[VALIDATION STUDY] Origin/Destination Model and License Plate Validation
Fehr & Peers has published an excellent and detailed project report here. They did all the work involved in the validation. We have merely summarized the results most relevant to StreetLight validation questions below. The validation results from this study were excellent, with better than 95% match. However, since the study period for this project (2013), the data sample has more than tripled and we’ve added more attributes to the O/D Metrics, indicating that future results will be even stronger.
Context for Study
The Napa County Transportation and Planning Agency (NCTPA) directed the Napa County Travel Behavior Study to gather information on the travel behavior of visitors, employees, residents, and students who make work and non-work trips in Napa County. The resulting data will provide the basis for multiple planning efforts by NCTPA and other planning agencies within Napa County. Such uses include the refinement of the Napa-Solano Travel Demand Model (NSTDM) and the update of the Countywide Transportation Plan. The data is also expected to be used to help direct the expansion of transit and paratransit services in Napa County.
Traditional Methods of Study
Data for trips that pass through the region are usually collected by a license plate survey, while data for trips that start from or end inside the region is usually collected by a roadside, mail, or telephone survey. These traditional survey methods tend to be very costly and generally provide very small sample sizes. They are also prone to human error during the data collection process as well as from the survey responders who may misinterpret the questions. Therefore, Fehr & Peers evaluated several alternative data sources, including StreetLight InSight® Travel Metrics as an alternative.
StreetLight Metrics Used
In order to support the various components of the project, Fehr & Peers and StreetLight ran several StreetLight InSight Metrics. All of these Metrics are derived from GPS data:
- Transportation Analysis Zone (TAZ) Origin-Destination Matrix. For each TAZ in Napa County (as well as surrounding counties) how many trips from each TAZ (eg. TAZ A) ended in another TAZ (e.g. TAZ B)?
- Select Link Origin-Destination and Internal/External Matrix for road segments crossing Napa County border. For each of the road segments, what percent of trips were Internal-to-External, External-to-Internal, or “Pass Through” Napa without stopping (External-to-External)? In addition, for those trips, what was their TAZ Origin or Destination?
- Select Link for Routing: For trips using a particular road segment, how many end up at another road segment later in the trip?
For this project, approximately 867,000 daily trips were measured, which was then used to create 108 stratified origin-destination trip tables, each consisting of approximately 440,000 cells of trips.
"For these inter-regional trips, StreetLight data was able to provide the county of origin and destination for trips that started or ended outside of Napa County, which is typically very difficult to obtain but required for SB 375” - F&P Final Report
In order to check the validity of several metrics, Fehr & Peers ran a license plate scan program on one Friday on certain road segments in Napa County. This allows Fehr & Peers to measure the percent of trips entering the county on various road segments that were “pass through” or “external-to-external trips.” The results from these sensors were compared to the StreetLight InSight results and found to be a very close match.
The Table below shows StreetLight InSight Pass-Through trips (designated as “Mobile Device Data”) compared to calibration data from a License Plate survey scanning various points in Napa County. For Friday-to-Friday comparison, the Personal and Total values are within 9% and 5% respectively.
Fehr & Peers also compared output from StreetLight InSight to the Napa STDM, and origin-destination matrix previously derived from modeling, surveys, and more. The results indicated that “the observed daily mobile device total daily trip data very closely resemble forecasted weekday daily total daily trip data from the 2010 CCTA Model.” The overall results are shown in Table 16 (using the report’s Table numbers) below.
Thanks to Fehr & Peers for their excellent work on this project!