Identifying Tall Buildings

We are going to look at how the methodology developed, and the common difficulties in identifying high rises. Initially we used address classification to identify residential properties.

This was achieved using a simple database query to join the OS MasterMap Topography Layer with Building Height Attribute, before narrowing the data down to residential properties by a join with AddressBase Premium. This utilises the address classification determined by local authorities.

This gave us a rough first estimate for high rises, but there were some mistakes. It did not account for buildings where the address fell outside of the building footprint, or if there were buildings with an incorrect height assigned to them. Lets look at this AddressBase Classification Method:

While this was a good first attempt, we realised the nature of addressing required a different approach to account for errors, of which included:

• Where the point for the address of the building fell outside of the building footprint, or was assigned to the wrong part of the building, such as in a porch or smaller area.
• Where the height for the building is incorrect.
• Where the address is not assigned to the correct buildings.

Ordnance Survey and Welsh Government then worked to create a SQL script to address these errors. This was an iterative process that identified missing buildings and gradually improved the accuracy of the initial analysis.

The method captured more high rises than previously. However, there were still some missing. To capture the remaining buildings three ways to capture tall buildings were developed:

• The Missing Method - identify tall buildings where the address data is located outside of the building footprint
• The TOID Method - identify tall buildings based on the height of the building taken from the relationship between the unique identifier TOID
• The PAO Method - identify tall buildings based on the parent child relationship, where there dependent secondary addresses under a primary address (e.g. Tower House, Flat 1 Tower House)

Using these methods we were able to create a more comprehensive picture of tall buildings.

We can see a 3D comparison created in Blender comparing the first method and final method below.

A final comparison highlights the difference between the first method and third method.

This story shows OS data can be used in a variety of ways to product a more complete picture of high rise buildings, and the complexities of addressing data.