Feasibility and Advantages of Estimating Local Road Vehicle Miles Traveled Based on GPS Travel Data

摘要

A critical part of the US national transportation data program is the Highway PerformanceMonitoring System and the reporting of vehicle miles traveled (VMT) on different levels ofroadways. While the amount of travel on higher-level roads can often be reliably estimated fromtraffic counts and other data sources, existing heuristic methods for estimating lower-level andlocal road VMT suffer from the lack of ground truth data. This paper develops a novel methodfor estimating local road VMT from GPS and other supplemental data sources, and investigatesthe associated statistical issues. The proposed method is applicable at the national, state, and anylocal levels, and is demonstrated in a case study in Maryland. GPS survey sample size andduration required for reliable VMT estimation are also analyzed within the case study. The casestudy and statistical analysis show that a 30-day GPS survey would reduce the required samplesize by 50~60% compared to a single-day survey, and a 15-day GPS survey with 670participating drivers could provide local road VMT estimates with a 5% margin of error at the 95%confidence level. Survey designers could employ longer survey duration to further reduce samplesize or recruit more participants for a shorter survey. These findings suggest GPS-based surveysare feasible and cost-effective options for VMT estimation on different levels of roadwaysincluding local roads. Federal, state and local agencies may just use GPS-surveys alreadyplanned for other purposes (e.g., travel demand modeling and planning applications) for VMTestimation.