1.1.1 The DfT has a number of national datasets, use of which, where practical, is recommended in order to assist the process of maintaining consistency between Studies:

the TEMPRO planning data;
Census journey-to-work trip matrices;
rail matrices from CAPRI data;
indexes and depositories of roadside interview survey datasets; and
traffic counts.

Brief details of these datasets are given below.

1.1.2 There then follows some brief notes on less familiar aspects of study-specific travel demand data collection - of public transport movements, and travel by members of households.

1.1.3 The last two parts of this TAG unit deal with transport supply data

2 National Datasets

2.1 Planning Data

2.1.1 The Multi-Modal Studies may be undertaken either:

using a transport model, with demand changes controlled by exogenously defined planning data (possibly with some feedback to land-use), or
using a land-use/transport interaction model, in which case transport demand changes will be controlled by the land-use 'planning data' forecast within the model.

2.1.2 It is highly desirable that the planning data used should at some level be consistent with the DfT TEMPRO projections, so as to ensure that different regions do not engage in competitive bidding by assuming local growth rates which are implausible when summed across the whole economy. In the case of a transport-only model, the TEMPRO system provides a number of variables at a relatively detailed level, and these may either be used directly, or as constraints on a locally agreed scenario. For a land-use/transport model, the TEMPRO variables should be used (in whatever way is appropriate to the particular model package) as constraints on the totals of households, population etc across the modelled area, with the model predicting location within the modelled area.

2.1.3 Note that whilst it is a requirement for DfT appraisal that scenarios should be consistent with TEMPRO it is also desirable that studies should test proposed strategies under alternative scenarios. This is needed to confirm that recommendations are robust, i.e. that they will still be appropriate if higher or lower rates of growth materialise.

2.1.4 TEMPRO itself is a program used to distribute the results of a set of forecasts known as the National Trip End Model, or NTEM. The existing NTEM system works mainly at the local authority district level. Forecasts of population, households, workforce, and jobs for each District over 30 years have been prepared by DfT, after consultation. These forecasts are used in a series of models which forecast car ownership, car trip ends and car traffic by district. The current set of land-use forecasts was produced in 1995/96, to a geography which pre-dates the introduction of unitary authorities in England and Wales.

2.1.5 In future, more use will be made of the forecasts at the NTEM Zone level rather than the District level. The 'NTEM zoning system' divides Great Britain into around 1200 zones. Each of these zones is a group of contiguous 1991 Census wards within one local authority district. The zoning system is designed in particular to distinguish settlements by size: several urban sizes are defined by bands of population. Settlements under 10,000 population are generally not distinguished but are included within the rural area (if any) of each district.

2.1.6 The whole system is at present being revised as part of the Department's programme of work towards the next round of National Road Traffic Forecasts (NRTF). Several contracts for projects within this programme have been let or are expected to be let shortly. The work in hand or under consideration includes:

the preparation of a new set of planning data;
new models to predict trip generation and trip attraction (on a multi-mode basis, rather than dealing only with car trips); and
a new model or models to convert trip ends into traffic volumes.

2.1.7 So far as planning data are concerned, the minimum output of this programme should be a new set of forecast values for the four variables listed above, at the NTEM zone level. It is also possible that some of the variables may be available in disaggregated forms, since some of the models intended to contribute to the new NRTF are likely to require particular classifications - notably household by composition for the car-ownership model, and employment by economic sector for the trip attraction model. The new planning data should be available in Spring 2000.

2.1.8 The new TEMPRO planning data is intended to be based upon and consistent with the latest round of official projections for population and households, and will probably make use of the employment projections produced by Cambridge Econometrics. All of these are available independently at greater or lesser degrees of spatial detail. Subject to consistency in the treatment of recognised difficulties (particularly students with two addresses, and the non-household population in general) it may be possible for studies to use those underlying projections directly rather than via TEMPRO, and hence to make use of additional detail they contain (for example about economic activity by sector).

2.1.9 The use of land-use modelling to provide the 'planning data' inputs in interaction with the transport model would obviously raise a different and more specialised set of data requirements which we do not attempt to consider here.

2.2 Census Journey to Work Trip Matrices

2.2.1 As an input to the Studies the DfT can supply software to produce Census Journey-to-Work matrix extracts, by mode, in a production and attraction format suitable for transport modelling. The user defines a zoning system, specifying each zone in terms of one or more regions, counties, districts or wards. Appropriate factors are applied to convert from the 'usual journey to work' in the Census to an estimated number of trips by each mode in the selected year (1991-1998), split by household car ownership if required.

2.3 Rail Passenger Trip Matrix from CAPRI Data

2.3.1 As a further input to the Studies, the DfT has commissioned the creation of rail passenger matrices based upon the railway ticketing data set CAPRI (Computer Analysis of Passenger Revenue Information). As an initial stage the demand estimate derived from the processing of CAPRI is a station to station trip matrix.

2.3.2 It should be noted that CAPRI information has a number of serious limitations:

it does not cover tickets bought for travel on the major light rail systems of the UK, including the London Underground;
CAPRI cannot assign a point to point journey where a multi-modal ticket is bought (e.g. rail/bus/LRT travelcards);
CAPRI cannot reflect any travel that takes place without a ticket (e.g. free concessionary travel, PTE rail only travelcards) and also has limitations with flat fare concessionary travel;
output CAPRI information covers all days of the week (a limitation brought about by the lack of recorded data concerning the return leg of return ticket purchases, and the timing of use of season tickets);
CAPRI information is not segmented by purpose or car ownership; and
CAPRI data are not collected in P/A format.

These limitations will need to be considered carefully before a decision is taken to use this data set.

2.3.3 The CAPRI data processing work for the DfT has developed a representation of full OD movements, taking account of the location of trip end points (referenced to wards) relative to rail stations. This has been carried out using category analysis techniques on survey data for a limited number of stations. Profiles for observed situations have been developed and applied across the network. The authors of the work recognise that this process has severe limitations, and detailed analysis of the data for the study area in question is recommended prior to a decision to make use of this source. Further data processing for the DfT has been conducted to segment the demand by purpose and car availability.

2.4 Index of Roadside Interview Survey Data

2.4.1 The DfT has established a national roadside interview survey (RIS) database. The database provides a list of site locations for existing RIS datasets held by UK public authorities, including summary data (contact name, telephone number, date of survey etc). Outputs from the database can be spreadsheet listings, printouts of GIS maps, or a GIS point dataset. Detailed maps showing site locations are available.

2.5 Traffic Counts

2.5.1 The DfT has an extensive database of traffic counts. The Highways Agency also has a large amount of traffic count data. It is expected that full use will be made of this information.

3 Study-Specific Travel Demand Data Collection

3.1 Roadside Origin-Destination Data

3.1.1 Advice on the collection of roadside interview data can be found in the Department's DMRB, Volume 12 and in Richardson et al, 1995. Analysts should ensure that data collected for the Studies includes, as a minimum, all those variables specified in the DMRB, Volume 12, section 6.5. Additional variables may be collected, though analysts should bear in mind the need to minimise the time required to conduct each interview. Addresses should be coded either to Ordnance Survey Grid References (at least six digits) or to Post Codes. Adopting these minimal standards will ensure that data can be used by others at a later date. Re-use of data in this way could be useful for the further analysis of projects identified during the Studies and subsequently accepted for further development.

3.1.2 Analysts should note that roadside interviews cannot be conducted on motorways. In addition, it may be impractical to conduct roadside interviews on some all purpose roads, especially where they carry high levels of traffic. Relocation of interview sites may address these problems - for example, interviewing on motorway slip roads may provide an alternative to interviewing on motorways.

3.2 Public Transport Passenger Origin-Destination Data

3.2.1 There is no equivalent of the Department's RSI databank for public transport surveys. Furthermore, there is no equivalent for public transport of the guidance on survey procedures given in the DMRB, Volume 12, and so an overview of the key issues is given below.

3.2.2 Procedures for public transport data collection are much less well understood and formalised than those for road (but see Richardson et al, 1994). However, issues concerning statistical accuracy of movement estimates can be addressed in the same manner.

3.2.3 Public transport passenger surveys are generally combinations of the approaches itemised below:

on-vehicle interviews and counts;
at stop/station interviews and counts;
interviewer conducted data collection;
self completion data collection;
full route surveys;
cordon surveys; and
town centre surveys.

3.2.4 Choice of format will be dictated by the requirements of the study. Generally speaking, data quality is higher for interviewer conducted surveys. Self-completion is more appropriate for rail than for bus, where short journeys and socio-economic factors tend to give limited and biased returns. Full route surveys give complete trip matrices, whilst other methods require techniques for estimation of OD pairs that have not been directly observed.

3.2.5 Data collected in public transport passenger surveys would typically be:

origin address;
origin purpose;
mode of access to the public transport system;
destination address;
destination purpose;
egress mode from the public transport system
household car ownership;
car availability for the trip made;
parking arrangements if journey made by car;
ticket type; and
gender/age.

3.2.6 The first seven items in the above list encompasses the data items that are essential for demand modelling. The type of survey undertaken will be influenced by the nature of the study and the associated transport model design.

3.3 Bus Operator Data

3.3.1 Most operators of local bus services now operate with electronic ticket machine (ETM) systems that allow a record to be kept of passenger journeys made. The advantages and disadvantages of this data source are similar in nature to the rail CAPRI data described above. Again the advantages include a close to a 100% sample over a significant time period. A number of consultancy firms have developed software to process ETM data into trip matrix format. Operators of scheduled coach services also have computerised ticket issuing systems, although the structure of the data tends to be company specific and would require development of tailored software.

3.3.2 However, the problems associated with use of bus ETM data tend to be greater than for rail:

ticket sales data generally relate to a journey leg rather than a full journey, giving major problems with representation of trips involving interchange;
the spatial reference is the fare stage, which can typically encompass many stops;
fare stage definitions may not be consistent between parallel services operated by different companies;
concessionary travel and pre-paid ticket schemes, which are generally poorly recorded can account for up to two thirds of travel;
availability of bus ETM data is subject to operator permission and active co-operation, and data format can vary considerably by operator; and
there is no equivalent of the DfT CAPRI processing work carried out for bus data.

3.3.3 Use of bus ETM data is therefore recommended as a supplement to the passenger survey based demand estimation described below, rather than as the primary source. It is particularly useful for estimation of minor movements for which surveys would be uneconomic, and for providing aggregate measures of passenger movements for use in survey expansion, matrix validation and annualisation.

3.4 Household Survey Travel Data Sets

3.4.1 Household surveys provide the most complete picture of travel by residents of a study area, including walking and cycling. Outputs from these surveys can be segmented by the key variables of household type, person type, trip purpose, mode and time period. However, building of trip matrices directly from household surveys is not generally practical on the ground of cost - the method is inherently expensive per person trip recorded. The primary application of this type of dataset is therefore in the segmentation of demand data collected from other sources, and for use in creating local car ownership and trip end models where the national models are not thought to be appropriate.

3.4.2 Many major urban areas have been the subject of major household surveys coinciding with the 1991 Census or more recently. Included in this are London (LATS), Greater Manchester (GMATS), Merseyside and Strathclyde.

4 Transport Supply Data

4.1 Highway Network Datasets

4.1.1 A significant number of national highway network data sets are now available and they can be of use in transport model development. They range in level of detail from:

detailed road centre line representations showing all highways and details of curvature, through to
skeletal representations of major roads used for 'journey planner' packages such as the AA Milemaster and Microsoft Autoroute products.

4.1.2 The detailed data sets are not generally suitable for direct incorporation into models, as they contain a vast amount of unnecessary detail that would complicate the model and increase memory requirements and processing times. Simplification is therefore required. The level of detail in journey planners is suited to inter-urban but not intra-urban modelling.

4.1.3 The data sets contain classifications such as road type and national classification number, making it possible to infer free flow speeds and capacities. The journey planner systems contain information on link travel times that are generally an inter-peak average.

4.1.4 NARNAS is a DfT geographically referenced dataset of motorways, trunk and principal roads. It currently contains congestion indices and 1996 rotating census traffic counts and is in the process of being updated to 1998. Whilst primarily intended as a problem identification tool, this network could be of value in the construction of the strategic highway element of road traffic models, and as a source of validation data.

4.1.5 None of the above data sets contain information on junctions. Other important sources of data for road traffic modelling include:

Ordnance Survey maps;
junction plans held by local authorities and the Highways Agency;
signal timings as held by traffic control units;
site surveys involving observation of traffic and parking behaviour;
aerial photographs; and
public transport models (as a source of fixed bus flow data);

4.2 Sources of Public Transport Supply Data

4.2.1 The main sources of data for the development of public transport network representations are published timetables and service maps and computerised public transport databases developed by operators and local authorities.

4.2.2 The nature of the deregulated bus market, with service changes permitted at short notice, means that precise coding of public transport supply from published information is often not possible. However, bus networks are now more stable than in the years immediately following deregulation. Changes that do occur are often limited to minor adjustments to service frequencies and stopping points, often below the level of significance that would require an adjustment to model networks. Within London, the absence of deregulation means that networks are relatively stable and published service information is generally up to date. Data for rail network modelling is more easy to obtain than for bus and the level of detail available is generally greater. A rail timetable gives precise station to station travel times for each departure, whereas for buses timetable timing points can be several stops apart.

4.2.3 Many local authorities have compiled computerised bus network databases to assist in the process of tendering for socially necessary bus services, and in calculating concessionary travel payments. Such databases have not been compiled to a common standard and there is great variety in content and format. Service data are coded at varying levels of detail, with stop level coding being the most useful with respect to transport model building. In general, the most sophisticated databases have been created for the Metropolitan Counties by the Passenger Transport Executives. These sources of data have been used successfully as inputs to many modelling exercises. Network monitoring carried out by local authorities can provide time series data to be used as the basis for assessing whether an update of the base public transport model network is required.

5 References

Richardson A J, Ampt E S and Meyburg AH (1995). Survey Methods for Transport Planning. Eucalyptus Press.

Highways Agency Design Manual for Roads and Bridges (DMRB)

6 Document Provenance

This Transport Analysis Guidance (TAG) Unit is based on Appendix D of Guidance on the Methodology for Multi-Modal Studies Volume 2 (DETR, 2000)

Technical queries and comments on this TAG Unit should be referred to:

Integrated Transport Economic Appraisal (ITEA) Division
Department for Transport
Zone 3/08 Great Minster House
33 Horseferry Road
London
SW1P 4DR
itea@dft.gsi.gov.uk
Tel 020 7944 6176
Fax 020 7944 2198

Updated: April 2009

Documents

Guidance documents - Expert

TAG unit3.1: Modelling

TAG unit 3.1.5: Data Sources

WebTAG

Department for Transport

Transport agencies

Transport bodies

UK Government

Key