Building and Building Access Feature Types
A Lightning Talk
Last updated
Was this helpful?
A Lightning Talk
Last updated
Was this helpful?
There are four feature types in the OS NGD Buildings theme. The first two have the same geometry as their equivalent features in OS MasterMap Topography Layer but all with new OS NGD attribution and are covered in the section. The third, added in September 2023, is comprised of merged Building Part features. The fourth, added in March 2025, is newly captured data about access points to certain key public buildings.
Polygon features representing either a complete separate building or a larger building where internal divisions have been dissolved to leave single feature. such as with a hospital or school.
The OS NGD Buildings data comes in two formats, CSV and GeoPackage (GPKG)
The GeoPackage format comes ready to load into the GIS package of the users choice with different gpkg files for each layer.
The CSV format is more suited for loading into a database. It has all of the same layers in csv format and all the same attribution as the GeoPackage, however instead of storing the geometry in a GIS-ready format, it is stored as a LINESTRING (Building Line), or POLYGON (Building Part/ Building) attribute with all of the vertices stored as co-ordinate pairs in the text file.
Building Features are a compound feature made up of all the connected Building Part features that comprise a large single feature, such as a hospital, school, industrial site, multi-storey car park, large residential block, sports stadium etc. This allows users to select a single Building feature that represents a complex building as opposed to selecting multiple individual Building Part features, similar in principle to Compound Structures.
There are also three cross reference tables that list the links between:
Address UPRN – Building
Building Part – Building
Site - Building
In the example here the Building feature for Raigmore Hospital is made up of 41 separate Building Parts
The Building Feature Type, new from 09/2023, represents a single compound feature for complex buildings made of multiple smaller Building Parts
There is a count of the number of Building Part features making up the larger compound Building feature
Attributes to say if the Building is in a Land Use site or not and if so, how many, in the example below there two
Attribution as to whether this is the main building and if not, what the ID of the main building is
New attribution about Building use, alongside the existing Land Use Tier A attribution from the Building Part
Counts of the number of Addresses and if they are Residential, Commercial or Other
Whether the building is connected to another building and if it is Standalone, Semi-Connected, End-Connected or Multi-connected and a count of the number of connected buildings
Definition: A text field giving a simple concise definition of what the building is.
This was added in Version 2 of Buildings in March 2024.
Provides a Building description attribute for all OS NGD Building features
This attribute is created using OS Address and Land Use Site data. There are three steps to this: identify Static Caravans; use Land Use data and Primary Sites; use addressing for all others.
The address-centric values use attribution from the address, sites, main building, building use, building type to derive a Building Description
Includes standard terminology for residential buildings e.g. detached house, mid-terraced house, end-of-terrace, etc.
Includes ‘Mixed’ description where the building has more than one use (also Building Use contains Mixed Use values)
From 03/2024 there has been an expansion of the attribution on the Building Feature Type:
Definition: The period of construction of the principal part of a building
This was added in Version 2 of Buildings in March 2024.
These attributes have been created using third party data from Verisk (pre-2021) and supplemented with OS data (2021 onwards).
It is only for NGD Buildings that contain an address. This means that if a Building Part has an address, and a construction period, and it is part of a larger Building feature with other Building Parts that do not have an address, the resulting Building feature will inherit the construction period of the single addressed part of the Building. This can create anomalies. Where a complex Building has multiple construction ages from different Building Parts, the oldest date is used.
It contains two main attributes – a period of construction (a date range) and year of construction (a specific year), where available.
The period of construction ranges vary for earlier periods and move to consistent decadal ranges from 1980 onwards. Earlier than 1919 is it difficult to identify commercial buildings so there is a catch-all range of pre-1919 as well as more defined ranges going back to 1837 mainly covering residential buildings
Additional metadata includes the third-party building age provenance which identifies the source and capture type used by Verisk e.g. Domestic EPC (extraction from another source) or Infill Modelling (the data has been modeled)
The additional metadata gives more information to make more informed decisions about the data
Definition: The material the outer skin of the principal part(s) of a building is made from.
This was added in Version 2 of Buildings in March 2024.
The primary construction material of the building which can be the observed external material, but in some cases the underlying construction, mainly depending on the source data.
This attribute is created using 3 sources: OS Data to identify Static Caravans; OS Field Surveyors for new or edited buildings and third-party data (Verisk) for all others
Only buildings that contain address are within scope. Where multiple Building Parts make up a single Building, the ‘constructionmaterial’ attribute will be derived from the Building Part with the largest area
The ‘constructionmaterial’ attribute can contain a material type, but also can be Unknown where we don’t know the material and null if the buildings are out of scope (domestic outbuildings like sheds and separate garages).
Additional metadata includes the third-party construction material provenance which identifies the source and capture type used by Verisk e.g. Domestic EPC (extraction from another source) or Infill Modelling (the data has been modeled)
Definition: An indicator whether a building has a basement or other underground occupiable space.
This was added in Version 2 of Buildings in March 2024.
This attribute is created from 3 sources: OS Surveyor, OS Address data (Address Authority) and 3rd party supplier (Verisk).
It provides two new attributes - Basement Presence (basementpresence) and Self-Contained flat presence (basementpresence_selfcontained).
Values for both these attribute are: Present, Not present, Unknown, Null. As only buildings that contain address are within scope, Null is used for buildings that are out of scope like garages, sheds, etc.
The ‘basementpresence_thirdpartyprovenance’ attribute shows where the data has come from and includes sources such as HMLR, VOA, EPCs, Historic England, CADW and Historic Environment Scotland as well as infill modelling and address analysis.
Definition: The maximum number of occupiable floors at or above ground level within a residential or office building.
This was added in Version 3 of Buildings in September 2024.
This attribute is created from 3 sources:
OS observed data (0.5% coverage) – Field / Remote Survey, Verified Customer Feedback data, or targeted desk-based data improvement
Energy Performance Certificate (EPC) data (1%) – currently only for England and Wales
OS modelled data (98.5%) – Uses a combination of building height data, Local Authority address data, EPC data (details no. of floors at UPRN level), and a Machine Learning Regression Model (utilises topographic data to more accurately predict number of floors in a locality) to derive number of floors in each building. This figure is then rounded up/down to the nearest whole number.​
The estimated conformance level is 95% for buildings with 1 to 2 floors. For buildings with 3-9 floors, 92% will have the correct number of floors recorded (±1 floor) when compared with the real world. Buildings with 10 or more floors, have an accuracy of 85% (±1 floor).
This enhancement provides a new attribute – ‘numberoffloors’ which is an integer output with values ranging from 1-99. This attribute excludes basements and plant rooms, etc. on the roofs of buildings. In addition:
Residential buildings must contain one or more residential address AND be the main building in an OS Land Use Site.
Office buildings must contain 50% or more office addresses.
Therefore, some residential buildings will not be in scope at this time or some unexpected buildings, such as cathedrals, may be in scope.
Definition: Height attribution for the building feature type.​
This was added in Version 4 of Buildings in March 2025.
This attribute is created in three steps: ​
Derive Height Data – data is derived using Ordnance Survey’s photogrammetrically created Digital Terrain Model (DTM) and Digital Surface Model (DSM).​
Derive Building Part Height – by intersecting the Building Part polygons with the DTM and DSM and, through using some additional calculations and algorithms, the associated height values for each building part can be extracted.​
Derive Building Height - ​The Building polygon inherits the most extreme absolute values from all the constituent Building Parts which make up the Building (e.g., the lowest absolute minimum value and highest absolute roof base/maximum value).​
The enhancement includes five height attributes. These are:​
height_absolutemin_m – minimum lowest point extracted from the DTM within the footprint of the Building.​
height_absoluteroofbase_m – maximum highest point extracted from the DSM where the external building walls intersect with the roof.​
height_absolutemax_m – maximum highest point extracted from the DSM within the footprint of the Building.​
height_relativeroofbase_m – the lowest ground point extracted to the tallest roof base point extracted from all constituent Building Parts combined.​
Height_relativemax_m – the lowest ground point extracted to the tallest roof maximum point extracted from all the constituent Building Parts combined.​
Definition: The primary roof material present on the building.​
This was added in Version 4 of Buildings in March 2025.
This attribute has been derived from Ordnance Survey’s imagery using a Machine Learning Model.​ An Automatic Feature Extraction algorithm has then been used to extract the primary material that the roof is comprised of
The scope is all buildings > 4sq m that are in OS topographic data. However, this is subject to the availability of imagery and height models.​
Definition: An indicator of whether solar panels are present on the roof of a building.​
This was added in Version 4 of Buildings in March 2025.
This attribute has been derived from Ordnance Survey’s imagery using a Machine Learning Model.​ An Automatic Feature Extraction algorithm has then been used to extract the presence of solar panels on building roofs.​
This enhancement provides a new attribute – ‘roofmaterial_solarpanelpresence’, which includes these values: ‘Present’, ‘Not Present’, ‘Unknown’ or ‘NULL’.​
‘roofmaterial_confidenceindicator’ is an attribute that indicates the quality of the data. These provide context on factors at time of capture which could have influenced the confidence in the data.​
The scope is all buildings > 4sq m that are in OS topographic data. However, this is subject to the availability of imagery and height models.​
Definition: Indicates whether the roof of the building is at least partially covered with vegetation. This is usually specifically installed on a growing medium with a waterproof membrane for environmental reasons.​
This was added in Version 4 of Buildings in March 2025.
This attribute has been derived from Ordnance Survey’s imagery using a Machine Learning Model.​ An Automatic Feature Extraction algorithm has then been used to extract the presence of green roofs.​
This enhancement provides a new attribute – ‘roofmaterial_greenroofpresence’, which includes these values: ‘Present’, ‘Not Present’, ‘Unknown’ or ‘NULL’.​
‘roofmaterial_confidenceindicator’ is an attribute that indicates the quality of the data. These provide context on factors at time of capture which could have influenced the confidence in the data.​
The scope is all buildings > 4sq m that are in OS topographic data. However, this is subject to the availability of imagery and height models.​
Definition: A simple description of the predominant form of the roof.​
This was added in Version 4 of Buildings in March 2025.
This attribute has been derived from Ordnance Survey’s imagery and Digital Surface Model (DSM).​ An Automatic Feature Extraction algorithm has extracted these shapes.​
‘roofshapeaspect_areapitched_m2’ and ‘roofshapeaspect_areaflat_m2’ provide a breakdown of the total roof area that are pitched or flat.​
‘roofshapeaspect_confidenceindicator’ is an attribute that indicates the quality of the data. These provide context on factors at time of capture which could have influenced the confidence in the data.​
The scope is all buildings > 4sq m that are in OS topographic data. However, this is subject to the availability of imagery and height models.​
Definition: A set of attributes that describes the key elements on the alignment(s) of a building roof.​
This was added in Version 4 of Buildings in March 2025.
This attribute has been derived from Ordnance Survey’s imagery and Digital Surface Model (DSM).​ An Automatic Feature Extraction algorithm has extracted the various aspect values.​
This enhancement provides eight new attributes, which contain the sum roof area in m2 which face in each direction, which indicate the general shape characteristic of the roof.​
‘roofshapeapspect_areainderterminable_m2’ is the sum of all roof face areas that have been unable to be calculated (for example, obscured by vegetation).​
‘roofshapeaspect_areatotal_m2’ is the sum of all roof areas, excluding those with indeterminable aspects.​
‘roofshapeaspect_confidenceindicator’ is an attribute that indicates the quality of the data. These provide context on factors at time of capture which could have influenced the confidence in the data.​
The scope is all buildings > 4m2 than are in OS topographic data. However, this is subject to the availability of imagery and height models.​
These are a new feature type within OS NGD Buildings that have been added in March 2025.​
Building Access Locations are points identified by OS where vehicles, pedestrians, or both can enter and/or leave a building.​
The data has been derived through field capture by OS Field Surveyors. and is based on externally identifiable access locations.​
They are captured for the following buildings:
Hospitals
Railway Stations
Major Transport Interchanges (airports, ferry terminals etc.)​
Stadiums and Sports Arenas (>5000 capacity)​
Shopping Centres (>50 retail units)
Conference Centres (>500 capacity)​
Concert Venues (>1000 people)​
‘access_mode’ defines the mode of transport that can use the access location. The values are ‘Pedestrian’, ‘Vehicular’, and ‘Pedestrian And Vehicular’
‘access_purpose’ defines the main function of the access location in normal circumstances. The values are ‘Emergency’, ‘Primary Public’, ‘Private’ and ‘Public’.
‘access_obstruction’ indicates whether there is a possible accessibility obstruction to the access location. The attribute can include a combination of ‘No Obstruction’, ‘Other Obstruction’, ‘Ramp’ and ‘Step’.
‘access_level’ is an indicator of the access location relative to the surface. The values are ‘Above Surface Level’, ‘Below Surface Level’ and ‘Surface Level’.
‘accessedbuildingid’ is a reference to the unique itendifier of the building that the access location is referring to.
Links that may be useful:
‘height’_confidencelevel’ is an attribute which indicates the quality of the data. It includes ‘High’, ‘Moderate’, ‘Low’, ‘Incomplete’ and ‘NULL’.​ There is on the docs platform.
This enhancement provides a new attribute – ‘roofmaterial_primarymaterial, which includes ten possible values which are
‘roofmaterial_confidenceindicator’ is an attribute that indicates the quality of the data. These provide context on factors at time of capture which could have influenced the confidence in the data. on this can be found on the docs platform.
'roofshapeaspect_shape’ indicates the predominant form of the roof and includes these values: ‘Flat, Pitched, Mixed, Unknown’.​ can be found on the docs platform
This content has been developed from what was originally a Lightning Talk PowerPoint slide set. These slides are available to PSGA members to view and download from the