I followed these steps to determine the capacity of land near stations to accommodate new development.
Step 1: Identify Every Station in England
The mapping data is from several sources, principally the Ordnance Survey Zoomstack polygon data which provides the railway lines, station positions, woodland, land, surface water, buildings, roads and urban areas. A separate dataset, called Open Green Space, provides the location, size and type of open space (golf courses, allotments, playing fields and so on). The Environment Agency has a polygon dataset showing areas of flood risk. MHCLG (or DLUHC) publishes a regularly updated map of the UK’s green belt. And, finally, the Office of Rail and Road (ORR) has a database of all UK stations providing station codes, passenger entrance and exit data, coordinates, and so on. This is the data I used to establish the locations of each station. Note that this excludes non-Network Rail-managed facilities: London Underground stations and Manchester’s Metrolink stations do not fall within this dataset.
In the examples on this page I’ve used Harold Wood station, located within the London Borough of Havering, but the same methodology has been applied to every station for which I have data.
Step 2: Establish 800m “Potential Development Zone”
The first step was to create a maximum potential development zone around every English station, with a radius of 800m (I’ve excluded stations in Scotland and Wales). 800m is generally accepted as an average 10-minute walking distance. For brevity, I’ll refer to this as the “PDZ”.
The entire area within a ten-minute walk of a station will not be developable. Much will be occupied by existing settlements, flood zones, lakes, rivers and so on. So I’ve whittled down each of these zones to extract any bits of land I consider to be undevelopable. Of course, this process results in some stations having no space close to them which could be used for housing. In these cases, the stations have been excluded from my data.
Step 3: Public Open Space
I don’t want to advocate building on public open space (I’ve spared golf courses too), so from this PDZ I’ve extracted from these areas land which is identified as green space within the Ordnance Survey’s Open Green Space dataset.
Step 4: Remove Areas at Risk of Flooding
Flooding is a major problem in the UK, and in general, it is unwise to build homes in areas which are at risk. The Environment Agency provides a dataset which identifies those areas which are vulnerable to flooding, grading them from 1 (least risky) to 3 (worst affected), based on the likelihood of such an event occurring. The next step is to use this dataset to remove any areas which fall inside Flood Zone 2 (which includes those in Flood Zone 3).
Step 5: Omit National Parks
National Parks tend to be some distance from population centres, so there’s little to be gained by proposing development within them. So I’ve extracted all National Parks from the potential areas for development.
Step 6: Remove Railway Lines
Given that the purpose of this exercise is to establish the development capacity of areas around stations, we would need to keep the railway lines connecting the stations to enable trains to reach them. While it’s possible to build close to railway lines, I’ve kept a 30m “safeguarding” zone either side of all existing railway lines just in case.
Step 7: Remove Urban Areas
Within the Ordnance Survey Zoomstack mapping data there’s a polygon set which identifies “built-up” areas in broad detail. This is generally the outline of existing cities, towns and villages – some rural hamlets and clusters of buildings do not fall within this. But as we’re trying to identify potential areas for new development in rural areas, it makes sense to extract any urban areas from our target zones.
There’s some complexity here. Because the Urban Areas polygon data is based on geography, rather than policy, it sometimes overlaps with green belt (some settlements fall entirely within the green belt, for example). However, I’ve decided that the green belt is fair game for the purposes of this exercise, so to ensure that we’re not excluding any useful areas for redevelopment that are protected by green belt designation, I’ve extracted green belt from urban areas where the two overlap. Existing homes are excluded from my calculations by a subsequent step.
Step 8: Omit Surfacewater
In step 4, above, we extracted any areas of land which were at the risk of flooding. The Ordnance Survey Zoomstack dataset also identifies rivers and lakes which we can use to make sure that we’re not proposing new homes surrounded by water. I’ve therefore excluded any space that sits within 10m of surfacewater, ie. a pond, lake, river, or other permanent watercourse as included within the OS data.
Step 9: Remove Areas Close to Roads
Roads are noisy and polluting, so it makes sense to avoid building new homes close to them, for all sorts of reasons. In developing land close to stations it’s unrealistic to expect major highways to be diverted or closed (for the time being, at least), so I’ve extracted any areas which are 25m from a “national” roadway, or 10m from a “regional” road, as defined by the Ordnance Survey Zoomstack data. As electric vehicles become every more ubiquitous, the risks to residents of air-borne pollution will diminish – but disturbance from noise will remain.
I’ve assumed that local roads will be reconfigured to enable new development of this scale.
Step 10: Omit Any Areas Not on Land
For coastal stations a radius of 800m will inevitably mean that some of our PDZ will fall within the sea. So it makes sense to remove these areas from the calculations. At the rate we’re going, floating cities will be a reality before too long, but not just yet.
The Zoomstack dataset includes a polygon layer called “foreshore”, which includes beaches and tidal areas. I’ve excluded these too as nobody wants to live in a house built on sand.
Step 11: Removing Existing Buildings
To avoid proposing development close to existing buildings that fall outside the “urban areas” identified by Ordnance Survey, I’ve created a 25m exclusion zone using the Zoomstack “district buildings” layer. This provides a rough approximation of existing buildings, although it doesn’t differential between types of buildings. Nevetheless, this helps avoid those smaller settlements which are located within the countryside. There’s a risk that this 25m zone around existing buildings – which is a standard distance to allow between private homes – might result in some private gardens being caught in the potential development zones, but given the myriad other constraints excluded from the data, this doesn’t seem too much to worry about.
Step 12: Remove Heritage Constraints and Protected Countryside.
I’ve excluded World Heritage Sites areas of Ancient Woodland, Areas of Outstanding National Beauty (AONB), Special Protection Areas, Ramsar wetlands and Sites of Special Scientific Interest (SSSI) from the DEFRA dataset.
Step 13: Remove Stations Serving Airports
Some of those stations with the highest number of passenger movements are the ones serving airports. Although many of the country’s airports are located within largely rural areas, and using the methodology described above, would qualify for development, it’s probably not a great place to live – and in some cases, the stations themselves are located directly within the airport boundaries. For this reason, we’ve removed all of the seven stations which directly serve airports, including Heathrow, Stansted, Manchester, Gatwick, Luton, Southampton, Teeside and Southend.
Step 14: Tidying Up
With all of the constraints listed above omitted, we end up with a series of polygons showing where new homes might be developed. The resultant geometry includes a few thin slivers of land and jagged corners which are unlikely to be sensible places to build. To clean this up I’ve applied a two-step process, called “buffering” in GIS, to first contract the boundaries of the geometry by 50m, and then expand it again by the same distance. This removes most of the small, untidy little polygons smaller than 50m x 50m (a quarter of a hectare). It has the added benefit of nicely rounding the polygons to make them a bit more pleasing to the eye.
I’ve also removed all of the resultant areas with an area of less than 5ha.
Using these final shapes I’ve then used QGIS to measure the area of each set of geometry left over. Using this area we can have a go at establishing the likely housing capacity.
Note that many stations are closer than 800m to each other, and in these cases the Potential Development Zones intersect. Although I’ve listed the potential capacity for each station separately, adding these figures together inflates the total, due to double-counting. To establish a total potential development capacity for the whole of England, I’ve “dissolved” these areas in QGIS to flatten the polygons to a single entity.