Guidelines

for Traffic
Counting
May 2013
Guidelines
for Traffic
Counting
Matthew Rodwell
Hastings District Council
Abbreviations
ADT Average Daily Traffic

AADT Annual Average Daily Traffic

HCV Heavy Commercial Vehicle

HPMV High Productivity Motor Vehicle

NZTA NZ Transport Agency

RAMM Road Asset and Maintenance Management

RCA Road Controlling Authority

TMS Traffic Monitoring System

VKT Vehicle kilometres travelled

GUIDELINES FOR TRAFFIC COUNTING 5

Contents
Abbreviations 5

1.0 Introduction 9
1.1 Background 9
1.2 Rationale 10

2.0 Why Collect Traffic Data? 11

3.0 Objectives 12

4.0 Principles 12

5.0 Traffic Counting Module 13

6.0 Improving traffic counting programmes’ efficiency 14

6.1 Increasing a traffic counting programme’s coverage 14
6.2 Improving the targeting of traffic counting programmes 15

7.0 Improving Traffic Estimates 16

7.1 Improving the estimation of AADT from short duration traffic counts 16
7.2 Improving the updating of estimates 16
7.3 Improving the estimates on uncounted road sections 17

8.0 Other Considerations for a Traffic Counting Programme 18

9.0 Where to go to learn how to use the RAMM Module 20

GUIDELINES FOR TRAFFIC COUNTING 7

1. Introduction
This guide is based on the results of research aimed at improving the quality of traffic flow data
collected by RCAs and stored in their Road Asset and Maintenance Management (RAMM) databases.
Traffic count data is important because it forms the basis for the national and local traffic flow
estimates used in many planning and decision-making contexts.

The research results propose a minimum standard for a traffic monitoring framework to significantly
improve the overall accuracy and efficiency (and therefore value for money) of typical traffic data
collection regimes.

The sampling framework’s key foundations are based around one key assumption and two types of
sample:
§§ A traffic link model based on the assumption that adjacent sections of road carrying similar
levels of traffic can be amalgamated or related. As this model reduces the sampling framework,
the number of road sections for which traffic flow data are required decreases.
§§ A core annual monitoring sample of between 3% and 7% of total traffic links (selected as a
stratified random sample based on expected Annual Average Daily Traffic (AADT) that allows
travel activity monitoring in the RCA as a whole)
§§ a rotational sample drawn to cover the bulk of road network travel (the top 20% of traffic links,
based on the contribution that each link makes to total vkt in the Road Controlling Authority (RCA)
– the 20% typically carries around 80% of travel activity). On some networks this sample may be
collected annually while on others the sample could be spread over two years.
The degree to which the new traffic counting framework overlaps with existing practices depends
greatly on the size and structure of the current monitoring strategy.

1.1 Background
Access to reliable data on the number and types of vehicles using New Zealand’s roads has been an
important input for a range of road asset management, transportation planning, and transport and
environmental policy activities. The primary source of traffic volume and vehicle travel data in New
Zealand is an aggregation of the individual Road Assessment and Maintenance Management (RAMM)
databases operated by RCAs. Primarily a road management tool, RAMM databases contain a variety
of information about the physical road asset, including traffic count data and estimates of the AADT.

The RAMM databases for the state highway network are populated with traffic volume data from
the NZ Transport Agency’s (NZTA) Traffic Monitoring System (TMS). TMS uses a combination of
continuous and non-continuous traffic counts to generate estimated AADT on every section of the
state highway network.

Unlike the arrangements for monitoring traffic on state highway network, each local authority
operates their own database, populating the traffic data fields with traffic volumes collected, typically
from short duration ( seven-day) traffic counts, or from estimates based on previous traffic counts,
traffic counts undertaken on adjacent road sections, or local knowledge. The average of the counted
data provides only an Average Daily Traffic (ADT), as the data is not adjusted seasonally.

The currency, quality and completeness of the traffic information contained in these databases is
extremely variable.

GUIDELINES FOR TRAFFIC COUNTING 9

1.2 Rationale
The clear message is the provision of up to date traffic count data on New Zealand’s local roads is
extremely variable and generally poor overall, making it unlikely to meet planning and policy needs.

The lack of consistent and reliable data on the traffic using New Zealand’s local authority roads is not
a new problem. Over the past 14 years there have been various attempts at creating a national traffic
monitoring system to meet the needs of a range of agencies for a range of purposes. Centrally driven,
these initiatives have lacked the resources necessary for a nationwide initiative and have failed to
provide on-going development and support. As a consequence, these national traffic databases have
not been maintained, and their usefulness has declined.

This guide represents a significant change in thinking, proposing a bottom-up process that seeks to
provide RCAs with the knowledge, tools and systems to improve their local traffic monitoring.

In general terms the approach seeks to improve and rationalise the traffic counting programmes
operated by individual RCAs, and to automate propagation of these traffic counts across the road
network models held within the RAMM databases. In doing so, the project seeks to improve traffic
monitoring practices by:
§§ Not placing undue pressures on the limited staff and financial resources of the RCAs
§§ Providing recognisable efficiency benefits to those using the system.
Improvements in the coverage and quality of local traffic monitoring will benefit the RCAs collecting
the data. It will also improve the basis for a national traffic monitoring system as well as network
performance monitoring and benchmarking.

10 GUIDELINES FOR TRAFFIC COUNTING

2. Why Collect Traffic Data?
The ADT or AADT is a figure used to describe the current traffic demand on the network (showing
how busy the roads are). A large number of asset management processes use the ADT or AADT figure
to provide decision making information.

There are two principal reasons for RCAs to have accurate estimates of road traffic volumes:
»» Funding
NZTA Planning and Investment group have auditors to check the accuracy of the information
provided to support funding applications.

»» Better Decision Making

More accurate information leads to more consistent and robust analysis and confidence in results.
This improves credibility and enables councils to better direct resources.

Table 1: Uses of traffic count data in a typical RCA.

Long term Network Renewal & Customer Service Operational

planning Development

Understanding route Project planning (e.g. Responding to Traffic management

use and hierarchy optimum working enquiries requirements
periods)

TSA & Predictive Deriving traffic loads for Development control Regulatory
condition modelling pavement design and planning requirements
(dTIMS etc)

Safety studies and Deriving traffic loads Network monitoring

crash analysis bridge design

Traffic modelling and Project economics Deriving traffic loads

simulation (BCR) for planning and
design of maintenance
intervention
strategies (e.g. reseal
programme)

Setting and measuring Enables a network by

levels of service network comparison.

Policy development

Vehicle Kilometres
Travelled (VKT)

HCV & HPMV

management

GUIDELINES FOR TRAFFIC COUNTING 11

3. Objectives
A traffic counting strategy’s primary objective is to ensure traffic volumes across the network are
known in sufficient detail and to an appropriate level of a ccuracy to inform the tools and processes
used by RCAs and NZTA that rely on traffic volumes. Traffic count data should be collected as
efficiently as possible to meet this objective.

4. Principles
Traffic counting needs vary depending on the intended use for the data. The following matrix
describes each core activity’s traffic volume requirements.

Table 2: Traffic Data Demands.

Traffic modelling & simulation

Maintenance intervention
Traffic management req.
Network understanding

Responding to enquiries

Development control
Condition modelling

Project economics
Project planning
Levels of Service

Regulatory req.
Design loadings
Safety studies
TSA

Volume • • • • • • • • • • • • •
Data Required

Classification • • • • • • • • • • • • • •

Class speed • • • • • • • • • •

Cycle counts • • • • • • •

The principle of the traffic counting strategy should be to collect data to inform the above activities in
a manner that accurately estimates, and is sensitive to the changes in:
§§ Network use (volume and classification)
§§ Speed and direction of travel
§§ Growth
§§ Seasonal variation.

12 GUIDELINES FOR TRAFFIC COUNTING

5. Traffic Counting Module
The ideal means of establishing the traffic usage of the network would be to count the traffic using
every road section, every day of the year. Clearly this is not practical. The alternative is to undertake
short duration traffic counts on a sample of road sections, and use this data to develop informed
estimates of the actual traffic usage on the network. However, this approach results in a number of
consistency and data accuracy issues.

To help address these issues, a traffic counting model has been developed by MWH New Zealand
Limited on behalf of the Ministry of Transport and the NZTA. It was implemented within the RAMM
dataset system by RAMM Software Limited. The RAMM module was commissioned to improve the
quality of RAMM’s traffic count data, and to address three issues identified in this study and previous
studies by:

§§ Improving traffic counting programme efficiency by:

§§ Increasing existing traffic counting programmes’ coverage
§§ Improving national cost effectiveness
§§ Improving traffic counting programme targeting

§§ Improving traffic estimates by:

§§ Improving AADT estimates from short term traffic counts
§§ Understanding network-wide traffic mix distributions better
§§ Improving estimate updating
§§ Improving uncounted road section estimates.

This guide uses the RAMM module process to describe and illustrate the proposed framework’s
principles. However, the principles and framework described could be equally applied to developing a
traffic counting strategy independent from RAMM.

GUIDELINES FOR TRAFFIC COUNTING 13

6. Improving traffic
counting programmes’
efficiency
6.1 Increasing a traffic counting
programme’s coverage
The RAMM database contains a schedule of road sections that make up the highway network. The
sections are typically defined in terms of the road asset’s physical characteristics, rather than the
traffic travelling on particular sections. For example, the road section between the two side roads
(see below) is broken into three carriageway sections, representing the changes in the physical
attributes of the road, the bridge and two sections on either side of it. However, the traffic using each
of the three carriageway sections is the same.

Figure 1: Traffic Link and Carriageway Section Relationship.

bridge

Carriageway Carriageway Carriageway

Section 1 Section 2 Section 3

Traffic Link 1

The traffic counting module introduces the concept of traffic links, or aggregations of RAMM road
sections that carry essentially the same traffic. This is similar to the method used by NZTA Highways
and Network Operations group on state highways. The principal reason for creating a traffic link
network is to provide a means of reducing the overall size of the database for which traffic data will
be required.

14 GUIDELINES FOR TRAFFIC COUNTING

There are various ways in which a traffic link network can be created:

§§ One way is to review each historic traffic count site and determine whether the carriageway
sections either side can be combined with that which has been monitored historically. This
process is continued working out from each site.
§§ Where the RCA operates a traffic model, this is effectively the traffic link network, and the
carriageway sections in the RAMM database can be combined to mirror the traffic model network.
§§ Another way is to use the routine developed within the RAMM Traffic Counting module to suggest
which RAMM carriageway sections might be combined, based on the current traffic estimates. A
pre-requisite for this process is to update all traffic estimates to the same base year.
Although creating a traffic link system is not essential, it will improve the traffic counting
programme’s efficiency. If in doubt, a couple of very short duration counts (an hour or so) may be
used to determine if sections can be combined. It is however, worth noting that while the decision to
combine adjacent sections may be easy to make where the vast majority of vehicles pass from one
section to another, more care is required where there is a lot of turning traffic. As this process is
based on traffic volumes, if traffic volumes and composition are expected to remain essentially equal
on adjacent sections, the sections can be combined to form traffic links.

When setting up the traffic link model, asset managers need to be diligent in incorporating network
knowledge and understanding of forecast growth and network change. This will ensure the traffic link
model appropriately reflects changes in network characteristics such traffic mix/composition as well
as volume.

6.2 Improving the targeting of

traffic counting programmes
The traffic data module does not specify detailed traffic counting programmes, but together with the
associated RAMM manual, guidance is given and tools provided that allow the user to create or revise
their traffic counting programme.

Two components of traffic counting strategies are suggested:

»» Core Sample – a randomly selected sample, typically of between 3% and 7% of traffic links, that
when surveyed annually will provide an estimated mean AADT. This can be multiplied by network
length to provide an estimate of network travel (vehicle kilometres of travel-VKT) that will be
within +/- 10% of the true volume of travel at least 90% of the time. The guide provides details of
the likely sample size for each RCA as well as the method of selection.

»» Rotational Sample – that portion of the traffic network, typically around 17% of traffic links) on
which more than 80% of total network travel is undertaken. The software provides the means by
which these links can be identified and managed.

In practice, it is possible to ensure the bulk of the Core sample is contained within the Rotational
Sample. Because the Rotational Sample covers the bulk of network travel, this sample will provide
network wide growth estimates as well as ensuring year by year information is collected on the key
traffic links.

GUIDELINES FOR TRAFFIC COUNTING 15

7. Improving
Traffic Estimates
The RAMM database holds estimates of the AADT for each section of road on a network. These
estimates are accessed and used in applications such as pavement deterioration modelling. The
estimated AADT figures are typically based on one or more short duration (typically seven-day) traffic
surveys, either on the road section in question or an immediately adjacent road section.

This creates confusion and raises three issues, confirmed by a survey of local authority traffic
monitoring practices:
§§ The “conversion” of a short duration traffic count to a “reliable” estimate of AADT
§§ The manner in which traffic counts on one road section may or could be used to estimate traffic
volumes on an adjacent road section
§§ Efficiently up-dating traffic estimates.

7.1 Improving the estimation of AADT

from short duration traffic counts
The volume of traffic using a particular road varies throughout the year depending on the road type.
In some cases this variation can be as much as +/- 30%. If a seven-day traffic count is undertaken in
late January on a road leading to a popular beach area, average daily traffic measured in the survey is
likely to be a significant over estimate of the AADT. Similarly, a traffic count undertaken during winter
months will under estimate the true AADT. While it is possible to undertake a series of short duration
traffic counts throughout the year, this simply increases the cost associated with obtaining a reliable
estimate.

To assist, the traffic counting module enables the user to allocate traffic count sites to a series of
seasonal profiles; a series of criteria are provided to aid allocation. The seasonal profiles are used to
obtain an estimate of AADT from the short duration traffic count. If there are multiple short duration
traffic counts undertaken in a year, each traffic count provides a separate of AADT and these are
averaged by the software to provide a more reliable estimate of AADT on the counted link. This
routine ensures that estimates are updated when new counts are entered.

7.2 Improving the updating of estimates

As the majority of local authority traffic count data in RAMM is typically incomplete, many associated
applications use the estimated AADT values, not the traffic count data. It is therefore important that
estimates are kept up to date.

The traffic data module enables the user to allocate the various traffic count sites/traffic links to
traffic growth groups. When the update routine is run, the traffic growth recorded for those links with
a traffic count history is reported and used to update those links in the group that were not counted in
the last year.

Currently there is no basis or routine for recommending traffic links and updating estimates on
anything other than traffic volume and count history. Other important factors already discussed in
Section 6.1 need to be considered when defining the traffic link model to ensure updated estimates
accurately reflect factors such as traffic mix (classification) as well as AADT.

16 GUIDELINES FOR TRAFFIC COUNTING

7.3 Improving the estimates on
uncounted road sections
The traffic link concept described in section 6 effectively links the carriageway sections so:

Estimate on section 1 = 100% of estimate on section 2.

This concept can be extended further to identify relationships between the traffic volumes on
adjacent or associated links and, in doing so, extend the coverage of the counting programme. In this
way, the traffic on other uncounted carriageway sections or traffic links is related to a traffic count
site and associated estimates. This means when a traffic count is undertaken at X, the estimated ADT
(or some portion of it) is automatically used to update the estimates on road sections linked to that
count site:

Figure 2: ADT Estimation on Uncounted Traffic Links

0.2X

0.2X

0.65X Traffic Link #1: AADT = X

0.95X
Count
Site

0.45X = Y 0.5Y
0.5Y

This ability to link road sections increases the network coverage of a particular traffic counting plan.

GUIDELINES FOR TRAFFIC COUNTING 17

8. Other Considerations
for a Traffic Counting
Programme
There is a fundamental difference in the relationship between links for private vehicles and heavy
commercial vehicles, particularly HPMVs that operate on permitted routes only. The RAMM module
is not currently designed to establish different link relationships for different vehicle classifications
so this element requires a manual audit after the annual update of link volumes. When auditing the
count programme, it is also important to consider the following (you may wish to designate certain
count stations as providing these additional functions).

»» Growth Node:
Sites tracking strategic growth trends. Also absorbing the representative sites for verifying low
volume road and growth estimation.

»» Seasonal:
Sites tracking key tourism and industrial fluctuations and for establishing seasonal profiles

»» Screenline:
It is important when providing traffic information for modelling purposes that counts are located
on the road network in such a way that they form screenlines and/or cordons. A screenline is a
notional line across the road network which delineates directional or sectoral movements across
the area. A cordon is a notional “ring” which delineates movement in and out of an enclosed area
such as a suburb or core part of a central business district (CBD). The figure below shows an
example of two sets of screenlines and two cordons. These screenlines and cordons were chosen
to both capture the primary movement patterns across the network while intersecting as few
roads as possible (thereby lessening the monitoring effort).

18 GUIDELINES FOR TRAFFIC COUNTING

Figure 3: Count Screenlines and Cordons.

Table 3: Example of a Traffic Counting Framework.

Count Purpose Location Criteria Frequency

Type

Core Monitoring Whole Network Random sample Annually

Monitoring travel activity
Sample on the network
as a whole

Rotational Monitoring Arterial, Collector, High Level Traffic Annually: randomly

Sample traffic flows to Distributor, VKT >5000 ADT + all selected count
capture 80% Regional Arterial Heavy Traffic VKT dates.
of VKT on the sites
network
Local, Local CBD Based on Low Level Bi-annually:
VKT <5000 ADT randomly selected
count dates.

Growth Track strategic Nominated Urban and Rural Annually: counted

Counts* growth trends development areas count sites selected in the same month
and representative on their strategic and each year.
sites for verifying base line value.
growth on low
volume roads.

GUIDELINES FOR TRAFFIC COUNTING 19

Count Purpose Location Criteria Frequency
Type

Seasonal Track key Nominated key Urban and rural Each count site
Counts* tourism and tourism and count sites selected used once every
industrial industrial roads on their strategic and five years.
fluctuations base line value. Seven-day counts
every month within
the count year.

Screenline Assist with Located on North/ All roads crossing As per Rotational
Counts* calibration of South and East/ screenlines and Sample sites
the strategic West screen lines cordons
regional and an inner and
transport outer cordon
model and
Hastings Urban
model.

Cycle Monitoring Routes as defined Urban and rural Annually: counted

Counts cycle flows to by cycling strategy count sites selected in the same month
capture use on their strategic and as the Growth
of designated base line value. Counts
cycle routes

* Note: Sites to be within rotational sample to minimise addition of counts to programme.

9. Where to go to learn how

to use the RAMM Module
The RAMM Traffic Count Estimation module uses the methodology described in this guide. It is
designed to enable you to combine historical traffic data with intelligent Carriageway Section linking.
This in turn enables you to produce a traffic counting and estimation programme which delivers
the most Network coverage, and the most accurate and up to date ADT Estimates for the minimum
number of counts.

To find out more on how to start using the RAMM Traffic Module, go to:
www.ramm.co.nz/manuals and select “Traffic Count Estimation”

20 GUIDELINES FOR TRAFFIC COUNTING