Nestoria Australia Blog

There’s something to be said for neatly pinning everything down to a point on a map.  It facilitates technological wonders like Augmented Reality and all sorts of useful ways of interacting with data.  But surely the step beyond pinning things to a point on a flat map is being able to pin things to a point on a 3D map…  Well guess what? It’s not something that just exists in my map-geek musings, it’s actually been rolled out.  In the seaside town of Bournemouth UK of all places, by a team from Ordnance Survey, which is Britain’s national mapping agency.

Bournemouth 3D

Glen Hart is Head of Research at Ordnance Survey, so the right person to interview to find out more about what they did, what it means and when you might realistically be able search for things on a 3D map of your own town.  Here’s what we learned:

Ordnance Survey has recently created a three dimensional map of Bournemouth, making it, in your opinion, one of the best mapped places on the planet.  Can you talk a bit first about the value of this kind of project and secondly about the laser technology involved and how it’s all put together?

The original aim of this project was to evaluate the possibility of extracting 3D building models using a merged aerial and terrestrial laser point cloud data. This project follows on from previous work where buildings have been captured using photogrammetric techniques and total station observations.
The source data was collected by aerial LiDAR from a helicopter in July 2006 which has a density of around 60 points per square meter. Also a terrestrial LiDAR data captured with 4 laser scanners mounted on a vehicle of the same area in February 2008. This was supplemented by Digital Aerial imagery captured in June 2006.
Both the aerial and terrestrial data were analysed and checked to ensure the two datasets were within the same geographic space and there were no obvious errors, in fact the fidelity of the combined laser point cloud was excellent having an RMSE of 4cm (based on high level GPS measurements).
In order to aid point and feature identification of the merged laser point cloud, it was necessary that the data was ‘painted’ using colour values from an accurate ortho-rectified digital aerial image.

The LiDAR lasers collected around 700 million individual points of light.

Bournemouth never looked so bright...

What’s order do you work in on this kind of project?  Do you start with the flat map of the ground and work upwards?  How long does it take on average to map a square metre for example?

Ordnance Survey already has a very detailed topographic map of the whole of Great Britain called OS MasterMap™.   This map captures all significant features from roads and buildings all the way down to individual mooring posts, fences and curbs.  It contains approximately 450 million individually mapped features.   We use this as the base and then work up from there.  Actually it would be more accurate to say that we then work down from the point cloud (whether it originates from a Lidar or Photogrammetric source).

It is difficult to say how much it costs per square metre as this was not the precise aim of this stage of the research at that time.   What can be said with a large degree of certainty was that it would be uneconomic at this time to capture the whole country at the level of detail that was require for Bournemouth.   One of the objectives of a future extension of this research is to determine what can be economically produced given certain sets of user requirements.   At present we are collecting and analysing a variety of user requirements for 3D data.

Can you also let us know why the team selected Bournemouth for the accolade?

There were three main reasons:
The Bournemouth area contains a good variety of topographic features within a fairly small area including a good variety of building shapes (it’s got a pier), vegetation cover and also a coastal area that includes both a beach and cliffs.
The area is very close to Ordnance Survey’s headquarters in Southampton.
The lead Researcher appears to have a particular liking for Bournemouth!

What are some of the values of creating maps with this level of detail?  And what’s the likelihood of being able to map the whole country or world this way?  And how long do you think it will be until that level of detail exists?

One of the main values of creating something with this level of detail is that it gives us a set of highly accurate data that can be used as a benchmark to compare future capture methods against.   In this respect it is unlikely that we would (or could) capture the whole country to this level of detail.  it simply would not be viable or economic to do so at present.   However, the data also enables us to automatically generate simplified versions of the model, and to test these simpler models against possible user needs, thus helping us to determine what would be both economic to capture and of value to end users. Whilst it would be foolish to say that such detail will never exist for the whole country I think it is safe to say that it won’t be for a good few years (if that degree of vagueness is allowed).

Of course it is.  Technology is obviously converging with cartography in an interesting way at the moment, do you have any predictions for future?  What do you think the next major strides forward will be?

I think what we will see initially (and indeed are seeing) are simple 3D building models that don’t fit well with the terrain.  These will be replaced by more detailed models that are better integrated with the terrain and topographic mapping and finally with internal building models.   However a limiting fact will be the cost of producing detailed building models, so it is very likely that we will see models that vary in detail with lots of buildings being represent by between one and four building heights produced automatically, and only more important buildings being modelled in much more detail, but produced by more manual methods.

I’ve recently been collecting geo-spatial data for Nestoria Australia, have you come across any interesting geo-spatial implementations recently?  Both practical ones and the weird and wonderful.

Moving away from 3D I think one of the most interesting things happening in the data world is the emergence of the Linked data web which, whilst not limited to geo-spatial data, could have a major impact on how we view and interact with geo-spatial (and other data) on the Web.   This will generate some very interesting mash-ups that go way beyond the “pins on maps” type mash-up that we see today.

Thanks very much for explaining a bit more about the reality of 3D mapping, Glen.  It’s a pretty exciting thing to think about, even if you don’t have a fixation with pinning things to maps.  You can have a look at the moving 3D model of Bournemouth here.

More about Glen Hart: Prior to joining Ordnance Survey Glen worked in real time control and command systems then moved into software reliability.  Once he’d moved into the Geographic Information Industry Glen conducted research into spatial data models that where later developed into some of the foundational aspects of OS MasterMap and the UK’s Digital National Framework.  More recently his research interests have included the Semantic Web, data integration and Vernacular Geography.

Try these similar posts

• Augmented Reality Interview with Rob Manson
• Cool Google Tools for Aus
• We’re not the only ones who love maps: Nestoria Interview with Julian Doherty
• Nestoria Australia Interview: Andrew Naish
• We love maps