Traffic Forecasting on Trunk Highways in Nonmetropolitan Areas

Transportation Research Synthesis: Traffic Forecasting on Trunk Highways in Nonmetropolitan Areas: A Survey of State Practice Presentation to MnDOT Technical Advisory Panel Wednesday, September 26, 2012 1:00 p.m. Christine Kline Patrick C. Casey CTC & Associates LLC Madison, Wisconsin

Background Traffic forecasting plays a critical role for MnDOT in corridor planning, geometric design, pavement design, safety analysis, access management and more. In previous decades, Minnesota has seen steady growth in vehicle miles traveled. In recent years, VMT growth in Minnesota has been flat and declined by 0.5% from 2009 to 2010.

TRS project scope MnDOT s Office of Transportation Data & Analysis wanted to know: How other states are dealing with what appears to be a nationwide trend in a leveling off or decline in VMT. The forecasting methods used by other state DOTs to project traffic volumes in nonmetropolitan areas with a population under 50,000.

Technical Advisory Panel Shirlee Sherkow, Project Coordinator Chu Wei, Technical Liaison Technical Advisory Panel Members Lynne Bly Jim Miles Gene Hicks Tom Nelson Jason Junge Paul Stine James McCarthy

Methodology We distributed an online SurveyMonkey survey to members of the AASHTO Standing Committee on Planning. Nine-question survey, reviewed and approved by the TAP, of open-ended and multiple-choice questions July 30 email announced the survey, with two weeks for responses (August 10 survey deadline) August 7 reminder email encouraged more responses September 7 final draft of report sent to TAP

Methodology The survey gathered information in four key areas: Methodologies and tools used to estimate future traffic volumes Data and factors used in forecasting Projection time periods Flattening or decrease in VMT

Survey response Thirty states provided survey responses. Survey Respondents Arizona Colorado Connecticut Florida Illinois Iowa Kansas Kentucky Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico New York North Carolina North Dakota Oregon Pennsylvania South Dakota Texas Utah West Virginia Wisconsin Wyoming

Survey response Rhode Island provided an email response All planning is performed by the state s metropolitan planning organization; Rhode Island DOT has no forecasting procedures for nonmetropolitan areas Not all respondents answered every question Percentages noted in survey results are adjusted accordingly

Survey results Question 1: Please indicate the methodology(ies) used in your agency s program to estimate future traffic volumes. Regression Models State Box-Cox linear regression WI Cubic regression NM Least squares regression AZ, KS, NM, PA Linear regression AZ, CT, FL, IA, IL, KS, KY, MA, ME, MI, MN, MO, MS, MT, NC, ND, NE, NV, NY, OR, SD, TX, UT, WI, WV, WY Logistic regression AZ, WI Multinomial regression (for mode split) CT Nonparametric regression FL, KY

Survey results Question 1: Please indicate the methodology(ies) used in your agency s program to estimate future traffic volumes. All responding states but four Colorado, Maryland, New Mexico and Pennsylvania use linear regression models to estimate future traffic volumes. Seven states make use of multiple regression models: Arizona (least squares, linear, logistic) Connecticut (linear, multinomial) Florida (linear, nonparametric) Kansas (least squares, linear) Kentucky (linear, nonparametric) New Mexico (cubic, least squares) Wisconsin (Box-Cox, linear, logistic)

Survey results Question 1: Please indicate the methodology(ies) used in your agency s program to estimate future traffic volumes. Respondents reported other methodologies used to forecast traffic volumes, including: Growth rate from similar sites (Nevada) Historical trend analysis (Colorado) Linear or parabolic growth rates based on the knowledge of local growth patterns (Maryland) Statewide or travel demand models (Arizona, Maine, Massachusetts, North Carolina, Oregon, Wisconsin)

Survey results Question 1a: Does your agency use a modeling software program to estimate future traffic volumes? Vendor Model/Program State Caliper Corporation TransCAD IA, MS, MT, NV, WY Citilabs Cube Voyager MD, ME, NY, UT Cititlabs TP+ (legacy system) MD, WI Citilabs Tranplan (legacy system) CT IHS Global Insight Statewide VMT macroeconomic model NY PTV America VISUM NM KY, MA, MD, MI, MN, OR,TX Not specified Statewide or travel demand models Only four states Colorado, Illinois, Kansas and South Dakota report no use of modeling software to forecast traffic volumes.

Survey results Question 2: How many years of historical data does your agency use when forecasting future traffic volumes?

Survey results Question 3: Please indicate the social and economic variables included in your agency s traffic forecasting methodology. Socioeconomic Variable Responses Number of Number of Responses Socioeconomic Variable Total population 22 Unemployment rate 3 Employment 21 Motor vehicle registration 3 Households 21 Fuel consumption 2 Personal income 9 Driving age population 2 Labor force 6 Population age 16 and over 2 Gas prices 4 Population age 65 and over 2 Only four states Colorado, Florida, Missouri and Nebraska do not report the use of socioeconomic variables in forecasting models.

Survey results Question 3a: If socioeconomic variables are included in your agency's traffic forecasting methodology, please describe how they are employed. Socioeconomic variables most often serve as inputs to respondents travel demand or statewide models. Socioeconomic variables are also used to: Analyze trends. Determine impact on current traffic. Determine trip generation/trip distribution/trip attraction. Develop growth profiles. Influence choice of growth rate.

Survey results Question 4: When forecasting traffic volumes, does your agency apply minimum and maximum growth factors? States Using Minimum Growth Factors States Using Maximum Growth Factors Factor Percentage Factor Percentage State (14) State (8) AZ, MS, ND, NM None specified AZ, MS, NM, WV None specified OR 0% MA 1.5% KS, MA, ME, MN, NV, WI 0.5% MN 3% MO 0.5 % to 1% MT 3.5% MT 1% TX 5% TX 2%

Survey results Question 5: Does your agency apply different growth rates to heavy commercial traffic versus total traffic volume? Twelve states apply different growth rates to heavy commercial traffic and total traffic volume. AZ applies different factors at statewide modeling level, not at a micro level. Arizona Illinois Kentucky Maine Maryland Michigan Nevada New Mexico New York North Dakota Oregon Wisconsin NV has four methods for truck forecasting. NM uses data from FAF3 to forecast truck trips. OR uses different factors for its statewide integrated freight model.

Survey results Questions 6 and 6a: Please indicate the time periods included in your agency s projections of future traffic volumes and describe why these time periods were selected. Reasons for selecting time periods: Number of Responses Number of Responses Time Period Time Period Pavement/project design (12 states) 20 years 26 5 years 6 Required for long-range transportation plan (4 states) 10 years 14 40 years 5 30 years 12 35 years 4 FHWA standards (3 states) 25 years 10 50 years 1 15 years 8 Required for statewide model (3 states)

Survey results Question 7: Has your state experienced a flattening or decrease in VMT in nonmetropolitan areas? Twenty-three states are experiencing a flattening or decrease in VMT; seven states are not. Yes Reponses No Responses Arizona, Colorado, Connecticut, Illinois, Iowa, Kansas, Kentucky, Maine, Maryland, Massachusetts, Michigan, Minnesota, Mississippi, Missouri, Nebraska, Nevada, New York, North Carolina, Oregon, Pennsylvania, Utah, West Virginia, Wisconsin Florida, Montana, New Mexico, North Dakota, South Dakota, Texas, Wyoming

Survey results Question 8: If your state has experienced a flattening or decrease in VMT in nonmetropolitan areas, is your agency considering changes in the methodology used to forecast traffic volumes, or have changes already been adopted? Description of Traffic Forecasting Program State Flattening or Decrease in VMT and Considering Forecasting Changes CT, IA, IL, KS, KY, MA, MI, MN, NC, NY, OR Flattening or Decrease in VMT and Not Considering Forecasting Changes AZ, CO, MD, ME, MO, MS, NE, NV, PA, UT, WI, WV No Flattening or Decrease in VMT FL, MT, ND, NM, SD, TX, WY

Survey results Question 8: If your state has experienced a flattening or decrease in VMT in nonmetropolitan areas, is your agency considering changes in the methodology used to forecast traffic volumes, or have changes already been adopted? States are considering or making changes in: Growth rates Statewide models

Summary Thirty-one states responded to our request (30 online surveys completed) Most of the responding states use linear regression models to estimate future traffic volumes. Less commonality is found in the software models states use to estimate future traffic volumes. Most respondents use socioeconomic variables, most often as inputs to travel demand or statewide models. Survey results indicate that many other states are also experiencing a flattening or decrease in VMT. Almost half of these states are considering or implementing changes in forecasting methods.

Thank you! Questions?