City Evolution - digital

UK Open building footprints

 

One of the greatest impediments to innovation in the UK in terms of sustainable city development, and metadata collection, continues to be restricted access to building footprint data. 

 

Lack of discussion around this subject is largely a result of the difficulty in accessing building footprint outlines, required to store, analyse and visualise spatial metadata relating to the building stock.

 

A growing number of city authorities however, including those in the Netherlands, the US,  Canada, Iceland and Slovenia are releasing simplified building footprints and other metadatasets. This reflects an increasing global awareness of the value of granular information relating to the urban fabric, in enabling residents, government departments, visualisation specialists and those involved in research, education and business to innovate and to collectively solve urban problems, particularly those relating to the reduction of energy and the conservation of urban resources.

 

Below we see an example of the wave of building age data visualisations enabled though the release, by city and central government authorities of footprint and urban fabric metadata

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

http://tiles.flother.is/2013/reykjavik-age/

 

Ordnance Survey and Open Street Map
Ordnance Survey (OS) is the UK's main mapping supplier and began as a defence ministry operation in 1747. Over the centuries it has maintained its reputation as a producer of mapping of the highest standard and by the 1990s was considered to have the most advanced digital mapping service in the world.

 

However in recent years, largely as a result of ongoing pressure from central government to maximum profits from commercial products, OS has failed to keep up with the world's open mapping movement. Despite recently releasing a number of open datasets, the most detailed, OS Open Map Local (2015), only offers vectorised polygons at streetblock level. As such OS products cannot currently be used to collect, visualise and release citywide metadata nor tap into our ability to generate new metadatasets relating to the city's fabric though crowdsourcing knowledge.

 

In 2004, in response to the absence of OS open footprint data Steve Coast initiated Open Street Map (OSM). This free product, created for and by the community, now has worldwide reach, and has had not only a profound influence on the debate over open mapping, but has also demonstrated the urgent need for it. However over a decade has passed since OSM's appearance, and despite considerable energy expended by the OSM community in the UK, building footprint data is still piecemeal. Meanwhile OS's only vectorised building outline product, MasterMap, continues to remain prohibitively expensive for many, with licences also heavily restricting online use.

 

How can we accelerate comprehensive UK open mapping?

At CASA we are currently looking at ways to accelerate the release of comprehensive open footprint data for the UK in order to facilitate greater innovation relating to metadata collection required to inform cities' sustainability strategies.

 

Our approach is to build on the availability and advantages of the OS Open Map Local product (currently in beta form) and to tackle its key weakness, its lack of block subdivisions. OS Open Map Local, covers the whole of the UK, is derived from highly reliable OS data and has the potential to be easily updated. It therefore has a number of advantages over OSM where all new blocks and buildings must be created from scratch and where systematic updating is not possible. Our aim has been to find a way of subdividing blocks as quickly and accurately as possible to create simplified building footprints, less detailed than MasterMap but more detailed than say the LIDAR derived footprints released for Vancouver. To then release both the methodology and open samples online, and with these test methods of crowdsourcing, analysing, visualising and releasing building fabric related metadata.

 

Our primary aim is not to substitute or challenge Open Street Map's work, but instead to support the organisation's original aim of proving the value of, and campaigning for, OS open footprint release. This is essential to enable our current research on rates of change in the building stock to proceed.

 

Methodologies described below enable demonstrate how metadata can be used in the analysis of demolition rates, energy assessments, building diversity impact analysis etc.

 

The first method looks at the release of simplified comprehensive open footprints for England and Wales. The second and third methods look at detailed map creation and ways in which accuracy and speed of production map be increased

 

Creating simplified polygons for England and Wales

The Land Registry's INSPIRE polygons product can be used to split the generalised polygons within OS's Open Map Local product (OML).The method was developed by Adam Dennett at CASA and allows for rapid and fairly comprehensive release of simplified footprint polygons for England and Wales.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Camden test using OML split using INSPIRE polygons. Adam Dennett

 

CASA has received confirmation from OS that polygons generated through this method can be released as open data. However OS has noted that the INSPIRE dataset restrictions means that any polygons derived from it cannot be used for commercial purposes, without OS permission. Polygons therefore, as we understand cannot be incorporated into OSM. 

 

INSIRE is available under an Open Government Licence (OGL). This includes a requirement that the data source needs to be acknowledged and a link to this source provided. (OML terms also need to be adhered to). OS also noted that polygons have not yet been released by the Registers of Scotland.

 

The OML/INSPIRE England and Wales footprint dataset also has limitations in that majority, rather than all, building footprints are included. INSPIRE polygons cover freehold titles, not leasehold, and define the Land Registry boundary of the title. They may in certain cases therefore be different to the Topographic data and several terraced houses for example may be grouped under a single title. (As OS noted these are not legally guaranteed as they are based on Land Registry's Index polygons - not the actual title plan polygons which are definitive and which can be purchased from a re-seller of the National Polygon Dataset ie LandMark). 

 

Building polygons created using OML/INSPIRE are also simplified - though also more detailed than OS's commercial OSVectorMap Local product.  For city metadata collection this level of simplification is not a significant issue. However for research/use where a detailed understanding of the physical make up of a locality is required, and for issues relating to planning, 3D modelling etc these outlines are insufficient.

 

One possible solution which would simultaneously address issues of comprehensive cover, footprint detail and restrictions on commercial use would be to automate the extraction of footprint outlines from LIDAR data and we anticipate advances in this area. 

 

Testing alternative methods for detailed open footprint production for local areas

OML, INSPIRE and LIDAR

Work with LIDAR outlines is currently being undertaken by Duncan Hay (working on the CASA/ Survey of London Whitechapel project).  OML polygons are split with reference to both the INSPIRE cadastral parcels and the LIDAR building outlines, creating a very accurate open vectorised footprint. LIDAR images have been found to be clearer to work with than Bing satellite imagery (for which tracing is permitted). They also use British National Grid, an therefore align better with the OS OML vectors in QGIS than with Bing which being in WGS84 needs to be reprojected.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Whitechapel sample using OML, INSPIRE and LIDAR. Duncan Hay

 

OML and OS historical mapping- footprints and demolition rates 

A second method of creating accurate subdivisions of OML uses out of copyright historical mapping and is being tested by Polly Hudson. The method also enables alterations to the built fabric to begin to be tracked, and demolition rates assessed.

 

OML vectorised polygon blocks here are split using the OS National Grid (NG) 1:2500 1943-1995 series, 1950s tiles (for pre WW11 buildings), with LIDAR outlines or Bing aerials used for postwar buildings. The NG 1950s tiles are over 50 years old and are therefore out of copyright. No restrictions therefore apply where maps are owned by the user and vectorised data derived from these, along with raster images, may be uploaded online. These may be purchased as originals from map suppliers or as raster tiffs from a map supplier ie Landmark Information Group.  (Annual licence and copyright restrictions apply for maps under 50 years old). Estimates are provided on application however the costs for base data to outline around 2,500 buildings in our inner London test area is approximately £170 for non-commercial use. For universities tiles are available via Edina with permission given by OS  for vectorisation of small areas.  Feedback on vectorisation of larger areas is awaited.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Archway Map. OML split using out of copyright OS historical maps divisions. OS National Grid 1:2500 1952-4 series courtesy Edina Digimap. Polly Hudson

 

In the current trial a sample of 2,500 buildings in Archway (Islington, London) was vectorised in a day and half, using the Edina Digimap NG 2,500 1952-54 map tiles shown above  Around 70% of current party walls of the sample stock matched those from the 1950s maps though minor alterations to extensions (ie removal of outside lavatories, or additional extensions) had often occurred. Raster scanning was not employed on this sample owing to the small sample size and clean up time involved, though this will be tested on larger areas.

 

Vectorising postwar footprints also enable the proportion of stock , and individual buildings, demolished since this time to be identified assessed, and issues relating to these debated.

 

​​Once complete the open footprint map will provide the 'filing cabinet' to enable local physical metadata, unable to be accessed elsewhere,  to be collected and collated. Attribute fields are currently being idenitifed and will be tested with the Better Archway Forum. Polygons will be submitted to Open Street Map, as unlike the OML/INSPIRE method, there are no restrictions with the OML/historical mapping on commercial or third party use. 

 

Interoperability issues

OML/INSPIRE simplified polygons will begin to be used to collect citywide metadata in summer 2016 if a revised version of OML has not yet been released.

 

Data releases such as OML, LIDAR, and INSPIRE are enabling us to build on and support Open Street Map’s long-term work developing comprehensive open footprint dataset for the UK. However significant amounts of time are now being spent mashing together different open products to try and create the single open dataset we all seek. In doing so, (and as more open products are released) we are likely to produce and release an ever greater number of open footprint datasets which are not interoperable.

 

As well as taking up time that could be put to better use, we are also it seems storing up even greater problems for the future, as large amounts of metadata begin to be collected. As each month goes by without a single standardised open footprint format,  time which could be used to merge and analyse data collected from multiple sources (to address sustainable city and other issues) will be wasted, and time needed to transfer data, in the long run, will increase. 

 

Potential solutions

OS is currently developing a new suite of APIs designed to give greater access to footprint data. For non-commercial use a number of free transactions are permitted per month. However, as each query has to be run individually for each building footprint, for even small local projects the transaction quota will be rapidly used up. 

 

The OML/INSPIRE method now offers the opportunity to create comprehensive open footprint data, more detailed than OS VectorMap Local. Accuracy will be further increased once footprints are able to be extracted from open LIDAR data. 

 

However what is obviously required is a single, standardised comprehensive, regularly updated,  OS open footprint product for the whole of the UK available to, and used by everyone.

 

OS's OpenMap Local is still a beta version. It could be released with building subdivisions and simplified outlines in a way that does not challenge the precision, quality, or market of its MasterMap product. 

 

This would allow UK metadata collection, visualisation and analysis to proceed immediately. Furthermore, as TfL has shown through its enlightened approach , it would also inevitably unleash a wave of innovation, within multiple sectors, which OS would benefit from through association/accreditation, and through income generation from add-on products.