The Land Use KPI classifies the land cover of a study site using the CORINE Land Cover (CLC) taxonomy and derives multiple sub-indicators: natural coverage, agricultural coverage, artificial/sealed coverage, water coverage, impermeable land share, trees per citizen, and green areas per citizen. It relies on a multi-source machine-learning pipeline that fuses up to ten global and European datasets at 10 m resolution into a single, consistent CLC class map. The result enables organisations to understand how their site footprint is distributed across habitat types, compare it against an equivalent control area, and track changes over time — supporting ESRS E3 (water) and E4 (biodiversity) disclosure requirements.
The land cover layer is produced by an ensemble machine-learning pipeline that ingests multiple raster and vector data sources, remaps each to the internal CLC class hierarchy, and generates a single categorical map per site. From this map, percentage breakdowns are computed for each macro-class.
Sub-indicator coverage percentages are calculated as follows:
where Category pixels = number of pixels classified in the target land cover category (natural, agricultural, artificial, or water), Total site pixels = total number of classified pixels within the site boundary
Data are sourced from 10 raster and vector datasets including ESA WorldCover, Sentinel-2, Google Dynamic World, and CORINE Land Cover at 10 m resolution.
Quality thresholds
| Sub-indicator | Grade A | Grade B | Grade C | Grade D | Grade E | Inverted |
|---|---|---|---|---|---|---|
| Natural Coverage | 80--100% | 60--80% | 40--60% | 20--40% | 0--20% | No |
| Artificial Coverage | 0--20% | 20--40% | 40--60% | 60--80% | 80--100% | Yes |
| Water Coverage | 60--100% | 40--60% | 20--40% | 10--20% | 0--10% | No |
| Agriculture Coverage | 0--20% | 20--40% | 40--60% | 60--80% | 80--100% | Yes |
Stacked Bar. A horizontal stacked bar chart displaying the proportional distribution of land cover categories for both Site (ROI) and Control Area (CA).
Purpose: How is the land cover distributed between natural, agricultural, artificial, freshwater, saltwater, and road surfaces — and how does the site compare to its surrounding control area?
Description: The chart shows two horizontal stacked bars — one for "Site" and one for "Control" — each divided into coloured segments proportional to the percentage share of each land cover category. A toggle above the chart switches between the "Chart" view and a detailed "Table" view. In Table mode, a three-column table lists each land cover class by Name, percentage (%), and area in Hectares, with a Copy button for export.
Land Use Distribution
Agricultural (30%)Natural (40%)Freshwater (5%)Artificial (15%)Roads (8%)Saltwater (2%)
How it's calculated: Each segment width is proportional to the percentage of the site (or control area) classified under that category. Categories are derived by remapping the full CLC class hierarchy to six display groups. Percentages are computed from the pixel counts of the generated CLC layer.
Legend:
| Category | Color | Meaning |
|---|---|---|
| Agricultural | ■ #DFBD00 | Cropland, pastures, arable land |
| Natural | ■ #228B22 | Forests, semi-natural areas, wetlands |
| Freshwater | ■ #ADD8E6 | Rivers, lakes, inland water bodies |
| Artificial | ■ #A9A9A9 | Sealed surfaces, urban fabric, industrial areas |
| Roads | ■ #AE0000 | Road network, transport infrastructure |
| Saltwater | ■ #00008B | Marine coastal and saltwater bodies |
Interpretation example:
If the Site bar shows 60% Natural and 30% Agricultural while the Control bar shows 20% Natural and 65% Agricultural, the site retains significantly more natural habitat than the surrounding landscape — a strong positive signal for biodiversity.
Progress Bar. A linear gauge comparing the natural habitat coverage percentage of the site against the control area, accompanied by a grade badge and a delta value.
Purpose: What share of the site's land surface is covered by natural habitat, and is this higher or lower than the surrounding control area?
Description: A horizontal scale from 0% to 100% shows the site score and the control area value as markers. A grade badge (A--E) appears alongside the site score. The delta value (Site minus Control) is shown as a signed number (e.g., −18.14). Below the gauge, a short description reads: "Land use distribution between the site and the control area."
Natural Coverage62%
How it's calculated: The site's natural coverage percentage is derived from the share of natural land pixels in the ROI. The grade is assigned using the thresholds: A >= 80%, B >= 60%, C >= 40%, D >= 20%, E < 20%. The delta is Site value minus Control value.
Legend:
| Level | Range | Color | Meaning |
|---|---|---|---|
| A | 80--100% | ■ #00A67A | Predominantly natural land — excellent habitat conditions |
| B | 60--80% | ■ #00DF80 | Mostly natural — good habitat quality |
| C | 40--60% | ■ #FFD21E | Mixed land use — moderate quality |
| D | 20--40% | ■ #FF8B16 | Limited natural cover — poor quality |
| E | 0--20% | ■ #FF367F | Very low natural cover — critical |
Interpretation example:
If this shows grade E with a value of 18.1% and a delta of −18.14, it means the site has substantially less natural coverage than the surrounding landscape — indicating high artificial or agricultural pressure.
Progress Bar. A linear gauge showing the proportion of soil that is waterproofed/sealed (impermeable), with site vs. control comparison.
Purpose: How much of the site surface is covered by impermeable material (concrete, asphalt, sealed rooftops), and does this exceed the surrounding area?
Description: A horizontal scale from 0% to 100% marks the site impermeability value and the control area value. A grade badge and a delta value (Site minus Control) are displayed. The description reads: "Soil waterproofing comparison site-control."
Impermeable Land Share35%
How it's calculated: Derived from the artificial coverage sub-indicator. Because higher impermeability is ecologically worse, this indicator is inverted: lower values receive higher grades. Thresholds: A <= 20%, B <= 40%, C <= 60%, D <= 80%, E > 80%.
Note: This indicator is inverted — lower values indicate better conditions.
Legend:
| Level | Range | Color | Meaning |
|---|---|---|---|
| A | 0--20% | ■ #00A67A | Minimal sealing — highly permeable soil |
| B | 20--40% | ■ #00DF80 | Low sealing |
| C | 40--60% | ■ #FFD21E | Moderate sealing |
| D | 60--80% | ■ #FF8B16 | High sealing — limited water infiltration |
| E | 80--100% | ■ #FF367F | Extreme sealing — critical for urban heat and flooding |
Interpretation example:
If the site shows 100.0% impermeability (grade E) versus a control of 79.9%, the site is entirely sealed — typical of dense urban infrastructure — representing a significant ecological liability.
Progress Bar. A linear gauge showing the share of the site area occupied by water bodies, compared against the control area.
Purpose: What percentage of the site is covered by water bodies (rivers, lakes, wetlands), and how does this compare to the surrounding landscape?
Description: A horizontal scale from 0% to 100% marks the site water coverage value and the control area value, with a grade badge and a signed delta. The description reads: "Comparison of water body presence control-site."
Water Coverage25%
How it's calculated: Derived from the water coverage sub-indicator (percentage of water body pixels in the ROI). Grade thresholds: A >= 60%, B >= 40%, C >= 20%, D >= 10%, E < 10%. The delta is Site value minus Control value.
Legend:
| Level | Range | Color | Meaning |
|---|---|---|---|
| A | 60--100% | ■ #00A67A | High water presence — strong freshwater habitat |
| B | 40--60% | ■ #00DF80 | Good water coverage |
| C | 20--40% | ■ #FFD21E | Moderate water presence |
| D | 10--20% | ■ #FF8B16 | Low water coverage |
| E | 0--10% | ■ #FF367F | Very low water presence |
Interpretation example:
If the site shows grade E with a value of 0% and a delta of −13.94, it means the site has no water bodies compared to a control area that has significant freshwater coverage.
Progress Bar. A linear gauge showing the number of trees per resident within the site area, compared to the control.
Purpose: How many trees are available per citizen in the site area, relative to the urban standard set by the control?
Description: A horizontal scale from 0 to 5,000 trees marks the site value and the control area value. A grade badge (A--E) and a signed delta (Site minus Control) are shown alongside. The description reads: "Comparison of tree density control-site."
How it's calculated: Tree count derived from the natural coverage and tree-cover CLC classes within the ROI, divided by the population estimate for the site. The delta is Site value minus Control value.
Interpretation example:
If this shows grade E with a delta of −1.84, the site has substantially fewer trees per resident than the surrounding area — typical of densely built urban districts.
Progress Bar. A linear gauge showing square metres of green space available per resident within the site.
Purpose: How much green area (m2) is accessible per citizen, and does it meet the urban greenery benchmark set by the control?
Description: A horizontal scale from 0 to 10,000 m2 shows the site value and the control value. A grade badge and a signed delta are displayed. The description reads: "Comparison of habitat density per citizen control-site."
How it's calculated: Total green area (natural and semi-natural CLC classes in the ROI, in m2) divided by the population estimate. Grade thresholds mirror the natural coverage scale. The delta is Site value minus Control value.
Interpretation example:
If the site shows grade E with a delta of −36.87 m2, it means the site has 36.87 m2 less green space per citizen than the surrounding control — well below urban greenery targets.
Line Chart. A monthly time-series chart showing the Normalized Difference Vegetation Index (NDVI) for the site, indicating vegetation density and health over the year.
Purpose: How does vegetation health and density at the site vary across the seasons, and what are the minimum, median, and maximum NDVI values?
Description: The chart shows a line plot of monthly NDVI values across the year (Jan--Dec), with minimum, median, and maximum statistics displayed as numeric badges above the chart. NDVI ranges from −1 to +1; values above 0.3 typically indicate active vegetation.
How it's calculated: NDVI is computed from Sentinel-2 annual composites for each month, aggregated over all pixels within the ROI:
where NIR = near-infrared reflectance, RED = red-band reflectance
Interpretation example:
If Min, Med, and Max all read 0.13, the site has uniformly low vegetation density year-round — consistent with a highly sealed urban surface with minimal greenery.
Map Layer. An interactive choropleth map displaying each pixel of the site area coloured by its CLC class, supported by a hierarchical class list in the left sidebar and area statistics in the right panel.
Purpose: Where exactly are the different land cover classes located within the site boundary, and what are the dominant classes by area?
Description: The map renders the CLC classification layer with each class in a distinct colour. The left sidebar shows a hierarchical list of detected CLC classes (e.g., "Urban fabric > Continuous urban fabric: 208.5 ha") with per-class hectare values and ecological indicator badges. A toolbar above the map toggles between CLC classes, natural area assessment, and impermeability overlays. The right panel shows area statistics: total area, number of classes detected, and the dominant class with its percentage. A ROI / CA toggle allows switching between site and control area views.
How it's calculated: The layer is the raw CLC classification map generated by the ML pipeline. Colours follow the CLC class colormap. Area values are computed from pixel counts multiplied by pixel area at the map resolution.
Legend: CLC classes are displayed in a continuous colour palette following the standard CORINE colour scheme. Key macro-classes:
| Macro-class | Color | Description |
|---|---|---|
| Artificial surfaces | ■ #ff092c | Urban fabric, industrial, transport |
| Agricultural areas | ■ #cb6fc0 | Arable land, permanent crops, pastures |
| Forest & semi-natural | ■ #bcff00 | Forests, scrubland, natural grassland |
| Wetlands | ■ #945566 | Inland and coastal wetlands |
| Water bodies | ■ #09d9ff | Rivers, lakes, sea, estuaries |
Interpretation example:
If the dominant class is "Continuous urban fabric" at 31.9% of the control area, with 15 different classes detected across 373.2 ha, the landscape is primarily urbanised with moderate diversity in land cover types.
Map Layer. An interactive map overlay showing the natural and agricultural areas classification — a simplified two-class view of natural versus non-natural land.
Purpose: What portion of the site and surrounding area is classified as natural or near-natural land, as opposed to agricultural or artificial?
Description: The map layer highlights natural areas in green tones and non-natural areas in grey or yellow tones. It can be toggled from the map toolbar alongside CLC and impermeability views. The right sidebar environmental KPIs panel shows the Natural Coverage score (grade and percentage) and other land-related metrics.
How it's calculated: Each pixel is classified as natural if its CLC class belongs to the natural macro-classes (forests, wetlands, semi-natural), and non-natural otherwise.
Interpretation example:
If the Natural Coverage card shows grade E at 18.1% for the CA, less than one-fifth of the broader landscape is natural — indicating strong agricultural or urban dominance.
Map Layer. An interactive map overlay showing the degree of soil sealing (impermeability) across the site and control area.
Purpose: Where are the most heavily sealed surfaces located, and how does sealing intensity vary across the study area?
Description: The map colours each pixel from low to high impermeability. A percentage value and grade card in the environmental KPIs panel (right sidebar) summarises the overall site impermeability. The toolbar toggle labelled "Impermeab." activates this layer.
How it's calculated: Pixels classified as artificial sealed surfaces are assigned high impermeability values. The percentage reflects the share of sealed pixels in the ROI.
Note: This indicator is inverted — lower impermeability values indicate better ecological conditions.
Interpretation example:
If the Impermeability KPI card shows grade D at 79.9% for the CA, nearly 80% of the surrounding landscape is sealed — a high environmental stress indicator.
| Source | Provider | Coverage | Resolution | Period |
|---|---|---|---|---|
| ESA WorldCover | ESA | Global | 10 m | 2020--2021 |
| Impact Observatory Annual LULC | Impact Observatory | Global | 10 m | 2017--2023 |
| Sentinel-2 SCL Annual Composite | ESA / Copernicus | Global | 10 m | 2017--present |
| Google Dynamic World Annual Composite | Global | 10 m | 2017--present | |
| OpenStreetMap Roads | OpenStreetMap | Global | Vector | Continuous |
| CLC+ Backbone | Copernicus | Europe | 10 m | 2018 / 2021 |
| CORINE Land Cover 2018 | Copernicus | Europe | Vector | 2018 |
| European Dominant Leaf Type | Copernicus | Europe | 10 m | 2018 |
| European Crop Map | JRC | Europe | 10 m | 2018 / 2022 |
| Lombardy Forest Type | Regione Lombardia | Lombardy | Vector | 2023 |
The land cover layer is generated by a weighted ensemble ML pipeline. For sites within Europe, all 10 data sources are fused; for sites outside Europe (global coverage), only the four global sources are used (ESA WorldCover, Impact Observatory, Sentinel-2, Google Dynamic World).
Each source raster is remapped to the internal 3Bee CLC class hierarchy. The ensemble model assigns a final CLC class to each 10 m pixel based on weighted agreement across sources. From the resulting categorical map, pixel counts per class are aggregated over the site polygon to produce percentage breakdowns for each macro-class.
The stacked bar chart groups the full CLC class hierarchy into six display categories: Agricultural (CLC 200), Natural (CLC 300 + CLC 400), Freshwater (inland water bodies), Artificial (CLC 100 excluding roads), Roads (extracted from OpenStreetMap), and Saltwater (marine and coastal water bodies).
The NDVI time series is computed separately from Sentinel-2 annual composites for each month, with values aggregated over all pixels within the ROI.
The "Trees per citizen" and "Green areas per citizen" sub-indicators divide the tree count or total green area (m2) within the ROI by the population estimate for the site.