17.3.5 Accessibility Modelling

Accessibility modelling can be categorised in many ways.  For the purposes of developing the STAG indicators of public transport network coverage and local access by walking and cycling (and their distribution), there are three main categories of model:

• Category 1 - Accessibility models which analyse walk times to different tiers of public transport systems or to local facilities.  In these models the PT systems are classified by mode, frequency, time of day, and types of destinations served and local facilities classified according to their function;

• Category 2 - Accessibility models which analyse travel times using PT systems.  In these models PT networks are defined using journey planning techniques and destinations are described in terms of activities, opportunities or places (e.g. town centres; and

• Category 3 - Models not primarily designed to calculate accessibility but which are used as part of the accessibility modelling process.  These include: demand models which calculate accessibility change for input to economic appraisal; land-use models which explain the spatial relationships in terms of accessibility; and activity based models which estimate behaviour based on accessibility opportunities.

Most models are developed using standard GIS, spreadsheet, database, and journey planner packages and customisation for the purposes of accessibility analysis can be helpful in some instances.  However it is also important to recognise that one of the strengths of accessibility analysis is its flexibility and custom solutions can sometimes hinder rather than help the accessibility analysis process.

Table 17.8 identifies a few of the well-known models under each category.

Table 17.8: Classification of Models by Category

Some of the main features of each category of model are described below.

Category 1 - Local Accessibility by Walking and Cycling

Local walking and cycling opportunities are usually considered by distance.  In their simplest form "crow fly" distances are used, and services are represented by some simple measure of availability.  However most models refine the assumptions about: the characteristics of the PT networks or services, including frequency, reliability, destinations served, wait times, and time of day; and the definition of the walking networks to access the services.

Computerised road network data can be used to define walking routes on footpaths following the road network.  GIS data sets can be purchased from various companies and custom approaches can be devised to define travel times using these networks.  Alternatively custom drive time software packages can be used to calculate times by setting speeds to suitable walking and cycling speeds.  In addition many of the category 2 models can also undertake these simple calculations.

Category 2 - Models which Analyse Transport Network Accessibility

Where transport network accessibility is being considered there are several main variants in modelling structure including:

The Transport System

• Representing the geography of the networks as a cost surface derived from public transport information data; and
• Representing travel time through the network as routes between zones based on journey planning algorithms or derived from other transport modelling.

People and Places

• Representing origins and destinations in terms of people and activities; and
• Representing origins and destinations as types of places e.g. town centre.

In their most highly developed form they can consider not just travel time though networks, but the scheduling of transport services and activities at trip destinations within time windows.  This level of complexity in scheduling has only been applied to date in rural situations.

The main disadvantage of these models is that their ability to represent supply/demand relationships is weak or non-existent, and they cannot assist with forecasting future changes in land-use or transport.

Category 3 - Models Designed for some other Purpose but which can be used in the Derivation of Accessibility Indicators

Demand Models

Most demand models can output some form of accessibility index.  Using time and cost skims between zones, accessibility indicators can be calculated by linking the land-use and transport data (generally using a logarithmic scale for the travel time/cost and readily available data on the land-uses and other activities available in each zone).

SPT have developed software to automate the links with the four-stage demand model SITM.  Alternatively these calculations can be undertaken on a spreadsheet/database or using proprietary software (Table 17.8)

Land-use Transport Interaction Models

Land-use transport interaction models seek to represent the most essential processes of spatial development including relevant land-uses.  A number of systems are in use and there are significant variations amongst the different models regarding overall structure, comprehensiveness, theoretical foundations, modelling techniques, dynamics, data requirements and model calibration.

Despite recent achievements in wider development and deployment these models, there remain some challenges to be met.  The transport sub-models used do not apply activity based modelling techniques but rather the traditional four-step travel demand model sequence.  They therefore have many limitations for modelling behavioural responses.

The derivation of the accessibility measures is therefore no more sophisticated than can be achieved with the four stage models but the models can output accessibility measures directly.