What Is a Building Footprint? Meaning, Floor Area
A building footprint is the polygon that marks where a building meets the ground, traced round the base of its outer walls. Most downloadable footprints follow the roof edge instead, traced from an aerial photo rather than surveyed on the ground. Planners, insurers and GIS teams use footprints to measure ground coverage, flood exposure and solar potential.
Almost none of the footprints you can download are that. They were traced from a photograph taken from above, so they follow the roof. That gap between the roof edge and the wall base decides which dataset answers your question, and it is where most of the trouble starts.
Footprint, roofprint, parcel and floor area: the four outlines
A footprint is a horizontal projection. It carries no height, no floors and no interior. Two buildings with the same footprint can be a bungalow and a tower.
Three other outlines get called a footprint and are not one, and each of them has cost me an afternoon at some point. A fourth question, whether a terrace is one building or eight, decides every count you make afterwards.
Footprint or roofprint
Aerial and satellite cameras see the roof. A model or a human tracing that image draws the roof edge, and the roof edge is not the wall base.
Three things pull them apart. Eaves and overhangs push the roof outward, usually by 30 to 80 cm on a house, which is a few percent of a small building's area. Lean, the sideways displacement of anything tall in an off-nadir image, moves a whole tower sideways by metres. And a car port, a veranda or a canopy has a roof with no walls under it at all.
Overture writes this into its own schema. Its buildings guide says the geometry is "expected to be the most outer footprint, roofprint if traced from satellite/aerial imagery". Google says the same thing about tall buildings in its FAQ: the model "is trained to detect building rooftop rather than base", so on high-rises the polygon sits where the roof appears, not where the building stands.
Footprint or parcel
A parcel, or cadastral boundary, is a legal object. It is the land somebody owns. The building sits inside it, usually covering a fraction of it. Parcel data comes from a land registry and changes when ownership changes, while footprints come from imagery or survey and change when the building does. Joining the two is a common and useful operation. Treating one as the other gives you a garden counted as a roof.
Footprint or floor area
Gross floor area is the footprint times the number of storeys, roughly, and it is what tax and planning rules usually care about. The building envelope is the 3D skin: footprint plus height plus roof shape. A footprint alone tells you neither. If you need floor area, you need a storey count or a height, and only some datasets carry one: IGN BD TOPO ships a height per building, and Microsoft has height on a subset.
One building, or several
Terraced housing is the case that breaks the most pipelines. A row of eight houses shares walls, so from above it is one continuous roof with no visible seam. A detector returns one polygon and a cadastre returns eight. Neither is wrong, they answer different questions, and you have to decide which one your analysis needs before you count anything.
Courtyards are holes, so a building around a courtyard is a polygon with an interior ring, and any tool that drops interior rings inflates its area. A church with a nave and a separate tower may arrive as a multipart geometry. And the canopies again: a petrol station forecourt roof is a roof over nothing.
Building footprint vs floor area vs building area
Four numbers get asked for at four different desks, and none of them stands in for another.
| Term | What it measures | Typical unit | Where it appears |
|---|---|---|---|
| Building footprint | Ground area the building covers, at wall base or roof edge | m² or ft² | GIS layer, cadastral or imagery-derived dataset |
| Building area, or building coverage | The same ground coverage, measured against the lot as a coverage ratio (also called lot or site coverage) | % of lot, or m² | Zoning code, setback and coverage rules |
| Gross floor area | Every enclosed level added up, roughly the footprint times the number of storeys | m² or ft² | Zoning code (floor area ratio), real-estate listing |
| Parcel, or lot area | The land the building sits on | m² or ft² | Cadastre, real-estate listing |
Take a two-storey house with a footprint of 90 m². Its gross floor area is roughly 90 × 2 = 180 m², before subtracting stairwells and wall thickness. Put that same house on a 300 m² lot: its building coverage ratio is 90 / 300, or 30%, and its floor area ratio is 180 / 300, or 0.6. Three different numbers, one building. Wikipedia's entry on floor area ratio sets out the coverage-ratio and floor-area-ratio definitions this table follows.
How to calculate a building footprint area
Three starting points get you there, and the arithmetic changes with each one.
From a plan
Multiply the outer length by the outer width of each rectangular section of the building, add the sections together, then subtract any courtyard or internal void. An L-shaped house splits into two rectangles. A building around a courtyard subtracts the courtyard on its own.
From a GIS polygon
Reproject the layer to a projected, metric CRS suited to the area, never EPSG:4326 (see the CRS check further down). Open the field calculator, add a new field, and enter the expression $area. QGIS returns the value in the CRS's linear unit squared, usually square metres.
From imagery, with no polygon yet
Digitize the outline by hand, or detect it first. Our building footprint extraction page covers the detection step, and our QGIS plugin runs it. Once you have a polygon, measure it with the GIS method above.
Why it matters
One building with the wrong outline changes nothing. The error compounds when you multiply it across every building in the area of interest, and by then it sits inside a number somebody is about to act on.
Solar potential wants the roof, so a roofprint is the right shape there. Ground sealing, flood exposure and setback compliance want the base, and a roofprint quietly overstates every one of them. Eaves add a few percent on a house and lean adds metres on a tower. A terrace counted as one polygon instead of eight breaks every per-building statistic downstream, and no later step in the pipeline will flag it.
What people use footprints for
Solar potential, by roof area and orientation. Flood and wildfire exposure, and the insurance pricing behind it. Population estimation where the census is old or absent. Property tax assessment and detection of unpermitted extensions. Telecoms and network planning, where a building is both an obstacle and a customer. 3D city models, extruded from footprint plus height. Damage assessment after an earthquake or a storm, by comparing two dates.
All of them need the same thing, one polygon per building, with an area you can trust.
Where to get the data
Five sources cover most needs. Everything in this table was read from the publisher's own page on 23 August 2026.
| Source | Coverage | Licence | Vintage | Geometry | How you get it |
|---|---|---|---|---|---|
| OpenStreetMap | Global, and as complete as the local mapping community made it | ODbL 1.0 | Live, whatever the map says at the moment you query it | Hand-drawn, so squared corners and real attributes, but traced from imagery in most places | QuickOSM in the plugin manager, or an Overpass query |
| Microsoft Global ML Building Footprints | Global. 1.4B buildings from Bing Maps imagery flown between 2014 and 2024, across 30,340 tiles in 225 regions | CDLA Permissive 2.0 | Refreshed 13 August 2026 | Model output. Height on a subset, confidence score on the newer footprints | dataset-links.csv per country and quadkey, or the Planetary Computer |
| Google Open Buildings v3 | Africa, South Asia, South East Asia, Latin America and the Caribbean. 1.8B detections over 58M km². Not Europe, not North America, not Australia | CC BY 4.0 or ODbL 1.0, your choice | Inference run in May 2023 | Model output, with a confidence score and a per-region precision threshold table | 178 GB of CSVs by S2 cell, Earth Engine, or 20 curated country files on HDX |
| Overture Maps buildings | Global, merged from the sources above plus Esri Community Maps and national data | ODbL | Monthly. Current release 2026-08-19.0 | Conflated. OpenStreetMap wins where it exists, model output fills the rest, matched at 50% overlap | GeoParquet on AWS or Azure, or the overturemaps CLI |
| IGN BD TOPO (France) | France and the overseas departments, the bâti theme | Licence Ouverte 2.0 | Quarterly, last published 31 July 2026 | Surveyed, metric precision, 3D with a height per building | data.gouv.fr, by department |
1.4B
buildings in Microsoft Global ML Building Footprints
README, read 23 August 2026
1.8B
detections in Google Open Buildings v3, over 58M km²
Dataset page, read 23 August 2026
2026-08-19
the current Overture buildings release, one a month
Release calendar, read 23 August 2026
Read the licences before you pick. ODbL is a share-alike licence, so a derived database goes back out on the same terms, and that clause has killed more than one client deliverable. CDLA Permissive 2.0 and CC BY 4.0 only ask for credit. Google publishes under both CC BY 4.0 and ODbL and lets you choose, which it did on purpose so that the OpenStreetMap community and everybody else could each use the data without arguing about compatibility.
The three global sources disagree about the same ground, and the disagreement is the useful part.
If you only want a file for one area, our free building footprint downloader reads all three global sources for a box you draw and hands you GeoJSON with the attribution inside it. No account, and the box caps at 25 km².
Four situations send you past all five of these, and I hit all four in a normal year. Your area is not covered, because Google stops at the edge of the Global South, Microsoft goes where Bing imagery went, and OpenStreetMap goes where somebody bothered. Or the data is older than the question: Google Open Buildings ran its inference in May 2023, so a subdivision finished in 2025 is not in it, and no amount of downloading changes that.
The other two are harder. Damage assessment and construction monitoring need two dates, and open data gives you one. And when you need a shape nobody publishes, such as roof planes for a solar study, or buildings out of your own 5 cm drone flight where a 30 cm global model has nothing useful to say, you extract your own.
How to get footprints for your own area in QGIS
Three of the four run inside QGIS. Same questions for each: what installs, whether it needs a GPU or an account, whether your imagery leaves your machine, what goes in, what comes out, the cost for real work, and one limit.
. Same questions for each: what installs, whether it needs a GPU or an account, whether your imagery leaves your machine, what goes in, what comes out, the cost for real work, and one limit.The open datasets, downloaded
- Install: Nothing, and no account.
- Runs on: Your machine, with no GPU. Nothing leaves it except the bounding box you ask about.
- In and out: In goes a box you draw. Out comes GeoJSON with the licence and the attribution written into the file.
- Price: Free.
- One limit: It returns only what the publishers already mapped, so a new estate stays missing.
Deepness, with a model you supply
- Install: Plugin manager, then it installs OpenCV and ONNX Runtime on first start. Around twenty minutes when it works. No account.
- Runs on: Your machine. CPU for small rasters, your own CUDA setup for speed.
- In and out: In goes any raster layer plus an ONNX model file you supply. Out comes a vector or raster layer.
- Price: Free, Apache 2.0.
- One limit: It ships no detection ability of its own, and a model trained on 25 cm imagery will read your 5 cm drone flight as noise without warning you.
Mapflow, by Geoalert
- Install: Plugin manager, then sign in. No GPU.
- Runs on: Their servers, so the imagery leaves your machine on every run.
- In and out: In goes an area you draw over one of their imagery sources. Out come vector layers: buildings with optional height, roads, fields, forest.
- Price: Buildings cost 10 credits per km² at $0.10 a credit, rounded up to whole square kilometres, so a 10 km² town is about $10 over your own imagery and about $13 on their Mapbox basemap, which adds 3 credits per km². The free plan gives 250 credits over at most 25 km², with export blocked.
- One limit: Terraced housing merges into one polygon, which their own screenshots show.
The cloud option
We build this one, AI Segmentation, so weigh it against the three above accordingly.
- Install: QGIS plugin manager, then a free account. No GPU, no weights download, no Python environment.
- Runs on: Our servers in Europe for Automatic and for the cloud save in Semi-Auto. The small Semi-Auto model runs locally and keeps the imagery on your machine.
- In and out: In goes a zone over any raster or basemap plus a word such as
building. Out comes a styled GeoPackage, one polygon per object, each with a class, a confidence score and an area. - Price: Free for 3 km² of automatic detection a month, then €39 a month for 200 km².
- One limit: It returns more than you asked for, so vents, skylights and courtyard sheds come back as correctly outlined polygons that are not buildings, and you filter them out by confidence and area.
Check the layer before you trust it
Five checks, in the order I run them. They take about fifteen minutes and they have saved me from shipping a wrong number more than once.
Count against a sample you digitise yourself
Pick one square kilometre. Digitise every building in it by hand. Compare the count to the layer. A ratio under 0.9 or over 1.1 means the layer answers a different question than you think, and you want to know that before the analysis rather than after.
Look hard at the attached rows
Zoom to the densest terraced street in your area of interest. Count the polygons against the front doors. This one check tells you whether the dataset splits shared walls, and it decides every per-building statistic you are about to compute.
Find the newest construction
Open the most recent development you know of. If it is absent, you have just measured the real vintage of the data, which is more honest than the date on the download page.
Check the CRS and the area units
Most of these datasets ship in EPSG:4326. Compute an area in degrees and you get a meaningless number that still looks like a number. Reproject to a metric CRS suited to the latitude, then compute.
Compare a roof to its walls
Pick one tall building. If the polygon sits off the base of the walls, you have roofprints with lean, and any ground-sealing or setback work needs a correction or a different source.
What to remember
A building footprint is the outline where the walls meet the ground. Almost every footprint you can download was traced from a photograph taken above, so it follows the roof instead.
Eaves push a roof edge 30 to 80 cm past the wall base on a house, and lean in an off-nadir image moves a tall building's roof sideways by metres.
Floor area adds up every enclosed level, roughly the footprint times the number of storeys, and a footprint itself carries no height, no floors and no interior.
An area computed in EPSG:4326 comes back in degrees and still looks like a number, so reproject to a metric CRS before you read $area.
A terrace counted as one polygon instead of eight breaks every per-building statistic downstream, and no later step in the pipeline flags it.
Try it free in QGIS, no card needed
Questions people ask
Is a building footprint the same as a roofprint?
No. The footprint is the outline where the walls meet the ground, and the roofprint is the outline of the roof above it. Eaves push the roof outward by 30 to 80 cm on a house, and lean in an off-nadir image can move a tall building's roof sideways by metres. Almost every downloadable dataset traced from imagery gives you the roofprint.
Where can I download building footprints for free?
OpenStreetMap, Microsoft Global ML Building Footprints, Google Open Buildings v3 and Overture Maps are all free, and IGN BD TOPO covers France under Licence Ouverte 2.0. Each has its own licence and its own vintage, both of which are in the table above. For one area you can also draw a box in our downloader and get GeoJSON from the three global sources at once.
Which dataset covers Europe?
Not Google Open Buildings v3, which covers Africa, South Asia, South East Asia, Latin America and the Caribbean and nothing else. In Europe you have OpenStreetMap, Microsoft, Overture, and national data such as IGN BD TOPO in France. Check the country list in the Google FAQ before you plan a project around it.
Can I use OpenStreetMap footprints in a client deliverable?
You can, under ODbL 1.0, but read the share-alike clause first. A derived database goes back out on the same terms, and that has killed more than one deliverable I have seen. CDLA Permissive 2.0 on the Microsoft data and CC BY 4.0 on the Google data only ask for credit, which is why the licence column matters as much as the coverage column.
What is the difference between building footprint and floor area?
A footprint is the ground area a building covers. Floor area adds up every enclosed level, roughly the footprint times the number of storeys. Building area, or building coverage, usually means the same ground area as the footprint, measured against the lot.
How do you calculate a building footprint area?
From a plan, multiply outer dimensions and subtract courtyards. From a GIS polygon, reproject to a metric CRS and read the $area expression. From imagery with no polygon yet, digitize or detect the outline first, then measure it the same way.
What do I do when no dataset covers my area?
Extract the footprints yourself from imagery you already have. Deepness runs an ONNX model you supply entirely on your machine, Mapflow runs on Geoalert's servers over their imagery, and our AI Segmentation takes a zone and the word building over any raster or basemap. All three are in the section above, with what each one costs.
Two pages take this further. GeoAI is the wider family these detection models belong to, and detecting building footprints in QGIS automatically is the step-by-step version of this page, with the filtering step that decides whether the layer is usable. For the tools themselves, including the local SAM-based options, see the AI plugin roundup and the QGIS AI hub.
