How we use 3D height data at OS

Guest blog by OS Surveyor Lee Harvey.

Imagine you’re buying a new house and are worried about the risk of flooding. Or you’re installing a mobile network, such as 5G, and want to know where to place transmitters. You’ll need to know the shape of the ground, buildings blocking line of sight, where will water flow, and a host of other things. OS create a set of 3D height products as well as our 2D maps and data.

Remote Sensing at OS is a big thing. The team spends Spring and Summer (when the weather is better… apparently!) flying up and down Great Britain capturing aerial imagery. These images are used to update our maps quickly and efficiently at head office in Southampton. By taking many overlapping images and using some air triangulation software (that’s the maths of measuring angles from the air), we match these images to their real-world location and work out the height of features on the ground. The software takes these height points and constructs a Digital Surface Model (DSM), which also forms the basis of our orthorectified imagery (a top-down perspective, map-like image). The DSM data includes buildings, trees, bridges and anything which exists at the time the photos are taken. A video game is a good example, as you move your character through any 3D environment, the hard surface that the game graphics display, will be draped over a DSM (although it’s called a Mesh in the 3D graphics industry, the fundamentals are the same). To complement this surface, and to allow lots of clever analysis of the real world, we also create a Digital Terrain Model (DTM).

A Digital Surface Model: all of the features are represented in the model.

A DTM shows the real-world terrain. This is the surface with all the objects removed. Think of it as the bare earth surface, if there was no nature or buildings. This is much more difficult to create, as with aerial imagery, there are lots of features blocking our view of the actual ground which need to be removed (although some man-made objects are kept, such as dams). Luckily at OS, we know where everything is. We can feed our DSM into data manipulation software which, using data from other products (such as buildings footprints from OS MasterMap), will remove these objects and fill in the holes. Surveyors in Remote Sensing will then model by hand where required, capturing the data in 3D stereo imagery to get accurate height values. Extra modelling could include making sure a bridge has been removed, that a dam is correctly represented, or that a road is flat.

A Digital Terrain Model: all the features have been removed from the DSM, leaving just the ground surface.

Thinking about checking a new house location for flood risk, a DTM allows computer modelling to be carried out which will show where water levels will reach and how it will travel in the landscape, mapping the chances of an area being flooded. The importance of the DTM is that in real life, a forest or building won’t stop water, so these need to be removed in the computer model too.

For engineers setting up the 5G network, a DSM will allow them to plan out where to position each transmitter. A Viewshed Analysis on a DSM shows which areas are visible from any location, by drawing a straight line away from the selected spot until it hits an object, highlighting “visible” locations. Planners can test a virtual network for signal coverage, before installing it in the real world, saving a lot of time and money. The two elevation surfaces work well together, for example, for finding out building heights. By comparing the height of a building in the DSM to the height of the ground in the DTM for that location, a quick building height can be measured.

St Paul’s Cathedral in London represented as a Digital Surface Model (image by Charley Glynn).

As the world becomes more reliant on 3D data, these surfaces become more and more important to geospatial analysis. Satellite and high-altitude imagery capture are becoming cheaper, clearer (with a higher resolution), and more accessible. A 3D surface is a vital component to exploiting this imagery, by helping computer software to automatically categorise features from images, allowing geographical analysis to be carried out faster, cheaper, and with more accuracy. The increasing use of LIDAR (a laser scanner that can find the gaps in tree cover and be flown at night) is allowing for more accurate surface capture. The Point Cloud data format allows us to attach more information to objects in the DSM, enhancing its capability for questioning the world around us. Data collected from road-based scanners and drones will increase the detail of our surfaces for use with self-driving vehicles. These digital surfaces already underpin much of the products we produce, including the 2D paper maps. Remote Sensing is already at the forefront of digital surface creation and will keep OS there for the foreseeable future.

Snowdon in Wales shown with OS imagery draped over an OS height model (image by Charley Glynn).


Find out more about the variety of being a field surveyor and learn about our flying unit.


You may also like

OS Flying Team keep mapping Britain’s changes
A glimpse into the role of an OS surveyor
Meet the team: Mark Cuthbert
Surveying coastal changes by drone 

3 Responses

  1. Alicja Karpinska

    Nicely put together and very interesting – great work!
    Also really cool visualisation of Snowdon by Charley Glynn.

Leave a Reply

Your email address will not be published. Required fields are marked *

Name* :

Email* :