Traffic Counts and Traffic Surveys

Author: Dr. Brian Slack

1. Traffic counting methods

Transport planning at all levels requires understanding of actual conditions. This involves determination of vehicle or pedestrian numbers, vehicle types, vehicle speeds, vehicle weights, as well as more substantial information such as trip length and trip purpose and trip frequency. The first group of data dealing with the characteristics of vehicle or people movement is obtained by undertaking traffic counts. Those related to measuring trips involving knowledge of origin and destination require more detailed surveys.

There is a wide range of counting methods available. It is useful to distinguish between intrusive and non-intrusive methods. The former include counting systems that involve placing sensors in or on the roadbed; the latter involve a remote observational techniques. In general the intrusive methods are used most widely because of their relative ease of use and because they have been employed for decades. The only widely used non-intrusive method is manual counting, which enjoys wide application because of its ease. Intrusive methods, however, have evolved little over the last decade, but in the US, with federal transport policy emphasis on IT solutions to traffic management, progress is being made in the development of non-intrusive methods.

The major intrusive methods include:

Bending plate: a weight pad attached to a metal plate embedded in the road to measure axel weight and speed. It is an expensive device and requires alteration to the road bed.
Pneumatic road tube: a rubber tube that is placed across the lanes that uses pressure changes to record the number of axle movements in a counter placed on the side of the road. The drawback is that it has limited lane coverage, may become displaced, and can be dislodged by snow ploughs.
Piezo-electric sensor: a device that is placed in a groove cut into the roadbed of the lane(s) being counted. This electronic counter can be used to measure weight and speed. Cutting into the roadbed can affect the integrity of the roadbed and decrease the life of the pavement.
Inductive loop: a wire embedded in the road in a square formation that creates a magnetic field that relays the information to a counting device at the side of the road. This has a generally short life expectancy because it can be damaged by heavy vehicles, and is also prone to installation errors.

The major non-intrusive methods include:

Manual observation: a very traditional method involving placing observers at specific locations to record vehicle or pedestrian movements. At its simplest, observers use tally sheets to record, but numbers, on the other hand there are mechanical and electronic counting boards available that the observer can punch in each time an event is observed. It can record traffic numbers, type and directions of travel. Manual counts give rise to safety concerns, either from the traffic itself or the neighborhoods where the counts are being undertaken.
Passive and active infra-red: a sensor detecting the presence, speed and type of vehicles by measuring infra-red energy radiating from the detection area. Typically the devices are mounted overhead on a bridge or pylon. The major limitation is the performance during inclement weather, and limited lane coverage.
Passive magnetic: magnetic sensors that count vehicle numbers, speed, and type are placed under or on top of the roadbed. In operating conditions the sensors have difficulty differentiating between closely spaced vehicles.
Microwave- Doppler/ Radar: mounted overhead the devices record moving vehicles and speed. With the exception of radar, devices they have difficulty in detecting closely spaced vehicles and do not detect stationary vehicles. They are not affected by weather.
Ultrasonic and passive acoustic: devices that sound waves or sound energy to detect vehicles. Those using ultrasound are placed overhead to record vehicle presence but can be affected by temperature and turbulence; the acoustic devices are placed alongside the road and can detect numbers and vehicle type.
Video image detection: use of overhead video cameras to record vehicle numbers, type and speed. Various software is available to analyze the video images. Weather may limit accuracy.

A recent study has examine the use of the various traffic count methods by State Departments of Transport in the US (Skszek 2001) found that less than half use any non-intrusive techniques. Part of the reason is the level of technical expertise required to operate the devices. Inductive loops are in use in all States, with very high levels of use (>90%) for pneumatic rubber tubes and piezo-electronic road sensors. Manual counts were used by 82%of the States. In terms of satisfaction with the methods, manual counts and inductive loops were rated highest. Despite the poor acceptance of the non-intrusive devices, their cost effectiveness was shown to be higher than the inductive loops (Skszek 2001, 14). This suggests that the newer devices may gain wider use once their cost effectiveness becomes more widely appreciated.

2. Surveys

Traffic counts may provide some precise information about numbers of vehicles, their type, weight or speed, but they cannot provide other data that are essential in transport planning, such as trip purpose, routing, duration etc. Collecting these data requires more extensive survey instruments. These instruments include:

Mailed questionnaires: can include a wide range of questions; is relatively cheap to administer to large numbers of people, although preparation can be expensive; the main problem is the generally low response rate.
Travel diaries: soliciting respondents to keep a diary of the trips undertaken, times, purposes, modes etc.; extremely useful instrument constrained largely by the number of people willing to complete such a detailed inventory.
Telephone surveys: with automated dialing this can achieve extensive coverage, but response rates are usually low.
Face-to-face home interviews: can overcome many of the errors based on misunderstanding of questions in mail surveys, but are extremely time-consuming and costly.

Extensive traffic surveys began to be developed in the 1950s. One of the earliest was the Chicago Area Transportation Study (CATS) that was undertaken in 1956, providing detailed O/D data on trip length, purposes, modes of travel, and travel patterns. This was followed in 1960 with the US Census’s first attempt to collect journey to work (JTW) travel data in urban areas. Other metropolitan areas in the US and Canada, including Detroit and Toronto, copied and extended the scope of such surveys in the 1960s. The growth of surveys was encouraged by the results that provided the first comprehensive snapshots of urban travel activities in a society rapidly adopting the automobile and undertaking new types of travel behavior. This was a boon to transport planning. Furthermore, much of academic understanding of travel activity in cities has been drawn from these surveys. Since then national censuses in many countries have included travel surveys in their decennial inventories, and many planning agencies update and extend the results from the national surveys with local investigations (see below).

All survey techniques represent a compromise between the objectives of the survey, the resources available, the coverage that is feasible, and the amount of data to be collected. The surveys instrument(s) that are employed depend largely on the resources available. Even national agencies find the costs of conducting national surveys are onerous. For example, it is estimated that the next daily trip survey to be undertaken in 2007 by the National Household Transport Survey (NHTS) in the US will be $14 million. Very common is the mail-back questionnaire. CATS, for example, uses a questionnaire along with a travel diary, which involves sending out a letter of introduction to selected households, distribution of the questionnaire and instructions, mailing out reminder letters, and a telephone follow-up to selected individuals to verify their information. The NHTS survey of 2007 will be based on a national telephone survey.

The degree of detail required in most travel surveys means that even the largest agencies have to rely on sampling. It is usual to target households rather than individuals, since the household is a good predictor of travel behavior. Fixing the size of the sample is an extremely important issue. Sample size determines the degree of reliability of the results, but these have to be conditioned by the resources available and the survey instruments to be employed. In its household surveys, CATS determined that 400 completed household responses would be sufficient to provide a statistically significant sample for each of the geographic units, and because it expected a 20% rate of response, it could plan for the distribution of 2,000 questionnaires in each zone. A clustered random sample of approximately 2,000 addresses in each zone was taken. For national surveys in the US, samples of 26,000 households are sought. Because of national surveys may not provide a sufficiently reliable or detailed set of data for the needs of individual States or planning agencies, these agencies frequently ‘back-on’ additional counts in their areas when national surveys are undertaken.

The main problems encountered in traffic surveys are:

Comparability between surveys. It is usually very important to compare survey results over time. This is frequently very difficult because of different sample sizes, different questions, different response rates, and different geographical collection units. These are usually major problems for studies trying to compare the results from different agencies.
Non response bias. There are significant variations in the response rates achieved by surveys. The larger the non response rate, the less reliable will be the results. A 60% response rate is sometimes considered as a threshold. Many surveys fail to achieve high rates of response, for example the 2001 NHTS survey only achieved 41%.
Coverage bias. The survey instruments frequently contain hidden biases. For example automatic telephone surveys exclude cell phone users and those without a land line connection.
Unreporting of trips. Research is now showing that surveys and travel diaries may be undercounting trips made. Some test surveys are using GPS devices to record trips and indicate that in the Kansas City survey 10% of trips were unreported and in the case of Laredo the figure was as high as 60%.