Model functionality

At its simplest level, calculating accessibility measures requires only some travel time data and some land use data so the two can be linked to calculate an integrated accessibility measure as shown in figure 4.2. Alternatively the accessibility analysis can be connected to highly sophisticated land use and transport analysis and forecasting models.

Figure 4.2     Classic stages in transport modelling


To illustrate the functionality of existing demand models, table 4.1 compares the capabilities of a selection of these models in terms of:

·         transport analysis – the functionality that allows journey times and costs to be calculated

·         land use analysis – the way location choice and land use affect travel including feedback loops from transport to land use

·         the ability of different groups of people to access particular activities.

The selection of models seeks to illustrate the range of features available in existing software products.


Table 4.1      Examples of demand models with accessibility indicator functions

Model

Estimation of travel time and cost

Land use interaction

Accessibility indicator calculation and output

Transport demand models

OmniTRANS (Omnitrans International – Netherlands)

Outputs average times by all modes and real-time simulations for traffic, but not clock-time public transport options. Costs estimated from distance.

Database functionality allows accessibility impacts of land use scenarios to be compared.

Optimised to compare access for different people groups using Cube functions and mapping interfaces.

Visual-TM
(Peter Davidson Consultancy, UK)

Outputs average times for trips by all modes. Costs estimated from distance.

Land use as an input but not interactive.

Use of map-point GIS software provides a visual interface and data management for comparing impacts on different groups of people.

Cube (Citilabs, UK)

Outputs average times by all modes and real-time simulations for traffic, but not clock-time public transport options. Costs estimated from distance.

Land use as an input but not interactive.

ArcGIS interface provides mapping options for indicators.

 

Although these demand models have fairly sophisticated accessibility indicator and mapping functions there are many potential users of accessibility analysis who are unable to justify the large expense of a demand model. As a result, markets have developed for a larger and more diverse group of tools to analyse other influences on accessibility. A selection of these is described in table 4.2.

The demand models rely on public transport frequencies and estimated travel times by time of day but the majority of models in table 4.2 use actual service schedules and output accessibility results for a specified time of day and day of the week.

Table 4.2      Examples of models with accessibility indicator functions but no demand modelling capability

Model

Estimation of travel time and cost

Accessibility indicator calculation and output

Destinations are specified by model users rather than being considered as trip attractors

Accession (Citilabs, UK)

Calculates journey times based on scheduled arrival and departure times.

Various contour and continuous functions are optimised for indicator and mapping outputs.

ICON (MCRIT, Spain)

Time and distance using average speeds using road networks.

GIS-based model to optimise regional accessibility indicator calculation.

AccessMAP – (CSIR Transportek, South Africa) the AccessMAP

Based on distance using GIS systems.

GIS based with indicators originally designed for planning new public facilities such as health centres but extended to investigate transport investment options.

ABRA (Colin Buchanan and Partners, UK)

Scheduled journey times from public transport timetables.

Spreadsheet based accessibility indicator calculation.

ACCALC (Scottish Executive, UK)

Travel times and costs not calculated but taken as outputs from transport models.

Functions to assist users to specify and output indicators for analysis and mapping.

Capital – ‘Calculator for public transport accessibility in London’ (TfL, UK)

Public transport times from strategic public transport model for London and walking times estimated using GIS from distance from origin to the nearest modelled node.

Links to London’s planning and development GIS for indicator calculation and output.

PTAM (West Yorkshire Passenger Transport Executive, UK)

A hierarchy of public transport nodes is determined and walk times to these from small local areas are calculated.

Travel time for users to services.

AutoPTpath

Highly optimised routing algorithms to be able to calculate optimal journey times for very large numbers of zones. Uses scheduled departure and arrival times for public transport.

Links to GIS for mapping.

WALC (University of Westminster, UK)

ArcGIS based with travel times for walkers being estimated and weighted based on obstacles faced (eg including steep hills).

Population catchment indicators are output based on a set of destinations.

Amelia (UCL, UK)

GIS used to allow user defined attributes to be allocated to links in the network to calculate travel times.

User consultations and focus groups being used to define parameters for indicators.

 

All the models in tables 4.1 and 4.2 include three main components:

·         a relational database

·         a GIS system

·         some bespoke programming or macros to optimise functionality for a perceived market or planning need.

The balance between the above components varies. The demand models rely more heavily on bespoke programming. In some of the simpler models, the database functionality within proprietary GIS packages has been sufficient to allow the first two elements to be combined. However, for more substantial applications a separate database is required.

Accessibility models can involve very large travel-time matrices in order to provide spatially detailed zoning systems. File size limits in the software restrict the capabilities of most of these models. For example the Citilabs Accession model can cope with about 2 million OD pairs (ie about 1400 zones when mapping access to other people).

Travel time is considered in some detail within most of the models. Cost is also sometimes included but most models use simple distance-based proxies rather than actual fares for the group being considered. Other deterrents on travel are largely ignored. Models such as Capital, PTAM, WALC, Amelia and ACCALC allow other deterrents to be included within the analysis (eg to exclude links without street lighting from trips at certain times of day).

When considering neighbourhood accessibility the non-time factors can be particularly important so it is worth considering the relevant parameters in more detail.