14367_02

CHAPTER 2
Traffic Data Collection Needs
2.1 Introduction
Traffic data collection programs vary widely. The scope and execution of these programs are functions of
both the needs of traffic data users and available resources. In order for traffic data programs to be run
efficiently, annual estimates are needed of the traffic counts that are collected and of the field and office
personnel required to collect and publish the data. Data collection agencies should document their needs
and perform periodic reviews to identify and make process refinements. This chapter addresses how an
agency should estimate its traffic data collection needs.
In designing and implementing a traffic data program, an agency must often balance Federal require-
ments for traffic data with customer needs while at the same time ensuring efficient resource allocation.
The two principal guidance documents for traffic data collection at the national level are the Traffic
Monitoring Guide (TMG) and the Highway Performance Monitoring System (HPMS) Field Manual.
Both documents are typically referenced in or provide the foundation for a state’s Traffic Monitoring
System for Highways (TMS/H).
2.1.1 Role of the Traffic Monitoring Guide (TMG)

The TMG, published by the Federal Highway Administration’s (FHWA) Office of Highway Policy In-
formation, provides recommendations for developing new or refining current traffic data programs. The
TMG’s intended audience is state departments of transportation (DOT), with metropolitan planning
organizations (MPO), and local transportation agencies as secondary users.
The TMG provides specific examples of how statewide data collection programs should be structured,
describes the analytical logic behind that structure, and provides information highway agencies require
to optimize the framework for their particular organizational, financial, and political structures. It is the
basic document for providing guidance about what traffic data to collect. The TMG is not a prescriptive
reference. Instead it recognizes states’ uniqueness and provides a flexible framework for states to develop
their programs in a sound statistical manner.
GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-1

CHAPTER 2 TRAFFIC DATA COLLECTION NEEDS
2.1.2 Role of the Highway Performance Monitoring System (HPMS)

Field Manual
The HPMS is an integrated database that contains data describing the nation’s highway system. It is
used by FHWA for a variety of public purposes, including reports to Congress and the apportionment
of Federal-aid highway funds to the states. Data in this database is provided primarily by the states,
using procedures specified by FHWA in the HPMS Field Manual.1 These data include summary traffic
data for each state and more detailed data for a sample of highway sections in the state. The HPMS
Field Manual is a procedural reference which defines the type of data to be collected, the data format for
transmittal, and the sampling rate. Traffic monitoring procedures are based on those presented in the
TMG and are discussed in one of the Field Manual’s appendices.
2.1.3 Growing Importance of Operations and ITS Data

Since the 1990s, the role of traffic operations and traffic management and traveler information centers
as data contributors to the statewide data collection program has increased. These centers collect large
amounts of data in very small time increments—some under one minute. These data typically include
volume data, but also may include speed or vehicle classification data. These data usually are not inte-
grated into traffic planning databases. The integration of operations data presents both technical issues
and institutional barriers and this lack of integration may represent opportunity losses and higher over-
all data collection costs as a result of duplication.
2.1.4 Chapter Organization

This chapter is organized into four topic areas: discussion of the types of traffic counts and needs (2.2);
process and considerations in estimating the traffic count needs (2.3); mechanisms to reduce count
needs (2.4); and estimation of staff and resource requirements (2.5).
2.2 Traffic Count Types and Needs
This section describes the reasons state departments of transportation need a traffic data collection
program. It also discusses the basics related to specific types of counts.
Data characterizing the use of public roadways have wide application in engineering, planning, finance,
and government. Agencies collect use characteristics by counting traffic volumes, classifying vehicle
types, weighing trucks and their axles, and monitoring point speeds. Table 2-1 provides examples of
some of the uses of this data.
1
FHWA. Highway Performance Monitoring System Field Manual,December 2000.
2-2 GUIDELINES FOR TRAFFIC DATA PROGRAMS

State traffic data programs collect and publish information for many customers. Customers range from
Federal, state, and local governments to the general public. The primary source of data for these custom-
ers is the coverage count program. The coverage count program is the predominant traffic count need.
Other needs include Federal/state research initiatives, project-related counts, special count requests,
performance measurement counts, and data obsolescence counts. Table 2-2 shows examples of the types
of data used by various customers.
2.2.1 Coverage Counts

The coverage count program involves the collection of short-term and continuous vehicle weight, clas-
sification, volume, and speed data for the production of system-level traffic estimates. System-level refers
to traffic estimation for the road system and classifications of roadways, rather than for site-specific traf-
fic estimates. The TMG and HPMS Field Manual provide requirements and guidance for structuring
the coverage count program.
Coverage count programs typically include weigh-in-motion, vehicle classification, volume, and speed
data collection programs. The short-term coverage count sample size recommended for traffic volume
and vehicle classification, is based on a system-wide statistical sample. The short-term volume, vehicle
classification, and weigh-in-motion counts, and supplemental counts taken within the special needs part
of the program comprise the coverage count program.
Weigh-in-Motion (WIM)
Weigh-in-motion systems record dynamic (in-motion) vehicle weights, vehicle classification, volume,
and speed. WIM systems require a significant investment relative to other traffic data collection systems.
WIM investments include the pavement structure, sensors and electronics, and equipment calibration
and maintenance. This automated equipment can collect data continuously or for short-term periods.

Table 2-1. Examples of Highway Traffic Data Uses*
Highway Traffic Vehicle Truck Speed

Activity Counting Classification Weighing Monitoring
Engineering Highway Geometry Pavement Design Pavement Design Traffic Control

Devices
Engineering Benefit of Highway Cost of Vehicle Benefit of Truck Costs Associated with
Economics Improvements Operation Climbing Lane Congestion
Finance Estimates of Highway Highway Cost Highway Cost User Travel-Time
Revenue Allocation Allocation Costs
Legislation Selection of Highway Speed Limits and Speed Limits
Routes Oversize Vehicle
Policy
Maintenance Selecting the Timing Selection of Pavement Work Zone Safety
of Maintenance Maintenance Management Measures
Activities
Operations Signal Timing Development of Weight Setting Speed Limits
Control Strategies Enforcement
Activities
Planning Location and Design Forecasts of Travel Congestion
of Highway Systems by Vehicle Type Monitoring Systems
Environmental Air Quality Analysis Forecasts of Emissions Project-level Analyses
Analysis Type of Vehicle
Safety Design of Traffic Safety Conflicts Due Design of Safety
Control Systems and to Vehicle Mix and Systems
Accident Rates Accident Rates
Statistics Average Daily Traffic Travel by Vehicle Average Weight 85th Percentile
Type by Vehicle Class
Private Sector Location of Service Marketing Keyed to Trends in Freight Accessibility to
Areas Particular Vehicle Movement Service Areas
Types
*Adapted from FHWA, Traffic Monitoring Guide, May 2001.

Table 2-2. Examples of Customer Needs for Traffic Data

Traffic Vehicle Truck Speed
Customer
Counting Classification Weighing Monitoring
General Public Site-Specific AADT Vehicle Mix Overweights Speed Distributions
Businesses Site-Specific AADT Percent Heavy
Commercial
Advertising Agencies AADT Speed Distributions
Travel Peaks
Realtors Site-Specific AADT
Journalists Busiest Roads/
Times of Year
Developers Peak-Hour Data
Tourism Bureau AADT Vehicle Mix
Travel Peaks
Economic Development AADT
Agencies
Chambers of Commerce Count Maps
Consulting Engineers AADT Vehicle Mix Speed Distribution
Turning Movements 85th Percentile
University Researchers Hourly Volumes Axle Factor Load Spectra Speed Distributions
Adjustment Factors Vehicle Mix
Traffic Engineers Hourly Volume Hourly Vehicle 85th Percentile
Mix 10 mph Pace
Highway Designers DHV Vehicle Mix
Directional Factor
Safety Engineers AADT Speed Distributions
Transportation Planners AADT
Pavement Designers Future AADT Truck Mix Load Spectra
Future AADTT 18K ESAL
Motor Carrier AADTT Percent Heavy Overweights
Commercial
Emergency Operations Real-Time Volumes Real-Time Speeds
Centers
Metropolitan or Rural AADT, AAWDT, Percent Heavy Speed Distributions
Planning Organizations AAWET Commercial
Road and Bridge AADT Vehicle Mix Load Spectra
Authorities
State Environmental VMT Truck Classes Speed Distributions
Agency
State Highway Patrols Truck ADT Vehicle Mix Overweight Speed Profiles
Vehicles
State Parks/ AADT Vehicle Mix Load Spectra
Wildlife Agency
Legislatures or VMT Axle and Gross 85th Percentile
Assemblies Vehicle Weights
Tribal Nations VMT Vehicle Mix Load Spectra
Federal VMT Vehicle Mix Load Spectra 85th Percentile

Vehicle Classification
The vehicle stream is classified through either manual observation or use of automated equipment.
Manual observation by its nature is short-term and is based on the number of axles and the vehicle’s
body style. Automated equipment can be used continuously or for short-term periods and classifies
vehicles through either axle distances or total vehicle length. Vehicle classification data also can be ag-
gregated to fulfill a volume count requirement. Some automated vehicle classification systems also can
provide speed data.
Volume
The most basic form of traffic data collection is the volume count. Volume counts are collected as a
combination of continuous counts and short-term counts. Continuous counts are collected from perma-
nently installed equipment in, along, or above the road bed that are operated continuously. Short-term
counts frequently are collected through the use of pneumatic tubes gathering axle impulses that are later
factored. Short-term volume counts, which can number in the thousands annually, represent the bulk
of any coverage count program. Both on- and off-state system roadways should be measured. A more
thorough discussion is available in Section 3.3.4.
It is recommended that, when physically and financially possible, vehicle classification counts be taken
in place of simple volume counts. While this may slightly increase the cost per count for data collection,
the need for truck volume data continues to grow. Most agencies will be well served to collect classifica-
tion data instead of simple volume data whenever the marginal cost difference between these two data
collection efforts is small. (See “Count Nesting” in Section 2.3.1 below.) While in many cases, collecting
classification data is not an acceptable option—and therefore a simple volume count is the appropriate
data collection choice—agencies are encouraged to increase the percentage of count locations at which
classification data are collected in place of simple volume count data.
Speed
Speed data can be captured through automated systems collecting weigh-in-motion, vehicle classifica-
tion, and some volume data (depending on the equipment used), or by using trailer-mounted radar
systems or floating car techniques. Speed data could be reported as an average speed for periods of the
day (e.g., morning peak) by day of week, month, or annual average. These would be calculated directly
as a weighted average of the daily averages. Speed distributions also can be reported, including percent
of traffic exceeding the speed limit. Speed data may be binned when collected and the volume within
each speed bin would be used to represent the distribution. Speed data also can be aggregated so that it
fulfills a volume count requirement.

2.2.2 Federal or State Research Initiatives

States may participate in cooperative research conducted by the state or Federal government. A Federal
example is the Long-Term Pavement Program (LTPP). The data collection discussion presented here
for this particular program could be applied to all cooperative data collection efforts supporting research.
Data reported for LTPP include the measurement of traffic on the pavement research lane. For many
state agencies, the benefit of acquiring data from all lanes of traffic at a site will outweigh the additional
expense of permanent sensor installation or field work on all lanes at the site. LTPP traffic counts, if not
included in the coverage count sample, will have to be added separately to the traffic data program. Some
states install permanent equipment and other states use portable equipment for LTPP and other ongo-
ing traffic monitoring.
The importance of LTPP to the state’s count program depends on the number of general and specific
project sites in each state. It is possible to add LTPP sites to the coverage count sample. For states that
have not incorporated LTPP counts into their coverage count program, some or all of the LTPP sites
will be an element of the count program.
2.2.3 Project-Related Counts

Agencies typically conduct traffic studies for project-level design if the project is not included in the
coverage count program. This effort is an important function of the traffic monitoring program. A
shared concern among agencies is to provide the needed quality and quantity of counts to support road
project development. Project-related counts are made beyond the needs of the coverage count program.
Project-related traffic counts are used widely from planning studies and preliminary engineering to final
geometric and pavement design.
To estimate the number of project-related counts, traffic data collection programs should open and
foster communications with project development and design engineers. The agency’s annual and multi-
year plan (Transportation Improvement Program, Statewide Transportation Improvement Program,
Long-Range Plans, or Metropolitan Transportation Plans) for upcoming projects should be reviewed.
Input should be sought from the project development and design engineers for the type and timing of
traffic data collection needed to support project development. Their responses will provide an indication
of the number of counts needed and the general timeframe in which they should be completed. Existing
traffic data may be used rather than recounting for highway projects if data have been collected within
a reasonable time and can be factored for growth. In most instances, agencies will collect vehicle volume
and vehicle classification data. It also should be noted that there is an increasing interest not only in site-
specific volume and classification counts, but also in site-specific vehicle weight counts in support of the
Mechanistic-Empirical Pavement Design Guide. Successful planning and collection for this component of
the count program is dependent upon communication with project personnel.

Project-related counts should be retained and used to produce system-level estimates. If these counts are
not used to supplement the coverage count program, greater benefits from investments in these counts
are lost.
2.2.4 Special Count Requests

Agencies receive requests for counts which are not directly related to coverage, research initiatives, or
road projects. Special count requests include rail crossing studies, bridge crossing studies, legislative
requests, air quality impact studies, signal timing, traffic flow maps, traffic simulation calibration and
validation, signal warrants, site impact studies, and capacity analyses.
It is recommended that state agencies include site-specific traffic data collection as part of all contract
and in-house pavement and materials research. If this decision has been or will be reached, these counts
should be included in the estimate of special count requests.
A three-year average of the number of special counts taken is recommended to estimate an annual need.
To estimate the annual number of special counts:
• List the types of special count requests made of the collection program;
• Record the number of special counts by type and by year within the past three years;
• Average the sums of the past three years of special counts requests; and
• This three-year average of requests is used to anticipate this portion of the count program.
If an agency has changed its policies (such as counts in support of materials research) or cyclical counts
are not accounted in the three-year average, these additional requests should be estimated and added to
the three-year average.
2.2.5 Performance Measurement

Performance measures are “objective measurements and observations to determine the degree of suc-
cess a project, program, or initiative has had in achieving its stated goals and objectives.”2 Measures are
usually focused on tracking inputs, outcomes, or outputs of individual day-to-day elements.3 Traffic data
collection programs may already, or may soon, be asked to support monitoring these measures.
2
Southwestern Pennsylvania Commission. Congestion Management Process Glossary. Pittsburgh,
Pennsylvania, August 2006.
3
TransTech Management, Inc. Strategic Performance Measures for State Departments of Transportation: A
Handbook for CEOs and Executives, Final Report, NCHRP Project 20-24(20). NCHRP, Washington, DC,
June 2003.

Data Types
There are three classes of performance measures. First, input performance measures relate to the quanti-
ties of resources used to produce goods or services and the demand for those goods and services. Out-
come measures are indicators of the actual impact or public benefit of a service. Finally, output perfor-
mance measures gauge the actual service or product delivered by an agency.4
Traffic data collection programs are likely resources to collect performance data for input and outcome
measures. Table 2-3 displays a sampling of the performance measures agencies may be interested in
tracking. The performance measures also are classified by type: input, outcome, or output. Some mea-
sures may be viewed as satisfying more than one performance type.
When agencies initiate performance measurement, the data collected to support them are considered
part of the special count program. Because of the regular monitoring frequency required for these
counts, the measurement needs should become part of the annual count projection.
Examples
Florida’s traffic data collection program contributes to several performance measures reported by Flor-
ida’s Mobility Performance Measures Program. Some of these measures include person-miles traveled,
truck-miles traveled, vehicle-miles traveled, maneuverability, and vehicles per lane-mile.
A regional congestion management process may identify volume/capacity (V/C) ratios as a performance
measure. Segment volumes are required to generate these measures. The Southwestern Pennsylvania
Commission Congestion Management Process monitors travel time in peak periods, average speeds,
delay per vehicle-mile, total delay, and total delay per mile.
An NCHRP report provides additional examples. Washington State monitors system performance for
daily vehicle hours of delay per mile, sample commutes measured by delay, and time-of-day distributions
of delay. Measuring reductions in peak-period work zone lane restrictions is important to Pennsylvania.
Finally, Minnesota monitors their interregional corridors to measure the percent of corridor miles meet-
ing speed targets, and hours and miles of congestion per day.
4
Florida Department of Transportation. FDOT Long-Range Program Plan for Fiscal Years 2006–2007 Through
2010–2011, revised September 28, 2005.

Table 2-3. Sample Performance Measures by Type

Performance
Agency Type Sample Performance Measures Measure Class
Output Outcome
State Transportation Agency Vehicle miles traveled •
Vehicles per lane-mile •
Truck miles traveled •
Person miles traveled •
Average travel time •
Person-miles of travel in congested ranges • •
Average travel rate •
Growth in annual delay per peak road traveler •

Metropolitan Planning Region Average travel time •
Travel reliability •
Person-miles of travel in congested ranges •
Person-hours of travel in congested ranges •
Person movement speed •
Total delay •
Person movement •
Delay per vehicle-mile •
Total delay •
Total delay per mile •

•
V/C ratios
Municipalities Average travel time •
Accident reduction •
Managed/HOT/HOV Lane Person movement •
Operator
Vehicle occupancy data •
Travel reliability •
Revenue •
Traffic Management Centers Lane occupancy data •
Duration of incident •
Time to incident detection •

2.3 Estimating Traffic Data Collection Needs
An effective traffic-monitoring program includes the collection of short-duration traffic counts, or “short
counts,” at a relatively large number of sites, and collection of continuous counts at a much smaller number
of sites. Short counts provide information about traffic volumes on a large number of road segments, or
“traffic count segments,” over a short time period, typically a 48-hour weekday period, but they do not
provide information on average volumes over the course of a year. The principal purpose of the continu-
ous counts is to provide the basis for converting short counts to estimates of annual average conditions
at the short-count sites.
Volume counts are collected at all short- and continuous-count sites. Classification counts should be col-
lected at all continuous-count sites and at a significant share of the short-count sites. WIM data should
be collected at a more limited number of sites, most or all of which are continuous sites, and all of which
also are used to produce volume and classification counts.
The use of one site to produce multiple types of traffic data (volume, classification, and WIM) is
referred to as “count nesting.” Count nesting makes efficient use of traffic monitoring resources, since it
allows a single site to be used to produce multiple types of data. Also, the more sophisticated equipment
used at WIM sites produces more accurate counts than equipment used at other sites (and classification
counters frequently produce more accurate counts than volume counters), so count nesting has the ad-
ditional benefit of improving the quality of counts collected at nested sites.
The first part of this section discusses the data collection sampling plan, the second discusses continu-
ous counts and seasonal factor groups, and the third discusses short counts.
2.3.1 The Sampling Plan

Agencies benefit from having well documented data collection procedures. One document is a formal
sampling plan. This document serves as a basis for future program revisions and as a support tool for
acquiring and retaining the resources needed. Good process documentation also provides for consistency
through periods of staff turnover. A sampling plan for the coverage count program, for instance, should
be based on statistical analysis of the system-level data and should properly document the target sta-
tistical confidence and precision levels for all relevant systems and for more discrete analysis units (e.g.,
seasonal factor groups).
The Development Process

In general the steps for developing or refining the sample size for each data type within the coverage
count program are:

• Define the Test Statistic—In creating or refining any sampling plan, the test statistic must be identi-
fied first. For volume counts, the appropriate test statistic is annual average daily traffic (AADT).
Other test statistics of interest include: monthly or seasonal factors from continuous volume sites;
axle correction factors for vehicle classification; or, for WIM, gross vehicle weight (GVW) or equiva-
lent single-axle loads (ESAL) for a specified vehicle class. Agencies should seek input from their
customers to help define test statistics.
• Define Stratification Groups—Stratification groups may be structured by geography, area type,

roadway use, functional classification, volume group, or other criteria important to the agency. Decid-
ing the stratification groups will have a direct effect on the total sample size.
• Define the Statistical Confidence and Precision for the System and Subsystems of Interest—For
each test statistic, a statistical confidence interval and precision should be selected. The HPMS Field
Manual provides guidance for these levels.5 Agencies should decide on the confidence and precision
levels that they desire to attain for the entire system and for the subsystems of interest. The precision
of estimates obtained for an entire system of roadways will be greater than the precision obtained for
any subsystems. Estimates obtained for individual strata may be of no interest in themselves, in which
case there is no need to specify confidence or precision levels for the strata.
• Process the Test Statistic’s Average and Standard Deviations—Currently available data should be
used to estimate averages and standard deviations for the test statistic.
• Calculate the Required Sample Size—The test statistic’s average and standard deviations with the
statistical confidence and precision levels are required to calculate the required sample size.
• Determine Additional Sampling Needs for Failed Counts—The required sample size estimates
should be slightly inflated to allow for samples that may not be available for the production of system-
level estimates due to roadway construction or equipment maintenance. This number is then added
to the calculated sample size. Methods are presented later in this section.
Statistical Basis
Many agencies collect traffic counts on a 100 percent sample of the higher functional systems and on
smaller samples of the lower systems. In both cases, the sample of segments consists of both segments
having sites that are continuously monitored and those at which only short counts are collected. That
is, for the purpose of understanding traffic (e.g., estimating VMT) on a particular system of roads, data
from both types of sites are used.
A particular concern in the sampling of road systems is that the sample be created and maintained in a
way that minimizes any bias in the resulting estimates of VMT for the state’s entire road system and for
5
FHWA. HPMS Field Manual, 2000, Appendices C and D.

all subsystems of interest. For this purpose, stratified sampling is useful. The strata that are used should
distinguish all subsystems of interest as well as other characteristics that are likely to have significant
influences on traffic volumes. Potentially useful strata include:
• Functional system;
• Geographic jurisdictions (such as counties or highway districts);
• Highway ownership (state, county, local, etc.); and
• Number of lanes.6
One of the advantages of stratification is that the strata can be designed to have more uniform values of
traffic volume and AADT than the system or subsystems to which they belong.7 This property means
that, for a given sample size, a stratified random sample generally will produce better estimates of average
AADT for the system and subsystems than will a simple random sample. Also, when stratified random
sampling is used, there is no disadvantage to increasing the sample size of certain strata of interest (such
as those corresponding to the State Highway System), as long as the sample sizes of the other strata
are maintained.
When stratified random sampling is used, the preferred procedure for estimating (daily or annual)
VMT for an entire system of roads involves the use of expansion factors. For this purpose, for each stra-
tum, an expansion factor is calculated by dividing the total length of roadway in that stratum by the total
length of the stratum’s traffic count segments that have been sampled (and, on which, traffic has been
counted). Multiplying AADT on each of the sampled segments by segment length produces estimated
(daily) VMT on the segment. Adding the resulting estimates of VMT for all segments in a stratum and
multiplying by the stratum’s expansion factor then produces estimated VMT for the stratum. Finally, an
unbiased estimate of VMT for the subsystem (or for the entire highway system) is obtained by sum-
ming the VMT estimates for all strata of the subsystem.
Count Nesting
As noted above, the use of higher data collection systems to provide information for lower data collec-
tion systems is termed count nesting. For example, weigh-in-motion systems collect data on weight,
speed, total traffic volume, and volume by vehicle class. Thus, in addition to providing WIM data, a
weigh-in-motion system can be used to provide a set of speed data, a volume count, and a classification
6
An alternative to stratification by ownership and number of lanes is stratification by volume group (as required
by HPMS). However, stratification by volume group presumes knowledge of the volume group to which all traf-
fic segments belong, which may not hold for the lower functional classes.
7
As an example, consider a road system that has 500 miles of four-lane road and 500 miles of two-lane road. An
unstratified random sample is likely to be split close to 50/50 between miles of four-lane road and miles of two-
lane road, but, most likely, the split will not be precisely 50/50. Such a slightly uneven split is likely to introduce
a small amount of bias into the resulting estimate of VMT for the road system. On the other hand, a stratified
random sample can be constructed to have exactly a 50/50 split between the two types of road, thus avoiding
this bias. Alternatively, a stratified random sample can be used in conjunction with “expansion factors” (described
below) that adjust for any over or undersampling of any of the strata.

count. Likewise, vehicle classification counts also provide volume counts.
Florida uses a simple graphic to convey this nesting, the size of the count program, and method of data
collection, as shown in Figure 2-1. The right side, or white area, of the pyramid represents permanent
data collection sites in, alongside, or above the road bed that can remotely transmit data back to the
office. The left side, or shaded area, of the pyramid represents short-term data collection sites that may
use portable equipment or manual data collection. Because WIM is able to collect weight, classification,
and count data, it resides at the top and signifies that the information collected can be used for both the
classification and count programs. The volume or coverage count program is limited to vehicle volume
and cannot be used to supplement the classification or WIM programs, thus representing the basic
foundation for traffic data collection.
Figure 2-1. Example of Count Nesting.8
Count nesting allows the number of required counts to be reduced, eliminating redundancy, improving
overall efficiency of the traffic data collection program, and freeing resources to be invested elsewhere.
Random Selection
Sample bias is eliminated through random selection of traffic count segments. Random selection is
performed most efficiently using electronic maps that provide the necessary roadway attributes. These
attributes include functional class, ownership (state, county, city, etc.), and area type (rural, small urban,
or urbanized). If key attributes are not available electronically, the process becomes less efficient, and
comparisons of hard copy maps may be required. The random selection process also may be structured
8
Florida Department of Transportation. Florida Project Traffic Forecasting Handbook, October 2002.

so that road segments are numbered, and a random number generator used to select potential segments
for the sample.
The list of potential segments may be reported as randomly selected. If random selection is automated,
sample selections should be spatially sorted in sequence by road. As an example, the sample may be
printed by road and milepoint. This will facilitate the review of the potential segments, selection of the
final sample, and scheduling of coverage counts.
There are cases in which random selection cannot be used. The phrase “potential segments” is used to
emphasize that the randomly selected segments must be reviewed for appropriate field data collection.
Effort is required to validate the physical condition of randomly selected segments for their appropri-
ateness for data collection. Most potential segments will include sites that are appropriate for use as
volume-count sites, but not necessarily for WIM sites or even for classification sites (as discussed in the
next subsection). It may be prudent to generate 10 percent more potential segments than that indicated
by the sample size determination. Alternatively, roadway segments may be sampled without replacement,
and additional segments randomly selected if required.
Classification Sites and WIM Sites

The above discussion of selecting sample segments and monitoring sites ignores some practical consider-
ations that affect WIM and classification sites.
Classification Counts and Classification Sites. The principal purposes of classification counts are to
learn how vehicles are distributed among vehicle classes and for developing axle factors for converting
axle counts collected using road tubes into vehicle counts. For these purposes, the best information is
road-specific information.
Ideally, all traffic counts would be classification counts, so that roadway agencies can fully understand
the truck volumes on their roadways. While this is currently impractical, classification count data should
be collected whenever practical. At a minimum, classification counts should be collected periodically on
all roads in the higher functional systems. In the case of longer roads on which vehicle-class distributions
are likely to change over the length of the road, these counts should be collected at multiple sites on the
road. Short-duration classification counts should be factored using class-specific seasonal and day-of-
week (DOW) factors (as described in Section 5.3.3) to produce estimates of AADT by vehicle class
(VC). For each classification count, these estimates, in turn, are used to produce distributions of AADT
by VC that are used on the same road (or on nearby parts of the road) to distribute estimates of total
AADT over the vehicle classes.
For the lower functional systems, it may be impractical to collect classification counts on all roads. For
such roads, the recommendation is that roads in a given area and functional system that are believed to

have reasonably similar vehicle-class distributions be grouped. For each of these groups, one classifica-
tion count site should be identified and measured to derive a group axle factor and a set of distributions
of AADT by VC for use at other short-term volume count sites within the same group.
The above discussion has some implications for selecting sites within a traffic count segment at which
volume and classification counts are collected. Accuracy requires that continuous classification counting
be performed only at locations where vehicle speeds are as constant as practical (minimum acceleration
and deceleration). Although this requirement is less applicable to short-count sites where only volume
counts are collected, it is still desirable to apply it, if practical, when selecting a site for volume counting
on a traffic count segment. Such a site would have the advantage of also being usable for classification
counting if it is subsequently decided to collect classification counts on the segment.
WIM Sites. The considerations affecting WIM sites are somewhat more complex. It generally is inef-
ficient to select WIM sites at random. WIM sites are relatively expensive to establish and maintain, and
they require strong, smooth, level pavement and straight sections of road. For this reason, it is generally
efficient to install new WIM sites only as part of resurfacing projects, and the particular sites considered
for such use should be limited to those that meet the criteria for producing reliable WIM data. Once
WIM equipment has been installed at a particular site, it generally should be used as a source of WIM
data as long as the site continues to meet the data accuracy criterion set by the collecting agency and is
confirmed by periodic calibration validation tests.
In order to make the maximum use of data from a new WIM site, any such new site should be used to
provide volume and class data as well as WIM data. If a new WIM site is located on a traffic segment
which is already part of a volume-count sample, then the new site simply replaces the existing site in the
sample. If not, it is recommended that the new site replace an existing site belonging to the same stratum
and, to the extent practical, having seasonal and day-of-week variations in traffic volume that are similar
to the new site.
Since WIM equipment generally produces higher quality classification data, WIM sites should be oper-
ated as continuous classification sites. For this purpose, the WIM sites are simply added to the appropri-
ate classification-count seasonal factor group. Also, if the data collection sensors continue to function
correctly, WIM sites should continue to be used as sources of continuous classification data after being
retired as WIM sites.
Field Review for Site Adequacy

Site adequacy is a paramount consideration in the deployment of both permanent and short-term data
collection devices. The field crew manager or assigned staff should review the randomly selected poten-
tial sample segments to determine if they contain sites that are adequate for field data collection. It is
important to recognize that simply identifying a road segment for sampling does not guarantee that an

appropriate traffic count site exists on the segment. Agencies should develop a list of minimum stan-
dards for each data type. These minimum standards may be based on manufacturer specifications or or-
ganized technical standards, such as those produced by the American Society for Testing and Materials.
The field review should evaluate the geometric and operational characteristics of the roadway. Randomly
selected segments may be on a horizontal curve, making them unsuitable for vehicle classification. Simi-
larly, the speed of vehicles on part or all of a randomly selected segment of roadway may be outside the
manufacturer-specified speed range for some vehicle classification devices. As previously noted, many
automated data collection systems do not collect accurate data under congested traffic conditions. These
conditions are unsuitable for vehicle classification and WIM systems. Site reviews should also include
roadway surface characteristics. Other considerations that should be included in site review may include
equipment tie-down, and the availability of telephone lines and power supply. Safety and equipment
limitations also should be reviewed on a site-by-site basis.
The site review can use existing engineering plans and other sources, but physical inspection of poten-
tial sites is recommended. A photo video system may be helpful in the process of reviewing randomly
selected traffic monitoring segments, but such a system is not an adequate replacement for physical
inspection of the segment.
Based on the review, the field crew manager or assigned staff may delete potential segments which do
not contain any acceptable sites. The reason for deleting potential segments must be documented and
provided to the Traffic Data Program Manager. Documentation can be via a printed form or via other
recorded notes identifying the characteristics that failed to meet minimum standards. The end product
of this process will be a randomly selected sample acceptable for field data collection.
Failed Counts—Machine/Collector Error or Malfunction

Sampling plans should provide for extra samples so that the desired statistical confidence and preci-
sion can be maintained even if there are a few failed counts. It is not uncommon for machines or staff to
generate data errors.
The number of failed short counts must be estimated so that these counts may be rescheduled without
disrupting the data collection program. Existing records should be used to estimate the rate of failed
counts for both short-term and continuous data collection. Estimates of failed short-term counts can
be derived from completed schedules, contractor invoices, or machine repair logs. Estimates of extra
samples for failed counts can be made from these historical records; e.g., by using a three-year average of
the failure rate.

Periodic Review
The goal of every traffic data collection program is the production of useful data for its customers. Peri-
odic adjustments to the plan are necessary to reflect the effects of changes in the highway system and in
the use of various roads and also to incorporate improvements in traffic monitoring procedures. It is rec-
ommended that traffic monitoring plans be reviewed at least once every five years. These reviews should
include a review of the seasonal factor groups and factoring procedures used, the continuous monitoring
sites providing data for these procedures, and the size and distribution of the short-count sample.
2.3.2 Continuous Counts

The principal purpose of continuous counts is to provide data on how traffic volumes vary by time
of year and by day of week. These data are used to convert short counts into estimates of AADT and
AADT by vehicle class (AADT by VC). Continuous counts also are used to produce estimates of
growth in traffic that are used to adjust AADT estimates developed in one year for growth in subse-
quent years.
In order to obtain reliable estimates of the seasonal and DOW patterns in traffic volume (or truck
volume), continuous counts should be collected on roads that have moderate to high traffic volumes.
Patterns on roads that have low traffic volumes are more likely to be affected by random events than pat-
terns on higher volume roads.
For the purpose of estimating AADT, each continuous count site and each short-count site is assigned
to one of several seasonal “factor groups.” Data from continuous counts collected at sites in a given group
is then used to develop a set of seasonal and DOW adjustment factors that are used to convert short
counts obtained at other sites in the group into estimates of AADT and AADT by VC. Current pro-
cedures for forming seasonal factor groups focus on total traffic volume and estimates of total AADT,
although the same factor groups usually are also used for developing factors to be used in developing
estimates of AADT by VC.
The development of seasonal and DOW factor groups is discussed below. The computation and appli-
cation of seasonal and DOW volume and vehicle-class factors and ratios is discussed in Section 5.2.1.
In developing factor groups, two goals should be kept in mind:
• To the extent practical, all sites in a given seasonal factor group should have reasonably similar sea-
sonal and DOW patterns in traffic volume; and
• To the extent practical, factor groups should be distinguished from each other on the basis of road-
way characteristics (such as functional class, location, or roadway direction) that can be applied to
road segments for which little or no previously collected traffic pattern data are available.

It is recommended that most seasonal factor groups contain five to eight continuous count sites. How-
ever, seasonal factor groups, such as recreational groups, that contain relatively few short-count sites may
be defined using data from fewer continuous count sites. In order to minimize the number of continu-
ous count sites required, seasonal factor groups should only be distinguished from each other if they
have seasonal and/or DOW traffic patterns that are at least moderately different from each other.
Developing Seasonal Factor Groups

In defining a set of seasonal factor groups, the principal goal is to group sites that have seasonal patterns
in traffic volume that are as similar as practical and DOW patterns that are as similar as practical.9 Fac-
tors developed using data from continuous sites at which traffic volumes drop on weekends (typical of
rural roads carrying relatively little long-distance traffic) should not be applied to short counts collected
at sites at which volumes tend to rise on weekends (typical of rural roads carrying significant amounts of
long-distance traffic).
Some observations that are useful in defining seasonal factor groups are:
• In rural areas, DOW variation usually differs between the interstate system, which tends to have high
through volumes) and the non-interstate system (with lower through volumes);
• Differences between seasonal and DOW patterns on the various non-interstate system rural func-
tional systems are small and probably do not warrant separate seasonal factor groups for the different
functional systems;
• In some large states, seasonal variations in volume on rural roads may differ in different parts of the
state and may warrant separate rural factor groups that are defined geographically;
• Roads in recreational areas usually have seasonal and DOW patterns that warrant separate factor
groups. In some states, these roads warrant separate recreational groups for different peak travel
periods (summer, winter, fall foliage, etc.);
• In urban areas, interstate system and non-interstate system roads are frequently assigned to separate
factor groups, though the seasonal and DOW variations on the two systems usually are quite similar;
• In urban areas, it may be worth distinguishing between roads that tend to have relatively low weekend
volumes (e.g., those serving employment centers that have little activity on weekends) from roads
serving shopping areas and other areas with significant weekend activity; and
9
This goal suggests that it might be desirable to assign road segments to factor groups on the basis of the predom-
inant use(s) of the segments. However, such a system of assignments is relatively difficult to implement because
roadway uses generally do not correlate with readily observable roadway characteristics (such as functional class).
Also, it should be observed that similar road uses (such as transport of different crops) do not necessarily mean
that seasonal patterns of traffic will be the same (since different crops may have different harvest seasons).

• There may be some value in establishing separate (“semi-urban”) factor groups for small urban areas
and/or for the urban fringe of major urbanized areas; the latter groups might include roads located
near the edge of urbanized areas and on either side of the boundaries of these areas.
Issues Relating to the Development of Seasonal Factor Groups

Similarity of Seasonal and DOW Patterns
Factoring procedures work best when all sites assigned to a given seasonal factor group have seasonal
and DOW patterns of traffic volume that are reasonably similar. However, in practice, this degree of
similarity is not always attainable. There will almost always be some sites that have volume patterns that
do not fit well in any group; and some short-count sites may be inadvertently assigned to the wrong
group. Such poor assignments have disadvantages for both short-count sites and continuous count sites.
In the case of short-count sites, a poor assignment adversely affects the quality of the AADT estimates
that are produced for the site by the factoring procedure. For example, if a site at which traffic volumes
rise on weekends is assigned to a group that otherwise consists entirely of sites at which volumes drop
on weekends, AADT estimates produced by factoring weekday counts collected at the site will be down-
wardly biased.
As a somewhat different example, consider a site on a recreational road on which traffic peaks during
the fall foliage season and assume that this site is assigned to a recreational group whose factors were
developed from continuous count sites on roads with summer peaks. For such a site, AADT estimates
produced from factored counts taken during the summer will be downwardly biased, while AADT
estimates produced from factored counts taken during the fall peak will be upwardly biased. For such a
site, AADT estimates produced from factored counts taken during the spring are likely to be better; but,
for various reasons, these estimates also may not be very good.
In the case of continuous count sites, poor assignments indicate a factor group that has been defined in a
way that results in the inclusion of sites that have moderately dissimilar seasonal and/or DOW patterns
of traffic volume. For such a group, it may not be possible to achieve the precision level for the seasonal
and/or DOW factors recommended by the TMG (pages 3-47 to 3-48) without increasing the number
of continuous monitoring sites above eight. This deficiency can be addressed either by redefining the
factor group (possibly by splitting it) in order to increase the similarity of the seasonal and DOW pat-
terns, by increasing the number of continuous sites belonging to the group or by forming different factor
groups for seasonal versus DOW factor types.
If practical, it is preferable to redefine the factor group. Increasing the similarity of the traffic patterns
will reduce or eliminate the number of short-count sites in the group that have seasonal or DOW
traffic patterns that are atypical of the group. Since poor estimates of AADT are more likely for sites
with atypical traffic patterns, reducing the number of such sites is a desirable goal of any traffic monitor-
ing program.

The second alternative, increasing the number of continuous count sites belonging to the group, im-
proves the precision of the factors used for the group and thus improves the resulting estimates of
AADT. However, only slight changes are likely in the AADT estimates for sites with atypical traffic
patterns, and not all of these changes will be improvements.
The third alternative, forming separate seasonal and DOW factor groups, allows continuous count sites
to be assigned independently to seasonal and DOW factor groups. This approach solves the problem of
continuous count sites that have different seasonal and/or DOW characteristics from other sites within
the group, while increasing the complexity of forming the groups.
The appropriate number of factor groups to establish depends on circumstances, including avail-
able resources and the degree to which different seasonal and DOW patterns in volume are readily
distinguished.
Recreational Roads
The development of procedures for monitoring traffic on recreational roads presents some issues alluded
to above that can only be partially solved under the typical resource constraints applied to Traffic Moni-
toring Systems.
One issue is the number of recreational groups to be distinguished. Recreational roads differ both in
the timing of their recreation-season peaks and in the intensity of these peaks. Some states define two
or three recreational groups representing the different peak seasons for recreational traffic in the state.
Other states have only a summer recreational season, but define separate groups for “high summer” and
“moderately high summer.” In concept, some states could create separate “high” and “moderately high”
groups for each of several peak seasons, but establishing that many groups is likely to stretch available
resources. In general, states simply accept AADT estimates for recreational roads that are less reliable
than estimates for other roads. Since the resulting errors in AADT estimates for different recreational
road sections are likely to partially cancel each other out, and since VMT on recreational roads generally
is a fairly small part of statewide VMT, these errors are generally considered to have an acceptably small
effect on estimates of statewide VMT.
A second issue relates to the assignment of sections of road containing short-count sites to recreational
groups. These assignments generally are made on the basis of local knowledge of whether or not the
peaking characteristics of traffic on these road sections follows a recreational pattern. In some cases,
extending a short count to include a full weekend during the recreational season could be helpful in
determining whether or not a site carries significant amounts of recreational traffic (and whether it
should be classified as “high” or “moderately high” recreational); volumes on rural non-Interstate System
roads generally drop on weekends, but, on recreational roads, these volumes rise on weekends during the
recreational season.

Truck Weight Roadway Groups
For the purpose of collecting data on vehicle weights, the 2001 TMG recommends that each state divide
its road system into truck weight roadway groups (TWRG) with the average gross weights of trucks in a
given class being reasonably similar on all roads in the group. The TMG suggests establishing between 2
and 10 TWRGs, depending on the size and diversity of a state’s roadway system.
A basic set of TWRGs for a state with relatively little geographic diversity might consist of three
TWRGs: urban; rural Interstate System; and rural other. For this set of TWRGs, it is recommended
that all WIM sites in the urban TWRG be located on the urban Interstate System, and all WIM sites in
the rural other TWRG be located on the rural other principal arterial system. There are several reasons
for this last pair of recommendations. In general, WIM data are of greatest interest when designing
pavements for the higher functional systems. Also, axle loads and the resulting pavement stresses are
likely to decline as one moves to lower functional systems, so it is a conservative approximation to use
WIM data collected at sites on a high-functional system when analyzing pavement stresses at a site on a
lower system. Finally, in urban areas, most of the appropriate locations for WIM sites are likely to be on
the Interstate System. The formation of TWRGs is discussed further in Section 5 of the TMG and in
NCHRP Report 538.10
If practical, there should be between three and eight WIM sites in a TWRG. However, one or two
WIM sites may be used for some small TWRGs. Three sites is the minimum number necessary to
provide some confidence that all sites in the TWRG have reasonably similar load spectra. On the other
hand, as the number of WIM sites in a TWRG grows, opportunities also grow for splitting the TWRG
so as to produce smaller TWRGs, each with more uniform sets of load spectra. Since it is difficult and
expensive to calibrate portable WIM equipment, it is recommended that all WIM sites use in-pavement
sensors. It also is recommended that the design lane of all WIM sites be monitored for at least one week
per month throughout the year. The TMG recommends that all new WIM installations have WIM
coverage in at least one lane in each direction, and that each TWRG contain a minimum of one site with
WIM coverage in at least two travel lanes in each direction.
2.3.3 Short Counts

The two principal types of short count are coverage counts and special counts.
For purposes of traffic counting, all roads should be divided into traffic count segments (also called “traffic
segments” or “traffic sections”), with traffic volumes on each segment being reasonably homogeneous. On
limited access roads, traffic count segments should extend from interchange to interchange. On other
10
Cambridge Systematics, Inc. Washington State Transportation Center, and Chaparral Systems Corporation.
Traffic Data Collection, Analysis, and Forecasting for Mechanistic Pavement Design, NCHRP Report 538, 2005,
pp. 2-19 to 2-20.

roads, the TMG recommends (pages 3-9 to 3-10) that, except in the case of the lowest volume roads,
traffic volume on each segment remain within a range of plus or minus 10 percent. For very low-volume
roads, the TMG suggests that a wider range may be appropriate. Traffic segments need not be broken at
urban/rural boundaries or at jurisdictional boundaries. For the purpose of factoring, judgment should
be used in choosing the seasonal factor group to which a segment that crosses an urban/rural boundary
is assigned.
For each highway system and subsystem of interest, coverage counts are collected periodically on a sample
of traffic count segments to estimate VMT on the system and to provide a more general understanding
of traffic on the system. The size of each sample is determined by the highway agency and reflects both
the availability of resources for collecting these counts and the agency’s intended uses of the data. Many
state highway agencies (SHA) collect coverage counts on a 100 percent sample of the interstate system,
and a few collect coverage counts on a 100 percent sample of the entire State Highway System (SHS).
Although not explicitly stated, all traffic count segments on which continuous count sites are located
are part of the coverage count sample. Most other traffic count segments in the coverage count program
are counted for short periods of time (“short counts”) once every few years, with growth factors used to
adjust all estimates of traffic on the segment for changes in traffic volume that are likely to occur in inter-
vening years. For specific segments, frequency of counting may vary with the importance of the segment.
The TMG recommends that most segments be counted at least once every six years, with important
segments and those experiencing unusual changes in traffic volume counted more frequently. FHWA
requires that HPMS sample and universe sections be counted at least once every three years.
For planning or design purposes, agencies frequently require traffic data that are not produced by the
coverage count program. These needs are met by collecting special counts. Such special counts may meet
needs for: counts on segments on which coverage counts are not collected; detail that is not provided by
the coverage counts (e.g., classification counts on segments on which only volume counts are available);
or more accurate estimates of current traffic volume than can be obtained by applying a growth factor to
a five-year-old count.
2.3.4 Estimating Overall Data-Collection Needs

In order to estimate the overall data collection needs, one should sum the count needs for the coverage
count program defined by the sampling plan, research initiatives, roadway project development, special
counts, performance measures, and replacement counts for field failures to estimate a comprehensive
traffic count total for the year. The count needs should distinguish data type and collection method for
later estimates of staff and resource needs.

2.4 Reducing Traffic Data Collection Needs
Traffic data collection needs can be reduced by means of count nesting and other multipurpose uses of
sites, cooperative data sharing, identifying obsolete sites, and managing special counts. Count nesting
was discussed earlier in this section, and cooperative data sharing (with substate jurisdictions and with
neighboring states) is addressed in a subsequent section.
Over time, data collection sites may cease to meet their original mission. Texas aggressively reduced the
number of its short-term volume counts.11 Many of the short-term volume locations had been selected
previously to fill needs that were rarely documented. Over time, these count locations were perpetu-
ated through tradition, and their original purpose lost. Texas eliminated almost all sites having 24-hour
volumes below 300 and not at bridges and at railroad at-grade crossings. These counts were replaced by
volume counts at a statistical sample of sites in five types of area (rural unincorporated, rural incorpo-
rated, small urban, small urbanized, and large urbanized). The randomly selected count locations result
in statistically valid mean values that are used to estimate volumes on road segments for which no counts
are performed.
2.4.1 Managing Special Counts

A helpful way to manage demand for traffic data collection is to establish and enforce procedures for
responding to special count requests. Some special requests may be satisfied by using existing traffic
data without requiring a new traffic count. Prior to scheduling and collecting a special count it should
be determined whether a count was taken recently on the same segment and if the customer will accept
either observed data or estimated data from the past count. In general, except for areas with significant
changes in traffic patterns and volumes and roadway characteristics, data from counts collected within
the past three years and factored for growth should be adequate for most planning needs.
2.5 Estimating Staff and Resource Requirements for Traffic

Data Collection
This section will assist in assessing current resources and guiding the delivery of new programs and
resource needs. As discussed below, state traffic data programs use a mixture of agency and contract
forces. In an ideal situation, an agency can design a program from scratch and estimate required re-
sources accordingly. However, in most cases, a traffic count program manager is faced with tailoring
the program’s counting capacity to its available financial or FTE resources. This section suggests guide-
lines for estimating staff and contracting requirements for a typical count program. Needs will vary
significantly for different programs. A careful assessment of customer needs for the traffic program is
highly recommended.
11
Frawley, W.E. Frawley. A Methodology for Randomly Selecting Traffic Count Locations in Texas—
Results and Benefits. Presented at the North American Travel and Monitoring Exposition and
Conference, June 27–July 1, 2004, San Diego, California.

2.5.1 Agency Forces

The types of staff required to manage a traffic count program include collection staff, inspectors, office
support, and site installation/maintenance support. These positions are discussed in more detail below.
Collection Staff
The productivity of collection staff is influenced by several factors. Staff are limited in the number of
counters they set each day by the complexity of the equipment, coordination with traffic control, travel
time between count locations, and the crew vehicle’s capacity for transporting counters and supplies.
Efficient count scheduling can maximize staff productivity. The following subsections discuss the factors
that must be considered in determining collection staff resources. The guidance could be used in estab-
lishing a new program or maximizing the productivity of an existing one.
Counters Set per FTE
Short-term data collection relies on field staff to deploy and retrieve the equipment and the collected
data. The number of full-time employees (FTE) required to conduct a given number of annual counts
(agency productivity level) depends in part on the number of counts that each FTE can take in a week.
The agency productivity level is dependent on several factors, including the devices used, roadway cross
section, safety factors, and even work schedules. Following are some examples:
• Multilane facilities require more than one counter for each data collection site.
• An agency estimate of field personnel productivity should consider the number of persons in a
field crew.
• Vehicle classification devices require somewhat longer to set up, adjust sensor distance, and calibrate.
• Portable WIM devices are more difficult to set up and monitor, even if the devices have self-calibrat-
ing features.
After estimating unique agency productivity levels, the next step is to identify the number of weeks
during the year in which traffic counts may be conducted. Because of different prevailing weather condi-
tions, the available weeks will vary among states. The number of available weeks for counting establishes
the number of weeks each year that each FTE may be employed in meeting traffic data collection needs.
The anticipated number of weekly FTE counts is then multiplied by the number of weeks available for
counting during the year. This is the estimated annual productivity for each FTE. Separate estimates
of FTE productivity should be developed for volume, classification, and WIM counts. Dividing the an-
nual FTE productivity into the estimated number of traffic data counts produces the number of FTEs

needed to meet data collection needs. The benefit of this approach is that it ties traffic data collection
FTEs to specific data collection needs. Increases or decreases in field personnel should be directly
related to specific data collection needs rather than being adopted independently of specific elements of
the traffic data program. Often agencies are faced with applying FTEs according to resource availability
rather than having the luxury of assigning an unlimited number of FTEs.
Travel Time
Travel times between count locations directly influence the number of counts taken each day. Produc-
tivity can be enhanced by arranging count schedules that minimize distances between count locations.
Because urban count locations are more closely spaced than rural locations, productivity in urban areas
could be twice that in rural areas.
Agencies that operate centralized data collection systems—where collection staff are all based in a single
central location—are subject to productivity losses due to the travel times between this location and the
areas where the counts are to be collected. Lower travel times can be obtained if there are several sepa-
rate data-collection teams based at district offices in various parts of the state.
Vehicle Capacity
Deployment of short-term data collection equipment is limited by several factors, including the cargo
capacity of the transport vehicle and the size and number of counters and materials. A factor in weekly
FTE count productivity is the vehicle fleet used to transport traffic volume and classification devices. For
example, if an agency uses a van to transport traffic recorders, the vehicle may not be capable of hauling
more than 50 portable devices, tubes, and tools for installation and calibration.
Inspectors
Inspectors are required to ensure that short-term data are collected on the scheduled dates, at the proper
locations, with equipment that is properly placed, using appropriate worker safety procedures, and that
data collection is properly performed.
Each agency should define the minimum level of inspection for their short-term counts and inspections
should be scheduled randomly. This may include inspections during overnight periods if manual crews
are responsible for data collection. Agencies may require more frequent inspection of new contractors,
while the agency builds trust in the contractor’s performance.
Safety Considerations
Attention should be given to all work schedules to guarantee worker safety. Additional staff may be war-
ranted in certain conditions—high-volume roadways, high-crime locations, and under certain geometric
sections where sight distance may be limited.

Office Support
Office support staff play an important role in traffic data collection. Office staff are responsible for
scheduling routine traffic counts and special requests, for reviewing and processing incoming data, for
generating reports, and for processing contractor billings. Central offices also typically have staff that
repair equipment.
Count Scheduling
Efficient count scheduling is an important factor in maximizing field data collection productivity. Suc-
cessful traffic data collection programs strive to improve the efficiency of their count scheduling process.
Office staff are needed to produce annual and special count data collection and maintenance schedules
for both state and contract forces. Constant communication between the staff scheduling counts and
the data analysts is vital. Analysts and collection personnel should know and understand each oth-
ers’ needs and limitations. This communication will produce reasonable expectations and prevent
misunderstandings.
To schedule traffic counts, agencies will find it helpful to map the count locations and to group all
routine coverage counts by geography or roadway route and by time of year. Also, when practical, special
counts should be scheduled to occur when coverage counts in the same area are being collected. If pos-
sible, deferring or advancing special counts for this purpose also will help to reduce program costs.
Schedules prepared by decentralized systems may have one FTE perform this function with other rou-
tine duties at each remote location.
Initial Processing of Data
Data collected either through machine or manual processes will be submitted to or retrieved by the
agency. Though the use of computers and communications to poll telemetry sites is common, the data
must be packaged, reviewed for completeness, screened for accuracy, and transmitted by data collection
office staff to the data analysts. Problems or errors identified in this initial screening may be resolved by
the screeners or passed on to others for further analysis and action.
The same staff also may be responsible for processing contractor billings. This function involves verify-
ing contractor production against submitted bills, and approving bills for payment.
Production of Traffic Data
The staff resources required to produce and publish traffic data are as important as the resources dedi-
cated to collecting the data. The number of office staff required are proportional to the number of traffic
counts made annually. For comparison, data analysis and report production in one small state has six
staff where in one large state eight staff are used.

Agencies maintain enough staff to produce annual reports, though agencies may desire additional staff
to perform more in-depth research or studies of its traffic data to uncover new trends or identify unique
characteristics. Without adequate analysis staff there is little or no time for these endeavors.
Agencies should strive to cross-train staff to avoid creating data analysis “silos.” This is especially impor-
tant for agencies with a staff nearing retirement or where there is a trend of increasing turnover. Staff
are one of an agency’s greatest assets and loss of institutional memory can have severe impacts on the
program’s efficiency.
Agencies may find efficiency gains by assigning staff to defined regions. For example, Texas assigns two
FTEs for each third of the state. There are several advantages to this staffing approach. Staff become
familiar with the seasonal and annual characteristics within their assigned region. Staff work with all
data types and gain proficiency. If vacancies occur on other regional teams, staff can be shared between
regions to ease the vacancies’ impact. One disadvantage to consider for cross-training is that the ability
to maintain analysis consistency, or to reduce the variability of professional judgment, is reduced or lost
with many analysts if business rules for traffic monitoring are not developed and followed.
Some agencies are beginning to create large database systems with defined business rules to identify sus-
pect data for further review. These systems free valuable staff time to concentrate on those counts that
fall outside of normal ranges. Agencies should document the business rules used by staff when exercising
their professional judgment. These documented business rules can then be programmed so that software
programs or database queries can be built. These systems require a significant investment, but can pro-
vide agencies with greater staff productivity and the possibility of conducting nonroutine research work
to improve their traffic data program.
Site Installation/Maintenance Support, Including Inspectors

High-quality traffic data are a product of both a strong sampling plan and aggressive maintenance pro-
grams for continuous count sites. Agencies should strive to identify mechanisms that promote efficiency
and improve maintenance response. For centralized programs, efficiency can be gained through coop-
erative agreements with other agency staff in remote areas performing light maintenance work. Some
states use a decentralized model where technicians are assigned regional work areas and are responsible
for maintaining sites within their region. If contractors are used to perform maintenance, agencies will
require a more modest level of resources for inspecting and verifying contractor performance.
Maintenance and repair of portable equipment is usually performed at a central facility.
States may benefit from support mechanisms, such as regional or district assistance from state DOT
offices or the use of contractors, for data collection equipment installation and, more importantly, main-
tenance. The benefits of using local staff are travel time and overnight travel cost savings for centrally
located staff to conduct minor or routine maintenance functions.

It is important to note that well-trained inspectors play just as an important role in monitoring
the installation of permanent count stations installed by contractors. Remote staff that perform part-
time inspector duties can be trained by central staff to ensure quality installation or maintenance
is performed.
Estimation Procedure
The state force staff estimate is the sum of the field data collectors; field inspectors; office data collection
support staff for scheduling, data processing, maintenance, installation and maintenance inspectors, and
production staff. There are opportunities to reduce the number of FTEs through the use of contractors,
and coordinated and cooperative data collection and sharing agreements. Use of these other available
forces are discussed in the next section.
2.5.2 Other Available Forces

There are effective ways of reducing traffic data collection needs. Agencies should consider the efficiency
benefits gained from using traffic data collected by other sources. Agencies may be hesitant to use data
from other sources when they cannot confirm the data were collected in a manner consistent with the
quality standards of their own agency. These issues should be identified and effort put toward their reso-
lution. Other available sources for traffic data collection are use of contractors, urban traffic management
centers, local agencies, and bordering states or countries.
Use of Contractors
Recent trends indicate that state traffic data collection programs will continue to use or shift more data
collection responsibility to contractors. These contracts are necessary with reductions in state agency
employees. Agencies cannot support the demand for traffic data collection with greatly reduced staff.
Contractors frequently are used for the collection of short-term, volume, and vehicle-classification
counts. Contractors also are used for the installation and maintenance of permanent sites.
Agencies should consider using performance-based specifications within their contracts where possible.
Illinois, for example, issues payment only for accepted short-term volume counts. Texas operates under a
similar payment policy but also includes provisions for half payments if it is not possible for the con-
tractor to physically setup a count (e.g., count schedule for a roadway under construction) or if there is
physical damage to the counter.
Installation or maintenance contracts can be structured in a similar manner. Installation contracts

should base payment on a successful demonstration that the installed equipment is working properly.
The demonstration method should be described in the contract. Maintenance contracts may be struc-
tured on a work order basis for on-demand repairs. Agencies should not guarantee quantities in the con-

tracts. Performance also should have established timelines. Ohio allows contractors to complete main-
tenance within 30 days of receipt of the work order. If there are problems with the work, contractors
should be required to correct the problem before payment is made. Contracts also should include war-
ranties. For example, Ohio requires a one-year warranty on all maintenance work performed.
Successful contract management requires validation through the use of inspectors, open dialogue with
the contractor, and a conflict-resolution process. Inspectors are important to validate contractor perfor-
mance in collecting data, installing permanent sites, or conducting routine or on-demand maintenance.
Agencies should maintain an open dialogue—pre-work meetings, telephone calls, and e-mail—with
their contractors to identify any problems or concerns before they become schedule delays or conflicts. If
conflicts do arise, the agency should seek to resolve them through a defined conflict resolution process.
Coordination with State Regional Traffic Management Centers

Urban areas are developing and operating regional traffic management centers. There is typically one
traffic management center for a region that monitors in real time the traffic flow conditions on
freeways and possibly some arterial streets. Traffic management centers are used to promote the op-
erations of the transportation system. Which is clearly a different mission than that of the traffic data
collection program.
These centers can collect data at very small time increments (e.g., 20 seconds) that could be aggregated
later into large time increments (e.g., hourly or daily). The data collected are volume, lane occupancy,
point speed, or travel time. Some centers archive this data that could possibly be used in generating
system-level estimates.
Agencies may wish to open discussion to create agreements for archiving this data. Additional discus-
sions should identify how the data will be shared from the traffic management center with the agency.
Agencies may question the quality—site locations, thoroughness/completeness of data, maintenance
procedures, or installation methods—of the traffic management center data for planning purposes.
Open discussion should address these quality concerns and negotiate to resolve them, if possible.
Suggested steps to incorporate these resources into the traffic data collection program include:
• Traffic data collection agencies meet for open discussion with traffic management center. This discus-
sion initiates the discovery phase for each group.
• Operations agencies identify the data collected, site locations, collection frequency, archive frequency,
and quality issues. Questions data collection agencies should ask at this phase are:
– Do the data collected complement the agency traffic data collection program?
– Are gaps in the system-level monitoring filled by the traffic management center?

– Are there other uses for traffic management center data (e.g., anchor points for the ramp
balancing process)?
• Data collection agencies negotiate issues to resolve differences between the monitoring systems.
Agencies may wish to offer maintenance support or other assets to balance the demands for the traffic
management center data.
• Data collection and operations agencies identify mechanisms for accessing or transferring the data
from traffic management center. Mechanisms for exchanging data may need to be designed for frequent
data transfer.
Local Agency Involvement

One way to reduce the need for resources is through cooperative traffic programs within states. Other
governmental agencies and the private sector may be collecting traffic data on some state roads, and
likely non-state roads. If the counts are in accord with documented agency guidelines or other citable
reference such as an American Society for Testing and Materials (ASTM) standard or practice, the data
are being collected and summarized in a way compatible with recommended state practice.
Data quality is improved by using local agencies. These local agencies have a vested interest in obtaining
quality data to accurately represent their system and for their use in engineering and planning functions.
Local agencies provide an invaluable knowledge base on current road conditions and special events so
that abnormal counts are avoided on these traffic segments and times of the year. Local agency data col-
lection also presents opportunities to improve cost efficiencies.
Local and regional agencies are working toward creating and populating data warehouses. These data
warehouses can be a central submission point for traffic counts in the local area or region. State agencies
also could participate by allowing links to or populating data within the warehouse.
Cooperative traffic monitoring within a state by all interested parties presents the possibility of satisfy-
ing all coverage, special, and data obsolescence needs within a state. It is recommended that dialogue be
established within each state to assure quality in standard data and to exchange both traffic base data
and summary statistics.
Cooperative Sharing Agreements with Border States/Countries

Agencies may reduce some data collection needs by sharing base data and summary statistics on roads
shared with bordering states and countries. Provided the guidelines are followed by neighboring agen-
cies, the issue of differing traffic volume summary statistics can be resolved by adjoining agencies using
the most recent traffic count at or near the shared border.

Factors to Consider in Negotiating Effective Traffic Data Partnerships

Agencies entering partnerships to collect traffic data must identify and resolve various issues. Some
issues to consider include:
• Equipment accuracy and confirmation of routine maintenance;

• Timeliness of data collection;
• Sufficient metadata (where, when, duration, raw count, factored count, factors, etc.);
• Uniform data format;
• Fiscal responsibility for installation and maintenance, or manual collection; and
• Potential for reallocation of total resources (when duplication is avoided).
Agencies must identify critical issues early in the process. By not doing so, negotiations may unex-
pectedly halt or progress very slowly. Critical issues to either party will require time to resolve, and may
impact data collection targets. Therefore, agencies should strive to negotiate to a cooperative agreement.
2.6 Conclusion
The recommended traffic data collection plan requires more than adoption of preferred field procedures.
It requires that field procedures take place in the context of a statistically sound and well-planned sam-
pling plan and efficiently scheduled data collection. Such a plan includes identifying the different types
of data collect needs, the annual extent of each need, and the number of employees required to plan,
collect, process, and analyze the data.
With this information, agencies have the ability to make informed decisions matching data with field
personnel needs. When field procedures are clarified and adopted, and a data collection plan established,
traffic data may be effectively and efficiently collected.

2.7 References
FHWA. Highway Performance Monitoring System Field Manual, December 2000.

FHWA. Traffic Monitoring Guide, May 2001.
Florida Department of Transportation. FDOT Long-Range Program Plan for Fiscal Years 2006–2007
Through 2010-2011, revised September 28, 2005.
Florida Department of Transportation. Florida Project Traffic Forecasting Handbook, October 2002.
Frawley, W. E. A Methodology for Randomly Selecting Traffic Count Locations in Texas—Results and
Benefits. Presented at the North American Travel and Monitoring Exposition and Conference,
June 27–July 1, 2004, San Diego, California.
Southwestern Pennsylvania Commission. Congestion Management Process Glossary. Pittsburgh,
Pennsylvania, August 2006.
TransTech Management Inc. Strategic Performance Measures for State Departments of Transportation:
A Handbook for CEOs and Executives, Final Report, NCHRP Project 20-24(20). NCHRP,
Washington, DC, June 2003.

14367_02

Uploaded by

Copyright:

Available Formats

14367_02

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

14367_02

Uploaded by

Copyright:

Available Formats

CHAPTER 2

Traffic Data Collection Needs

2.1.1 Role of the Traffic Monitoring Guide (TMG)

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-1

2.1.2 Role of the Highway Performance Monitoring System (HPMS)

2.1.3 Growing Importance of Operations and ITS Data

2.1.4 Chapter Organization

2.2 Traffic Count Types and Needs

2-2 GUIDELINES FOR TRAFFIC DATA PROGRAMS

2.2.1 Coverage Counts

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-3

Table 2-1. Examples of Highway Trafﬁc Data Uses*

Highway Trafﬁc Vehicle Truck Speed

Engineering Highway Geometry Pavement Design Pavement Design Traﬃc Control

*Adapted from FHWA, Traffic Monitoring Guide, May 2001.

2-4 GUIDELINES FOR TRAFFIC DATA PROGRAMS

Table 2-2. Examples of Customer Needs for Trafﬁc Data

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-5

2-6 GUIDELINES FOR TRAFFIC DATA PROGRAMS

2.2.2 Federal or State Research Initiatives

2.2.3 Project-Related Counts

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-7

2.2.4 Special Count Requests

2.2.5 Performance Measurement

2-8 GUIDELINES FOR TRAFFIC DATA PROGRAMS

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-9

Table 2-3. Sample Performance Measures by Type

Vehicles per lane-mile •

Truck miles traveled •

Person miles traveled •

Average travel time •

Person-miles of travel in congested ranges • •

Average travel rate •

Growth in annual delay per peak road traveler •

Person-miles of travel in congested ranges •

Person-hours of travel in congested ranges •

Person movement speed •

Delay per vehicle-mile •

Total delay per mile •

Average travel time •

Traﬃc Management Centers Lane occupancy data •

Average travel time •

Time to incident detection •

2-10 GUIDELINES FOR TRAFFIC DATA PROGRAMS

2.3 Estimating Traffic Data Collection Needs

2.3.1 The Sampling Plan

The Development Process

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-11

• Define Stratification Groups—Stratification groups may be structured by geography, area type,

2-12 GUIDELINES FOR TRAFFIC DATA PROGRAMS

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-13

count. Likewise, vehicle classification counts also provide volume counts.

Figure 2-1. Example of Count Nesting.8

2-14 GUIDELINES FOR TRAFFIC DATA PROGRAMS

Classification Sites and WIM Sites

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-15

Field Review for Site Adequacy

2-16 GUIDELINES FOR TRAFFIC DATA PROGRAMS

Failed Counts—Machine/Collector Error or Malfunction

GUIDELINES FOR TRAFFIC DATA PROGRAMS 2-17