Intersections and Non-Intersections: A Protocol for Identifying Pedestrian Crash Risk Locations in GIS

Area of Science:

• Transportation Engineering
• Geographic Information Science
• Public Health

Background:

• Pedestrian crash analysis often uses intersection and non-intersection locations as spatial units.
• Comparing crash-risk studies is challenging due to varied data and identification methods.
• A standardized approach is needed for consistent spatial analysis in pedestrian safety.

Purpose of the Study:

• To develop a systematic and replicable protocol in GIS for creating consistent spatial units for pedestrian crash modeling.
• To compare the reliability of different algorithms and search radii for identifying crash-risk locations.

Main Methods:

• Utilized four public datasets: roadway intersection points, lanes, speed limits, and crash records.
• Developed and tested two algorithms with five search radii (20-100m) to identify unique intersection and non-intersection locations.
• Assessed protocol reliability using agreement rates between algorithm results and actual pedestrian crash data.

Main Results:

• The developed GIS protocol successfully identified 87.2% of pedestrian crash locations within a 20m radius.
• High agreement rates were achieved: 94.1% for intersections and 98.0% for non-intersections.
• A 20m buffer size demonstrated the highest performance in capturing crash-risk areas.

Conclusions:

• The protocol provides an efficient, reliable, and cost-effective method for creating spatial analysis units for pedestrian crash modeling.
• This standardized approach facilitates consistent and comparable research in pedestrian safety.
• Further research should explore additional search radii to optimize the identification of crash-risk locations.