APPENDIX B: SURVEY RESPONSES

Appendix B summarizes the individual agency responses to each question of the survey.

RESPONDENTS

A. DATA COLLECTION

1) How many short-term count sites do you have?

^a The table includes the states that have at least one short-term count site.

2) How many short-term counts do you conduct every year?

^a The table includes the states that conduct at least one short-term count every year.

3) What days of the week do you conduct short-term traffic counts? (please select all that apply)

4) Do you conduct short-term counts all year long?

5) Which months of the year do you take short-term counts as part of your coverage count program (not the special needs count program)? (please select all that apply)

The following comments were provided:

• ID: We stop counting when the snow falls. Sometimes this is mid-October. Sometimes it’s mid-November.
• IL: On rare occasions counts have been taken in November.

• NE: February and November are not always counted every year as we count as early and late as the weather allows.
• OR: We allow March counts in urban areas since the seasonal factors are flatter there and fewer studs on tires.
• PA: Count season begins March 1 (weather dependent) and ends the Thursday prior to Thanksgiving.
• SC: We attempt to have all data collected by 10/31/21 but will allow counts into the 2nd week of November if needed.

6) What is the typical duration of your short-term counts conducted as part of your coverage count program? (please select all that apply)

The following comments were made by state respondents who selected “Other (please specify)”:

• DC: ramps-2 days, volume-3 days, class- 7 days.
• IA: We do some 7-hour turning movement counts.
• IL: All counts are 24 hours in duration except for HPMS segments on roads having less than 5,000 AADT which are counted for 48 hours per FHWA recommendations.
• NJ: Turning movement counts (TMCs) for intersections, the duration is minimum of 8- hours and Maximum of 12- hours. the majority of the TMCs duration are 12-hours.
• OR: 24 hours if we have to count it using video.
• SC: Special counts of volume only for 24 hours. 48 hour vehicle classification and speed for the majority of short-term counts.
• WA: Note: 12-hour manual counts for classification are conducted and expanded for locations where mechanical counts for classification are not feasible to perform.

7) How many active permanent/continuous sites do you have?

^a The table includes the states that have at least one permanent site.

8) Are you considering to use probe data/estimates (e.g., probe AADT estimates) in the future?

The following comments were made by state respondents who selected “Yes”:

• AL: We are looking into it for the Urban locations only. Not seeing good Rural estimates at this time.
• AR: We are reviewing probe data to evaluate how accurate the data is on low ADT roadways. We are evaluating the data on micro levels because we only purchased a limited number of zones.
• AZ: We are piloting a project to investigate this possibility.
• CO: It’s something that we’ll consider with the MIRE requirement for an AADT estimate on all roads by 2026; unsure of the timeframe for when that will be implemented.
• DC: Maybe, depending on the accuracy and coverage. I don’t think we will use it in the near future. There is no mature product yet.
• IA: We are exploring the options.
• MI: Yes, based on recent update from FHWA regarding the acceptance of Probe Data it is something that we’ll be looking into for our Traffic Monitoring Program.
• MN: Pending approval from FHWA.
• MT: Yes, we would like to use it for lower volume rural roads and currently probe data is not approved for use on roads with less than 5000 AADT.
• NE: We would like to use probe AADT estimates in high-volume urban settings that are difficult to set up counters.
• NJ: Not yet, it’s proposed to be used through an Innovative Concepts tasks in the future.
• NV: This is a need. It has to be part of our plan for the future.
• OR: Probably we are. But it will take quite a bit of consideration to test out what roads we feel comfortable using that for. We actually have a meeting this week with the ODOT personnel that use RITIS.
• PA: For the local road counts we have looked into StreetLight, but the accuracy they have on lower volume roads is not there yet. The issue seems to be the continued improvements in accuracy is geared more toward higher volume roads (over 5,000).

• VT: We are interested in using probe data to estimate AADTs. We do not have a vendor in mind, yet. We use the MS2 traffic data management system and I think MS2 is looking into it. We will see where that goes.
• WA: Considering the use of probe data for high-volume urban areas that cannot be collected using permanent recorders (due to logistical or resource concerns) or short-duration mechanical counts (due to safety concerns). This would be a relatively small subset of our overall count program.
• WI: Depending on advancements in technology.

The following comments were made by state respondents who selected “No”:

• ID: While we are using probe data for congestion analysis, Idaho is predominantly a state with lower volumes. Right now there are too many concerns about the accuracy level of probe data on lower AADT roads, combined with the lower saturation levels in many parts of the state.
• IL: Possibly in the future, but not at this time.
• MS: We have looked into it, but haven’t made active plans yet.
• ND: Unknown at this time.
• NY: Probe Data estimates proposed to NYSDOT varied widely particularly under 10000 AADT. Some estimates shown were AADTs around 2000 were estimated at +/-100 percent or greater.
• OH: We already have an in-house estimation process for HPMS, our modeling & forecasting section estimates AADTs on FC 6R and FC 7 for VMT summaries.
• RI: Probe data is not as accurate as our radar detector data.
• SC: Currently, we do not plan to use probe data. As new research is conducted, we will reevaluate.
• WV: Our Traffic Monitoring Program recently lost all experienced members and is now managed by members with a year or less in experience. We are currently focusing on more important matters, but we may research this concept in the future.

9) Has your agency subscribed to or purchased probe data/estimates from one or more data vendors?

10) How many months of probe data do you have from each vendor?

11) Would you be able to support Phase II of this NCHRP project by providing us with probe data? (In Phase II, the research team will use raw point, trajectory, or trip count data from different states and vendors to develop and examine the accuracy of probe-based adjustment factors)

The following comments were made by state respondents who selected “Maybe (please explain)”:

• GA: Will need to discuss further.
• ND: Unknown at this time, I would have to follow-up.
• OH: Our office is not using the probe data, it’s being used by modeling and forecasting, so they would need to be available to provide the data.
• OR: Our unit is not the one that deals with RITIS directly. That question would have to go to our Analysis unit.
• WI: A different area of WisDOT would have to supply the data.

12) Have you (or another agency in your state) computed or examined the accuracy of adjustment factors developed from probe data?

The following comments were made by state respondents who selected “Yes (please provided more information below)”:

• NV: Reviewed FHWA research showing the accuracy with their calculations on various volumes of traffic.
• PA: Unsure. Another agency has INRIX data, but we are unsure if they have computed or examined the accuracy of adjustment factors.
• RI: Select comparisons from probe data and radar detector counts showed a disparity.

B. ADJUSTMENT FACTOR GROUPS

13) How do you calculate adjustment factors for each permanent site? (please select all that apply)

The following comments were made by state respondents who selected the third option “In-house, we use our own program/system (please provide name of program/system)”:

• AL: TDM(Traffic Data Manager) by PMG.
• AZ: We use GIS, MS2 and SQL.
• HI: TMS.
• MN: R statistical program.
• ND: Traffic Data Editing and Analysis (TDEA).
• NE: Day of week factors are created using a mainframe report each month.
• NJ: Statistical analysis.
• SC: Traffic Data Management Application (TDMA).
• WA: Historically used adjustment factors computed by Mainframe traffic data system; moving to SQL-based computation.

14) Which grouping method(s) do you use to develop adjustment factor groups? (please select all that apply)

The following comments were made by state respondents who selected “Other (please specify)”:

• AL: Inherited Geographical system, where state is divided up by interstate system. Working on updating factor groups by FC & Rural/Urban areas.
• CO: Our Urban Non-Interstate and Rural Non-Interstate contain permanent stations in FC 2 through 5. We also have two groups of summer seasonal and winter/seasonal factor groups that incorporate both interstate and non-interstate segments.
• IA: Roadway system by rural/urban: rural interstate, municipal interstate, rural primary, municipal primary, county secondary and city street.
• IL: We have factors for the 6 counties surrounding the City of Chicago (IDOT District 1) and we have factors for the remainder of the state (IDOT Districts 2 thru 9).
• NE: The rural interstate is broken down into groups between major cities/interchanges.
• NY: Per TMG - Coefficient of Variance (of MADT) >.1 between .1 and .25 and over .25 are the groupings used. Outliers are removed.
• SC: We combine roadway functional classification into what we call volume factor groups.

15) How do you develop adjustment factor groups? (please select all that apply)

The following comments were made by state respondents who selected “Other (please specify)”:

• MN: Current staff inherited past groups for day of week and month of year, but axle factor groups are now coded to functional class and urban/rural route coding.
• NE: Our factor groups were developed decades ago and have not been redeveloped since.
• OR: Varies quite a bit by factor type. MS2 program helps with some.

16) In the case of cluster analysis, which inputs do you use to develop clusters of permanent sites? (please select all that apply)

The following comments were made by state respondents who selected “Other (please specify)”:

• ID: Functional Class.
• MI: Current factor groups were developed by the previous Planner and the specifics of the analysis aren’t available.
• MN: 36 monthly day-of-week factors (Tues, Wed, Thurs times 12 months).
• WI: Truck percent.

17) In the case of cluster analysis, how do you determine the final number of clusters?

• AZ: The Traffic Section table contains a field called “ReferenceEquation,” which provides a method for relating data from specific traffic count stations onto the traffic sections. Many reference equations create a one to one relationship between a single traffic station and a traffic section. However, there are many instances where one count station can be used for multiple traffic sections. Similarly, there are situations where multiple traffic count stations can be used on a single traffic section. In these cases, algebraic equations can be used to establish the relationship between the count stations and the section. For example, a one-way count location can be multiplied by two in order to calculate the AADT for a two-way road. Another example may be taking the average of two count locations.
• ID: Trial and error and weighting things like seasonality.
• MI: Current Factor Groups were developed by the previous Planner and the specifics of the analysis aren’t available.
• MN: Current staff inherited the current number of cluster groups. These were likely last determined in 2004 using WARDs cluster analysis and engineering judgment.
• NV: The primary driver was there needs to be enough sites in each cluster to make it a robust group.
• VT: We have been using the same seasonal factor groups for many years and we find that our sites tend to fit in one or another of the existing groups, so we aren’t looking to add or remove groups.
• WA: Unsure.
• WI: K-means cluster analysis in R.

18) Do you have adjustment factor groups for lower FCs (6R, 7R, 7U)? If so, which grouping method(s) do you use to develop them? (please select all that apply)

The following comments were made by a state respondent who selected “Other (please specify)”:

• AL: Given lack of Permanent Stations to develop adequate groups, we include these routes with our FC 4 & 5 routes.
• OR: Depends on the factor type. Seasonal and growth factors are only from ATRs, so we have no lower FC factor groups. Axle factors can come from lower FC short-term counts, so they can have groups for those factors.
• WV: Our lower FCs are included in the same factor groups as any other roadway, aside from interstates. Interstates have their own factor group, one for rural, and one for urban.

C. ASSIGNMENT OF SHORT-TERM COUNTS TO ADJUSTMENT FACTOR GROUPS

19) Do you assign short-term counts to adjustment factor groups?

20) Which characteristic(s) do you use to assign counts to factor groups? (please select all that apply)

The following comments were made by state respondents who selected “Other characteristics (please specify)”:

• IA: Roadway system by rural/urban: rural interstate, municipal interstate, rural primary, municipal primary, county secondary and city street.
• ID: Local knowledge.
• MI: Professional judgment based upon local traffic characteristics.
• MN: Analyst review, engineering judgment, nearby traffic generator characteristics. As of 2020 axle factors are assigned by functional class and urban/rural.
• MT: Seasonal variation.
• NE: Again, we separate the rural interstate into segments between major cities/interchanges.
• NJ: Seasonal attractions regions of the state.
• NY: Most Urban Sites are assigned to the less varied group, rural typically assigned to medium variance. Regional folks and short counts are used to check locations for inclusion in the highest variance group Time of Day patterns, Region and Geography are used during this analysis, but not automated.
• OR: Truck percentages for some factors.
• VT: Day of week pattern helps discriminate between rural and recreational.
• WA: Factor groups are applied in our areas with moderate and strong influences due to recreational and agricultural travel.
• WI: Truck percent.

21) Are there other variables that you have considered using in the assignment process? If so, what are those variables?

• AL: None.
• AR: No.
• AZ: We also use mileage as a category.
• CO: N/A.
• GA: None at this time. Planning to a review factor group analysis in the near future.
• IA: None.
• ID: Local knowledge (ski resorts, high rec summer areas, etc.).
• IL: Not at this time.
• MI: This is a future project that MDOT is looking into starting so that we can confirm our current seasonal factor group assignments to the short count sites is appropriate.
• MN: The difficulty of doing this task on a routine basis due to the lack of geographic and count data.
• MS: none.
• MT: No.
• ND: no.

• NE: We break up the rural interstate into 6 sections for short-term count adjusting.
• OH: Yes, we have considered creating a recreation factor group for areas near Lake Erie that have heavy summer traffic.
• OR: Use ternary graphs to explore seasonal classification factors.
• PA: Not at this time.
• WA: The variables listed for question 20 – functional class (really Interstate vs. non-Interstate), geographic region, rural/urban designation, recreational & agriculturally impacted areas – serve as a starting point for the assignments, but assignments are subsequently manually maintained as count data comes in and provides more information – e.g., a rural area near an urban growth boundary may be assigned an urban factor group if patterns observed in the count data suggest that the area behaves more like an urban area than a rural one.
• WV: When it comes to our Rural locations, we also break it down to if the county is agricultural or non-agricultural.

22) Which group adjustment factors do you use to annualize short-term traffic volume counts? (please select all that apply)

The following comments were made by state respondents who selected “Other (please specify below)”:

• AL: For our 48 hour data, we utilize HoD, DOW, and MoY factors. For our 168 hour data, we utilize MoY factor only, since you only need 24 hours or each day of the week in a month to produce a MADT for that month.
• GA: Axle-correction factors – developed based on data that represents all seasons of the year which is applied to raw counts that register two axle impacts as one vehicle. Growth factors – calculated for each factor group by using historical data and linear regression analysis.
• MI: Axle adjustment factors.
• MN: Axle-correction factors.
• NY: NYSDOT Monthly Factors used also correct for Day of Week adjustments. NYSDOT collects short counts on multiple days, then uses only data from Monday 6 a.m.–Friday 12 p.m. to create an “Average Weekday” for the count, this value is adjusted to an annualized AADT.

23) Which software/system do you use to assign short-term counts to factor groups? (please select all that apply)

The following comments were made by state respondents who selected the third option “In-house, using our own program/system (please provide name of program/system)”:

• AL: TDM(Traffic Data Manager) by PMG.
• HI: TMS.
• ID: ArcGIS Python Tools developed by High Street.
• MN: Review each site spatially in ArcGIS. As of 2020 axle factors are assigned by functional class and urban/rural.
• ND: TDEA.
• NE: Since our factor groups have remained static for a long time, we only periodically review locations where either the road or land use have changed.
• PA: Assigned automatically through the Roadway Management System.
• SC: Traffic Data Management Application (TDMA).

24) Do you factor short-term counts taken on lower FCs (6R, 7R, 7U)? If yes, how? (please select all that apply)

The following comments were made by state respondents who selected “Yes, we use other characteristics (please specify)”:

• MN: Analyst review, engineering judgment, nearby traffic generator characteristics to assigned the group. As of 2020 axle factors are assigned by functional class and urban/rural.
• SC: Our volume factor groups are combined based on roadway functional classification which includes rural/urban designation.
• WA: The response for question 20 applies here as well. In this case lower functional classes will generally receive the non-interstate assignment related to the region, urban/rural, and agriculture/recreational area, and will be updated as short-term count observations provide better information related to the patterns at the location.
• WI: Cluster analysis.

25) Do you use any alternative assignment method (e.g., statistical or machine learning method)?

26) Who applies the alternative assignment method stated in the previous question and how? (please select all that apply)

27) How often do you review or change the assignment of a short-term count site to a factor group?

The following comments were made by state respondents who selected “Other frequency (please specify)”:

• MN: Never, unless we notice the factor is causing a problem with the data at a site or that spatially the network of factor assignments does not make sense.
• NE: Since our factor groups have remained static for a long time, we only periodically review locations where either the road or land use have changed.
• NY: Typically when a count is taken in an odd time of year, the seasonal Factor Group will be verified that the AADT is calculating as desired, and a group will be adjusted as needed.
• OR: Growth factors are reviewed every year. Other factors are reviewed on longer intervals. We are reviewing all pretty intensely now since we are new to MS2 and it does the process differently than our old program.
• PA: RMS assigns the group so if the variables that determine the group changed, then the factor group could change.
• SC: If the roadway functional classification changes, our system will adjust accordingly.
• VT: We do a quick check when we load the data but otherwise only when we notice something off about the AADT.
• WA: Our intent is to review all factor group compositions (and consequently assignments) every three years. However, as mentioned before, assignment of count locations is reviewed as part of the count review process (continually) and updated as necessary to accommodate patterns observed in the count data.

D. IMPLEMENTATION PROCESS

28) How likely is each of the following scenarios that involve implementing a new statistical assignment method? (please make a selection for each scenario)

29) Please order (by dragging and dropping each item) the following potential barriers associated with the implementation of a statistical method (1 is the most important and 8 is the least important).

30) What kind of information and research products (e.g., Guide; windows application; analytics website; code in R, Python, or other language; training material, etc.) would help you select and adopt an assignment method?

• AL: Windows application and/or analytics website.
• AR: Training Material and Python & R training.
• AZ: Python, R, SAS, SQL, ArcGIS.
• CO: Guide, code, training materials would help in selecting and adopting an assignment method.
• DC: SAS Code.
• GA: FHWA -Traffic Monitoring Guide, AASHTO Guide.
• HI: analytics website.
• IA: Guide, training.
• ID: We use ArcGIS Pro heavily and Python is our tool development language of choice.
• IL: We use ACCESS databases and SQL databases to store and analyze our traffic data.
• LA: FHWA guide.
• MN: Approved land use datasets (consistent and approved by TMG/FHWA), guide/training analytics website; code in R or python.
• MS: Guide, training.
• MT: Endorsement by FHWA and incorporated into MS2 software.
• ND: Unknown at this time.
• NJ: Guide, standardized software program, further understanding in coding, training material.
• NV: analytics website or training material would be helpful. Windows application would be even more helpful if we could get it.
• OR: A guide foremost, with good conceptual descriptions. The others would also be helpful.
• PA: If a guide was available to help users with the various options would be greatly beneficial. The software side is not an issue. There are resources internally that can figure out software related barriers if those occurred.
• WA: Outlines of existing statistical assignment methods; example R/Python code to implement these methods.
• WI: Guide and windows application.
• WV: I think a written guide would help most with adopting a new assignment method. We currently have limited knowledge on coding.

31) Can you please provide examples of Guidebooks or other research products that you used in the past to select and implement a new method or process?

• AL: FHWA Traffic Monitoring Guide (TMG).
• AR: N/A, only been in this position for 3 years and the methods have not changed.
• AZ: Traffic Monitoring Guide and HPMS Manual.
• CO: The traffic monitoring guide would be best resource we have used in the past to implement new methods or processes.
• GA: FHWA -Traffic Monitoring Guide, AASHTO Guide.
• ID: We hired High Street to research and develop our tool.

• IL: n/a.
• LA: FHWA guide and TMG.
• MT: TMG, FHWA WIM manual.
• ND: N/A.
• NJ: Purchased coding books for reference help (Python, SQL).
• NV: We read the guides from the FHWA and TMG on how we are supposed to be doing our factoring and grouping and then we ran the stats on our sites in Excel to start with.
• OH: Traffic Monitoring Guide, HPMS Field Manual.
• OR: Since we just moved to MS2, we are still in the process of documenting a lot of this. Since the seasonal classification is from scratch (our old program didn’t do that), I have a little more for that. But even that needs more description in it.
• PA: We referenced the calculations for traffic computations from the FHWA pocket guide.
• SC: We are finalizing a research project with the University of SC for estimating non-coverage AADT. The majority of non-coverage counts in SC are on local functional classified roads.
• VT: We refer to the Traffic Monitoring Guide and FHWA HQ for the most up-to-date guidance.
• WA: n/a.
• WI: TMG and HPMS Manual.
• WV: Traffic Monitoring Guide. HPMS Field Manual.