G4M cells without reprojecting anything.

  • Dataset ID: ibiom_crosswalks_spatial_units

  • Format: Parquet

  • Contents: per-unit area tables for each system; all pairwise unit-to-unit tables; and EEA 1 km cell to unit tables

  • Key columns: area_ha total shared area, land_area_ha excluding LUM water classes 511, 512, 521, 522 and 523, water_area_ha, and the corresponding area shares

  • Calibrated against: not calibrated. Derived by area from the zone rasters and the LUM v5 2018 land and water split

  • Key figures: 436 916 317 hectares total, of which 10 923 739 hectares are water. 9 011 SimUs are split across G4M cells; 0 SimUs are split across LUs; 1 983 G4M cells contain more than one LU

Known limitations

Only aggregation is supported: iBIOM grid to model units. Downscaling requires an allocation rule and is not covered by these tables.

Weight by land area unless water is meant to count. Total-area weights are diluted by lakes in the gap-filled SimUs.

Salt marshes, salines and intertidal flats, LUM classes 421 to 423, are counted as land. Intertidal flats are a judgement call and the tables can be rebuilt with a different water class list.

Under the pixel-centre rule, where two unit systems share a boundary a few hectares can fall on the other side. Those rows are retained so that totals remain exact.

All crosswalk tables must be rebuilt whenever a zone raster or the LUM mask changes. The accompanying summary file records the inputs and the water classes used for each build.

Status

The iBIOM-EU Database is in active development. Datasets listed under Data Available below have been produced and are accessible to IIASA researchers via the Accelerator. Datasets listed under Future Developments are in progress and will be added to this page as they are released.

Feedback from researchers and modellers using the database is welcome and actively shapes its development.

Data Available

Land Use and Management (LUM)

European land use and management at 100 m resolution for the years 2000, 2010, and 2018. The classification distinguishes management intensity within each broad land use category and covers EU27 plus the United Kingdom.

The LUM dataset has the European Wetland Map integrated directly into it and is not published as a separate LUM-EWM product. 12.6 million hectares of wetland were added and 20.7 per cent of wetland pixels were reclassified with more detailed wetland categories. Source: Tegetmeyer et al. (2025), DOI 10.5281/zenodo.14745285.

  • Resolution: 100 m

  • Coordinate system: EPSG:3035 (ETRS89-LAEA Europe)

  • Years: 2000, 2010, 2018

  • Coverage: EU27 + UK

  • Files: ibiom_LUM_europe_100m_[YEAR].tif

  • Calibrated against: FAO Forest Resources Assessment (FRA) national totals for forest area, and Eurostat for cropland and grassland area

Known limitations

The LUM aggregated files are consistent with official statistics from the Forest Resources Assessment of the Food and Agriculture Organization (FRA-FAO) for forest areas, and from Eurostat for cropland and grassland areas. This is described in LAMASUS Deliverable 2.1, “The LUM Geodatabase and Area Estimates of Land Use Change to 2018,” section 3 (pp. 12-34): https://www.lamasus.eu/wp-content/uploads/LAMASUS_D2.1_LUMGeodatabase.pdf.

FRA national totals for Spain and Portugal include the outermost regions, which fall outside the iBIOM grid extent. The reference target used for calibration excludes them explicitly; a user comparing an iBIOM forest area total against a published FRA national figure should expect that difference.

Wetland integration changes wetland pixels only. Agricultural and forest pixels are unchanged, so the wetland enhancement does not alter any agricultural or forest area total.

See LUM Classification: Full Code List for the full list of LUM classes.

Forest Aboveground Biomass 2010

JRC forest aboveground biomass data redistributed onto the LUM forest mask using NUTS2 proportional redistribution. Currently available for the year 2010 only.

  • Resolution: 100 m

  • Coordinate system: EPSG:3035

  • Year: 2010

  • Units: Mg/ha

  • Files: ibiom_forest_biomass_2010.tif

  • Calibrated against: NUTS2 biomass totals from the JRC source, redistributed across LUM forest classes in proportion to their area.

Known limitations

Validation against Avitabile gives an NUTS2 RMSE of 14.84 t/ha and correlation R = 0.921 against the reference, compared with raw JRC R = 0.997. The integrated product is closer in European total and looser per region; that trade is deliberate.

The JRC source is 2010 and the LUM forest mask is 2018, an eight-year temporal mismatch. The product is approximately 5.26 per cent conservative against the source because it preserves the LUM spatial extent.

Tree Species Shares - Brus EU_TreeMap

Tree species shares are derived from the Brus et al. EU_TreeMap, with 18 genus groups harmonised onto the LAMASUS LUM forest mask. The year in the dataset name denotes the LUM forest mask the species map was harmonised onto, not an observation year.

  • Resolution: 1000 m (1 km)

  • Coordinate system: EPSG:3035 (ETRS89-LAEA Europe)

  • Years: 2000 and 2010, referring to the LUM forest mask

  • Units: percent (0-100); the 18 shares sum to the pixel’s LUM forest fraction x 100

  • Files: ibiom_species_brus_europe_1km_2000.tif and ibiom_species_brus_europe_1km_2010.tif

  • Provenance files: the corresponding _provenance.tif single-band rasters

  • Format: one multi-band Cloud-Optimized GeoTIFF, one band per genus group, plus a separate single-band provenance raster

  • Band order: frozen alphabetical order: AbiesSpp, AlnusSpp, BetulaSpp, CarpinusSpp, CastaneaSpp, EucalyptusSpp, FagusSpp, FraxinusSpp, LarixSpp, PiceaSpp, PinesMisc, PinusPinaster, PinusSylvestris, PopulusSpp, PseudotsugaMenziesii, QuercusMisc, QuercusRoburPetraea, RobiniaSpp

  • Calibrated against: National Forest Inventory statistics and the Forest Map of Europe, inherited from the Brus source rather than applied by iBIOM. The iBIOM pipeline preserves the Brus totals and does not re-scale them to national statistics

The frozen alphabetical band order is documented because downstream code reads bands by position.

Known limitations

The NFI calibration is the Brus source’s, not iBIOM’s. Brus et al. mapped species proportions by compositional kriging where NFI plot data exist, for eighteen countries, and by multinomial multiple logistic regression elsewhere, using a soil map, a biogeographical map and temperature and precipitation bioindicators as predictors. The regression predictions were then iteratively scaled to National Forest Inventory statistics and the Forest Map of Europe. The iBIOM iterative proportional fitting enforces only the per-pixel constraint and preserves those source shares.

Two consequences for use:

  • Accuracy is spatially structured. Brus et al. report an overall accuracy of 43 per cent for the predicted dominant species: 57 per cent inside the area with NFI plot data and 33 per cent outside it. The authors attribute that difference mainly to plot density rather than to the prediction method. The product carries no flag distinguishing the two subareas.

  • National compliance is not pixel accuracy. The source authors report that the scaling to NFI statistics did not significantly decrease the Bhattacharyya distance between predicted and observed proportions at 230 held-out plots, with a median of 1.712 inside the NFI subarea and 2.131 in the ICP Level-I subarea. Totals that reconcile nationally can still be wrong at a pixel.

Because the NFI calibration is upstream and inherited, this product cannot be validated independently against national inventory species statistics; doing so would test the source against its own calibration target. Where an independent national check is required, use the JRC variant below, whose iterative proportional fitting is applied against national statistics inside the iBIOM pipeline.

Tree Species Shares - JRC tree_spec

The JRC product provides 113 tree species shares harmonised onto the LUM 2000 forest mask. The year denotes that mask and not an observation year.

  • Resolution: 1000 m (1 km)

  • Coordinate system: EPSG:3035 (ETRS89-LAEA Europe)

  • Reference period: single 2000 snapshot, referring to the LUM 2000 forest mask

  • Units: percent (0-100); the 113 shares sum to the pixel’s LUM forest fraction x 100

  • Source: JRC tree_spec (113 species), redistributed and calibrated onto the LAMASUS LUM forest mask

  • Format: one multi-band Cloud-Optimized GeoTIFF, one band per species, plus a separate single-band provenance raster

  • Band order: frozen alphabetical order defined in the production README

  • Calibrated against: national inventory statistics, applied per country and skipped where the source numbers are flagged unreliable or carry no species breakdown

The JRC variant is delivered as two files: the multi-band Cloud-Optimized GeoTIFF ibiom_species_jrc_europe_1km_2000.tif, with 113 percent-share bands, and its single-band provenance COG ibiom_species_jrc_europe_1km_2000_provenance.tif.

Known limitations

The JRC product is harmonised to the LUM 2000 forest mask; the year denotes that mask and not an observation year. IPF is skipped for countries whose source inventory numbers are flagged as unreliable or lack a species breakdown, so those countries retain the gap-filled map rather than being calibrated against distrusted national statistics. The source inventory statistics cover the 1980s-90s, so they do not represent a single common observation year.

Climate

CHELSA-W5E5 daily climate

The CHELSA-W5E5 v1.0 product is daily near-surface climate data for Europe, derived from the EU subset of the source daily netCDF and reformatted into one Apache Parquet file per year. It contains land cells only.

  • Dataset ID: ibiom_chelsa_w5e5_eu_parquet_1979_2016

  • Temporal coverage: 1979-2016; all 38 yearly files are present

  • Temporal resolution: daily; date is a YYYY-MM-DD string

  • Variables and units: tas, tasmax, and tasmin in K; pr in kg m-2 s-1; hurs in percent; rsds in W m-2; and sfcWind in m s-1

  • Spatial coverage: WGS84 extent [-24.9585, 34.0415, 44.9582, 72.9582]

  • Resolution: 300 arcsec (0.083333 degrees, approximately 9 km)

  • CRS: EPSG:4326; coordinates are explicit latitude/longitude columns

  • Format: Apache Parquet with zstd compression, one file per year

  • Files: CHELSA_EU/<year>.parquet

  • Source: CHELSA-W5E5 v1.0, DOI 10.48364/ISIMIP.836809; converted with netcdf_to_parquet.py

  • Calibrated against: not calibrated. Source values are reformatted from netCDF to Parquet without adjustment; any calibration is the source product’s own

Known limitations

Reprojection to EPSG:3035 remains an open harmonisation decision, so this product is served in its native EPSG:4326 and consumers working on the iBIOM 100 m grid must reproject it themselves.

A converted year has not yet been spot-checked against the source netCDF.

ISIMIP3b monthly climate

The ISIMIP3b product is monthly near-surface climate data for Europe in scenario/GCM/year Parquet files. It contains historical and projected data; the README records finite values over sea cells, so consumers requiring land only must apply a land mask.

  • Dataset ID: ibiom_isimip3b_eu_parquet

  • Temporal coverage: 1850-2100, scenario-dependent; historical 1850-2014 and SSP projections 2015-2100

  • Temporal resolution: monthly; date is a YYYY-MM string

  • Variables and units: tas, tasmax, and tasmin in K; pr and prsn in kg m-2 s-1; hurs in percent; huss in kg kg-1; ps in Pa; rlds and rsds in W m-2; and sfcwind in m s-1

  • Scenarios: historical, ssp126, ssp245, ssp370, and ssp585

  • GCMs: GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI-ESM2-0, and UKESM1-0-LL

  • Spatial coverage: WGS84 extent [-33.25, 25.25, 40.75, 75.25]

  • Resolution: 0.5 degrees

  • CRS: EPSG:4326; coordinates are latitude/longitude columns

  • Format: Apache Parquet with zstd compression, one file per year under <scenario>/<gcm>/<year>.parquet

  • Source: ISIMIP3b monthly EU netCDF, converted with netcdf_to_parquet.py

  • Calibrated against: not calibrated. Source values are reformatted from netCDF to Parquet without adjustment; any bias adjustment is the ISIMIP3b product’s own

Known limitations

Reprojection to EPSG:3035 and the treatment of the land mask remain open harmonisation decisions. Values over sea cells are finite rather than missing, so consumers requiring land only must apply their own mask.

Future Developments

The following data products are in active development and will be added to the database as they are released:

Forestry

  • Forest aboveground biomass for additional years beyond 2010

  • Forest increment data

  • Age class distribution. Currently a documented limitation rather than active work: no gridded prior exists in the reference data and only 4 of 25 countries yield a usable area-by-age distribution, so the source decision is deferred

Crops

Spatial crop distributions are produced with the DownScalR methodology and validated at NUTS2 against Eurostat, currently delivered to the GLOBIOM team as model parameter files and not yet published as documented database products. Nine GLOBIOM-Global crops have no European coverage; this is an accepted, documented gap at about 1.5 per cent of EU area.

Other datasets

  • Protected area data cube: in preparation

  • Soil (HWSD2 and ISRIC): in preparation

  • EU-CORDEX bias-adjusted: in preparation

  • CHELSA extreme events: in preparation

  • Water (HydroATLAS): in preparation

Access

The iBIOM-EU Database is hosted on the IIASA Accelerator. Access is currently available to IIASA researchers.

Format, coordinate system, and resolution vary by dataset; see the individual entry above and its README for the specifics of each dataset.

For platform or access questions, please contact Rupesh Singh (singhr@iiasa.ac.at). For data or integration questions, please contact Sara Pruckner.

Citation

Persistent DOI assignment for the iBIOM-derived products via the IIASA data repository is in progress. Upstream source datasets have their own DOIs and should be cited directly where the iBIOM product is a reformatting rather than a derivation. Please contact the database manager for current citation guidance when using the data in publications.

Provisional references:

  • LUM baseline: See, L. et al. (2025). LAMASUS Land Use Management 100m, European Coverage [YEAR]. IIASA, Laxenburg, Austria. DOI: 10.5281/zenodo.21787459.

  • Forest biomass: Pruckner, S. (2025). iBIOM Integrated Forest Biomass 2010. iBIOM-EU Database, IIASA, Laxenburg, Austria.

Feedback and Contributing

The iBIOM-EU Database is built to serve the research community. Feedback, suggested datasets, and integration requests are welcome. If you are working on a dataset that could be integrated into iBIOM-EU, or if you would like to use the database in your research, please get in touch.

Contact

IIASA, Schlossplatz 1, 2361 Laxenburg, Austria.