Elevating analysis: new roof attribution data released
My role at OS is to be a data steward and subject matter expert on our Buildings data. I work alongside my Product colleagues, to understand customer use cases and needs, and translate them into data requirements for rest of the business to work from.
Such user requirements can include the information required in the data, expected quality levels, metadata information, such as how and when the data was collected, and how it should be updated.
I also collaborate with my colleagues in Products, Data Science, Data Architecture, and Technology to design the Buildings data, establishing how it should look at different stages in the data pipeline.
New OS NGD data: roof attribution
March 2025 saw the release of new roof data into the OS NGD Buildings feature type. New roof attribution includes information about:
- primary roof material
- presence of solar panels and green roofs
- roof shape
- roof aspect
Almost all the attribution data listed above was automatically derived from the imagery OS captures, as part of our National Cyclic Revision (NCR) programme.
The NCR programme is a systematic process for updating topographic data across Great Britain. It involves the aerial imagery capture and remote surveying of all 243,241km2 of Great Britain over a three-year cycle, with more than 80,000 km2 updated annually.
The new roof attribution data can be used by our diverse customer base, for a variety of purposes. The insurance sector can use it for better estimating and validating building insurance premiums. The Department for Energy Security and Net Zero (DESNZ) can estimate the total power output of solar installations on building roofs. Start-ups using the material and shape outputs for modelling building heat loss will benefit from this data too.
The breadth of uses for this data is incredible, and while we are excited to see how our users will integrate it into their own solutions, it did provide a challenge in data design. A high-level summary of roof information may be suitable for one customer, but more technical users may require even higher levels of detail, to meet their use cases. Trying to strike the balance was tricky, but I believe we’ve done a great job with this data release.
It was an exciting and innovative project, because it was one of the first instances of OS using artificial intelligence (AI) technology such as Automated Feature Extraction (AFE) and Machine Learning (ML), to capture data that feeds into our licensed data products.
With the dataset released, looking ahead we will pay close attention to customer feedback, to ensure we can make improvements where they’re most needed. We’re already looking into the possibility of publishing a more granular dataset, to meet our more technical users’ needs.
New roof data for over 40 million buildings
The roof attribution data is part of a larger release that will enhance an already comprehensive buildings dataset:
What is the OS NGD?
The OS NGD is a single store of all Ordnance Survey’s authoritative data for Great Britain. The OS NGD delivers the richer data you need for better analysis and our download service, OS Select+Build, makes it easier to find, package and analyse the data you need, such as address and building data for emergency planning.
Our highly accurate geospatial data and printed maps help individuals, governments and companies to understand the world, both in Britain and overseas.