• No se han encontrado resultados

Paper

4. Conclusions

Measures of distance and accessibility are used frequently in spatial analysis to assess spatial relationships and the potential for interaction. Network

representations allow transportation infrastructure to be represented and more accurate network accessibility measures based on actual routes and services (rather than straight-line distances) can be calculated. The basic structure of a network consists of nodes and links. Links determine the connectivity between nodes, and have attributes describing the cost of traversing the link, such as distance or time. Public transport modes are the most directly amenable transport system for network representation, with stations/stops represented as nodes and public transport services represented as links between these nodes.

Data on the spatial location of public transport interchange points and service timetables can be sourced from public transport agencies. There has been significant progress in standardising the data formats of these databases and increasing availability over the web with initiatives such as NAPTAN and TransXchange (Department for Transport, 2010). These datasets do not yet include public transport fares information, and monetary costs have been excluded from the accessibility analysis here.

The representation of road and street transport infrastructure in graph form is less straightforward than rail transport. Essentially road and street infrastructure is shared by multiple modes: private cars, taxis, buses, cycles and pedestrians.

The competition for space and accessibility between these modes has many implications for the functioning of cities (as discussed in earlier chapters) as well as for the representation this system in network form. Firstly network connectivity is mode-specific, and consequently the same infrastructure may need to be abstracted differently depending on the transport mode of interest.

For instance an A-road provides high accessibility to motorised transport, but may have low accessibility for cycles and can act as a barrier for pedestrian travel. Secondly the complexity of the network and flexibility in movement of private transport leads to representational ambiguities in how certain features are to be abstracted. Irregular junctions and public space features such as squares and parks, do not directly translate into node and link structures.

Road centre-line data is a common mapping product for the representation of road networks. Essentially this vector data represents roads as lines, and junctions as nodes. The rules defining how complex junctions are represented depend on the scale of the representation. Road lines are given attributes such as the number of carriageways and the road type classification. The primary applications for this data are for vehicle routing, thus the network representation is designed for private vehicle travel. Information on infrastructure for other modes (such as pavement provision, bus lanes and cycle lanes) is generally absent, as are links that do not serve private vehicles, such as pedestrianised streets and paths1. While road centre line networks have not been designed specifically to cater for non-motorised modes, they do describe the core of the street network and can be modified to provide a useful base network for pedestrian, bus and cycle travel. For example a pedestrian network can be created by excluding road types without pavement provision (such as

motorways) and augmenting the network by adding pedestrian paths, provides a reasonable approximation of pedestrian accessibility. This approach does however overlook micro-scale issues such as severance, road crossings, bridges and underpasses. Cycle networks can also be represented in a similar vein. Data on cycle infrastructure is currently poorly catered for in the UK from

commercial products, and crowd-sourced data products such as OpenStreetMap provide a useful alternative.

Road centre line data is by no means the only possible network representation of urban transport infrastructure. A contrasting body of work has emerged from the architectural research field of space syntax (Hillier, 1996). The aim of this research is to create a network based on cognitive perceptions of urban public space. The network is formed by lines of sight, with the intention of

representing pedestrians‟ cognitive maps of street networks. Travel cost or

1 To address this issue, the Ordnance Survey released an early version of an urban pedestrian paths layer for their Integrated Transport Network data product in 2010 which was unfortunately too late to be used in

accessibility is defined by the route complexity (e.g. the number of turns represents travel cost) in contrast to distance-cost accessibility approaches.

While there have been critiques regarding how robustly and unambiguously line of sight networks can be defined (Batty, 2004), space syntax does successfully emphasise the important role that spatial cognition plays in pedestrian

accessibility.

4.4.4 Summary

The physical built-environment can be directly represented using iconic spatial models. Sources of iconic spatial data such as topographic mapping and remote sensing have been greatly improving in accuracy and sophistication in recent years. This has stimulated the development of digital city models. GIS can be used as a platform for developing built-environment models, integrating the iconic data sources with socio-economic data. This relies on accurate spatial referencing models, and spatial address referencing has also been improved and integrated with topographic modelling products. A particularly important resource for this research is real-estate data, which offers a distinct empirical perspective on urban form, function and development that can be used for detailed built-environment measures of urban density and land-use.

Network representations are a very common and powerful means of abstracting transportation infrastructure allowing topological and connectivity analysis to be performed. Road centre line data is the most common geographical network product. This is designed for motorised travel routing, and needs modification to consider non-motorised modes. Once a network has been created a range of analysis functions are possible, as described in Section 4.6.

4.5 Indicator Datasets Summary: Strengths and