Towards Successful Suburban Town Centres
A study of the relationship between morphology, sociability, economics and accessibility
Please note that this project has finished. The current project is Adaptable Suburbs.
Seven selection criteria
The first criterion was to exclude centres located close to the major arterial roads of the M25 or the north/south circular, this is because their increased accessibility which results from their proximity to these roads would exert considerable influence on the town which is not experienced by the suburban towns lying more inland of these boundaries.
The confounding issue of edge effects is a problem commonly faced by all a lot of scenarios subject to boundaries. Unduly and biased results are often the consequence of influences that result from nearness to edges/boundaries. Thus the next step was to define a geographical filter to remove the influence of increased accessibility likely to be experienced by town centres close to the major roads of the north/south and M25. A filter was developed to match this criterion. An inner buffer was developed to remove centres from the sample that were located within a defined distance of the edges of the historical doughnut. An inner buffer is an area smaller than the original area (in this case the historical doughnut) by a given distance, in this example the given distance was 1500m (1.5km). Figure 5, shows the extent of the inner buffer filter for the historical doughnut, it is the region yellow in colour.
80 town centres were identified located entirely within this region. 30 town centres were excluded from the sampling process at this stage because they were located within 1.5km from the arterial circular London roads; this included centres such as Wembley in the north, Richmond and Ealing in the west, Bromley in the south and Ilford to the east.
The second, third, forth and fifth criteria used to stratify the sample was concerned with land use activity of industry and offices occurring in the town centres and its approximate walkable neighbourhood (hinterland). The identification of industrial activity in the centres hinterland supports one of the key project hypotheses: that the presence of industry and offices around a town centres edge increases the levels of town centre activity and contributes to greater co-presence (movement) on the street. These criteria were used to develop filters and sorting variables to facilitate the stratification of the town centres sample.
Address Layer 2, an Ordnance Survey product, was used to identify a number of land uses occurring in and around the town centres. Address Layer 2 is a product derived primarily from the royal mail postal address file (PAF), whereby each postal address (delivery point) has been allocated a unique reference and national grid reference. Supplementary to these data Address Layer 2 also incorporates information about geographically derived address locations that do not have specific postal addresses. The resultant dataset contains address locations (of which there are over 29 million in Great Britain (England, Scotland and Wales, but not the Isle of Man, the Channel Islands or Northern Ireland) which have a corresponding classification; residential or commercial. This classification is further broken down into more detailed classifications which were sourced from 3 different data records.
- OS field surveyor's allocation (OS base function), of which there are 1500 categories;
- Valuation Office Agency's special category code (SCAT), of which there are 400;
- National Land Use database - of which there are 13 Land Use order descriptions which are further categorised into 41 more detailed Land Use Group descriptions.
- the sampling process land use group descriptions were used to determine the number of addresses present for both Town centres and a 250 m buffer from the edge of the town centre boundary.
- The sampling process was specifically concerned with identifying the presence of manufacturing (Land use group description: U101 ) and offices (Land use group description:U102). Counts of address locations were calculated and used as a proxy indicator for the number and types of businesses located in the town centre and within a short walk (approximately 20 - 25 minutes) from its boundary.
These counts were then used to develop detailed sorting filters. In the case of the example shown in Figure 6, Wealdstone has 6 address locations categorised as manufacturing in the hinterland around the town centre and 4 address locations classed as offices in the town centre. Wealdstone also appear to have some manufacturing located within its core. Only town centres that met the outlined criteria were considered suitable for entry into our final sample and were included in the final random sampling process.
The sixth criterion used to create the sample was concerned with removing centres that were either very small or very large, in the main the project wanted to exclude large metropolitan centres such as Croydon or Watford which were extreme centres. To identify these extremes the chosen variables (Table 3) were categorised according to their inter-quartile ranges. Each town centre was assigned a classification for each variable which corresponded to their location position in variable data distribution: top quartile, inter-quartile range or bottom quartile. To give an example the total count of address locations in the town centres categorised as manufacturing had the following quartile range; lowest 25% (less than or equal to 1 address), middle 25 to 75% (greater than 1 and less than 4 addresses) and the highest 25% (greater than or equal to 4 addresses). Thus centres with four of more manufacturing addresses or one or fewer were excluded from the sample.
Using this selection all the centres that consistently fell within the top 25% of values and bottom 25% of values were excluded. So from this filter ensured a further 9 town centres were excluded from the final sampling process.
The seventh criterion used to stratify the town centre dataset ensured an even distribution across London: north, south, east and west. This was done by creating adding a flag to indicate whether the centres were located North/South of the river Thames were used to ensure the selection was fairly evenly distributed across London. A random selection was then drawn from the remaining town centres by assigning each town centre a unique identification number. A random generator was then used to select a list of 20 numbers, which corresponded to the town centres sample. The result of this sampling process produced 20 sample centres which are explored in the profiler.