Using Spatial Analysis to Estimate Link Travel Times on Local Roads

Abstract

Link travel times are not only an index of congestion, but also a key variable in route choice research and an important component of impedance functions in trip assignment models. Global Positioning System (GPS) devices are increasingly being used as probes to collect real-time travel time data but usually on main arterial roads. Limited research has focused on the estimation or transfer of travel times to links without GPS probe data. This research developed a systematic method to estimate link travel times for the entire road network including links with GPS probe data and links without. The model used sparse GPS probe data such as that which would be collected in a household travel survey using GPS. This paper investigates the feasibility of using the sparse GPS data to estimate link travel times for local road types with very little coverage relatively to others in the network. Categorization of road links based on time, day of week and other spatial characteristics was determined and link travel time estimation for the 4 categories of local roads with only 36.5% GPS coverage are the focus in this paper. A geostatistical Kriging model was employed for these link categories and the accuracy of the link-based results was similar to that found in previous models with more probe data (a percentage difference of 19.2% between the estimated link travel time and the observed link travel time on local roads was found).

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 18 citations exist upstream.