Topics

• The Big Picture
• Model Input
  • Data and Surveys
    • Vehicle Speed Data
    • Vehicle Count Data
    • Household Travel Surveys
    • Work Place Travel Surveys
    • Stated Preference Surveys
    • Revealed Preference Surveys
    • Intercept Surveys
    • Special Generators
    • Land Use Data
    • Socioeconomic Data
    • Zonal Hierarchy
    • Transit Onboard Surveys
    • Transit Boarding and Alighting Counts
    • New Data Sources
    • Transit Speed Data
    • Transit Ridership Data
    • Data Transferability
    • Establishment Surveys
    • Visitor-Airport Surveys
  • Networks
• Model Development
• Model Application

Vehicle Count Data

Summary 

Traffic data is critical for validating and running travel demand models.  One of the many interesting features of traffic data is that the data is always just a sample.  Even continuous collection sites (such as ATRs) usually have some error built in (missing days/hours; errors due to the equipment).  All of the traffic data subtopics are related to an extent since they use similar data collection equipment.  Essentially though there are three main areas:

•Volume data;

•Vehicle classification data

•Speed data - covered in another  topic.

The other areas (data periods, equipment/technologies and screenline counts) are subsets of the above.

It is important to mention that traffic data is used for many other purposes than providing data to travel demand models and for traffic forecasting.  Some other uses of traffic data include:

•Pavement management;

•Construction management/lane closures

•Locating businesses (restaurants)

•Determination of funding (use of VMT for allocating federal matches);

•Determining speed limits;

•Safety analysis;

•Crash  analysis;

•Determination of level of service and other performance measures for highway facilities;

•Use for traffic impact studies;

•Use for signalizing intersections;

•Use for reporting highway conditions to Congress (VMT, VHT, etc.);

•Use by HPMS for assessing future highway demand (apart from similar efforts  using travel demand models).

•Use for air quality.

Key Issues: Traffic count variability, traffic data availability, traffic data trend analysis, hourly data.. ODME

Practical Tips:  Based on a recent model developed by WSA for a project in Louisville, Kentucky (Louiville Southern Indiana Ohio River Bridges project travel demand model development), a brief case study has been prepared.

Study Area: For this model, the study area spanned five counties in two different states. 

Data Needs: The needed data included ADT, truck percentages, hourly data, historical count data and new "special" counts on ramps and other locations not already covered. The map shows the location of the data points used.  There wee 1,391 data points with 534 locations having combined ADT & Truck, 749 locations having ADT & hourly data and 269 locations having ADT & Truck and hourly data.  INSERT MAP

Tips:

Station numbers for ID: In order for the linkage between the model network file (the repository of the count data) and the source data it is recommend to use the count station numbers (usually a six digit number). The list below shows  the traffic data attributes organized into ID info, core traffic data attributes and hourly data attributes.

ID Info:ID, Longitude, Latitude, Source, Count_Stat_ID, Dualized, CountyName, State

Core Traffic Data: Base_ADT,Base_ADT_Year,Truck Percentage, Count Type, SUT_Percentage, MUT_Percentage, S/M,

Hourly Volume Percentages:AM Percent, MD Percent, PM Percent, NT Percent, TRK_ADT, AM_Count, PM_Count, Drop, Revise_Loc, Hourly Volume %s (24 fields)

Data reconcilaation: When doing a model involving two states and a MPO, it is to be expected to have different standards and practices for the data collection so those will need to be reconciled. ADTs are usually the easiest data to collect for models but still may need some kind of factoring system to bring all of the data to the base year. In the LSIORB model the base year is 2008 so older data had to be factored forward and newer data possibly backcast.  In our case, with the recession impact, the model volumes between 2007 to 2010 were treated as the same year.

Truck data and hourly data: Truck percentage data is not as easily found and there were not databases in both states that could be easily used.  Hourly data is usually not archived except via pdf files so using that level of data required significant manual copying of data.It is likely better to work with default hourly or  period truck data rather  than try to disaggregate.  Also, it is not practical to work with  all of the vehicle classification types.

Special counts: New counts collected for the project or "special" counts need to be factored using the state procedures.

Data Requirements: For most travel demand models, a minimum of 24-hour volumes (ADTs) and truck percentages are needed.  Time of day models also require hourly or period data.  Most truck models also use the categories of heavy trucks, light trucks and commercial vehicles.

Visualization:  Visualization is a critical component of understanding traffic data.  Traffic count data is usually viewed in traffic count flow maps (see example from the Kentucky Transportation Cabinet for 2008 ADTs for the City of Lexington, Kentucky (http://www.planning.kytc.ky.gov/maps/count_maps/maps/lexington.pdf)..  Traffic data is also frequently shown in band widths to show the relative value of the traffic volumes.  The  dirurnal aspect of traffic counts is another frequently used tool to understand traffic data over a 24-hour period or longer periods (an example from Kentucky ATRs is shown on p. 28 of the referenced report:http://www.planning.kytc.ky.gov/maps/count_maps/maps/lexington.pdf) . Vehicle classification data is more difficult to visualize because of the numerous categories and is most frequently depicted using the flow maps, band widths and diurnal distibutions for truck percentages.

Reasonableness Checks: Due to the importance of the role of traffic data's use in model validation and model accuracy, it is critical to perform some review of the source data.  Typical checks might include corridor level review for consistency, progression and anomolies, a historial review of the data (at least 20 years back, possibly more due to the impact of the recession on traffic counts between 2006 and 2010), missing segments in network (especially ramps), and a review to see if the counts were performed properly (were they factored for seasonality/day of week, were they made for at least 24 hours, what are the limitations of the counting technology used).

History: Traffic data has been an integral part of evaluating highways since the Bureau of Public Roads started planning activities.  An example of 1918 flow map from Portland:

http://www.portlandonline.com/transportation/index.cfm?a=65817&c=36416http://www.portlandonline.com/transportation/index.cfm?a=65817&c=36416

A paper by David Albright gives a brief overview of traffic statistics:http://pubsindex.trb.org/view.aspx?id=365623

Sources of Good Work: Diurnal Distribution of Traffic Data: NCHRP 255: Highway Traffic Data for Urbanized Area Project Planning and Design ( http://pubsindex.trb.org/view.aspx?id=188432) provides tables for look-up values for the dirurnal distribution of traffic for various population groups.  This report  also gives default k-factors and directional distribution factors.  This standard for  traffic forecasting is in the process of being updated as a part of NCHPR 8-83.

K-factors and D-factors: KYTC Traffic Forecasting Report http://www.planning.kytc.ky.gov/traffic/traffic_files/Forecast%20Report%204-25-08_dah.pdf

Count programs and factoring: Traffic Monitoring Guidehttp://www.fhwa.dot.gov/ohim/tmguide/ gives best practice for traffic counting programs and for factoring individual data.

Validation/accuracy: Model Validation and Reasonableness Checking Manualhttp://tmip.fhwa.dot.gov/resources/clearinghouse/1397 gives insight into the recommended accuracy of modeling by volume group