The primary purpose of SCOOT is to control traffic signals in urban networks to optimise overall traffic performance in accordance with the traffic management policies of the local authority. However, in the process of optimisation, the traffic model within SCOOT generates a large quantity of on-line traffic data, such as flow, delay and congestion. Use of the data has been facilitated by the development of ASTRID hyperlink, which automatically collects, stores and processes traffic information for display or analysis. Data are available at an individual link level on all signal controlled links.
SCOOT detector processing
The way that SCOOT processes data from its detectors affects the information available. In SCOOT detectors are required on every link under control and so flow data is available for every link. However, to minimise costs, the flow and occupancy data from detectors is processed in a way that allows good control with only one loop per 2 lanes on a link. The result of the processing is traffic demand on the link in SCOOT's internal units, link profile units (LPUs). Whilst LPUs are good for traffic control the relationship between LPU and vehicle flow is link dependant. A good measure of vehicle flow is directly available for those links that have one loop per lane (at least all single lane approaches). For other links a basic conversion value of 17 LPU per vehicle is provided as standard, but for more accurate flow information a link specific LPU to vehicle factor should be manually measured.
Information available and data storage
ASTRID is the database used to store information derived from SCOOT systems. A standard setup will store the following data directly from SCOOT:
- Flow: flow in vehicles per hour as modelled by SCOOT
- Flow: flow in vehicles per hour derived from detectors (best for links with one detector per lane)
- Delay: total delay in vehicles per hour
- Congestion: percentage of 4 second intervals when a detector is occupied by traffic. (This value is independent of the SCOOT model)
- Emissions estimates (version 4.0 onwards only)
The current value of any of these items is also available to the user in the form of SCOOT messages.
In addition to the data directly output, ASTRID calculates and stores the following information:
- Vehicle delay: mean delay per vehicle
- Journey time: obtained by adding vehicle delay to a pre-measured 'cruise time'
- Speed: derived from link length, cruise time and vehicle delay
- Congestion index: derived from vehicle delay and cruise time
The data is available at the level of link, node, region, area or route ('route' is any pre-defined set of links). Both current and historic data is available.
Accuracy of data
The accuracy of SCOOT data has been assessed in a study for TRL carried out by Southampton University15. The main conclusions from that study were that:
SCOOT's prediction of average delay per vehicle was accurate provided that SCOOT is well validated to reflect on-street conditions. This result is robust over a wide range of link and traffic conditions predominating in Southampton. The accuracy may be expected to reduced to some extent with extreme conditions of congestion, exit blocking and mixed traffic (buses, taxis etc.) as might be found, for example, on some links in central London.
SCOOT's estimate of flow is accurate provided that link specific LPU/flow conversion flow factors are used.
N.B. The study was undertaken before the facility to derive accurate vehicle flow directly from the detector output on links with a loop per lane had been introduced.
SCOOT data is available comprehensively throughout the controlled network and can be used with some confidence for traffic management purposes provided its limitations are recognised. One such limitation is that SCOOT only models delay between the detector(s) and the stopline and does not therefore 'see' queues extending upstream of the detector(s).
Comparison with traditional traffic data
The data from SCOOT may not be as accurate as traditional statistical data at a specific count site, except for single lane links. However, the SCOOT data in an urban area will provide a much denser coverage than traditional counters. Where the standard 17 LPUs per vehicle is used, the flows should not be taken to be an exact vehicle count, but proportional changes over time should be a good representation of the changes in flow. In addition, SCOOT estimates delay and congestion on all controlled links.
It should also be noted that the delay estimate from SCOOT is an average for all vehicles on a link. It is not possible to provide separate estimates for different traffic movements unless they are separately staged.
The SCOOT total delay estimate does have the advantage over an estimate based on floating car data that the delay of all vehicles is taken into account whereas, usually only vehicles making a particular movement (e.g. straight ahead) are sampled in floating car surveys.