Home About ATSIP Program Awards Venue Sponsors/Exhibitors

Signalized Intersection Database for Safety Analysis of Left Turn Signal Phasing in Wisconsin


Abstract: As part of the research project NCHRP 03-118: Decision Making Guide for Traffic Signal Phasing, the University of Wisconsin – Madison was in charge of providing safety data representative of the Midwest region. Other jurisdictions committed to the project include Arizona, Texas, Florida, and North Carolina. Sampling and data collection are complex tasks. In Wisconsin, the Southeast region keeps a well-documented inventory of signalized intersections which have available digital copies of volume count and signal time plan reports throughout the years. The inventory consisted of 549 signalized intersections. Each intersection was carefully reviewed and 127 intersections were selected. The selection criteria consisted of crash data availability, four-leg approaches, skew angle less than 15 degrees, proximity to adjacent intersections, and significant operational/geometric/signal changes during 2005 and 2016.

The data collection was conducted for the 127 intersections sampled. The data collection consisted of intersection information, geometry, traffic, signal type, and crashes. Since there was not an integrated database, several data sources were required. The intersection information was collected from the inventory which contained detailed information of the coordinates and name of approach legs to the intersections. The geometry was collected using Google maps aerial and street views over different periods of time. The geometric variables include the number of lanes, median type, speed limit, and left turn offset.

The traffic data was collected from volume count reports usually available every three years. A total of 636 intersection traffic reports were individually reviewed to obtain peak our volumes, peak hour factors, pedestrian volumes, and AADTs. Intersection signal phase information was collected from signal time plan reports. The reports provided detailed information of changes made throughout the years. A total 1,483 signal time plan reports were individually reviewed to obtain information of the presence of video detection, vehicle/pedestrian actuation, exclusive pedestrian phasing, left turn phasing type, left turn phasing schedule, left turn phasing lead/lag, right turn overlap, and concurrent pedestrian indication.

The crash data was not readily available with intersection assignments. Thus, using reference point coding maps and GIS application, crashes within 500 ft. of the center of the intersection were collected and assigned to the corresponding intersection. Each crash contained a distance to the center of the intersection for future data processing. A total of 22,588 crashes were assigned to the sampled intersections. The development of the signalized intersection database in Wisconsin provides an overview of the complexity of assembling and integrating a consistent database from different sources for rigorous statistical safety analysis.

Boris Claros - Postdoctoral Research Associate, University of Wisconsin - Madison



Session Material
Quick Links


Chris Osbourn


fb_logo   linkedin